|Home | About | Journals | Submit | Contact Us | Français|
A spontaneous mutant of Rhodobacter sphaeroides f. sp. denitrificans IL-106 was found to excrete a large amount of a red compound identified as coproporphyrin III, an intermediate in bacteriochlorophyll and heme synthesis. The mutant, named PORF, is able to grow under phototrophic conditions but has low levels of intracellular cysteine and glutathione and overexpresses the cysteine synthase CysK. The expression of molybdoenzymes such as dimethyl sulfoxide (DMSO) and nitrate reductases is also affected under certain growth conditions. Excretion of coproporphyrin and overexpression of CysK are not directly related but were both found to be consequences of a diminished synthesis of the key metabolite S-adenosylmethionine (SAM). The wild-type phenotype is restored when the gene metK encoding SAM synthetase is supplied in trans. The metK gene in the mutant strain has a mutation leading to a single amino acid change (H145Y) in the encoded protein. This point mutation is responsible for a 70% decrease in intracellular SAM content which probably affects the activities of numerous SAM-dependent enzymes such as coproporphyrinogen oxidase (HemN); uroporphyrinogen III methyltransferase (CobA), which is involved in siroheme synthesis; and molybdenum cofactor biosynthesis protein A (MoaA). We propose a model showing that the attenuation of the activities of SAM-dependent enzymes in the mutant could be responsible for the coproporphyrin excretion, the low cysteine and glutathione contents, and the decrease in DMSO and nitrate reductase activities.
Rhodobacter sphaeroides is a photosynthetic purple bacterium that is able to grow under phototrophic or chemoheterotrophic conditions. Anoxygenic photosynthetic growth requires the synthesis of a large amount of bacteriochlorophyll (Bchl) via the tetrapyrrole pathway. Tetrapyrrole biosynthesis is a central anabolic pathway leading to the formation of essential compounds such as heme, Bchl, siroheme, and vitamin B12 from simple precursors (Fig. (Fig.1).1). In photosynthetic purple bacteria, the first seven enzymatic reactions leading to protoporphyrin IX are common to heme and Bchl biosynthesis. Addition of Fe2+ to protoporphyrin IX yields heme, whereas addition of Mg2+ yields Mg-protoporphyrin, which subsequently produces Bchl (for a review, see reference 46).
The first branch point in tetrapyrrole biosynthesis directs uroporphyrinogen III toward corrinoid production. The first reaction of the corrinoid branch is catalyzed by the cobA gene product, a methyltransferase that transfers two S-adenosylmethionine (SAM)-derived methyl groups to generate precorrin-2, an intermediate common to the cobalamin and siroheme pathways (for a review, see reference 44). Siroheme is the prosthetic group of some assimilatory nitrite and sulfite reductases. In Rhodobacter sphaeroides 2.4.1, the sulfite reductase, which is involved in the cysteine biosynthesis pathway that reduces sulfite into sulfide, is encoded by cysI (34). Sulfide is then incorporated into O-acetyl-l-serine to produce cysteine. This step is catalyzed by either O-acetylserine (thiol)-lyase A or O-acetylserine (thiol)-lyase B, encoded by the genes cysK and cysM, respectively. Both enzymes are able to synthesize cysteine from O-acetylserine and sulfide, but only CysM can utilize thiosulfate (21).
As tetrapyrroles are precursors for several pathways, they are essential compounds in the cell. Heme is an essential cofactor in cells, playing a key role as an electron carrier under both aerobic and photosynthetic conditions. In contrast, Bchl synthesis is inhibited under aerobic conditions, as free Bchl and porphyrin intermediates produce toxic free radicals in the presence of light and oxygen (27). When oxygen is limiting, the need for a high level of Bchl drives tetrapyrrole synthesis toward Bchl, increasing overall tetrapyrrole production by up to 100-fold (23). Tetrapyrrole synthesis is thus strictly regulated (for a review, see reference 50). There are two major points at which the biosynthetic pathway is controlled as a function of oxygen tension (50). One is at the first reaction shown in Fig. Fig.1,1, the condensation of glycine and succinyl coenzyme A (succinyl-CoA) to give 5-aminolevulinic acid (ALA). The second control point is the conversion of coproporphyrinogen III to protoporphyrinogen IX. Two structurally different enzymes catalyze this reaction; one is active only under aerobic conditions and the other only under anaerobic conditions (16, 48). The dimeric aerobic coproporphyrinogen III oxidase (encoded by hemF) uses molecular oxygen as an electron acceptor for the decarboxylation of propionyl groups to vinyl groups, while anaerobic coproporphyrinogen III oxidase (encoded by hemN or hemZ), a monomeric iron-sulfur protein, requires SAM for catalysis (24). In R. sphaeroides 2.4.1, a single gene, hemF (RSP_0682), encodes the aerobic coproporphyrinogen oxidase, while two genes encode anaerobic coproporphyrinogen oxidases. One of the latter was initially described and named hemF (9), but this name is now reserved for the aerobic coproporphyrinogen oxidase gene, so we refer to it here as hemN. Another gene flanking fnrL, named hemZ, has