Inhibition of Fructose-1,6-bisphosphatase by Aminoimidazole Carboxamide Ribotide Prevents Growth of Salmonella enterica purH Mutants on Glycerol*

The enzyme fructose-1,6-bisphosphatase (FBP) is key regulatory point in gluconeogenesis. Mutants of Salmonella enterica lacking purH accumulate 5-amino-4-imidazole carboxamide ribotide (AICAR) and are unable to utilize glycerol as sole carbon and energy sources. The work described here demonstrates this lack of growth is due to inhibition of FBP by AICAR. Mutant alleles of fbp that restore growth on glycerol encode proteins resistant to inhibition by AICAR and the allosteric regulator AMP. This is the first report of biochemical characterization of substitutions causing AMP resistance in a bacterial FBP. Inhibition of FBP activity by AICAR occurs at physiologically relevant concentrations and may represent a form of regulation of gluconeogenic flux in Salmonella enterica.

AICAR has been shown to impact different metabolic pathways in diverse organisms. Bochner and Ames (1) proposed that 5amino-4-imidazole carboxamide riboside 5Ј-triphosphate formed from the accumulation of AICAR is an alarmone for C1-folate deficiency in S. enterica, although this conclusion, in its simplest form, was not supported by subsequent work (2). AICAR has been shown to be a negative effector of cytochrome terminal oxidase cbb 3 production in Rhizobium etli (3) and was recently shown to affect regulation of purine biosynthesis genes in Saccharomyces cerevisiae (4). Previous work from our laboratory demonstrated that AICAR accumulation inhibits the conversion of aminoimidazole ribotide to 4-amino-5-hydroxymethyl-2-methyl pyrimidine phosphate in the biosynthesis of thiamine pyrophosphate in S. enterica (5).
In mammalian systems, AICAR has been shown to activate AMP-activated protein kinase (6), which plays a critical role in energy balance and the response to metabolic stress (7). Through activation of AMP-activated protein kinase, AICAR inhibits proliferation of several types of cancer cells (8,9). Furthermore, AICAR has been shown to inhibit gluconeogenesis in mammals due to inhibition of fructose-1,6bisphosphatase (10,11).
Fructose-1,6-bisphosphatase (D-fructose-1,6-bisphosphate 1-phosphohydrolase, EC 3.1.3.11; FBP) catalyzes the hydrolysis of fructose 1,6-bisphosphate (Fru-1,6-P) to fructose 6-phosphate and inorganic phosphate (12). FBP is important for the regulation of flux between glycolysis and gluconeogenesis (Fig.  1B); as the enzyme is inhibited by AMP and fructose 2,6bisphosphate (Fru-2,6-P) (13). Fru-2,6-P is a competitive inhibitor with respect to the substrate Fru-1,6-P, which binds at the active site (14,15), whereas AMP binds to an allosteric site (16,17), which causes a structural transition from the active R state of the enzyme to the inactive T state (18). FBP from mammalian sources has been extensively studied at the biochemical level. In Escherichia coli, FBP activity is required for growth on glycerol, succinate, or acetate as sole carbon and energy source (19). E. coli FBP has been shown to be activated by Mg 2ϩ and inhibited by AMP and Fru-2,6-P, similar to the mammalian enzymes, although Mg 2ϩ activation and AMP inhibition did not display cooperativity, as in the mammalian enzymes (20). The physiological relevance of Fru-2,6-P inhibition of E. coli FBP is not clear; Fru-2,6-P inhibition is not synergistic with AMP inhibition as in mammalian systems (20,21), and to date Fru-2,6-P has not been found in bacterial cells (22). Phosphoenolpyruvate has been shown to relieve AMP inhibition of E. coli FBP (21,23). Recently, the crystal structure of E. coli FBP was reported; these authors also demonstrated activation of FBP by phosphoenolpyruvate and postulated a novel allosteric binding site for this compound (24).
The studies described here demonstrate that purH mutants of S. enterica are unable to use glycerol or other gluconeogenic carbon sources in the absence of exogenous histidine. Results show that this growth phenotype is due to inhibition of FBP by the accumulation of AICAR. AICAR-resistant alleles of fbp were isolated and the mutant proteins were purified and characterized in vitro. FBP is inhibited by AICAR at physiologically relevant concentrations and thus this interaction may represent a form of regulation relevant under some growth conditions, such as folate stress.

