|Home | About | Journals | Submit | Contact Us | Français|
The purple anoxygenic photosynthetic bacterium Rhodobacter capsulatus is capable of growing in aerobic or anaerobic conditions, in the dark or using light, etc. Achieving versatile metabolic adaptations from respiration to photosynthesis requires the use of tetrapyrroles such as heme and bacteriochlorophyll, in order to carry oxygen, to transfer electrons and to harvest light energy. A third tetrapyrrole, cobalamin (vitamin B12), is synthesized and used as a cofactor for many enzymes. Heme, bacteriochlorophyll and vitamin B12 constitute three major end products of the tetrapyrrole biosynthetic pathway in purple bacteria. Their respective synthesis involves a plethora of enzymes several that have been characterized and several that are uncharacterized as described in this review. To respond to changes in metabolic requirements the pathway undergoes complex regulation to direct the flow of tetrapyrrole intermediates into a specific branch(s) at the expense other branches of the pathway. Transcriptional regulation of the tetrapyrrole synthesizing enzymes by redox conditions and pathway intermediates is reviewed. In addition, we discus the involvement of several transcription factors (RegA, CrtJ, FnrL, AerR, HbrL, Irr) as well as the role of riboswitches. Finally, the interdependence of the tetrapyrrole branches on each others synthesis is discussed.
Cyclic tetrapyrroles encompass porphyrins, such as heme, chlorophyll and bacteriochlorophyll (Bchl), and porphynoids, which are more reduced. The porphynoid class consists in the corrinoids (cobalamin), siroheme, heme d1 and coenzyme F430 (Frankenberg, Moser and Jahn 2003). Many species are capable of synthesizing numerous tetrapyrrole end products. One of the best studied organisms is Rhodobacter capsulatus, an anoxygenic photosynthetic bacterium of the α-proteobacteria subfamily. It is capable of growing in a variety of different environmental conditions such as dark aerobic, dark and light anaerobic, etc. This versatile metabolism requires the capability of synthesizing the tetrapyrroles heme and Bchl that are needed for respiratory or photosynthetic growth. Heme is needed for electron transfer during both respiration and photosynthesis while Bchl is the major pigment involved in collecting and converting light energy into biochemical energy during photosynthesis. In addition to synthesizing heme and Bchl, R. capsulatus also synthesizes cobalamin [vitamin B12 (vitB12)]. VitB12 and its derivatives are the most complex tetrapyrrole synthesized in nature, and are involved in various cellular functions such as DNA repair and methionine synthesis (Martens et al. 2002). The synthesis of siroheme in a R. capsulatus strain has been reported as well (Olmo-Mira et al. 2006).
The four tetrapyrrole end products synthesized by R. capsulatus, are derived from different branches that offshoot from a common tetrapyrrole biosynthetic pathway “core” (fig. 1). A common trunk of the pathway used by all three branches involves a multistep conversion of δ-aminolevulinic acid (δ-ALA) into uroporphyrinogen III (uro’gen III). At uro’gen III, the “corrin branch” offshoots leading to the synthesis of cobalamin. Uro’gen III is also the potential substrate in siroheme synthesis. The central core of the pathway continues to synthesize intermediates up to protoporphyrin IX (proto IX) that are common to both the heme and Bchl branches. At proto IX, the heme and Bchl branches differentiate either through the insertion of Fe to form heme or the insertion of Mg to form Mg-proto IX which is the first committed step in the Mg-branch that leads to the production of Bchl.
The biosynthesis of heme, vitB12 and Bchl is tightly controlled in order to achieve the adequate Bchl/heme/vitB12 ratio that meets metabolic requirements of the cell. Heme and cobalamin levels appear to be maintained at rather stable levels while the amount of Bchl swings dramatically depending on the presence or absence of environmental oxygen. Moreover, a number of the intermediates in each of the three branches are potentially toxic as they can generate reactive singlet oxygen as a byproduct of light absorption. Tuning the relative amount of Bchl, heme and vitB12 as a function of the metabolic requirements, while coping with the potential toxicity of intermediate products requires this pathway to be highly regulated. Early work on the regulation of pigment synthesis in purple bacteria was initiated in the 1950’s by Cohen-Bazire (Cohen-Bazire, Sistrom and Stanier 1957). More recently it has been demonstrated that the tetrapyrrole biosynthetic pathway in purple bacteria is highly complex and its regulation extremely sophisticated. This review is centered on the model species R. capsulatus and attempts to identify the actors involved in this pathway, both synthesizing enzymes and regulators.
Examination of the complete genome of R. capsulatus has enabled the identification of the genes involved in each branch of tetrapyrrole synthesis (cf. annotations available at http://www.ergo-light.com and http://onco.img.cas.cz/rhodo/results/index.html). While genes involved in the vitB12 and Bchl synthesis are found in large operons, genes encoding the enzymes involved in heme synthesis are scattered over the genome (Young et al. 1989, McGoldrick et al. 2002, Warren et al. 2002, Smart, Willett and Bauer 2004). This feature is not unusual, though a few organisms show an operon organization for some of their hem genes (Avissar and Moberg 1995). The one exception in R. capsulatus are the hemC and hemE genes that are divergently transcribed and separated by 111 bp (Smart, Willett and Bauer 2004).
