Hfq Is a Limiting Factor for sRNA Activity.

Having established the necessity of Hfq for the optimal activity of RyhB, DsrA, and MicC, we examined the effect of additional Hfq. If Hfq is a limiting factor, increasing the Hfq concentration will increase sRNA activity. In contrast, if Hfq is in excess, increasing its concentration will have little effect. Therefore hfq was cloned onto a plasmid under control of the pLlacO-1 promoter and transformed into a strain with chromosomal hfq (Fig. 1D). We measured the expression of sodB::gfp, ompC::gfp, and rpoS::gfp in the presence of their complementary sRNA (RyhB, MicC, and DsrA, respectively) with and without extra Hfq. From these measurements, we calculated the percentage change in expression caused by the additional Hfq (Fig. 1D), which showed that silencing of sodB::gfp and ompC::gfp and activation of rpoS::gfp increased. The measurements were also performed without the sRNAs, and this showed a much smaller change in the expression of these target mRNAs (Fig. 1D).

To examine the relationship between the Hfq level and sRNA activity further, we placed the above circuits in a strain without chromosomal hfq and added IPTG to the media to vary Hfq production (Fig. 2A). To estimate the relative amount of Hfq produced at different IPTG concentrations, we placed gfp under the control of the pLlacO-1 promoter. We found that the amount of silencing and activation occurring with only chromosomal hfq (no induction of plasmid hfq) can be achieved at ≈10–20% of the maximal induction of plasmid hfq (Fig. 2 B–D, gray lines). Western blotting confirmed that the Hfq concentration increases with IPTG induction, and it showed that the level of Hfq produced by the chromosome is relatively low (Fig. 2 E and F).

Open in a separate window

Fig. 2.

Hfq is a limiting factor for sRNA silencing and activation. Error bars indicate the SEM. (A) Experimental system. (B–D) Target gene expression at varying Hfq levels in strains with the chromosomal hfq deleted. The gray lines indicate the level of Hfq production resulting in the same level of target gene expression as the strain with chromosomal hfq (green). (E) Hfq quantitation by Western blotting. Hfq and L9 (loading control) had multiple nonspecific bands that are not shown. The amount of Hfq produced by chromosomal hfq in E. coli MG1655 (“chr. hfq”) was typically less than in this image. (F) Relative amount of Hfq measured by Western blotting normalized to the maximum level. Hfq quantification was performed in triplicate.

Increasing the Hfq concentration above the level provided by the chromosomal copy produced an approximately linear increase in MicC silencing (i.e., decreased ompC::gfp expression) and DsrA activation (i.e., increased rpoS::gfp expression) (Fig. 2 B and C). This shows that the amount of Hfq was limiting the amount of silencing and activation. Increasing the Hfq concentration also increased RyhB silencing (Fig. 2D), suggesting that Hfq is also limiting in this system; however, it is not as clear in this case because the amount of silencing was also limited by the target mRNA concentration (i.e., sodB::gfp was transcribed from a weak promoter because of toxicity at high transcription levels).

Effect of Hfq on sRNA and Target mRNA Concentrations and Translation.

Hfq binding has been shown to decrease the degradation of RyhB (32, 33), DsrA (27), and probably MicC (26) but not OxyS (34). We sought to confirm this in our system and also to examine the effect of Hfq on the stability of the target mRNAs and their translation (in the absence of sRNAs, and therefore duplex formation). The concentrations of the sRNAs and target mRNAs were measured by quantitative RT-PCR in strains with and without chromosomal hfq that expressed only the sRNA or only the target mRNA. The target mRNAs were measured in strains that also lacked the chromosomal copy of the partner sRNA. We found that RyhB, DsrA, and MicC had reduced concentrations and that the OxyS concentration was unchanged in the absence of Hfq (Fig. 3A). This is in agreement with previous studies (26, 27, 32, 33). In contrast, the stability and translation of the target mRNAs were not substantially different with and without Hfq (Fig. 3 B–D). The expression of fhlA::gfp was very low; therefore, the relative change in fluorescence was not considered to be accurate (Fig. 3 C and D).

Open in a separate window

Fig. 3.