Bacterial Strains, Media, and Chemicals
All strains used in this study are derivatives of S. enterica LT2 and are listed with their genotypes in Table 1. The purH355 allele contains a thymidine inserted at position 716, resulting in a frameshift and generating a truncated PurH protein of 266 amino acids. MudJ refers to the Mud1734 insertion element (25) and Tn10d(Tc) refers to the transposition-defective mini-Tn10 described by Way et al. (26). No-carbon E medium supplemented with 1 mM MgSO 4 (27,28) and glucose (11 mM) or glycerol (22 mM) was used as a minimal medium. Difco nutrient broth (8 g/liters) with NaCl (5 g/liters) and Luria-Bertani broth were used as rich media. Difco BiTek agar was added (15 g/liter) for solid medium. When present in the media, supplements were provided at the following final concentrations: thiamine, 10 or 100 nM; adenine, 0.4 mM; histidine, 74 M. When needed, antibiotics were added to the following concentrations in rich/ minimal media: tetracycline, 20/10 g/ml; kanamycin, 50/125 g/ml; chloramphenicol, 20/4 g/ml; and ampicillin, 50/15 g/ml. Restriction enzymes and DNA ligase were purchased from Promega (Madison, WI). Cloned Pfu DNA polymerase was purchased from Stratagene (La Jolla, CA).

Genetic Methods
Phenotypic Analysis-Nutritional requirements were assessed on solid medium with soft agar overlays and by quantification of growth in liquid media. Protocols for each have been previously described (29,30).
Transduction Methods-The high-frequency generalized transducing mutant version of bacteriophage P22 (HT105/1 int-201) was used in all transductions (31,32). The methods for transduction and purification of transductants have been previously described (33).
Isolation of Linked Insertions-Transposons (Tn10d(Tc)) genetically linked to the mutation allowing utilization of glycerol as sole carbon source in a purH mutant were isolated by standard genetic techniques (34). The chromosomal location of each insertion was determined by sequencing using a PCRbased protocol (35,36). A DNA product was amplified with degenerate primers and primers derived from the Tn10d(Tc) insertion sequence as described and sequenced at the University of Wisconsin Biotechnology DNA Sequence Facility.

Molecular Biology Techniques
The fbp gene was amplified from wild-type and mutant strains using genomic DNA as a template and the forward primer, 5Ј-TACTTTGCTGGCGCGATTCACCTG-3Ј, and the reverse primer, 5Ј-TGTTTTCGTTTCCGCCTCATCTGG-3Ј. The resulting PCR products were purified and ligated into SmaI-cut pSU19. Plasmids were transformed into E. coli strain DH5␣ and screened for vectors containing inserts. The identity of the resulting plasmids was confirmed by sequencing. Plasmids for expression of His-tagged FBP were constructed as described above using plasmid DNA as the template and the forward primer 5Ј-GCAGGGAAAGTCATATGAAAACG-3Ј and the reverse primer 5Ј-CTCGAGCGCGTCCGGGTAT-TCGCGGATAAA-3Ј (engineered restriction sites underlined). These PCR products were ligated into pET20 (Novagen) digested with NdeI and XhoI and screened as described above. Plasmids are listed in Table 1.

Purification of FBP
Cultures (10 ml) of DM9297 containing an fbp-expression plasmid (Table 1), grown overnight in LB with 11 mM glucose, chloramphenicol, and ampicillin, were used to inoculate 1 liter of LB with 2.5 mM glucose, chloramphenicol, and ampicillin. The culture was grown at 37°C with shaking (200 rpm) to an optical density at 650 nm of 0.6 -0.7 and induced with 1 mM arabinose. After 3 h at 37°C, cells were harvested by centrifugation (7000 ϫ g, 15 min, 4°C). Cell pellets were resuspended in 15 ml of buffer A (50 mM sodium phosphate, 300 mM NaCl, 50 mM imidazole, pH 7.0) containing protease inhibitor mixture (Sigma P-8849) and disrupted with a French pressure cell (15,000 lb/inch 2 ). The crude extract was clarified by centrifugation (39,000 ϫ g, for 30 min at 4°C). The clarified extract was loaded onto a HisTrap FF column (Amersham Biosciences) equilibrated with buffer A, then the column was washed with 15 column volumes of buffer A. Protein was eluted with a linear gradient (20 column volumes) from 50 to 500 mM imidazole in buffer A. Fractions containing FBP as judged by SDS-PAGE analysis were pooled, concentrated to 7-10 mg/ml, exchanged into buffer B (10 mM potassium phosphate, 20% glycerol, 1 mM EDTA, 1 mM dithiothreitol, 0.4 mM fructose 1,6-bisphosphate, pH 7.0), frozen in liquid N 2 , and stored at Ϫ80°C.