The biosynthetic pathway to produce uro’gen III from δ-ALA is very conserved in nature and is therefore considered as universal. However there are two different pathways to synthesize δ-ALA, which is the precursor of all known tetrapyrroles. One is the C-5 pathway where two enzyme reactions transform glutamyl-tRNAGlu into δ-ALA. The other is the C-4 or Shemin pathway that involves the condensation of glycine and succinyl-CoA to produce δ-ALA, CO2 and CoA (Avissar and Moberg 1995, Frankenberg, Moser and Jahn 2003). As a typical representative of the α-proteobacteria subfamily, R. capsulatus synthesizes δ-ALA using the C-4 pathway. This reaction is performed by the δ-ALA synthase (EC 188.8.131.52, ALAS), encoded by the hemA gene, that was identified and cloned as in 1988 (Biel, Wright and Biel 1988). R. capsulatus ALAS is a piridoxal-5′-phosphate-dependent homodimer, consisting in two 44kDa subunits. It shows 49% of identity and 70% of similarity with the human erythroid-specific ALAS. The structure solved in 2005 by Astner et al., provided numerous information on mutations that generate dysfunctional forms of the human eALAS (Astner et al. 2005). Moreover, despite a very low primary sequence identity of 18%, ALAS shows remarkable three-dimensional homology with the Glutamate-1-semialdehyde-2,1-aminomutase (GSAM) of Thermosynechococcus elongatus. The GSAM is the last enzyme of the C-5 pathway and, as such, synthesizes δALA. It is likely that the C-4 pathway originated from the evolution of GSAM into ALAS, followed by the loss of the C-5 pathway in common ancestors of α-proteobacteria and of early mitochondria (Schulze et al. 2006). An ORF encoding a putative GSAM, hemL, was found in the genome of R. capsulatus. In prokaryotes, the occurrence of two uncoupled δALA biosynthetic pathways was shown in Streptomyces nodosus subsp. asukaensis, where the C-5 pathway is producing δ-ALA for heme synthesis, while the C-4 pathway generates δ-ALA targeted to secondary metabolism (Petricek et al. 2006). In R. capsulatus there is no evidence of an alternative δ-ALA synthesis pathway as the first enzyme of the C-5 pathway, glutamyl-tRNA reductase, was not found in the genome. Furthermore, the annotation of this ORF as a hemL gene has to be confirmed experimentally as GSAM encoding genes shares a lot of similarity with genes coding for other types of aminotranferases (Panek and O’Brian 2002). Interestingly, a δ-ALA-requiring mutant containing a single-insertion was characterized, showing 10% of the ALAS activity of the parental strain remaining (Wright, Cardin and Biel 1987). Finally, the synthesis of δ-ALA appears as one of the crucial regulation point of the tetrapyrrole biosynthesis pathway, the transcription of hemA being up-regulated under low oxygen tension (Wright et al. 1991, Smart, Willett and Bauer 2004) and down-regulated in the presence of excess of exogenous heme (Smart and Bauer 2006).
The first common step in tetrapyrrole synthesis is achieved by porphobilinogen synthase (EC 184.108.40.206, PBGS), or δ-ALA dehydratase, that converts two δ-ALA molecules into a monopyrrole porphobilinogen. In R. capsulatus, hemB encodes a 35.8 kDa protein which presents unusual characteristics among the PBGS studied so far. Indeed, most PBGS are known to be homooctameric metalloenzymes presenting a high variability regarding their cations requirement in terms of catalysis (Zn2+, Mg2+ or K+ binding site at the active site) and activity regulation, most of them containing an extra allosteric Mg2+ binding site. The biochemical characterization of R. capsulatus revealed that its enzyme is a homohexamer clearly independent of cations (Zn2+, Mg2+, K+) or chelators (EDTA, 1,10-phenathroline). R. capsulatus PBGS seems to switch between a highly active homohexameric form and a less active homodimeric form (Nandi and Shemin 1973, Bollivar et al. 2004). Regarding metal-dependence, this feature was partially predicted from the primary sequence. Indeed, among 130 PBGS sequences representing the entire tree of life, only the ones from the Rhodobacter genus, namely R. capsulatus and R. sphaeroides, lack the cysteine-rich motif involved in Zn-binding at the active site, and the conserved arginine and glutamate residues that coordinate the allosteric Mg. It is postulated that PBGS containing Zn at the active site evolved to other configurations (metal-free, Mg and/or K containing active site) during the set up of the photosynthetic process and the early synthesis of Bchl. Indeed, Zn is a competitor of Mg in the metallation step of proto IX (Jaffe 2003, Frère et al. 2005). If the PBGSs of the Rhodobacter genus fit well in this model regarding their active site, the evolutionary pressure(s) that let them lose the totality of their metal content, including the Mg in the allosteric site, remain(s) unknown. Moreover, despite porphobilinogen accumulation under low oxygen tension in R. capsulatus, the activity of PBGS is unregulated by oxygen, proto IX and hemin in R. capsulatus. In addition, no or only a slight increase in hemB transcription was observed upon decreasing the oxygen tension (Biel et al. 2002, Smart, Willett and Bauer 2004). However, the conversion of δ-ALA into porphobilinogen seems to be a major control point in the tetrapyrrole biosynthesis pathway (Biel 1992). This may be achieved by controlling the level of porphobilinogen through the availability of the substrate of the PBGS, i.e. δ-ALA. Indeed, the vast majority of the δ-ALA synthesized is transformed into aminohydroxyvalerate by an oxygen-dependent δ-ALA dehydrogenase (Biel et al. 2002). Lastly, although not occurring in a manner as dramatic as for hemA, hemC or hemE, it was shown that the transcription of hemB is repressed in the presence of exogenous heme (Smart and Bauer 2006).
The transformation of monopyrrole porphobilinogen into the cyclic tetrapyrrole uro’gen III is catalyzed by the enzymes porphobilinogen deaminase (EC 220.127.116.11, PBGD) and uro’gen III synthase (18.104.22.168, UROS) encoded by hemC and hemD, respectively. During the first step, the PBGD polymerizes four molecules of porphobilinogen into a linear tetrapyrrole, pre-uro’gen or 1-hydroxymethylbilane, and four NH3. The main product, pre-uro’gen, is very unstable and, if exposed to the solvent, will be converted chemically into a toxic cyclic porphyrin, uro’gen I. To avoid toxicity pre-uro’gen is passed to UROS that presumably interacts with PBGD to limit exposure of the cell to free pre-uro’gen (Shoolingin-Jordan 1995). UROS catalyzes the formation of uro’gen III, by circularizing the linear tetrapyrrole while inverting ring D. An asymmetric ring D is a signature of biologically functional porphyrins.
