Analysis of ThiC Variants in the Context of the Metabolic Network of Salmonella enterica

Abstract

In bacteria, the 4-amino-hydroxymethyl-2-methylpyrimidine (HMP) moiety of thiamine is synthesized from 5-aminoimidazole ribotide (AIR), a branch point metabolite of purine and thiamine biosynthesis. ThiC is a member of the radical S-adenosylmethionine (AdoMet) superfamily and catalyzes the complex chemical rearrangement of AIR to HMP-P. As reconstituted in vitro, the ThiC reaction requires AdoMet, AIR, and reductant. This study analyzed variants of ThiC in vivo and in vitro to probe the metabolic network surrounding AIR in Salmonella enterica. Several variants of ThiC that required metabolic perturbations to function in vivo were biochemically characterized in vitro. Results presented herein indicate that the subtleties of the metabolic network have not been captured in the current reconstitution of the ThiC reaction.

INTRODUCTION

Thiamine pyrophosphate is an essential cofactor used by enzymes in central metabolism such as pyruvate dehydrogenase and α-ketoglutarate dehydrogenase to stabilize the acyl carbanion (22). Thiamine is composed of two independently synthesized moieties, 4-methyl-5-(2-hydroxyethyl)-thiazole (THZ) and 4-amino-5-hydroxymethyl-2-methylpyrimidine (HMP). In all bacteria characterized to date, the pyrimidine of thiamine is synthesized from a branch point in purine biosynthesis. In a single step, ThiC (HMP-P synthase) converts the branch point metabolite 5-aminoimidazole ribotide (AIR) to HMP-P (10, 27). HMP-P is sequentially phosphorylated and combined with THZ-P to generate thiamine monophosphate (2, 31) before a final phosphorylation generates the active cofactor (43) (Fig. 1).

Fig 1.

Thiamine and purine biosynthesis in S. enterica. A schematic of thiamine synthesis is illustrated, with the gene products involved in purine biosynthesis shown near the relevant position in the pathway. The inset displays the fate of individual atoms in the synthesis of HMP-P from AIR on the basis of labeling studies (9, 16–18, 26). Metabolic processes that impact ThiC activity in vivo are depicted. Stains lacking PurR are derepressed for all purine genes (not depicted). Abbreviations: AIR, 5-aminoimidazole ribotide; THZ-P, 4-methyl-5-(2-hydroxyethyl)-thiazole HMP-P, 4-amino-5-hydroxymethyl-2-methylpyrimidine phosphate; TPP, thiamine pyrophosphate.

In vivo, thiamine synthesis is integrated with purine biosynthesis by the shared branch point metabolite AIR. During growth in minimal medium, the majority of flux through AIR is used for purine biosynthesis and a minor portion is diverted to thiamine biosynthesis. Nutritional studies suggest that a 1:1,000 ratio of flux to thiamine and purine synthesis would satisfy both requirements (1). Despite its role as a precursor to HMP-P, the synthesis of AIR is not regulated by thiamine; this fact raises the question of how Salmonella enterica ensures that the thiamine requirement of the cell is satisfied under conditions where purine biosynthesis is repressed. Significantly, the natural gut habitat of Salmonella has excess purines (39). This suggests that during steady-state growth in the gut, PurF (amidophosphoribosyltransferase, EC 2.4.2.14) is allosterically inhibited and the flux through the purine pathway is low (44).

Results of a number of studies have shown that when purine biosynthetic flux is reduced, thiamine synthesis is maintained, suggesting that the metabolites flow to thiamine synthesis prior to purines. Mutants with less than 1% wild-type activity of the purine biosynthetic enzyme PurI (AIR synthetase, EC 6.3.3.1) require purines but not thiamine (1). Furthermore, several pathways recruited to feed into the purine pathway, by forming either phosphoribosylamine (PRA) or AIR, satisfy the thiamine but not purine requirement of the cell (4, 15, 24, 30, 32, 33). In other words, in a wild-type cell, the partitioning of flux to thiamine biosynthesis is relatively insensitive to changes in flux to AIR. This type of insensitivity is thought to be characteristic of systems when the substrate concentration is well above the enzyme K_m value (5). Thus, the ThiC K_m for AIR is expected to be lower than the AIR concentration in the cell; however, this hypothesis cannot be addressed with current enzymatic assays of ThiC. It is possible that ThiC variants can be used to probe the AIR branch point because the sensitivity of the branch point can amplify relatively small changes in enzyme characteristics (25). It follows that this system could also be used to monitor subtle changes in the metabolic network.

