The OxyS regulatory RNA integrates the adaptive response to hydrogen peroxide with other cellular stress responses and protects against DNA damage. Among the OxyS targets is the rpoS-encoded σs subunit of RNA polymerase. σs is a central regulator of genes induced by osmotic stress, starvation and entry into stationary phase. We examined the mechanism whereby OxyS represses rpoS expression and found that the OxyS RNA inhibits translation of the rpoS message. This repression is dependent on the hfq-encoded RNA-binding protein (also denoted host factor I, HF-I). Co-immunoprecipitation and gel mobility shift experiments revealed that the OxyS RNA binds Hfq, suggesting that OxyS represses rpoS translation by altering Hfq activity.
During exponential growth, the expression of several hydrogen peroxide-inducible proteins is controlled by the transcriptional regulator, OxyR. We recently showed that OxyR also induces the expression of an abundant, 109 nucleotide, untranslated regulatory RNA, denoted OxyS (Altuvia et al., 1997). This small RNA acts as an antimutator and a regulator of as many as 40 genes in Escherichia coli. One target of OxyS repression is the rpoS-encoded σs subunit of RNA polymerase. σs levels are normally increased during transition into stationary phase or upon starvation, as well as under other stress conditions such as acid shock and osmotic upshift (for reviews, see Loewen and Hengge-Aronis, 1994; Hengge-Aronis, 1996). The increased levels of σs lead to the induction of genes that have diverse functions including protection against DNA damage, resistance against oxidative, thermal, acid and osmotic stresses, as well as virulence in Salmonella. We found that the OxyS RNA represses rpoS expression. In strains carrying deletions of oxyS, an rpoS–lacZ fusion is derepressed after treatment with hydrogen peroxide. Since σs and the OxyR transcription factor activate some of the same antioxidant genes, we proposed that OxyS RNA repression of rpoS evolved to prevent redundant induction of oxidative stress genes (Altuvia et al., 1997).
We investigated the mechanism of OxyS regulation of rpoS and found that the small RNA represses rpoS at a post-transcriptional level. We also discovered that OxyS repression of rpoS translation requires Hfq and that the OxyS RNA binds the Hfq protein in vivo and in vitro. These results have led us to propose that OxyS binding alters Hfq activity and thereby prevents rpoS mRNA translation.
Several OxyS-regulated genes are also controlled by σs
We previously observed that several OxyS-regulated genes are induced in stationary phase (Altuvia et al., 1997). Therefore, we examined whether these genes are regulated by σs and whether OxyS repression of these targets is mediated by σs. We compared the expression of fhlA–lacZ, yhiV–lacZ, yhiM–lacZ, gadB–lacZ and dps–lacZ fusions in wild-type and rpoS::Tn10 mutant backgrounds, and examined the effects of constitutively expressing OxyS (poxyS) (Table I). For the strains carrying the pKK177-3 vector control, the activities of all of the fusions were lower in the rpoS mutant background compared with the wild-type cells; therefore, σs is required for expression of these genes. The extent of OxyS repression of the fhlA–lacZ fusions was identical in the wild-type and rpoS::Tn10 mutant backgrounds, indicating that OxyS repression of fhlA is not mediated by σs. In contrast, OxyS repression of yhiV, yhiM, gadB and dps was significantly decreased in the mutant backgrounds, suggesting that OxyS represses these genes, in part, by decreasing rpoS expression. We still observed some OxyS repression of yhiV–lacZ and gadB–lacZ, and surprisingly, the expression of the yhiM–lacZ fusion was induced rather than repressed by OxyS in the rpoS mutant background. These results suggest that OxyS is also acting by rpoS-independent mechanisms at yhiV, yhiM and gadB.
Table 1. Effect of rpoS::Tn10 on OxyS-regulated genes
Average (in Miller units) of three independent experiments. Cells were diluted 1:100 and grown in LB with ampicillin (amp) for 12 h.
