Small RNA binding-site multiplicity involved in translational regulation of a polycistronic mRNA

Abstract

In animal systems, mRNAs subject to posttranscriptional regulation by small RNAs (sRNAs) often possess multiple binding sites with imperfect complementarity to a given sRNA. In contrast, small RNA–mRNA interactions in bacteria and plants typically involve a single binding site. In a previous study, we demonstrated that the Escherichia coli sRNA SgrS base pairs with a site in the coding region of the first gene of a polycistronic message, manXYZ. This interaction was shown to be responsible for translational repression of manX and to contribute to destabilization of the manXYZ mRNA. In the current study, we report that translational repression of the manY and manZ genes by SgrS requires a second binding site located in the manX–manY intergenic region. Pairing at this site can repress translation of manY and manZ even when mRNA degradation is blocked. Base pairing between SgrS and the manX site does not affect translation of manY or manZ. Pairing at both sites is required for optimal SgrS-mediated degradation of the full-length manXYZ mRNA and for a particular stress phenotype. These results suggest that bacterial sRNAs may use target-site multiplicity to enhance the efficiency and stringency of regulation. Moreover, use of multiple binding sites may be particularly important for coordinating regulation of multiple genes encoded in operons.

Now accepted as fundamentally important players in regulation of gene expression, regulatory RNAs are present in organisms across all domains of life. In many eukaryotic organisms, siRNAs and microRNAs (miRNAs) control gene expression at the posttranscriptional level in diverse pathways (1–3). In bacteria, small RNAs (sRNAs) similarly control gene expression posttranscriptionally, and the principles (if not the details) of sRNA regulatory mechanisms have much in common with miRNA and siRNA regulation. Like miRNAs, many bacterial sRNAs function by base pairing with mRNA targets to affect their translation and stability. Both miRNAs and bacterial sRNAs usually regulate multiple mRNAs through interactions involving short regions (7–10 bases) of imperfect complementarity. In most instances, base pairing between sRNAs or miRNAs and their mRNA targets negatively affects target expression by modulating translation and mRNA stability (2, 4). Bacterial sRNAs usually repress target mRNA translation by base pairing with and sequestering sequences of the ribosome-binding site (RBS) in the 5′ UTR of the mRNA, making it unavailable for ribosome binding. Subsequent mRNA degradation is initiated by the endoribonuclease RNase E and its associated proteins, collectively referred to as the “degradosome” (4).

SgrS is a well-studied sRNA found in enteric bacteria and is expressed in response to a metabolic stress known as “glucose-phosphate” (GP) stress (5, 6). GP stress is a condition associated with imbalanced glycolytic flux resulting in the accumulation of sugar phosphates. The stress occurs when certain phosphosugars, including early glycolytic intermediates or nonmetabolizable phospho-glucose analogs such as α-methyl glucoside-6-phosphate (αMG6P) or 2-deoxyglucose-6-phosphate (2DG6P), accumulate in the cytoplasm (7–11). The activity of SgrS allows cells to cope with stress and continue growing, whereas mutants lacking sgrS are severely growth inhibited under stress conditions (6, 10). SgrS possesses two separate functions that can contribute independently to recovery from stress (5, 12). The first function is base pairing with several mRNA targets (13–15), including ptsG and manXYZ, which encode sugar transporters of the phosphoenolpyruvate phosphotransferase system. SgrS inhibits translation and promotes degradation of these mRNAs (9, 10, 13). This base pairing-dependent regulation stops new synthesis of sugar transporters, and this inhibition is hypothesized to alleviate stress by reducing intracellular accumulation of nonmetabolizable and potentially toxic phosphosugars. The second function of SgrS is encoding the small protein SgrT, which prevents glucose uptake through PtsG transporters (15).

The molecular mechanisms by which sRNAs regulate their target mRNAs are of considerable interest, and studies continue to identify new intricacies of translational and stability control. SgrS-mediated repression of ptsG mRNA proceeds via what now is considered a canonical mechanism for sRNA repressors. SgrS base pairs with sequences of the ptsG mRNA RBS and directly inhibits ribosome binding. This translational repression is concomitant with but does not require RNase E-mediated degradation (9, 14). We recently described SgrS-dependent regulation of manXYZ mRNA, a polycistronic message encoding a broad-substrate sugar transporter. We established that SgrS base pairs with sequences in the coding region of manX to inhibit translation. The SgrS–manX interaction stimulates degradation of the manXYZ message by the RNase E degradosome; however, as with ptsG, translational repression of manX does not require degradation (9). We hypothesized that manY and manZ translation would be regulated indirectly by SgrS as a result of the SgrS-dependent degradation of manXYZ mRNA.

In this study, we describe the surprising finding that SgrS base pairs with sequences at a second site on the manXYZ polycistronic mRNA. The second base-pairing interaction involves sequences in the manX–manY intergenic region, and this interaction allows SgrS to repress translation of manY by a mechanism that is independent of manX translational regulation. Furthermore, we show that, although pairing at each site is sufficient for translational repression of each gene individually, interactions at both sites are required for SgrS-dependent mRNA degradation and maximal recovery from stress, implying that the interactions somehow synergize to enhance the efficiency of regulation of this target. Binding-site multiplicity, that is, the interaction of an sRNA at multiple sites on a single mRNA target, also is used by miRNAs in animals. This study therefore reveals an example of the evolution of analogous mechanisms of RNA-based regulation in very divergent species.

