Hfq affects mRNA levels independently of degradation
© Le Derout et al; licensee BioMed Central Ltd. 2010
Received: 19 August 2009
Accepted: 18 February 2010
Published: 18 February 2010
Skip to main content
© Le Derout et al; licensee BioMed Central Ltd. 2010
Received: 19 August 2009
Accepted: 18 February 2010
Published: 18 February 2010
The bacterial Lsm protein, Hfq, is an RNA chaperone involved in many reactions related to RNA metabolism, such as replication and stability, control of small RNA activity and polyadenylation. Despite this wide spectrum of known functions, the global role of Hfq is almost certainly undervalued; its capacity to bind DNA and to interact with many other proteins are only now beginning to be taken into account.
The role of Hfq in the maturation and degradation of the rpsO mRNA of E. coli was investigated in vivo. The data revealed a decrease in rpsO mRNA abundance concomitant to an increase in its stability when Hfq is absent. This indicates that the change in mRNA levels in hfq mutants does not result from its modification of RNA stability. Moreover, a series of independent experiments have revealed that the decrease in mRNA level is not a consequence of a reduction of translation efficiency and that Hfq is not directly implicated in translational control of rpsO expression. Reduced steady-state mRNA levels in the absence of Hfq were also shown for rpsT, rpsB and rpsB-tsf, but not for lpp, pnp or tRNA transcripts. The abundance of chimeric transcripts rpsO-lacZ and rpsB-lacZ, whose expression was driven by rpsO and rpsB promoters, respectively, was also lower in the hfq null-mutants, while the β-galactosidase yield remained about the same as in the parent wild-type strain.
The data obtained suggest that alteration of rpsO, rpsT and rpsB-tsf transcript levels observed under conditions of Hfq deficiency is not caused by the post-transcriptional events, such as mRNA destabilization or changes in translation control, and may rather result from changes in transcriptional activity. So far, how Hfq affects transcription remains unclear. We propose that one of the likely mechanisms of Hfq-mediated modulation of transcription might operate early in the elongation step, when interaction of Hfq with a nascent transcript would help to overcome transcription pauses and to prevent preliminary transcript release.
Hfq is an RNA binding protein initially identified as a host factor required for the replication of the phage Qβ RNA . It was then demonstrated to belong to the Sm-like protein family involved in many RNA processing events in eukaryotes . The Hfq-encoding gene is widely conserved in bacteria and found in many sequenced bacterial genomes . Hfq is a highly abundant protein considered to act as a global regulator of gene expression [4, 5]. It has recently received much attention because of its crucial role in diverse cellular processes controlled by small non-coding RNAs (ncRNAs), where Hfq facilitates pairing of ncRNAs with their target mRNAs [6–9]. This explains, at least partly, why inactivation of the hfq gene causes pleiotropic phenotypes . Some of changes in the gene expression pattern are related to reduced translation efficiency of the rpoS mRNA, encoding the major stress sigma factor σS [4, 10], others to the induction of the σE-mediated envelope stress response [11–13] and the deficiency of the σH-mediated cytoplasmic stress response .
In addition, Hfq may affect some processes through its direct interaction with RNAs, e.g. with Qβ phage RNA during replication [14, 15] or with its own mRNA where it acts as a translational autorepressor [16, 17]. Hfq interactions with tRNAs [18, 19] and tRNA precursors  have also been reported.
Hfq function in the cell may also be mediated by protein-protein contacts. Hfq has been reported to interact with numerous proteins including ribosomal proteins, RNases, helicases, Rho-factor, RNA polymerase (in the presence of S1 protein), protein H-NS and poly(A)polymerase (PAP I) [21–25]. Related to the last observation, we have also shown that Hfq stimulates PAP I mediated synthesis of poly(A) tails by promoting the processivity of the enzyme and by protecting the poly(A) tails from exoribonucleolytic degradation [26–29]. Finally, Hfq has also been identified as a DNA binding protein that preferentially binds curved DNA and affects negative supercoiling [5, 30, 31]. However, in a cell, the majority of Hfq is located in the cytoplasm, presumably in association with the translation machinery, and only a minor fraction is associated with the nucleoid .