56 ± 4
1.0 ± 0.5
27 ± 1
0.5 ± 0.2
147 ± 19
0.1 ± 0.1
4 ± 1
0.1 ± 0.1
667 ± 56
0.5 ± 0.1
0.7 ± 1.0
14 ± 4
28 855 ± 2247
135 ± 58
211 ± 40
15 ± 4
40 967 ± 4580
15 948 ± 9868
2038 ± 125
1596 ± 89
OxyS represses stationary-phase and osmotic induction of rpoS
To examine the effect of OxyS on rpoS expression, we measured the expression of an rpoS–lacZ translational fusion in strains carrying a vector control (pKK177-3) or constitutively-expressing OxyS (poxyS). We first monitored rpoS–lacZ expression throughout growth in both rich [Luria-Bertani (LB)] and minimal (M63) media. As shown in Figure 1A, the growth rates of OxyS-expressing cells and control cells were identical. However, the level of β-galactosidase activity in the OxyS-expressing cells was dramatically decreased, showing that OxyS represses the stationary-phase induction of rpoS expression in both rich and minimal media. Similarly, we found that rpoS induction by high salt was abolished in the poxyS strain (Figure 1B), indicating that the OxyS RNA also modulates the osmotic regulation of rpoS. OxyS repression of the stationary phase induction of σs was confirmed by immunoblots (Figure 2; Altuvia et al., 1997). We found that OxyS expression caused a decrease in σs levels in the late-exponential and stationary phase cells in both rich and minimal media.
OxyS inhibits rpoS expression at a post-transcriptional level
Since rpoS expression is regulated at multiple levels, we examined the effect of constitutive OxyS expression (poxyS) on transcriptional and translational rpoS–lacZ fusions (Table II). The extent of repression observed for the translational fusions was significantly higher than the extent of repression observed for the transcriptional fusion, indicating that OxyS acts primarily at a post-transcriptional level. The translational fusion rpoS379–lacZ (RO90) contains the translation initiation region (TIR), while the translational fusion rpoS742–lacZ (RO91) carries both the TIR domain and an σs internal turnover element such that the expression of the rpoS742–lacZ fusion is regulated at both the level of translation and protein stability (Muffler et al., 1996b; Figure 3). We found that the OxyS RNA repressed both of these translational fusions but observed 5-fold stronger repression with the rpoS742–lacZ fusion in minimal medium. These findings suggest OxyS inhibits rpoS translation in both rich and minimal medium, and may destabilize σs in minimal medium.
Average (in Miller units) of three independent experiments. Cells were diluted 1:100 and grown in LB or M63 glucose medium with chloramphenicol (cm) for 12 h.
17 ± 1
15 ± 2
423 ± 85
41 ± 16
552 ± 59
129 ± 17
10 ± 2
5 ± 1
62 ± 12
1 ± 1
132 ± 10
11 ± 2
OxyS regulation of rpoS is dependent on Hfq
Several factors, including the RNA-binding protein Hfq, the small RNA DsrA, the histone-like protein H-NS, the response regulator RssB and the ClpX/ClpP protease, have been demonstrated to modulate rpoS expression at a post-transcriptional level (Barth et al., 1995; Yamashino et al., 1995; Brown and Elliott, 1996; Muffler et al., 1996a,b; Pratt and Silhavy, 1996; Schweder et al., 1996; Sledjeski et al., 1996). To determine whether the OxyS RNA acts through any of these regulators, we moved the corresponding mutant alleles into a host strain carrying the rpoS742–lacZ fusion and examined the effect of constitutive OxyS expression in these strain backgrounds in both rich and minimal media (Table III). The OxyS RNA repressed rpoS expression in all strains except the hfq1::kan mutant. Thus, OxyS appears to be acting through the Hfq protein. OxyS repression of rpoS was 4-fold stronger in minimal medium than in rich medium (30-fold versus 7-fold), consistent with the results in Table II. Interestingly however, OxyS-mediated repression was reduced 5-fold in the rssB::Tn10, clpX::kan and clpP::kan backgrounds in the M63 medium, suggesting that OxyS affects σs stability in minimal medium through RssB and the ClpX/ClpP protease.
Table 3. Effect of chromosomal mutations on OxyS repression of rpoS742–lacZ
Average (in Miller units) of three independent experiments. Cells were diluted 1:100 and grown in LB or M63 glucose medium with cm for 12 h.