Results

SgrS Repression of manY and manZ Translation Is Independent of manXYZ mRNA Degradation.

Our earlier study demonstrated that base pairing between the manX mRNA coding sequence and SgrS represses manX translation and stimulates degradation of full-length manXYZ mRNA (9). Repression of manX translation is not dependent on manXYZ mRNA degradation (9). The manY gene is located 62 nt downstream of the manX stop codon and has its own putative RBS, whereas manZ is located only 3 nt downstream of the manY stop codon and is not preceded by a recognizable RBS (Fig. 1A, Upper). The distance between manX and manY would not allow translational regulation of manY via manX–manY translational coupling (16, 17), suggesting that the SgrS–manX interaction would not inhibit manY or manZ translation directly. Instead, we hypothesized that SgrS-dependent regulation of manY and manZ would require SgrS-dependent manXYZ mRNA degradation (Fig. 1A, Lower). To test this hypothesis, translational lacZ reporter fusions to manY and manZ were constructed at the native chromosomal manXYZ locus. These are referred to as “manXY′-′lacZ” and “manXYZ′-′lacZ,” respectively (Fig. 1B). When SgrS production was induced (ectopically, from a plasmid-borne P_lac-sgrS), translation of manY and manZ reporter fusions was repressed by approximately fourfold and by ∼2.5-fold, respectively (Fig. 1C). To test whether degradation was required for SgrS-mediated regulation of manY or manZ, reporter fusions were constructed in wild-type and degradosome-deficient (rne131 allele) backgrounds, and activities of the fusions were measured in the presence and absence of SgrS. The rne131 allele encodes a C-terminally truncated RNase E that does not allow degradosome assembly (18). We showed previously that in an rne131 mutant manX translation is repressed by SgrS even though manXYZ mRNA is not degraded (9). Our hypothesis predicted that manY and manZ would not be regulated by SgrS in the absence of manXYZ mRNA degradation in the rne131 host. Surprisingly, however, SgrS still repressed manY and manZ translation in the degradosome-deficient background. Specifically, manY translation was repressed by SgrS approximately fourfold in wild-type cells, compared with approximately threefold in the rne131 background, whereas manZ translation was repressed approximately threefold in wild-type cells compared with approximately twofold in the rne131 background (Fig. 1D). These results suggest that SgrS-dependent regulation of manY and manZ is not dependent on degradosome assembly (Fig. 1D) and, importantly, does not require manXYZ mRNA degradation (9).

Fig. 1.

Regulation of manY and manZ translation by SgrS. (A) (Upper) The manXYZ operon is organized so that 62 nt separate the stop codon of manX and the start codon of manY, and 3 nt separate the stop codon of manY and the start codon of manZ. Ribosomes are represented by stacked blue ovals. Ribosome-binding sites are indicated by red boxes labeled “RBS” upstream of manX and manY. (Lower) When exposed to the nonmetabolizable glucose analogs αMG or 2DG, SgrS is produced, base pairs with the coding sequence of manX, and promotes degradation of the manXYZ transcript (9). We hypothesized that SgrS-dependent degradation of manXYZ mRNA was responsible for manY and manZ regulation. (B) Chromosomal lacZ translational fusions were constructed at the native manXYZ locus. (C) manXY'-'lacZ (JH136) and manXYZ'-'lacZ (JH130) carrying an empty vector or P_lac-sgrS were grown to early log phase, and IPTG was added. Samples were harvested 60 min after IPTG addition and assayed for β-galactosidase activity. Specific activities were normalized to the levels in the strain carrying the vector control to yield percent relative activity (reported in the graph). Specific activity values in Miller units are reported below the graph. (D) Activities of manXY'-'lacZ, manXYZ'-'lacZ, and manX_GUGY'-'lacZ in strains with an empty vector or plasmid-borne P_lac-sgrS were analyzed in wild-type or rne131 backgrounds. IPTG induction, sample harvesting, normalized specific activities, and Miller units are reported as in C.

SgrS Regulation of manYZ Is Independent of manX Regulation.

The results described above could be explained if regulation of manX translation caused polarity on manY and manZ, for example, if translational repression at manX caused premature transcription termination in the operon. To test whether manY translational regulation is dependent on SgrS base pairing at the manX site, a mutation in manX that prevents this pairing (9) was made in the manY reporter fusion (manX_GUGY′-′lacZ, Fig. 1D). When SgrS was expressed, the manX_GUGY′-′lacZ fusion was repressed by 4.5-fold compared with the vector-only control (Fig. 1D). This result strongly suggests that SgrS regulation of manY translation occurs by a mechanism independent of base pairing and translational repression at manX.

The manXYZ mRNA is processed in a degradosome-dependent manner, forming a manYZ species (9). To assess the role of the processing in SgrS-dependent regulation of manY, we first examined abundance of full-length and processed RNAs in wild-type cells grown under several different conditions and in mutant strains defective for RNA degradosome assembly (rne131) or RNase E catalytic activity [rne3071, temperature-sensitive (19)]. Northern blots showed that the full-length manXYZ and processed manYZ mRNAs were present at similar ratios in wild-type cells grown with several different carbon sources (Fig. S1A). Likewise, both RNAs were present in wild-type cells at both permissive and nonpermissive temperatures (Fig. S1B). In contrast, in the host carrying the rne3071 allele, the full-length manXYZ RNA was greatly increased in abundance relative to the manYZ RNA at the nonpermissive temperature. Similarly, manYZ RNA was virtually undetectable in the degradosome-deficient strain, whereas the full-length manXYZ mRNA was present at levels similar to wild-type cells (rne131) (Fig. S1B). Together, these results confirm that processing of manXYZ mRNA to yield the manYZ RNA depends on the catalytic activity of RNase E as well as on the ability to form the degradosome complex.

