Altered Pathway Routing in a Class of Salmonella enterica Serovar Typhimurium Mutants Defective in Aminoimidazole Ribonucleotide Synthetase

Abstract

In Salmonella enterica serovar Typhimurium, purine nucleotides and thiamine are synthesized by a branched pathway. The last known common intermediate, aminoimidazole ribonucleotide (AIR), is formed from formylglycinamidine ribonucleotide (FGAM) and ATP by AIR synthetase, encoded by the purI gene in S. enterica. Reduced flux through the first five steps of de novo purine synthesis results in a requirement for purines but not necessarily thiamine. To examine the relationship between the purine and thiamine biosynthetic pathways, purI mutants were made (J. L. Zilles and D. M. Downs, Genetics 143:37–44, 1996). Unexpectedly, some mutant purI alleles (R35C/E57G and K31N/A50G/L218R) allowed growth on minimal medium but resulted in thiamine auxotrophy when exogenous purines were supplied. To explain the biochemical basis for this phenotype, the R35C/E57G mutant PurI protein was purified and characterized kinetically. The K_m of the mutant enzyme for FGAM was unchanged relative to the wild-type enzyme, but the V_max was decreased 2.5-fold. The K_m for ATP of the mutant enzyme was 13-fold increased. Genetic analysis determined that reduced flux through the purine pathway prevented PurI activity in the mutant strain, and purR null mutations suppressed this defect. The data are consistent with the hypothesis that an increased FGAM concentration has the ability to compensate for the lower affinity of the mutant PurI protein for ATP.

In Salmonella enterica serovar Typhimurium LT2, purines and thiamine are synthesized via a divergent pathway where 5-aminoimidazole ribonucleotide (AIR) is the intermediate at the branch point (Fig. 1). Since the cellular purine requirement is approximately 10³-fold higher than the thiamine requirement (based on auxotrophic requirements), this pathway provides a model to address control of an important metabolic branch point. Previous genetic and molecular analyses demonstrated that even 1% of the wild-type level of AIR synthetase was sufficient to supply the cellular requirement for thiamine but not purines (J. L. Zilles and D. M. Downs, submitted for publication), indicating that thiamine synthesis can be maintained even when flux through the common pathway is severely reduced. Under this condition, thiamine synthesis could continue if levels of the substrate (presumed to be AIR) remained above the K_m for the first committed thiamine enzyme or if there were metabolite channeling between PurI and the thiamine enzyme (thought to be ThiC). Mutational analysis of purI, encoding the final common enzyme AIR synthetase, was pursued to clarify the parameters controlling thiamine synthesis with changing flux through the purine pathway.

FIG. 1.

Biosynthesis of purines and thiamine. Purine enzymes are noted below the arrow representing the reaction that they catalyze. THZ-P, 4-methyl-5-(β-hydroxyethyl)thiazole monophosphate; HMP-PP, 4-amino-5-hydroxyethyl-2-methylpyrimidine pyrophosphate.

The synthesis of AIR, via the pathway common to purine and thiamine synthesis, appears to be regulated only in response to exogenous purines. There are three known levels of regulation on this pathway: (i) transcription of pur genes is repressed by PurR (with its corepressors hypoxanthine and guanine) (17, 18, 22, 28, 33, 39), (ii) allosteric inhibition of the first committed step in purine biosynthesis (phosphoribosylpyrophosphate amidotransferase, PurF) by AMP and GMP (24), and (iii) control of the levels of phosphoribosylpyrophosphate (PRPP), a substrate for the PurF enzyme. The level of PRPP in the cell drops substantially in the presence of exogenous purines (2, 19). Labeling studies suggest that exogenous adenine reduces flux through the purine biosynthetic pathway to 10% of that on minimal medium (32).