458 ± 49
62 ± 17
75 ± 26
91 ± 22
233 ± 11
49 ± 15
747 ± 58
100 ± 14
835 ± 78
69 ± 10
986 ± 88
172 ± 22
867 ± 71
205 ± 33
1101 ± 70
185 ± 45
M63 glucose medium:
60 ± 17
2 ± 2
6 ± 3
2 ± 1
34 ± 4
2 ± 3
358 ± 62
30 ± 8
345 ± 48
24 ± 3
291 ± 81
45 ± 17
250 ± 81
49 ± 23
410 ± 98
70 ± 32
A-rich linker region is important for OxyS repression of rpoS
Hfq is a 12 kDa, heat-stable, RNA-binding protein. This protein was originally identified as having a role in the replication of the RNA bacteriophage Qβ but was recently found to have more pleiotropic effects including a role in rpoS translation (Tsui et al., 1994; Brown and Elliott, 1996; Muffler et al., 1996b, 1997). Studies of Hfq binding to the Qβ and R17 phage RNAs indicate that the protein preferentially binds A-rich sequences (Senear and Steitz, 1976). The OxyS RNA may act on rpoS by either affecting the Hfq levels or modulating Hfq action. To test the first possibility, we examined the Hfq levels by immunoblot analysis (Figure 2). No apparent differences were detected in cells carrying poxyS compared with the vector control in both rich and minimal media at all stages of growth, indicating that the OxyS RNA has no effect on the Hfq levels.
Several models of the rpoS mRNA secondary structure (Brown and Elliott, 1997; S.Bouche, D.Traulsen and R.Hengge-Aronis, unpublished data) predict that the rpoS ribosome binding site is occluded, and that Hfq binding to A-rich sequences in this region is required for a conformational change which allows efficient translation of the rpoS mRNA. Deletion studies of the OxyS RNA showed that the 27-nucleotide linker region between stem–loops b and c of the OxyS RNA is required for OxyS repression of rpoS (Altuvia et al., 1998). Intriguingly, this linker region contains five AA repeats. To test whether this A-rich region was important for OxyS repression of rpoS, we constructed mutants carrying AA to GG substitutions at four positions (Figure 4). All of the strains carrying the mutant plasmids showed less repression than the strain carrying poxyS. Although none of the effects of the individual AA mutations is dramatic, the reduced rpoS repression is consistent with a role of the linker region. We propose Hfq binds this region, possibly at multiple sites.
OxyS co-immunoprecipitates with Hfq
To test for an interaction between the OxyS RNA and the Hfq protein in vivo, we carried out immunoprecipitation experiments. We first constructed an Hfq derivative carrying a C-terminal Myc tag (pSU-hfq-myc). The Myc-tagged Hfq protein appeared to have wild type activity since the protein could restore normal levels of rpoS–lacZ expression in an hfq1::kan mutant strain (data not shown). MC4100/pSU-hfq-myc cells carrying pKK177-3 or poxyS were grown to early stationary phase in LB medium. Immunoblots of lysates from these cells showed that, as expected, σs levels were reduced in the poxyS strains compared with the pKK177-3 strain (Figure 5A). In contrast, both strains had equally high levels of the Myc-tagged Hfq protein. α-Myc monoclonal antibodies were then used to immunoprecipitate the Myc-tagged Hfq protein. Equal quantities of Hfq protein were precipitated from the pKK177-3 and the poxyS strains (Figure 5B). The RNA in the immunoprecipitated complexes was extracted and assayed for the presence of OxyS. The OxyS RNA was clearly co-immunoprecipitated from the poxyS extracts (Figure 5B). The co-immunoprecipitation was specific since the OxyS RNA was not precipitated by the cells expressing Hfq without a Myc tag (data not shown). As another specificity control, we assayed for the presence of the 362-nucleotide 10Sa RNA which also carries several AA repeats and whose levels are 3-fold higher than the OxyS RNA (Altuvia et al., 1997). While the 10Sa RNA was clearly present in the cell lysate, we could not detect this control RNA in the precipitated complex (Figure 5B). Together, these results indicate that the OxyS RNA binds Hfq in vivo.
OxyS binding to the Hfq protein may prevent rpoS translation by competing with the rpoS mRNA. Alternatively, Hfq may simultaneously bind OxyS and the rpoS mRNA, leading to the formation of a ternary complex which cannot be translated. We tested for the presence of the rpoS mRNA in the complex which immunoprecipitated with Hfq. Interestingly, equal amounts of rpoS mRNA were precipitated from cells carrying both pKK177-3 and poxyS (Figure 5B). Thus, rpoS binding to Hfq is not affected by high levels of OxyS, suggesting that the OxyS and rpoS RNAs do not compete for the same site on Hfq.