In the present work, our initial goal was to investigate the role of Hfq in the maturation and degradation of the rpsO mRNA of E. coli in vivo. The degradation pathway of the rpsO transcript, coding for ribosomal protein S15, is one of the best understood decay-pathways in E. coli. The genes encoding ribosomal protein S15 (rpsO) and polynucleotide phosphorylase (pnp) occupy adjacent positions and are oriented in the same direction on the E. coli chromosome. The two genes have their own promoters, P1 and P2 respectively, and can be expressed as monocistronic transcripts or as an rpsO-pnp dicistronic transcript. In this latter case, an endonucleolytic cleavage by RNase III produces a P1-RIII rpsO mRNA slightly longer than the P1-t1 rpsO monocistronic transcript [33, 34]. The initial step in the degradation of the rpsO mRNA is an RNase E cleavage that generates RNA molecules lacking the Rho-independent terminator, which are then rapidly degraded by the 3' to 5' exonucleases PNPase and RNase II [35–37]. The rpsO mRNA is stabilized by efficient translation because terminating ribosomes occlude the site for the rate-limiting RNase E cleavage located 10 nucleotides downstream of the translation termination codon . Since ribosomal protein S15 autoregulates its synthesis at the posttranscriptional level by repressing its own translation  and thereby decreasing the number of ribosomes translating the rpsO mRNA, it has been proposed that the coupling of the mRNA stability to translation allows the cell to adapt the amount of the rpsO mRNA to the need for ribosomal protein S15. Besides the RNase E-mediated pathway, exonucleolytic poly(A)-dependent degradation also plays a significant role, which becomes predominant when RNase E is inactivated [40–42]. Although the regulation of the rpsO gene expression was studied primarily at the posttranscriptional level, its transcription is most likely also modulated. However, while transcriptional start points have been identified precisely , transcriptional control remains largely unexplored.
Experiments were initiated to examine whether Hfq interferes with the rpsO mRNA decay mediated by RNase E or polyadenylation in vivo, as was previously shown in vitro . Unexpectedly, we found that Hfq deficiency induces a decrease in rpsO mRNA abundance concomitant to an increase in its stability. We present here a set of data suggesting that in the case of rpsO and some other cases, modulation of gene expression observed upon Hfq deficiency may result from changes in transcriptional yield.
Variation and stability of the rpsO, rpsO Δ and rpsT transcripts as a function of Hfq quantity.
rpsO mRNA (**)
rpsO Δ mRNA (**)
t1/2 (min) (*)
Relative RNA abundance at T = 0
t1/2 (min) (*)
Relative RNA abundance at T = 0
t1/2 (min) (*)
Relative RNA abundance at T = 0
Previous investigations suggested that Hfq could affect rpsO mRNA stability by modulating both the poly(A)-dependent and the RNase E-mediated degradation pathways. Besides its stimulating effect on the synthesis of poly(A) tails, Hfq was shown to protect poly(A) tails from exoribonucleolytic degradation in vitro and to affect the length and the frequency of oligo(A) tails in vivo [26–29]. Moreover, Hfq was found to protect the rpsO mRNA from RNase E cleavage in vitro . To evaluate the role of Hfq in the poly(A)- and the RNase E-dependent degradation pathways in vivo, we compared the stability of the rpsO transcript in hfq+ and hfq-deficient strains when either of these pathways was inactivated. For this purpose, we used the pcnB null-mutation to fully inactivate poly(A) polymerase and the rne 3071 allele allowing inactivation of thermosensitive RNase E at the non permissive temperature.
These data suggest that Hfq does not significantly affect the RNase E dependent degradation pathway, which accounts for RNA decay in the pcnB null-mutant . Moreover, if extended to other strains, this conclusion implies that the stabilization that occurs in hfq- cells may not result from the impairment of the RNase E-mediated pathway. It is therefore likely, that changes of rpsO mRNA stability and concentration described above require active PAP I; a prediction which prompted us to verify how Hfq affects the decay-rate of the rpsO transcript when thermosensitive RNase E was inactivated.
We have ascertained that polyadenylation contributes to degradation of the rpsO mRNA in the absence of RNase E. As expected, the rpsO P1-t1 transcript was significantly stabilized when thermosensitive RNase E was inactivated . Moreover, Hfq had a strong effect on the rpsO mRNA abundance. While the stability of the P1-t1 transcript rose from 6.99 +/- 0.14 min. to 11.18 +/- 0.14 min. the steady-state amount of the transcript was reduced 7.7 times in the absence of Hfq (Fig. 2B). These data reinforce the hypothesis above that Hfq deficiency has a stabilizing effect on the rpsO mRNA, which in theory should cause an accumulation of RNA. However, this stabilization is completely masked by a drop in mRNA concentration whose origin is independent of RNA stability. Moreover, they also confirm that the extent of the impact of Hfq on RNA stabilization and concentration depends on the presence of PAP I in the cell. It is worth noting again, that this effect is also observed in wild type strains where the degradation of the rpsO mRNA is mostly carried out by RNase E. It must be recalled here, that the RNase E and PAP I dependent pathways are somehow related and can substitute for one another .
Since mRNAs are generally protected by translating ribosome, we reasoned that the stabilization of the rpsO transcript in the Hfq deficient strains may result from an increase in translation efficiency. One could imagine that the autoregulation loop which adjusts rpsO translation efficiency to the need for S15, may convert the drop in mRNA resulting from Hfq deficiency into a stimulation of translation aiming to compensate for the poor levels of S15 synthesis. Alternatively, Hfq-mRNA interactions could directly affect translation or the autogeneous repression by S15.
