The panE Gene, Encoding Ketopantoate Reductase, Maps at 10 Minutes and Is Allelic to apbA in Salmonella typhimurium

Abstract

In Salmonella typhimurium, precursors to the pyrimidine moiety of thiamine are synthesized de novo by the purine biosynthetic pathway or the alternative pyrimidine biosynthetic (APB) pathway. The apbA gene was the first locus defined as required for function of the APB pathway (D. M. Downs and L. Petersen, J. Bacteriol. 176:4858–4864, 1994). Recent work showed the ApbA protein catalyzes the NADPH-specific reduction of ketopantoic acid to pantoic acid. This activity had previously been associated with the pantothenate biosynthetic gene panE. Although previous reports placed panE at 87 min on the Escherichia coli chromosome, we show herein that apbA and panE are allelic and map to 10 min on both the S. typhimurium and E. coli chromosomes. Results presented here suggest that the role of ApbA in thiamine synthesis is indirect since in vivo labeling studies showed that pantoic acid, the product of the ApbA-catalyzed reaction, is not a direct precursor to thiamine via the APB pathway.

Salmonella typhimurium can synthesize thiamine in the absence of de novo purine biosynthesis by the alternative pyrimidine biosynthetic (APB) pathway (5, 7). Several mutants unable to utilize this pathway have been isolated, and apbA was the first locus characterized that was essential for thiamine synthesis in the absence of purine biosynthesis (1, 2, 6, 8, 15). apbA mutants were found to be prototrophic in a Pur⁺ background, as expected for a locus involved in an alternative pathway. Two conditions were identified under which mutations in apbA cause thiamine auxotrophy: first, if adenine is present in the medium and second, if the purF gene (whose product catalyzes the first step in de novo purine biosynthesis) is deleted. In both of these cases, the growth defect of the strain can be corrected by the addition of exogenous thiamine or pantothenate. The ability of pantothenate and thiamine to satisfy the same growth defect had been previously described but was not well understood (4, 8, 12). The apbA gene maps to 10 min and encodes a protein predicted to contain a flavin adenine dinucleotide-NAD binding site, a motif common to oxidoreductases in various organisms (6).

We recently showed that ApbA catalyzes the reduction of ketopantoic acid to pantoic acid, a reaction in the biosynthetic pathway for pantothenate (16). Two gene products had been implicated in the ketopantoic acid reductase activity in the cell. The ilvC gene product was shown to catalyze the reduction of ketopantoic acid in crude extracts under some conditions, although the primary reaction catalyzed by this enzyme is the reduction of α-acetolactate or α-aceto-α-hydroxybutyrate in the synthesis of branched-chain amino acids (16). Mutations which removed the residual ketopantoic acid reductase activity in ilvC mutants of Escherichia coli were isolated, and the lesion was mapped to 87 min. These mutations defined the panE locus and caused pantothenate auxotrophy only in strains defective for acetohydroxy acid isomeroreductase (IlvC) (10). Similar mutations were isolated in Salmonella typhimurium, and crude extracts of the panE ilvC double mutant had no ketopantoic acid reductase activity, consistent with the nutritional requirement of the strain (13, 16). From this work, it was suggested that both IlvC and PanE have ketopantoic acid reductase activity and that together they account for 100% of the cellular activity.

Our identification of a ketopantoic acid reductase activity associated with purified ApbA caused us to re-evaluate the past work on panE. We report here that panE mutations previously reported to map to 87 min are allelic with apbA and map at 10 min on both the S. typhimurium and E. coli chromosomes. Results presented here clarify the position of the final pantothenate biosynthetic gene and raise interesting questions as to the connection between pantothenate and synthesis of thiamine by the APB pathway.

The identification of a ketopantoic acid reductase activity of ApbA (9) meant that either PanE and IlvC do not provide 100% of the cellular activity or, alternatively, that apbA and panE are allelic. To distinguish between these possibilities, we obtained one of the strains containing an original panE mutation (DU6611 ilvC8 panE61) from the Salmonella Genetic Stock Center. As expected, this strain required both branched-chain amino acids (ilvC) and pantothenate (panE). The results described below demonstrated that apbA and panE are allelic.

panE61 was transductionally linked to apbA.

Phage P22 was grown on an apbA⁺ strain containing Tn10d(Tc) (17) linked to apbA [DM86; zba-8007::Tn10d(Tc)] (6). This lysate was used to transduce strain DU6611 (panE61 ilvC8) to tetracycline resistance (Tc^r), and the transductants were scored for growth in the absence of pantothenate (panE⁺). Of the 50 Tc^r transductants, 32 required only branched-chain amino acids, indicating that the zba-8007::Tn10d(Tc) insertion was linked to a region sufficient to restore growth without pantothenate to the panE61-containing strain. From these results, we concluded that the panE mutation in strain DU6611 maps at 10 min, not at 87 min, as had been previously reported.

panE61 causes an APB phenotype.

