The Small RNA GcvB Regulates sstT mRNA Expression in Escherichia coli

Abstract

In Escherichia coli, the gcvB gene encodes a nontranslated RNA (referred to as GcvB) that regulates OppA and DppA, two periplasmic binding proteins for the oligopeptide and dipeptide transport systems. An additional regulatory target of GcvB, sstT, was found by microarray analysis of RNA isolated from a wild-type strain and a gcvB deletion strain grown to mid-log phase in Luria-Bertani broth. The SstT protein functions to transport l-serine and l-threonine by sodium transport into the cell. Reverse transcription-PCR and translational fusions confirmed that GcvB negatively regulates sstT mRNA levels in cells grown in Luria-Bertani broth. A series of transcriptional fusions identified a region of sstT mRNA upstream of the ribosome binding site needed for negative regulation by GcvB. Analysis of the GcvB RNA identified a sequence complementary to this region of the sstT mRNA. The region of GcvB complementary to sstT mRNA is the same region of GcvB identified to regulate the dppA and oppA mRNAs. Mutations predicted to disrupt base pairing between sstT mRNA and GcvB were made in gcvB, which resulted in the identification of a small region of GcvB necessary for negative regulation of sstT-lacZ. Additionally, the RNA chaperone protein Hfq was found to be necessary for GcvB to negatively regulate sstT-lacZ in Luria-Bertani broth and glucose minimal medium supplemented with glycine. The sstT mRNA is the first target found to be regulated by GcvB in glucose minimal medium supplemented with glycine.

The Escherichia coli chromosome encodes 50 to 100 small, nontranslated RNAs (sRNAs) (excluding tRNAs and 5S RNA) that function using diverse mechanisms under a variety of cellular conditions (2, 7, 36-39). A number of these sRNAs have been found to function as regulators in response to different growth conditions including various osmolarities, temperatures, oxidative stresses, and iron levels (37). A large class of sRNAs bind to the RNA chaperone protein Hfq (12, 28). Many of these sRNAs that bind Hfq regulate expression of target genes posttranscriptionally by base pairing with target mRNAs (12, 28). Base pairing between the sRNA and target mRNA can result in an alteration in the stability or translation of the message (7, 29, 37). It is unclear how extensive base pairing between an sRNA and a target mRNA must be, but research indicates that one or two regions of 8 to 9 bp are sufficient for regulation (7). Single mutations made in the target mRNA ompC in the region of complementarity with the sRNA MicC or with mutations made in the target mRNA ptsG in the region of complementarity with the sRNA SgrS interfered with regulation (5, 12). These sRNAs were found to regulate these target mRNAs by base pairing (5, 12). Other examples of sRNAs that bind Hfq and regulate target mRNAs by base pairing include MicF, DsrA, RprA, RyhB, DicF, OxyS, and Spot42 (5, 7, 8, 28). In most cases base pairing results in negative regulation of translation activity and decreased stability of the target mRNA (7).

The gcvB gene encodes an sRNA of 206 nucleotides that, in turn, regulates oppA and dppA mRNAs, which encode the oligopeptide and dipeptide periplasmic binding proteins, respectively (33). The sRNA RyhB is known to regulate at least 18 different target mRNAs (16), and many other sRNAs are also predicted to regulate more than one target. In addition, in Salmonella, GcvB has been shown to regulate livK, livJ, argT, gltI, dppA, and oppA mRNAs by direct interactions (24). Therefore, we predicted that GcvB in E. coli has additional regulatory targets other than the oppA and dppA mRNAs. We compared RNA isolated from a wild-type strain and a gcvB deletion strain grown to mid-log phase in Luria-Bertani (LB) broth by microarray analysis to identify any additional regulatory targets of GcvB. One potential target identified by microarray analysis was sstT, which encodes a Na⁺/l-serine and l-threonine transport protein. Similar to DppA and OppA, SstT also functions to transport nutrients into the cell (19) and is the principal serine transporter thought to be constitutively expressed in E. coli K-12 (13, 18). A GU-rich region in GcvB previously identified to base pair with dppA and oppA mRNAs (21) and shown to be necessary for regulation of seven target mRNAs in Salmonella (24) was found to be complementary with sstT mRNA upstream of the ribosome binding site. On the basis of the dppA and oppA regulation by GcvB in E. coli (21), we hypothesized that GcvB and mRNA pairing is also the regulatory mechanism employed by GcvB for negative regulation of sstT mRNA and that this regulation is dependent upon Hfq. This study shows that GcvB negatively regulates sstT mRNA in an Hfq-dependent mechanism in cells grown in both LB medium and defined glucose minimal medium.

MATERIALS AND METHODS

Bacterial strains, plasmids, and phage.

The E. coli strains, plasmids, and bacteriophages used in this study are listed in Table 1 or described in the text. Strains GS1148 and GS1149 were constructed by transduction. P1clr phage grown on the hfq-1::ΩCm^r strain GS081 was used to transduce strains GS162 and GS1132 to Cm^r, generating strains GS1148 and GS1149, respectively. Plasmid pGS609, carrying the E. coli hfq gene, was constructed as follows. In a PCR, HindIII primer 1 hybridized to a region beginning 992 bp upstream of the Hfq translation start site, and HindIII primer 2 hybridized to a region beginning 109 bp downstream of the Hfq translation stop codon. PCR amplification was carried out using Vent DNA polymerase (New England Biolabs, Inc., Beverly, MA). Cycling conditions were as follows: 1 min and 30 s of denaturation at 95°C, 1 min of annealing at 55°C, and 2 min of extension at 72°C for 30 cycles. The 1,409-bp PCR-generated HindIII fragment was cloned into the HindIII site in plasmid pACYC177 (4) and was verified by DNA sequence analysis at the DNA Core Facility of the University of Iowa.

TABLE 1.