In S. enterica, a number of metabolic perturbations have been shown to impact the conversion of AIR to HMP-P by ThiC. In vivo experiments found that compromised iron-sulfur ([Fe-S]) cluster metabolism reduced ThiC activity, results that contributed to the identification of the [Fe-S] requirement of this protein (13). In other studies, strains with reduced coenzyme A (CoA) levels and those that accumulated the purine intermediate 5-aminoimidazole-4-carboxamide ribotide (AICAR) have less ThiC-dependent growth in vivo (1, 20). Neither CoA nor AICAR had a demonstrable effect on the ThiC reaction in vitro, raising the question of how the respective perturbations were affecting the activity of the enzyme in vivo (N. C. Martinez-Gomez and D. M. Downs, unpublished data).

Work on ThiC in the past 5 years has established an in vitro assay and begun to define some of its biochemical parameters. ThiC is a dimeric member of the radical S-adenosylmethionine (SAM or AdoMet) superfamily of proteins that use an [4Fe-4S] cluster and SAM or AdoMet to carry out radical-mediated catalysis (10, 27, 34). In the presence of AdoMet and reductant in vitro, ThiC converts AIR to HMP-P, carbon monoxide, and formate (9). A structure of Caulobacter crescentus ThiC without the [Fe-S] cluster was solved (10), and modeling both the cluster and the active site has provided insights into the reaction mechanism of this protein. A catalytic mechanism for the conversion of AIR to HMP-P has been proposed on the basis of data that suggested that the 5′-deoxyadenosyl radical sequentially abstracts two hydrogen atoms from AIR in the reaction (9). Despite the recent progress in our understanding of HMP-P synthesis, the ThiC reaction mechanism has not been fully characterized and enzymatic turnover has not yet been achieved. We suggest that better understanding of the metabolic environment relevant to ThiC in vivo may aid in improving enzyme activity in vitro.

This study was initiated to use our ability to analyze the in vivo system and characteristics of the in vitro ThiC assay to query changes in the cellular environment that are caused by specific metabolic perturbations. Specifically, we sought to use ThiC variants that required perturbations in metabolism to function in vivo as a means to better understand how ThiC is integrated in the metabolic network and whether the in vitro assay adequately reflects the cellular properties relevant to the function of ThiC.

MATERIALS AND METHODS

Strains, media, and chemicals.

The strains used in this study were derived from S. enterica serovar Typhimurium LT2 and are listed in Table 1. The indicated thiC alleles and a linked marker [zxx-8039::Tn10d(Tc), where Tn10d(Tc) refers to the transposition-defective mini-Tn10 described by Way et al. (42)] are present in each of four strain backgrounds: wild type, purE3043, purR3090, and purE3043 purR3090.

Table 1.

Strains used in this study^a

thiC allele	Protein variant	Mutagenb	Conditional auxotrophc	Strain
				Wild type	purE	purR	purE purR
Wild type			NA	DM7778	DM13262	DM13850	DM13274
thiC1076	G486D	HA	N	DM4003	DM13264		DM13276
thiC1128	E281K	HA	Y	DM7777	DM13292	DM13851	DM13295
thiC1129	V267 M	HA	Y	DM7779	DM13293	DM13852	DM13296
thiC1130	S247F	HA	N	DM7781	DM13263		DM13275
thiC1140	R397H	HA	N	DM13314	DM13265		DM13277
thiC1141	G481S	HA	N	DM13315	DM13286		DM13290
thiC1142	D468N	HA	N	DM13316	DM13266		DM13278
thiC1143	G479R	HA	N	DM13317	DM13287		DM13291
thiC1144	H501Y	HA	N	DM13318	DM13267		DM13279
thiC1145	G472D	HA	N	DM13319	DM13268		DM13280
thiC1146	G92D	HA	Y	DM13320	DM13269	DM13853	DM13281
thiC1147	G273N	HA	Y	DM13321	DM13288	DM13854	DM13297
thiC1158	A527T	NTG	Y	DM13322	DM13294	DM13855	DM13298
thiC1159	D509G	NTG	Y	DM13323	DM13289	DM13856	DM13299
thiC1161	P498L	NTG	Y	DM13324	DM13270	DM13857	DM13282
thiC1163	G355D	NTG	N	DM13325	DM13271		DM13283
thiC1171	R544C	NTG	N	DM13326	DM13272		DM13284
thiC1172	D61N, G513E	NTG	N	DM13327	DM13273		DM13284

^athiC mutant alleles are listed with the predicted protein variant, mutagen used to generate the allele, and corresponding strain numbers in various genetic backgrounds.

^bGenerated by hydroxylamine (HA) (23) or nitrosoguanidine (NTG) (12) mutagenesis.

^cDepicts whether the relevant allele can support thiamine synthesis sufficient for growth under any tested condition. NA, not applicable; Y, yes; N, no.

