Antagonistic Roles for GcvA and GcvB in hdeAB Expression in Escherichia coli

In E. coli, the periplasmic proteins HdeA and HdeB have chaperone-like functions, suppressing aggregation of periplasmic proteins under acidic conditions. A microarray analysis of RNA isolated from an E. coli wild type and a ΔgcvB strain grown to mid-log phase in Luria-Bertani broth indicated the hdeAB operon, encoding the HdeA and HdeB proteins, is regulated by the sRNA GcvB. We wanted to verify that GcvB and its coregulator Hfq play a role in regulation of the hdeAB operon. In this study, we show that GcvB positively regulates hdeA::lacZ and hdeB::lacZ translational fusions in cells grown in Luria-Bertani broth and in glucose minimal media + glycine. Activation also requires the Hfq protein. Although many sRNAs dependent on Hfq regulate by an antisense mechanism, GcvB regulates hdeAB either directly or indirectly at the level of transcription. GcvA, the activator of gcvB, negatively regulates hdeAB at the level of transcription. Although expression of gcvB is dependent on GcvA, activation of hdeAB by GcvB occurs independently of GcvA's ability to repress the operon. Cell survival and growth at low pH are consistent with GcvA negatively regulating and GcvB positively regulating the hdeAB operon.


Introduction
Acid resistance is important for the ability of enteric bacteria to survive the low pH environment encountered in the gastrointestinal tract of mammalian hosts and other natural environments [1]. Enteric bacteria have five systems of acid resistance [2][3][4][5][6][7]. The first system, AR1, is least understood. When cells are grown in LB at pH 5 to stationary phase, they survive dilution into minimal medium at pH 2.5, which kills cells grown at pH 8. The stationary phase sigma factor RpoS and cyclic-AMP receptor protein are required to develop acid tolerance [2,6]. The other four systems, AR2 AR3, AR4, and AR5, are decarboxylate/antiporter-dependent acid resistance systems that require glutamate, arginine, lysine, and ornithine, respectively [2,[4][5][6][7][8][9][10]. Additional acid protection comes from the periplasmic proteins HdeA and HdeB that have chaperone-like functions, suppressing aggregation of periplasmic proteins under extreme acidic conditions [11][12][13]. Both hdeA and hdeB mutants show reduced viability upon acid stress and HdeA/HdeB expressing plasmids restore viability close to wild type, suggesting both proteins are necessary for protection of the bacterial periplasm against acid stress [14]. Regulation of the hdeAB operon is complex. The hdeAB operon in E. coli is acid inducible and regulation involves GadE, RpoD, RpoS, H-NS, MarA, and several other regulators [6,7,[15][16][17][18].
The E. coli gcvB gene encodes a sRNA of 206 nucleotides [19]. Expression of gcvB is activated by the GcvA protein when cellular glycine is high and repressed by GcvA and GcvR when glycine is limiting [19]. In both E. coli and Salmonella enterica serovar Typhimurium, GcvB regulates genes involved in the transport of small peptides and polar and branched amino acids [19][20][21][22][23][24]. Recently, it was shown GcvB enhances the ability of E. coli to survive low pH by upregulating RpoS [25]. In addition, microarray data suggested the hdeAB operon is positively regulated by GcvB [22]. Results from this study establish a role for GcvA in repressing the hdeAB operon and GcvB in activating the operon. Hfq, an RNA chaperone required for GcvB regulation of known target genes [20,22,23,26], is also required for activation. However, the results suggest GcvB and Hfq do not function as an antisense RNA system to upregulate hdeAB translation, 2 ISRN Microbiology but act at the level of transcription. The results also suggest GcvA, the activator for gcvB, negatively regulates hdeAB at the level of transcription.

Materials and Methods
2.1. Bacterial Strains, Plasmids, and Phage. The E. coli strains, plasmids, and phage used are listed in Table 1 or described in the text.