One of the clones from the above-described transduction that had retained its pantothenate requirement [DM4593; ilvC8 zba-8007::Tn10d(Tc) panE61] was saved, and a P22 lysate was grown on it. This lysate was used to transduce wild-type strain LT2 to Tc^r and the transductants were screened for the ability to grow on minimal glucose medium in the presence of adenine. Sixty-four percent of the Tc^r transductants were unable to grow on this medium unless either exogenous thiamine or pantothenate was provided; i.e., they had the phenotype previously described for apbA mutants (6). In addition, the panE61 mutation prevented thiamine synthesis in a purF mutant background, the diagnostic phenotype for a mutant defective in the APB pathway (data not shown).

After it was determined that panE mutants resulted in an APB⁻ phenotype, we addressed the question of whether null alleles of apbA caused the phenotype previously reported for panE mutants. To do this, we constructed an ilvC apbA double mutant by transducing ilvC2104::Tn10 into DM63 (apbA1::MudJ). As expected, the double mutant required not only branched-chain amino acids but a source of pantothenate.

If the apbA locus was allelic with panE, then the nutritional correction of apbA mutants by pantothenate should require only the pantoic acid moiety. When strains DM63 (apbA) and DM587 (purF apbA) were tested under thiamine-requiring conditions, at a 500 nM final concentration, either pantoic acid or pantothenate allowed these strains to grow, while no stimulatory effect was seen with the addition of β-alanine or other intermediates in the pantoate branch, including ketoisovalerate and ketopantoic acid. These nutritional data were consistent with the ketopantoic acid reductase activity of ApbA.

Taken together, these results allowed us to conclude that (i) the panE gene maps to 10 min and (ii) a panE mutation results in the phenotypes described for apbA mutants and vice versa. It was not surprising that this phenotypic connection was not previously identified, since manifestation of either phenotype required unusual growth conditions or genetic backgrounds, i.e., purines or an ilvC or purF mutation. The simplest interpretation of these results was that apbA and panE are allelic, a conclusion supported by the sections below.

panE and apbA are allelic.

To confirm that apbA and panE are allelic, the apbA gene was sequenced from an isogenic set of PanE^+/− mutant strains. Primers were designed both upstream and downstream of apbA based on the previously determined sequence (GenBank accession no. U09529). These primers were used to amplify the apbA gene from the chromosomes of strains DM4593 (zba-8007:Tn10d panE61 ilvC8) and DM4594 (zba-8007:Tn10d ilvC8) by PCR as has been described (19). With each strain, the fragments resulting from three independent amplifications were sequenced by the University of Wisconsin-Madison Biotechnology Center. This sequence analysis identified a single base change in the coding sequence of apbA in DM4593, the panE-containing strain, compared to DM4594. This change resulted in a predicted alanine-to-threonine change at amino acid 119, a residue conserved in the three identified ApbA homologs (E. coli, S. typhimurium, and Archaeoglobus fulgidus). To reflect these findings, ApbA and PanE are used interchangeably throughout the remainder of this report.

Ketopantoic acid reductase activity is reduced in apbA mutant backgrounds.

To confirm previous reports that IlvC and PanE account for all of the ketopantoic acid reductase activity in the cell, this activity was measured in crude extracts of various mutant strains. Table 1 shows the ketopantoic acid reductase levels in crude extract of wild-type and mutant strains. Appropriate strains were grown overnight in rich medium at 37°C with shaking, and cell extracts (ca. 5 mg of protein per ml) were generated by sonication. The assay for ketopantoic acid reductase was performed as previously described (9, 11). Briefly, the assay mixture contained 2.5 μmol of ketopantoic acid, 0.5 μmol of NADPH, and 200 μl of crude extract as a source of enzyme in a total volume of 2.0 ml of 100 mM potassium phosphate buffer, pH 7.5. Mixtures were incubated at 42°C for 2 min prior to initiation of the reaction by addition of ketopantoic acid. NADPH oxidation was monitored at 340 nm to determine the rate of the reaction. As can be seen in Table 1, an apbA mutant strain contained ∼40% of the ketopantoic acid reductase activity of a wild-type strain. This remaining activity was eliminated by an ilvC mutation, as indicated by the lack of reductase activity in strain DM3498 (ilvC apbA). These assays showed that, together, the ApbA (PanE) and IlvC proteins account for all of the detectable ketopantoic acid reductase activity in the cell. Taken together, the above results clearly demonstrate that apbA and panE are allelic and encode the pantothenate biosynthetic enzyme ketopantoic acid reductase.

TABLE 1.

Ketopantoic acid reductase activities in mutant strains^a

Strain (genotype)	Ketopantoic acid reductase sp actb
LT2 (wild type)	0.033 ± 0.0012
TT82 [ilvC2104::Tn10d(Tc)]	0.032 ± 0.0023
DM63 (apbA1::MudJ)	0.013 ± 0.0011
DM3498 [ilvC2104::Tn10d(Tc) apbA1::MudJ)	0.002 ± 0.0015

^aStrains were grown in minimal gluconate pantothenate medium supplemented with each of the branched-chain amino acids at 0.004% (wt/vol). Specific activities are expressed as micromoles of ketopantoic acid reduced per minute per milligram. 