Strains, plasmids, and bacteriophages

Strain, plasmid, or bacteriophage	Relevant genotypeb	Source or reference
Strainsa
GS081	MC4100 hfq-1::ΩCmr	G. Storz
GS162	Wild-type	This laboratory
GS1132	GS162 Δ(gcvA gcvB)::ΩaadA	33
GS1144	GS162 ΔgcvB::ΩCmr	This lab
GS1148	GS162 hfq-1::ΩCmr	This study
GS1149	GS162 Δ(gcvAgcvB)::ΩaadA hfq-1::ΩCmr	This study
Plasmids
pGS311	Single-copy-number vector	This laboratory
pGS341	Single-copy-number vector with wild-type gcvA	11
pGS571	Multicopy-number vector with wild-type gcvB	10
pGS594	Single-copy-number vector with wild-type gcvB	This laboratory
pGS595	pGS594 with a change in gcvB at bp +79 to +87 of TGTTGTGTT to ACAAgAgAA [gcvB(+79ACAAgAgAA)]; previously called gcvB Con-7	21
pGS596	pGS594 with a change in gcvB at +71 to +73 of TGT to CCC [gcvB(+71CCC)]	21
pGS597	pGS594 with a deletion in gcvB of G at bp +75 in [gcvB(+75ΔG)]	21
pGS602	pGS594 with a change in gcvB at bp +76 to +78 of TGT to AAA [gcvB(+76AAA)]	21
pGS603	pGS594 with a deletion in gcvB at bp +83 to +93 GTGTTTGCAAT [gcvB(+83Δ11)]	21
pGS605	pGS594 with a change in gcvB at bp +65 and +66 of TT to AA [gcvB(+65AA)]	21
pGS609	Multicopy-number vector carrying hfq	This study
pGS618	pGS594 with a change in gcvB at bp +73 of T to C [gcvB(+73C)]	21
pGS619	pGS594 with a change in gcvB at bp +71 and +72 of TG to CC [gcvB(+71CC)]	21
pGS625	pGS594 with a change -in gcvB at bp +79 of T to C [gcvB(+79C)]	This study
pGS626	pGS594 with a change in gcvB at bp +80 of G to C [gcvB(+80C)]	This study
pGS627	pGS594 with a change in gcvB at bp +81 of T to C [gcvB(+81C)]	This study
pGS628	pGS594 with a change in gcvB at bp +82 of T to A [gcvB(+82A)]	This study
pGS629	pGS594 with a change in gcvB at bp +79 to +82 of TGTT to CCCA [gcvB(+79CCCA)]	This study
pGS630	pGS594 with a change in gcvB at bp +83 to +87 of GTGTT to TCTCC [gcvB(+83UCUCC)]	This study
Bacteriophages
λsstT-lacZ	Lambda vector carrying wild-type sstT-lacZ translational fusion	This study
λsstT(−66)-lacZ	Lambda vector carrying sstT(−66)-lacZ transcriptional fusion	This study
λsstT(−40)-lacZ	Lambda vector carrying sstT(−40)-lacZ transcriptional fusion	This study
λsstT(−24)-lacZ	Lambda vector carrying sstT(−24)-lacZ transcriptional fusion	This study
λsstT(−18)-lacZ	Lambda vector carrying sstT(−18)-lacZ transcriptional fusion	This study
λsstT(−13)-lacZ	Lambda vector carrying sstT(−13)-lacZ transcriptional fusion	This study
λsstT(−11UUU)-lacZ	Wild-type λsstT-lacZ fusion with a change of AAA to UUU at bp −11 to −13 in sstT	This study
λsstT(−18AGAGG)-lacZ	Wild-type λsstT-lacZ fusion with a change of CACAA to AGAGG at bp −18 to −22 in sstT	This study
λsstT(−14UGUUcUcUU)-lacZ	Wild-type λsstT-lacZ fusion with a change of GTAACACAA to UGUUcUcUU at bp −14 to −22 in sstT	This study

^aAll strains also carry the pheA905 thi araD129 rpsL150 relA1 deoC1 flbB5301 ptsF25 rbsR mutations.

^bNumbering for gcvB mutations is based on the transcriptional initiation site as +1. Numbering for the sstT fusions and mutations is based on the A residue in the AUG translation initiation codon as +1, with bases upstream assigned negative values.

Five λsstT-lacZ transcriptional fusions were constructed by synthesis of DNA fragments with either BamHI or EcoRI sites at bp −386 upstream of the sstT translational start codon and five different fusion points with HindIII sites at bp −66, −40, −24, −18, and −13 upstream of the translation start codon. Following digestion with BamHI and HindIII or with EcoRI and HindIII, the DNA fragments were gel purified and ligated into the BamHI and HindIII or the EcoRI and HindIII sites of plasmid pgcvB(+50)-lacZ (where +50 is the nucleotide position of the fusion in gcvB) (33), replacing the gcvB fragment with the sstT fragments and generating intermediate plasmids psstT(−66)-lacZ, psstT(−40)-lacZ, psstT(−24)-lacZ, psstT(−18)-lacZ, and psstT(−13)-lacZ. An EcoRI-MfeI DNA fragment carrying each sstT-lacZ fusion and the lacYA genes was gel purified and ligated into the EcoRI site of phage λgt2. The phage isolates were designated λsstT(−66)-lacZ, λsstT(−40)-lacZ, λsstT(−24)-lacZ, λsstT(−18)-lacZ, and λsstT(−13)-lacZ (Table 1 and Fig. 1A). The λsstT-lacZ translational fusion phage was constructed in a similar manner except that the upstream primer contained an EcoRI site at bp −386, and the downstream primer contained an SmaI site at codon 9 within the sstT structural gene. Following EcoRI-SmaI digestion, the DNA fragment was gel purified and ligated into the EcoRI and SmaI sites of plasmid pMC1403 (3), fusing the sstT structural gene in frame with the eighth codon of the lacZYA genes. The intermediate plasmid was designated psstT-lacZ. The translational fusion was cloned into λgt2 as described above, and the resulting phage was designated λsstT-lacZ (Table 1 and Fig. 1A). All fusion sequences were verified by DNA sequence analysis at the DNA Core Facility of the University of Iowa. The phage were used to lysogenize various E. coli host strains as described previously (32). Each lysogen was tested to ensure that it carried a single copy of the λ chromosome by infection with λcI90c17 (25). All lysogens were grown at 30°C since all fusion phage carry the λcI857 mutation, resulting in a temperature-sensitive λcI repressor (20).

FIG. 1.

(A) The sstT gene sequence including the promoter −35, −10, and +1 start sites; the Shine-Dalgarno (SD); and the ATG translational start site (19). The sites of the translational fusion and five transcriptional fusions are indicated by arrows below the sequence. (B) Comparison of GcvB between nucleotides +60 and +100 with the sstT mRNA. The AUG translational start site for sstT mRNA is underlined. The Shine-Dalgarno sequence is overlined and labeled SD. A 13-base region of GcvB from nucleotide +79 to +91 conserved in GcvB homologs from the genera Escherichia, Salmonella, Yersinia, Haemophilus, Vibrio, Pasteurella, Shigella, Klebsiella, and Photorhabdus is underlined. Regions of complementarity are indicated with dots between the sequences. GU base pairs are indicated by a line. The transcriptional fusions sstT(−24)-lacZ, sstT(−18)-lacZ, and sstT(−13)-lacZ are indicated with red arrowheads. Changes made to GcvB are indicated below the sequence. A 1-nucleotide and an 11-nucleotide deletion are indicated with Δs. A 7-nucleotide random substitution in the conserved region, designated GcvB(+79ACAAgAgAA), is indicated in brown. An AAA mutation made to GcvB, designated GcvB(+76AAA), is shown in pink. Single mutations made to the 4-nucleotide region in the conserved region are shown in black below the brown sequence. These mutations include GcvB(+79C), GcvB(+80C), GcvB(+81C), and GcvB(+82A). A combined 4-nucleotide CCCA mutation is shown in dark green and is designated GcvB(+79CCCA). A 5-nucleotide UCUCC mutation is shown in light blue and is designated GcvB(+83UCUCC). Compensatory mutations in sstT to restore base pairing with mutants GcvB(+76AAA), GcvB(+83UCUCC), and GcvB(+79ACAAgAgAA) are shown above the sstT sequence in the corresponding colors. The A nucleotide of the translation initiation codon in sstT mRNA is designated +1. Mutations upstream of the translation initiation sites are defined as negative numbers, and mutations downstream of the translation initiation sites are defined with positive numbers.

Mutagenesis.