Construction of Recombinant Phages and Plasmids.
The λhdeA::lacZ translational fusion was constructed by PCR synthesis of a DNA fragment using an upstream primer with an EcoRI site that hybridized to DNA beginning 223 bps upstream of the hdeA transcription start site and a downstream primer with a SmaI site that hybridized to DNA beginning at codon 7 within the hdeA gene. The PCR amplified DNA fragment was digested with EcoRI + SmaI and the 303 bp EcoRI-SmaI fragment ligated into the EcoRI-SmaI sites of plasmid pMC1403 [37], fusing the first 7 codons of the hdeA gene in frame with the 8th codon of the lacZYA genes in pMC1403 (Figure 1(a)). The cloned sequence was verified by DNA sequence analysis at the DNA Core Facility of the University of Iowa. The plasmid was designated phdeA::lacZ. A 5,574 bp EcoRI-MfeI fragment from phdeA::lacZ carrying the hdeA::lacZYA fusion was then ligated into the EcoRI site of phage λgt2 [30], generating λhdeA::lacZ. A λhdeB::lacZ fusion was constructed using the same upstream primer and a downstream primer with a SmaI site that hybridized to DNA beginning at codon 9 within the hdeB gene. The 757 bp EcoRI-SmaI fragment was then used as described above, generating plasmid phdeB::lacZ and phage λhdeB::lacZ (not shown). A λhdeA::lacZ transcriptional fusion was constructed using the same upstream primer and a downstream primer with a HindIII site and that hybridized to DNA at bp −36 relative to the hdeA translation start site (Figure 1(a)). Following digestion with EcoRI and HindIII, the DNA fragment was ligated into the EcoRI and HindIII sites of plasmid pgcvB-lacZ +50 [19], replacing the gcvB fragment with the hdeA fragment, generating plasmid phdeA −36 ::lacZ. The cloned sequence was verified by DNA sequence analysis. A 5,538 bp EcoRI-MfeI fragment from phdeA −36 ::lacZ carrying the hdeA −36 ::lacZYA fusion was then ligated into the EcoRI site of phage λgt2 [30], generating λhdeA −36 ::lacZ. The 3 fusion phages were used to lysogenize E. coli host strains as described [38]. Each lysogen was tested to ensure it carried a single-copy of the λ chromosome by infection with λcI90c17 [39]. All lysogens were grown at 30 • C since all fusion phage carry the λcI857 mutation, resulting in a temperature-sensitive λcI repressor [30]. The λP BAD ::hdeA::lacZ fusion, where hdeA transcription is under control of the P BAD promoter, was constructed as described in Figure 1 Plasmid pGS611 (pgcvA 3+ ), carrying the E. coli gcvA gene on a 1,155 bp EcoRI fragment, was constructed as follows. In a PCR reaction, an upstream primer was used containing an EcoRI site and that hybridized to a region beginning 121 bp upstream of the gcvA transcription start site and a downstream primer containing an EcoRI site and hybridized to a region beginning 44 bp downstream of the gcvA translation stop codon. The EcoRI sites were added as parts of the primers. The PCR-generated fragment was digested with EcoRI and cloned into the EcoRI site in plasmid pACYC184 [40] and verified by DNA sequence analysis (Figure 1(c)). Plasmid pGS624 (pgcvA 3+ gcvB 3+ ), carrying both the gcvA and gcvB genes, was constructed in the same way except the upstream primer hybridized to DNA 51 bps after the gcvB transcription terminator and the downstream primer hybridized to DNA 44 bps after the gcvA translation stop codon, generating a 1,347 bp EcoRI fragment (Figure 1(c)).

Enzyme
Assay. β-galactosidase assays were performed on mid-log phase cells (OD 600 ∼ 0.5) using the chloroform/SDS lysis procedure [41]. Results are the averages of two or more assays with each sample done in triplicate.
2.6. Acid Sensitivity Assay. WT, an isogenic ΔgcvAB strain and the two strains transformed with either plasmid pgcvB 2+ (constitutively produces GcvB), pgcvA 3+ or pgcvA 3+ gcvB 3+ were grown for 24 hr at 30 • C in LB and then tested for acid resistance by dilution into LB at pH 2.0. Samples of 0.2 mL were taken at 0, 1, 2, and 4 hr and diluted in 2 mL of LB at pH 7. The final pH of the diluted cultures was ∼7.0. Cell viability was determined by plate counts. Percent survival is the titer of colony forming units of acid-tested cells compared to the zero-time point (Figure 2).

