Induction of the Sugar-Phosphate Stress Response Allows Saccharomyces cerevisiae 2-Methyl-4-Amino-5-Hydroxymethylpyrimidine Phosphate Synthase To Function in Salmonella enterica

ABSTRACT

Thiamine pyrophosphate is a required cofactor for all forms of life. The pyrimidine moiety of thiamine, 2-methyl-4-amino-5-hydroxymethylpyrimidine phosphate (HMP-P), is synthesized by different mechanisms in bacteria and plants compared to fungi. In this study, Salmonella enterica was used as a host to probe requirements for activity of the yeast HMP-P synthase, Thi5p. Thi5p synthesizes HMP-P from histidine and pyridoxal-5-phosphate and was reported to use a backbone histidine as the substrate, which would mean that it was a single-turnover enzyme. Heterologous expression of Thi5p did not complement an S. enterica HMP-P auxotroph during growth with glucose as the sole carbon source. Genetic analyses described here showed that Thi5p was activated in S. enterica by alleles of sgrR that induced the sugar-phosphate stress response. Deletion of ptsG (encodes enzyme IICB [EIICB] of the phosphotransferase system [PTS]) also allowed function of Thi5p and required sgrR but not sgrS. This result suggested that the role of sgrS in activation of Thi5p was to decrease PtsG activity. In total, the data herein supported the hypothesis that one mechanism to activate Thi5p in S. enterica grown on minimal medium containing glucose (minimal glucose medium) required decreased PtsG activity and an unidentified gene regulated by SgrR.

IMPORTANCE This work describes a metabolic link between the sugar-phosphate stress response and the yeast thiamine biosynthetic enzyme Thi5p when heterologously expressed in Salmonella enterica during growth on minimal glucose medium. Suppressor analysis (i) identified a mutant class of the regulator SgrR that activate sugar-phosphate stress response constitutively and (ii) determined that Thi5p is conditionally active in S. enterica. These results emphasized the power of genetic systems in model organisms to uncover enzyme function and underlying metabolic network structure.

INTRODUCTION

Thiamine pyrophosphate (TPP) is a cofactor for many central metabolic enzymes and is required at low levels by all organisms. Humans require dietary intake of thiamine, which is biosynthesized by a variety of plants, bacteria, and fungi. TPP is composed of two independently synthesized moieties, 5-(2-hydroxyethyl)-4-methylthiazole phosphate (THZ-P) and 2-methyl-4-amino-5-hydroxymethylpyrimidine phosphate (HMP-P). In bacteria and plants, the first steps of the HMP-P biosynthesis pathway are shared with purine biosynthesis (Fig. 1) (1–3). In these organisms, the radical S-adenosylmethionine (SAM) enzyme ThiC catalyzes an intramolecular rearrangement of the purine intermediate 5-aminoimidazole ribotide (AIR) to the pyrimidine HMP-P (4, 5).

FIG 1.

Schematic of Thi5p-dependent thiamine biosynthesis in S. enterica. The relevant enzymes are shown. The origins of atoms in HMP-P are indicated, with thick black lines and type indicating PLP and gray lines and type indicating histidine. Abbreviations: AIR, 5-aminoimidazole ribotide; His, histidine; PLP, pyridoxal phosphate; HMP-P, 2-methyl-4-amino-5-hydroxymethylpyrimidine phosphate; THZ-P, 4-methyl-5-hydroxyethylthiazole phosphate.

In contrast, fungi do not contain a ThiC homolog; under aerobic conditions, HMP-P is synthesized by the Thi5p enzyme family. In vivo labeling in yeast implicated histidine and pyridoxine as precursors to HMP-P biosynthesis under aerobic conditions (Fig. 1) (6–8). There are four members of the THI5 gene family in Saccharomyces cerevisiae (THI5 [YFL058w], THI11 [YJR156c], THI12 [YNL332w], and THI13 [YDL244w]), and other species have between zero and five copies of this gene (9). Genetic analysis in S. cerevisiae demonstrated that the four enzymes are functionally redundant, as only the quadruple mutant displayed thiamine auxotrophy under aerobic conditions (9). A less efficient HMP-P biosynthetic pathway operates under anaerobic conditions and does not rely on pyridoxine or Thi5p family enzymes (9, 10).

The in vivo labeling pattern for aerobic HMP-P synthesis in S. cerevisiae suggested that unique chemistry was involved. The Candida albicans Thi5p activity has been reconstituted (11). These authors reported that the Thi5p protein was oxygen sensitive, but it required oxygen, iron, and pyridoxal phosphate (PLP) for HMP-P synthesis. The Lai et al. study (11) further reported that the addition of histidine was not required for activity. Rather, their data suggested that residue His66 of the Thi5p protein served as the histidine substrate, implying that Thi5p was a single-turnover enzyme (11). The importance of His66 for Thi5p activity was corroborated by Coquille et al. (12), who presented the structure of Thi5p and reported that His66 was required for Thi5 activity in vivo in S. cerevisiae. The report that Thi5p is a single-turnover enzyme raised questions about potential differences in the physiology between fungi and bacteria/plants that might select for the maintenance of a single-turnover enzyme when a catalytic mechanism is available (i.e., ThiC). Maintaining such an enzyme mechanism would appear to have significant energy cost implications for a dynamic metabolic network.