Using plasmid pGS571 (gcvB) and plasmid psstT-lacZ as templates, the PCR megaprimer mutagenesis procedure (23) was used to create nucleotide changes in the gcvB gene and sstT gene (Fig. 1B and Table 1). Mutations were verified by DNA sequence analysis at the DNA Core Facility of the University of Iowa. The gcvB mutations were subcloned into the single-copy-number plasmid pGS341, replacing the gcvA gene. The sstT mutations were cloned into λgt2, and the phage was used to lysogenize appropriate host strains as described previously (32).

Media.

The complex medium used was LB (17). Agar was added at 1.5% (wt/vol) to make solid medium. The minimal medium used was the salts of Vogel and Bonner (35) supplemented with 0.4% (wt/vol) glucose (glucose medium [GM]). Supplements were added at the following concentrations (in μg ml⁻¹): ampicillin (Amp), 50; chloramphenicol (Cm), 20; glycine, 300; inosine, 50; l-serine, 400.

DNA manipulation.

The procedures for plasmid DNA purification and restriction enzyme digestion were as described previously (22). PCR amplifications were carried out under standard reaction conditions using Vent DNA polymerase (New England BioLabs, Inc., Beverly, MA). Restriction enzymes and other DNA modifying enzymes were from New England Biolabs, Inc. (Beverly, MA).

Enzyme assays.

β-Galactosidase assays were performed on mid-log-phase cells (optical density at 600 nm of ∼ 0.5) using a chloroform-sodium dodecyl sulfate lysis procedure (17). Results are the averages of two or more assays, with each sample done in triplicate.

RNA isolation and Northern blot analysis.

Appropriate E. coli strains were grown in 5 ml of LB medium to an optical density at 600 nm of ∼ 0.5 and treated with a 1/5 volume of stop solution (95% ethanol-5% acidic phenol); cultures were centrifuged at 2,000 × g for 5 min, and cell pellets were frozen at −80°C. Total RNA was isolated by phenol extraction (15), and each RNA sample was DNase I treated for 1 h at 37°C. RNA was quantified using a NanoDrop ND-1000 Spectrophotometer. For Northern blot analysis, 10 μg of total RNA was run on an 8 M urea-8% polyacrylamide gel and electroblotted to a positively charged nylon membrane (Roche, Mannheim, Germany). The blot was hybridized with a DNA probe specific to nucleotides +1 to +134 of the E. coli gcvB gene labeled using a PCR-DIG Probe synthesis kit (Roche, Mannheim, Germany). Hybridization was performed at 42°C as described previously (6), and the membrane was exposed to film and imaged using a Fujifilm LAS-1000 camera and Intelligent Dark Box. Quantification of RNA was performed using the software Image Gauge, version 3.12. The membrane was subsequently stripped using 0.1% (wt/vol) sodium dodecyl sulfate-2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), heated to ∼95°C, and rehybridized using a digoxigenin-labeled DNA probe specific to 5S RNA from nucleotides +2 to +112. The relative amount of GcvB expressed from each mutant gcvB allele was determined by taking the ratio of the GcvB to 5S RNA detected for each sample.

Serine growth curve.

Wild-type (GS162), ΔgcvB (GS1144), and complemented ΔgcvB pgcvB⁺ (GS1144 [pGS594]) strains were grown overnight in 2 ml of GM. Overnight cultures were used to inoculate 5 ml of GM or 5 ml of GM plus l-serine (Amp was added to the transformed strain for both conditions). l-Serine was always made the day before a growth curve and added at 400 μg ml⁻¹ immediately prior to the growth curve experiment. Cell density was read every 30 min using a Klett-Summerson reader until cultures reached mid-log phase of growth. The growth curve was repeated three times, and the averages are reported.

RESULTS

GcvB represses sstT mRNA and an sstT-lacZ translational fusion in LB medium.

Microarray analysis of RNA isolated from a wild-type strain and an otherwise isogenic ΔgcvB strain grown in LB medium suggested that sstT mRNA is negatively regulated ∼8.5-fold by GcvB (data not shown). To initially confirm that sstT mRNA is negatively regulated by GcvB, we performed semiquantitative reverse transcription-PCR. We observed a higher level of DNA corresponding to the sstT gene fragment amplified from cDNA generated from total RNA isolated from the ΔgcvB strain than from either the wild-type or the complemented ΔgcvB strain (data not shown), indicating that GcvB negatively regulates sstT mRNA in cells grown in LB medium to mid-log phase of growth.

To further confirm that GcvB negatively regulates sstT mRNA, we made a translational fusion of sstT with lacZ (Fig. 1A), cloned this fusion into λgt2, and lysogenized the wild-type, ΔgcvB, and the complemented ΔgcvB pgcvB⁺ strains. These lysogenic strains were grown in LB medium (with Amp for the complemented strain) to the mid-log phase of growth and assayed for β-galactosidase activity. Expression of sstT-lacZ in the wild-type lysogen was ∼12-fold lower than expression in the ΔgcvB lysogen (Fig. 2A, lanes 1 and 2). Repression of sstT-lacZ was restored in the complemented ΔgcvB pgcvB⁺ lysogen (Fig. 2A, lane 3). These results confirm that sstT mRNA is an additional regulatory target of GcvB in wild-type cells grown in LB medium.

FIG. 2.

Effects of GcvB on sstT-lacZ translational and transcriptional fusions. (A) β-Galactosidase levels from the λsstT-lacZ translational fusion from the wild-type (GS162), ΔgcvB (GS1144), ΔgcvB pgcvB⁺ (GS1144 [pGS594]), or ΔgcvB strain complemented with mutated gcvB alleles shown in Fig. 1B. (B) β-Galactosidase activity levels from the transcriptional fusions λsstT(−66)-lacZ, λsstT(−40)-lacZ, λsstT(−24)-lacZ, λsstT(−18)-lacZ, and λsstT(−13)-lacZ in the wild-type (black), ΔgcvB (white), or ΔgcvB pgcvB⁺ (gray) strain. (C) β-Galactosidase levels from the λsstT-lacZ translational fusion from the wild-type, ΔgcvB, ΔgcvB pgcvB⁺, or ΔgcvB strain complemented with a mutated gcvB allele shown in Fig. 1B. (D) β-Galactosidase activity levels of λsstT-lacZ, λsstT(−18AGAGG)-lacZ, and λsstT(−14UGUUcUcUU)-lacZ translational fusions from either the wild-type, ΔgcvB, or the ΔgcvB lysogen complemented with pgcvB⁺, pgcvB(+83UCUCC), or pgcvB(+79ACAAgAgAA) alleles. (E) β-Galactosidase activity levels of the λsstT(−11UUU)-lacZ translational fusion from the wild-type, ΔgcvB, or the ΔgcvB lysogen complemented with pgcvB⁺ or pgcvB(+76AAA) alleles. (F) β-Galactosidase levels from the λsstT-lacZ translational fusion from the wild-type, Δhfq (GS1148), ΔgcvAB Δhfq (GS1149), and Δhfq phfq⁺⁺ (GS1148 [pGS609]) lysogens. Results are the averages of two or more assays, with each assay performed in triplicate.

GcvB regulates sstT-lacZ in cultures grown in GM with glycine.