The current data on the purine-thiamine branch point are consistent with a model in which the flux to each branch of the purine-thiamine pathway depends on the concentration of AIR and the kinetic properties of the enzymes competing for AIR as a substrate. The primary phenotypic consequence of reduced flux through the common pathway is a purine requirement (Zilles and Downs, submitted). However, mutations that result in a thiamine (but not purine) requirement when flux through the purine pathway is reduced have been isolated and characterized (4, 5, 13, 30, 31). In general, these mutations appear to indirectly affect the thiamine biosynthetic pathway subsequent to the purine-thiamine branch point. The identification of mutations in the biosynthetic gene purI (encodes AIR synthetase in S. enterica, homologous to purM in Escherichia coli) that caused a conditional requirement for thiamine was unexpected. One of these mutants is the subject of this paper.

AIR synthetase catalyzes the conversion of formylglycinamidine ribonucleotide (FGAM) to AIR, ADP, and P_i (39) and has been purified and characterized from both chicken liver and E. coli (34, 35). Kinetic studies with the E. coli enzyme suggested a sequential mechanism in which ATP bound first and ADP was released last (35). The structure of AIR synthetase from E. coli has recently been solved, and the enzyme is believed to represent a new class of ATP-binding proteins (21, 27). The ATP-binding site in AIR synthetase was identified based on sequence alignments, structural considerations, and studies with an ATP affinity label (27). In this report we present the isolation of one purI mutant that can support purine synthesis but requires thiamine under some growth conditions. Biochemical analysis of the mutant PurI protein identified a defect in ATP binding that, in combination with the sequence analysis, supported the proposed location for the ATP-binding site of AIR synthetase (21, 27). Phenotypic and suppressor analyses indicated that high levels of FGAM were required for function of the mutant enzyme in vivo, suggesting that increased levels of FGAM can compensate for the decreased affinity of the mutant enzyme for ATP.

MATERIALS AND METHODS

General procedures.

All strains used in this study are derivatives of S. enterica LT2 and are listed with their genotypes in Table 1. Unless otherwise indicated, strains were part of the lab collection or were constructed during the course of this work. Transductions were performed as described previously (12). Plasmid pPurF was isolated from a plasmid library (pBR328) of S. enterica DNA by its ability to complement a purF deletion mutant. PCR analysis confirmed the presence of the purF gene.

TABLE 1.

Strains used^a

Strain	Relevant genotype
LT2	Wild type
DM42	purI2152::MudJ
DM2148	purI1757::Tn10
DM3593	purI2152::MudJ p103 (Cmr)
DM3995	purI2152::MudJ pTE1 (Cmr)
DM4794	purI3098
DM4825	purI3098 purR2319::Tn10d(Tc)
DM4826	purI3098 add-2346::MudJ
DM4829–DM4834	purI3098 purR3105–purR3110
DM5045	purI3098 zdx-9103::MudJ
DM5171	purI3098 purR2319::Tn10d(Tc) purA1881::MudJ
DM5172	purI3098 purR2319::Tn10d(Tc) purC2156::MudJ
DM5174	purI3098 pPurF1 (Ampr)
DM5175	purI3098 pJS187 (Ampr)
DM5176	purI3098 p40 (Ampr)
DM5180	purI3098 purG2324::MudJ
DM5181	purI3098 purG2324::MudJ purR2319::Tn10d(Tc)
DM5182–DM5187	purI3098 purG2324::MudJ purR3105–purR3110
DM5495	purI2152::MudJ pPurI-his (Cmr)
DM5496	purI2152::MudJ pPurI3098-his (Cmr)

^aMudJ is used throughout to refer to the Mu dI 1734 transposon described previously (9). Tn10d(Tc) is used throughout to refer to the transposition-defective mini-Tn10 (Tn10del-16 del-17) described by Way et al. (38). pJS187 is a previously described purD clone (36), and p40 is a previously described purG clone (40). 