PBGD was purified from R. capsulatus and the N-terminal sequence was determined to identify its corresponding gene, hemC. The latter was also expressed heterologously in Escherichia coli, confirming this ORF is sufficient to produce an active form of PBGD (Biel et al. 2002). No biochemical characterization of the R. capsulatus enzyme has been performed so far but the analysis of the sequence indicates that hemC corresponds to a 34.1 kDa protein with 47% of homology with PBGDs from E. coli and Pseudomonas aeruginosa, and 43% with the Bacillus subtilis homolog. In addition, previous characterizations of PBGDs highlighted the use of a unique cofactor, dipyrromethane which consists in two porphobilinogen molecules. This cofactor acts as a primer during the polymerization of a hexapyrrole, which finally releases the tetrapyrrolic reaction product and the intact cofactor. R. capsulatus PBGD contains the conserved cysteine residue that is involved in the binding of the cofactor. More work has been done regarding the genetics. Indeed, characterization of an hemC mutant confirmed it was incapable of synthesizing neither siroheme nor vitB12 and grew only in media supplemented with cysteine and methionine, in addition of hemin (Biel et al. 2002). In addition, the expression of hemC was shown to be regulated by oxygen, as seen by a 5-fold increase of the transcription level from aerobic to anaerobic growth conditions (Smart, Willett and Bauer 2004). Finally, the transcription of hemC was repressed two thirds upon the addition of exogenous hemin (Smart and Bauer 2006).
Characterization of UROS is very unclear, as it was never purified from R. capsulatus. A high variability in the primary sequences of UROSs has made it difficult to identify the hemD gene in the R. capsulatus genome. In addition, the lack of genetic organization such as a hemCD operon or a hem gene cluster, like in E. coli or B. subtilis, renders the task even harder (Avissar and Moberg 1995). Recently, an ORF in the R. capsulatus genome was annotated as encoding an UROS. Because hemD genes have been mis-annotated in many other organisms, experimental work has to be undertaken to confirm this annotation (Panek and O’Brian 2002).
The first step after the uro’gen III crossroad consists in the synthesis of coproporphyrinogen III (copro’gen III). This activity is performed by uro’gen III decarboxylase (EC 22.214.171.124, UROD) that catalyzes the sequential decarboxylation of the four acetate side chains into methyl groups. The reaction is ordered, starting with the ring D and then decarboxylating rings A, B and finally ring C. Numerous UROD encoding genes from various biological sources have been identified showing a rather high level of conservation and presumably a common mechanism of action (Frankenberg, Moser and Jahn 2003, Heinemann, Jahn and Jahn 2008). The gene coding for UROD, hemE, was identified in 1995 from R. capsulatus, (Ineichen and Biel 1995). The deduced protein sequence shows more than 34% of identity with formerly characterized UROD and contains a signature sequence of the class of enzyme (PXWXMRQAGR) at the amino-terminal. No biochemical work has been performed with R. capsulatus UROD but a bit is known regarding the genetics and its regulation. An interesting feature of hemE is its linkage to hemC, while the other hem genes are scattered on the genome. Those two genes are divergently transcribed and separated by an intergenic region of 111 bp. Furthermore, the transcription of hemE is strongly dependent on redox conditions, as seen using hemE::lacZ fusion. Expression of hemE increases by 112- and 52-fold under semi-aerobic and anaerobic conditions, respectively (Smart, Willett and Bauer 2004). As the case of hemC, the transcription level of hemE is decreased by two thirds in the presence of exogenous heme (Smart and Bauer 2006).
An interesting phenomenon about this step of the pathway is that the cells excrete copious amounts of the product of the reaction, copro’gen III under certain growth conditions. Cooper highlighted this feature in R. capsulatus, using cells grown under iron starvation or excess of exogenous methionine (Cooper 1956, 1963). More recently, various R. capsulatus mutants, either unable to synthesize Mg-protoporphyrin monomethyl ester or lacking c-type cytochromes, were studied and showed such a phenotype. Moreover, it was found out that coproporphyrin is excreted as a complex with a 66 kDa protein (Biel and Biel 1990, Biel 1991) that turned out to be the major outer membrane porin (Bollivar and Bauer 1992).
The antepenultimate and penultimate steps of heme synthesis are achieved by the copro’gen III oxidase (EC 126.96.36.199, CPO) and proto’gen IX oxidase (EC 188.8.131.52, PPO), respectively. These steps lead to the formation of protoporphyrin IX that is a precursor for the synthesis of heme and Bchl. Oxygen-dependent and oxygen-independent CPOs conventionally named HemF and HemN (or HemZ), respectively, are known to catalyze this reaction. HemF is prevalent in eucaryotes and in a few bacterial groups (primarily cyanobacteria and proteobacteria) while HemN is widely distributed among prokaryotes. These two forms of CPOs share no obvious primary sequence homology indicating that they arose from distinct evolutionary processes. Both enzymes convert copro’gen III into proto’gen IX by decarboxylating the propionate side chains of rings A and B consecutively to yield vinyl groups (Panek and O’Brian 2002, Frankenberg, Moser and Jahn 2003, Heinemann, Jahn and Jahn 2008). Careful examination of the genome of R. capsulatus revealed no HemF encoding gene but three ORFs that may code for putative HemNs. This is contrasted by the genome of the phylogenetically related species R. sphaeroides that contains one hemF and three hemN genes (locus tags in img.jgi.doe.gov: RSP_0682, RSP_0699, RSP_1224, RSP_0317).