Because degradosome-deficient cells contain predominantly full-length manXYZ mRNA (Fig. S1B), we used this strain to determine whether formation of the manYZ RNA played a role in SgrS-dependent regulation of manY when pairing at the manX site was eliminated. The mutant fusion manX_GUGY′-′lacZ still was repressed by SgrS by approximately threefold compared with empty vector control in the rne131 host (Fig. 1D), suggesting that formation of manYZ RNA does not influence SgrS regulation of manY translation significantly. Additional experiments confirm that SgrS regulates manY and manZ translation similarly and by a mechanism independent of the manX pairing site when expressed from its native locus under stress conditions (Fig. S2). Together, these data indicate that SgrS regulation of manX is not polar on manY; rather, SgrS independently regulates manY translation.

Defining the Region Required for SgrS-Dependent manY Regulation.

The results presented thus far indicate that degradation of manXYZ mRNA, SgrS pairing, and regulation at manX are not required for SgrS to regulate manY or manZ. To define the determinants of SgrS-mediated regulation of manXYZ, several translational lacZ fusions expressed from a synthetic constitutive promoter Cp19 (20) were constructed (Fig. 2A). A fusion containing the full manX–manY intergenic region (62-manY′; Fig. 2B) was repressed by SgrS to a level similar to that of a fusion containing all the upstream manX sequence (manXY′; Fig. 2B), demonstrating again that manY translational regulation by SgrS does not depend on manX sequences. Truncation of the intergenic region to 40 nt (40-manY′) or 25 nt (25-manY′) eliminated SgrS-dependent repression (Fig. 2B). These results imply that SgrS requires sequences in the manX–manY intergenic region to regulate manY translation.

Fig. 2.

The SgrS-binding site in manX is not required for manY translational regulation. (A) Chromosomal lacZ translational fusions were constructed at the native locus. The native promoter of manX was replaced with the constitutive Cp19 promoter (21). Cp19 also was placed at various locations upstream of the manY start codon, thereby deleting sequences of manX or manY. IGR, intergenic region. (B) Strains with Cp19-manXY'-'lacZ (JH314), Cp19-62-manY'-'lacZ (JH316), Cp19-40-manY'-'lacZ (JH346), or Cp19-25-manY'-'lacZ (JH320) carrying an empty vector or P_lac-sgrS were analyzed for β-galactosidase activity as described in Fig. 1C.

The 40-manY′ fusion has the same 5′ end as the longest processed manYZ product we have detected (Fig. 3A and ref. 9). The fact that SgrS cannot regulate a fusion with this 5′ end suggests that SgrS regulates manY translation in the context of the full-length manXYZ mRNA. This suggestion also is consistent with the observation that SgrS regulates manY translation in the rne131 host where the manYZ RNA species is not produced (Fig. 1D and Fig. S1B).

Fig. 3.

SgrS base pairs with a site in the manX–manY intergenic region. (A) Alignment of manY and SgrS RNA sequences shows complementarity between the manX–manY intergenic region and the 3′ region of SgrS. Pairing that was confirmed by footprinting analysis is indicated by vertical lines. The 5′ ends of the processed manYZ species identified previously (9) are denoted by bold arrows. Boxes around manY bases −41 and −43 and SgrS bases 173–174, 176, and 178 indicate positions that were mutated to test base pairing. (B) In vitro-transcribed SgrS was labeled with ³²P (SgrS*), incubated with unlabeled manY where noted, and treated with RNase T1 (T1), alkaline hydrolysis (OH), or lead acetate (PbAc). The positions of G residues are shown. (C) Cp19-manX'-'lacZ (JH175), Cp19-62-manY'-'lacZ (JH316), and Cp19-manY_S1'-'lacZ (JH332) strains carrying an empty vector, P_lac-sgrS, P_lac-sgrS1, or P_lac-sgrS24 were analyzed for β-galactosidase activity as described in Fig. 1C.

SgrS Base Pairs with Sequences in the manX–manY Intergenic Region to Regulate manY Translation.

The single-stranded 3′ region of SgrS is involved in regulation of two targets, ptsG and manX mRNAs, via base pairing with sequences overlapping the RBS or within the coding region, respectively (9, 10, 13). Based on the analyses of truncated fusions (Fig. 2), we predicted SgrS would pair with nucleotides upstream of the manY RBS in the manX–manY intergenic region. A region of complementarity between SgrS and this region was identified by “sequence gazing” (Fig. 3A). The SgrS nucleotides participating in this predicted interaction overlap with those that pair with ptsG and manX mRNAs (Fig. 3A) (9, 10, 13). In vitro footprinting analysis demonstrated a manY mRNA-dependent region of protection spanning SgrS residues G168 to U179 (Fig. 3B), exactly matching the prediction shown in Fig. 3A.

