Phosphoribosylpyrophosphate synthetase (PrsA) variants alter cellular pools of ribose 5-phosphate and influence thiamine synthesis in Salmonella enterica

Abstract

Phosphoribosylamine (PRA) is the first intermediate in the common purine/thiamine biosynthetic pathway and is primarily synthesized by the product of the purF gene, glutamine phosphoribosylpyrophosphate (PRPP) amidotransferase (E.C. 2.4.2.14). Past genetic and biochemical studies have shown that multiple mechanisms for the synthesis of PRA independent of PurF are present in Salmonella enterica. Here, we describe mutant alleles of the essential prsA gene, which encodes PRPP synthetase (E.C. 2.7.6.1), that allow PurF-independent thiamine synthesis. The mutant alleles resulted in reduced PrsA activity in extracts, caused nutritional requirements indicative of PRPP limitation and allowed non-enzymic formation of PRA due to a build-up of ribose 5-phosphate (R5P). These results emphasize the balance that must be reached between pathways competing for the same substrate to maintain robustness of the metabolic network.

INTRODUCTION

The metabolic network in a bacterial cell consists of a multitude of biochemical processes, both specialized and general, that are integrated to generate an effective cellular metabolism. Thiamine biosynthesis in Salmonella enterica has proven to be an easy model system for exploration of connections and definition of paradigms of robustness in the metabolic network. Thiamine is an essential vitamin that must be synthesized or acquired exogenously by all living cells. The low level requirement for thiamine makes growth in its absence a sensitive indicator of flux through its biosynthetic pathway.

Phosphoribosylamine (PRA) is an intermediate metabolite in the common pathway for thiamine/purine biosynthesis (Fig. 1). Glutamine phosphoribosylpyrophosphate (PRPP) amidotransferase (E.C. 2.4.2.14), the product of the purF gene in S. enterica, catalyses the synthesis of PRA from PRPP and glutamine. Strains lacking PurF are purine auxotrophs, but can grow in the absence of thiamine in some growth conditions and with some genetic backgrounds (Downs & Roth, 1991; Petersen et al., 1996). Thiamine-independent growth in the absence of PurF reflects the existence of alternative mechanisms to generate PRA. Analysis of this redundancy has described several connections radiating from PRA (Enos-Berlage et al., 1998; Petersen et al., 1996; Ramos & Downs, 2003) and led to the identification of proteins of unknown function that contribute to the metabolic network (Beck et al., 1997; Beck & Downs, 1998; Downs & Petersen, 1994; Enos-Berlage et al., 1998; Gralnick & Downs, 2001).

Fig. 1.

Biosynthetic pathways of purines, thiamine and tryptophan shown schematically, with the enzymes catalysing various steps indicated above the arrows. Metabolites relevant to PRA formation are shown. PRPP is formed from R5P and ATP in a reaction catalysed by PrsA and the metabolic fates of PRPP are represented. PRA, Phosphoribosylamine; PRPP, phosphoribosylpyrophosphate; TPP, thiamine pyrophosphate; HMP-PP, hydroxymethylpyrimidine pyrophosphate; THZ-P, thiazole phosphate; R5P, ribose 5-phosphate; Ant, anthranilate; PR-Ant, phosphoribosylanthranilate; AIR, 5-aminoimidazole ribotide.

Perturbing metabolic flux in a distinct pathway can generate PRA sufficient for thiamine biosynthesis. For instance, disruption of flux in the tryptophan biosynthetic pathways allowed PRA synthesis (Ramos et al., 2008). In this case, constraining flux at the TrpC-catalysed step led to the accumulation of the biosynthetic intermediate phosphoribosylanthranilate. This accumulation resulted in the concomitant build up of ribose 5-phosphate (R5P) and anthranilate as a result of natural breakdown (Creighton, 1968; Ramos et al., 2008). The resulting increase in R5P levels was predicted to generate PRA sufficient for thiamine biosynthesis by non-enzymic reaction with ammonia (Ramos et al., 2008) (Fig. 1).

