INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

There are two pathways for the production of one-carbon (C₁) units in Escherichia coli. Serine hydroxymethyltransferase, the glyA gene product, catalyzes the cleavage of serine to glycine and the transfer of a C₁ unit to tetrahydrofolate to form 5,10-methylenetetrahydrofolate and is the primary source of C₁ units (24, 26). The glycine cleavage (GCV) enzyme system catalyzes the oxidative cleavage of glycine to form CO₂, NH₃, and 5,10-methylenetetrahydrofolate, providing a secondary pathway for C₁ units (19). The C₁ units produced by these pathways are used in cellular biosyntheses of methylated products such as methionine, thymine, and purines (26). It has been proposed that the physiological role of the GCV system may be to balance a cell's need for glycine and C₁ units.

The GCV enzyme system is composed of the GcvT, GcvH, and GcvP proteins, encoded by the gcv operon, and lipoamide dehydrogenase, encoded by the unlinked lpd gene. The regulation of the gcv operon is not fully understood, but there are four proteins known to affect gcv expression. The leucine-responsive protein, Lrp, is a global regulator of genes involved in amino acid metabolism (5) and is required for activation of the gcv operon (22, 41). The PurR protein is a negative regulator of nucleotide metabolic genes (14, 20, 31) and mediates a twofold repression of a gcvT-lacZ fusion when cells are grown in the presence of the purine nucleoside inosine (44). The GcvA protein is responsible for controlling gcv operon expression in two distinct ways. GcvA activates gcv expression when cells are grown in the presence of glycine and mediates a PurR-independent repression of gcv when cells are grown in the presence of inosine but without glycine (44, 45). A fourth protein, GcvR, is a GcvA-dependent negative regulator of gcv expression (12). However, GcvR has not been shown to bind to DNA, and its mechanism of regulation is unknown. Here we report a fifth protein that is involved in controlling gcv expression. The cyclic AMP (cAMP) receptor protein (CRP) mediates a fourfold positive effect on gcv expression as measured from a gcvT-lacZ fusion. In vitro binding experiments and a mutational analysis suggest that CRP binds to a site centered at bp 313 relative to the transcriptional start site for gcv. In addition, the CRP effect is dependent on a functional gcvA gene and its role may be to antagonize GcvA's repression of the gcv operon.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, phages, and plasmids. All strains used are listed in Table 1 and were constructed by P1 clr transduction (25). The gcvA-lacZ (46) and gcvR-lacZ (13) fusion phages were described previously. The gcvA-lacZ +15G phage carries a base pair change at position +15 relative to the transcription start site that results in a loss of GcvA-mediated autoregulation and about a sevenfold increase in gcvA-lacZ expression (15a). The gcvT-lacZ phage (39) used in earlier studies includes 466 bp upstream of the transcriptional start site. A derivative, gcvT-lacZ₃₄₁, was constructed in this study and extends upstream only to bp 341; this 125-bp deletion does not alter regulation. Strains were lysogenized with lambda phages as previously described (42). Other gcvT-lacZ phages carrying mutations in the gcv control region were constructed during this investigation and are described below. Plasmids used are listed in Table 1 or were constructed during this investigation. Plasmids pYZcrp, p19A, p52N, p158A, and p162C were gifts from R. Ebright.

Media. Glucose minimal (GM) medium or glycerol minimal medium was Vogel and Bonner minimal salts (43) supplemented with 0.4% glucose or 0.4% glycerol, respectively. Supplements were added at the following concentrations: phenylalanine, 50 µg/ml; inosine, 50 µg/ml; thiamine, 1 µg/ml; glycine, 300 µg/ml; ampicillin, 30 µg/ml for single-copy plasmids and 100 µg/ml for all other Ap^r plasmids; chloramphenicol, 40 µg/ml; and kanamycin, 20 µg/ml. GM and glycerol minimal media were always supplemented with phenylalanine and thiamine since all strains used carry the pheA905 and thi mutations.

DNA manipulations. Isolation of plasmid DNA, restriction enzyme digestions, ligations, and plasmid transformations were performed as described previously (32).