Transductions.
The gcvB gene is linked to the argA gene and hfq is linked to the cycA gene, with predicted phage P1 cotransduction frequencies of ∼78% and ∼67%, respectively. P1clr phage prepared on GS854 (argA81::Tn10) was used to transduce ΔgcvB::ΩCM R λhdeA::lacZ to TC R and transductants scored on CM versus TC plates. A TC R CM S transductant was purified. P1clr prepared on GS776 (cycA::Tn10) was used to transduce Δhfq-1::ΩCM R λhdeA::lacZ to TC R resistance and transductants scored on CM versus TC plates. A TC R CM S transductant was purified.

Results and Discussion
3.1. GcvA/GcvB Role in Acid Sensitivity. Microarray data suggested the hdeA and hdeB mRNAs are 1.9-and 2.7-fold higher in WT than a ΔgcvB strain grown in LB, respectively [22]. These genes were not reported to be regulated by GcvB in that study because they fell below the 3-fold cut-off level used for GcvB-regulated genes. Since HdeA and HdeB are necessary for protection of the bacterial periplasm against acid stress [14,17], we tested if GcvB plays a role in cellular acid resistance. WT and an isogenic ΔgcvAB strain were grown for 24 hr at 30 • C in LB and tested for acid resistance by dilution into LB at pH 2.0 [43]. The WT was killed significantly more readily at pH 2 than the ΔgcvAB strain ( Figure 2, compare black and gray lines). However, when the ΔgcvAB strain was transformed with pgcvB 2+ that constitutively expresses GcvB [19], we did not see complementation that restored acid sensitivity ( Figure 2, green line). When transformed with the multi-copy plasmid pgcvA 3+ gcvB 3+ , both the WT and the ΔgcvAB transformants were more acid sensitive ( Figure 2, compare the black and blue lines and the gray and purple lines). Plasmid pgcvA 3+ , which carries only the gcvA gene, also complemented the ΔgcvAB mutation, increasing acid sensitivity ( Figure 2, compare the gray and red lines). The results suggest it is the absence of GcvA that is responsible for increased acid resistance in the ΔgcvAB strain. It was reported previously that GcvB plays a positive role in acid resistance [25]. Our failure to observe a significant effect on acid resistance is possibly due to the assay conditions. We tested for acid resistance after 24 hours of growth in LB, whereas in the earlier study acid resistance was tested after 5 hr of growth in LB [25]. Although the precise stage of growth was not stated in the earlier study, it is possible cells were still in log phase. In E. coli and Salmonella grown in LB, GcvB was only detected through early stationary phase, with the highest levels observed at the mid-exponential phase [20,26]. Thus, GcvB regulation of target genes involved in acid resistance is likely during log phase and if GcvB plays a role in stationary phase, it is its absence that is important for allowing an appropriate regulatory response.

Effects of GcvB on λhdeA::lacZ Expression in LB Grown
Cells. Although GcvB had no effect in the acid sensitivity assay, we made and tested expression of λhdeA::lacZ and λhdeB::lacZ translational fusions. Expression of the hdeA::lacZ fusion was 2.7-and 4-fold higher in WT grown in LB compared to ΔgcvB and Δhfq strains (Figure 3 Figure 1: (a) The hdeAB control region. The hdeAB promoter −35, −10 and transcription start site and the SD sequence and translation start site are indicated above the sequence [31]. The transcription start site for the divergently transcribed hdeD gene is also shown [32]. Binding sites for H-NS [16] and MarR/SoxS [15] are indicated above the sequence with arrows. The binding site for GadX/W [33] is below the sequence in blue and for GadE above the sequence in green [18,34]. In addition, there are putative binding sites for the transcriptional regulators Lrp and TorR (not shown) [34]. The consensus GcvA binding site is T-N 11 -A containing a 5 -CTAAT-3 sequence [35]. Two putative GcvA binding sites are indicated in red. The fusion sites for the λhdeA::lacZ translational fusion, the λhdeA −36 ::lacZ transcriptional fusion, and the λP BAD ::hdeA::lacZ fusion (see below) are indicated with black, red, and green arrows, respectively. (b) Construction of a λP BAD ::hdeA::lacZ promoter fusion. The WT P BAD and P hdeA promoters are shown in the top line. The transcription start sites are in red [31,36]. Small case letters show bases added during PCR amplification of the P BAD and P hdeA promoters. The P BAD promoter was amplified with an upstream primer containing an EcoRI site at bp −272 relative to the transcription start site (not shown) and a downstream primer with a BsaI site (blue). The P hdeA promoter was amplified with an upstream primer containing a BsaI site (blue) and a downstream primer containing a SmaI site at codon 10 in the hdeA gene (not shown). The arrows indicate cut sites for BsaI. The amplified products were cut with BsaI, mixed, and ligated, generating a fusion of the P BAD promoter with the +1G residue of the P hdeA promoter. The fragment was then digested with EcoRI + SmaI and ligated into the EcoRI-SmaI sites of plasmid pMC1403, and subsequently subcloned into λgt2 as described [19]. (c) The gcvA gcvB region of the E. coli chromosome. The regions amplified by PCR and cloned into pACYC184 to generate pGS611 (pgcvA 3+ ) and pGS624 (pgcvA 3+ gcvB 3+ ) are indicated with bars. See Section 2.2 for details.