To assess the relevance of this interaction for regulation in vivo, mutant sgrS alleles were analyzed for their abilities to regulate manX and manY reporter fusions. Mutant SgrS molecules contained G176C, G178C (sgrS1) or A173T, C174G (sgrS24) substitutions. These mutations were predicted to interfere with regulation of manY but not manX (Fig. 3A and ref. 9). As shown previously (9), SgrS1 (G176C, G178C) repressed manX translation as well as wild-type SgrS (Fig. 3C). In contrast, SgrS1 was deficient in repression of manY (Fig. 3C). SgrS24 (A173T, C174G) displayed the same pattern of regulation: It regulated manX as well as wild-type SgrS but was deficient in regulation of manY. The fact that SgrS1 and SgrS24 retained the ability to regulate manX suggested that each of these mutants retained the appropriate structure. Loss of regulation of manY by this allele indicates that distinct residues of SgrS are required for regulation of two different cistrons on the same mRNA.

To obtain further evidence that base pairing between SgrS and sequences in the manX–manY intergenic region is responsible for translational repression of manY, we tested regulation of a manY fusion (62-manY_S1′-′lacZ) containing mutations (C41G and C43G) (Fig. 3A) that should disrupt pairing with wild-type SgrS but restore pairing with mutant SgrS1. As predicted, wild-type SgrS lost the ability to regulate 62-manY_S1′-′lacZ, whereas the compensatory mutations in SgrS1 largely restored regulation of 62-manY_S1′-′lacZ (Fig. 3C). These results confirm that SgrS base pairs with sequences in the manX-manY intergenic region to control manY translation independent of manX translational regulation.

SgrS-manZ Translational Regulation Is Dependent on manY.

Only three nucleotides separate the stop codon of manY and the start codon of manZ (Fig. 1A), making these two genes candidates for translational coupling (16, 17). We hypothesized that SgrS-dependent translational repression of manY would be polar on manZ as a result of translational coupling. To test this hypothesis, a stop codon was inserted into manY in the context of a manZ reporter fusion (Cp19-62-manY_STOPZ′-′lacZ) (Fig. 4A). Only very low basal levels of β-galactosidase activity were detected in the Cp19-62-manY_STOPZ′-′lacZ background in the presence or absence of SgrS (Fig. 4C), confirming that manZ translation is coupled to manY translation. To show that SgrS regulation via pairing upstream of manY affects manZ translation, a Cp19-62-manYZ′-′lacZ fusion (Fig. 4B) was analyzed for regulation by SgrS and SgrS24, which respectively do or do not regulate manY (Fig. 3C). As predicted, wild-type SgrS repressed manZ translation, but SgrS24 lost the repressive ability (Fig. 4C). These results are consistent with SgrS regulation of manZ being dependent on the SgrS–manY interaction and on coupled translation of manY and manZ.

Fig. 4.

SgrS–manZ translational regulation is dependent on SgrS–manY regulation. (A and B) Chromosomal lacZ translational fusions were constructed at the native locus. (C) The Cp19-62-manY_STOPZ'-'lacZ (JH347) strain carrying an empty vector or P_lac-sgrS plasmid and the Cp19-62-manYZ'-'lacZ (JH353) strain carrying an empty vector, P_lac-sgrS, or P_lac-sgrS24 were grown to early log phase, and 0.1 mM IPTG was added. Samples were harvested 60 min after IPTG addition and assayed for β-galactosidase activity as described in Fig. 1C. Activities are reported in Miller units, shown below the bar graph.

SgrS Base Pairing at both Sites Is Required for Regulation of manXYZ mRNA Stability.

The data presented here and in our previous study (9) demonstrate the existence of two SgrS-binding sites on the manXYZ mRNA. In mammalian systems, miRNAs often recognize multiple binding sites on a single mRNA target, and multiple binding events act cooperatively to achieve maximal translational regulation of the target (21–24). For SgrS regulation of manXYZ, there is no evidence of cooperativity for translational regulation, because binding at the manX site alone allows repression of manX (Fig. 3C and ref. 9), and binding at the manX–manY intergenic site alone (when the manX site is mutated or removed) allows regulation of manY translation (Figs. 1D and 3C). Nevertheless, we reasoned that multiple binding sites might operate synergistically with respect to SgrS-dependent degradation of the full-length manXYZ mRNA. To investigate this possibility, mutations that abrogated SgrS pairing at either the manX site (manX_GUGYZ) or the manX-manY intergenic site (manXY_S1Z), or at both (manX_GUGY_S1Z), were incorporated at the native manXYZ locus on the chromosome. Levels of manXYZ mRNA were monitored by quantitative real-time PCR before and after stress. Incorporation of GUG and S1 mutations, individually or in combination, had no significant effect on steady-state manXYZ mRNA levels in the absence of stress (Fig. S3). Stress caused manXYZ mRNA levels to increase slightly (by ∼1.3-fold relative to prestress levels) in a strain lacking sgrS (Fig. 5A). This result was expected, because Northern blots had shown that manXYZ mRNA levels increase in stressed cells in the absence of SgrS (9). In cells producing wild-type SgrS and manXYZ mRNA, levels of the manXYZ mRNA were reduced to approximately one-third the prestress levels (Fig. 5A), confirming the SgrS-dependent degradation of manXYZ mRNA (9). Levels of mutant mRNAs that contained only one of the SgrS-binding sites (manX_GUGYZ and manXY_S1Z) remained unchanged upon stress (Fig. 5A), suggesting that pairing at only one of the two SgrS sites is not sufficient to permit significant SgrS-dependent degradation. When both binding sites were disrupted, i.e., in cells with manX_GUGY_S1Z, there again was slight accumulation of the manXYZ mRNA (∼1.3-fold increased) after stress. In sum, although binding at an individual site permits SgrS to regulate translationally the appropriate genes, binding at both sites is required to allow degradation of the manXYZ mRNA.