No-carbon E medium supplemented with 1 mM MgSO₄ (11, 37) and glucose (11 mM) as a carbon source was used as a minimal medium, and Difco nutrient broth (8 g/liter) with NaCl (5 g/liter) was used as a rich medium. Difco BiTek agar was added (15 g/liter) for solid medium. Adenine and thiamine were included in media as needed to the final concentrations of 0.4 mM and 50 or 500 nM (liquid or solid medium), respectively. Antibiotics were added as needed to the following concentrations in rich/minimal media: carbenicillin, 50/50 μg/ml; kanamycin, 50/125 μg/ml; and chloramphenicol, 20/4 μg/ml. Phenotypic tests were performed by replica printing or by monitoring A₆₅₀ in a 96-well microtiter plate using the Spectramax Plus plate reader (Molecular Devices, Sunnyvale, Calif.).

Formylglycinamide ribonucleotide (FGAR) was produced using purified E. coli purine biosynthetic enzymes and isolated as described (25, 26). FGAM was produced from FGAR using purified E. coli FGAR amidotransferase and isolated as described (26, 35). IPTG (isopropyl-β-d-thiogalactopyranoside) was purchased from Fisher Biotech. Other chemicals were purchased from Sigma Chemical Co. (St. Louis, Mo.). Restriction enzymes and DNA ligase were purchased from Promega (Madison, Wis.).

To identify the wild-type purI sequence from S. enterica, the purI coding region was sequenced from a previously identified clone (p42) at least twice on both strands (GenBank accession number U68765) (40). For analysis of mutant alleles, the purI coding region was amplified from the chromosome of the appropriate strains using primers based on the wild-type sequence. Amplification was performed using Vent (exo⁻) polymerase (New England Biolabs) in a Thermolyne Temp-Tronic thermocycler. Plasmid templates were purified using QIAprep spin miniprep kits (Qiagen), and PCR products were purified using the Qiaquick gel extraction kit (Qiagen). For each allele or construct, the complete coding region of purI was sequenced from both strands from at least two independent PCRs. Sequencing was performed by the University of Wisconsin Biotechnology Center–Nucleic Acid and Protein Facility (Madison, Wis.). Sequence data were examined using EditView (ABI Prism; Perkin Elmer) and aligned using SeqEd (Applied Biosystems).

Strain construction. (i) Thiamine-requiring purI mutants.

The purI coding region was amplified using the error-prone Vent (exo⁻) polymerase. PCR conditions were as follows: initial denaturation at 95°C for 5 min followed by 30 cycles of denaturation at 95°C for 1 min, annealing at 62°C for 1 min, and extension at 72°C for 1.5 min. Resulting PCR products were purified and blunt-end ligated into SmaI-cut pSU19, and the plasmids were moved into a purI null mutant (DM42) by electroporation as described (14), selecting for pSU19-encoded chloramphenicol resistance. The resulting colonies were screened for a purine-sensitive phenotype: growth on minimal medium and minimal medium supplemented with adenine and thiamine but no growth with adenine alone. Plasmids from putative purine-sensitive mutants were purified and reelectroporated to verify that the causative mutations were contained on the plasmid. To ensure independence, only one purine-sensitive mutant was saved from each amplification.

(ii) Site-directed mutagenesis.

Site-directed mutagenesis was performed using the megaprimer method as described previously (3). The resulting PCR product was blunt-end ligated into SmaI-digested pSU19 and electroporated into a purI insertion mutant (DM42), selecting for chloramphenicol resistance, and replica printing to assess growth on minimal medium. Putative clones were confirmed by sequencing, and their phenotype was assessed by replica printing.

(iii) Generating chromosomal alleles of purI mutant.

The plasmid-encoded purine-sensitive purI allele was removed from p103 by restriction digestion, gel purified, and ligated into the pMAK705 vector, which encodes a chloramphenicol resistance marker and has a temperature-sensitive origin of replication (16). The construct of interest was identified by electroporating into a purI null mutant (DM42) at 30°C (the permissive temperature), selecting for both chloramphenicol resistance and growth on minimal medium. A phage lysate was then grown on a strain containing the purI allele in pMAK705 and used to transduce a purI insertion mutant (DM2148) to prototrophy at 42°C (the restrictive temperature). Transductants were screened for chloramphenicol sensitivity, tetracycline sensitivity, and purine sensitivity, indicating loss of the plasmid, loss of the purI insertion, and presence of the mutant allele, respectively. The presence of the mutant allele was confirmed by sequence analysis.