Assay of FBP Activity
FBP activity was determined using a coupled, spectrophotometric assay (37). Aliquots of FBP were thawed and diluted to 0.05 g/ml in buffer C (10 mM potassium phosphate, 20% glycerol, pH 7.0) immediately prior to assay. Reactions (1 ml) contained: 100 mM Tris-Cl, pH 7.5, 0 -100 M Fru-1,6-P, 0 -25 mM MgCl 2 , 2 mM ammonium sulfate, 0.2 mM NADP ϩ , 50 M EDTA, 5 mM ␤-mercaptoethanol, 0.8 units of glucose-6-phosphate dehydrogenase, 0.8 units of phosphoglucose isomerase, and FBP. Reactions were initiated by the addition of FBP (250 ng), and the reduction of NADP ϩ was monitored at 340 nm (⑀ ϭ 6.22 mM Ϫ1 cm Ϫ1 ). The linear steady-state portion of the A 340 versus time plot was used to determine the kinetic parameters k cat , K m (Fru-1,6-P), and Hill coefficient by fitting initial velocity data to Equation 1 (R 2 Ն 0.99), where is velocity, S is either the concentration of varied cofactor or substrate, V max is the velocity at saturating substrate and cofactor, K m is the Michaelis constant for the varied cofactor or substrate, and H is the Hill coefficient. The velocity data as plotted in the double-reciprocal form was fit by the following function, where x is 1/[substrate], and y is 1/velocity. The initial slope is defined as b Ϫ ae, the slope of the asymptote is c/e, the y intercept is a, and the asymptote intercept is (be Ϫ c)/e 2 .
Inhibition by AMP was studied by using saturating Fru-1,6-P (100 M), five concentrations of MgCl 2 (0.5-8 mM), and five concentrations of AMP (0 -8 M for FBP(wt), 0 -0.6 mM for FBP(T19I), and 0 -8 mM for FBP(G20D) and FBP(T23I)). Inhibition by AICAR was studied by using saturating Fru-1,6-P (100 M), five concentrations of MgCl 2 (0.5-8 mM), and five concentrations of AICAR (0 -120 M for FBP(wt) and 0 -10 mM for mutant FBP enzymes). Inhibition data were fit to the following equation (Equation 3) for nonlinear mixed-type inhibition (R 2 Ն 0.95), where is the velocity, V max is the velocity with saturating substrate and cofactor with no inhibitor present, K a is the Michaelis constant for Mg 2ϩ , S is the concentration of Mg 2ϩ , K ic is the competitive dissociation constant for inhibitor from the enzyme-inhibitor complex, K iu is the uncompetitive dissociation constant for inhibitor from the enzyme-inhibitor complex, I is the concentration of inhibitor, and H is the Hill coefficient for Mg 2ϩ derived from Equation 1. Linear and nonlinear regression analysis was preformed using SigmaPlot 9.0 (Systat Software).