been characterized (51). hemN and hemZ (RSP_0317 and RSP_0699, respectively) are both expressed under anaerobic conditions (with very low expression under aerobic conditions) and are under the control of FnrL and PrrA (30). HemN and HemZ belong to the family of “radical SAM” proteins (40). A third gene (RSP_1224) encodes a putative anaerobic coproporphyrinogen oxidase. Despite having only 20% identity with the other two enzymes, key residues involved in SAM binding and the 4Fe4S cluster are conserved. Radical SAM proteins transfer one electron from an iron-sulfur cluster to the SAM cofactor, which is then cleaved into methionine and a highly oxidizing radical. This catalytic radical abstracts one hydrogen atom from the substrate's propionate chain, giving rise to a vinyl group with the elimination of CO2. According to Fontecave et al. (15), SAM is the second most prevalent enzyme substrate in cells after ATP. In addition to its role in radical SAM enzymes, it is the major methyl donor for essential methylation reactions (4, 7) and serves as a substrate in polyamine biosynthesis (15). SAM is synthesized in a two-step reaction from ATP and l-methionine by SAM synthetase. This tetrameric metalloenzyme is encoded by metK. This gene is unique in some bacteria and has been shown to be essential for development, in particular in Escherichia coli, Bacillus subtilis, or Myxococcus xanthus (38, 45, 49). One SAM transporter has been identified in Rickettsia prowazekii and allows the growth of strains with an inactivated metK gene (12, 42).
Regulation of the tetrapyrrole biosynthesis pathway is complex and involves several regulatory systems. It is nevertheless an efficient process in bacteria; despite the heavy metabolic demands, which vary according to the growth conditions and affect different branches of this pathway, there is no intracellular accumulation of intermediates (52). However, several mutants affected in one of the steps of the tetrapyrrole synthesis pathway accumulate porphyrins (25, 35-37, 48). In R. sphaeroides, one mutant unable to grow under photosynthetic conditions excretes coproporphyrin III into the growth medium (9). Synthesis of Bchl and photosynthetic growth are recovered by introducing the hemN gene in trans.
While studying selenite reduction in R. sphaeroides, we isolated several spontaneous mutants showing increased resistance to selenite (not explored further here). Several of these mutants excreted large amounts of a red compound that we show here to be coproporphyrin III. We present a detailed analysis of one of these mutants, which we named PORF (for porphyrin). This mutant is also affected in cysteine synthesis and molybdoenzyme activity. We propose a model showing that this phenotype results from SAM depletion due to a single point mutation in metK, the gene encoding SAM synthetase.
R. sphaeroides f. sp. denitrificans IL-106 (a generous gift from T. Satoh) was grown at 30°C in Sistrom minimal medium supplemented with succinate (39) in Schott bottles sealed with rubber septa and sparged with nitrogen. For standard phototrophic growth, the light intensity was 180 μmol of photons·m−2·s−1. Where mentioned, the cells were grown under low-light conditions (17 μmol of photons·m−2·s−1) or in the presence of 20 mM nitrate or 60 mM dimethyl sulfoxide (DMSO). E. coli was grown in Luria-Bertani medium. When required, 25 μg/ml kanamycin, 50 μg/ml spectinomycin, or 50 μg/ml streptomycin was included in the medium.
DNA isolation, plasmid purification, and restriction analysis were carried out using standard methods. For sequencing metK, R. sphaeroides f. sp. denitrificans genomic DNA was first PCR amplified with PfuUltra (Stratagene) using the primers 5′-TACAAAGATGTCGCACTGC-3′ and 5′-GAGGTGCAGATCTCGCTCG-3′. DNA sequencing was performed by GATC Biotech, France.
The cysK gene, encoding cysteine synthase, was amplified from R. sphaeroides f. sp. denitrificans genomic DNA with the primers 5′-CACCTCTAGACAGGAGACGCATCATGGATG-3′ and 5′-TCAGAGCTGGAAGCTCGGCGT-3′. The PCR product was cloned into pENTR/D-TOPO (Invitrogen) and sequenced. An omega cartridge encoding resistance to streptomycin and spectinomycin (31) was then cloned into the PstI site of cysK. The resulting plasmid was digested with EcoRV and XbaI, and the fragment containing the disrupted cysK gene was cloned into pJQ200SK (32). The resulting plasmid, which was unable to replicate in R. sphaeroides, was transferred from E. coli by conjugation. The occurrence of a double-crossover event was confirmed by Southern blot analysis and absence of the protein from the SDS-PAGE profile.
A R. sphaeroides 2.4.1 genomic DNA library (13) was transferred from E. coli to the PORF mutant by biparental mating. Cells were grown under phototrophic conditions in 96-well plates and screened for the lack of coproporphyrin excretion into the medium. Only cosmid pUI8308 restored a wild-type phenotype. This cosmid was digested with SacI, and each fragment was cloned into the plasmid pBBR1MCS2 (20). All the resultant plasmids were transferred to PORF and scored for the absence of coproporphyrin excretion. Among them, only plasmid pMS931, containing a 1,817-bp SacI fragment, restored a wild-type phenotype, and it was then sequenced.