(iv) Isolation and mapping of suppressor mutations.

Six independent, spontaneous suppressor mutations were identified by plating saline cell suspensions of six independent overnight cultures of a strain carrying the purine-sensitive allele purI3098 (DM4794) on minimal glucose adenine and incubating at 37°C for 2 days. The suppressor phenotype was confirmed by replica printing and growth curves, and one suppressor mutant was saved from each culture (strains DM4829 to DM4834). To facilitate characterization of these spontaneous mutants, an insertion linked to a suppressor mutation was identified as follows. A suppressed strain (DM4829, purI3098 purR3105) was transduced to kanamycin resistance using phage grown on a MudJ pool. The transduction was performed on minimal medium to eliminate purine auxotrophs. Transductants were then screened for a purine-sensitive phenotype, indicating that the suppressor mutation had been replaced with a wild-type allele. The linkage between a MudJ insertion and the suppressor mutation was verified during reconstruction. The resulting insertion, zdx-9103::MudJ, was subsequently found to be linked to all six spontaneous suppressor mutations.

The chromosomal location of the suppressor mutations was identified by first mapping the linked insertion zdx-9103::MudJ. A MudJ-specific primer and an arbitrary primer were used to amplify sequences flanking the MudJ insertion as previously described (8, 29). The resulting PCR product was sequenced and compared to the E. coli genome sequence using the Blast program (1). The positions of the MudJ insertion and the suppressor mutations at 37.3 min were verified by testing them for linkage to an insertion mutation in the pykF gene, which is located nearby in E. coli.

Biochemical analysis. (i) Protein purification.

Plasmids were constructed containing the appropriate purI alleles in the pET20b vector, which added 13 amino acids to the carboxyl terminus of AIR synthetase (KLAAALEHHHHHH) (Novagen). The constructs were confirmed by phenotypic analysis for purine sensitivity in a purI null mutant of S. enterica and by sequence analysis before being electroporated into the E. coli strain BL21/λ(DE3) for overexpression. For unknown reasons, the plasmid containing the mutant allele was not stable in this background; induction of the R35C/E57G protein was therefore performed in the presence of an additional plasmid, pLysS (Novagen), which reduced preinduction expression and allowed maintenance of the plasmid. Cultures (12 ml) grown overnight in Luria-Bertani (LB) with carbenicillin (LB-Cb) were used to inoculate 400 ml of LB-Cb. The resulting culture was grown at 37°C with shaking to 80 Klett units (red filter), at which point solid IPTG (Fisher Biotech) was added to a final concentration of 0.4 mM. After 2 h of additional growth, the cultures were collected by centrifugation at 4°C at 5,000 × g for 10 min. Cell extracts were prepared as follows. Cell pellets were resuspended in 4 ml of buffer (500 mM NaCl, 20 mM Tris-HCl [pH 7.9], 5 mM imidazole), sonicated on ice, spun at 4°C at 39,000 × g for 20 min to remove debris, and filtered through a 0.45-μm filter (Gellman Laboratory).

The histidine-tagged proteins were purified on a nickel affinity column at 4°C according to the manufacturer's instructions (Novagen), with two modifications. First, the wash volume was increased from 15 to 25 ml. Second, while PurI protein was found both in the elution (1 M imidazole, 500 mM NaCl, and 20 mM Tris-HCl [pH 7.9]) and in the strip buffer wash (100 mM EDTA, 500 mM NaCl, and 20 mM Tris-HCl [pH 7.9]), only the protein in the strip buffer wash was used in subsequent analysis. The EDTA in 10 ml of strip buffer was removed by dialyses against two changes of 2 liters of buffer containing 20 mM Tris (pH 8), 100 mM KCl, 10% glycerol, and 5 mM β-mercaptoethanol (calculated equilibrium concentration, 2.5 μM EDTA). The protein was concentrated using Ultrafree-15 centrifugal filter devices (Millipore), quantified by the Bradford assay (7), frozen in aliquots on dry ice, and stored at −80°C.