Enzyme assays. -Galactosidase assays were performed by the method of Miller (25), by using the chloroform-sodium dodecyl sulfate lysis procedure. All results are the averages of results from two or more assays, with each reaction being performed in triplicate.

Site-directed mutagenesis and construction of lysogens. Starting with plasmid pGS239 as the template, bp 139 and 152 relative to the +1 transcription initiation site were changed to an A and a T, respectively (Fig. 1), by the PCR megaprimer mutagenesis method (33). The new plasmid was designated pGS484. Starting with plasmid pGS362 as the template, bp 306, 307, and 308 relative to the transcription initiation site were changed to a T, G, and T, respectively (Fig. 1). The new plasmid was designated pGS485. The specific base pair changes were verified by DNA sequence analysis. The approximately 5,400-bp EcoRI-MfeI fragment carrying each mutant gcvT-lacZ fusion along with the lacY and lacA genes was isolated from each plasmid and ligated into the EcoRI site of phage gt2 (28). The phages generated were single plaque purified and designated gcvT-lacZ139A152T and gcvT-lacZ₃₄₁306T307G308T. The extensions after each fusion indicate the nucleotide changes and positions relative to the +1 transcription initiation site. Appropriate strains were lysogenized with the above-described phages, and the lysogens were verified to carry a single copy of  by infection with phage cI90c17 (38).

View larger version (12K): [in this window] [in a new window]   FIG. 1.   CRP binding sites in the gcv control region. The transcription start site for gcvT is indicated as +1. The nucleotide sequences of the CRP binding sites centered at bp 313 and 140 relative to the transcription initiation site are shown. The inverted repeat sequence known to be important for CRP binding is in capital letters. Nucleotides conserved with respect to the CRP consensus site are underlined. The arrows indicate the nucleotide changes in the mutants gcvT-lacZ341306T307G308T and gcvT-lacZ139A152T. The consensus CRP binding site is indicated for comparison.

CRP. The purified CRP used in the DNA mobility shift and DNase I footprinting assays was a gift from E. P. Greenberg.

Gel mobility shift assay. The gel mobility shift assay used was based on the methods described by Fried and Crothers (9) and Garner and Revzin (10). A 759-bp EcoRI-BamHI fragment from pGS239 and a 606-bp EcoRI-BamHI fragment from pGS258 were ³²P labeled at the EcoRI ends with T4 polynucleotide kinase (32). Samples of less than 22 ng of the labeled DNA fragments were included in 20-µl reaction mixtures containing DNA binding buffer (10 mM Tris HCl [pH 7.5], 50 mM KCl, 0.5 mM EDTA, 5% glycerol, 1 mM dithiothreitol), 125 µg of bovine serum albumin per ml, and cAMP as indicated in the figures. Reaction mixtures were incubated for 5 min at 37°C, and 2 µl of purified CRP diluted in DNA binding buffer was added to the mixtures as indicated in Fig. 2 and 3. Incubation was continued for 15 min at 37°C, the reactions were stopped by the addition of 1 µl of loading buffer (0.1% xylene cyanol and 50% glycerol in H₂O), and the samples were loaded on a 5% polyacrylamide gel and run at approximately 12 V/cm. The gels were transferred to Whatman 3MM paper, dried, and autoradiographed.

View larger version (40K): [in this window] [in a new window]   FIG. 2.   Gel mobility shift assay for the binding of CRP to gcv DNA. The wild-type 759-bp gcv fragment was used as target for lanes 1 to 5. The 5'-end-truncated 606-bp fragment was used as the target for lanes 6 to 10. Where indicated, 20 mM cAMP was included. The CRP dimer was added at a concentration of either 10 nM (+) or 100 nM (++).

View larger version (76K): [in this window] [in a new window]   FIG. 3.   Gel mobility shift assay for the binding of CRP to gcv DNA. The wild-type 759-bp gcv fragment was used as the target. The CRP dimer was added at the following concentrations: 0, 2.5, 5.0, 10, 25, 50, and 100 nM (lanes 1 to 7, respectively). cAMP was included in all reaction mixtures at a final concentration of 2 mM. The arrow denotes the unbound fragment.