Effect of Hfq availability on sRNA and target mRNA concentrations and translation. Error bars indicate the SEM. (A and B) Relative sRNA and target mRNA levels in strains without Hfq as measured by quantitative RT-PCR. The sRNA and target mRNA concentrations were normalized to the 5S RNA and to the concentration in the strain with Hfq. The complementary sRNA is deleted in all strains in which the target mRNA::gfp fusion was measured. (C) Relative fluorescence of the samples shown in B. (D) Translational efficiency of target mRNAs in the presence and absence of Hfq. The fluorescence level of each target mRNA (C) was divided by its mRNA concentration (B) to determine the average amount of translation per target mRNA.

The absence of any change in the expression of the target mRNA fusions with the deletion of Hfq is unlikely to be due to the target mRNAs not binding to Hfq because coimmunoprecipitation studies with Hfq indicate that ompC and rpoS bind to Hfq (35), in vitro studies show that Hfq binds to the rpoS (36) and sodB (37) mRNAs, and our competition experiments also suggest that target mRNAs bind to Hfq (see below). Instead, the results indicate that the binding of target mRNAs to Hfq (in the absence of sRNAs) does not alter their degradation and translation rates. This is an important point for the competition experiments below because it indicates that altering the free Hfq concentration does not directly affect the expression of the target mRNAs; therefore, any changes in expression are the result of altered sRNA activity.

Loss of sRNA Activity Due to the Transcription of Unpartnered sRNAs and Target mRNAs.

Given that Hfq availability can be limiting, we investigated whether the transcription of competing sRNAs and target mRNAs that bind to and form complexes with Hfq can decrease its availability, and consequently decrease sRNA activity. To test this, we transcribed competing sRNAs and target mRNAs without their partner (i.e., “unpartnered”) and measured their effect on RyhB and MicC silencing and DsrA activation (Fig. 4A). For the competing target mRNAs, only the sequence necessary for sRNA activity was fused to the mCherry gene. mCherry was found to interact with rpoS::gfp; therefore, the first one-third of the T7 RNA polymerase gene (T7RNAP) was used instead of mCherry for the competition experiments with rpoS::gfp.

Open in a separate window

Fig. 4.

Competition for Hfq by unpartnered sRNAs and unpartnered target mRNAs. Error bars indicate the SEM. Gene fusions are abbreviated to the target mRNA name. (A) The experimental system. (B) The effect of competing sRNAs and target mRNAs on RyhB silencing of sodB::gfp. The competitors were transcribed from the constitutive pcon promoter, and the competing target mRNAs were fused to mCherry (RFP). #As an additional control, part of the T7 RNA polymerase gene (T7RNAP) was fused to RFP (instead of the target mRNA), and this showed no effect on RyhB silencing. *Samples measured separately with controls that enabled the fluorescence values to be calibrated across experiments. (C) The effect of competing sRNAs and target mRNAs on MicC silencing of ompC::gfp. The competitors were transcribed using a weaker promoter (pLtetO-1), which enabled sodB::mCherry to be transcribed without toxicity. The competing target mRNAs were fused to RFP. (D) The effect of competing sRNAs and target mRNAs on DsrA activation of rpoS::gfp. The competitors were transcribed from the pcon promoter and the competing target mRNAs were fused to the first third of T7RNAP. (E) The expression of additional Hfq in the RyhB-sodB::gfp system. All strains have chromosomal hfq. ◊ indicates strains with additional hfq on a plasmid under pLtetO-1 control.

We found that the activity of all sRNAs could be disrupted by at least one unpartnered sRNA or target mRNA. RyhB silencing of sodB::gfp was decreased (resulting in increased sodB::gfp expression) when unpartnered sRNAs (DsrA, MicC, and OxyS) and unpartnered target mRNAs (ompC::mCherry and fhlA::mCherry) were transcribed [Fig. 4B, compare level with unpartnered sRNAs (blue-shaded bars) and target mRNAs (pink-shaded bars) with the level without them (red dashed line)]. We demonstrated that this is due to competition and not to the unpartnered sRNAs and target mRNAs directly regulating sodB::gfp expression because expression does not increase in the absence of RyhB. Furthermore, nontarget mRNA sequences, which are not thought to bind to Hfq (mCherry and part of the T7RNAP gene) do not substantially effect RyhB silencing. Therefore, the observations are consistent with the unpartnered sRNAs and target mRNAs competing for Hfq.