GcvB has a negative role in controlling oppA-phoA and dppA-lacZ expression in cells grown in LB medium (33). However, when cells were grown in GM supplemented with glycine, GcvB showed no significant repression of either oppA-phoA or dppA-lacZ fusions, with 1.1- and 1.6-fold repression observed, respectively (33). Previous results found gcvB-lacZ to be expressed ∼8-fold higher in wild-type cells grown in GM with glycine than in GM alone, ∼25-fold higher than in wild-type cells grown in GM with inosine (33), and ∼4-fold higher than wild-type cells grown in GM with l-serine (data not shown), suggesting that GcvB is differentially regulated in different GM combinations. Additionally, by Northern blot analysis we observed GcvB RNA from wild-type cells grown GM with glycine but not from wild-type cells grown in GM or GM supplemented with inosine (S.C. Pulvermacher, L. T. Stauffer, and G. V. Stauffer, unpublished data). The amount of GcvB detected in wild-type cells grown in GM with glycine was ca. eightfold lower than wild-type cells grown in LB medium (Pulvermacher et al., unpublished). Because of the dramatic differences in GcvB expression observed among wild-type cells grown in these different media, we hypothesized that there are regulatory targets for GcvB in wild-type cells grown in GM with glycine. Therefore, we tested whether GcvB regulates the sstT-lacZ translational fusion in GM supplemented with glycine. The wild-type, ΔgcvB, and the complemented ΔgcvB pgcvB⁺ lysogens were grown in GM with glycine to the mid-log phase of growth and assayed for β-galactosidase activity. Expression of sstT-lacZ was approximately twofold lower in the wild-type lysogen than in the ΔgcvB lysogen, and repression of sstT-lacZ was restored in the complemented ΔgcvB pgcvB⁺ lysogen (Fig. 3A). The results suggest that although slight, GcvB is able to regulate sstT-lacZ in GM supplemented with glycine, but the range of regulation is considerably less than for cells grown in LB medium.

FIG. 3.

Effects of GcvB and Hfq on the sstT-lacZ translational fusion in defined medium. (A) β-Galactosidase levels from the λsstT-lacZ translational fusion in the wild-type (GS162), ΔgcvB (GS1144), Δhfq (GS1148), ΔgcvAB Δhfq (GS1149), ΔgcvB pgcvB⁺ (GS1144 [pGS594]), or Δhfq phfq⁺⁺ (GS1148 [pGS609]) strain grown in GM with glycine. (B) β-Galactosidase levels from the wild-type λsstT-lacZ translational lysogen grown in the indicated media. Results are the averages of two assays, with each assay performed in triplicate.

Since there are lower levels of gcvB expression in GM, GM supplemented with inosine, and GM with l-serine than in GM supplemented with glycine (33), we hypothesized the lower levels of GcvB under these conditions would result in higher levels of expression of the sstT-lacZ fusion. To test this hypothesis, the wild-type λsstT-lacZ lysogen was grown in GM, GM with inosine, GM with glycine, and GM with l-serine to the mid-log phase of growth and assayed for β-galactosidase activity. There was a small increase in sstT-lacZ expression in GM and GM supplemented with inosine compared to expression in GM with glycine (Fig. 3B). There was also higher expression of sstT-lacZ in GM supplemented with l-serine than in GM with glycine (Fig. 3B). Expression in GM plus l-serine was slightly lower than sstT-lacZ levels from GM or GM with inosine. Although the change is small, the results suggest that repression of gcvB does increase sstT-lacZ expression.

Region of sstT mRNA is important for negative regulation by GcvB.

GcvB has a region of complementarity with the dppA and oppA mRNAs, two negatively regulated targets of GcvB (21). We hypothesized that a region of complementarity would exist between GcvB and sstT mRNA, but we were unable to identify a region of significant complementarity using the TargetRNA program (30) and Clone Manager, version 6, software. Therefore, to determine which region of sstT mRNA is required for negative regulation by GcvB, we made a series of transcriptional fusions of sstT with lacZ (Fig. 1A). The fusions were cloned into λgt2, and the phages were used to lysogenize the wild-type, ΔgcvB, and complemented ΔgcvB pgcvB⁺ strains. The lysogens were grown in LB medium to the mid-log phase of growth and assayed for β-galactosidase activity. We did not observe any significant regulation by GcvB with the λsstT(−66)-lacZ, λsstT(−40)-lacZ, or λsstT(−24)-lacZ fusion (Fig. 2B). However, there were decreases of approximately four- and eightfold in β-galactosidase activity in the λsstT(−40)-lacZ and λsstT(−24)-lacZ lysogens, respectively, compared with the λsstT(−66)-lacZ lysogens. These results suggest that a region of the sstT mRNA between nucleotides −66 and −24 is important for a GcvB-independent mechanism of regulation. In contrast to the −66, −40, and −24 lysogens, there were decreases of 3- and a 3.5-fold in β-galactosidase activity from the λsstT(−18)-lacZ and λsstT (−13)-lacZ fusions, respectively, in both the wild-type and the complemented ΔgcvB pgcvB⁺ lysogens compared with the ΔgcvB lysogen (Fig. 2B). This suggests that the region of sstT mRNA between the λsstT(−24)-lacZ and λsstT(−13)-lacZ fusions is necessary for negative regulation by GcvB. We did not observe the 12-fold range of regulation observed using the translational fusion (Fig. 2A), suggesting that additional sequence from the sstT mRNA is needed for full GcvB regulation.

Effects of mutations in gcvB on sstT-lacZ expression in vivo.

Most analyses of base pairing between sRNAs and target mRNAs began by either creating a series of large deletions in the sRNA or mutating a large region of the sRNA to determine the region essential for regulation (1, 5, 12). Once a region was identified in the sRNA as being important for regulation, specific single nucleotide mutations were made and analyzed, often leading to loss of regulation of the respective target mRNA. For GcvB, a different approach was taken. A region of GcvB was identified as complementary to dppA and oppA (21). In addition, eight different single-copy-number plasmids were constructed carrying mutations in gcvB to analyze GcvB regulation of dppA and oppA mRNAs (21). These mutations were generated to disrupt the predicted region of complementarity between GcvB and dppA or oppA mRNA. Many of the mutations resulted in only minor alterations in GcvB's ability to regulate oppA-phoA and dppA-lacZ (21); however, the GcvB(+79ACAAgAgAA) (where uppercase letters indicate a 7-nucleotide substitution at nucleotide +79, and lowercase letters represent wild-type sequence) and GcvB(+71CCC) mutant sRNAs resulted in an inability to fully repress dppA-lacZ or oppA-phoA as the β-galactosidase levels were 4- and 4.5-fold higher, respectively, and the alkaline phosphatase levels were 2-fold higher with both mutant sRNAs than levels in controls (21). Additionally, the GcvB(+65AA) mutant sRNA was found to affect only regulation of dppA-lacZ, with β-galactosidase levels threefold higher than in controls (21). We tested if any of these gcvB mutant alleles failed to regulate the sstT-lacZ translational fusion. The ΔgcvB strain lysogenized with λsstT-lacZ was transformed with each mutant plasmid and assayed for β-galactosidase activity (Fig. 2A). Two GcvB mutants, GcvB(+79ACAAgAgAA) and GcvB(+76AAA), were unable to repress sstT-lacZ (Fig. 2A). The GcvB(+79ACAAgAgAA) mutant sRNA resulted in a complete loss of repression of sstT-lacZ, while the GcvB(+76AAA) mutant sRNA, previously shown to have no significant effect on regulation of dppA-lacZ and oppA-phoA (21), resulted in β-galactosidase levels fivefold higher than the wild-type λsstT-lacZ lysogen (Fig. 2A, compare lane 1 with lanes 4 and 7). We previously showed by Northern analysis that the mutations did not significantly alter the levels of GcvB (21), suggesting that the differences in regulation of sstT-lacZ are not a result of a difference in GcvB RNA levels.