Rich medium (NB) is Difco nutrient broth (8 g/liter) with NaCl (5 g/liter). Superbroth (SB) is tryptone (32 g/liter), yeast extract (20 g/liter), and NaCl (5 g/liter) with NaOH (0.05 N). Minimal medium is no-carbon essential (NCE) medium supplemented with MgSO₄ (1 mM), trace minerals (0.1×; adapted from Balch et al. [3]), and glucose (11 mM). Difco BiTek agar was added (15 g/liter) for solid medium. When present in the medium, supplements were provided at the following concentrations: thiamine, 100 nM; HMP, 100 nM; adenine, 0.4 mM; methionine, 0.3 mM; pantothenate, 50 μM. Antibiotics were added at the following concentrations in rich and minimal media: ampicillin (Ap), 150 mg/liter and 30 mg/liter, respectively; tetracycline (Tc), 20 mg/liter and 10 mg/liter, respectively; kanamycin (Km), 50 mg/liter and 12.5 mg/liter, respectively. All chemicals were purchased from Sigma-Aldrich, St. Louis, MO.

Mutant isolation and strain construction.

A lysate of P22 (HT 105/1 int201 [36, 40]) was generated on strain DM851 that had been mutagenized with nitrosoguanidine (NTG) (12). Alternatively, a P22 lysate grown on strain DM851 was mutagenized by hydroxylamine (HA) (23). DM851 carries zxx-8039::Tn10d(Tc), which is ∼80% linked to the thi locus. Each lysate was used to transduce a wild-type strain to Tc^r. Tc^r transductants were screened for those that required exogenous thiamine or HMP for growth under one of three conditions: (i) in the presence of exogenous adenine, (ii) in a purF mutant background, or (iii) at increased temperature. After isolation, the thiC alleles were transduced into the appropriate backgrounds by P22. An isogenic pair (thiC⁺ and thiC mutant alleles) from each transduction was purified to be phage-free and saved (8). In the backgrounds where the thiC alleles did not display thiamine auxotrophy, the presence of the mutation was confirmed by backcrosses. Accumulation of the ThiC variant proteins in the wild-type background was verified by Western blot analysis.

Growth analysis.

Cells from overnight cultures in NB medium were pelleted and resuspended in an equal volume of saline (0.85% NaCl), and a 5-μl aliquot was used to inoculate 195 μl of the appropriate minimal medium. Growth was quantified in a 96-well plate using a microplate reader (model EL808; Bio-Tek Instruments). Cell density was measured as the optical density (OD) at 650 nm (OD₆₅₀), and growth was reported as a specific growth rate [μ, which is equal to ln(X/X₀)/T, where X is OD₆₅₀ and X₀ is the initial OD₆₅₀ value of the period analyzed during exponential growth and T is the time (in h)].

Expression and purification of Fpr/FldA.

Plasmids to express S. enterica flavoprotein reductase (Fpr) and flavodoxin (FldA) were constructed with pET-28b(+) (Novagen) using the NheI and XhoI sites. Fpr and FldA were individually expressed in Escherichia coli strain BL21(DE3) (Novagen). In each case, a 10-ml overnight culture grown in SB with Ap was used to inoculate two 2.8-liter Fernbach flasks, each containing 1.5 liters of SB with Ap. The culture was incubated at 37°C with shaking to an OD₆₅₀ of approximately 0.9, isopropyl-β-d-thiogalactopyranoside (IPTG) was added to 1 mM, and incubation was continued overnight at 28°C with shaking. The cells were harvested by centrifugation and resuspended to 30 ml buffer A (50 mM HEPES, pH 7.5). Lysozyme (30 mg) and DNase (2 mg) were added, and the cells were lysed by two passages through a French pressure cell at 1,500 lb/in². The lysate was clarified by centrifugation at 41,400 × g for 45 min, filtered with a 0.45-μm-pore-size polyethersulfone filter (Whatman), and loaded on an Ni-nitrilotriacetic acid (NTA) Superflow (Qiagen) column (7 ml). For both Fpr and FldA preparations, the column was washed with 75 ml buffer A and then 25 ml with 97% buffer A and 3% buffer B (50 mM HEPES, pH 7.5, 500 mM imidazole, 500 mM Na₂SO₄). The proteins were then eluted with a linear gradient (3 to 100% buffer B) over 10 column volumes. The proteins were concentrated and flavin adenine dinucleotide and flavin mononucleotide were added to 1 mM final concentrations to Fpr and FldA, respectively. The proteins were then dialyzed into 2× 1 liter buffer A and then 1 liter freezing buffer (50 mM HEPES, pH 7.5, 10% glycerol). The proteins were dropped into liquid nitrogen, and the beads were stored at −80°C. The protein concentration was determined by Pierce 660 (Pierce) assay using bovine serum albumin (BSA) as a standard. The Fpr concentration was 9.2 mg/ml (0.3 mM), and the FldA concentration was 6.7 mg/ml (0.3 mM).

Expression and purification of ThiC.