Reduced hdeA::lacZ Expression in ΔgcvB and Δhfq Strains
Is due to the Absence of GcvB and Hfq. Due to the failure of pgcvB + and phfq 3+ to fully complement the gcvB and hfq mutations (Figure 3), we wanted to verify the reduced levels of hdeA-lacZ expression are due to the absence of GcvB and Hfq. We transduced the ΔgcvB and Δhfq lysogens with WT alleles using linked Tn10 markers. The gcvB + and hfq + transductants showed about the same levels of expression as the WT lysogen (Figure 3(a), compare lanes 1, 12 and 13). Thus, despite the failure of pgcvB + and phfq 3+ to fully complement, the results support the reduced levels of expression are due to the absence of GcvB and Hfq.

Effects of GcvA on λhdeA::lacZ Expression in LB Grown
Cells. The acid sensitivity assay showed GcvA plays a role in acid resistance (Figure 2). In addition, putative GcvA binding sites can be identified in the hdeA promoter region (Figure 1(a)). Thus, we tested the effects of a spontaneous gcvA mutation in strain GS1198 (which is phenotypically GcvB − [19]), on hdeA::lacZ expression. Expression of hdeA::lacZ was ∼1.5-fold higher in WT than in the gcvA mutant ( Figure 3, lanes 1 and 6). However, expression was 2-fold higher in the gcvA lysogen than in the ΔgcvB lysogen (Figure 3(a), compare lanes 2 and 6). The results could be explained if GcvA, in addition to activating expression of gcvB, which encodes a positive regulator for hdeA, also has a negative role to keep HdeAB levels low. The intermediate level of expression would result from the absence of GcvB to upregulate the hdeA::lacZ fusion and the absence of GcvA to negatively regulate the fusion. To test this hypothesis, we transformed the gcvA mutant with multi-copy pgcvA 3+ . In the gcvA[pgcvA 3+ ] lysogen, one would expect high GcvA levels, but GcvB would also be produced. Expression of hdeA::lacZ was 2.5-fold higher in WT than in the gcvA[pgcvA 3+ ] transformant (Figure 3

Effect of GcvA, GcvB, and Hfq on hdeA::lacZ Expression in GM + Glycine.
In E. coli, GcvB represses dppA::lacZ, oppA::phoA, cycA::lacZ, and sstT::lacZ fusions when cells are grown in LB, but does not significantly repress these fusions when grown in GM + glycine [19,22,23]. However, gcvB is differentially regulated over a 25-fold range in GM supplemented with inosine versus glycine [19]. We hypothesize some genes respond to GcvB levels in GM media. Microarray data suggested the hdeA mRNA is 1.6fold higher in WT than a ΔgcvB strain grown in GM + glycine [22]. In GM + glycine, hdeA::lacZ expression was significantly higher than for cells grown in LB (Figure 3, compare a and b). In addition, although there are small differences in fold regulation for individual strains, there was a similar regulatory pattern in GM + glycine as observed in LB. The results suggest GcvB positively regulates hdeA in LB and GM + glycine. The results are important since they confirm GcvB does regulate in GM + glycine. In addition, acid resistance mechanisms are most active in the stationary phase in rich media [6,7,44,45]. Our results suggest GcvA and GcvB could play important roles in acid resistance during the log phase of growth in both rich and minimal medium.