PRPP synthetase (E.C. 2.7.6.1) is the product of the prsA gene in S. enterica and catalyses the formation of PRPP from ATP and R5P. PRPP is then used in the de novo synthesis of purines, pyrimidines, histidine, tryptophan and pyridine nucleotides (Jensen, 1983). The prsA gene is essential in all freely living organisms; while an Escherichia coli mutant with a null allele of prsA has been isolated, the strain required additional mutations and several metabolites to be viable (Hove-Jensen, 1988). The role of PrsA has been studied in vivo primarily through the isolation and characterization of temperature-sensitive alleles in E. coli and S. enterica (Hove-Jensen, 1985; Post et al., 1991). The study herein is part of our continuing efforts to probe the robustness of the metabolic network anchored at PRA formation. One class of mutations that restored PRA synthesis in a purF mutant consisted of recessive alleles of prsA. Data presented here suggest that the accumulation of R5P in prsA mutants supported non-enzymic formation of PRA that could be manipulated by medium conditions. Further results extend our understanding of the interplay between cellular pathways and PRPP levels in addition to describing a positive selection for recessive alleles of prsA.

METHODS

Medium and chemicals.

Culture medium was obtained from Difco, chemicals from Sigma-Aldrich and restriction enzymes from Promega. No-carbon E medium (NCE) (Davis et al., 1980; Vogel & Bonner, 1956) supplemented with 1 mM MgSO₄, trace minerals (Balch & Wolfe, 1976) and 11 mM glucose (or 20 mM ribose) was used as minimal medium. Nitrogen- and carbon-free salts medium (N⁻C⁻) (Gutnick et al., 1969), supplemented with trace minerals (Balch & Wolfe, 1976), 11 mM glucose and 1 mM glutamine was used as a limiting ammonium minimal medium. Difco nutrient broth (8 g l⁻¹) with NaCl (5 g l⁻¹) was used as rich (NB) medium. Luria broth was used for experiments involving molecular biology, protein purification and plasmid isolation. Difco BiTek agar was added (15 g l⁻¹) for solid media. When present in the culture medium, and unless otherwise stated, compounds were used at the following final concentrations: adenine, 0.4 mM; thiamine, 100 nM; histidine, 0.1 mM; tryptophan, 0.1 mM; methionine, 0.3 mM; guanosine, 0.1 mM; uridine, 0.1 mM; and Difco vitamin assay casamino acids, 0.2 %. The final concentrations of the antibiotics in rich medium were as follows: tetracycline, 20 μg ml⁻¹; kanamycin, 50 μg ml⁻¹; ampicillin, 150 μg ml⁻¹; and chloramphenicol, 20 μg ml⁻¹.

Bacterial strains.

All strains used in this study are derivatives of S. enterica serovar Typhimurium strain LT2 and are listed with their genotypes in Table 1. MudJ refers to Mud1734 (Castilho et al., 1984) and Tn10d(Tc) refers to the transposition-defective mini-Tn10(Tn10Δ16Δ17) (Way et al., 1984). A deletion of the trpEDCBA operon (trp-3618) was generated by replacement and has been described previously (Browne et al., 2006; Datsenko & Wanner, 2000). The his-712 deletion removes hisDCBHAFI and has been described by Hartman et al. (1971).

Table 1.

Bacterial strains used in this study

MudJ refers to the Mud1734 transposon (Castilho et al., 1984) and Tn10d(Tc) refers to the transposition-defective mini-Tn10(Tn10Δ16Δ17) (Way et al., 1984).

Strain	Genotype
DM10000	Wild-type
DM728	purF2085 gnd-181
DM3871	purF2085 gnd-181 yjgf3 : : MudJ
DM9890	purF2085 gnd-181 trpC3620
DM10373	purF2085 trp-3618 gnd174 : : MudJ his-712
DM11257	purF2085 trp-3618 gnd174 : : MudJ his-712 zxx-4106 : : Tn10d(Tc) prsA500
DM11461	purF2085 gnd-181 zxx-4106 : : Tn10d(Tc) prsA500
DM11523	zxx-4106 : : Tn10d(Tc)
DM11524	zxx-4106 : : Tn10d(Tc) prsA500
DM11538	zde-1859 : : Tn10d(Cm) prsA100
DM12033	zxx-4106 : : Tn10d(Tc) prsA501
DM12034	zxx-4106 : : Tn10d(Tc)
DM12035	zxx-4106 : : Tn10d(Tc) prsA503
DM12037	zxx-4106 : : Tn10d(Tc) prsA505
DM12062	zxx-4106 : : Tn10d(Tc) prsA502
DM12064	zxx-4106 : : Tn10d(Tc) prsA504
DM12067	purF2085 gnd-181 zxx-4106 : : Tn10d(Tc) prsA501
DM12069	purF2085 gnd-181 zxx-4106 : : Tn10d(Tc) prsA502
DM12071	purF2085 gnd-181 zxx-4106 : : Tn10d(Tc) prsA503
DM12073	purF2085 gnd-181 zxx-4106 : : Tn10d(Tc) prsA504
DM12075	purF2085 gnd-181 zxx-4106 : : Tn10d(Tc) prsA505
DM12076	purF2085 gnd-181 zxx-4106 : : Tn10d(Tc)
DM12131	gnd174 : : MudJ zxx-4106 : : Tn10d(Tc) prsA500
DM12141	gnd174 : : MudJ zxx-4106 : : Tn10d(Tc) prsA504
DM12616	E. coli BL21(AI) pET-prsA
DM12619	E. coli BL21(AI) pET-prsA505

Genetic techniques

Mutant isolation.