(ii) Assays.

AIR synthetase activity was assayed by detecting formation of AIR using a previously described modification of the Bratton-Marshall assay for diazotizable amines (34, 35). To conserve reagents, the assays were performed in small volumes. In a typical experiment, a volume of 40.5 μl contained 50 mM HEPES (pH 7.7), 20 mM MgCl₂, 150 mM KCl, and various concentrations of ATP and FGAM. The reaction mixture was equilibrated at 37°C, and the reaction was initiated by the addition of enzyme: 75 ng of wild-type AIR synthetase or 300 ng of mutant AIR synthetase in 4.5 μl. Enzymes were diluted in buffer containing 50 mM HEPES (pH 7.7), 50 mM KCl, and 10% glycerol. After 2 min, the reaction was stopped by the addition of 10 μl of 20% trichloroacetic acid–1.33 M potassium phosphate (pH 1.4). Remaining reagents (sodium nitrite [0.016% (wt/vol) final concentration], ammonium sulfamate [0.08% final concentration], N-(1-naphthyl)ethylenediamine dihydrochloride [0.01% final concentration]) were added as described to a final volume of 85 μl (34, 35). After allowing the color to develop for 30 min, absorbance was read at 500 nm in a quartz microtiter plate on the Spectramax Plus plate reader and corrected to a pathlength of 1 cm (Molecular Devices). For the determination of ATP kinetic constants, FGAM was constant at 100 μM and ATP was varied from 25 μM to 5 mM (wild-type enzyme) or 0.2 to 7.5 mM (mutant enzyme). For the determination of FGAM kinetic constants, ATP was constant at 0.5 mM (wild-type enzyme) or 5 mM (mutant enzyme), determined in an earlier experiment (data not shown) to be saturating (five times the K_m) for each enzyme, and FGAM was varied between 10 and 200 μM (wild-type enzyme) or 15 and 160 μM (mutant enzyme). At low substrate concentrations, one or more of the following was doubled: reaction time, amount of enzyme, or reaction volume. Control experiments showed the expected linear dependence of the amount of AIR produced on reaction time and amount of enzyme. Kinetic constants were calculated using the programs of Cleland (10) and plotted using Kaleidagraph (Abelbeck Software).

RESULTS

Isolation of purI mutations resulting in a conditional Thi⁻ phenotype.

To address whether enzymes at the purine-thiamine branch point could alter the distribution of the common metabolite, purI was mutagenized. Strains in which the only source of AIR synthetase was a plasmid-encoded, PCR-mutagenized purI gene were grown on minimal medium and screened for thiamine auxotrophy in the presence of exogenous adenine. Two independent mutants were isolated and reconstructed, verifying the involvement of the plasmid-encoded purI allele in the phenotype. Plasmid p103 (strain DM3593) contained three mutations predicted to cause amino acid substitutions R35C, E57G, and R198W in AIR synthetase. The second plasmid, pTE1 (strain DM3995), also contained three mutations in the purI coding sequence, corresponding to amino acid changes of K31N, A50G, and L218R.

The mutant enzyme encoded by p103 (R35C/E57G/R198W) was characterized further. Three plasmids, containing each of the respective single mutations, were constructed by site-directed mutagenesis and verified by sequencing. None of the plasmid-encoded single mutants displayed the original phenotype, indicating that more than one mutation in purI was required to generate a conditional thiamine requirement. To avoid possible contributions of multicopy expression to the phenotypic analysis, the original mutant purI allele was recombined into the chromosome (see Materials and Methods), resulting in strain DM4794 (16). The fortuitous location of the recombination events resulted in a recombinant that contained the R35C and E57G substitutions but not the R198W substitution. Since the chromosomal allele of purI in strain DM4794 carried only two mutations and resulted in a purine-dependent thiamine auxotrophy, we concluded that these two mutations were both necessary and sufficient for this phenotype. Unless noted, subsequent analyses were performed with this purI3098 allele of AIR synthetase (R35C/E57G).