RESULTS
Salmonella purH Mutants Are Unable to Use Glycerol as a Sole Carbon and Energy Source-The observation was made that a purH mutant strain of S. enterica (DM2) was unable to grow on minimal medium containing glycerol as sole carbon and energy source and supplemented with adenine and thiamine (Table 2). Further experiments demonstrated that this growth defect was eliminated by the addition of exogenous histidine. This finding led to the hypothesis that the accumulation of the purine biosynthetic intermediate, AICAR, was inhibiting an enzyme(s) required for growth on glycerol medium. AICAR is also generated as a byproduct of histidine biosynthesis; therefore addition of exogenous histidine reduces the intracellular concentration of AICAR in a purH mutant (5).
Mutations That Allow a purH Mutant to Grow on Glycerol Medium in the Absence of Histidine Map to the fbp Locus-A purH mutant (DM2) was plated on minimal glycerol medium supplemented with adenine and thiamine and mutagenized with diethyl sulfate (3 l spotted in the middle of a plate containing 10 8 colony forming units). Colonies appeared within 48 h and were concentrated in a ring around the spot of diethyl sulfate; 24 colonies were purified and re-tested for the glycerolpositive phenotype. One of these mutants was a purH ϩ revertant; and it was discarded. Twelve of the remaining strains were chosen at random for further study. The causative mutation in each was found to be genetically linked to a transposon located near the fbp locus. Six of these mutants were reconstructed and the fbp gene was amplified by PCR, cloned into pSU19, and sequenced. Three unique mutations were identified, causing the following amino acid substitutions: T19I, G20D, and T23I. Mutations causing the T19I and T23I substitutions were independently isolated multiple times. Isogenic strains containing each of the three mutations on the chromosome were constructed and tested for their ability to grow on minimal glycerol medium supplemented with adenine and thiamine ( Table 2). The presence of any of these three mutations allowed a purH mutant to grow on minimal glycerol medium supplemented with adenine and thiamine. Strains containing null mutations in fbp also fail to grow on other gluconeogenic carbon sources such as acetate and succinate (19). In addition to glycerol, purH mutants failed to use acetate and succinate as sole carbon and energy sources in the absence of exogenous histidine (data not shown).
The Mutant FBP Enzymes Are Resistant to Inhibition by AICAR and AMP in Vitro-Wild-type and mutant FBP enzymes were purified as His 6 -tagged fusion proteins via Ni 2ϩ affinity chromatography. The purified enzymes were judged to be ϳ90% pure by image analysis (Phoretix TotalLab) (Fig. 2). Kinetic parameters were determined from initial velocity data for the purified enzymes as described under "Experimental Procedures." Fig. 3 Table 3. The values obtained for the  NOVEMBER 10, 2006 • VOLUME 281 • NUMBER 45 wild-type enzyme were similar to those reported for the E. coli enzyme (20). Inhibition by AMP and AICAR was examined for the wildtype and variant enzymes. The double-reciprocal plots of the versus [Mg 2ϩ ] were found to be nonlinear and inhibition visually appeared to be competitive relative to [Mg 2ϩ ]. The plots were nonlinear because of the cooperative behavior of Mg 2ϩ binding to the enzymes. Linearity was obtained by raising the [Mg 2ϩ ] to the power of the Hill coefficient of the specific enzymes, which was determined using Equation 1 (Table 3). A representative double reciprocal plot of wild-type FBP inhibition by AICAR is shown in Fig. 4. The type of inhibition exhibited for AMP and AICAR was determined by graphing initial rate data in the form of double reciprocal plots at various inhibitor concentrations and Dixon plots (1/ versus [I]). By comparing actual data to theoretical values obtained by fitting the nonlinear data to various inhibition equations it was found that the data are best described by Equation 3 for mixed-type inhibition. The inhibition constants are shown in Table 4. This means that AMP and AICAR bind to both the E and E⅐Mg complex. The high K iu values and low K ic values obtained from the fits indicated that AMP and AICAR bind more tightly to the free enzyme and weaker to the E⅐Mg complex. Shown in the inset of Fig. 4 is a replot of the slopes for the theoretical data in Fig. 4 versus [AICAR] 1 . The relationship is linear indicating that AICAR and AMP (data not shown) do not inhibit the enzymes in a cooperative manner. As predicted from the phenotypic data, all three mutant forms of FBP had significantly higher K i values than the wild-type enzyme (K ic ϳ 120 -1100-fold for AMP, ϳ45-60-fold for AICAR; K iu ϳ 100 -1500-fold for AMP, 20 -40-fold for AICAR).