Here we describe the characterization of S. cerevisiae Thi5p activity in vivo in the model organism Salmonella enterica. This study was initiated with two goals: (i) to begin to describe metabolic network differences between two well-studied organisms of different domains and (ii) to improve future studies of Thi5p enzymatic activity. Metabolic networks are thought to be composed of conserved metabolic modules (13), and general network organizing principles are broadly conserved (14). However, it remains unclear the degree to which metabolic network structure is impacted by component substitution. Thiamine biosynthesis in S. enterica is an established system for metabolic dissection (reviewed in reference 3), suggesting that analysis of Thi5p activity in S. enterica could uncover metabolic differences between S. enterica and S. cerevisiae in addition to informing mechanistic studies of the Thi5p enzyme.

Although this study was initiated before in vitro Thi5p activity was reported, the proposed Thi5p mechanism suggested that further study in vivo could uncover enzymatic properties important for Thi5p activity. Historically, physiological studies have led to successful reconstitution or improvement of in vitro activity assays for enzymes that have been difficult to study. For example, the ThiC bacterial HMP-P synthase activity in vivo was linked to methionine and iron sulfur cluster metabolism in Salmonella enterica (15, 16), leading to its subsequent reconstitution of activity and identification as a radical SAM enzyme (4, 5). Similarly, although many reports described biotin synthase BioB as a single-turnover enzyme in vitro, in vivo work in Escherichia coli showed that BioB was capable of multiple turnovers and inhibited by its product 5′-deoxyadenosine (17, 18). These findings informed later studies demonstrating that BioB was capable of multiple turnovers in vitro when product inhibition was alleviated (19).

The analysis herein has uncovered unexpected integration between the sugar-phosphate stress response regulator and Thi5p activity in S. enterica. The sugar-phosphate stress response is coordinated by the transcription factor SgrR (formerly YabN) in enteric bacteria (20, 21). SgrR was defined for its activation of the small-RNA sgrS gene (20, 22), which destabilizes transcripts of genes encoding phosphotransferase system (PTS) transporters (ptsG [20] and manXYZ [23]) and stabilizes transcripts of a sugar phosphatase (yigL [24]). The small-RNA sgrS gene also encodes a small peptide, SgrT, which inhibits enzyme IICBGlc (EIICBGlc) activity posttranslationally (20, 22). Further, SgrR regulates additional genes in E. coli including those encoding a sugar efflux pump (setA [25]), a glutamic-pyruvic transaminase (alaC, formerly yfdZ [21]), and itself (21). The results described herein suggest that one mechanism to allow Thi5p function in S. enterica is by remodeling the metabolic network associated with the sugar-phosphate stress response regulator.

MATERIALS AND METHODS

Strains, media, and chemicals.

The minimal medium was no-carbon E medium (NCE) (26) supplemented with MgSO₄ (1 mM), trace minerals (0.1×) (27), and carbon sources based on 66 mM available carbon units with the following exceptions: pyruvate (50 mM; prepared from powder as needed), acetate (as noted), malate (40 mM), succinate (20 mM), and fumarate (50 mM). Thiamine was supplemented at 100 nM. Rich media were Difco nutrient broth (NB) (8 g/liter) with NaCl (5 g/liter), lysogeny broth (LB), or superbroth (SB) (tryptone [32 g/liter[, yeast extract [20 g/liter], NaCl [5 g/liter] with NaOH [0.05 N]). Solid media contained 1.5% agar. Antibiotics were added at the following concentrations in rich media, unless otherwise noted: chloramphenicol (Cm), 20 mg/liter; ampicillin (Ap), 150 mg/liter; tetracycline (Tc), 20 mg/liter; kanamycin (Kn), 50 mg/liter. All chemicals were purchased from Sigma-Aldrich, St. Louis, MO. The strains used in this study were derivatives of Salmonella enterica strain LT2 and were generated for this study or were part of the laboratory collection, and their relevant genotypes are listed in Table 1.

TABLE 1.

S. enterica strains used in this study

Strain	Relevant genotypea
DM13623	ΔthiC1225 ΔaraCBAD/pDM1336
DM13631	ΔthiC1225 ΔaraCBAD zxx10167::Tn10d(Tc) sgrR1/pDM1336
DM13632	ΔthiC1225 ΔaraCBAD zxx10167::Tn10d(Tc)/pDM1336
DM14419	ΔthiC1225 ΔaraCBAD/pDM1381
DM14422	ΔthiC1225 ΔaraCBAD ΔsgrR3::Kn/pDM1381
DM14426	ΔthiC1225 ΔaraCBAD zxx10167::Tn10d(Tc) sgrR1/pDM1381
DM14427	ΔthiC1225 ΔaraCBAD zxx10167::Tn10d(Tc)/pDM1381
DM14430	ΔthiC1225 ΔaraCBAD zxx10167::Tn10d sgrR2/pDM1381
DM14431	ΔthiC1225 ΔaraCBAD zxx10167::Tn10d(Tc)/pDM1381
DM14443	ΔthiC1225 ΔaraCBAD pgi::Tn5(Kn)/pDM1381
DM14517	ΔthiC1225 ΔaraCBAD ptsG4152::Tn10d(Tc)/pDM1381
DM14531	ΔthiC1225 ΔaraCBAD zxx10167::Tn10d(Tc)/pDM1402
DM14532	ΔthiC1225 ΔaraCBAD zxx10167::Tn10d(Tc) sgrR1/pDM1402
DM14533	ΔthiC1225 ΔaraCBAD zxx10167::Tn10d(Tc) sgrR2/pDM1402
DM14534	ΔthiC1225 ΔaraCBAD ΔsgrR3::Kn/pDM1402
DM14540	ΔthiC1225 ΔaraCBAD ΔpfkA::Kn/pDM1381
DM14668	ΔthiC1225 ΔaraCBAD ptsG4152::Tn10d(Tc) ΔsgrR3::Kn/pDM1381
DM14896	ΔthiC1225 ΔaraCBAD ptsG4152::Tn10d(Tc) ΔsgrS6::Kn/pDM1381
DM14932	ΔthiC1225 ΔaraCBAD ptsG4152::Tn10d(Tc) ΔalaC251::Kn/pDM1381
DM14980	ΔthiC1225 ΔaraCBAD pgi::Tn5(Kn)/pDM1402
DM14985	ΔthiC1225 ΔaraCBAD ΔpfkA::Kn/pDM1402

^aTn10d(Tc) refers to the transposition-defective mini-Tn10(Tn10Δ16Δ17 Tet^r) (28).