Overnight cultures of strain DM10373 (purF2085 trp-3618 gnd174 : : MudJ his-712) in NB medium were centrifuged and the cells were resuspended in an equal volume of saline. From each of 10 cultures, 100 μl was spread on NCE glucose medium supplemented with adenine, tryptophan, histidine and methionine before incubation at 37 °C for 72 h. Spontaneous mutations allowing growth arose at a frequency of ∼4×10⁻⁸. After phenotypic confirmation, a single colony derived from each culture was saved for further characterization. The strains used for the mutant selection were deleted for the trp and his operons to prevent isolation of mutations in these genes that were previously shown to respond to a similar selection. Supplementing the medium with tryptophan and histidine achieved the same thing. The gnd deletion was present to increase the stringency of the selection.

Strain construction.

Transductional crosses were performed using the high-frequency general transducing mutant of bacteriophage P22 (HT105/1, int-201) (Roberts, 1978; Schmieger, 1972). Methods for transduction and purification of transductants from phage have been described previously (Downs & Petersen, 1994).

Growth quantification.

Cells from overnight cultures in NB medium were pelleted and resuspended in an equal volume of saline (85 mM NaCl). A 150 μl aliquot of this suspension was used to inoculate 5 ml appropriate minimal medium. Cultures were incubated at 37 °C with shaking at 200 r.p.m. and cell density (OD₆₅₀) was monitored every hour in a Thermo Electron Corporation Spectronic 20D+ apparatus. Alternatively, 5 μl cell suspension was used to inoculate 195 μl appropriate minimal medium in each well of a 96-well microtitre plate. Growth at 37 °C with shaking at intensity level 2 was monitored using a microplate spectrophotometer Spectra-Max Plus. In each case, the specific growth rate was determined as μ=ln(X/X_o)/T, where X is the OD₆₅₀ value during the exponentially linear portion of growth (routinely between OD₆₅₀ 0.2 and 0.7) and T is time in hours. Doubling time (h) was determined as ln(2)/μ. Growth curves are presented in a linear format to visualize the growth lag and final OD₆₅₀. These features, in addition to growth rates, have been helpful in distinguishing growth properties of various strains.

Molecular biology.

Amplification of the prsA gene from chromosomal DNA of strain DM11258 (wild-type prsA) or DM11457 (prsA505) was performed by PCR using Herculase II fusion DNA polymerase. The primers used for amplification of prsA were 5′ PrsA (5′ CCTGATATGAAGCTTTTTGCTGG 3′) and PrsA3′ PstI (5′ GAGACTGCAGTCAATGCTCGAACATG 3′). PCR conditions were as follows: denaturation at 95 °C for 30 s, annealing at 53 °C for 30 s and extension at 72 °C for 30 s. The resulting 955 bp product was purified using a Qiagen PCR purification kit. The PCR product was then digested overnight with PstI and ligated into HincII- and PstI-digested pET45b (Novagen). The ligation reaction was electroporated into E. coli BL21(AI) and plated on selective medium. Sequence analysis confirmed the presence of an N-terminal 6×His tag and that the sequence of prsA was identical to that of wild-type (pET-prsA) or prsA505 (pET-prsA505).

Protein purification.

E. coli strain BL21(AI) containing pET-prsA (DM12616) or pET-prsA505 (DM12619) was grown in 4 l 2× LB+ampicillin medium at 37 °C with shaking. When the cell density (OD₆₅₀) reached 0.6, arabinose (0.2 %) and IPTG (1 mM) were added to the medium and the culture was moved to 28 °C and incubated overnight with shaking. Approximately 7 g cells was harvested by centrifugation and the cell paste was stored at −80 °C.