OxyS binds Hfq in vitro
To further verify OxyS binding to Hfq, in vitro-synthesized OxyS RNA was incubated with purified Myc-tagged Hfq protein and examined by a gel mobility shift assay (Figure 6). Incubation with Hfq clearly led to gel retardation of the oxyS transcript showing that Hfq interacts with the oxyS RNA. To test for specificity we added a 500-fold excess of unlabeled RNAs. These experiments showed that the OxyS–Hfq interaction is specific since the binding was strongly competed by the OxyS1→109 transcript, but much less efficiently by the similarly-sized 10Sa1→137, 10Sa138→299 or rpoS−126→44 transcripts.
The E.coli OxyS is a small untranslated RNA, induced by oxidative stress (Altuvia et al., 1997). This novel RNA acts as a pleiotropic regulator of the expression of as many as 40 genes. We have examined OxyS regulation of its target genes and found that the regulatory RNA represses a subset of the target genes, yhiV, yhiM, gadB and dps, through the RNA polymerase σs subunit encoded by rpoS (Table I). OxyS repression of dps is completely dependent on σs, while OxyS repression of yhiV and gadB is partially σs-dependent. The effect of OxyS on yhiM expression is more complex. In an rpoS+ background OxyS represses yhiM, while the RNA activates yhiM in an rpoS::Tn10 mutant strain. This observation suggests that OxyS regulates yhiM expression by two different mechanisms, one being σs-dependent and the second being σs-independent. Further characterization of the two mechanisms should be interesting, especially since another OxyS-regulated gene recently identified also shows opposing effects of OxyS in the presence and absence of rpoS (T.Schar and G.Storz, unpublished data).
Previous studies have shown that rpoS expression is controlled on multiple levels, and several regulators have been characterized. Our results suggest that OxyS represses rpoS translation in both rich and minimal medium. OxyS also appears to affect σs stability in minimal medium. The effect on σs stability may be mediated through the response regulator RssB and/or the ClpX/ClpP protease, but not through the histone-like protein H-NS. It is interesting to note that a mutation in rssB was isolated in a screen for mutations that affected the ability of OxyS to repress the expression of a yhiV–lacZ fusion (Muffler et al., 1996a). Why the effect on proteolysis is more predominant in minimal medium and how OxyS affects σs stability is not clear. Possibly, OxyS destabilizes σs by affecting the levels of RssB or ClpX/ClpP, or by modulating the activities of these proteins.
Our genetic studies showed that OxyS repression of rpoS translation is mediated through the RNA-binding protein, Hfq. Since OxyS had no effect on the Hfq levels, the RNA must repress rpoS by affecting Hfq activity. The Hfq protein preferentially binds A-rich sequences (Senear and Steitz, 1976), and we observed that the linker region between stem–loops b and c of OxyS contains five AA repeats. We propose that one or more Hfq molecules binds this linker region since mutations of the AA sequences decrease OxyS repression of rpoS. The essential role of the linker region is supported by deletion studies of Altuvia et al. (1998). While a mutant carrying a 5′-deletion of stem–loops a and b (poxySΔ1–63) still repressed rpoS, a mutant carrying a 5′-deletion of stem–loops a and b, and the linker region (poxySΔ1–90) did not repress rpoS even though this deletion mutant was still able to regulate fhlA, an OxyS target found to be repressed by an antisense mechanism.
Co-immunoprecipitation and gel mobility shift experiments revealed that the OxyS RNA binds the Hfq protein. We previously calculated there to be ∼4500 molecules of oxyS per cell after treatment with hydrogen peroxide (Altuvia et al., 1997). Hfq has also been reported to be abundant although the exact levels of the protein are controversial (Carmichael et al., 1975; Kajitani et al., 1994). While the nature of the OxyS–Hfq interaction needs to be analyzed, it is reasonable to expect that the observed levels of OxyS are adequate to modulate Hfq activity.
Based on our current results and other rpoS structure-and-function studies (Brown and Elliott, 1997; S.Bouche, D.Traulsen and R.Hengge-Aronis, unpublished data), we propose the following model to explain OxyS RNA repression of rpoS translation. In the absence of other regulators, a secondary structure encompassing the rpoS mRNA Shine–Dalgarno sequence interferes with ribosome binding. Hfq recognizes and binds A-rich sequences in the vicinity of the ribosome binding site. This binding releases the rpoS Shine–Dalgarno and allows translation. When the OxyS RNA is induced by oxidative stress, OxyS binds Hfq and prevents Hfq from acting. Interestingly, the DsrA RNA contains a sequence complementary to the rpoS message. Thus, DsrA may act to induce rpoS expression by pairing with the rpoS message and melting the secondary structure (Majdalani et al., 1998).