HemN contains a [4Fe-4S] cluster for catalysis and requires S-adenosyl-L-methionine (SAM) as a cofactor or co-substrate. As such, this enzyme belongs to a family of “radical SAM” enzymes that carry the signature sequence CXXXCXXC, where the cysteine residues are involved in the coordination of the Fe atoms (Layer et al. 2005). This motif is present in the deduced protein sequences of the three putative CPOs. Occurrence of multiple hemN genes in one organism has been reported and several bacterial oxygen-independent CPOs have been shown to be expressed and active in the presence of oxygen (Rompf et al. 1998, Schobert and Jahn 2002). The expression of hemN1, designed hemZ in the referred articles, was studied as a function of aeration, which showed only a moderate influence. Indeed, a roughly two-fold variation was recorded, with a maximum transcription level under semi-aerobic conditions (Smart, Willett and Bauer 2004). In addition, the presence of exogenous heme in these growth conditions decreased the expression of the gene by a two-fold factor, making it back to its basal transcription level (Smart and Bauer 2006). Finally, in silico analysis of α-proteobacterial genomes revealed an “Iron-Rhodo-box” in the promoting area of the hemN2 gene. This predicts hemN2 as part of the iron regulon (Rodionov et al. 2006).
The next step in the pathway is the aromatization of proto’gen IX by removal of six electrons to yield proto IX. This is performed by proto’gen IX oxidase (PPO) that exists as two forms, oxygen-dependent (HemY) or oxygen-independent (HemG). Most knowledge about PPOs deals with eukaryotic enzymes that are oxygen-dependent and inhibited by diphenyl ether herbicides in plants (Dailey 2002, Heinemann, Jahn and Jahn 2008). In prokaryotes, the situation is very unclear as the PPO encoding gene is unidentifiable in many genomes and, so far, hemG seems to be limited to six genera’s within the γ-proteobacteria group (Panek and O’Brian 2002, Frankenberg, Moser and Jahn 2003). As is the case of many other heme synthesizing prokaryotes, a PPO encoding gene is not identified in the genome of R. capsulatus nor in R. sphaeroides (cf. img.jgi.doe.gov). Even if the oxidation of proto’gen IX into proto IX can happen chemically, an enzyme-mediated catalysis is most probably required. Indeed, on the one hand, hemY and hemG mutants have been shown to be heme defective in B. subtilis and E. coli, respectively (Panek and O’Brian 2002). On the other hand, this oxidation has to occur in anaerobic conditions as well, when the need for Bchl peaks. So there must be either an unknown PPO-encoding gene or an already known gene carrying such an activity. The ERGO annotation of the R. capsulatus genome recently indicated an ORF encoding a “NAD(FAD) utilizing dehydrogenase with a possible PPO activity”. This has yet to be experimentally confirmed.
The synthesis of heme from proto IX is performed by the enzyme ferrochelatase (EC 184.108.40.206, FC) that chelates ferrous iron and insert it into the center of the tetrapyrrole ring. The core of the enzyme is very conserved between bacteria, plants and mammals, although bacterial FCs do not contain [2Fe-2S] clusters (Frankenberg, Moser and Jahn 2003, Heinemann, Jahn and Jahn 2008). Interestingly, FC and the anaerobic cobalt chelatase CbiK are both class 2 metal ion chelatases that exhibit a similar 3-dimensional structure (Schubert et al. 2002). Recently, Masoumi et al confirmed in vivo what Koch et al predicted in silico that PPO and FC form a stable PPO-FC complex (Koch et al. 2004, Masoumi et al. 2008). Such a configuration may ensure the channeling of the potentially toxic proto IX to FC. Finally, it has been suggested that this complex might include the CPO, to form a tripartite association CPO-PPO-FC (Dailey 2002, Koch et al. 2004).
In R. capsulatus, the FC encoding gene, hemH, was described and cloned in the 1990’s with the deduced protein sequence exhibiting 41% (59%) and 21% (44%) of identity (similarity) with E. coli and B. subtilis FCs (Kanazireva and Biel 1996). Moreover, homologous expression of a second copy of hemH induces a dramatic reduction of the tetrapyrrole pool, with 2.3-fold and 5-fold decreases of the porphyrin and the Bchl concentrations, respectively. In addition, the ALAS activity is also decreased by a factor 2.3 in these conditions (Kanazireva and Biel 1995). This phenotype is a signature of a negative feedback operated by heme on the whole pathway, as it was studied more recently (Smart and Bauer 2006). The transcription of hemH was analyzed as well, highlighting a unique responsive behavior to oxygen among the other R. capsulatus hem genes. Indeed, while the expression level is rather constant between aerobic and anaerobic conditions, it is three-fold lower in semi-aerobic condition (Smart, Willett and Bauer 2004). Finally, in semi-aerobic conditions, hemH transcription was found independent of the addition of exogenous heme (Smart and Bauer 2006).
The RegA-RegB system is a global regulator in R. capsulatus that has been shown to regulate synthesis of numerous cellular processes as a function of redox conditions (reviewed in Elsen et al. 2004, Wu and Bauer 2008). As shown in figure 2, studies of the transcription of various hem genes in a regA mutant strain highlighted that hem genes are part of the RegA-RegB regulon (Smart, Willett and Bauer 2004). Indeed, the most significant effect was observed on hemE and hemH in which transcription was activated 5–10-fold by RegA. A strong activating role of RegA was also observed with hemA but only in semi-aerobic and anaerobic conditions. Moderate influences were reported on hemC and hemZ. Finally, the transcription of hemB does not seem to be regulated by RegA (Smart, Willett and Bauer 2004). Overall, the dramatic transcriptional activation of hemE and hemH highlight the crucial presence of the RegA-RegB system at key steps of the tetrapyrrole synthetic pathway. Activating the transcription of these genes ensures the constitution of a constant stock of mRNA for a quick synthesis of UROD and FC when the flow of tetrapyrrole synthesis needs to be directed towards the porphyrin branch rather than the corrin branch, and when heme needs to be produced at the expense of Bchl. Finally, the RegA-induced increase in the transcription of hemA under semi-aerobic and anaerobic conditions underlines a function of the RegA-RegB system, namely stimulating the production of the general tetrapyrrol precursor δ-ALA in coordination with the synthesis of the pigment-binding proteins of the photosystem.