Frozen cell paste was resuspended in an equal volume of binding buffer (50 mM Tris pH 7.4, 100 mM NaCl and 5 % glycerol) containing DNase and lysozyme (both at 0.01 mg ml⁻¹) and passed through a French pressure cell (15 000 p.s.i.) four times at 4 °C. The resulting extract was clarified by centrifugation (42 000 g) for 45 min at 4 °C. Cell-free extract was loaded on a column with Superflow Ni²⁺ resin (Qiagen) (4 ml column volume) equilibrated with buffer (50 mM Tris pH 7.4, 500 mM NaCl, 5 % glycerol and 10 mM imidazole). Twenty column volumes of buffer were passed through the column prior to eluting PrsA with a linear gradient from 0 to 100 % elution buffer (50 mM Tris pH 7.4, 500 mM NaCl, 5 % glycerol and 500 mM imidazole) over 20 column volumes. Fractions containing recombinant PrsA at >95 % purity were pooled and concentrated at 30 p.s.i. under Argon gas using a 10 000 kDa molecular mass cut-off membrane (Amicon YM10). The protein was dialysed in binding buffer, frozen as beads in liquid nitrogen and stored at −80 °C. Protein concentration was determined using BCA kit (Pierce) according to the manufacturer's instructions. The pH of all Tris buffers was determined at room temperature.

PRPP synthetase assays

Purified protein.

Purified PrsA protein was assayed by a protocol modified from that of Ferrari et al. (1978). The assay quantifies PrsA-generated AMP using an NADH-coupled enzyme system. A 1 ml reaction mixture contained 50 mM potassium phosphate buffer (pH 7.5), 10 mM MgCl₂, 10 mM KCl, 5 mM R5P, 5 mM ATP, 1.5 mM PEP, 0.2 mM NADH, 20 U pyruvate kinase, 5 U myokinase and 10 U lactate dehydrogenase. The pH of the assay was routinely 8.1. Purified recombinant PrsA was added to initiate the reaction and the oxidation of NADH was measured by monitoring absorption at 340 nm using a Lambda Bio 40 spectrophotometer (Perkin Elmer Life Sciences) at room temperature. The specific activity of PrsA is expressed as μmol min⁻¹ (mg NADH oxidized)⁻¹, using the molar extinction coefficient of 6220 M⁻¹ cm⁻¹ for NADH.

Crude extract.

A 0.4 ml aliquot from an overnight NB culture was used to inoculate 4.6 ml NB medium. Strains assayed were DM11524, 12033, 12034, 12035, 12037, 12062 and 12064. Cultures were incubated with shaking at 37 °C until the OD₆₅₀ was ∼1.0. Pelleted cells were resuspended in 200 μl buffer (50 mM potassium phosphate buffer pH 7.5), disrupted by sonication, clarified by centrifugation (42 000 g, 45 min) and dialysed in resuspension buffer with a Microdialyser System 100 (Pierce). Buffer was added to bring the volume of each extract to 600 μl prior to centrifugation (42 000 g) for 45 min. NaF was added to 25 mM to inhibit cellular ATPases, and PrsA activity was determined using the assay described above. The R5P-independent rate of NADH oxidation was subtracted from the rate determined in the full reaction. Assays were done with two independent cultures each done in duplicate.

RESULTS

Mutant alleles of prsA allow PRA formation

A strain carrying deletions of purF, gnd, hisDCBHAFI and trpEDCBA (DM10373) failed to grow on medium containing adenine, tryptophan and histidine in the absence of thiamine. No colonies were detected with or without diethylsulfate mutagenesis, despite screening more than 10¹⁰ cells. The addition of 0.2 % vitamin-free casamino acids to the medium resulted in colonies arising at a frequency of ∼4×10⁻⁸. The component of the casamino acids which allowed thiamine-independent revertants to arise was methionine. Subsequent screens containing methionine in place of casamino acids resulted in a similar frequency of revertant colonies. After isolation, methionine was not necessary for thiamine-independent growth and the basis of the requirement during selection was not pursued.

Six independent colonies of DM10373 that arose on medium with adenine, tryptophan, histidine and methionine were isolated and phenotypically confirmed. An insertion 59 % linked to the causative mutation [zxx-4106 : : Tn10d(Tc)] in one of the isolates was identified. Sequence analyses located the insertion between STM1785 and STM1786 on the chromosome, ∼6 kb clockwise from the prsA gene, which encodes PRPP synthetase (E.C. 2.7.6.1). Insertion zxx-4106 : : Tn10d(Tc) was genetically linked to the causative mutation in each of five additional revertants of DM10373. The mutation allowing growth in each of the six strains was identified (Table 2). Five of the six mutations generated a variant PrsA protein and one allele (prsA503) was a change in the predicted ribosome-binding site (RBS), moving the site away from the consensus GAGGT to TAGGT (Bower et al., 1988).