DNase I protection assay. The DNase I protection assay was a modified version of the method of Schmitz and Galas (34) as previously described (47). The 759-bp ³²P-labeled fragment used in the gel mobility shift assay was used in the DNase I footprint assay. Less than 44 ng of labeled DNA was added to 18-µl reaction mixtures containing DNA binding buffer, 125 µg of bovine serum albumin, and 2 mM cAMP. The reaction mixtures were incubated for 5 min at 37°C, 2-µl samples of serial dilutions of CRP were added to the mixtures, and incubation continued at 37°C for 15 min. A 2-µl sample of a DNase I solution (0.1 U of DNase I per µl in 20 mM ammonium acetate-32 mM CaCl₂) was added for 30 s, reactions were stopped with the addition of 5 µl of stop solution (3 M ammonium acetate, 0.17 M EDTA, 33 µg of sheared calf thymus DNA per ml), and the samples were precipitated with ethanol. The DNA pellets were resuspended in DNA sequence loading buffer (0.1 M NaOH, 5 M urea, 1 mM EDTA, 0.05% xylene cyanol-bromophenol blue) and loaded onto a 5% polyacrylamide-7 M urea sequencing gel alongside the Maxam and Gilbert (23) A+G and C+T sequencing reaction mixtures loaded on the same labeled fragment.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

CRP involvement in gcvT-lacZ expression. Previous results showed that a deletion that ends at bp 313 upstream of the gcvT-lacZ transcription initiation site results in reduced levels of expression of the fusion (41). An analysis of the DNA sequence in this region identified a possible CRP binding site, with 14 of 22 bp matching the consensus binding sequence (11) (Fig. 1). Therefore, we tested if the CRP protein plays a role in regulation of the gcv operon. The wild-type strain and the crp deletion strain were lysogenized with the gcvT-lacZ₃₄₁ phage, and the lysogens were grown in GM medium with the appropriate supplements and assayed for -galactosidase activity. The crp deletion caused more than a fourfold decrease in -galactosidase levels compared to the level in the control strain when cells were grown in GM medium (Table 2). In contrast, the crp deletion had only a 1.5-fold effect on gcvT-lacZ expression when the cells were grown in GM medium supplemented with glycine and no significant effect when the GM medium was supplemented with inosine.

341

CRP requires cAMP to regulate gcv expression. Since CRP does not bind specifically to DNA in the absence of cAMP (21), we tested whether CRP's regulation of gcvT-lacZ requires cAMP. Strain GS1079 carries the (cyaA1400)::Kn^r allele and is defective in the production of cAMP (36). This strain was lysogenized with gcvT-lacZ₃₄₁ phage, the lysogen was grown in GM medium, and -galactosidase activity was measured. The -galactosidase level was not significantly different from the level measured in the crp deletion strain (Table 2). Since both the crp deletion and the cyaA deletion caused about a fourfold decrease in gcvT-lacZ expression compared to the level of expression in the wild-type strain when it was grown in GM medium, we wanted to confirm that the cAMP effect was mediated through CRP. Thus, we constructed a crp cyaA double mutant. This strain was lysogenized with the gcvT-lacZ₃₄₁ phage, the lysogen was grown in GM medium, and -galactosidase activity was measured. The -galactosidase levels were not significantly different from the levels measured in either the crp deletion lysogen or the cyaA deletion lysogen (Table 2).

341

CRP binds to the gcv control region. To test whether the putative CRP site centered at bp 313 can be bound by CRP in vitro, gel mobility shift assays were performed with purified CRP and two different DNA templates. One template was the 759-bp EcoRI-BamHI fragment carrying wild-type gcv DNA extending from bp 466 to +293 relative to the gcv transcriptional start site. The second template was a 606-bp EcoRI-BamHI fragment carrying gcv DNA extending from bp 313 to +293. This 606-bp fragment lacks half of the potential CRP binding site centered near bp 313 (Fig. 1). CRP dimer at a concentration of 10 nM and cAMP at a concentration of 20 mM resulted in a shift of the 759-bp wild-type fragment to a single band of slower mobility (Fig. 2, compare lanes 1 and 2). Binding of CRP to DNA at this concentration was dependent on the presence of cAMP (Fig. 2, compare lanes 2 and 4). At 100 nM CRP dimer, all of the wild-type DNA fragment was shifted in the presence and absence of cAMP, probably the result of nonspecific binding by CRP. The truncated DNA fragment did not show a specific band shift at 10 nM CRP in the presence or absence of cAMP (Fig. 2, lanes 7 and 9). However, at 100 nM CRP dimer all of the truncated template shifted in the presence and absence of cAMP (Fig. 2, lanes 8 and 10). These results suggest that CRP, in the presence of cAMP, binds specifically to the wild-type DNA template but not when the DNA fragment lacks half of the putative CRP binding site.