Fig. 5.

Pairing of SgrS at both sites on the manXYZ mRNA is required for maximum manXYZ degradation. Cells containing the Cp19-manXYZ (JH124), Cp19-manX_GUGYZ (JH341), Cp19-manXY_S1Z (JHDB05), and CP19-manX_GUGY_S1Z (JHDB07) fusions in wild-type sgrS⁺ backgrounds and the Cp19-manXYZ fusion in a ΔsgrS background (JH125) were grown to midlog phase, and total RNA was extracted from cultures (−αMG). Cells then were treated with 0.5% αMG, and total RNA was extracted 30 min after αMG addition (+αMG). RNA was DNase-treated and reverse transcribed, and the cDNA was subjected to quantitative real-time PCR using manX-specific primers. Levels of the manXYZ transcripts from each of the fusion backgrounds with αMG treatment (+αMG) were calculated relative to the −αMG RNA samples. (B) Total RNA was extracted from cells containing the Cp19-manXYZ (JH124), Cp19-manXY_RBSZ (JHDB11), Cp19-manXY_RBS+S1Z (JHDB15) fusions in wild-type sgrS⁺ backgrounds and the Cp19-manXYZ fusion in a ΔsgrS background (JH125) and was subjected to quantitative real-time PCR as described in A. Levels of the manXYZ transcript from each of the fusion backgrounds αMG treatment (+αMG) were calculated relative to the −αMG RNA samples.

A recent study by Prévost et al. (25), demonstrated that pairing of RyhB sRNA near the RBS of certain targets induces RNase E-dependent cleavage at distal downstream sites. This report led us to test whether SgrS-dependent degradation of manXYZ mRNA requires pairing at both sites, because translation of all three genes must be blocked to expose RNase E cleavage sites. If translation of all three genes must be blocked to expose RNase E cleavage sites, we predicted that RNase E recruited by SgrS–manX pairing would promote manXYZ mRNA degradation even in the absence of pairing at the other site (in the manXY_S1Z mutant) if manYZ translation was blocked by other means. A manY RBS mutation was incorporated on the chromosome in the context of wild-type manXYZ (manXY_RBSZ) and the S1 mutant (manXY_RBS+S1Z). The RBS mutation alone had no effect on SgrS-dependent repression of manXYZ mRNA levels upon stress induction (Fig. 5B), suggesting that blocking translation of manYZ does not inherently change SgrS-dependent degradation of manXYZ mRNA. In contrast, levels of the manXY_RBS+S1Z-mutant RNA were not reduced upon stress (Fig. 5B), indicating that inhibiting manY translation did not restore SgrS-dependent repression when the manX–manY intergenic (S1) pairing site was mutated. This result suggests that pairing at the two sites plays a somewhat more complicated role in recruiting or activating degradosome-dependent cleavage of manXYZ mRNA.

In Vivo Relevance of Binding-Site Multiplicity.

To assess the impact of regulation via multiple binding sites on the physiology of cells growing under stress conditions, we examined growth of cells that were stressed by exposure to the glucose analog 2DG. Exposure of cells to 2DG induces the GP stress response, and ManXYZ is responsible for transporting this stress-inducing molecule (9). We hypothesized that SgrS-dependent translational repression of all three cistrons and subsequent degradation of manXYZ mRNA would make the regulation more efficient by making repression irreversible and that this increased efficiency would facilitate recovery in the presence of 2DG. Strains with wild-type or mutant sgrS and manXYZ alleles were grown on fructose minimal medium with or without 2DG. In the absence of stress, all strains grow similarly on this medium (Fig. 6A). When cells were stressed with 2DG, the strain carrying chromosomal copies of wild-type sgrS and wild-type manXYZ (WT sgrS WT manXYZ; Fig. 6B) and the strain lacking manXYZ (WT sgrS ΔmanXYZ; Fig. 6B) both grew well and formed single colonies. In contrast, cells expressing manXYZ but lacking sgrS failed to grow (ΔsgrS WT manXYZ; Fig. 6B). These data are consistent with previous observations that ManXYZ is the primary transporter for 2DG, and so, when ManXYZ levels are reduced via SgrS-mediated repression or eliminated by manXYZ deletion, cells recover from stress (9). We next tested growth of two different strains expressing RNAs in which SgrS-intergenic site pairing was disrupted (sgrS1 WT manXYZ and WT sgrS manXY_S1Z). The strain with sgrS1 and wild-type manXYZ showed a growth phenotype intermediate between those of wild-type and ΔsgrS strains (sgrS1 WT manXYZ) (Fig. 6B). The same defective growth phenotype was observed for the strain with wild-type sgrS and mutant manXY_S1Z (Fig. 6B). Taken together, these results imply that, although pairing of SgrS at one of the two binding sites on the manXYZ mRNA allows enough repression to provide some relief from stress, base pairing at both sites is required for maximal repression of ManXYZ synthesis and full recovery of cells from stress.