The purine dependence of the thiamine requirement in strain DM4794 is illustrated by the growth data in Fig. 2. Two points were noted from these data. First, the mutant strain grew with a nearly wild-type rate on minimal medium. This finding demonstrated the ability of the mutant enzyme to provide adequate AIR for both purine and thiamine synthesis. Second, the presence of adenine in the medium prevented growth unless thiamine was also supplied. Growth in the absence of thiamine was similarly impaired by the addition of adenosine, hypoxanthine, or inosine (data not shown). Since exogenous purines are known to inhibit flux through the purine pathway (32), a simple model was that the mutant enzyme was unable to function when cellular levels of purine intermediates, including the substrate FGAM, were reduced. If this model is correct, the mutant protein might be expected to have an increased K_m for FGAM.

FIG. 2.

Purine-sensitive phenotype of R35C/E57G AIR synthetase mutant. Growth of the wild-type (A) and PurI R35C/E57G mutant (B) strains is shown. Inocula were grown to full density in minimal glucose medium before subculturing into minimal medium (○) or minimal medium with 0.4 mM adenine (▵). Solid symbols represent growth with 50 nM thiamine added. A representative experiment is shown; similar results were seen with at least four independent cultures.

Guanine did not have a significant effect on the growth of DM4974 in the absence of thiamine. This difference between adenine and guanine has been observed for other conditional thiamine auxotrophs, perhaps reflecting the different ability of purine sources to repress the purine biosynthetic pathway (M. Frodyma and D. M. Downs, unpublished data). Although purine flux has not been measured directly in the presence of exogenous adenine or guanine, two previous reports suggest that guanine affects purine flux less than adenine. First, guanine repressed a purE-lacZ transcriptional fusion less than a variety of other purine sources, including adenine (22), and second, levels of PRPP, a substrate of the first committed step in purine biosynthesis, were reduced 90% by exogenous adenine but only 50% by exogenous guanine (19). However, this difference between guanine and adenine could also reflect a specific inhibition of the mutant AIR synthetase by adenine nucleotides.

R35C/E57G mutant AIR synthetase has an increased K_m for ATP.

Thirteen amino acids (including six histidine residues) were fused to the carboxy terminus of wild-type and R35C/E57G mutant AIR synthetase, and the proteins were purified using a nickel affinity column. The purified fusion proteins were assayed for AIR synthetase activity as described previously (34, 35). Both the wild-type and mutant proteins were stable and active after purification. The kinetic plots are shown in Fig. 3, and the resulting kinetic constants are shown in Table 2. As anticipated from the high degree of sequence identity (92%), the kinetic constants of the wild-type S. enterica AIR synthetase were similar to those reported for the wild-type E. coli enzyme (35).

FIG. 3.

Kinetic analysis of purified S. enterica AIR synthetase. Data for the wild-type (open circles) and the R35C/E57G mutant (solid circles) strains are shown. (A) Concentration of ATP was fixed at approximately 5 K_m, 0.5 mM (wild type) or 5 mM (mutant). (B) Concentration of FGAM was fixed at 0.1 mM and the ATP concentration was varied. The lines are least-squares fits of the kinetic data (solid, wild type; dotted, R35C/E57G mutant) to a hyperbolic rate expression, calculated using the programs of Cleland (10).

TABLE 2.