Effect of GcvA, GcvB, and Hfq on hdeB::lacZ Expression.
The hdeB gene is the second gene in the hdeAB operon. We tested if hdeB is regulated in a manner similar to the hdeA gene. There were small differences in the levels of hdeB::lacZ expression in response to GcvB, GcvA, and Hfq compared to hdeA::lacZ in both LB and GM + glycine (compare Figures 3(a)

and 3(b) with Figures 3(c) and 3(d)
). Qualitatively, however, the λhdeB::lacZ fusion showed essentially an identical pattern of expression compared to the hdeA::lacZ fusion, suggesting both genes of the operon are regulated in a similar manner by GcvA, GcvB, and Hfq. It is worth noting that Δhfq lysogens consistently showed lower levels of hdeA::lacZ and hdeB::lacZ expression than ΔgcvB lysogens in both LB and GM + glycine (Figure 3). Two other sRNAs, DsrA, and GadY, are known to play roles in regulation of acid-resistance genes [46,47]. Since both sRNAs require Hfq, it is not surprising the absence of Hfq has a more dramatic effect on hdeAB expression than the absence of GcvB.

High Levels of GcvA, GcvB, and Hfq in WT Alters
hdeA::lacZ and hdeB::lacZ Expression. To verify GcvA negatively regulates and GcvB and Hfq positively regulate the hdeAB operon, we transformed WTλhdeA::lacZ, and WTλhdeB::lacZ lysogens with plasmids carrying gcvA, gcvB, both gcvA + gcvB, or hfq. We hypothesized high GcvB and Hfq would increase expression and high GcvA would repress expression. The lysogens were grown in LB and assayed for β-galactosidase. The presence of pgcvB 3+ resulted in a small increase in hdeA::lacZ expression and about a 2-fold increase in hdeB::lacZ (Figures 4(a) and 4(b), lines 1 and 2).   The lysogens were also grown in GM + glycine. The pattern of regulation was similar to the LB grown lysogens with one exception. The pgcvB 3+ transformant did not show increased expression of hdeB::lacZ as in LB (Figure 4(b), compare lines 1 and 2 with lines 6 and 7). It is possible that in WT grown in GM + glycine GcvB is already in excess for regulation. Nevertheless, the results are in agreement with GcvB and Hfq positively regulating the hdeAB operon and GcvA negatively regulating the operon.
To determine at what step in regulation of hdeA GcvA,    GcvB and Hfq function, we constructed a λP BAD ::hdeA::lacZ fusion where transcription from the P BAD promoter begins at the +1 start site of the hdeA gene (Figure 1(b)). We initially lysogenized a WT strain with the fusion, the lysogen was grown in LB + arabinose (0.0 to 0.2% concentrations) and assayed for β-galactosidase. There was a 379-fold induction (2.3 units versus 872 units of activity) at 0.0% and 0.05% arabinose, respectively. This is similar to the level observed from the λhdeA::lacZ lysogen grown in LB ( Figure 3) and confirmed the fusion is inducible by arabinose. We then lysogenized WT, ΔgcvB, ΔgcvAB, and Δhfq strains. The WT lysogen was also transformed with the plasmids indicated in Figure 5 As a complement to the above experiment, we constructed a transcriptional fusion of the hdeA promoter 36 bps upstream of the hdeA translation start site to a promoterless lacZYA operon (Figure 1(a)). Expression of hdeA −36 ::lacZ was 3.3-and 5.6-fold higher in WT grown in LB compared to ΔgcvB and Δhfq strains ( Figure 5(b), lanes 1-3) and activation was partially restored in the ΔgcvB[pgcvB + ] and Δhfq[phfq 3+ ] complemented strains ( Figure 5(b), compare lanes 2 and 4 and 3 and 5). The results show GcvB and Hfq still regulate the fusion. Expression in the WT was ∼2-fold higher than in the ΔgcvA and ΔgcvAB lysogens ( Figure 5(b), lanes 1, 6 and 7). In the ΔgcvA[pgcvA 3+ ] and ΔgcvAB[pgcvA 3+ ] lysogens, there was a 2.4-and 3.9-fold reduction of hdeA −36 ::lacZ expression compared to WT ( Figure 5(b), compare lanes 1, 8 and  9), showing GcvA negatively regulates the fusion. We also transformed the WT lysogen with plasmids phfq 3+ , pgcvA 3+ , and pgcvA 3+ gcvB 3+ . The presence of phfq 3+ resulted in a 1.4-fold increase in expression ( Figure 5 show GcvB and Hfq function during log phase to positively regulate hdeAB at the transcriptional level, counterbalancing the negative effect of GcvA on downregulating these genes. GcvB is known to bind Hfq [49]. It is possible GcvB binds to and sequesters Hfq during exponential growth, and the effects observed are due to decreased levels of Hfq to alter regulation of genes such as rpoS or the activity of sRNAs such as DsrA and GadY that play roles in acid resistance. Additional studies will verify if GcvA directly binds the hdeAB promoter region and how GcvB and Hfq activate the operon.