Fig. 6.

Pairing of SgrS at both sites on the manXYZ mRNA is required for maximum relief from stress. Strains with wild-type sgrS and manXYZ (DJ480), ΔsgrS and wild-type manXYZ (CS104), sgrS1 and wild-type manXYZ (CS123), wild-type sgrS and manXY_S1Z (JHDB01), and wild-type sgrS and ΔmanXYZ (YS208) were plated on minimal M63 fructose medium without (A) and with (B) 0.5% 2DG.

Discussion

Our previous (9) and current studies are congruent with a model in which SgrS regulation of manXYZ occurs by base pairing at two sites on the manXYZ transcript, as illustrated in Fig. 7. Interactions of SgrS with a site in the manX coding region and another in the manX–manY intergenic region cause translational repression of manX and manY-manZ, respectively. Although pairing at each site individually is sufficient for translational regulation of the cognate genes, individual pairing interactions do not promote significant RNase E-dependent manXYZ mRNA degradation. Rather, pairing at both sites appears to be a prerequisite for inducing mRNA degradation. Further, although pairing at a single site provides sufficient regulation to allow partial recovery from GP stress, pairing at both sites apparently enhances the efficiency of regulation in a way that facilitates better recovery of cells from stress.

Fig. 7.

The SgrS–manXYZ regulatory mechanism. The data presented are consistent with the model shown in which SgrS binds to sites in the manX coding sequence and to the manX–manY intergenic region to inhibit translation of manX and manYZ, respectively. Both interactions are required to promote efficient RNase E-mediated degradation of the manXYZ mRNA.

Although we have not yet elucidated the molecular mechanisms of translational repression carried out by SgrS base pairing at each site, we can speculate that they are not the canonical sRNA mechanism of direct interference with ribosome binding. Studies mapping ribosome contacts with translation initiation regions (26) and elucidating the window for sRNA binding that permits direct RBS occlusion (27) suggest that the SgrS-binding sites are too far downstream and upstream from the manX and manY RBSs, respectively, to allow direct interference with ribosome binding. Footprinting of manX mRNA with SgrS and the RNA-chaperone Hfq revealed Hfq-dependent structural changes around the manX RBS (Fig. S4). This observation is interesting in light of a recent study that revealed that Hfq itself can repress translation initiation by binding to an RBS-proximal site and inhibiting binding of ribosomes (28). In that case, Hfq is recruited to its binding site on the mRNA by an sRNA that base pairs further upstream. In principle, the manX-binding site for SgrS similarly could allow SgrS to recruit Hfq to bind near the manX RBS and inhibit translation. As for the manX–manY intergenic site, we noted that its composition is rich in CA repeats, reminiscent of putative translational enhancer elements (29). CA-rich elements in several mRNAs are target pairing sites for another bacterial sRNA, GcvB (30). In support of the idea that these elements have a stimulatory effect on manY translation, we note that a truncated manY′-′lacZ fusion lacking one of these sites (40-Y′-′lacZ) (Fig. 2B) had reduced basal levels of translation compared with the fusion with both sites (62-Y′-′lacZ) (Fig. 2B; compare Miller units for strains carrying vector). Moreover, the activity of the 25-Y′-′lacZ fusion, which lacks both CA-rich elements, was less than 1/10th that of the fusion with both elements. Because SgrS base pairs with these CA-rich elements (Fig. 3A), we hypothesize that SgrS pairing at the intergenic site inhibits manY translation by interfering with translational enhancers.

This study raises interesting questions regarding the coupling of sRNA-mediated translational regulation and mRNA degradation. For example, why should interactions at two sites be required for efficient SgrS-dependent manXYZ mRNA degradation? Hfq associates with the C-terminal scaffold region of RNase E, and this association is thought to help recruit RNase E to sRNA–mRNA duplexes (31). However, not all Hfq-dependent sRNA–mRNA duplexes are targeted for degradation, so other factors must modulate the efficiency of RNase E recruitment or the accessibility of RNase E cleavage sites. The experiment shown in Fig. 5B suggests that the role of the manX–manY intergenic SgrS pairing site is more than simply inhibiting manY–manZ translation to unmask RNase E-sensitive sites. Perhaps the nature of Hfq interactions with manXYZ mRNA dictates binding at both sites for efficient Hfq-mediated recruitment of RNase E. More work is needed to elucidate the properties of sRNA–mRNA ribonucleoprotein complexes that determine the RNAs’ stability.