A cell-free supernatant of the PORF mutant grown under phototrophic conditions in Sistrom medium was first desalted on an SPE C18 cartridge, diluted in methanol (MeOH), and analyzed by direct infusion in an electrospray ionization-mass spectrometer (ESI-MS) at a flow rate of 10 μl·min−1 using a syringe pump (Harvard Apparatus, Cambridge, MA). It was introduced using a 250-μl glass syringe with a stainless steel needle (Hamilton), which was cleaned thoroughly with solvent between injections. Molecular mass analyses were performed with an ion trap (model LC-Q; Thermoelectron). The source parameters, such as temperature, cone voltage, and gas pressure, were optimized to give the best intensity for the base peak at m/z 655.2. Spectra were acquired at 5 scans/s over a mass range of 50 to 2,000 with an acquisition time of 1 min. Tandem mass spectrometry (MS/MS) was done until MS8, which allowed the identification of a structural scheme of a compound belonging to the porphyrin family.
Cells were grown under phototrophic conditions in Sistrom medium until the stationary phase. Cultures were centrifuged, and the supernatant was diluted 200-fold in 0.1 M HCl. High-performance liquid chromatography (HPLC) analysis was as described by Wright and colleagues (47). A Symmetry C18 analytical column (particle size, 5 μm; 250 mm by 3 mm) (Waters) was used for separation. The mobile phase was 26% acetonitrile in 1 M ammonium acetate buffer with a flow rate of 0.7 ml/min. Fluorescence was monitored with a Waters 464 detector (λexcitation = 405 nm; λemission = 620 nm). Standards coproporphyrin I and coproporphyrin III were purchased from Sigma and solubilized in 0.1 M HCl.
Cells were grown under phototrophic conditions in Sistrom medium until the exponential phase (24 h) or stationary phase (96 h). For each condition, 5 ml each of three independent cultures was centrifuged and the pellets frozen (−80°C) until analysis. Nonprotein thiols were extracted by disrupting cells by sonication in 1 ml of extraction buffer [6.3 mM diethylenetriamine pentaacetic acid (DTPA), 0.1% trifluoroacetic acid (TFA), 1 mM Tris (2-carboxyethyl)phosphine (TCEP)]. As an internal standard, 40 μM N-acetyl-l-cysteine was added. The homogenate was centrifuged at 10,000 × g for 15 min at 4°C, and the supernatant was filtered (0.22-μm pore diameter). The derivatization procedure was modified from that of Rijstenbil and Wijnholds (33). Filtered extracts (125 μl) were mixed with 225 μl of reaction buffer, 0.2 M 4-(2-hydroxy-ethyl)-piperazine-1-propanesulfonic acid (pH 8.2) containing 6.3 mM DTPA, and 5 μl of 25 mM monobromobimane dissolved in acetonitrile. The reaction mixture was incubated for 15 min at room temperature in the dark, and the reaction was stopped by adding 150 μl of 1 M methanesulfonic acid. The samples were stored in the dark at 4°C until being analyzed by HPLC. The bimane derivatives were separated on a reverse-phase Symmetry C18 analytical column (particle size, 3.5 μm; 4.6 mm by 75 mm) (Waters). The mobile phase was aqueous trifluoroacetic acid (0.1%)-acetonitrile (90:10, vol/vol) with a flow rate of 1 ml·min−1. Fluorescence was monitored with a Waters 2998 photodiode array detector (λexcitation = 380 nm; λemission = 470 nm). For quantification, we used standard solutions of cysteine and glutathione (Sigma-Aldrich). The total protein content of sonicated homogenates was measured with the BC assay (Interchim) in 2% SDS.
The HPLC method was adapted from that of Dever and Elfarra (11). Cells were grown under phototrophic conditions in Sistrom medium to exponential phase (24 h). For each strain, three independent cultures were used; 45 ml of each culture was centrifuged and the pellet frozen (−80°C) until analysis. The cell pellet was sonicated in 2 ml water and centrifuged at 10,000 × g for 15 min at 4°C. Samples (0.5 ml) were then deproteinated with 3 volumes of ice-cold ethanol and centrifuged at 13,000 × g for 20 min. The supernatant was removed and dried down in a miVac concentrator (Genevac Inc.). The residue was dissolved in 200 μl of 0.4% aqueous TFA and filtered (pore diameter, 0.22 μm). HPLC was carried out using a Sunfire C18 reverse-phase column (particle size, 5 μm; 3 mm by 250 mm) (Waters). The mobile phase was composed of 0.4% TFA (solvent A) and methanol-0.4% TFA (3:1, vol/vol) (solvent B). A 2998 photodiode array detector (Waters) was used for UV detection at 220 nm and 254 nm. Metabolites were eluted using a gradient method with an initial concentration of 2% solvent B for 8 min. The solvent B concentration was increased to 50% over 12 min and maintained at that level for an additional 5 min. The solvent B concentration was then increased to 100% over 5 min. For quantification we used standard solutions of S-adenosylmethionine (Sigma). The total protein content of sonicated homogenates was measured with the BC assay (Interchim) in 2% SDS.