Wild-type and mutant thiC were cloned into pET-28b(+) using the NheI and XhoI sites. These plasmids were coelectroporated into E. coli strain BL21AI (Novagen) with plasmid pDB1282, which expresses Azotobacter vinelandii [Fe-S] cluster-loading genes (iscSUA, hscBA, fdx, orf3, ndK) (11). Two 2.8-liter Fernbach flasks with 1.5 liters SB, Ap, and Km were each inoculated with 10 ml of an overnight culture of the appropriate strain. The cultures were incubated at 37°C with shaking to an OD₆₅₀ of 0.3. Arabinose was added to 0.2% to induce expression of the nif genes, and FeCl₃ was added to 100 μM. Incubation continued for 3 h, before expression of thiC was induced by adding IPTG to a final concentration of 1 mM, and incubation was continued with shaking overnight at 15°C. Cells were harvested by centrifugation at 16,000 × g for 15 min. Cell pellets were routinely stored at −80°C until use.

All manipulations with ThiC proteins were carried out in a Coy glove box maintained at <2 ppm oxygen, as detected by an oxygen sensor (Coy Laboratory Products, Inc.). The cell pellet from 3 liters culture (generally, 15 to 25 g cell wet weight) was resuspended to 30 ml with anoxic buffer A (50 mM HEPES, pH 7.5, 200 mM Na₂SO₄, 12% glycerol). Cells were incubated on ice with lysozyme (4 mg/ml) and DNase (2 mg) for a minimum of 10 min before sonication. Sonication was performed on ice, with 30 s of sonication and 1-min rests for 15 min using a 60 Sonic Dismembrator (Fisher Scientific) emitting 15 to 25 W. The cell lysate was clarified by centrifugation at 48,000 × g for 45 to 60 min in sealed Oakridge tubes outside the glove box. The clarified lysate was returned to the glove box, filtered, and applied to an Ni-NTA Superflow resin column (6 ml) equilibrated in buffer A. The column was washed with approximately 10 column volumes of buffer A, followed by approximately 5 column volumes of 97% buffer A, 3% buffer B (50 mM HEPES, pH 7.5, 200 mM Na₂SO₄, 12% glycerol, 500 mM imidazole). ThiC was eluted by a linear gradient of 3 to 100% buffer B over 5 column volumes. Peak elution of wild-type ThiC and the variants was at approximately 45% buffer B or ∼230 mM imidazole. Brown fractions were pooled and desalted into buffer A by a PD-10 Sephadex G-25 column (GE Healthcare). The desalted protein was concentrated in a 50,000-molecular-weight-cutoff centrifugal filter unit (Amicon) at 2,400 × g in sealed centrifuge tubes outside the glove box. To allow comparison of iron loading between the ThiC variants, the proteins were not reconstituted. An aliquot of protein was removed from the glove box. The concentration was then measured by Pierce 660 assay using BSA as a standard, and the concentration was adjusted to 500 μM. ThiC was then aliquoted into sealed serum vials (Wheaton) for storage at −80°C. Protein concentration was validated by amino acid analysis and found to be within 10% of that measured by the Pierce 660 assay with the BSA standard (Molecular Structure Facility, University of California, Davis, CA). Wild-type ThiC and the variants were analyzed by SDS-PAGE and visualized by Coomassie stain. Each preparation was >95% pure, as analyzed by TotalLab gel analysis software, and the dominant band was at the position expected for the molecular weight.

Synthesis of AIR.

AIR was synthesized using a modification of a previous protocol (28). In a typical synthesis, 36 mg AICAR (Sigma) was dissolved in 5 ml 4 N LiOH and saponified to 5-amino-4-imidazolecarboxylic acid ribotide (CAIR) by heating in a 100-ml round-bottom flask attached to a reflux condenser in a paraffin oil bath maintained at 120°C for 4 h. After it was removed from the heat, condensate was collected with 10 ml double-distilled H₂O, and the pH of the mixture was adjusted to 7.0 with 1 N acetic acid. The neutralized mixture was frozen in liquid nitrogen in two 50-ml conical tubes and lyophilized to a white powder. The powder was dissolved in 25 ml 100% ethanol and vortexed for 1 min. After 5 min, the insoluble CAIR was pelleted by centrifuging at 48,000 × g for 5 min, and the supernatant was discarded. This wash step was repeated four times before the final pellet was resuspended in 10 ml ethanol, aliquoted into eppi tubes, and dried by centrifugation under vacuum. Powdered CAIR was stored at −20°C until use.