Genetic methods.

The high-frequency generalized transducing mutant of bacteriophage P22 (HT105/1, int-201) (29) was used for all transductional crosses. Transduction and subsequent purification was performed as previously described (30). Mutations in sgrR, alaC, and sgrS were generated by using lambda Red recombination using primers listed in Table 2 (31).

TABLE 2.

Plasmids and primers used in this study

Plasmid or primera	Description or sequence
Plasmids
pDM1336	pSU18-THI5 (encodes Thi5pM37I, A138V, G152D)
pDM1381	pSU18-THI5
pDM1361	pBAD18S-THI5
pDM1373	pET14b-THI5
pDM1402	pFZY1-sgrSp
Primers
Sc THI5 for SacI pBAD18	AAGAGCTCTGATGTCTACAGACAAGATC
Sc THI5 Rev HindIII	TTTAAGCTTTTAAGCTGGAAGAGCCAATC
sgrSp_KpnI_F	ATGGTACC CATAAAAGGGGAACTC
sgrSp_BamHI_R	ATAGGATCC CGAAAGATATTATTGGC
S. cerevisiae Thi5 for 5′	CATATGATGTCTACAGACAAGATCAC
S. cerevisiae Thi5 rev 3′	CTCGAGTTAAGCTGGAAGAGCCAATC
pfkA_wanner_for	CATTCCAAAGTTCAGAGGTAGTCATGATTAAGAAAATCGGGTGTAGGCTGGAGCTGCTTC
pfkA_wanner_rev	CCATCAGGCGCGCAAAAACAATCAGTACAGTTTTTTCGCGCATATGAATATCCTCCTTAG
yabN (sgrR) wanner for	TTTTCATCGGAGTTCCCCTTTTATGCCCTCAGGTCGCCTGGTGTAGGCTGGAGCTGCTTC
yabN (sgrR) wanner rev	CAGCAATCAAGAGCTGGCGTTAAGGATCTGGCGGCGCAAACATATGAATATCCTCCTTAG
yabN(sgrR)_NcoI_for	ATATCCATGGCAATGCCCTAGGTCGC
yabN(sgrR)_XbaI_rev	GCGATCTAGATTAAGGATCTGGCGGCGC
sgrS wanner for	GCAATTTTATTATCCCTATATTAGGCCAATAATATCTTTCGTGTAGGCTGGAGCTGCTTC
sgrS wanner rev	ATTTCGGCTGTTTCTGGATGACGATGATGGGACGGCGTTTCATATGAATATCCTCCTTAG
alaC(yfdZ) wanner for	ACGTTAATCTGAGGATATTATGGCTGACTTCCGCCCTGAAGTGTAGGCTGGAGCTGCTTC
alaC(yfdZ) wanner rev	GGGGCTCCTGTTTGCTTTTCTACTCCGTGCTCGCTTCAACCATATGAATATCCTCCTTAG

^aIn the primer names, Sc stands for S. cerevisiae, F and for stand for forward, and R and rev stand for reverse.

Molecular techniques.

Plasmids were constructed using standard molecular techniques. DNA was amplified using Herculase (Agilent, Santa Clara, CA) or Q5 (New England BioLabs, Ipswich, MA) DNA polymerase. Primers were purchased from Integrated DNA Technologies, Coralville, IA. Plasmids were isolated using the Wizard Plus SV miniprep kit (Promega, Madison, WI), and PCR products were purified using the PCR purification kit (Qiagen, Venlo, Limburg, The Netherlands). Restriction endonucleases were purchased from New England BioLabs, Ipswich, MA, and ligase was purchased from Thermo Scientific, Waltham, MA.

The plasmids and primers are listed in Table 2. The THI5 yeast glutathione S-transferase (GST)-tagged plasmid from Thermo Open Biosystems (Huntsville, AL) was isolated using the modified Qiagen miniprep protocol for yeast and used as the template for amplification of THI5. Later analysis revealed the cloned YDR155C plasmid included a mixed population of wild-type and variant THI5 (cloned in pDM1336) with the following substitutions: M37I, A138V, and G152D. We did not detect any functional differences between the variant THI5 or wild-type THI5 gene. pSU18 is a pACYC184 derivative compatible with pBR322 that uses the lac promoter and encodes Cm^r (32). To construct pDM1336 and pDM1381, THI5 was amplified using the S. cerevisiae Thi5 for 5′ primer (for stands for forward) and S. cerevisiae Thi5 rev 3′ primer (rev stands for reverse); the resulting PCR products were purified and blunt end ligated into pSU18 digested with SmaI.

pBAD vectors are pBR322 derivatives that use the ara promoter and encode Amp^r; pBAD18S includes the N-terminal sequence of AraB Met-Ala-Ile-Ala-Gly prior to the multiple cloning site (33). THI5 was cloned into SacI/HindIII sites in pBAD18S to construct pDM1361. pFZY1 is a mini-F derivative (averages one or two copies per cell) with a multiple cloning site upstream of a promoterless galK9-lacZYA reporter segment (34). The promoter region of sgrS was cloned into pFZY1 using the KpnI/BamHI sites to construct pDM1402 (sgrSp-lacZ [sgrSp is the sgrS promoter]).