Effect of pH on gcvB Expression. Our results suggest
GcvB plays a role in acid resistance during log phase of growth in rich and minimal media. Therefore, we tested if pH plays a role in regulating gcvB expression. A WTλgcvB::lacZ fusion was grown to mid-log phase in LB at different pH values from 5.0 to 9.0 and assayed for β-galactosidase. There was no significant effect from pH 7 to pH 9 on gcvB::lacZ expression ( Figure 6). However, there was a 3-fold increase as the pH was lowered from pH 7 to pH 5 ( Figure 6). Since GcvB activates hdeAB, an increase in gcvB expression at low pH is likely to play a role in final HdeAB levels and in controlling acid resistance.

Effect of GcvA and GcvB on Cell Growth at Low pH.
We carried out studies to show the effects of high GcvA and GcvB levels on growth at low pH. In a ΔgcvAB strain transformation with pgcvA 3+ or pgcvA 3+ gcvB 3+ did not significantly alter generation times (GTs) in LB at pH 7 (Table 2). At pH 4.5, GTs of both the WT and ΔgcvB strains were significantly increased ( Table 2, compare rows 1 and 2, pH 7.0 versus pH 4.5). In addition, in the WT[pgcvA 3+ ] strain, with high GcvA and low GcvB, there was a significant increase in the GT compared to the non-transformed WT strain ( Table 2, compare rows 1 and 3, pH 4.5). In the gcvAB[pgcvA 3+ ] transformant, with high GcvA and no GcvB, there was an additional increase in the GT ( Table 2, compare rows 3 and 4, pH 4.5 column). In the WT[pgcvA 3+ gcvB 3+ ] and ΔgcvAB[pgcvA 3+ gcvB 3+ ] strains, with high GcvA and GcvB, the GTs were not significantly different than in the non-transformed strains ( Table 2, compare rows 1 and 2 with rows 5 and 6, pH 4.5). The results are consistent with GcvA negatively regulating acid resistance genes and GcvB overcoming the negative effect of GcvA. The results also show GcvA and GcvB affect acid resistance in log phase cells and could play important roles in the ability of enteric organisms to colonize the GI tract.

Role of GcvB in Cell
Physiology. In E. coli, GcvB negatively regulates SstT, CycA, OppA, and DppA levels, the serine transporter, glycine transporter and the oligopeptide, and dipeptide periplasmic binding proteins, respectively [19,22,23]. These proteins not only transport amino acids and peptides to provide nutrients, but possibly toxins and antibiotics [50,51]. If conditions that favor relatively high levels of amino acids and small peptides also favor the presence of small toxic compounds, the decreased expression of transport systems for these small molecules by GcvB could prevent transport of toxic compounds into the cell [22]. Our results show that GcvB also positively regulates genes involved in acid resistance. In addition, GcvA, the activator for gcvB expression, negatively regulates genes involved in acid resistance. These findings suggest GcvB and GcvA play important roles in the ability of E. coli to survive low pH conditions. Recently, in a screen of a sRNA gene knockout library, GcvB was shown to enhance E. coli survival at low pH [25]. Thus, GcvB likely allows E. coli to respond to and survive two stress conditions, the presence of toxic compounds and low pH environments. Both of these conditions are encountered as E. coli moves from an external environment into the GI tract. Understanding the biological roles of GcvB and GcvA in acid resistance and their mechanism(s) of regulation will provide insights as to how cells respond to environmental challenges to infect host organisms.