These hypotheses led us to look at the behaviour of the constitutively translated rpsO transcript, which does not bear the operator recognizable by S15 and hence whose translation is independent of the S15 yield. This was achieved by using a plasmid-borne rpsO gene in which the first 122 nucleotides of the coding region were deleted, seven of these nucleotides being involved in the formation of the translational operator. This plasmid, referred to as pΔS15AUG, encodes a truncated S15 polypeptide unable to participate in the 30S assembly and to repress translation of the wild-type rpsO message . The stabilities of mRNAs originating from the chromosome (rpsO) and from the plasmid (rpsO Δ) were determined in hfq+ and hfq 1 backgrounds. As was shown previously in wild-type cells the constitutively translated rpsO Δ transcript was more stable than the native rpsO mRNA due to higher translation levels . The greater stability of the rpsO Δ transcript was also observed in the hfq mutant (Table 1). Moreover, just as in the case of the wild-type rpsO mRNA transcript originating from the chromosome, Hfq inactivation caused both a drop in the rpsO Δ mRNA level and an increase in its stability (Table 1). These data indicate that the consequences of Hfqdeficiency on mRNA level and stability do not rely on the autoregulation of S15 synthesis. Indeed, the observed stabilization of the unregulated rpsO Δ mRNA in the hfq mutant rules out the possibility that a drop in a functional rpsO mRNA concentration provokes increased translation efficiency through the autoregulation loop, resulting in stabilization of mRNA.
As one can notice (Fig. 3B), the reduction of the rpsO mRNA level in the hfq- strain is more pronounced in strains with empty vector (lanes 1 and 4) than for the pS15 bearing strains (lanes 2 and 5) indicating that the gene dose plays a role in this effect. As expected, the drop in the rpsO mRNA abundance was complemented by the plasmid pTX381 (pHfq), a pACYC 184-derivative expressing the wild-type hfq gene from its own promoter.
In order to determine whether the effect of Hfq deficiency on the rpsO mRNA could be also observed for other transcripts, we probed total RNA isolated from cells containing wt and mutant hfq alleles for rpsT, rpsB and lpp mRNAs and for the Leu1 and MetY2 tRNAs. While it has been reported that Hfq binds the rpsO mRNA [16, 29] and tRNAs [18, 19]in vitro, the other three transcripts were not examined. However, the fact that mRNAs from many ribosomal protein operons co-immunoprecipitate with Hfq  argues in favor of such a possibility for rpsB and rpsT. The Northern blots revealed that hfq inactivation did not modify lpp mRNA or tRNA abundance (Fig. 4). In contrast, the two rpsT mRNA species were less abundant in the hfq mutant, the stronger effect being observed for the P2-t1 rpsT transcripts initiated at promoter P2. Just as in the case of rpsO, the decreased level of rpsT transcripts was accompanied by a significant increase in their stability (Table 1).
The experiments described above show that Hfq deficiency may cause modifications of both mRNA abundance and stability, but these effects are not related. Indeed, stabilization of the rpsO mRNA takes place even in the absence of translational regulation, which presumably occurs in response to a drop in mRNA levels in the cell, and the decrease in mRNA concentration is accompanied by RNA stabilization expected to cause its accumulation. The requirement of PAP I activity for both stabilization and decrease of rpsO transcripts upon Hfq inactivation suggests that polyadenylation and Hfq act synergistically. The fact that mRNA stabilization in the absence of Hfq is only observed if PAP I is active might be explained by assuming that Hfq can no longer stimulate the elongation of oligo(A) extensions used as recognition sites by exoribonucleases. At the same time, the competition between Hfq and RNase E for the same site on the rpsO mRNA, which was observed in vitro , does not seem to play a role in the control of mRNA degradation under conditions of exponential growth in rich medium (growth conditions used in our experiments): indeed, the RNase E pathway of decay is not affected by Hfq inactivation. We cannot explain right now why Hfq has a significant impact on the rpsO mRNA abundance if PAP I is active. Nevertheless, the data above showing that the drop in rpsO mRNA concentration does not result from Hfq-mediated posttranscriptional events led us to the conclusion that it may reflect a negative impact of the loss of Hfq on transcription yield.
Recent microarray analysis revealed down-regulation of many ribosomal protein (r-protein) operons caused by lack of Hfq . Moreover, earlier work also reported the co-immunoprecipitation of Hfq with mRNAs of r-protein operon in hfq+ cells, and their general loss in hfq mutants . No plausible explanation for these effects was proposed. The implication of Hfq-dependent ncRNAs in positive regulation of house-keeping genes is very unlikely. The finding that a similar negative impact on abundance of the rpsO and rpsB transcripts (Fig. 3B and 4), on one hand, and on the corresponding reporter constructs rpsO-lacZ and rpsB-lacZ (Fig. 5), on another hand, implies that the observed reduction of the steady-state RNA level in hfq- strains may result from changes in promoter recognition or other early steps of transcription.