Western blotting.

Cells containing pDM1381 or pSU18 were grown overnight in 10 ml of minimal medium containing glucose or minimal medium containing ribose. Cultures were centrifuged (10 min at 17,000 × g) and resuspended in 14.5 mM NaCl before sonication (1-s pulses for 1 min) (Fisher Scientific). Lysate was clarified by centrifugation (10 min at 17,000 × g), and total protein was quantified by the bicinchoninic acid (BCA) assay (Pierce) using bovine serum albumin (BSA) as a standard. Protein samples were denatured in a loading buffer (60 mM Tris Cl [pH 6.8], 2% SDS, 0.1 M dithiothreitol [DTT], 10% glycerol) at 85°C for 15 min before sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using a 12% gel. Proteins were then transferred to a polyvinylidene fluoride (PVDF) membrane (Bio-Rad Laboratories) and detected using a rabbit polyclonal antibody for Thi5p from Harlan Bioproducts (1/1,000 dilution) as a primary antibody and a horseradish peroxidase-conjugated polyclonal goat anti-rabbit IgG (1/50,000 dilution) as the secondary antibody (Thermo Scientific Pierce). The blot was then visualized using an ECL Plus Western blotting substrate (Pierce) on a Typhoon Trio+ imager (GE Healthcare).

For antibody production, THI5 was cloned into the pET14b vector (Novagen) using the NdeI/XhoI sites and transformed into E. coli BL21AI. Cells were grown, expression was induced, and protein was purified by laboratory protocols used routinely for purifying His-tagged proteins using nickel-nitrilotriacetic acid (Ni-NTA) Superflow resin (Qiagen). Protein aliquots were stored at −80°C prior to use. Purified His₆Thi5p was used as a control in the Western blot procedure.

Suppressor isolation.

Ten independent cultures of S. enterica DM13623 in nutrient broth with chloramphenicol (NB Cm) were incubated overnight with shaking at 37°C. A 100-μl sample (∼1 × 10⁸ cells) was spread on a minimal glucose medium plate, and 5 μl diethyl sulfate (DES) was spotted in the center of the plate as indicated. Suppressor frequency was analyzed after incubation at 37°C for 2 days. Six colonies per plate were streaked for individual colonies on selective medium (minimal glucose medium) and then on nonselective medium (NB Cm), before patching onto NB Cm and printing to minimal glucose medium for confirmation of the selected phenotype.

The resultant strains were separated into classes based on growth in minimal glucose medium. For each class, a transposon [Tn10d(Tc)] genetically linked to the causative mutation was isolated by standard genetic techniques and used to reconstruct the mutant for phenotypic confirmation. The chromosomal locations of relevant insertions were determined by sequencing using a PCR-based protocol (35). A DNA product was amplified with degenerate primers and primers derived from the Tn10d(Tc) insertion sequence and sequenced. The genomes of the reconstructed strains were sequenced to identify the causative mutations (see below). Alternatively, putative loci were PCR amplified and sequenced by GeneWiz (South Plainfield, NJ). The strains were reconstructed in the DM14419 background.

Genome sequencing.

Whole-genome sequencing was used to identify the causative suppressor mutations in S. enterica DM13631 and DM13718 (ptsI611). The protocol used for preparation of genomic DNA has been described previously (36).

Genomic DNA was submitted to the Georgia Genomics Facility (GGF) at the University of Georgia (Athens, GA) for paired-end (250-bp) sequencing using the Illumina MiSeq platform. DNA samples were fragmented and tagged with sequencing adapters using the Nextera XT DNA sample preparation kit (Illumina, San Diego, CA). Processing and assembly of the sequencing data were performed by the Georgia Advanced Computing Resource Center (GACRC) at the University of Georgia. Briefly, the raw sequencing data were cleaned up using Trimmomatic (research laboratory of B. Usadel, Max Planck Institute, Germany) with a read length cutoff of 100 bp, resulting in >300-fold coverage of the 4.95-Mb S. enterica LT2 genome (37). Trimmed reads were mapped to the published genome using Bowtie 2 (Source Forge). Variant calling was performed using the Genome Analysis Toolkit (Broad Institute, Cambridge, MA), and single nucleotide polymorphisms (SNPs) were identified using the Integrative Genomics Viewer (Broad Institute).

β-Galactosidase assays.

β-Galactosidase assays were performed with modifications of previously described methods (38, 39). To determine β-galactosidase activity in NB medium, a 50-μl sample of overnight culture grown in NB medium with ampicillin (30 mg/liter) was inoculated into 5 ml medium with ampicillin (30 mg/liter). The cultures were incubated at 37°C with shaking until an optical density at 600 nm (OD₆₀₀) of 0.5 to 0.7 was reached, at which time samples were removed and incubated with α-methylglucoside (α-MG) (0.5%) or with an equal volume of double-distilled water (ddH₂O) in a 96-well plate. After the cells were incubated in the presence or absence of α-MG for 45 min at 37°C with shaking, the OD₆₀₀ of 175-μl cell samples was determined by a SpectraMax 385 Plus microplate spectrophotometer (Molecular Devices, Sunnyvale, CA). Twenty-microliter samples of cells were then added to 80 μl permeabilization solution that contained Na₂HPO₄ (100 mM), KCl (20 mM), MgSO₄ (2 mM), hexadecyltrimethylammonium bromide (1.6 mM), deoxycholic acid sodium salt (0.9 mM), and β-mercaptoethanol (77 mM). After at least 10-min incubation, 25 μl permabilized cell mixture was added to 150 μl substrate solution that contained Na₂HPO₄ (60 mM), NaH₂PO₄ (40 mM), o-nitrophenyl-β-d-galactoside (ONPG) (3.3 mM), and β-mercaptoethanol (39 mM). The reaction mixtures were incubated at 30°C, and ONP product formation was monitored at A₄₂₀ over time. Rates (ΔA₄₂₀/min) were determined by fitting the data to a linear equation with outlier elimination in GraphPad Prism 6.0d (La Jolla, CA). Specific activity A was calculated by using the formula 1,000 × rate (ΔA₄₂₀/min)/OD₆₀₀ using OD₆₀₀ as measured from 175-μl cell samples in a flat-bottom 96-well plate using a SpectraMax 385 spectrophotometer.