Inspection of the sstT and GcvB sequences using results from both of the transcriptional fusions and the gcvB(+79ACAAgAgAA) and gcvB(+76AAA) alleles revealed a small region of complementarity upstream of the translational start site for sstT that was largely disrupted by the gcvB(+79ACAAgAgAA) mutation and partially altered by the gcvB(+76AAA) mutation (Fig. 1B). This region of the sstT mRNA also spans the λsstT(−18)-lacZ and λsstT(−13)-lacZ transcriptional fusion sites that identified the region as necessary for GcvB regulation of sstT-lacZ (Fig. 1 and 2B). To determine which bases in the gcvB(+79ACAAgAgAA) mutation are responsible for loss of repression, we constructed several new gcvB mutations. Because sRNA of the GcvB mutant with a deletion of 11 nucleotides at position +83 [GcvB(+83Δ11)] showed wild-type levels of repression of sstT-lacZ and partially overlaps nucleotides mutated in GcvB(+79ACAAgAgAA), we assumed that bp +84, +86, and +87 in gcvB that were deleted in the gcvB(+83Δ11) allele are not necessary for regulation of sstT-lacZ (Fig. 1B). To determine which of the remaining four nucleotides defined by the GcvB(+79ACAAgAgAA) mutant sRNA are necessary for regulating sstT mRNA, we mutated each nucleotide individually in gcvB, generating the plasmids pGS625 [gcvB(+79C)], pGS626 [gcvB(+80C)], pGS627 [gcvB(+81C)], and pGS628 [gcvB(+82A)], as well as mutating all four nucleotides together, generating the plasmid pGS629 [gcvB(+79CCCA)] (Table 1 and Fig. 1B). We also mutated the five nucleotides that overlap between the mutant sRNAs GcvB(+79ACAAgAgAA) and GcvB(+83Δ11), generating the plasmid pGS630 [gcvB(+83UCUCC)] (Table 1 and Fig. 1B). The ΔgcvB strain lysogenized with the λsstT-lacZ translational fusion was transformed with each of these six mutant plasmids, and the transformants were grown in LB medium with Amp to the mid-log phase of growth and assayed for β-galactosidase activity. Four mutant sRNAs, GcvB(+79C), GcvB(+80C), GcvB(+81C), and GcvB(+82A), were able to repress sstT-lacZ similarly to both the wild-type and ΔgcvB pgcvB⁺ complemented lysogens (Fig. 2C, compare lanes 1 and 3 with lanes 4 to 7). The ΔgcvB lysogen carrying the mutant sRNA GcvB(+79CCCA) showed ∼1.5-fold higher activity than the wild-type, while the ΔgcvB lysogen carrying the mutant sRNA GcvB(+83UCUCC) had fivefold higher β-galactosidase activity than either the wild-type or complemented ΔgcvB pgcvB⁺ lysogens (Fig. 2C, compare lanes 1 and 3 with lanes 8 and 9).

Effects of mutations made in gcvB on GcvB RNA production.

It is possible that the normal repression of sstT-lacZ observed for the mutant gcvB(+79C), gcvB(+80C), gcvB(+81C), and gcvB(+82A) alleles and the loss of repression of sstT-lacZ observed for the mutant gcvB(+79CCCA) and gcvB(+83UCUCC) alleles are results of overexpression or underexpression of the mutant RNAs, respectively. To determine if each of the newly generated gcvB mutant alleles produces comparable amounts of GcvB, Northern blotting was performed. The blot showed roughly equivalent amounts of GcvB detected for each of the eight RNA samples tested (Fig. 4). These results and the previously reported Northern blot analysis of other gcvB mutant alleles (21) suggest that the normal repression or loss of repression observed for sstT-lacZ is likely not due to changes in the synthesis or initial stability of GcvB but is a direct effect of GcvB's role in regulation of sstT mRNA.

FIG. 4.

Northern blot analysis of GcvB expressed from either wild-type cells or from plasmids carrying either a wild-type or mutant gcvB allele transformed in the ΔgcvB strain. Total cell RNA was isolated from each strain, run on an 8 M urea-8% polyacrylamide gel, and probed with either a digoxigenin-labeled GcvB or a 5S RNA-specific DNA probe. Total RNA extracts are as follows (strain and plasmid): wild-type (GS162), gcvB⁺ (GS1144 [pGS594]), gcvB(+79C) (GS1144 [pGS625]), gcvB(+80C) (GS1144 [pGS626]), gcvB(+81C) (GS1144 [pGS627]), gcvB(+82A) (GS1144 [pGS628]), gcvB(+79CCCA) (GS1144 [pGS629]), and gcvB(+83UCUCC) (GS1144 [pGS630]) (Fig. 1B and Table 1). RNA was quantified using the software Image Gauge, version 3.12. RNA from each mutant gcvB allele was compared to the GcvB RNA expressed from the wild-type strain.

Effects of mutations in sstT-lacZ that restore base pairing to mutant GcvB on GcvB repression.

The above results identified a region of GcvB complementary to sstT mRNA required for normal regulation. To determine whether GcvB regulates sstT mRNA by base pairing, a series of seven, five, or three nucleotides were changed in sstT-lacZ to restore base pairing with the gcvB(+79ACAAgAgAA), gcvB(+83UCUCC), and gcvB(+76AAA) mutant alleles (Fig. 1B). The changes UGUUcUcUU at nucleotides −14 to −22, AGAGG at nucleotides −18 to −22, and UUU at nucleotides −11 to −13 were made in sstT-lacZ (Fig. 1B). The fusions were cloned into λgt2, and the phages were used to lysogenize wild-type and ΔgcvB strains. The lysogens were grown in LB medium to the mid-log phase of growth and assayed for β-galactosidase activity. The wild-type strain lysogenized with λsstT(−18AGAGG)-lacZ produced β-galactosidase activity at a level similar to that of the wild-type strain lysogenized with λsstT-lacZ (Fig. 2D, compare lanes 1 and 3). However, the ΔgcvB λsstT(−18AGAGG)-lacZ lysogen showed β-galactosidase activity approximately twofold higher than that of the wild-type λsstT(−18AGAGG)-lacZ lysogen (Fig. 2D, compare lanes 3 and 4). The results suggest that the endogenous wild-type GcvB still regulates the sstT(−18AGAGG)-lacZ fusion over a twofold range, and the low level of expression from λsstT(−18AGAGG)-lacZ in the ΔgcvB strain is because the 5-nucleotide mutation to sstT affects the stability or translation of the fusion. We then tested to see if restoring base pairing between GcvB and λsstT(−18AGAGG)-lacZ would cause better repression by transforming the ΔgcvB λsstT(−18AGAGG)-lacZ lysogen with either pgcvB⁺ or pgcvB(+83UCUCC) and assaying for β-galactosidase activity. There was a modest decrease (∼1.2-fold) in β-galactosidase activity when the strain was transformed with pgcvB⁺ or pgcvB(+83UCUCC) (Fig. 2D, compare lanes 4, 5, and 6), suggesting that both the wild-type and mutated GcvB(+83UCUCC) strains regulate equally and that the modest decrease was not due to the restoration of base pairing.