OxyS binding to Hfq may compete with rpoS mRNA binding. Alternatively, Hfq may simultaneously bind OxyS and the rpoS mRNA leading to the formation of a ternary complex which cannot be translated. The findings that high levels of OxyS do not affect rpoS co-immunoprecipitation with Hfq, and that a rpoS transcript encompassing the putative Hfq binding site did not compete with OxyS binding in a gel mobility shift assay, suggest that Hfq may simultaneously bind the OxyS and rpoS transcripts, but additional binding studies are warranted. Regardless, it is clear that the OxyS RNA specifically binds the Hfq protein.
It is intriguing how many regulators act to modulate σs levels in response to starvation, osmotic shock, acid shock, low temperature, and in the case of OxyS, oxidative stress. The finding that the hydrogen peroxide-induced OxyS RNA represses rpoS expression appeared counterintuitive since σs controls the expression of several genes that protect against oxidative stress. We proposed that OxyS regulation of rpoS may provide a mechanism to fine tune the expression of antioxidant activities and prevent the redundant induction of katG, gorA and dps, by both the hydrogen peroxide-specific OxyR transcription factor and the general stress factor σs (Altuvia et al., 1997). In addition, OxyS repression of rpoS would prevent the induction of many broadly protective σs-dependent proteins whose synthesis would be costly and unnecessary as long as the OxyR-regulated response can alleviate the stress condition. An important direction for future work will be to clearly elucidate the interactions among all of the regulators of rpoS under different stress conditions.
Materials and methods
Bacterial strains and growth conditions
The bacterial strains used in this study are listed in Table IV. Cultures were grown under aeration at 37°C in LB rich medium or in M63 minimal medium supplemented with 2 mg/ml glucose, 20 μg/ml vitamin B1 and 1 mg/ml casamino acids (Miller 1972). Growth was monitored by measuring the optical density at 600 nm (OD600). Ampicillin (amp, 50 μg/ml) or chloramphenicol (cm, 25 μg/ml) was added where appropriate. The rpoS::Tn10, hfq1::Ω, ΔdsrA5…Tn10, hns8::Tn10, hns24::Tn10, rssB::Tn10, clpX::kan, and clpP::kan mutant alleles (Loewen and Triggs, 1984; Shi et al., 1993; Tsui et al., 1994; Muffler et al., 1996a; Sledjeski et al., 1996) were moved into the desired strain backgrounds by P1 transductions (Silhavy et al., 1984).
All DNA manipulations were carried out using standard procedures. The OxyS mutants carrying the specific nucleotide changes were generated by using the QuikChange Site-Directed Mutagenesis Kit (Stratagene). poxyS (pGSO85, Altuvia et al., 1997) DNA was subjected to PCR using primers to create the desired mutations; poxySA65G+A66G [#663 (5′-GAG TTT CTC AAC TCG GGT AAC TAA AGC CAA CG) + complementary #664], poxySA72G+A73G [#665 (5′-TCA ACT CGA ATA ACT GGA GCC AAC GTG AAC TT) + complementary #666], poxySA78G+A79G [#667 (5′-CGA ATA ACT AAA GCC GGC GTG AAC TTT TGC GG) + complementary #668], and poxySA84G+A85G [#669 (5′-ACT AAA GCC AAC GTG GGC TTT TGC GGA TCT CC) + complementary #670]. The presence of the mutations was confirmed by DNA sequencing. The levels of the mutant RNAs were examined by Northern blot analysis: total RNA isolated from cells grown for 12 h was separated on a 6% polyacrylamide–urea gel, transfered to a nylon membrane by electroblotting, and probed with a γ-32P-end-labeled primer (5′-CCT GTG TGA AAT TCT TAT CC, corresponding to pKK177-3 sequence upstream of the oxyS coding sequence).
β-galactosidase activity was assayed by use of o-nitrophenyl-β-D-galactopyranoside (ONPG) as a substrate (Miller, 1972).