Present in the photosynthesis gene cluster is a redox-responding transcription factor called CrtJ or PpsR depending on the species. Studies demonstrated that CrtJ/PpsR is an aerobic repressor of heme (hem), Bchl (bch), and carotenoid (crt) biosynthesis genes in R. capsulatus (Ponnampalam and Bauer 1997, Elsen, Ponnampalam and Bauer 1998, Smart, Willett and Bauer 2004). CrtJ also aerobically represses synthesis of light harvesting-II peptides (puc) that binds Bchl, and carotenoids in these species (Ponnampalam, Buggy and Bauer 1995). CrtJ cooperatively binds to two copies of the palindromic sequence TGT-N12-ACA that are present in many hem promoter regions (Ponnampalam and Bauer 1997, Elsen, Ponnampalam and Bauer 1998, Ponnampalam, Elsen and Bauer 1998, Smart, Willett and Bauer 2004).
The characterization of a crtJ mutant strain has revealed its involvement in the transcriptional control of several hem genes, summarized in figure 2 (Smart, Willett and Bauer 2004). The strongest regulation is the activation of hemH expression, whatever the redox condition. CrtJ is also required to induce the translation of hemZ under semi-aerobic conditions. In the same redox conditions, CrtJ exhibits a dual role of repressor of hemC and activator of hemE at the same time. This is puzzling as these genes share a common promoting area, containing a CrtJ-binding site (Smart, Willett and Bauer 2004). Elucidating how, in constant redox conditions, the interaction with this promoter induces two opposite events will be a challenge.
The redox regulator FnrL appeared to have only moderate effect on transcription of hem genes (fig. 2), as either an activator or a repressor, depending on the gene and/or on redox conditions (Smart, Willett and Bauer 2004). This makes it hard to clearly define its mode of action and function in the regulation of tetrapyrrole synthesis. On the one hand, FnrL significantly represses hemA in aerobic conditions and hemB and hemC in semi-aerobic and anaerobic conditions. On the other hand, it dramatically activates the transcription of hemE in semi-aerobic conditions. FnrL seems to be mostly involved in the early steps of the tetrapyrrole synthesis, the universal trunk and the first step of the porphyrin branch. It was shown in other photosynthetic bacteria, such as R. sphaeroides and Rubrivivax gelatinosus, to have a crucial role on the transcription of hemZ, which did not appear to occur in R. capsulatus (Smart, Willett and Bauer 2004, Ouchane et al. 2007).
The aerobic repressor AerR (Dong et al. 2002) has a limited although important role in the transcriptional regulation of the tetrapyrrole synthesis, consisting in the activation of hemE and hemH whatever the redox conditions and the repression of hemC in semi-aerobic conditions (fig. 2). Overall, AerR seems to back up CrtJ in its target genes, with the exception of hemZ (Smart, Willett and Bauer 2004).
The Heme-binding regulatory LysR-type (HbrL) protein was identified in 2006 from the screening of a cosmid library for suppressors defective in heme synthesis using hemB as a reporter (Smart and Bauer 2006). From these experiments, an 804 bp ORF was isolated and cloned and the deduced amino-acid sequence showed a LysR-type Transcriptional Regulator Type (LTTR) signature: a Helix-Turn-Helix DNA-binding domain at the N-terminal followed by a LTTR substrate-binding motif, or co-inducer-binding motif. In addition to be ubiquitous in prokaryotes, the LTTRs represent the largest family of DNA-binding protein. They regulate functions as diverse as metabolism, cell division, quorum-sensing, oxidative stress, virulence, nitrogen fixation, etc… (Maddocks and Oyston 2008). As shown in figure 2, HbrL appeared to be involved in the heme-mediated control of the expression of hemA, hemB and hemN1 (hemZ). Whereas it acts as a strong activator of hemA and hemN1 in the absence of exogenous heme, it is a moderate repressor of hemB. The recombinant HbrL was obtained from heterologous expression in E. coli and purification. It exhibited heme-binding properties either in vivo by supplementing the expression medium with δ-ALA or in vitro by adding heme to the crude lysate containing the apoprotein. Finally, the recombinant HbrL was found to interact with the promoting region of the genes it regulates and the DNA-binding properties were affected by the presence or absence of heme (Smart and Bauer 2006). Interestingly, out of seven strains from the laboratory freezer, only four displayed an intact HbrL encoding gene (unpublished observation).
The Iron response regulator (Irr) is a transcription factor related to the Ferric-uptake regulator (Fur) family of metalloregulators. It has been extensively studied in Bradyrhizobium japonicum where it is a global regulator of iron homeostasis that acts in particular as a repressor of heme synthesis in iron-limited conditions. It is constitutively expressed and its activity mostly conditioned by post-translational modification. Indeed, its degradation is triggered by the binding to both redox status of heme: the ferric form binds to an N-terminal Heme Regulatory Motif (HRM) while the ferrous forms interacts with a Histidine-rich domain (Small, Puri and O’Brian 2009). In the genome of R. capsulatus, an ORF was annotated as a Fur transcription factor (rcc02670 in http://onco.img.cas.cz/rhodo and RRC01140 in http://www.ergo-light.com/). Analysis of the sequence revealed it belongs to the Irr subgroup rather than to the Fur one stricto sensu, as seen from divergence in key residues involved in metal coordination in the Fur group (Rudolph, Hennecke and Fischer 2006). Compared to the well-characterized B. japonicum Irr, it does not carry any HRM, which indicates a presumably different mechanism of action. On the opposite, the DNA-binding helix is very well conserved. Rather importantly, it is noteworthy that, among the α-proteobacteria subfamily, R. capsulatus is the only representative that carries Irr as the only regulator involved in the iron-responsive network. The usually master regulators Fur and RirA are indeed missing (Rodionov et al. 2006). This fact indicates a potential major role of Irr in R. capsulatus regarding the management of iron metabolism. Preliminary studies pointed out that the irr-deleted strain of R. capsulatus presents abnormal levels of heme: approximately 25% less and 25% more than the wild-type strain in aerobic and photosynthetic conditions, respectively (unpublished data).