Cells were resuspended in 50 mM Tris-HCl buffer (pH 8) in the presence of 0.45 M sucrose, 1.3 mM EDTA, and 0.6 mg/ml lysozyme. After incubation for 60 min at 30°C, the suspension was centrifuged for 10 min at 8,000 × g. The supernatant (periplasmic fraction) was centrifuged at 200,000 × g for 60 min to remove cell wall debris. The pellet from the low-speed centrifugation was washed with the same buffer and then resuspended vigorously in 50 mM Tris-HCl (pH 8) to break the spheroplasts. The suspension was centrifuged for 10 min at 8,000 × g, and the supernatant was again centrifuged for 60 min at 200,000 × g. The soluble fraction contains cytoplasmic proteins, while the pellet contains membrane proteins. Protein concentrations were determined using the BC assay (Interchim).
Proteins were separated by polyacrylamide gel electrophoresis (PAGE) with a 10% acrylamide gel. For the first dimension, the nondenaturing buffer 20 mM Tris-200 mM glycine was used. A strip of the first-dimension gel was excised and incubated at 60°C for 20 min in sample buffer consisting of 60 mM Tris-HCl (pH 6.8), 2% SDS, 40 mM dithiothreitol (DTT), and 0.02% bromophenol blue. For the second dimension, the strip was placed on top of a 10% polyacrylamide gel and the denaturing buffer used was 20 mM Tris, 200 mM glycine, and 0.1% SDS. Matrix-assisted laser desorption ionization-time-of-flight (MALDI-TOF) analysis of the protein samples was done by PAPPSO, INRA, Jouy en Josas, France (http://pappso.inra.fr/index.html).
Cells were grown until late exponential phase in Sistrom medium under several conditions: anaerobic dark conditions with 60 mM DMSO or photoheterotrophic conditions (180 μmol of photons·m−2·s−1) in the absence or the presence of 20 mM nitrate or 20 mM DMSO. Cells were centrifuged, resuspended in 50 mM HEPES (pH 7), and disrupted (One shot; Constant Systems). The suspension was centrifuged for 60 min at 200,000 × g. The supernatant containing the soluble proteins was concentrated. Dimethyl sulfoxide (DMSO) and nitrate reductase activities were assayed spectrophotometrically using reduced benzyl viologen as the electron donor (2). The reactions were carried out under anaerobic conditions at 30°C in a 50 mM HEPES buffer (pH 7) containing 0.2 mM dithionite-reduced benzyl viologen and 5 mM DMSO or 2 mM nitrate as substrates. Oxidation of benzyl viologen was assayed spectrophotometrically at 600 nm (extinction coefficient, 10.4 mM−1 cm−1 ).
The R. sphaeroides f. sp. denitrificans IL-106 metK and cysK sequences have been submitted to the DDBJ/EMBL/GenBank databases under accession number FN552708 and GU296109, respectively.
In the course of a study of selenite reduction, we isolated several spontaneous R. sphaeroides f. sp. denitrificans mutants showing increased resistance to selenite. Among them, some excreted large amounts of a red compound when grown in Sistrom minimal medium under phototrophic conditions (Fig. (Fig.2A).2A). We studied one of these, designated PORF, in more detail. Spectral analysis of a cell-free supernatant (Fig. (Fig.2B)2B) suggested that the substance excreted could belong to the porphyrin family. The compound was analyzed by tandem mass spectrometry (MS/MS), and the fragmentation pattern (m/z 655, 637, 596, 583, 577, 565, 551, 537, 523, 510, 479, and 464) allowed us to identify it as coproporphyrin (10), arising from the spontaneous oxidation of coproporphyrinogen. HPLC analysis of the cell-free supernatant (Fig. (Fig.3)3) confirmed this result and was used to discriminate between the isomers coproporphyrin I and coproporphyrin III. Coproporphyrin I is also found in the plasma of humans presenting a congenital erythropoietic porphyria or in bacteria lacking an active uroporphyrinogen III synthase (5). As shown in Fig. Fig.3,3, the PORF mutant excretes coproporphyrin III, an intermediate in the tetrapyrrole pathway. Excretion of coproporphyrin III continues for several days after bacteria have reached the stationary phase, and the coproporphyrin concentration in the growth medium can attain 50 mg/liter. Unlike the R. sphaeroides mutant N1 (9), which excretes coproporphyrin III but does not synthesize Bchl, the PORF mutant grows under phototrophic conditions and synthesizes Bchl (Fig. (Fig.4).4). Under standard phototrophic conditions (180 μmol of photons·m−2·s−1), the PORF mutant growth is similar to that of the wild type and the amounts of B800-850 complexes are very similar in the two strains (Fig. 4A and B). It is well known that under low light intensity, R. sphaeroides synthesizes more intracellular membrane and B800-850 complexes (8, 19). When grown in such low-light conditions (17 μmol of photons·m−2·s−1), both the cell growth rate and the maximal cell density of the PORF mutant are less than those of the wild type (Fig. (Fig.4C).4C). Moreover the amount of light-harvesting pigments B800-850 is slightly less than that in the wild type (Fig. (Fig.4D).4D). This is also the case under dark anaerobic conditions in the presence of DMSO (Fig. (Fig.4F).4F). Under these conditions, there is a marked lag in PORF mutant growth, but the same maximal cell density is attained (Fig. (Fig.4E).4E). The cytochrome c2 content of cells was measured (absorption at 552 nm) after reduction of soluble extracts with sodium dithionite, and no difference between wild-type and PORF extracts was observed.