Table 2.

prsA alleles that allow synthesis of PRA

nd, Not detected.

Strain	Allele*	DNA change†	Protein change	PRA‡	Specific activity§
DM12076	Wild-type	–	–	–	0.094±0.013
DM11461	prsA500	C542A	A181D	1.19	0.054±0.014
DM12067	prsA501	G149A	G50D	–||	nd
DM12069	prsA502	G916A	E306K	3.12	0.032±0.001
DM12071	prsA503	G−12T	None	1.24	0.041±0.012
DM12073	prsA504	A671C	D224A	2.26	0.027±0.005
DM12075	prsA505	G626T	G209V	1.67	0.016±0.001

*A purF gnd strain carrying any of the listed alleles is able to grow in the absence of thiamine. Assayed strains contained only a prsA mutant allele and are listed in Table 1.

†From the annotated LT2 genome, NCBI GeneID 1253299. Numbering starts at the first nucleotide of the coding sequence for PrsA. The protein change is indicated in the next column.

‡The ratio of doubling times for the indicated strain in adenine : adenine plus thiamine supplemented medium.

§Specific activity of PrsA in μmol NADH oxidized min⁻¹ (mg protein)⁻¹ in crude extract.

||Growth in defined liquid medium was not sufficient to allow phenotypic analysis.

Suppressor alleles of prsA reduce PRPP synthetase activity

Three results enabled us to conclude that the prsA mutations isolated here resulted in reduced PRPP synthetase activity. First, the prsA mutations were recessive. When provided in trans, the wild-type prsA gene reversed the behaviour of the mutant strains to that of the parent (data not shown). Second, when the prsA alleles were transduced into a wild-type genetic background, they generated a growth defect. Each of the resulting strains grew on minimal medium to some extent (Fig. 2a) but in all cases, growth was enhanced by the addition of PRPP-derived metabolites (Fig. 2b). Mutations in prsA that compromise PRPP synthetase activity have been isolated (Jochimsen et al., 1985; Post & Switzer, 1991) and a previously characterized temperature-sensitive allele (prsA100) displayed growth properties similar to the newly isolated alleles (Fig. 2).

Fig. 2.

Growth of prsA mutants is stimulated by PRPP-derived metabolites. Growth of strains containing mutant prsA alleles and a wild-type control are shown. Cultures were grown as described in Methods in minimal medium (a) and in minimal medium supplemented with guanosine (0.1 mM), uridine (0.1 mM), histidine (0.1 mM) and tryptophan (0.1 mM) (b). The strains contained prsA100 (•), prsA500 (○), prsA502 (▴), prsA503 (▵), prsA504 (⧫) and prsA505 (◊); wild-type (▪). Strain DM11524 (prsA500) grew to an OD₆₅₀ of ∼0.55 by 24 h in minimal medium.

Finally, biochemical assays in crude extract determined that the mutant strains had reduced PRPP synthetase activity compared with a wild-type strain. The PRPP synthetase activity in each of the mutant strains was less than 55 % of that found in the wild-type parental strain (Table 2), with no detectable activity in strain DM12033 (prsA501). These data could not distinguish between catalytically defective PrsA variants and unstable or under-expressed protein. For instance, the specific activity of strain DM12035 (prsA503) was ∼44 % that of wild-type extracts despite encoding wild-type PrsA. This result is consistent with the prediction that changed RBS in this strain decreases the amount of protein synthesized.

The biochemical data from crude extracts was complemented by assays with purified proteins. Biochemical data with purified proteins were available for two of the variants. The variant protein encoded by prsA505 was purified to >95 % homogeneity and assayed for PRPP synthetase activity. PrsA_G209V had a specific activity of 2.0±0.26 μmol NADH oxidized min⁻¹ (mg protein)⁻¹, 45-fold lower than that of the wild-type protein [94±7.6 μmol NADH oxidized min⁻¹ (mg protein)⁻¹]. Further, a PrsA_D224A variant protein (like that encoded by prsA504) from E. coli was previously shown to have a 10–15-fold lower specific activity than the wild-type protein in vitro (Willemoes et al., 1996).

R5P is the limiting metabolite for PRA formation on NCE medium

The non-enzymic formation of PRA from R5P and ammonia has been established in vitro and suggested in vivo (Nierlich & Magasanik, 1965; Ramos et al., 2008). A working model suggested constrained PRPP synthetase activity caused by the prsA mutant alleles resulted in accumulation of R5P, which could react non-enzymically with ammonia to form PRA sufficient for thiamine-independent growth (Fig. 1).