Alternatively, to determine β-galactosidase activity in minimal glucose medium, a 50-μl sample of overnight culture grown in NB medium with ampicillin (30 mg/liter) was inoculated into 5 ml minimal glucose medium with ampicillin (15 mg/liter) and with α-MG (1%) as indicated. The cultures were incubated at 37°C with shaking until mid-logarithmic-phase growth, and the OD₆₀₀ was 0.5 to 0.7 in borosilicate tubes (18 by 150 mm) as measured by a Spectronic 20+ spectrophotometer (Thermo Scientific, Waltham, MA). Then, 20-μl samples were added to 80 μl permeabilization solution, and β-galactosidase activity was measured as described above. Specific activity B was calculated by using the formula 1,000 × rate (ΔA₄₂₀/min)/OD₆₀₀ using OD₆₀₀ as measured from 5-ml cell samples in borosilicate tubes (18 by 150 mm) using a Spectronic 20+ spectrophotometer.

RESULTS AND DISCUSSION

Thi5p functions conditionally in S. enterica.

The THI5 gene was cloned into the pSU18 vector (pDM1336 and pDM1381) with expression from the lac promoter. S. enterica lacks the lactose utilization operon, and therefore, expression in the pSU18 vector was constitutive. THI5 expression did not allow growth of an S. enterica thiC mutant strain on minimal glucose medium (Table 3 and Fig. 2A). The addition of 2 or 5 mM cyclic AMP (cAMP) did not alter growth, indicating that the glucose-specific effect was not due to catabolite repression (data not shown). Further analysis assessed Thi5p-dependent growth of strain DM14419 (ΔthiC/pTHI5) on a suite of carbon sources (Table 3). Thiamine allowed full growth of this strain on all carbon sources, as expected. Significant Thi5p-dependent thiamine synthesis was observed on a limited number of permissive carbon sources, including ribose, xylose, and mannose. Even on these permissive carbon sources, the growth rates in the absence of thiamine were low; however, a high final density was reached. In contrast, Thi5p did not support significant growth in the absence of thiamine on glucose and other carbon sources (e.g., pyruvate). Expression of THI5 from pDM1361 (pBAD18S-THI5) induced with arabinose (0.02% or 0.2%) did not change the growth pattern (data not shown). These results indicated that the differential Thi5p-dependent growth was not the result of poor gene expression on different carbon sources. This conclusion was further demonstrated by Western blot analyses that showed the accumulation of Thi5p on a nonpermissive (glucose) and permissive (ribose) carbon source was not significantly different (Fig. 3). Taken together, these data showed that Thi5p function was impacted by components of the metabolic network that differed based on carbon source.

TABLE 3.

Thi5p-dependent growth on different carbon sources^a

Carbon sourceb	Growth ratec (h−1)		Final cell yieldd
	Without thiamine	With thiamine	Without thiamine	With thiamine
Acetate (50 mM)	0.01 ± 0.01	0.16 ± 0.02	0.10 ± 0.01	0.60 ± 0.06
Fructose	0.07 ± 0.04	0.55 ± 0.02	0.23 ± 0.07	0.74 ± 0.01
Fructose-6-P	0.04 ± 0.01	0.64 ± 0.02	0.15 ± 0.01	0.65 ± 0.01
Fumarate (50 mM)	0.03 ± 0.03	0.07 ± 0.05	0.16 ± 0.01	0.21 ± 0.01
Galactose	0.03 ± 0.01	0.54 ± 0.04	0.13 ± 0.01	0.75 ± 0.01
Gluconate	0.02 ± 0.01	0.57 ± 0.05	0.11 ± 0.01	0.52 ± 0.01
Glucose	0.01 ± 0.01	0.59 ± 0.03	0.12 ± 0.02	0.76 ± 0.02
Glucose-6-P	0.06 ± 0.02	0.61 ± 0.08	0.12 ± 0.02	0.60 ± 0.01
Glycerol	0.12 ± 0.06	0.50 ± 0.03	0.25 ± 0.09	0.81 ± 0.01
Malate (40 mM)	0.01 ± 0.01	0.40 ± 0.02	0.12 ± 0.03	0.64 ± 0.01
Mannose	0.19 ± 0.05	0.53 ± 0.01	0.59 ± 0.09	0.74 ± 0.01
Pyruvate (50 mM)	0.02 ± 0.01	0.41 ± 0.03	0.12 ± 0.01	0.86 ± 0.01
Ribose	0.15 ± 0.01	0.43 ± 0.04	0.51 ± 0.03	0.67 ± 0.01
Succinate (20 mM)	0.05 ± 0.03	0.12 ± 0.04	0.19 ± 0.05	0.26 ± 0.08
Xylose	0.15 ± 0.01	0.36 ± 0.01	0.64 ± 0.01	0.73 ± 0.01

^aStrain DM14419 (ΔthiC1225 ΔaraCBAD/pTHI5) was grown at 37°C in minimal medium for 36.5 h with each carbon source.