Under aerobic conditions, the mutant does not excrete coproporphyrin III (data not shown), and its growth is similar to that of the wild type. As mentioned in the introduction, two structurally different enzymes catalyze the conversion from coproporphyrinogen III to protoporohyrinogen IX, the aerobic coproporphyrinogen oxidase HemF and SAM-dependent anaerobic coproporphyrinogen oxidases encoded by hemN and hemZ (RSP_0317 and RSP_0699, respectively, in Rhodobacter sphaeroides 2.4.1). The HemN and HemZ sequences share 46% identity. A third gene (RSP_1224) encodes a putative anaerobic coproporphyrinogen oxidase. A mutation in one of the three genes encoding anaerobic coproporphyrinogen oxidases or in one of the numerous regulators of the tetrapyrrole pathway (50) could be responsible for the phenotype of the PORF mutant. Because of the complexity of this biosynthesis pathway, we chose global approaches of proteomics and genomic complementation to first elucidate the phenotype of the PORF mutant.
The phenotype of the PORF mutant was first investigated at the protein level. Periplasmic and cytoplasmic extracts of the wild type and the PORF mutant were prepared and separated by 2D PAGE. Several proteins seemed to be overexpressed in the cytoplasmic extract of the PORF mutant (Fig. (Fig.5).5). They were analyzed by MALDI-TOF. Among them were some periplasmic proteins due to slight contamination of the cytoplasmic extract, which can vary from one preparation to another. The major difference observed was the overexpression of a 38-kDa cytoplasmic protein in the PORF mutant extracts. This protein was identified as a cysteine synthase. In R. sphaeroides 2.4.1, two genes, RSP_1109 and RSP_2147, encode O-acetylserine-lyases that have only 30% identity. Both genes are probably expressed, since an RSP_2147 null mutant remains prototrophic for cysteine (29). The cysteine synthase overexpressed in the PORF mutant is encoded by a homologue to RSP_1109. We sequenced the gene; 99.4% of its residues are identical to those of R. sphaeroides 2.4.1 RSP_1109.
Are excretion of coproporphyrin and overexpression of cysteine synthase related? We observed that addition of 1 mM l-cysteine to the Sistrom minimal medium of wild-type R. sphaeroides and the PORF mutant attenuates the synthesis of light-harvesting pigments under phototrophic conditions (see Fig. S1 in the supplemental material). This suggests that cysteine and tetrapyrrole syntheses are somehow indirectly connected, leading us to focus on the overexpression of the cysteine synthase.
The overexpression of cysteine synthase observed in the mutant could cause an increase in intracellular cysteine and glutathione concentrations, as is observed when the Staphylococcus aureus cysM gene is expressed in E. coli (26). Alternatively, it could result from a lack of cysteine, as when the amount of intracellular cysteine decreases during sulfur starvation, the cysteine regulon (which includes the genes for cysteine synthesis and uptake of sulfur compounds) is upregulated due to the accumulation of O-acetylserine and N-acetylserine, the latter being the inducer of transcription (21). To distinguish between these two hypotheses, we measured the cysteine and glutathione contents of wild-type and mutant cell extracts by liquid chromatography of thio-bimane derivatives (Fig. (Fig.6).6). As early as the exponential phase, the PORF mutant contains less cysteine and glutathione than the wild type. This deficit increases with time, and during the stationary phase the cysteine content and glutathione content of mutant extracts are only 50% and 13% of those of wild-type extracts, respectively. These results demonstrate that overexpression of cysteine synthase is related to a shortage of intracellular thiols in the PORF mutant.