Siroheme, the prosthetic group of various reductases, was reported to occur in R. capsulatus strain E1F1. A 17 kb region contains genes encoding an assimilatory nitrate reduction system, which consists in: (i) putative regulatory genes nsrR and nasTS; (ii) an ABC-type nitrate transporter coded by nasFED; (iii) genes coding for the apo-nitrate and -nitrite reductases, nasA and nasB; (iv) a gene encoding a siroheme synthase, cysG, responsible of synthesizing the cofactor of the nitrite reductase (Pino et al. 2006). Although the siroheme synthase was not biochemically studied, the nitrite reductase NasB was heterologously expressed in E. coli with purified recombinant protein exhibiting a spectral signature of siroheme (Olmo-Mira et al. 2006). CysG (EC 220.127.116.11) was characterized in Salmonella enterica and in E. coli where it was shown to be a bimodular homodimer that catalyses the four reactions that converts uro’gen III into siroheme (Stroupe et al. 2003).
Cobalamin synthesis requires more than thirty enzymatic steps with genes that encode these enzymes taking up approximately 1% of a typical bacterial genome (Roth et al. 1993, Martens et al. 2002). There are two distinct cobalamin biosynthesis pathways that exist in different microorganisms; the well-studied aerobic pathway, represented by Pseudomonas denitrificans, and the partially-resolved anaerobic pathway, represented by Salmonella typhimurium, Bacillus megaterium and Propionibacterium shermanii (Scott and Roessner 2002). The two pathways diverge at precorrin-2, a compound synthesized by dimethylation of uro’gen III (Raux et al. 1999). These two pathways primarily differ in their requirement for molecular oxygen and the timing of cobalt insertion. In the aerobic pathway, molecular oxygen is incorporated before the ring contraction, where the carbon bridge (C20) between the rings A and D is excised, and the cobalt chelation occurs nine steps after the synthesis of precorrin-2. In contrast, in the anaerobic pathway, cobalt is inserted directly into precorrin-2. Ring contraction is likely facilitated by the different oxidation states of the cobalt ion (Martens et al. 2002, Warren et al. 2002). The existence of distinct cobalamin synthesis pathway is also reflected at the genetic level with organisms containing genes unique to one pathway or the other. Interestingly, R. capsulatus carries genetic hallmarks for an aerobic pathway although cobalamin synthesis occurs under both aerobic and anaerobic conditions (McGoldrick et al. 2002). A cobG gene, encoding a mono-oxygenase, is missing in R. capsulatus. This mono-oxygenase catalyzes the contraction of the corrin ring, which is a characteristic reaction in the aerobic pathway. Instead, an isofunctional enzyme, CobZ was identified (McGoldrick et al. 2005), which may mediate the ring contraction process under both aerobic and anaerobic conditions. Details of enzymes at additional steps of this pathway are beyond the scope of this review but can be obtained from a recent review of this complex branch in the pathway by Warren (Warren and Deery 2009).
The regulation of cobalamin biosynthesis is mostly reported to occur post-transcriptionally via an RNA structure called a riboswitch. Riboswitches are metabolite-binding regions located within the 5′ UTR of messenger RNAs that function as regulatory elements for target genes (Mandal et al. 2003). Riboswitches are widely spread in prokaryotes and interact with various target metabolites. In Bacillus subtilis, seven different metabolites, including guanine, adenine, lysine, thiamine pyrophosphate, FMN, SAM and adenosylcobalamin, regulate at least 68 genes via a riboswitch mechanism (Mandal et al. 2003). In cobalamin biosynthesis, adenosylcobalamin is known to be an effector regulating translation of several cobalamin genes (Nahvi et al. 2002, Rodionov et al. 2003). A B12 binding element exists within the 5′-UTR of btuB mRNA in E. coli and within the 5′-UTR of cobalamin biosynthesis operon in S. typhimurium that binds adenosylcobalamin (Nahvi, Barrick and Breaker 2004). At elevated adenosylcobalamin concentrations, the translation of cobalamin-transport protein (BtuB) and cob operon is repressed by adenosylcobalamin binding to the B12 box (Nahvi, Barrick and Breaker 2004). Such interactions induces a conformational change in the mRNA secondary structure at the B12 box that prevents ribosome binding (Nou and Kadner 2000). Rodionov et al employed the combination of a comparative approach of gene regulation, positional clustering and phylogenetic profiling for identifying cobalamin regulated biosynthesis/transport genes. In this work, they identified a conserved RNA structure called the B12 element that is widely distributed in eubacteria (Rodionov et al. 2003). Forexample, 13 B12 elements are present in UTR regions of the R. capsulatus genome with potential target genes including cobW, cbiMNQO, btuFCD, btuD, and btuB, as well as upstream of several non-cobalamin biosynthesis/transport genes (Rodionov et al. 2003).
Examples of cobalamin genes regulated at the transcription level are not abundant. In S. typhimurium the global regulators Crp/cAMP and ArcA/ArcB, are responsible for indirect redox and carbon controls of cobalamin biosynthesis by controlling synthesis of the positive regulatory protein, PocR. PocR subsequently uses propanediol as an inducer of cob gene expression (Rondon and Escalante-Semerena 1992, 1996). Examples other global regulators directly regulating cob genes are not documented.