Location(s) of the CRP binding site(s) in the gcv control region. DNase I footprinting assays were performed to determine where CRP binds in the gcv control region (see Materials and Methods). As the CRP concentration was increased from 5 to 100 nM, two regions were protected from DNase I cleavage; one site centered near bp 313 as expected and the other site centered at bp 140 (Fig. 4). The protected region centered near bp 313 extends over about 27 bp, from bp 299 to 326 relative to the transcription initiation site, and was designated site 1. This protected region contains 14 bp that match base pairs in the 22-bp CRP consensus binding site (Fig. 1) (11). The second CRP-protected site, designated site 2, extends from about bp 131 to 158 and contains 12 bp that match base pairs in the 22-bp consensus CRP binding site (Fig. 1). Site 1 has a two- to fourfold higher affinity for CRP than site 2 (Fig. 4), likely due to the higher degree of sequence conservation in site 1. 

View larger version (84K): [in this window] [in a new window]   FIG. 4.   Protection from DNase I digestion of gcv DNA by CRP plus cAMP. The 32P-labeled wild-type 759-bp gcv fragment was incubated with dilutions of CRP and digested with DNase I (see Materials and Methods). cAMP (2 mM) was included in all reaction mixtures. The digestion products were electrophoresed on a denaturing 5% polyacrylamide gel adjacent to the Maxam-Gilbert sequencing reaction mixtures of the labeled DNA probe (not shown). (A and B) Long and short runs, respectively, of the digestion products. Lane 1, no protein; lanes 2 to 6, 5, 10, 25, 50, and 100 nM CRP dimer, respectively. The brackets indicate the two sites protected from digestion by DNase I.

Genetic analysis of the CRP binding sites. Although the DNase I footprint analysis identified two binding sites for CRP, the results from the gel mobility shift assay suggested that CRP binds to a single site in the gcv control region (Fig. 2 and 3). To determine whether one or both sites were required for CRP-mediated activation of the gcvT-lacZ fusion, we carried out a genetic analysis of the two binding sites. A triple mutation (306T307G308T) was created in CRP binding site 1, in the downstream half of the inverted repeat known to be important for CRP binding (11) (Fig. 1). A double mutation (139A152T) was created in CRP binding site 2, with one change being in each half of the inverted repeat (Fig. 1). CRP was unable to bind and protect these two mutated binding sites from DNase I digestion (data not shown). gcvT-lacZ phage carrying the 306T307G308T and the 139A152T mutations were used to lysogenize the wild-type strain and the crp deletion strain. The lysogens were grown in GM medium, and -galactosidase levels were determined. The 306T307G308T triple mutation in site 1 caused about a twofold decrease of gcvT-lacZ expression in the wild-type strain (Table 4). In the crp deletion strain these changes did not cause a further decrease in gcvT-lacZ expression compared to that of the wild-type strain. These results suggest that the mutations in binding site 1 eliminated CRP's regulatory role in controlling gcvT-lacZ expression. The 139A152T double mutation in binding site 2 decreased gcvT-lacZ expression 1.6-fold in the wild-type strain (Table 4). However, the crp deletion caused a further 2.6-fold decrease in expression (Table 4), suggesting that binding site 2 has no significant role in controlling gcvT-lacZ expression in vivo. The small decrease in -galactosidase levels observed with the 139A152T double mutation is possibly due to an alteration in Lrp-mediated regulation of gcvT-lacZ, as the changes are within the Lrp binding region (41). CRP binding to site 2 observed in vitro is likely due to the sequence similarity between the region and the consensus CRP binding site.