AIR was synthesized enzymatically from CAIR by Treponema denticola AIR carboxylase (TdPurE; EC 5.4.99.18). TdPurE was expressed from pJK376 and was purified as previously described (41). The TdPurE (7.0 mg/ml, 0.4 mM) was frozen in beads in liquid nitrogen prior to storage at −80°C. To synthesize AIR, an aliquot of CAIR was dissolved in 50 μl reaction buffer (50 mM N-tris(hydroxymethyl)methyl-3-aminopropanesulfonic acid sodium-potassium salt, [(2-hydroxy-1,1-bis(hydroxymethyl)ethyl)amino]-1-propanesulfonic (TAPS) buffer, pH 8.0) and incubated with 20 nM TdPurE for 30 min at 30°C. The reaction was stopped by heat treatment (80°C, 3 min), and protein was pelleted by centrifugation at 21,100 × g for 1 min. The AIR-containing supernatant was retained and used in ThiC assays within 12 h.

ThiC assays.

Fpr, FldA, methylthioadenosine nucleosidase (MTAN; 3.2.2.16), and AIR were degassed with nitrogen for 10 min in sealed Eppendorf tubes prior to entering the glove box. SAMe (NatureMade) was used as the AdoMet source because high-pressure liquid chromatography (HPLC) analysis determined that it is 88% biologically active S,S-AdoMet (data not shown), a significant improvement over the AdoMet available from Sigma, which is approximately 43% pure and contaminated by inhibiting AdoMet-related biomolecules (19). SAMe was crushed to form a powder and then moved into the glove box, resuspended in anaerobic reaction buffer, and passed through a Spin-X filter (Corning). Aliquots of AIR and AdoMet were removed from the glove box, and their concentrations were determined using a Nanodrop spectrophotometer (Thermo Scientific) using the extinction coefficients ε₂₆₀ equal to 1,570 M⁻¹ cm⁻¹ and ε₂₅₉ equal to 15,400 M⁻¹ cm⁻¹, respectively (6, 29). Because SAMe was 88% pure AdoMet, the concentration in the assay was calculated accordingly.

In a typical reaction, ThiC (50 μM), MTAN (800 nM), Fpr (10 μM), FldA (20 μM), and NADPH (1 mM) were incubated for 10 min at room temperature before adding AdoMet (25 μM to 250 μM) and AIR (25 μM to 250 μM) to a final volume of 50 μl. Fpr, FldA, and NADPH were provided to reduce the ThiC [Fe-S] cluster. MTAN was provided to remove 5′-deoxyadenosine from the reaction mixture, which has been shown to inhibit other radical SAM enzymes (7, 19). The reaction mixtures were incubated at 37°C in the anaerobic chamber for 1 h, the reaction was stopped by heat treatment (85°C for 3 min), and the reaction mixtures were frozen at −20°C before analysis, if necessary.

Quantification of HMP by HPLC.

The heat-denatured protein was pelleted from the reaction mixture by centrifugation at 21,100 × g for 1 min, and the supernatant was moved to a new eppi tube. In our hands, the ThiC reaction often produces a mixture of phosphorylated and nonphosphorylated HMP; in order to convert all product to HMP, rAPid alkaline phosphatase (Roche) was added to the supernatant (1 μl alkaline phosphatase/50 μl assay mixture) and the mixture was incubated for 15 min at 37°C. The mixture was then heat treated (3 min at 80°C) to inactivate the alkaline phosphatase. The denatured enzyme was pelleted by centrifugation at 21,100 × g for 1 min, and the supernatant was transferred to an HPLC autosampler vial (Macherey-Nagel). Five-microliter injections of the reaction mixtures were separated by reversed-phase HPLC with an LC-20AT delivery system (Shimadzu) equipped with a 250- by 4.6-mm Luna C₁₈2, 5-μm chromatography resolution column (Phenomenex). The column was equilibrated with 90% mobile phase A (20 mM potassium phosphate buffer, pH 7.0) and 10% mobile phase B (methanol). The separation used a flow rate of 1 ml min⁻¹ with 90% mobile phase A and 10% mobile phase B for 10 min, followed by a linear gradient to 50% mobile phase B over 20 min. Components eluted from the column were monitored with a photodiode array detector (wavelengths, 190 to 350 nm), with data extracted at 235 nm. The HMP peak was identified by (i) the biological activity of the collected fractions, (ii) a UV spectrum matching that of authentic HMP, (iii) coelution with authentic HMP, and (iv) mass spectrometry. For mass spectrometry, the University of Wisconsin Biotechnology Center Mass Spectrometry/Proteomics Facility performed standard mass spectrometry in positive mode. HMP was quantified by comparison to a standard curve of authentic HMP (provided by JoAnn Stubbe, Massachusetts Institute of Technology), which was determined using the extinction coefficient ε₂₆₈ equal to 4,266 M⁻¹ cm⁻¹ (M. Parker and J. Stubbe, personal communication).