DISCUSSION
Results herein showed that the inability of S. enterica purH mutants to grow on gluconeogenic carbon sources in the absence of exogenous histidine is due to inhibition of FBP by the accumulation of the purine biosynthetic intermediate AICAR. Revertant mutations restoring growth of the purH mutant on glycerol mapped to fbp, and resulted in variants of FBP that were resistant to inhibition by AICAR and AMP.
The three variant proteins characterized in this work (T19I, G20D, and T23I) had lesions located in the AMP binding site of the FBP protein as determined from the crystal structure of the pig kidney enzyme (17). As shown in the partial alignment in Fig. 5, the AMP binding site is highly conserved between both bacterial and mammalian enzymes. S. enterica FBP is 97% identical and 98% similar to E. coli  FBP, 43% identical and 63% similar to the pig kidney enzyme, and 44% identical and 63% similar to the human liver enzyme. Like the mammalian enzyme, but unlike the E. coli enzyme, FBP from S. enterica displayed cooperative kinetic behavior that had not been reported (mammalian h Mg(II) ϳ 2 (38)) (20). Interestingly, S. enterica FBP also displayed cooperative behavior for Fru-1,6-P that has not been reported for any FBP studied to date. Moreover, the S. enterica enzyme did not show cooperative kinetic behavior for AMP or AICAR inhibition, whereas this behavior has been shown for these inhibitors in the mammalian enzyme (mammalian h AMP ϭ 2 (16,38,39)).
Thr 19 corresponds to Thr 27 in the mammalian enzymes; the side chain of this residue contacts the phosphate group of AMP in the crystal structure (17), and replacement of Thr 27 in the porcine enzyme with alanine increased the K i for AMP 1300fold (40). In this work, the T19I substitution increased the K ic for AMP 118-fold and the K iu 98-fold. Thr 23 corresponds to Thr 31 in the mammalian enzymes; the side chain of this residue forms a hydrogen bond with the purine base of AMP in the crystal structure, and site-directed mutagenesis studies have confirmed its importance in AMP inhibition (39,41). Gly 20 corresponds to Gly 28 of the mammalian enzymes, which forms a hydrogen bond to a water molecule in the AMP binding site but has not been studied by site-directed mutagenesis (42). Additional residues known to be important in AMP inhibition of the mammalian enzymes include Arg 22 , Ala 24 , Glu 29 , Ala 54 , Lys 112 , Tyr 113 , Gly 122 , and Arg 140 (39 -41, 43-45). A previous report on an AMP-insensitive fbp mutant of E. coli found that growth was unaffected by the presence of this mutation unless the allele was highly expressed, but further characterization of the enzyme was not reported (46).
The purine biosynthetic intermediate AICAR was shown to bind to the AMP binding pocket of human liver FBP by crystallographic studies (42) and shown to inhibit FBP in rat liver extracts (10,11). All three substitutions described in this work significantly reduce inhibition of the enzyme for both AMP and AICAR. It is interesting to note that a perfect correlation does not exist between the effects of each substitution on inhibition by AMP and AICAR despite the fact that both compounds bind at the same site. In particular, the T23I substitution increases the K ic for AMP 1100-fold, but only increases the K ic for AICAR 31-fold. This observation may be explained by crystallographic data demonstrating that AICAR binds in a slightly different orientation than AMP; specifically, the imidazole ring of AICAR is rotated 180°relative to AMP in the crystal structure (42). The side chain of this residue forms hydrogen bonds to the    purine ring and a water molecule in the AMP-bound structure, but in the AICAR-bound structure, the side chain hydrogen bond to the water molecule is replaced by a backbone nitrogen hydrogen bond from the same Thr residue. If the substitution of Ile for this Thr residue still allows the backbone hydrogen bond while eliminating the side chain hydrogen bond, this could explain the different effects of this substitution with respect to the two inhibitors. The K ic determined for AICAR with the wild-type enzyme (37.3 Ϯ 5.6 M) agrees well with the IC 50 determined by Iversen et al. (42) for the human liver enzyme. This is significant because it indicates that AICAR inhibits FBP at a physiologically relevant concentration. Bochner and Ames (47) determined the intracellular concentration of AICAR in Salmonella to be 105 M in minimal glucose medium, whereas Rohlman and Matthews (2) measured concentrations of 40 -150 M in E. coli strain W3110. AICAR has many physiological effects in various organisms. These data suggest that in enteric bacteria AICAR has the potential to affect flux through gluconeogenesis, as FBP is a key control point for this pathway. The concentration of AICAR would be expected to decrease in response to exogenous purines or histidine due to feedback inhibition of the biosynthetic pathways for these compounds. It is known that AICAR levels increase significantly in response to folate stress (ϳ3-6-fold) due to inhibition of the PurH enzyme (1,2). Genetic evidence suggests that the PurH enzyme is sensitive to high temperature (48). Considering the in vitro kinetic parameters and in vivo pool sizes, it seems likely that flux through gluconeogenesis may be altered under growth conditions that impact the concentration of AICAR exemplified by blocking PurH by mutation.
In summary, the work described here has identified three amino acid substitutions in FBP from Salmonella that cause insensitivity to both AMP and AICAR. An additional potential regulatory role for the small molecule AICAR has been identified, although further work is necessary to determine whether this effect is physiologically relevant. Finally, the use of purH mutants could provide a powerful genetic selection for AMPinsensitive mutants of FBP, including substitutions that might not be predicted based on structural information.