CRP's role is dependent on the GcvA protein. Of the many CRP-regulated promoters, the binding sites for CRP vary in their distances from the transcriptional start site and can often be correlated to CRP's mode of regulation (21). A distant upstream binding site, such as CRP binding site 1 for gcv, often indicates that CRP regulates in conjunction with another regulatory protein. To determine whether regulation is dependent on GcvA, a gcvA crp double mutant was constructed. This strain was lysogenized with the gcvT-lacZ₃₄₁ fusion, the lysogen was grown in GM medium, and the -galactosidase level was measured. The crp deletion caused approximately a threefold decrease in expression when the wild-type gcvA gene was present (Table 5). However, in the gcvA mutant background, a deletion of the crp gene had no effect on gcvT-lacZ expression. These results suggest that CRP is dependent on GcvA for its regulatory role and that the lower level of expression in the crp strain than that in the crp gcvA strain was due to GcvA.

CRP regulates expression of gcvA but does not regulate gcvR. When CRP-mediated regulation is dependent on a second protein, CRP often regulates expression of the gene encoding the second regulatory protein (21). A possible explanation for CRP's dependence on GcvA is that CRP regulates expression of the gcvA gene and indirectly affects expression of gcvT-lacZ. In addition, GcvA is known to require the gcvR gene product for its role as a repressor (12), raising the possibilities that CRP also regulates expression of gcvR and indirectly alters gcvT-lacZ expression. These possibilities were tested by lysogenizing the wild-type strain and the crp deletion strain with the gcvA-lacZ (46) and gcvR-lacZ (13) fusions. The lysogens were grown in GM medium, and the cells were assayed for -galactosidase activity. The deletion of the crp gene had no effect on gcvR-lacZ expression but caused a twofold decrease in expression of the gcvA-lacZ fusion (Table 5).

gcvA crp

gcvA crp

CRP requires the repressor function of GcvA to regulate gcvT-lacZ expression. GcvA activates and represses expression of gcvT-lacZ, and it is likely that both functions of GcvA are responsible for the basal levels of gcvT-lacZ expression in GM medium. Due to the dual action of GcvA, it is possible that CRP interferes with repression by GcvA or facilitates activation by GcvA, since both scenarios can explain the phenotypes of the crp and the crp gcvA strains. To distinguish between these two possible roles for CRP, the positive-control (PC) gcvA mutant gcvAF31A (16) was used to separate the activation and repression functions of GcvA. In a purR gcvA strain, a single-copy plasmid carrying the gcvA PC allele allows binding to the gcv control region and repression by GcvAF31A but not activation, and in a purR gcvA gcvR strain, GcvAF31A has virtually no activity since it cannot activate efficiently due to the amino acid change and cannot repress in the absence of GcvR (16). If CRP-mediated regulation is dependent on the activator function of GcvA, a crp deletion would be expected to have no effect if the gcvAF31A allele is the only gcvA present in the cell. Conversely, if CRP's role is to interfere with repression by GcvA, then CRP would be expected to regulate normally when the repressor function is intact but would have no role when GcvAF31A cannot repress (in a gcvR strain). A crp deletion resulted in a fourfold decrease in gcvT-lacZ expression in a purR gcvA strain carrying the gcvAF31A allele (little activation) (Table 7). However, in the purR gcvA gcvR strain carrying the gcvAF31A allele (little activation and no repression), the crp deletion had no effect (Table 7), suggesting that CRP's role in the regulation of gcvT-lacZ is dependent on GcvA's ability to repress.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study CRP was shown to be a positive regulator of the gcv operon. A deletion of the crp gene resulted in a three- to fourfold decrease in -galactosidase expression from a gcvT-lacZ fusion that was relieved by the introduction of a single-copy plasmid bearing the wild-type crp gene. A deletion of the cyaA gene also resulted in a fourfold decrease in -galactosidase levels (Table 2), indicating that CRP requires cAMP for regulation of the gcvT-lacZ fusion. Although the addition of cAMP to GM medium restored CRP-mediated regulation in the cyaA mutant (Table 3), its addition resulted in no further increase in gcvT-lacZ expression in a wild-type strain, indicating that the cAMP level in glucose-grown cultures is sufficient for CRP-mediated regulation of gcv. In an early study of CRP-cAMP binding to DNA, a fragment with the CRP consensus site was shown to bind CRP with an increased affinity compared to that of naturally occurring CRP binding sites, even in the absence of high levels of cAMP (11). However, the CRP binding site on gcv diverges from consensus at many positions (Fig. 1) and it is not clear how CRP can achieve its maximum range of regulation of the gcv operon at the cAMP levels found in GM medium-grown cultures.