We next tested if a larger number of nucleotides are necessary for base pair regulation between GcvB and sstT by lysogenizing the wild-type and ΔgcvB strains with λsstT(−14UGUUcUcUU)-lacZ and assaying for β-galactosidase activity. The wild-type λsstT(−14UGUUcUcUU)-lacZ lysogen produced ∼2.5-fold more β-galactosidase activity than the wild-type λsstT-lacZ lysogen, suggesting a loss of regulation by endogenous wild-type GcvB (Fig. 2D, compare lanes 1 and 7). The ΔgcvB λsstT(−14UGUUcUcUU)-lacZ lysogen produced approximately twofold higher levels of β-galactosidase than the wild-type λsstT(−14UGUUcUcUU)-lacZ lysogen (Fig. 2D, compare lanes 7 and 8). The low level of expression from λsstT(−14UGUUcUcUU)-lacZ in the ΔgcvB strain compared to the λsstT-lacZ lysogen is likely due to mutations in sstT that alter the stability or translation of the fusion (Fig. 2D, compare lanes 2 and 8). Despite lower overall levels of β-galactosidase activity, the ΔgcvB λsstT(−14UGUUcUcUU)-lacZ lysogen was transformed with pgcvB⁺ or pgcvB(+79ACAAgAgAA), and the transformants were assayed for β-galactosidase activity to see if restoration of base pairing was able to restore GcvB repression of sstT mRNA. When the ΔgcvB λsstT(−14UGUUcUcUU)-lacZ lysogen was transformed with pgcvB⁺ and pgcvB(+79ACAAgAgAA), there were 1.2- and 1.7-fold lower levels of β-galactosidase, respectively (Fig. 2D, compare lanes 8, 9, and 10).

Next, wild-type and ΔgcvB strains were lysogenized with λsstT(−11UUU)-lacZ, and the lysogens were assayed for β-galactosidase activity. The wild-type λsstT(−11UUU)-lacZ lysogen produced much lower β-galactosidase activity levels than previously assayed strains (Fig. 2E). This lower β-galactosidase activity is likely a result of the three mutated nucleotides in sstT that occur near the translational start site (Fig. 1B) and are predicted to alter the Shine-Dalgarno sequence as well as alter the secondary structure of sstT mRNA, allowing the Shine-Dalgarno sequence to possibly form part of a stem-loop structure (Fig. 5A and B). Interestingly, β-galactosidase levels from the ΔgcvB λsstT(−11UUU)-lacZ lysogen decreased approximately twofold compared with the wild-type λsstT(−11UUU)-lacZ lysogen (Fig. 2E). When ΔgcvB λsstT(−11UUU)-lacZ was transformed with pgcvB⁺, a further decrease in β-galactosidase activity was observed (Fig. 2E). When the ΔgcvB λsstT(−11UUU)-lacZ lysogen was transformed with pgcvB(+76AAA), which restored base pairing with the mutant sstT(−11UUU)-lacZ lysogen, β-galactosidase activity increased ∼3.8-fold compared to the ΔgcvB λsstT(−11UUU)-lacZ lysogen and ∼2-fold compared with the wild-type λsstT(−11UUU)-lacZ lysogen (Fig. 2E).

FIG. 5.

The predicted secondary structure of wild-type sstT (A) and mutated sstT(−11UUU) (B) mRNAs starting from the transcriptional start site of sstT 81 nucleotides upstream of the AUG translational start codon. Both secondary structures are based on Mfold predictions (42). Bases shown with a gray background indicate nucleotides predicted to base pair with GcvB (A) or GcvB(+76AAA) (B). The predicted ΔG values are −17.2 kcal mol⁻¹ for wild-type sstT and −21.6 kcal mol⁻¹ for mutated sstT(−11UUU).

Effects of inactivation of hfq or of both gcvB and hfq on λsstT-lacZ expression.

It has been reported that many sRNAs that use RNA pairing to regulate require the Hfq RNA chaperone protein (26, 40). We observed that a deletion of hfq results in reduced repression of dppA-lacZ and oppA-phoA in E. coli (Pulvermacher et al., unpublished). A similar loss of regulation of dppA-gfp was reported in an E. coli hfq mutant (31), and the deletion of hfq in Salmonella enterica serovar Typhimurium resulted in a loss of regulation of OppA (24). Additionally, microarray analysis comparing the wild-type with a Δhfq strain found that sstT mRNA was negatively regulated in the wild-type strain (9), suggesting that Hfq is necessary for regulation of sstT mRNA. Since the above results suggest that GcvB regulates sstT mRNA, we tested if Hfq is necessary for GcvB to regulate sstT mRNA. The wild-type, Δhfq, ΔgcvAB Δhfq, and the complemented Δhfq phfq⁺⁺ λsstT-lacZ lysogens were grown in LB medium to the mid-log phase of growth and assayed for β-galactosidase activity. β-Galactosidase activity expressed from sstT-lacZ in the wild-type lysogen was approximately eightfold lower than in the Δhfq lysogen, and repression was restored in the complemented Δhfq phfq⁺⁺ (multicopy) lysogen (Fig. 2F). Additionally, when both gcvB and hfq were inactivated, β-galactosidase activity was nearly 11-fold higher than in the wild-type λsstT-lacZ lysogen (Fig. 2F). These results confirm that Hfq is required for GcvB-mediated repression of sstT-lacZ in cells grown in LB medium.

We also tested if Hfq is required for GcvB regulation of sstT in GM supplemented with glycine. The wild-type, Δhfq, ΔgcvAB Δhfq, and complemented Δhfq phfq⁺⁺ lysogens were grown in GM with glycine to the mid-log phase of growth and assayed for β-galactosidase activity. β-Galactosidase activity in the wild-type lysogen was approximately twofold lower than in the Δhfq lysogen, and repression was restored in the complemented Δhfq phfq⁺⁺ lysogen (Fig. 3A). When both gcvB and hfq were inactivated, β-galactosidase activity from sstT-lacZ was approximately threefold higher than the levels produced from the wild-type lysogen (Fig. 3A). These results suggest Hfq is required for GcvB-mediated regulation of sstT mRNA in both LB medium and GM supplemented with glycine.

Effects on growth of wild-type and ΔgcvB strains in GM supplemented with serine.

We have shown that GcvB negatively regulates sstT mRNA, encoding a protein that transports l-serine into the cell. In addition, we have found that GcvB negatively regulates livJ mRNA, encoding a periplasmic binding protein that also transports l-serine into the cell (data not shown) (14). Growth of E. coli in GM is inhibited by the addition of a low concentration of serine (less than 1 mM) (18). Because two of the four known serine transport mechanisms are negatively regulated by GcvB, we hypothesized that a ΔgcvB strain would have increased l-serine transport and enhanced growth inhibition compared with either the wild-type strain or the complemented ΔgcvB pgcvB⁺ strain when grown in GM supplemented with l-serine. Initially we tested if l-serine alters the level of gcvB expression. A low level of gcvB expression might mimic a ΔgcvB allele. β-Galactosidase levels from the wild-type strain carrying a λgcvB-lacZ transcriptional fusion were approximately twofold higher when cells were grown in GM supplemented with l-serine than when cells were grown in GM or GM with inosine (data not shown), suggesting that transcription of GcvB is slightly induced in the presence of l-serine but not to the same extent as in GM with glycine. Growth curves were then performed for wild-type, ΔgcvB, and complemented ΔgcvB pgcvB⁺ strains grown in GM and GM supplemented with l-serine. There was no significant difference in the growth of the three stains in GM (Fig. 6A). However, the ΔgcvB strain had a longer lag and grew considerably more slowly than the wild-type and the ΔgcvB pgcvB⁺ strains in GM with l-serine (Fig. 6B).