Medium composition was altered and PRA synthesis was monitored by thiamine-independent growth of various strains. Levels of R5P and ammonia (the two components of non-enzymic PRA synthesis) could be independently altered (Table 3). As previously reported, a purF gnd mutant (DM728) grows independently of thiamine on NCE medium with ribose as the sole carbon source (Petersen et al., 1996). NCE medium has high levels of ammonia (50 mM) and growth on ribose is expected to generate increased internal levels of R5P, an intermediate in the anabolism and catabolism of ribose (Davis et al., 1980; Sprenger, 1995; Vogel & Bonner, 1956). Consistently, reducing either R5P levels (by changing carbon source) or the level of ammonia (using N⁻C⁻ medium with 1 mM glutamine) prevented thiamine-independent growth of DM728 (purF gnd). N⁻C⁻ medium with ribose as the sole carbon source (1 mM glutamine as N source) did not allow thiamine-independent growth of DM728 (data not shown). From this control, we concluded that thiamine-independent growth in different media could provide insight into the mechanism used to generate PRA, distinguishing active synthesis from non-enzymic formation.

Table 3.

Growth medium impacts non-enzymic formation of PRA

Values given are the mean doubling time (h) of two independent cultures. ng, No change in OD₆₅₀ over a 24 h period. Ade, 0.4 mM adenine; Thi, 100 nM thiamine.

Strain	Mutant allele*	NCE glucose		NCE ribose		N−C− glucose
		Ade	Ade Thi	Ade	Ade Thi	Ade	Ade Thi
DM728	None	ng	0.85	1.52	1.44	ng	1.47
DM11461	prsA500	4.61	3.08	1.50	1.48	ng	2.93
DM12069	prsA502	3.51	1.41	2.91	2.90	ng	2.33
DM12071	prsA503	2.87	2.45	2.17	2.29	ng	2.68
DM12073	prsA504	3.07	1.80	2.56	2.58	ng	2.78
DM12075	prsA505	3.63	2.73	2.65	2.78	ng	2.42
DM3871	yjgF3	1.68	0.97	2.04	2.05	2.19	1.31
DM9890	trpC3620	1.91	1.64	2.49	2.85	ng	1.91

*All strains contain a deletion in purF and gnd and the indicated mutant allele that allows for thiamine-independent growth in NCE medium.

The trpC3620 allele allows PRA formation in a purF gnd mutant by causing accumulation of phosphoribosylanthranilate, which breaks down to R5P and anthranilate (Creighton, 1968; Ramos et al., 2008). The thiamine-independent growth of DM9890 (purF gnd trpC3620) was predicted to be due to non-enzymic PRA formation from the accumulated R5P (Ramos et al., 2008). Consistent with this scenario, strain DM9890 was unable to grow in the absence of thiamine when 1 mM glutamine was used as the sole source of ammonia (Table 3). A similar pattern of medium-dependent growth was displayed by the prsA mutant strains, supporting a model in which the prsA mutations increase cellular R5P pools that result in non-enzymic PRA formation when excess ammonia is present. Consistent with this interpretation was the finding that neither the trpC- nor the prsA-containing strains were able to grow in the absence of thiamine if N⁻C⁻ ribose medium was used.

Thiamine-independent growth correlates with the severity of the growth defect in minimal medium

Mutants lacking purF and gnd with each of five prsA alleles were assessed for growth in glucose adenine medium with or without thiamine. The ratio of growth of a strain in these two media provided a qualitative measure of PRA synthesis. A ratio close to 1 would indicate that addition of thiamine did little to improve growth and thus PRA synthesis was not the limiting factor for growth. Using this rationale, the four prsA alleles tested fell into two classes. The growth data for strains containing prsA503 (DM12071) and prsA500 (DM11461) are shown in Fig. 3a. For these strains, the ratios of doubling times in adenine vs adenine and thiamine medium were less than 1.25, indicating that there was little growth stimulation by thiamine. In contrast, the strains with prsA502 (DM12069) or prsA504 (DM12073) had ratios greater than 2.25 when the doubling times with and without thiamine were compared (Fig. 3b). Thus, in the latter set of strains, PRA synthesis limited growth since it was significantly increased by exogenous thiamine.

Fig. 3.

PurF-independent PRA synthesis is allowed by prsA mutant alleles. The growth of purF strains with mutant prsA alleles in minimal glucose adenine medium (open symbols) compared with growth with the addition of thiamine (filled symbols). (a) Strains with prsA500 (circles) and prsA503 (triangles) mutations and wild-type prsA (squares). (b) Strains with prsA502 (squares), prsA504 (circles) and prsA505 (triangles) mutations.