There are a number of examples in which bacterial sRNAs mediate discoordinate regulation of polycistronic transcripts. For example, Spot42 targets the galETKM mRNA and inhibits galK translation but does not affect the expression of galET or the stability of the mRNA (32, 33). RyhB selectively inhibits translation of iscS in the iscRSUA operon and promotes degradation of the iscSUA portion of the transcript, whereas iscR mRNA is stable (34). There are certainly hints from transcriptome studies that bacterial sRNAs commonly target entire operons for coordinate up- or down-regulation (35, 36). We suggest that, in at least some of these cases, sRNAs use multiple binding sites to accomplish coordinated regulation of genes encoded on polycistronic mRNAs. In fact, there is evidence implying sRNA-binding site multiplicity for regulation of the opp3BCDFA mRNA in Staphylococcus aureus. The S. aureus sRNA RsaE appears to regulate mRNAs encoding nutrient transport and metabolism functions (37). RsaE was demonstrated to pair directly with a site overlapping the RBS of the first gene in the operon, oppB. Another study (38) predicted pairing with opp3A, the distal gene in the operon, and demonstrated that RsaE inhibits opp3A translation in vitro, although the specific pairing determinants were not elucidated. Together, these studies suggest that the opp3 mRNA contains at least two RsaE-binding sites. More studies regarding the molecular mechanisms governing this regulation are required to determine how the existence of multiple binding sites affects the production of the proteins encoded by the mRNA and the physiological role for this regulation. There are likely to be some interesting differences between this case and SgrS–manXYZ, because S. aureus sRNAs do not appear to require Hfq, and the data thus far suggest that RsaE does not affect opp3 mRNA levels (37, 38).

One of the central challenges in sRNA research is identifying bona fide mRNA targets: Because complementarity of both miRNAs and bacterial sRNAs with their targets is so short and often imperfect, discerning real targets from false positives is difficult using computational methods. Two major strategies have been used to help reduce the number of false positives obtained in computational searches. One is to constrain the search area to sequence regions of mRNAs that are known to be targeted by small RNAs, i.e., the 3′ UTR in eukaryotes and the 5′ UTR in bacteria. Another approach is to require that putative sRNA target sites be conserved among different species. Both strategies have helped identify many legitimate sRNA-binding sites, but they are not without significant caveats. It is clear from recent studies that there are numerous exceptions in both animals and bacteria to the localization of sRNA-binding sites in UTRs, and sites located within coding sequences of mRNAs have been verified (3, 39). Moreover, it is becoming apparent that many conserved sRNAs may regulate distinct subsets of targets, namely, those that are conserved members of the sRNA’s regulon and those that are regulated in some but not all of the organisms that possess that particular sRNA (9, 40, 41). Our study adds yet another layer of complexity to this issue. The methods used to validate individual predicted sRNA–mRNA binding site interactions usually involve reporter fusions (in vivo) or footprinting experiments (in vitro) in which an individual site is isolated from its normal sequence context. Thus, regulatory contributions of additional unsuspected binding sites are lost. Our earlier study describing the SgrS–manX coding sequence interaction (9) illustrates this concept; only when we probed deeper to test our original model for regulation of manYZ (Fig. 1A) did it become apparent that there were additional important regulatory determinants on the manXYZ mRNA. This example illustrates the fact that we are still uncovering mechanistic variations that regulatory RNAs use to influence gene expression on a global scale.

Materials and Methods

Strain and Plasmid Construction.

The strains and plasmids used in this study are listed in Table S1, and oligonucleotides are listed in Table S2. The standard plasmid vector was pHDB3 (42). The sgrS mutant host used in most experiments was CS104 and was described previously (43). Strains with hfq and rne alleles were made as described in refs. 44 and 45, respectively. Alleles were moved between strains by P1 transduction or inserted via lambda Red recombination (46). Translational LacZ fusions were constructed as described previously (47). Strain XL10 Gold (Stratagene) was used for the QuikChange Mutagenesis procedure. LacI^q (harbored in several strains, including JH111; Table S1) was used to control expression from the P_lac promoter.

The manXY′-′lacZ translational fusion (Fig. 1 B–D) was created as described previously (47) using primers O-JH133/O-JH134 to fuse lacZ to the 109th codon of manY. The fusion was transduced into various backgrounds (Table S1) to create JH135, JH247, and JH256. Similarly, the manXYZ′-′lacZ translational fusion (Fig. 1 B–D) was constructed as described previously (47) using primers O-JH135/O-JH136 to fuse lacZ to the 230th codon of manZ. The fusion was transduced into various backgrounds (Table S1) to create JH130, JH274, and JH275. Chromosomal mutations in manX were made using the following method: A tet^R cassette (generated with oligonucleotides O-JH194/O-JH255) was inserted into manX so that nucleotides to be targeted for mutation were deleted. The resulting strain was named “JH245.” A PCR product (obtained using oligonucleotides O-JH211/O-JH134) containing the CAC to GUG manX mutation and a kan^R cassette at the desired lacZ fusion junction was recombined into JH245 creating JH249. The manX_GUGY′-′lacZ translational fusion (Fig. 1D) then was constructed as described previously (47) by replacing the kan^R cassette with ′lacZ to create JH250.

Translational lacZ fusions with various portions of the manX and manY sequence were constructed in the context of another manXY ′-′lacZ translation fusion (Fig. 2). This manXY′-′lacZ translational fusion was created as described previously (47) using primers O-JH134/O-JH250 to fuse lacZ to the 22nd codon of manY and was transduced into NM200 to create JH298. A tet^R-Cp19 PCR product generated using oligonucleotides O-JH161/O-JH175 was used as a PCR template to generate tet^R-Cp19-manXY’ (O-JH163/O-JH173), tet^R-Cp19-′manXY’ (O-JH163/O-JH239), tetR-Cp19-62-manY′ (O-JH163/O-JH240), tet^R-Cp19-40-manY′ (O-JH163/O-JH288), and tet^R-Cp19-25-manY′ (O-JH163/O-JH273). These PCR products were recombined into JH298 to create JH314, JH315, JH316, JH346, and JH320, respectively.