FIG. 6.

The wild-type (GS162), ΔgcvB (GS1144), and complemented ΔgcvB pgcvB⁺ (GS1144 [pGS594]) strains were grown overnight in GM. Cultures were reinoculated in either 5 ml of GM (A) or 5 ml of GM with l-serine (B) (400 μg ml⁻¹; transformants also had Amp added), and bacterial growth was monitored every 30 min by using a Klett-Summerson reader. Growth curves were conducted three separate times, and the averages are presented here.

DISCUSSION

Microarray analysis identified the sstT mRNA as a putative regulatory target of GcvB in E. coli cells grown in LB medium. Reverse transcription-PCR and lacZ fusions verified that sstT mRNA is negatively regulated by GcvB in LB medium. SstT is a transport protein involved in transporting l-serine into cells (13, 18, 19). DppA and OppA, two other known targets of GcvB regulation in E. coli, are the periplasmic-binding protein components of the two major peptide transport systems that are repressed in cells grown in rich medium (33). Preliminary results suggest that the cycA gene, encoding the E. coli glycine transport system, is also negatively regulated by GcvB in LB medium (S. C. Pulvermacher and G. V. Stauffer, unpublished results). It is possible that in E. coli GcvB plays a general role in repressing transport systems for small peptides and amino acids when nutrients are plentiful. In Salmonella, GcvB also directly interacts with seven mRNAs that encode different periplasmic binding proteins of ABC transporters (24). However, in neither organism was any gene previously identified that was regulated by GcvB in a defined medium. Results with the sstT-lacZ translational fusion are the first case of a target of GcvB regulated in defined medium. Although regulation in GM with glycine is not as robust as in LB medium (∼2-fold versus ∼12-fold), the results suggest that GcvB plays a role in regulating serine transport in defined medium.

Transcriptional fusions at bp −66, −40, and −24 relative to the translational start site of sstT were totally insensitive to regulation by GcvB (Fig. 2B). However, a transcriptional fusion at −18 in sstT showed 3-fold negative regulation by GcvB, and a transcriptional fusion at −13 in sstT showed 3.5-fold negative regulation by GcvB, suggesting that this region is required for normal GcvB control of sstT mRNA (Fig. 2B). Additional sstT sequence not carried in any of the five transcriptional fusions must be necessary to observe the ∼12-fold regulation observed with the λsstT-lacZ translational fusion. An analysis of the GcvB sequence identified a short 7-nucleotide region complementary to the sstT mRNA overlapping the −18 and −13 transcriptional fusion sites (Fig. 1B). This is the same region of GcvB found to be complementary to, and necessary for, GcvB regulation of both the dppA and oppA mRNAs (21). The GcvB(+79ACAAgAgAA) mutant sRNA that resulted in a complete loss of GcvB regulation of sstT-lacZ disrupts a large portion of the predicted region of complementarity with sstT mRNA; this same GcvB mutant resulted in reduced repression of both dppA-lacZ and oppA-phoA (21). The GcvB(+76AAA) mutant sRNA resulted in a partial loss of GcvB regulation of sstT-lacZ and immediately flanks the 7-bp region and disrupts two additional nucleotides complementary to sstT mRNA near the Shine-Dalgarno sequence (Fig. 1B). Interestingly, the GcvB(+76AAA) mutant sRNA showed normal regulation of both dppA-lacZ and oppA-phoA (21). These results suggest that the nucleotides defined by these changes are involved in GcvB regulation of sstT mRNA. The GcvB(+83Δ11) mutant sRNA, which removes part of the 7-bp region of complementarity, as well as several nucleotides defined by the GcvB(+79ACAAgAgAA) mutant sRNA (Fig. 1B) had no affect on GcvB regulation of sstT-lacZ. The result suggests that nucleotides +76 to +82 that are changed by the GcvB(+79ACAAgAgAA) mutant sRNA but are not removed by the GcvB(+83Δ11) mutant sRNA are necessary for negative regulation of sstT mRNA. However, when nucleotides +79 to +82 were each changed individually, there was no effect on the repression of sstT-lacZ compared to wild-type GcvB (Fig. 2C, compare lane 3 with lanes 4 to 7). When all 4 nucleotides were changed together [pgcvB(+79CCCA)], there was a modest increase in sstT-lacZ expression compared to the wild-type (Fig. 2C, compare lanes 3 and 8). The other 5 nucleotides altered in GcvB(+79ACAAgAgAA) were mutated together in gcvB [pgcvB(+83UCUCC)] and resulted in a fivefold increase in β-galactosidase activity from sstT-lacZ compared to the wild-type (Fig. 2C, compare lanes 3 and 9). None of the various combinations of mutations in gcvB was able to produce the complete loss of regulation of sstT-lacZ observed for the GcvB(+79ACAAgAgAA) mutant sRNA. Even though Mfold predictions indicate that the gcvB(+79ACAAgAgAA) allele does not alter secondary structure, it is possible that in vivo the secondary structure is altered. If this occurs, the resulting loss of repression of sstT-lacZ would not depend on changed nucleotides of GcvB(+79ACAAgAgAA) no longer being able to base pair with sstT mRNA. Because GcvB(+79ACAAgAgAA) is still able to partially repress dppA-lacZ and oppA-phoA (21), we do not believe that the complete loss of regulation of sstT-lacZ by this mutant allele is due exclusively to potential changes in the secondary structure. We tested strains transformed with each plasmid carrying a mutant gcvB allele by Northern analysis to verify that equal amounts of GcvB RNA were produced. Since essentially equal amounts of GcvB were produced in each transformant, it is unlikely that changes in GcvB levels are responsible for any changes observed in sstT-lacZ expression. It seems that a large region of gcvB needs to be altered in order to produce a change in the ability of GcvB to regulate sstT-lacZ. The results do not rule out the possibility that the changes in gcvB alter regulation of an additional factor necessary for sstT regulation. Results with the transcriptional fusions suggest that an additional GcvB-independent factor is required for full sstT expression (Fig. 2B).

An alternative reason for the loss of repression of sstT-lacZ with the GcvB(+79ACAAgAgAA) mutant sRNA and not the GcvB(+83Δ11) mutant sRNA is that, despite the deletion of 11 nucleotides, the GcvB(+83Δ11) mutant sRNA remains GU rich while GcvB(+79ACAAgAgAA) has much of the GU-rich region changed to CA rich (Fig. 1B). This GU region has been shown to be important for regulation in Salmonella (24). It is possible that the GcvB(+83Δ11) mutant sRNA is still able to base pair with and regulate sstT. This GU-rich region of GcvB also becomes more CA rich in the GcvB(+83UCUCC) mutant sRNA, possibly explaining why a loss in regulation of sstT-lacZ is observed even though this region was not originally predicted to have an effect based on the results of the GcvB(+83Δ11) mutant sRNA. We did not observe the full loss of regulation of sstT-lacZ by GcvB(+83UCUCC) compared with GcvB(+79ACAAgAgAA), likely because fewer interactions are predicted to be disrupted by GcvB(+83UCUCC).