RESULTS AND DISCUSSION

Eighteen mutant alleles of thiC were isolated, sequenced, and reconstructed in four different genetic backgrounds (Table 1). Eleven of the 18 alleles of thiC caused a requirement for HMP or thiamine under all conditions tested. Three of these variants had substitutions in residues in the predicted active site (10): ThiC^G355D, ThiC^R397H, and ThiC^H501Y (Fig. 2). The isolation of these alleles in an unbiased screen for loss of function supported the assignment of these residues to the active site by the modeled ThiC structure (10). The other eight null variants had substitutions that were near the active site in the tertiary structure. Four of these variants, ThiC^G355D, ThiC^G472D, ThiC^G481S, and ThiC^G486D, had substitutions of glycine residues in α helices, and the disruption of helical structure could explain their enzymatic defect.

Fig 2.

Primary sequence of ThiC with position of variants. Residues in the predicted active site, based on C. crescentus ThiC (10), were identified in S. enterica ThiC by using Clustal (version 2.1) program sequence alignment and are outlined with a dashed line. Residues that are substituted in variants identified in this study are highlighted on the basis of the phenotype that their presence causes in vivo: null phenotype (black) and conditional thiamine auxotrophs (gray). Asterisks above the residues indicate that their null phenotype was characterized by site-directed mutagenesis in a previous study (13).

The remaining seven alleles of thiC supported thiamine synthesis in vivo under certain conditions and were characterized further to understand the impact of the cellular environment on their activity. Specifically, the variants were characterized for the ability to support thiamine synthesis in genetic backgrounds and with growth conditions that had predictable consequences for the metabolic network.

Purine pathway flux impacts the activity of some ThiC variants.

In an otherwise wild-type genetic background, three thiC alleles allowed growth at or near wild-type levels in minimal medium, but the strains were unable to grow when adenine was present in the medium, unless thiamine was added (Table 2). On the basis of this phenotype, these variants, ThiC^E281K, ThiC^V267M, and ThiC^G92D, were designated “adenine sensitive.” Because purine biosynthesis is regulated at both transcriptional and posttranslational levels (37, 44), addition of exogenous adenine is expected to decrease flux through the pathway and therefore lower the AIR concentration in the cell (30). On the basis of this assumption, these data suggested that the adenine-sensitive ThiC variants required more AIR than the wild-type protein to function in vivo.

Table 2.

Growth reflects that thiamine synthesis is allowed by thiC alleles

Alleleb	Protein	Growth ratea (h−1)
		Ade	None	purE Ade	purE purR Ade	purR Ade
	Wild type	0.58 ± 0.01	0.68 ± 0.06	0.53 ± 0.01	0.50 ± 0.04	0.64 ± 0.04
thiC1128	ThiCE281K	0.07 ± 0.04	0.62 ± 0.08	0.21 ± 0.01	0.43 ± 0.02	0.56 ± 0.03
thiC1129	ThiCV267 M	0.07 ± 0.03	0.54 ± 0.06	0.20 ± 0.07	0.49 ± 0.01	0.16 ± 0.03
thiC1147	ThiCG273N	0.02 ± 0.03	0.01 ± 0.01	0.02 ± 0.00	0.08 ± 0.01	0.02 ± 0.01
thiC1158	ThiCA527T	0.01 ± 0.01	0.02 ± 0.01	0.13 ± 0.02	0.39 ± 0.03	0.02 ± 0.02
thiC1159	ThiCD509G	0.05 ± 0.04	0.01 ± 0.01	0.09 ± 0.03	0.24 ± 0.01	0.01 ± 0.01
thiC1161	ThiCP498L	0.01 ± 0.00	0.03 ± 0.01	0.17 ± 0.02	0.30 ± 0.07	0.01 ± 0.01
	Wild typec	0.26 ± 0.01	0.31 ± 0.02	0.35 ± 0.04	0.33 ± 0.02	0.45 ± 0.02
thiC1146	ThiCG92Dc	0.00 ± 0.01	0.24 ± 0.02	0.26 ± 0.02	0.24 ± 0.01	0.13 ± 0.01

^aGrowth rate is reported as μ, which is equal to ln(X/X₀)/T, where X is OD₆₅₀, X₀ is the initial OD₆₅₀ value of the period analyzed during exponential growth, and T is the time (in h). Under all conditions, the addition of thiamine resulted in a growth rate of 0.42 to 0.87 h⁻¹ (data not shown). Data shown are the average and standard deviation of three independent cultures. Data from each condition were collected on the same day. Strains were grown at 37°C in minimal medium with no supplement or with adenine (Ade; 0.4 mM), as indicated.

^bEach allele was present in a strain with the indicated additional genotype.

^cStrains were grown at 30°C.