CRP binding to the gcv control region protected two sites from DNase I digestion, one centered near bp 313 (site 1) and the other centered at bp 140 (site 2) relative to the transcriptional start site (Fig. 4). A mutational analysis of the two binding sites demonstrated that CRP binding is required only at site 1 to effect regulation of gcvT-lacZ (Table 4), consistent with the results of the gel mobility shift assay that showed only a single shifted species that required an intact site 1 (Fig. 2 and 3). We cannot explain why CRP binds to site 2 in the footprinting assay and not in the gel mobility shift assay. It should be noted that site 2 is totally within the region shown previously to be protected from DNase I digestion by the Lrp protein (41), and we believe this site is probably not accessible to CRP in vivo due to Lrp binding to this sequence.

CRP-mediated regulation from site 1 centered at bp 313 is interesting; in other CRP-regulated genes, the binding sites are located from bp 40 to 200 relative to the transcription start sites (21). CRP binding at 313 does not appear to activate any upstream promoters since S1 nuclease mapping experiments and a genetic analysis of the gcv control region did not reveal any additional promoters (40). Although our results indicate that CRP's role is to inhibit repression by GcvA at the gcv promoter, it is still possible that direct contact occurs between CRP and RNAP via DNA bending at gcv. This possibility may be unlikely since four crp mutants, each with an amino acid change in a known activating region, were able to complement the crp deletion and restore gcvT-lacZ expression to near the wild-type level.

It was demonstrated in several other systems such as the ara and mal regulons that CRP regulates specific promoters in conjunction with another regulator and, in addition, regulates the expression of the coregulatory protein itself (21). CRP-mediated regulation of gcvT-lacZ is dependent on the GcvA protein, and CRP stimulates expression of a gcvA-lacZ fusion about twofold. However, the reduced level of GcvA in a crp deletion strain does not appear to be responsible for the reduced expression of gcvT-lacZ. The results from this study are consistent with a model where CRP's role in the regulation of gcv is to interfere with repression by GcvA, rather than to activate transcription via interactions with RNAP.

There is no clear definition of antirepression, although several modes of regulation have been described as antirepressive (15, 17, 35, 37). In our system, antirepression is characterized by the antirepressor (CRP) having no function in the absence of the repressor (GcvA). There are two similar examples of antirepression that have been described for E. coli. In the first example, the global regulator integration host factor binds to the aceBAK operon and appears to inhibit repression by IclR, but no mechanism for this antirepression has yet been characterized (29). The second example of antirepression occurs at the pap genes in E. coli. A single CRP binding site is located 115.5 and 215.5 bp upstream of the divergent transcription start sites for the papI and papB genes, respectively, and CRP positively regulates expression of these genes by inhibiting binding of the repressing H-NS protein (8). It is difficult to visualize CRP utilizing this mechanism at the gcv operon since CRP's binding site does not overlap the GcvA binding sites. How then can CRP binding at site 1 antagonize GcvA-mediated repression? Evidence indicates a requirement not only for GcvA and its three binding sites but also for both GcvR and Lrp (12, 13, 41, 44, 47), suggesting that a nucleoprotein complex may form with these regulatory components to cause repression (Fig. 5). The role of CRP, then, may be to antagonize the formation or function of the complex. Since repression by GcvA, GcvR, and Lrp is poorly understood, further elucidation of the roles of these proteins in the regulatory mechanism is necessary for further investigation into CRP's role in controlling the gcv operon.

View larger version (14K): [in this window] [in a new window]   FIG. 5.   Hypothetical model for repression of gcv. There is evidence that GcvA must bind to its three target sites for repression (47). GcvR is also required for repression (12, 13), but it is unknown if GcvR-GcvA contacts are required or if GcvR performs some other function. Lrp binds gcv DNA in the region depicted and bends DNA (29), possibly to allow the formation of a nucleoprotein complex.