Kinetic parameters for S. enterica AIR synthetase^a

Substrate	AIR synthetase	Km (mM)	Vmax (μmol min−1 mg−1)	V/K (U/mg/mM) (% of wild-type value)
ATP	Wild type	0.126 ± 0.033	5.2 ± 0.4	41
	R35C/E57G	1.68 ± 0.89	2.7 ± 0.5	1.6 (4)
FGAM	Wild type	0.049 ± 0.012	6.8 ± 0.6	139
	R35C/E57G	0.066 ± 0.022	2.1 ± 0.3	32 (23)

^aAIR synthetase assays were performed as described in the text, and kinetic constants were determined using the programs of Cleland (10, 34, 35). Values are means ± standard deviations. 

The R35C/E57G AIR synthetase showed a 13-fold increase in the K_m for ATP compared to the wild-type enzyme from S. enterica, while retaining a wild-type K_m for FGAM. Although the R35C/E57G AIR synthetase showed a two- to threefold decrease in V_max, we considered it unlikely that this difference was significant in vivo. The difference in K_m for ATP was likely to be significant in vivo, since concentrations of ATP in E. coli or S. enterica cells grown on minimal medium are reported to be 1.8 to 3 mM (6, 39). This puts the intracellular level of ATP well above the K_m of the wild-type enzyme but near the K_m of the mutant. Since both R35 and E57 are near the proposed ATP binding site (Fig. 4), this biochemical analysis is consistent with the proposed ATP-binding site of AIR synthetase (21, 27).

FIG. 4.

Location of mutated residues in S. enterica AIR synthetase. A model of S. enterica AIR synthetase was generated automatically by Swiss-Model (15) using the coordinates for E. coli AIR synthetase (PDB entry 1CLI) as input (21). As the two proteins are 92% identical (314 of 341 crystallographically located residues), it is anticipated that this is a highly accurate model. Each active site is defined by a groove composed of residues from both subunits in the dimer. The groove, which runs into the plane of the page, is bracketed by the ATP-binding loop (labeled ATP, at the front), initially identified by crystallography and affinity labeling (27), and an inorganic sulfate (at the back), proposed to locate the binding site for FGAM (21). The N termini of each subunit are labeled; in subunit A the first residue is Thr-5, and in subunit B the first residue is Ala-21. (Top) Locations of mutations in the R35C/E57G protein are shown in space-filling models. Note that R198W, found in strain DM3593, is not shown because it is unnecessary for the purine-sensitive phenotype. (Bottom) Locations of mutations in the K31N/A50G/L218R protein (strain DM3995) are shown in space-filling models. This figure was prepared using Molscript (20) and Raster3D (23).

Increased flux through the purine pathway suppresses the in vivo defect of R35C/E57G AIR synthetase.

The biochemical parameters of the mutant enzyme were inconsistent with the initial model suggested by the in vivo phenotype, which predicted an increased K_m for FGAM. Exogenous purines have been reported to increase ATP levels slightly, which if anything might be expected to increase the activity of the mutant protein (2). However, in vivo the mutant protein appeared to be inhibited, either directly or indirectly, by exogenous purines. To resolve this paradox, we isolated and characterized the most frequent class of mutations that suppressed the purine-dependent thiamine requirement.

Six independent, spontaneous mutations allowing the R35C/E57G AIR synthetase mutant (DM4794) to grow in the presence of purines were isolated (strains DM4829 to DM4834). Each of these mutations also allowed growth of the K31N/A50G/L218R AIR synthetase mutant under similar conditions. Standard genetic techniques were used to place these mutations at 37 min on the S. enterica chromosome. Significantly, one gene in this chromosomal region is purR, a global repressor whose regulon includes the purine biosynthetic genes (39). Two reconstruction experiments suggested that these six mutations were null alleles of purR. First, the presence of any of the six suppressor mutations increased the expression of a purG-lacZ fusion 8- to 10-fold above the wild-type level (e.g., 390 versus 46 Miller units, respectively, in a representative experiment), similar to the effect of a purR insertion mutation (31). Second, a known purR insertion mutation exhibited the suppressor phenotype and restored growth of the purI3098 mutant strain in the presence of adenine.