While exogenous adenine is expected to reduce AIR levels in vivo, it potentially has additional effects on the metabolic network. To test the hypothesis that the adenine-sensitive variants were in fact sensitive to AIR levels, growth was tested in genetic backgrounds that retained high levels of AIR, despite the presence of adenine. PurK [5-(carboxyamino) imidazole ribonucleotide synthase, EC 6.3.4.18] and PurE [5-(carboxyamino) imidazole ribonucleotide mutase, EC 5.4.99.18] catalyze the enzymatic steps in the purine biosynthetic pathway that consume AIR. The purE3043 allele is expected to result in the accumulation of AIR in the presence of exogenous adenine (14, 30). The purE3043 allele restored thiamine synthesis in strains with ThiC^E281K and ThiC^V267M variants, as judged by the growth rate of ∼0.20 h⁻¹ compared to one of ∼0.07 h⁻¹ in adenine medium (Table 2). ThiC^G92D is a temperature-sensitive variant, and at 30°C in the presence of purE3043, the growth rate in adenine was fully restored to that in minimal medium (∼0.24 h⁻¹). Interrupting the gene encoding the purine repressor (PurR) further increased the growth rate of these strains in adenine medium, consistent with a need of the variants for higher AIR accumulation (30, 38). Finally, the ThiC^D509G variant supported growth only in the purE purR genetic background, suggesting that it had the highest requirement for AIR among the variants. We interpreted these results to mean that the adenine-sensitive variants required increased levels of AIR to function in vivo. Taken together, these data suggested that the ThiC^E281K, ThiC^V267M, ThiC^D509G, and ThiC^G92D variants would require a higher concentration of AIR than the wild-type enzyme to function in vitro.

Exogenous methionine or increased expression of AdoMet synthetase restores function to some ThiC variants.

ThiC requires AdoMet as a cosubstrate for radical catalysis (10, 27), so changes to the metabolic network that impact AdoMet were predicted to alter growth of some thiC mutant strains. Because addition of exogenous methionine to the medium has been shown to elevate AdoMet levels in Saccharomyces cerevisiae and Candida utilis yeasts (21), the thiC mutant strains were screened for growth on minimal medium with exogenous methionine. The growth of two strains, encoding variants ThiC^G273N and ThiC^P498L, was greatly improved by the addition of methionine. Significantly, growth of these same strains was improved by the presence of the plasmid pMETK2, which encodes MetK (AdoMet synthetase, EC 2.1.5.6) (Fig. 3). The presence of pMETK2 increased AdoMet in the cell by 5-fold (S. Wang and P. A. Frey, personal communication). Plasmid pMETK2 did not improve the growth of a strain containing the null ThiC^H501Y variant, supporting the conclusion that AdoMet, and not thiamine, was increased by the presence of this plasmid (data not shown). These results suggested that addition of exogenous methionine increases AdoMet levels in S. enterica, as it was previously shown to do in yeasts (21). Further, these in vivo phenotypes suggested that the ThiC^G273N and ThiC^P498L variants would have a higher requirement for AdoMet than wild-type ThiC when assayed in vitro.

Fig 3.

Either methionine or induction of S-adenosylmethionine biosynthesis allowed growth of strains containing ThiC^G273N or ThiC^P498L. (A) Growth of the strain containing ThiC^G273N (thiC1147, DM13321); (B) growth of the strain containing ThiC^P498L (thiC1161, DM13224). Strains were grown in minimal medium and arabinose (0.2%) with no additions (squares), addition of methionine (triangles), or addition of thiamine (circles). Strains containing an empty vector are depicted with open symbols, while those containing pMETK2 are depicted with filled symbols. Data shown are the average and standard deviation of three independent cultures.

Exogenous pantothenate stimulates the growth rate of ThiC^A527T.

Conversion of AIR to HMP-P in vivo is sensitive to the CoA levels in the cell (1, 20). CoA has no demonstrable effect on ThiC activity in vitro; therefore, we predicted that the CoA effect was mediated through another reaction component, i.e., AIR, AdoMet, or the [Fe-S] cluster. Each thiC mutant strain was screened for growth stimulation by pantothenate because exogenous pantothenate elevates endogenous CoA levels (20). Growth of a single strain, DM13294, which carried both purE3043 and thiC1158, was stimulated by exogenous pantothenate (Fig. 4). This result suggested that the ThiC^A527T variant responded to the reaction component affected in vivo by CoA levels. The finding that pantothenate stimulated growth only when the purE3043 mutation was present implicated the AIR concentration in the effects of CoA on ThiC activity in vivo.

Fig 4.

The growth rate of a strain containing ThiC^A527T in a purE mutant background is increased by exogenous pantothenate. Strain DM13294 (thiC1158 purE3043) was grown in NCE glucose medium supplemented with adenine (0.4 mM) and no additions (squares), addition of pantothenate (diamonds), and addition of thiamine (circles). Data shown are the average and standard deviation of three independent cultures.

ThiC variants are generally defective in activity in vitro.