Recently, we observed that the global RegA-RegB signal transduction cascade not only controls expression of the Bchl and heme branches, but also controls expression of enzymes at two steps in the cobalamin branch, the cobK gene and the cobWNHIJ operon (Li 2009). Thus, expression of cobalamin enzymes appears to be regulated by feedback control via a riboswitch mechanism as well as in response to redox via the RegA-RegB signal transduction cascade.
Analysis of genes involved in the Mg-branch of the tetrapyrrole biosynthesis pathway was primarily advanced by the work of Marrs and coworkers who used a generalized transducing agent (GTA) from R. capsulatus to map the location of Bchl biosynthesis genes (Marrs 1974, Yen and Marrs 1976). These studies established that carotenoid and Bchl biosynthesis genes were present in the R. capsulatus chromosome in a tight linkage order. The transduction mapping studies were followed by the isolation of an R’ plasmid that was shown by marker rescue and transposition mapping analyses to contain all of the essential genetic information needed to synthesize Bchl (Marrs 1981, Biel and Marrs 1983, Taylor et al. 1983, Zsebo and Hearst 1984). Structural and functional information regarding these loci was refined by complete sequence analysis of the R’ by Hearst and coworkers (Alberti, Burke and Hearst 1995) and by construction of defined interposon mutations in each of the sequenced open reading frames (Giuliano et al. 1988, Young et al. 1989, Yang and Bauer 1990, Bollivar et al. 1994a, Bollivar et al. 1994b). Similar clustering of Bchl biosynthesis genes occurs in other purple bacterial species (Igarashi et al. 2001, Jaubert et al. 2004).
Several of the enzymes encoded by one or more of the Bchl biosynthesis genes have been expressed in E. coli and shown to exhibit enzymatic activity at specific steps of the Mg-branch. The first committed step of the Mg branch involves insertion of Mg into Proto IX by the enzyme Mg chelatase (EC 18.104.22.168) to form Mg-proto IX. Gibson et al (1995) were the first to successfully heterologously express subunits of Mg-chelatase coded by the bchH, bchI and bchD genes in E. coli to definitively establish the subunit composition of this enzyme. Activity is dependent on Mg, ATP and proto IX. The multimeric Mg chelatase is a typical class 1 metal chelatase (Schubert et al. 2002) where BchH binds Proto IX (Gibson et al. 1995) while the structures of BchI and BchD show that they both have ATPase domains of the AA+ class (Fodje et al. 2001).
The next step of the pathway involves SAM dependent methylation of the carboxyl group of Mg-proto IX by the enzyme SAM Mg-proto IX-O-methyltransferase (EC 22.214.171.124) to form Mg-proto IX monomethyl ester (MPE). Confirmation that the R. capsulatus bchM gene codes for this enzyme occurred when this gene was heterologously expressed in E. coli with cell lysate extracts exhibiting activity (Bollivar et al. 1994a, Bollivar et al. 1994b).
Oxidative cyclization to form the fifth ring of Bchl is catalyzed by the enzyme MPE oxidative cyclase (EC 126.96.36.199) that is coded by the bchE gene to form the product protochlorophyllide (PChlide). This enzyme has not been well characterized biochemically owing to the difficulty in observing activity in vitro. However, analysis of the R. capsulatus BchE peptide sequence indicates that it may be a SAM enzyme that also uses vitB12 as a cofactor (Gough, Petersen and Duus 2000). Indeed depletion of vitB12 in vivo leads to accumulation of MPE in this branch of the pathway (Gough, Petersen and Duus 2000).
Reduction of the D ring of PChlide to form chlorophyllide (Chlide) occurs by the enzyme NADPH- PChlide oxidoreductase (EC 188.8.131.52). In nature there are two unrelated enzymes that can catalyze this reduction. One is a light dependent version that is present in cyanobacteria, alga, gymnosperms and angiosperms, while the other is light independent and present in anoxygenic bacteria, cyanobacteria, alga and gymnosperms (Fujita and Bauer 2003, Heyes and Hunter 2005). The dark operative form is present in R. capsulatus, encoded by bchL, bchN and bchB. Interestingly, it has a high degree of primary sequence similarity to nitrogenase (Fujita and Bauer 2003). Biochemical characterization of the R. capsulatus enzyme indicates that it contains iron sulfur centers and requires ATP, ferridoxin and a reducing agent for catalysis (Fujita and Bauer 2003, Nomata et al. 2005). Reduction of Ring B of Chlide by the enzyme Chlide a reductase (EC 184.108.40.206), encoded by bchX, bchY and bchZ, also uses an enzyme that is very similar to dark PChlide reductase (Nomata et al. 2006).
There are several latter steps of the pathway that have not been biochemically characterized. The one exception is the esterification of the propionate on ring IV by the enzyme Bchl a synthase, encoded by bchG. This enzyme is membrane bound and difficult to isolate. However, it was heterologously expressed in E. coli with membrane fractions shown to exhibit this activity (Oster, Bauer and Rudiger 1997).
Studies demonstrated that CrtJ/PpsR is an aerobic repressor of Bchl genes, bch, in R. capsulatus (Ponnampalam and Bauer 1997, Elsen, Ponnampalam and Bauer 1998). As discussed below, anaerobic induction of Bchl, carotenoid and light harvesting genes also requires phosphorylated RegA. So together, CrtJ and RegA regulate synthesis of Bchl biosynthesis genes by coordinating aerobic repression, and anaerobic activation, respectively (figure 2).