When 1 mM cysteine is added to Sistrom minimal medium, the PORF mutant still excretes porphyrins. Thus, the decrease in cysteine concentration is not responsible for the porphyrin excretion. This does not exclude the possibility that overexpression of RSP_1109 is the cause of this excretion. Besides its enzymatic role in cysteine biosynthesis, it has been shown in Bacillus subtilis that CysK is a global regulator of genes involved in sulfur metabolism (1). Could CysK also be involved in the regulation of tetrapyrrole synthesis? To test this possibility, we created null mutations of RSP_1109 by inserting an omega cartridge in both the wild-type strain and the PORF mutant. We verified that the protein was not expressed in the resultant strains by 2D PAGE (data not shown). The PORF RSP_1109 null mutant still excretes a similar amount of coproporphyrin as the PORF mutant, so this characteristic cannot be due to overexpression of cysteine synthase.
These RSP_1109 null mutants are still phototrophic for cysteine, probably because R. sphaeroides f. sp. denitrificans has an RSP_2147 homologue. We evaluated the three other strains of R. sphaeroides whose genomes have been sequenced (ATCC 17029, ATCC 19025, and KD1131). All of them have only two genes encoding O-acetylserine (thiol)-lyase, one a homologue of RSP_1109 (94 to 99% identity) and the other a homologue of RSP_2147 (92 to 99% identity). This is most probably the case in R. sphaeroides f. sp. denitrificans too. Analysis of the RSP_1109 null mutants did, however, allow us to determine whether RSP_1109 encoded O-acetylserine (thiol)-lyase A or B. In a modified Sistrom medium containing no other source of sulfur, 1 mM thiosulfate supports RSP_1109 null mutant growth. This mutant probably contains one cysteine synthase, encoded by a homologue of RSP_2147, which allows growth on thiosulfate as the unique sulfur source. This suggests that RSP_2147 encodes CysM (since CysK cannot utilize thiosulfate) and hence that RSP_1109 encodes CysK. This is consistent with structural data on O-acetylserine (thiol)-lyase A and B from Salmonella enterica serovar Typhimurium (6). There is a major difference between the active sites of these two isoenzymes: two ionizable residues (C280 and D281) in the B isoenzyme are replaced by neutral residues (P299 and S300, respectively) in the A isoenzyme. The equivalent residues are C303 and D304 in RSP_2147 and P323 and D324 in RSP_1109, reinforcing the idea that RSP_2147 encodes CysM [O-acetylserine (thiol)-lyase B] and RSP_1109 encodes CysK.
Both the tetrapyrrole pathway and cysteine biosynthesis are affected in the PORF mutant, but these two phenotypes are not directly linked. With the aim of finding the mutation responsible for such a phenotype, we complemented the PORF mutant with R. sphaeroides 2.4.1 genomic DNA by mobilizing a cosmid library (13) from E. coli into PORF. A lack of coproporphyrin excretion was the phenotype used to score for complementation. Only one cosmid (pUI8308) prevented the excretion of coproporphyrin in the PORF mutant. This cosmid was subcloned into pBBR1MCS2, and all subclones were mobilized into the PORF mutant by conjugation. One of these subclones (pMS931), containing an 1,813-bp SacI insert, was able to prevent the expression of coproporphyrin, while the empty plasmid (pBBR1MCS2) was not. pMS931 fully restores the wild-type phenotype. Specifically, the PORF mutant containing pMS931 in trans (i) does not excrete coproporphyrin, (ii) does not overexpress RSP_1109, and (iii) contains amounts of thiols similar to or greater than those in the wild type (Fig. (Fig.6).6). The 1.8-kb DNA fragment contains the entire metK (RSP_3595) gene (1,167 bp) and starts 322 bp upstream of the metK start codon. The metK gene encodes SAM synthetase, which catalyzes the synthesis of SAM from methionine and ATP. Despite its fundamental role in the cell, E. coli K-12 and the four R. sphaeroides strains sequenced each have a single metK gene.