The different phenotypic behaviors of the thiC mutant strains suggested that the biochemical properties of the ThiC variants would be distinguishable from each other and from those of the wild-type protein. Due to limitations of HMP detection at low enzyme concentrations and the inability to achieve enzyme turnover, which is common for SAM radical proteins (35), Michaelis-Menten kinetic models could not be used to analyze the assay results. Wild-type ThiC and four variants of ThiC (V267M, G273N, A527T, D509G) were purified, and the activity was determined in assays using 50 μM protein, 250 μM AIR, and 250 μM AdoMet over 1 h. The data in Table 3 show that with these reaction conditions, each of the variants was significantly compromised in activity. The iron content of the variants was similar to that of wild type (data not shown); however, a conformational difference between the [Fe-S] cluster of the wild-type and mutant proteins could explain the activity difference in vivo. There was no condition where the growth rates (Table 2) of the strains expressing the respective thiC alleles were proportional to the ThiC variant activity in vitro.

Table 3.

ThiC variants have reduced activity^a

ThiC variant	Amt of HMP produced (nmol)
Wild type	2.59 ± 0.31
ThiCV267 M	0.73 ± 0.02
ThiCG273N	0.38 ± 0.05
ThiCD509G	0.43 ± 0.06
ThiCA527T	1.44 ± 0.30

^aEach reaction mixture contained the relevant ThiC protein at 50 μM, and both AdoMet and AIR were present at 250 μM. Reaction mixtures were incubated for 1 h at 37°C and heat quenched at 85°C for 3 min; detailed reaction conditions are described in Materials and Methods. Data for wild-type protein are the average and standard deviation of eight replicates. Data for each variant protein are the average and standard deviation of four replicates.

Similarly, independent titrations of AIR and AdoMet concentrations with purified variant protein failed to provide insights into how the activity correlated with in vivo results. The activity of variants ThiC^G273N and ThiC^D509G was not significantly different over the range of AIR or AdoMet concentrations used (data not shown). This result was unexpected for ThiC^G273N, which allowed wild-type levels of in vivo growth when internal AdoMet levels were elevated by either methionine or pMETK (Fig. 3). The constant weak activity of ThiC^D509G was not unexpected, given the complex nutritional environment needed to generate wild-type growth with this variant. In contrast, the activities of both ThiC^V267M and ThiC^A527T were affected by the concentration of AIR and AdoMet (Fig. 5). In the case of ThiC^V267M, the activity did not require an elevated level of AIR, as predicted from the in vivo growth. The variants' lower specific activity could be caused by thermal instability in the 1-h incubation; however, instability would not explain the lack of the expected response to the substrate concentration.

Fig 5.

ThiC variants are compromised for HMP formation regardless of substrate concentration. Total HMP formed in 1 h by 50 μM ThiC proteins with various concentrations of substrate is shown. Data are shown for ThiC (triangles), ThiC^V267M (squares), and ThiC^A527T (circles). (A) Reaction mixtures contained 250 μM SAM and the indicated concentration of AIR; (B) reaction mixtures contained 250 μM AIR and the indicated concentration of SAM. Data points are the average and standard deviation of two replicates.

ThiC^A527T had wild-type activity with low concentrations of either AIR or AdoMet, yet the best in vivo growth supported by this variant required conditions thought to increase AIR. Exogenous pantothenate had a positive impact on the growth of strains carrying this variant, which suggested that the biochemical characteristics of the ThiC^A527T variant would reveal the role of CoA. However, the ThiC^A527T variant's response to substrate concentrations and its iron content were similar to those of wild-type ThiC; thus, no particular reaction component could be implicated in the CoA effect. Further studies into the role of CoA in the ThiC reaction will screen specifically for variants that require higher CoA levels and may provide greater insight into this metabolic connection.

Conclusions.

The different in vivo behaviors of thiC mutant strains indicated that the variant proteins had different capacities and/or requirements for HMP synthesis. On the basis of knowledge of the relevant pathways and metabolites, simple predictions about characteristics of the variant proteins were made. The finding that the variants failed to act as predicted in vitro suggested that (i) simplistic explanations or the metabolic changes caused by mutations and/or nutrients were incorrect or (ii) the in vitro reconstitution of ThiC activity was not adequately representing in vivo conditions. These two scenarios are not mutually exclusive, and we expect that the lack of correlation between in vivo growth behavior and in vitro activity reflected a combination of the two scenarios. For example, the metabolic network may be providing unanticipated metabolites and/or proteins that are crucial for optimal ThiC activity in vivo but have not yet been identified and are therefore not present in the in vitro assay. Regardless of the specific explanation for the lack of correlation between in vivo and in vitro results, this study highlights the need for caution when extrapolating between in vitro and in vivo conditions. When conditions that allow ThiC to turn over in vitro have been identified, refinement of the assay can strive to detect the expected correlation between in vivo behavior and in vitro properties.