CrtJ cooperatively binds to two copies of the palindromic sequence TGT-N12-ACA that is present in all of the characterized bch promoters (Alberti, Burke and Hearst 1995, Ponnampalam and Bauer 1997, Elsen, Ponnampalam and Bauer 1998, Ponnampalam, Elsen and Bauer 1998). The palindrome sequence is found either eight base pairs apart, or at sites that are distantly separated. For example, the R. capsulatus bchC promoter region has a CrtJ recognition palindrome that spans the −35 promoter region and a second CrtJ palindrome located 8 bp away that spans the −10 promoter region (Ponnampalam, Elsen and Bauer 1998). Binding to these two palindromes is cooperative so if the eight base pair space between the two palindromes in the bchC promoter region is altered by the addition or deletion of just a few base pairs, then CrtJ is unable to bind to either palindrome effectively (Ponnampalam, Elsen and Bauer 1998).
CrtJ also cooperatively binds to two palindromes at other promoters but these palindromes are separated by more than 100–150 base pairs (Elsen, Ponnampalam and Bauer 1998). An example of this type of binding occurs in the intergenic region between crtA and crtI that contains two promoters, one that is responsible for driving expression of the crtA-bchI-bchD operon, and a second divergent promoter >100 bp away that is responsible for expression of the crtI-crtB operon (Elsen, Ponnampalam and Bauer 1998). The promoter for the crtA-bchI-bchD transcript has a single CrtJ binding site that spans the −10 promoter sequences while the crtI-crtB promoter also has a single CrtJ recognition sequence that spans the −35 recognition sequence. Cooperative binding of CrtJ to these two palindromes coordinately represses expression of both the crtA-bchI-bchD and crtI-crtB operons (Elsen, Ponnampalam and Bauer 1998). This affects synthesis of both Bchl and carotenoids since BchI and BchD are subunits of Mg-chelatase, (Bollivar et al. 1994a, Bollivar et al. 1994b) and crtI and crtB code for phytoene dehydrogenase and phytoene synthase that are enzymes for the first two committed steps of carotenoid biosynthesis, respectively (Armstrong et al. 1990). Presumably, binding to distant sites involves looping of the DNA so that tetrameric CrtJ can bind cooperatively to both of the recognition palindromes (Elsen, Ponnampalam and Bauer 1998).
The Mg-branch of the tetrapyrrole biosynthetic pathway is also anaerobically activated by the global two-component system RegA-RegB (Willett, Smart and Bauer 2007). Specifically, expression of the large bchEJG-orf428-bchP-idi operon that encodes numerous enzymes in Bchl biosynthesis, as well as an enzyme involved in carotenoid biosynthesis, requires phosphorylated RegA for maximal expression (Bollivar et al. 1994a, Bollivar et al. 1994b, Alberti, Burke and Hearst 1995, Hahn, Baker and Poulter 1996, Suzuki, Bollivar and Bauer 1997, Bollivar 2006, Willett, Smart and Bauer 2007). In addition, phosphorylated RegA is also required for expression of the crtA-bchIDO operon that codes for early enzymes involved in the Bchl branch as well as early enzymes in the carotenoid biosynthesis pathway (Alberti, Burke and Hearst 1995).
The trifurcated tetrapyrrole biosynthetic pathway not only share common early intermediates, but also form an intricate network where the endproducts cobalamin and heme are involved in each other and Bchl biosyntheses. For example, in the Bchl branch, Gough et al identified a cyclase encoded by bchE gene that catalyzes the conversion of MPE to PChlide. The sequence similarity between this MPE-cyclase and a cobalamin-dependent P-methylase from Streptomyces hygroscopicus indicates an involvement of cobalamin in MPE-cyclase activity. Although no in vitro cyclase assays have been reported, MPE-cyclase activity was demonstrated to be cobalamin dependent in vivo as both cobalamin depletion or a bchE knockout results in an accumulation of MPE (Gough, Petersen and Duus 2000).
In addition to the dependence of cobalamin for Bchl biosynthesis, the synthesis of cobalamin is dependent on the presence of heme. Specifically, one of the most complicated steps in the R. capsulatus cobalamin branch is the corrin ring contraction that is catalyzed by the cofactor-rich enzyme, CobZ. CobZ contains heme as a cofactor as well as flavin and two Fe-S centers (McGoldrick et al. 2005). The dependence of heme as a cofactor in CobZ thus makes cobalamin synthesis heme dependent.
Finally, synthesis of heme is also cobalamin dependent as heme synthesis requires S-adenosylmethionine methionine as a methyl group donor and its synthesis involves a cobalamin-dependent enzyme (Drennan et al. 1994, Layer et al. 2003, McGoldrick et al. 2005). The interdependent relationship among all three tetrapyrrole compounds as discussed in ref (McGoldrick et al. 2005) suggests that codependence of these different tetrapyrrole branches on different tetrapyrrole end products is a form of regulation used to coordinate the levels of these different tetrapyrrole derivatives.
The tetrapyrrole biosynthetic pathway from R. capsulatus is quite complex involving several branches that are responsible for synthesis of heme, vitB12 and Bchl a. The flow of tetrapyrroles into each branch in the trifurcated pathway is highly regulated with an intricate network of transcriptional and post-transcriptional events controlling the synthesis of these various end-products. This includes the control of enzyme activity by interaction with tetrapyrrole endproducts, transcriptional control either in response to redox poise or in response to an interaction with a tetrapyrrole. There is also a Riboswitch control of translation via interaction of mRNA with a tetrapyrrole. Many players in this complex regulatory web have been identified, although it will not be surprising to find that there are additional players involved in controlling this complex pathway.
Among many challenges going forward will be to determine how multiple transcription factors coordinately regulate the synthesis of individual enzymes in the pathway. For example the hemC promoter appears to be controlled by at least four different activators/repressors. Where these factors bind to the hemC promoter and how binding of individual transcription factors affect binding of other transcription factors remains to be determined. The study of this complex pathway has been an ongoing for over 5 decades, and owing to its complexity, will likely remain an area of focus for some time.