^bEach carbon source at 66 mM available carbon units unless indicated otherwise. Fructose-6-P, fructose-6-phosphate; glucose-6-P, glucose-6-phosphate.

^cGrowth rate is reported as μ, which is equal to ln(X/X₀)/T, where X is OD₆₅₀, X₀ is the initial OD₆₅₀ value for the period analyzed during exponential growth, and T is the time (in hours). Values are averages ± standard deviations from three independent cultures.

^dFinal cell yield is reported as the maximum OD₆₅₀. Values are averages ± standard deviations from three independent cultures.

FIG 2.

Suppressor alleles allow Thi5p-dependent growth on minimal glucose medium. The thiC/pTHI5 strain and derivative carrying sgrR1 were grown in minimal glucose medium without thiamine (empty symbols) and with thiamine (filled symbols). The strains shown are DM13623 (ΔthiC/pTHI5) (A) and DM13631 (ΔthiC sgrR1/pTHI5) (B). Growth was monitored by optical density at 650 nm with shaking at 37°C. Values are averages ± standard deviations (error bars) for three independent cultures.

FIG 3.

Thi5p accumulates in minimal medium with glucose or ribose as the carbon source. The strain containing pSU18 expressing THI5 (DM14419; lanes 3 to 5 and 7 to 9) and the strain carrying the empty vector (DM14148; lanes 6 and 10) were grown overnight in minimal medium with chloramphenicol (5 mg/liter) and glucose or ribose, as indicated. Thiamine was included when necessary for growth, as indicated (+Thi). Lane 1 contained 12.5 ng purified His₆Thi5p, and lane 2 was empty. αThi5p, anti-Thi5p antibody.

A genetic approach identified metabolic factors that allowed Thi5p activity in S. enterica. Suppressor mutations that allowed growth of a ΔthiC1225 ΔaraCBAD/pTHI5 strain on minimal glucose medium were isolated. Colonies arose at a rate of ∼1 × 10⁻⁷ when strain DM13623 was plated on minimal glucose medium, but none retained this phenotype after purification. Addition of diethyl sulfate (DES) to the selection plate increased the number of colonies and resulted in stable revertants. Four independent, stable derivatives of S. enterica DM13623 that grew on minimal glucose medium plates were further analyzed. Genetic mapping, along with targeted and whole-genome sequencing, identified one causative mutation in the ptsI locus and two in the sgrR locus. A fourth mutation (zxx10175) was not identified and was genetically unlinked to either the ptsI or sgrR locus. The sgrR suppressor strains were chosen for further investigation, and their growth patterns were characterized in minimal glucose medium with and without thiamine. Representative data are shown in Fig. 2. The data showed that, as expected, the growth rate of the parent strain DM13623 was increased by the addition of thiamine (0.04 ± 0.03 h⁻¹ and 0.60 ± 0.03 h⁻¹ without and with thiamine, respectively). In contrast, the strain carrying mutant allele sgrR1 (Fig. 2B) grew approximately as well with or without thiamine supplementation. It was noted that the growth rate of the sgrR1-carrying strain with thiamine was less than the growth rate of the parental strain (0.33 ± 0.01 h⁻¹). This result suggested that the suppressing mutations carried a fitness cost to the strains on glucose medium, a result not atypical when the metabolic network is remodeled to compensate for a perturbation or to generate a new function (40–44).

Alleles of sgrR allow Thi5p function in S. enterica.

Two alleles of sgrR (sgrR1 and sgrR2) were isolated as suppressor mutations that allowed growth of a ΔthiC1225 ΔaraCBAD/pTHI5 strain on minimal glucose medium. The same strain lacking pTHI5 failed to grow, indicating that Thi5p was satisfying the thiamine requirement of the cell. SgrR mediates the sugar-phosphate stress response that has been characterized in E. coli. SgrR belongs to a novel class of transcription regulators (COG4533) and has a predicted N-terminal DNA-binding domain and C-terminal solute-binding domain (20, 21). Analysis of S. enterica SgrR in the Interpro database (45) identified the N-terminal binding domain as amino acids 5 to 118, with the solute-binding domain beginning at amino acid 163. The sgrR1 allele encodes a variant with a substitution in the putative solute-binding domain (SgrR with the G-to-R change at position 525 [SgrR^G525R]), while the sgrR2 allele encodes a variant with a substitution immediately following the predicted DNA-binding domain (SgrR^R119W). A null allele of sgrR did not allow Thi5p-dependent growth on minimal glucose medium, indicating that the suppressing alleles encoded variants with altered function (data not shown).

Expression from the sgrS promoter (sgrSp) has been used as a reporter of SgrR activity (20) and was used to probe the function of the suppressing variants of SgrR. A vector (pDM1402) in which the sgrS promoter drives expression of the lacZ gene was constructed. This plasmid allowed β-galactosidase activity to be assayed as a proxy for expression from the sgrS promoter. The nonmetabolizable glucose analog α-methylglucoside (α-MG) is a gratuitous inducer of sugar-phosphate stress and SgrR activity (20). The effects of SgrR variants on expression of lacZ from pDM1402 with and without induction by α-MG were measured in nutrient medium. The data in Fig. 4 showed that both SgrR^G525R and SgrR^R119W increased transcription in the absence of inducer. These data suggested that the SgrR variants were constitutively active and resulted in the continual induction of the sugar-phosphate stress response. In the presence of the α-MG inducer, expression of lacZ did not increase to the same level allowed by the wild-type protein. The above results were consistent with the hypothesis that the variants of SgrR that allowed function of Thi5p had an altered ability to sense the appropriate signal.