As addition of metK in trans is able to restore all the wild-type characteristics in the PORF mutant, its phenotype is likely due to a mutation in metK. After PCR amplification of genomic DNA, we sequenced the metK genes from the wild type, the PORF mutant, and two other mutants (MS737 and MS794) that we isolated under the same conditions and that have the same phenotype as PORF (resistance to selenite and excretion of coproporphyrin). The deduced sequence of PORF MetK has one mutation, H145Y (Fig. (Fig.7).7). The other two mutants with similar phenotypes each have one mutation in their respective metK genes, leading to the changes H145Y and P138L in the encoded proteins. The structure of E. coli SAM synthetase, in which 54% of the residues are identical to those of the R. sphaeroides enzyme, has been published (41). Although the two residues H145 and P138 are highly conserved among SAM synthetases, they are distant from the active site. It has been shown that SAM synthetase activity is essential for the development of several bacteria (38, 45, 49). Mutations causing a total absence of SAM synthetase are lethal in E. coli (45), but some mutants showing only a decrease in SAM synthetase activity have been characterized (28). Moreover, SAM is implicated in so many essential reactions (15, 40) that it is improbable that a strain with an inactive SAM synthetase could survive, unless it possessed a SAM transporter, as Rickettsia prowazekii does (12, 42). We verified that addition of 1 mM SAM to the growth medium of the PORF mutant does not restore a wild-type phenotype. Thus, it is likely that the PORF mutant does not have a SAM transporter and that the SAM synthetase is still active in the PORF mutant but less efficient because of the point mutation. To demonstrate that the H145Y mutation affects SAM synthetase activity and consequently the SAM pool, we measured intracellular SAM content by HPLC (Fig. (Fig.8).8). The SAM concentration in the PORF mutant cells is only 31% of that in the wild type. When R. sphaeroides 2.4.1 metK is introduced in trans on the plasmid pMS931 into the PORF mutant, the SAM concentration increases to 0.92 nmol per mg of total protein, higher than the wild-type value of 0.60 nmol/mg protein. This higher value could be explained by the copy number of the plasmid pMS931 (pBBR1MCS2 derivative), which is between 9 and 14 (18). The wild-type and mutated R. sphaeroides f. sp. denitrificans metK genes were cloned into pBBR1MCS2 and introduced into the wild type and the PORF mutant by conjugation. The strain containing the R. sphaeroides f. sp. denitrificans wild-type metK gene in trans does not excrete any coproporphyrin, like the strain containing the R. sphaeroides 2.4.1 metK gene. The plasmid containing the mutated gene only partially restores the wild-type phenotype. The PORF mutant bearing this plasmid excretes only one-third of the amount excreted by the PORF mutant bearing the empty plasmid. This can be explained again by the copy number of the plasmid. The lower specific activity of the H145Y enzyme could be partially balanced by the increase in the amount of enzyme synthesized when the encoding mutated gene is present in multiple copies.
As the PORF mutant excretes large amounts of coproporphyrin, we hypothesized that it could overexpress a transporter in the external membrane. External membrane extracts were therefore treated with 0.1% Triton X-100 and separated by 2D electrophoresis. This protocol is not ideal for analyzing intrinsic membrane proteins but allows proteins which are bound to the membrane (plus some periplasmic contaminants) to be visualized. Under these electrophoretic conditions, we did not find an overexpressed membrane-bound protein; however, there seemed to be less of two proteins in the PORF mutant (see Fig. S2 in the supplemental material). MALDI-TOF analysis of the proteins showed them to be an aerobic monoxide dehydrogenase (encoded by RSP_2877 in R. sphaeroides 2.41) and a putative sulfite oxidase (encoded by RSP_1410) homologous to E. coli YedY (3). These proteins are not membrane bound but interact with membrane-bound partners, and both are molybdoenzymes. We found (Fig. (Fig.4E)4E) that when the PORF mutant grows at the expense of DMSO reduction, the lag phase is longer and the growth rate lower than for the wild type. As DMSO reductase is also a molybdoenzyme, we hypothesized that DMSO reductase synthesis could be affected in the PORF mutant like the synthesis of other molybdoenzymes. We measured the activities of two molybdoenzymes, DMSO reductase and nitrate reductase, under different growth conditions (Table (Table1).1). These activities vary a lot depending on the growth conditions, since both substrates are able to induce the synthesis of their respective enzymes. Under phototrophic conditions, the presence of nitrate induces a 10-fold increase in nitrate reductase activity, while the presence of DMSO induces a 7-fold increase in DMSO reductase activity. In the presence of DMSO (under both dark and light conditions), the activity of nitrate reductase is very low and is similar in the wild type and in the PORF mutant. However, under phototrophic conditions, in the absence of DMSO the nitrate reductase activity is decreased in the PORF mutant compared to the wild type. This is clearly the case in the presence of nitrate, where a 30% decrease in nitrate reductase activity is observed in the PORF mutant. Concerning DMSO reductase activity, the amounts of enzyme are similar in the two cultures, except under dark anaerobic conditions, where a 60% decrease in DMSO reductase activity is observed in the PORF mutant compared to the wild type. These results clearly show that the PORF mutant is able to synthesize molybdoenzymes to the same extent as the wild type when the amount of these enzymes remains low; however, it cannot sustain a high demand for molybdoenzyme synthesis or maturation.
In this paper we have characterized a spontaneous mutant of Rhodobacter sphaeroides f. sp. denitrificans, PORF, which (i) excretes large amounts of coproporphyrin III, (ii) overexpresses cysteine synthase CysK, (iii) is deficient in glutathione and cysteine, and (iv) synthesizes fewer molybdoenzymes.
The wild-type phenotype can be restored when the gene metK, encoding SAM synthetase, is provided in trans, which suggests that the mutant is affected in SAM synthesis. Indeed, we found that the intracellular SAM content is only 30% of that of the wild type. Moreover, we showed that the metK gene of the PORF mutant bears a mutation leading to a single amino acid change (H145Y) in the encoded protein. Although the enzymatic activity of the H145Y MetK enzyme has not been directly measured, this set of evidence indicates that the phenotype of the PORF mutant results from the H145Y mutation in the SAM synthetase, leading to shrinkage of the intracellular SAM pool.