To determine the cellular levels of σs and Hfq, MC4100 cells carrying pKK177-3 or poxyS (pGSO5) were grown for 2, 6 and 12 h in LB, or 4, 8 and 12 h in M63 medium. Aliquots (1 ml) were centrifuged, and equal absorbance units were suspended in Laemmli buffer (1 OD600/100 μl). The proteins (5 μl) were separated on SDS–12% polyacrylamide gels and transferred to nitrocellulose membranes by electroblotting. The blots were probed with a 1:4000 dilution of α-σs antibody (Lange and Hengge-Aronis, 1994) or a 1:4000 dilution of α-Hfq antibody (Kajitani et al., 1994).
Primer extension assay
RNA samples were subjected to primer extension assays using primers specific to OxyS [#188 (5′-GCA AAA GTT CAC GTT GG)] and 10Sa [#620 (5′-TTG CGA CTA TTT TTT GCG GC)]. The RNA samples were incubated with 0.5 pmol of a γ-32P-endlabeled primer for 5 min at 65°C, and then quick-chilled on ice. After the addition of dNTPs (1 mM each) and AMV reverse transcriptase (10 U, Life Sciences) the reactions were incubated for 1 h at 42°C. The cDNA products were then analyzed on an 8% polyacrylamide–urea gel.
RNase protection assay
Fragments carrying the rpoS coding sequence (from +24 to +304, relative to A of the initiating AUG) were generated by PCR [#683 (5′-GGC AAG CTT CCA GAC GCA AGT TAC TCT CGA) and #684 (5′-GGC GAA TTC TCA TGA TTT AAA TGA AGA TGC)]. The DNA fragments were digested with EcoRI and HindIII, and cloned into the corresponding sites of pGEM2 (Promega). The [α-32P]rpoS antisense transcript was made from the EcoRI-linearized plasmid by in vitro transcription with T7 RNA polymerase. The RNase protection assays were carried out using the RPAII kit (Ambion). The RNA samples were hybridized with 1.5×106 c.p.m. of [α-32P]rpoS antisense transcripts overnight at 45°C and then digested with RNase A (0.5 U) and RNase T1 (20 U) for 30 min at 37°C. Subsequently, the samples were treated with proteinase K, extracted with phenol, precipitated with ethanol in the presence of 5 μg yeast tRNA, and analyzed on a 6% polyacrylamide–urea gel.
To construct pSU-hfq-myc, fragments carrying the E.coli hfq gene linked to the myc tag at the C-terminus were generated by PCR [#654 (5′-ACG AAT TCG ATG GCT AAG GGG CAA TCT) and #655 (5′-CCA AGC TTT CAA TTC AAG TCC TCC TCG CTG ATC AGC TTC TGC TCC ATT GAT TCG GTT TCT TCG CTG TCC T)]. To construct the control plasmid pSU-hfq, hfq was amplified by PCR [#654 and #680 (5′-CCA AGC TTA TTC GGT TTC TTC GCT GTC C)]. The DNA fragments were then digested with EcoRI and HindIII and cloned into the corresponding sites of pSU18 (Bartolomé et al., 1991). Cultures of MC4100/pSU-hfq-myc carrying pKK177-3 or poxyS (pGSO5) were grown to early stationary phase (OD600 = 1.5) in LB medium with amp and cm. The cells from 30 ml of culture were collected and lysed by three freeze-thaw cycles in 1.5 ml of lysis buffer (10 mM Tris–HCl pH 8.0, 250 mM NaCl, 10 mM MgCl2 and 1 mM dithiothreitol) with 150 μg lysozyme. The samples were then treated with DNase I (740 U) and RNasin (200 U) for 20 min on ice. After centrifugation, the supernatant was treated as follows: one aliquot (0.25 μl) was analyzed by immunoblotting as described above. Total cellular RNA from a second aliquot (490 μl) was isolated by phenol extraction, and then 2 μg of RNA was subjected to primer extension assays using primers specific to OxyS and the 10Sa RNA, or to RNase protection assays with rpoS antisense transcript. A third aliquot (500 μl) was subjected to immunoprecipitation as follows: protein A–Sepharose beads (250 μl of a 1:1 slurry; Pharmacia) were incubated with α-Myc monoclonal antibodies (25 μl of 9E10; Santa Cruz Biotechnology) and lysis buffer (1 ml) for 2 h at room temperature, and then washed three times with lysis buffer. The washed beads (50 μl of a 1:1 slurry) was mixed with the supernatant (500 μl) and incubated for 2 h at 4°C. The beads were subsequently collected and washed three times with lysis buffer. Finally, the pellet was resuspended in 500 μl of lysis buffer and 2.5 μl was analyzed by immunoblotting. RNA was isolated from 490 μl (with 100 μg yeast tRNA as carrier) and 10 μg of RNA subjected to primer extension or RNase protection assays.