MicC silencing also decreased (causing increased ompC::gfp expression) when an unpartnered sRNA (DsrA) or an unpartnered target mRNA (sodB::mcherry) was transcribed (Fig. 4C). Again, the unpartnered DsrA and sodB::mCherry did not directly increase ompC::gfp expression, and control sequences that are not thought to bind to Hfq did not interfere with MicC silencing. These results are also consistent with Hfq competition.

We observed that DsrA activation decreased when an unpartnered sRNA (MicC, OxyS, and perhaps RyhB) or target mRNA (ompC::T7 RNAP) was transcribed (Fig. 4D). Unfortunately, it is difficult to determine which of these unpartnered sRNAs and target mRNAs are competing for Hfq as opposed to acting directly on the rpoS::gfp target mRNA because the expression of rpoS::gfp is low without DsrA and further decreases in expression are difficult to measure. At least one of the sRNAs, OxyS, is known to bind directly to rpoS (34). The transcription of the T7RNAP sequence, which is not thought to bind to Hfq, did not affect DsrA activation.

Expression of Additional Hfq Reverses the Effect of a Competing Unpartnered sRNA.

If the observed loss of sRNA activity caused by the transcription of unpartnered sRNAs is due to Hfq competition, increasing Hfq should reverse the effect, causing sRNA activity to improve. In contrast, if the unpartnered sRNAs are causing loss of sRNA activity due to an off-target effect or by binding directly to the partnered sRNA or target mRNA, the loss of sRNA activity would be expected to be the same or worse with the additional Hfq. The hypothesis was tested in the RyhB-sodB::gfp system, where the transcription of unpartnered DsrA caused a profound loss of RyhB silencing. We found that the loss of RyhB silencing was largely reversed by increasing the production of Hfq, as predicted (Fig. 4E). This experiment provides compelling evidence to support the existence of competition for Hfq.

Dual Pathways for sRNA Activity.

In some sRNA-target mRNA pairs, both the sRNA and the target mRNA appear to compete for Hfq and impair sRNA activity. For example, MicC and its target mRNA (ompC) can separately decrease RyhB silencing of sodB::gfp as well as DsrA's activation of rpoS::gfp (Fig. 4 B and D). The binding of both the sRNA and its target mRNA to Hfq indicates the potential for duplexes to form via dual pathways: one where the sRNA binds first to Hfq, followed by target mRNA binding, and another where the target mRNA binds first to Hfq, followed by sRNA binding. This is consistent with other studies that have found the sRNA and target mRNA in the DsrA-rpoS (36) and RyhB-sodB pairs (37) can separately bind to Hfq. For generality, all sRNAs are considered to act via a dual pathway, although it is clear that the rate constants substantially favor one pathway over the other in some cases.

Data Collection and Analysis.

Single-cell measurements of GFP expression were collected with a Beckman–Coulter EPICS XL-MCL, and the data were analyzed using Flow explorer 4.1 (R. Hoebe, University of Amsterdam, Amsterdam, The Netherlands) and custom programs written with Matlab (MathWorks) software. Measurements were performed on exponentially growing cultures grown at 37 °C in LB with the appropriate antibiotic and concentration of IPTG (SI Materials and Methods).

Western Blotting.

Proteins were extracted from 4-mL aliquots of exponentially growing cells (≈0.3 OD₆₀₀) and then separated on 10% wt/vol Bis-Tris gel in Mes buffer. The gels were blotted onto PVDF membranes, blocked, and then incubated overnight with anti-Hfq using standard protocols (SI Materials and Methods). L9 was used as a loading control (43). Anti-Hfq and anti-L9 antibodies were kindly provided by Udo Blasi, Branislav Vecerek, and Isabella Moll (Max F. Perutz Laboratories, University of Vienna, Vienna). The secondary antibodies were donkey anti-rabbit IgG HRP and donkey anti-goat IgG-HRP. The antibodies were visualized with ECL Plus Western Blotting Detection Reagents and radiographic film (both from GE Healthcare Life Sciences). Digital images were captured by transillumination of the film using the Gel Doc XR imaging system, and the bands were quantified with Quantity One Analysis software (both from Bio-Rad). The Western blots were performed in triplicate.

We thank David Adamson for helpful discussions and comments. We are grateful to Rajaa Hussien for her assistance with the Western blots and Dr. George Brooks for allowing us to use his resources. This work was supported by University of California Berkeley Power Excellence Award W5646 and the Hellman Family Fund.