We constructed sstT(−14UGUUcUcUU)-lacZ and sstT(−18AGAGG)-lacZ fusions to test if base pairing restored the ability of the mutant gcvB alleles to repress. Restoration of base pairing of sstT(−14UGUUcUcUU)-lacZ mRNA with the GcvB(+79ACAAgAgAA) sRNA allowed better repression than GcvB (Fig. 2D, compare lane 8 with lanes 9 and 10). However, restoration of base pairing between sstT(−18AGAGG)-lacZ and GcvB(+83UCUCC) did not increase the ability of GcvB to repress (Fig. 2D, compare lane 4 with lanes 5 and 6). Similarly, when compensatory mutations were made in dppA-lacZ or oppA-phoA to restore base pairing with GcvB mutants that failed to fully repress, in most but not all cases we observed better repression when base pairing was restored (21). Our results show that a large number of nucleotides need to be disrupted between GcvB and sstT mRNA before a loss of regulation is observed. Thus, GcvB differs in its ability to regulate target mRNAs compared with the sRNA SgrS and the target mRNA ptsG or the sRNA OxyS and the target mRNA fhlA, where a loss of regulation is observed with single nucleotide mutations made in either the sRNA or the target mRNA (1, 12). When compensatory mutations were made to restore base pairing between SgrS and ptsG or between OxyS and fhlA, regulation was fully restored (1, 12). Because multiple nucleotides need to be mutated in GcvB to observe a loss in repression of sstT-lacZ, the results suggest that there may be a level of redundancy in the GcvB sequence, allowing regulation to be maintained despite mutations in GcvB.

The sstT(−11UUU)-lacZ fusion has three nucleotides changed at the translational start site (Fig. 1B). These changes either destabilize sstT mRNA or reduce the translational efficiency as the β-galactosidase activity levels in the wild-type λsstT(−11UUU)-lacZ lysogen are only ∼2 units, and activity in the wild-type λsstT-lacZ lysogen is ∼40 units (Fig. 2D and E). Mfold predictions suggest that the secondary structure of sstT mRNA containing these three changes is altered compared with wild-type sstT, sstT(−14UGUUcUcUU), or sstT(−18AGAGG) mRNAs. Part of the Shine-Dalgarno sequence for sstT(−11UUU)-lacZ now potentially base pairs while the Shine-Dalgarno sequence for wild-type and other mutant alleles is predicted to be unpaired (Fig. 5 and data not shown). Interestingly, when complementary base pairing was restored between sstT(−11UUU)-lacZ and GcvB(+76AAA) in the ΔgcvB lysogen, we observed higher levels of β-galactosidase activity than in the ΔgcvB λsstT(−11UUU)-lacZ lysogen. When the pgcvB⁺ or pgcvB(+65AA) alleles, which do not restore complementarity, were transformed into the ΔgcvB λsstT(−11UUU)-lacZ lysogen, we observed a decrease in β-galactosidase activity compared with the ΔgcvB λsstT(−11UUU)-lacZ lysogen (Fig. 2E and data not shown). The results suggest that complementary base pairing between GcvB(+76AAA) and sstT(−11UUU)-lacZ is important for the positive regulation of sstT(−11UUU)-lacZ. Even though sstT(−11UUU)-lacZ is an artificial construct, it is unclear why we now observe positive regulation by GcvB. Although base pairing between GcvB(+76AAA) and sstT(−11UUU)-lacZ would disrupt the secondary structure that potentially sequesters the Shine-Dalgarno sequence, one would predict that this base pairing would also hinder ribosomes from binding to the mRNA (Fig. 5B). Further analysis needs to be performed to determine if other circumstances lead to positive regulation of sstT by GcvB. The results raise the possibility that GcvB both represses and activates certain genes under different environmental conditions. If changes in temperature, osmolarity, pH, etc., change the secondary structures of mRNA transcripts such as sstT, the changes could allow GcvB to assume a different regulatory role.

Previous results showed that sRNAs that regulate their target mRNAs by base pairing often require the RNA chaperone protein Hfq for regulation. Additionally, sstT mRNA was differentially regulated in the wild type compared to a Δhfq strain (9), suggesting that Hfq is involved in sstT regulation. However, Hfq was not shown to directly bind sstT mRNA (41). We found that Hfq is necessary for negative regulation of sstT-lacZ in cells grown in LB medium and GM supplemented with glycine. In addition, we observed an increase in β-galactosidase activity from the ΔgcvAB Δhfq double deletion strain compared to the ΔgcvB and Δhfq single deletion strains grown in LB medium and GM with glycine (Fig. 2F and 3A). The increase in β-galactosidase activity was not additive, suggesting that GcvB and Hfq likely regulate sstT-lacZ by the same regulatory mechanism. However, the increased β-galactosidase activity in the ΔgcvAB Δhfq double deletion suggests that an additional factor possibly plays a minor role in regulating sstT-lacZ. Although the ΔgcvAB Δhfq double deletion also removes the gcvA gene, we observed that β-galactosidase expression in the ΔgcvAB λsstT-lacZ lysogen is similar to that of the ΔgcvB lysogen (data not shown). Thus, the increased activity is not due to the ΔgcvA mutation.

Because sstT mRNA is negatively regulated by GcvB, we hypothesized that loss of GcvB would increase l-serine transport in E. coli. The addition of l-serine to GM is known to temporarily inhibit growth in E. coli (34) due to an elevated rate of serine catabolism to pyruvate (27). High pyruvate levels cause acetohydroxy acid synthase isozyme I to synthesize almost entirely 2-acetolactate (on the pathway to valine-leucine) and synthesis of only 1 to 2% 2-aceto-2-hydroxybutyrate (on the pathway to isoleucine), leading to isoleucine restriction in the cell (27). Additionally, homoserine dehydrogenase I, an enzyme involved in threonine and isoleucine synthesis, has also been reported to be inhibited by l-serine (27). The inhibition of this enzyme in isoleucine synthesis would also contribute to l-serine sensitivity (27). We predicted that increased serine transport due to a gcvB deletion would enhance growth inhibition in GM with l-serine. As shown in Fig. 6, we observed a longer lag and a slower growth rate for the ΔgcvB strain than for both the wild-type and the complemented ΔgcvB pgcvB⁺ strains. This result is consistent with our finding that GcvB is involved in negatively regulating sstT-lacZ expression in GM supplemented with glycine in E. coli (Fig. 3). Since multiple amino acid transport systems, as well as the di- and oligopeptide transport systems, are either predicted to be or have been shown to be regulated by GcvB (21, 33) (Pulvermacher and Stauffer, unpublished data), it is likely that additional growth phenotypes occur in an E. coli ΔgcvB strain grown in GM supplemented with different amino acids and peptides. Transport systems often play roles in the regulation of gene expression (27); therefore, regulation of different transport systems is essential. Our results suggest that GcvB is likely a global regulator in E. coli for genes encoding proteins for the transport of these small nutrient molecules into the cell in both rich medium and in GM.