FIG 4.

Effects of sgrR alleles on transcription from the sgrS promoter. β-Galactosidase assays were performed on cultures grown in NB medium containing ampicillin (30 mg/liter) to mid-log phase and then incubated with α-methylglucoside (α-MG) (+) and without α-MG (0.5%) (−) for 45 min before permeabilization. The data are reported as specific activity A (1,000 × A₄₂₀ min⁻¹ OD₆₀₀⁻¹, where the OD₆₀₀ was determined from a 175-μl cell sample in a flat-bottom 96-well plate using a SpectraMax 385 spectrophotometer) and are averages ± standard deviations from three independent cultures. These results are representative of four independent experiments.

In the above-described assays, the cells were grown in nutrient medium, potentially masking a response that was relevant for the ability of the sgrR alleles to allow Thi5p to function in vivo. Strains with representative sgrR alleles and the sgrSp-lacZ reporter construct were grown in minimal glucose medium with thiamine. The β-galactosidase activity was measured during mid-logarithmic growth to assess transcription mediated by SgrR^G525R encoded by sgrR1. The data in Fig. 5 showed that sgrR1 increased expression from the sgrS promoter on minimal thiamine medium when glucose was the carbon source. Therefore, a variant SgrR that allowed Thi5p-dependent thiamine synthesis activated the sugar-phosphate stress response constitutively during growth on glucose.

FIG 5.

Suppressor mutations affect transcription from the sgrS promoter. β-Galactosidase assays were performed on cultures grown in minimal glucose medium containing ampicillin (15 mg/liter), thiamine, and 1% α-MG as indicated to mid-log phase before permeabilization. The data are reported as specific activity B (1,000 × A₄₂₀ min⁻¹ OD₆₀₀⁻¹, where the OD₆₀₀ was determined from a 5-ml culture in borosilicate tubes [18 by 150 mm] using a Spectronic 20+ spectrophotometer) and are averages plus standard deviations from three independent cultures.

Disruption of glycolysis allowed Thi5p-dependent growth on glucose.

Glycolytic flux has been tied to the induction of the sugar-phosphate stress response (46). Mutations that eliminated phosphoglucose isomerase activity (encoded by pgi) or phosphofructokinase A activity (encoded by pfkA and responsible for ∼95% of the phosphofructokinase activity in E. coli [47]) were introduced into a Thi5p-containing strain, and thiamine synthesis was assessed by growth on minimal medium. Eliminating either pgi or pfkA allowed Thi5p-dependent growth on glucose (Table 4). Previous reports found that E. coli strains disrupted in pgi and pfkA had destabilized ptsG mRNA (48, 49), presumably through the sugar-phosphate stress response (20, 21). Therefore, the strains carrying pgi or pfkA mutations were assessed for their sugar-phosphate stress response induction by monitoring the expression of sgrSp-lacZ during growth on glucose medium. The data in Fig. 5 showed that elimination of either locus induced the expression of sgrS, again consistent with the conclusion that induction of the sugar-phosphate stress response was associated with the activation of Thi5p in S. enterica.

TABLE 4.

Thi5p-dependent growth on glucose in strains disrupted in glycolysis^a

Strain	Relevant genotype	Growth ratea (h−1)		Final cell yieldb
		Without thiamine	With thiamine	Without thiamine	With thiamine
DM14419	thiC/pTHI5	0.01 ± 0.01	0.60 ± 0.01	0.11 ± 0.01	0.84 ± 0.01
DM14443	pgi thiC/pTHI5	0.09 ± 0.01	0.10 ± 0.01	0.66 ± 0.01	0.87 ± 0.01
DM14540	pfkA thiC/pTHI5	0.11 ± 0.03	0.14 ± 0.01	0.65 ± 0.07	0.90 ± 0.02

^aGrowth rate is reported as μ, which is equal to ln(X/X₀)/T, where X is OD₆₅₀, X₀ is the initial OD₆₅₀ value for the period analyzed during exponential growth, and T is the time (in hours). Values are averages ± standard deviations from three independent cultures.

^bFinal cell yield is reported as the maximum OD₆₅₀. Strains DM14419, DM14443, and DM14540 were grown with shaking at 37°C in minimal medium with 11 mM glucose. Values are averages ± standard deviations from three independent cultures.

Induction of the sugar-phosphate stress response is required to decrease PtsG activity.

The data above supported a general model in which the constitutive induction of the sugar-phosphate stress response allowed Thi5p function in S. enterica on minimal glucose medium. α-MG was used as a gratuitous inducer of the sugar-phosphate stress response to further test this conclusion. Titration of α-MG determined that when 1% α-MG was added to the medium, growth of strain DM14419 (ΔthiC/pTHI5) occurred on glucose in the absence of thiamine (growth rate of 0.139 ± 0.01 h⁻¹ with α-MG and of <0.01 ± 0.01 h⁻¹ without α-MG). Expression from the sgrSp-lacZ reporter plasmid in strain DM14531 (ΔthiC sgrSp) during mid-log growth in minimal glucose medium containing 1% α-MG was at a low level of expression, similar to that induced by sgrR1 on minimal glucose medium (Fig. 5).