Purification of Myc-tagged Hfq
Cell pellets from 25 ml of cultures (grown to OD600 = 1.5 in LB) of MC4100/pSU-hfq-myc or MC4100/pSU-hfq (as a control) were resuspended in 1.5 ml lysis buffer (10 mM Tris–HCl pH 8.0, 250 mM NaCl, 10 mM MgCl2, 1 mM EDTA) containing 150 μg lysozyme, and lysed by three freeze-thaw cycles and subsequent sonication. The lysates were then treated with DNase I (740 U) and RNase A (15 μg) for 15 min on ice. After centrifugation, the cleared lysates were incubated with protein A–Sepharose cross-linked to α-Myc monoclonal antibodies (150 μl of a 1:1 slurry, prepared as described in Harlow and Lane, 1988) for 2 h at 4°C. The immunoprecipitated complexes were washed three times with lysis buffer and the proteins were eluted in 150 μl of lysis buffer by heating for 30 min at 95°C. An aliquot was examined on an SDS–polyacrylamide gel stained with Coomassie Blue as well as by immunoblotting to verify that the Myc-tagged Hfq protein isolated from the MC4100/pSU-hfq-myc cells was purified to near homogeneity.
Gel mobility shift assay
The RNAs used for the mobility shift were obtained as follows. PCR was used to amplify OxyS1→109 [#689 (5′-CTT GAA TTC TAA TAC GAC TCA CTA TAG GGA AAC GGA GCG GCA CC) and #690 (5′-TAC AAG CTT GCG GAT CCT GGA GAT CCG CAA AAG TT)], 10Sa1→137 [#694 (5′-CTT GAA TTC TAA TAC GAC TCA CTA TAG GGG CTG ATT CTG GAT TCG) and #695 (5′-TAC AAG CTT GCT CTA AGC AGG TTA TTA AGC TGC TA)], 10Sa138→299 [#696 (5′-CTT GAA TTC TAA TAC GAC TCA CTA TAG GGC CTC TCT CCC TAG CCT) and #697 (5′-TAC AAG CTT GTC AGT CTT TAC ATT CGC TTG CCA GC)], and rpoS−126→+44 (from −126 to +44 relative to A of the initiating AUG) [#698 (5′-CTT GAA TTC TAA TAC GAC TCA CTA TAG GGC ATT TTG AAA TTC GTT AC) and #699 (5′-TAC AAG CTT GCA TCT TCA TTT AAA TCA TGA ACT TT)]. The DNA fragments were then digested with EcoRI and HindIII, and cloned into the corresponding sites of pSP64 Poly(A) (Promega). The [α-32P]OxyS transcript and the unlabeled competitor RNAs were prepared from the HindIII-linearized plasmids by in vitro transcription with T7 RNA polymerase (Gibco-BRL).
The gel mobility shift assays were carried out as follows. Purified Myc-tagged Hfq (2 pmol) and [α-32P]OxyS1→109 transcript (2 fmol, 3×104 c.p.m.), without or with unlabeled OxyS1→109, 10Sa1→137,10Sa138 →299, or rpoS−126→+44 transcript (1 pmol), were incubated in a 20 μl reaction in RNA binding buffer (10 mM Tris–HCl pH 8.0, 50 mM NaCl, 50 mM KCl, 1 mM MgCl2) for 10 min at 37°C, and mixed with 2 μl loading dye (50% glycerol, 0.1% bromophenol blue, 0.1% xylene cyanol). The mixtures were analyzed on 5% polyacrylamide gels in 0.5× TBE buffer at 150 V for 2 h, and subjected to autoradiography.
We thank A.Ishihama for the Hfq antibodies, S.Drake for α-Myc-protein A–Sepharose, S.Gottesman, D.Sledjeski and M.Winkler for strains, and H.Zgurskaya and W.Zhang for technical advice. We also appreciate the experimental suggestions of the anonymous reviewers and the editorial comments of H.Giladi, S.Gottesman, P.Loewen and K.Wassarman. This work was supported by the intramural program of the National Institute of Child Health and Human Development, by grant number 95-00092 (BSF) from the United States–Israel Binational Science Foundation, and by internal funds from the Hebrew University to S.A.