How can a decrease in the intracellular SAM concentration lead to the PORF phenotype? We propose a model (Fig. (Fig.9)9) which shows that the PORF phenotype could be explained by a decrease in the activity of SAM-dependent enzymes. The SAM content is expected to affect the activity of the SAM-dependent enzymes; whenever the SAM concentration is below the saturation level for these enzymes (i.e., in the range of the Km), their activities will decrease. Among these numerous enzymes, three are known to be involved in the tetrapyrrole pathway (coproporphyrinogen oxidase HemN or HemZ [HemN/Z]; SAM:magnesium-protoporphyrin IX O-methyltransferase, BchM; and uroporphyrinogen III methyltransferase, CobA), and one is involved in molybdocofactor synthesis (molybdenum cofactor biosynthesis protein A, MoaA) (40).
A decrease in HemN/Z activity would result in the accumulation of the substrate coproporphyrinogen III and in diminished heme and bacteriochlorophyll synthesis. It would explain why the PORF mutant does not excrete coproporphyrin under aerobic conditions, since under such conditions decarboxylation of coproporphyrinogen is catalyzed not by the SAM-dependent HemN/Z enzyme but rather by HemF.
For BchM, the decreased activity would be expected to lead to an accumulation of the substrate magnesium protoporphyrin IX and to a deficit in bacteriochlorophyll synthesis. Under the phototrophic conditions used (180 μmol of photons·m−2·s−1), we found that PORF synthesizes the same amount of Bchl and grows at a similar rate as the wild type. However, it cannot sustain a high demand for Bchl, as when the light intensity is low, its growth is slowed down. Under low-light conditions, the amount of B800-850 complexes per cell is just slightly reduced, showing that the cell's response to the Bchl shortage is to grow more slowly rather than to diminish the light-harvesting capacity of the photosynthetic apparatus. Two steps in Bchl synthesis, catalyzed by HemN/Z and BchM, respectively, may be affected by a drop in SAM content. For BchM, the Km for SAM has been measured as 106 μM (17), whereas the Km for HemN/Z is not known. We did not observe any accumulation of Mg protoporphyrin IX (the BchM substrate) as would be expected if BchM activity was abrogated. However, since the pathway is limiting upstream, at the coproporphyrinogen oxidation step it is possible that BchM activity is reduced but that its substrate does not accumulate because it is itself produced in smaller amounts.
A decrease in CobA activity due to SAM depletion is likely to result in a drop in the concentration of siroheme, the prosthetic group of sulfite reductase, and hence in the activity of sulfite reductase (Fig. (Fig.9).9). Consequently, insufficient sulfide would be produced for cysteine synthesis and O-acetylserine would accumulate. With little available cysteine and glutathione, the cysteine regulon would be upregulated, resulting in CysK overexpression (21). Finally, the consequence of a decrease in MoaA activity would be a drop in molybdenocofactor synthesis, which would affect the maturation and activity of molybdoenzymes (43).
The decrease in SAM-dependent enzymes activity could account for the excretion of coproporphyrin III, the drop in thiol content, the overexpression of CysK, the decrease in molybdoenzyme synthesis, and the decrease in bacteriochlorophyll synthesis under some conditions. However, one could expect more drastic effects. Indeed the PORF mutant is not impaired in cytochrome synthesis, and under standard growth conditions (180 μmol of photons·m−2·s−1), it is not impaired in bacteriochlorophyll synthesis. However, under low-light conditions, its growth is slowed down because the mutant cannot sustain a high demand for bacteriochlorophyll. In a similar manner, it cannot sustain a high demand for molybdocofactor synthesis, but it is able to synthesize nitrate and DMSO reductases to some extent. How can these features be explained? The question of how a modified activity of some enzyme within a complex network affects the amounts of metabolites is the subject of metabolic control analysis (14). Experimental and theoretical studies in this domain have shown that the response in terms of formation rate or steady-state level of some end products is often weaker than the activity modulation of the modified enzyme. When the amount or efficiency of an enzyme is diminished, this results in a higher steady-state level of its substrates, which tends to increase the flux and partly makes up for the diminished activity. In addition to such simple responses, more complex regulatory feedbacks may intervene. It is thus no surprise that, as described above, the lowered performance of the SAM synthetase turns out to affect the various SAM-dependent pathways to different extents.
As mentioned in the introduction, the PORF mutant was isolated on selenite-containing medium and shows an increased resistance to selenite. The reason for this phenotype is far from obvious, since the reduction of selenite by the mutant is strongly diminished while, as shown, the glutathione content is quite low. It might rather be expected, therefore, to be more sensitive to selenite. Experiments to elucidate this unexpected feature, which could result from diminished import (or increased export) caused by SAM depletion, are under way. To conclude, we note that the PORF mutant, because of its diminished SAM synthesis, could be a useful tool for studying phenotypic effects influenced by SAM concentration and the many SAM-dependent pathways.
We thank J. Lavergne and P. Arnoux for critical review of the manuscript and Didier Chevret (PAPPSO, Jouy en Josas, France) for MALDI-TOF analysis.
Published ahead of print on 28 December 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.