The data from Figs 2 and 3 indicated that the prsA alleles that resulted in a more severe general growth defect allowed PRA formation that did not limit growth. For instance, strains with prsA500 (DM11461) and prsA503 (DM12071) showed a severe growth defect on minimal medium and grew similarly with and without thiamine, indicating that PRA formation was not limiting. Strains with prsA502 (DM12069) or prsA504 (DM12073) had the opposite pattern; there was little growth defect on minimal medium but in the absence of purF, strains were unable to generate enough PRA to allow full thiamine-independent growth. The above data are consistent with a simple scenario in which PrsA activity is inversely correlated with the R5P accumulation and resulting PRA formation. The PrsA activities measured in crude extracts were inconsistent with this simple model, since the strains with the lowest activity (prsA502 and prsA504) had the lowest PRA formation and the least severe growth defect on minimal medium. Thus, although a correlation between activity and growth phenotypes existed, it was an inverse, rather than the expected direct correlation.

Based on these data, it is possible that a certain level of R5P is sufficient for PRA formation and other aspects of cell health may become compromised if the PrsA activity is further reduced. Two mutants support this general model. The first is prsA503; since this strain has wild-type protein, the growth phenotypes reflect a decreased level of protein not kinetic differences. This strain has a severe growth defect on minimal medium and makes sufficient PRA to allow thiamine-independent growth. Based on these data, an approximately 2.3-fold reduction in PRPP synthetase activity is sufficient to allow PRA formation for growth. Strains that have still lower PRPP synthetase activity and yet grow better may have activated other systems or possibly picked up suppressor mutations. Finally, allele prsA505 did not fit well into either class since it caused poor general growth and poor PRA formation. The prsA505 allele causes one of the most severe growth defects in a wild-type strain and encodes a variant with a specific activity 45-fold lower than wild-type enzyme. It is possible that this allele represents the level of activity at which the correlations identified above are no longer true, because the metabolic network cannot absorb this level of deficiency in PRPP synthesis.

Lesions in gnd and purF impact the need for PrsA

Deletion of the gnd gene [encodes 6-phosphogluconate dehydrogenase (E.C. 1.1.1.44)] prevents carbon flux through the oxidative pentose phosphate pathway (OPP) (Sprenger, 1995). R5P is an intermediate in the OPP and disruption of this pathway could impact the growth defect of a prsA mutant strain (Sprenger, 1995). Previous work suggested R5P synthesis via the OPP is involved in PurF-independent PRA synthesis (Enos-Berlage & Downs, 1996).

The impact of two mutations (purF and gnd) on the prsA mutant strains was assessed by growth on glucose adenine medium. Several prsA alleles were analysed in three relevant genetic backgrounds and representative data are shown in Fig. 4. The behaviour of the prsA504 allele was representative of most alleles, with the gnd and purF mutations having additive deleterious effects on growth. Strain DM12141 (gnd prsA504) had an increased lag compared with DM12064 (prsA504), and the presence of a purF deletion reduced both the growth rate and the final OD₆₅₀ further. This growth pattern can be explained in the context of our model. The R5P levels are lowered (due to the gnd deletion) causing a decrease in substrate needed to generate sufficient PRPP with a compromised enzyme. When purF is also deleted, thiamine must be synthesized from R5P and ammonia, requiring the strain to produce both PRA and PRPP with lower levels of R5P.

Fig. 4.

Disrupting flux through the OPP and/or PurF impacts growth of prsA mutants. Strains were grown in minimal glucose adenine medium. (a) Wild-type (▪), prsA504 (□), gnd prsA504 (▴), purF gnd prsA504 (▵). (b) Wild-type (▪), prsA500 (□), gnd prsA500 (▴), purF gnd prsA500 (▵).

A distinct pattern of growth was found with the prsA500 or prsA501 (data not shown) alleles in the three genetic backgrounds. Both of these alleles caused a severe growth defect in a wild-type background (Fig. 2). Interestingly, growth of the strains with these alleles was negatively impacted by a lesion in gnd, but positively impacted by a lesion in purF (Fig. 4b). Thus, lack of purF and gnd was better than lack of gnd only. One interpretation of these data is that the growth allowed by the presence of adenine is reversed by the loss of the R5P generated by the OPP (via a mutation in gnd). These data suggest that a lesion in purF indirectly made more PRPP available than the addition of adenine. Such a result would be consistent with the allosteric inhibition of PurF by exogenous adenine being less than 100 %. These results indicate that blocking flux through the purine pathway spares PRPP despite the fact that generation of ATP from exogenous adenine requires PRPP. Further, since these PrsA variants result in a severe growth defect, the requirement for PRA formation when purF is deleted may not put an additional burden on the strain (Fig. 3).