^bThe data shown are averages of three independent experiments ± the standard error. 

Is PanE involved in thiamine biosynthesis?

Previous results showed that thiamine and pantothenate were equally proficient at correcting the nutritional requirement of apbA strains in either a Pur⁺ or a purF genetic background (6, 8a). With the identification of apbA as the pantothenate biosynthetic gene panE, our focus shifted from understanding how pantothenate satisfies the growth requirement of a putative thiamine biosynthetic mutant to addressing how a lesion in pantothenate biosynthesis could result in a defect in the APB pathway for thiamine synthesis.

panE mutants are unique among the pantothenate biosynthetic mutants since they grow prototrophically due to the redundant function of IlvC. Thus, we considered how panE mutants could be defective in the APB pathway while not phenotypically defective in pantothenate biosynthesis. It was possible that PanE was involved in pantothenate biosynthesis and also required for the APB pathway, performing either ketopantoic acid reduction or an alternative reaction in this pathway for thiamine synthesis. Either of these scenarios seems to predict that a purF ilvC panE triple mutant should require both pantothenate and thiamine (in addition to the clear requirement for branched-chain amino acids and a source of purines). In fact, the triple mutant strain (DM3500) had wild-type growth rates in the presence of adenine, branched-chain amino acids, and pantothenate. While this result was apparently inconsistent with the above scenarios, it did not rule them out because of previous observations by us and others that pantothenate can correct the growth of other mutants defective in thiamine synthesis by a mechanism not yet understood but unlikely due to a defect in pantothenate biosynthesis (5a, 8). In fact, the growth defect of many, but not all, mutants that have been identified as defective in the APB pathway is corrected by thiamine or pantothenate.

We considered three models for the involvement of PanE in the APB pathway: (i) a direct role for ketopantoic acid reductase; (ii) an additional function of PanE, required for the APB pathway, and (ii) an indirect role for PanE, possibly reflecting an involvement of coenzyme A (CoA) in the APB pathway. In the latter scenario, panE mutants would have reduced CoA levels due to the partial block in pantothenate biosynthesis, and this would prevent thiamine synthesis when the cell was relying on the APB pathway. It is important to emphasize that there is no thiamine requirement of panE mutants in a wild-type genetic background. Thus, such a requirement for CoA would be specific to the APB pathway and would provide a focus for future work on the nature of that pathway.

Pantoic acid is not a precursor to thiamine via the APB pathway.

In the first model suggested above, pantoic acid would be a precursor to thiamine via the APB pathway, demanding that the pantoic acid synthesized by PanE donate carbon to both pantothenate and thiamine. To initiate dissecting the role of PanE in thiamine biosynthesis, we synthesized uniformly labeled ¹⁴C-pantoic acid to address the question of whether pantoic acid is a direct precursor to thiamine via the APB pathway. First, [U-¹⁴C]ketoisovalerate was synthesized from [U-¹⁴C]valine by the method of Meister (14). [U-¹⁴C]pantoic acid was then synthesized from the [U-¹⁴C]ketoisovalerate by the method of King et al. (11). Identification and radiochemical purity were confirmed by multiple descending paper chromatography systems as described by Benson et al. (3). A 0.009-μmol sample of [U-¹⁴C]pantoic acid was added to a culture of purF apbA ilvC cells growing in 10 ml of minimal medium containing adenine and branched-chain amino acids. After overnight growth at 37°C with shaking, the thiamine was extracted from the cells and quantified by using a previously published procedure (18). No radioactivity was found to be associated with the thiamine purified from purF apbA ilvC cells grown in this manner, although in a control experiment, radiolabeled pantothenate was recovered. Based upon the pantoic acid specific activity (0.59 mCi/μmol) and the quantity of thiamine isolated, we expected to detect at least 12,000 counts associated with the thiamine-containing fraction if pantoic acid donated at least one carbon atom to thiamine. From these experiments, we concluded that pantoate is not a precursor to thiamine via the APB pathway.

Although a direct role for ketopantoic acid reductase in thiamine synthesis seemed unlikely from the above results, it was feasible that PanE has a second enzymatic role that is required for the APB pathway. Work to address this possibility is limited by our poor knowledge of potential substrates and products for such a reaction. The model we favor is that the role of PanE in the APB pathway is to ensure adequate pools of CoA. This model suggests that reduced CoA levels prevent thiamine synthesis by the APB pathway before they produce a growth requirement for pantothenate. Consistent with this model is our finding that panE mutants grown on minimal glucose medium accumulate a fivefold smaller CoA pool compared to a wild-type strain under the same conditions (8a).

Work presented here has contributed to our understanding of both pantothenate biosynthesis and the alternative pyrimidine biosynthetic pathway for thiamine synthesis. Significantly, results presented here have clarified the final gene in pantothenate biosynthesis and suggested a testable model to explain how mutants lacking a pantothenate biosynthetic gene (panE) could be defective in thiamine synthesis via the APB pathway.