The primary target of the sugar-phosphate stress response, PtsG, was tested for an effect on Thi5p function in minimal glucose medium. An insertion mutation in ptsG allowed Thi5p-dependent thiamine synthesis in a ΔthiC strain, as shown in Fig. 6A. The effect of the ptsG mutation was independent of sgrS, consistent with a role for sgrS in decreasing PtsG activity (Fig. 6C). In contrast, sgrR was required for Thi5p-dependent thiamine synthesis allowed by the ptsG mutation (Fig. 6B). Together, these results indicated that the suppression observed in the sgrR1 and sgrR2 mutant strains had two components: the constitutive expression of sgrS resulted in decreased PtsG activity and an additional, undefined role of SgrR. Further, the role of SgrR was not to express alaC, which is the only known SgrR target in S. enterica that is not regulated via sgrS (Fig. 6D). Finally, these data showed that the secondary role of SgrR did not require its “activation,” since the allele present was the wild-type allele and no known inducing conditions were needed. Consistent with this interpretation, the presence of a ptsG mutation did not affect the basal level of expression from the sgrSp-lacZ reporter or its ability to be induced by α-MG (data not shown). It was noted that the mutation in alaC increased the lag before growth. The specific cause for this is not obvious, but it could indicate a subtle role of this gene in the metabolic network required to support Thi5p-dependent thiamine synthesis.

FIG 6.

An insertion in ptsG allows Thi5p-dependent growth on glucose. The thiC ptsG/pTHI5 strain and derivatives with insertions or deletions of sgrR, sgrS, and alaC were grown in minimal glucose medium without thiamine (empty symbols) and with thiamine (filled symbols). The strains shown are DM14517 (thiC ptsG/pTHI5) (A), DM14668 (thiC ptsG sgrR/pTHI5) (B), DM14896 (thiC ptsG sgrS/pTHI5) (C), and DM14932 (thiC ptsG alaC/pTHI5) (D). Growth was monitored by optical density at 650 nm with shaking at 37°C. Data are averages ± standard deviations for two independent cultures and are representative of three experiments.

Induction of the sugar-phosphate stress response is not necessary for Thi5p activity.

Together, the results above suggested that induction of the sugar-phosphate stress response was involved in at least one mechanism that allowed Thi5p function in S. enterica during growth on glucose. The data showed that one relevant component of this response was the induction of sgrS that led to decreased PtsG. Other observations indicated that the stress response, and in fact SgrR, was not required for all mechanisms that allowed Thi5p function in S. enterica. For instance, when the parental strain (DM14419) was grown on a permissive carbon source (ribose), expression of the sgrSp-lacZ reporter was at a low basal level, and deletion of sgrR did not prevent Thi5p-dependent growth (data not shown). These results indicated that induction of the sugar-phosphate response was not the only mechanism that allowed function of Thi5p in S. enterica and implicate a role for the different carbon source-specific metabolic networks in Thi5p activity in S. enterica. In addition, the ptsI allele that restored Thi5p-dependent growth on minimal glucose medium did not require sgrR or sgrS for this effect (data not shown).

Conclusions.

S. cerevisiae HMP-P synthase Thi5p is capable of supporting thiamine-independent growth in a ΔthiC mutant strain of S. enterica under certain conditions. The conditionality of Thi5p function in S. enterica provided an opportunity to dissect the metabolic components involved. Two suppressor mutations that allowed Thi5p function on minimal glucose medium were alleles of sgrR, which encodes a transcriptional regulator. The suppressing alleles resulted in constitutive expression of sgrS, suggesting a link between Thi5p and the sugar-phosphate stress response. This link was extended when null mutations in pgi (encodes phosphoglucose isomerase) or pfkA (encodes phosphofructokinase A) also (i) induced sgrS expression and (ii) allowed Thi5p-dependent thiamine synthesis on minimal glucose medium. Further, inducing the sugar-phosphate stress response with the gratuitous inducer α-MG allowed Thi5p-dependent thiamine synthesis on minimal glucose medium.

The finding that disrupting ptsG independently restored thiamine synthesis revealed one mechanism connecting Thi5p function and induction of the sugar-phosphate stress response. The results associated with strains lacking ptsG supported the working model in which one of the mechanisms of Thi5p activation requires decreased PtsG function and an undefined role of SgrR. The expectation is that the role of SgrR is to regulate a gene whose product is required to remodel the metabolic network to allow Thi5p function. Interestingly, the data suggest that the relevant role of SgrR does not depend on conditions that are known to activate the regulator. Thus, this work has the potential to expand our understanding of SgrR and how it contributes to metabolism beyond its role in the sugar-phosphate stress response.

When considered in total, the data suggest that Thi5p must be activated in order to function in S. enterica, possibly by phosphorylation or another posttranslational modification. This scenario would demand a Thi5p-activating enzyme in S. cerevisiae with no S. enterica ortholog. Instead, robustness in the S. enterica metabolic network would support Thi5p activation under certain conditions, including those generated by growing on a permissive carbon source or by remodeling the metabolic network by inducing the sugar-phosphate stress response during growth on glucose. Alternative hypotheses include the hypothesis that Thi5p activity is inhibited during growth on glucose or that a relevant cofactor is not available. In either case, the limitation can be alleviated through activation of the sugar-phosphate stress response, specifically the decrease in PtsG activity. Regardless of which scenario is correct, further definition of this system will have implications for in vitro studies on Thi5p. Production of Thi5p in minimal glucose medium (as reported by Lai et al. [11]) could deleteriously affect its activity in reconstituted assays. In total, the findings here emphasize that metabolism is characterized by complex metabolic networks that are not always amenable to “plug-and-play” approaches, even if all required substrates are present. The data further presented a number of provocative questions about the enzymology of Thi5p and the sugar-phosphate stress response regulator. The results reported here identify a new area of investigation that has the potential to yield unexpected knowledge about central metabolic processes and the integration of heterologous components to a network.