The above results indicate that while the non-oxidative branch of the pathway is not the major pathway for R5P synthesis (Shimizu, 2004), its contribution can be critical. This notion was previously suggested by genetic studies (Petersen et al., 1996) and metabolic flux analysis (Shimizu, 2004) despite the lack of a clear growth defect caused by a gnd mutation.

DISCUSSION

The metabolic network is a complex system with a high level of efficiency and robustness. To maintain these properties, the system must achieve a balance of metabolites that are generated and consumed by the integrated biochemical pathways in the cell. The formation of PRA, the first metabolite in purine/thiamine biosynthesis, provides a node to probe metabolic integration. This study identified mutant alleles of prsA that disrupted central metabolism and resulted in the formation of PRA. The general conclusion from this work was that constraining PRPP synthesis causes the accumulation of R5P that can non-enzymically combine with ammonia to generate sufficient PRA to support thiamine-independent growth. Each of the six prsA alleles isolated was shown to reduce PRPP synthetase activity in the cell. Consistent with results in crude extract, two of the protein variants had significantly decreased specific activity when assayed in vitro.

Fig. 5 shows an alignment of PrsA homologues, including the one from Bacillus subtilis, for which a structure has been solved (Eriksen et al., 2000). In most cases the mutant alleles isolated herein affect conserved residues. PrsA_A181D, PrsA_D224A and PrsA_G209V all have changes in conserved residues located in or near the active site of the protein. However, PrsA_G50D and PrsA_E306K do not have changes in conserved residues, rather these changes are located in the region of the protein that interacts with additional subunits for oligomerization. While these residues are not conserved, the location of the changes suggests that they may result in a protein with a defect in oligomerization, leading to decreased activity.

Fig. 5.

Alignment of PrsA sequences from S. enterica strain LT2 (Se), Bacillus subtilis strain 168, Methanocaldococcus jannaschii strain DSM2661 and Homo sapiens. The numbering refers to residues from the S. enterica strain LT2 PrsA. The locations of the protein variants indentified in this study are shown.

Somewhat surprisingly, beyond the general conclusion stated above, the correlation between growth phenotypes and cellular PRPP synthetase activity was not upheld. Distinct from thiamine synthesis, the growth defect of the prsA mutants was inversely correlated with the PRPP synthetase activity in crude extracts. This was an unexpected result if the growth defect is a direct consequence of the level of PRPP formed. Rather, the growth defect may reflect a complex regulatory/metabolic network that can redirect flux. In this scenario, severely reduced PRPP synthetase (e.g. prsA505) would compromise the system beyond its ability to respond. The similar growth phenotypes of mutants with the same range of PRPP synthetase activity make it unlikely that suppressor mutations are arising and suggest that this behaviour is a consequence of interactions in the metabolic network that we don't yet understand.

The mutations in prsA exemplify the compromise that a cell must make in altering one component of metabolism to address a problem elsewhere in the system. In the case of prsA, the mutations disrupted an essential enzyme, leading to severe metabolic consequences. However, this metabolic disruption had the positive effect that it provided for the cellular requirement of thiamine. Importantly, the prsA alleles isolated had to result in a flux balance that allowed growth both when flux to thiamine via this disruption was needed and when it was not. The strains that thrived in the first environment struggled in the second; thus the balance achieved is a reflection of the robustness that exists in the metabolic network.

The results herein have identified and defined the consequences of metabolic integration between PRPP and thiamine biosynthesis. This connection allowed us to define two opposing consequences of disrupting a single pathway. Several observations from this work are worth noting. First, this study defines a condition where the non-enzymic synthesis of a metabolite is essential for growth of S. enterica. Additionally, this study is a new report of a positive selection for mutations that negatively affect the enzymic activity of PrsA. This method could be used for further study of this key essential enzyme, defects in which have been implemented in several clinical diseases (Yen et al., 1978). Finally, this study shows that a single reaction in the metabolic network cannot be considered in isolation, since the substrate and/or product may be working beyond the described metabolic role in altering the metabolic network and differentially impacting growth in different environments.

Abbreviations

• NCE, no-carbon E medium

• OPP, oxidative pentose phosphate pathway

• PRA, phosphoribosylamine

• PRPP, phosphoribosylpyrophosphate

• R5P, ribose 5-phosphate