The purR mutation might restore thiamine-independent growth by increasing expression of any gene in the purR regulon. Analysis with insertion mutations in purC or purA and multicopy clones of purG or purD eliminated these genes as targets for this effect (data not shown). However, introduction of a plasmid containing purF (pPurF1) into the purI3098 mutant restored wild-type growth in the presence of exogenous purines. This result suggested that the purR mutations suppressed the conditional thiamine auxotrophy by increasing levels of the PurF enzyme, PRPP amidotransferase. Because PurF catalyzes the first biosynthetic step and is the primary site of allosteric regulation of the pathway, it was reasonable that increased levels of this enzyme would allow more flux through the purine pathway in the presence of purines (2, 17–19, 22, 24, 28, 33, 39). Although it was formally possible that suppression of the purine-dependent thiamine auxotrophy by purR mutations resulted from increased transcription of the mutant purI allele, the ability of a purF clone to suppress the same phenotype suggested that elevated flux through the purine pathway resulted in suppression.

In related work, reconstruction experiments determined that a mutation in add (encoding adenosine deaminase) was able to suppress the conditional thiamine auxotrophy of the purI3098 mutant. Adenosine deaminase is required for the synthesis of hypoxanthine, a corepressor for PurR (18, 22, 39). Thus, the add mutation was likely to prevent full activation of the PurR protein and result in derepression of the purine genes. Taken together, the ability of a purR mutation, a purF clone, and an add mutation to eliminate the conditional thiamine auxotrophy supports the conclusion that increased flux through the purine pathway is required for efficient function of the R35C/E57G AIR synthetase mutant.

DISCUSSION

We report here the initial genetic and biochemical analyses of one phenotypic class of AIR synthetase mutants. Although the mutants were prototrophic on minimal medium, in the presence of exogenous purines they exhibited thiamine auxotrophy. This phenotype was unexpected, since previous analysis demonstrated that as little as 1% of the wild-type level of AIR synthetase was sufficient for growth in the presence of exogenous purines but not for growth on minimal medium (Zilles and Downs, submitted). The conditional thiamine requirement described here was suppressed by mutations that increased flux through the common pathway in the presence of purines. A simple interpretation of this genetic analysis suggested that the mutant enzyme might have an increased K_m for FGAM. However, kinetic analysis of the mutant AIR synthetase showed no change in the K_m for FGAM but a 13-fold increase in the K_m for MgATP.

While this work was in progress, new structural data on AIR synthetase became available. The AIR synthetase ATP-binding site was proposed on the basis of structural constraints and ATP analog affinity labeling (21, 27). Structurally, the mutant residues in this work (R35C and E57G) are near the proposed ATP binding site (Fig. 4), consistent with the mutations affecting the K_m for MgATP. Although a thorough analysis of the second reported purI mutant was not performed, it is worth noting that it also contained two amino acid substitutions near the proposed ATP binding site (Fig. 4). The mutant described in this work represents the first analysis of a mutant AIR synthetase enzyme with significant impairment of the ATP site. The structural analyses shown in Fig. 4 also show that the mutant residues are surface exposed and appear to lie in a single groove.

The connection between the high K_m for MgATP of the mutant enzyme and the resulting in vivo phenotype was not immediately apparent. It is likely that the nutritional requirement caused by the R35C/E57G mutant is a consequence of in vivo metabolite levels affecting the activity of the mutant PurI enzyme. Efforts to characterize these effects more thoroughly or to mimic them in vitro have been hampered by the difficulties involved in obtaining accurate measurements of metabolite pool sizes, particularly for unstable metabolites such as FGAM. Several possible models could explain the nutritional phenotype of the mutant in the context of biochemical parameters determined for this enzyme in vitro. These models include (i) increased FGAM altering the reaction order of the mutant enzyme, (ii) disruption of metabolite channeling in the mutant, (iii) stabilization of the mutant PurI by FGAM, and (iv) inhibition of the mutant PurI by elevated nucleotide levels.