The Cp19-manY_S1′-′lacZ fusion (Fig. 3C) was created by the same method used to create the manX chromosomal mutants: A tet^R cassette (generated with oligonucleotides O-JH274/O-JH275) was inserted into manY to create strain JH321. A PCR product (obtained using oligonucleotides O-JH134/O-JH276) containing the G39C and T35A mutations, and a kan^R cassette at the desired lacZ fusion junction was recombined into JH321. The manXY₆₆′-′lacZ translational fusion then was constructed as described previously by flipping out the kan^R cassette and replacing it with ′lacZ (47) to create JH323. A tet^R-Cp19-manY_S1 cassette (generated with oligonucleotides O-JH163/O-JH283) was recombined into JH323 to create JH332.

Various mutations at the chromosomal manX and manY loci (Fig. 5) were constructed by first inserting a tet^R cassette using a PCR product generated by primers O-JH285/O-JH286; the resulting strain was JH339. To make the strain JH340 containing the manX_GUG mutation under the control of the Cp19 promoter, a PCR product containing the mutation linked to a kan^R-Cp19 cassette (amplified using O-JH104/O-JH287) was recombined into JH339, and colonies were screened for tet sensitivity. The resulting kan^R-Cp19-manX_GUGYZ cassette was transduced to create JH341 and JH342. To create the strain JHDB01 containing the manXY_S1Z construct, a P_BAD-kan^R-ccdB cassette (generated using oligonucleotides O-JH294 and O-JH295) was inserted into manY so that the manY nucleotides targeted for mutagenesis were deleted. The resulting kanamycin-resistant and arabinose-sensitive strain was called “JH349.” Then a PCR product (obtained using oligonucleotides O-JH276 and O-JH289) containing the manY_S1 mutation was recombined into JH349; strain JHDB01 was selected for arabinose-resistance and screened for kanamycin sensitivity. The Kan^R-Cp19 cassette (amplified using oligonucleotides O-JH104 and O-JH286) was recombined into JHDB01 as described (12) to create JHDB02 (Kan^R-Cp19-manXY_S1Z). This cassette then was transduced into a wild-type background and verified by sequencing to create strain JHDB05. The Kan-Cp19-manX_GUGmanY_S1Z was created starting with JHDB01. A tet^R cassette (generated with oligonucleotides O-JHDB01 and O-JH287) was inserted in manX to delete the nucleotides targeted for mutagenesis. This strain is JHDB03. A PCR product containing the CAC to GUG manX mutation linked to a kan^R-CP19 promoter (obtained using O-JH104 and O-JH287) was recombined into JHDB03 to replace the tetracycline cassette with Kan^R-Cp19-manX_GUG. The resulting JHDB04 strain contained kan^R-Cp19-manX_GUGmanY_S1Z. A manXYZ deletion (Fig. 6) was created by Yan Sun in our laboratory by moving the ΔmanXYZ::kan construct into DJ480.

Plasmid pBRJH24, containing the A173T and C174G mutations in SgrS (Figs. 3C and 4C), was created using the QuikChange Mutagenesis Kit (Stratagene) with primers O-JH253 and the P_lac-sgrS (pLCV1) plasmid as the template.

Media and Reagents.

Bacteria were cultured in LB medium or on LB agar plates at 37 °C unless otherwise noted. Bacto Tryptone medium (BD), supplemented with 100 μg/mL ampicillin where indicated, was used for β-galactosidase assays. Isopropyl β-d-1-thiogalactopyranoside (IPTG) was used at a concentration of 0.1 mM for induction of P_lac-sgrS. Morpholine-propanesulfonic acid–rich defined medium (Teknova) with 0.2% glucose, 0.2% mannose, or 0.4% glycerol as a carbon source was used for culturing cells for RNA extraction where noted.

β-Galactosidase Assays.

β-Galactosidase assays were performed as described previously (9). Activities (in Miller units) were normalized to ΔsgrS or empty vector control to give the percentage relative activity for experimental samples.

In Vitro RNA Footprinting.

In vitro transcription templates were generated by PCR using gene-specific oligonucleotides containing the T7 promoter sequence. Oligos O-JH251/O-JH120 were used to generate the manY template (from −62 to +107 relative to the translational start site) and oligos O-JH219/O-JH119 were used to generate an SgrS template. In vitro transcription and footprinting reactions were carried out as described previously (9).

Growth Assays.

The strains containing wild-type or various manXYZ and sgrS chromosomal alleles described in Fig. 6 were streaked on M63 fructose medium with and without 0.5% 2DG. Plates were imaged after 20 h of incubation.

RNA Extraction and Quantitative Real-Time PCR.

RNA extraction was performed as described previously (9) using hot-phenol extraction (48) both before (0 min) and 30 min after αMG addition (28). Reverse transcription was performed with 2 μg DNase (Ambion)-treated RNA using SuperScript III Reverse Transcriptase (Invitrogen). The manXYZ gene product was amplified from this resulting cDNA using Power SYBR-Green (Applied Biosystems) with oligonucleotides O-MBP18F1 and O-MBP18R1. The cycle threshold (Ct) values of the manXYZ transcript were normalized to that of the housekeeping gene, rrsA, encoding the 16S rRNA. The amount of manXYZ at 0 min was used as the calibrator (and was normalized to 1.0) to calculate relative abundance of manXYZ mRNA after αMG treatment.