It is unlikely that Hfq discriminates promoters directly. However, there remains a possibility that Hfq may affect promoter recognition indirectly. The E. coli RNA polymerase holoenzyme is composed of the core enzyme (consisting of α2, β, β' and ω subunits) tightly bound with one of seven σ subunits which determine sequence specific contacts with promoter DNA. While the level of the housekeeping σ70 subunit is constant under all growth conditions, the intracellular concentrations of the six other σ subunits vary as a function of the growth phase, growth conditions and upon exposure to environmental stress . There is considerable evidence for competition between sigma factors for core RNA polymerase in vivo; therefore alterations in the level of one sigma factor may influence the competitor properties of others [46, 47]. At least two sigma factors, σS and σE, are affected by Hfq inactivation, indicating that the relative abundance of different holoenzymes varies between hfq+ and hfq- strains. Hfq deficiency decreases rpoS (σS) expression at the translational level [4, 48]. At the same time the rpoE mRNA encoding σE is up-regulated in a hfq mutant , thus enhancing the competitiveness of σE for the core RNA polymerase. However, the transcription of pnp as that of the tRNA genes, all of which are insensitive to Hfq deficiency (Fig. 4 and 7), are σ70-dependent, arguing against the possibility that competition between sigma factors is responsible for the down-regulation of the r-protein mRNAs. Alternatively, it is possible that Hfq-mediated modulation of transcription may occur at an early elongation step. Indeed, during the past decade, transcription elongation has been appreciated as a regulated phase of transcription. Regulation of transcript elongation occurs mainly due to transcriptional pausing that can be stabilized by several mechanisms: hairpin-dependent pausing (e.g. well-characterized his leader pause site), backtracked pauses, and σ-stabilized RNA polymerase stalling downstream from promoters [49–53]. As concerns the potential role of Hfq in regulation of transcript elongation via hairpin-dependent pausing, most pause signals have been found in the leader regions of certain amino acid biosynthetic operons. The reported co-immunoprecipitation of Hfq with mRNAs encoding leader peptides of these operons may be indicative of a role of Hfq in modulating attenuation . Moreover, this role should be positive, as microarray analysis revealed that lack of Hfq decreased the levels of mRNAs implicated in threonine (thrABC) or histidine (hisG, D, C, B, H) biosynthesis pathways regulated by attenuation .
The positive role for Hfq in transcription in vivo we propose here is consistent with a reported stimulatory effect of Hfq on the overall yield of transcription in vitro . Hfq may play a chaperoning role in co-transcriptional folding of a nascent RNA chain, counteracting transcription pausing or arrest and preventing premature release of the transcript. Biologically important RNAs often solve their folding problem using the assistance of chaperone and cofactor proteins . In this respect, it should be mentioned that folding of long structured mRNA leaders of the rpsO and rpsB operons plays a crucial role in regulation of their expression [43, 55]. In addition to the folding problems, transcriptional pauses within the mRNA leaders of r-protein mRNAs may also be stabilized by rebinding of σ70 to the elongating RNA polymerase, when -10-like elements located downstream from promoters are exposed in transcriptional bubble [50, 52, 53]. Indeed, -10-like elements followed by three GC base pairs (consensus for σ70-dependent pausing) are easily recognized within the rpsO (TACACTGGG, positions +40 to +48 from transcription start) and rpsB (TAATATGGG, positions +110 to +118) mRNA leaders. Taking this into account, it is tempting to suppose that Hfq may play a positive role in the co-transcriptional folding and/or maintaining of the nascent RNA chain during transcriptional pauses, preventing transcription arrest and formation of abortive elongation complex. If this is the case, the portion of abortive transcription on the pause-inducing sites should increase in the absence of Hfq in the cell, leading to a reduced steady-state level of the mRNAs produced. The mRNA transcripts that escape abortive events will be less abundant but more stable either because they are translated more efficiently to produce sufficient amount of r-proteins for ribosome biosynthesis, or due to inefficiency of poly(A)-dependent degradation pathway mediated by Hfq. We believe that the proposed here potential role of Hfq in maintaining nascent transcripts to prevent premature transcript release at the pausing sites is the most plausible explanation for the observed down-regulation of r-protein mRNAs in hfq null-mutants.
We present here several lines of evidence indicating that rpsO, rpsB, rpsB-tsf and rpsT mRNAs are down-regulated in hfq null mutants and that the reduction of the mRNA levels upon Hfq deficiency cannot be explained by destabilization of mRNAs and relates rather to changes in transcription efficiency. Our results provide an explanation for the recent microarray analysis which has revealed down-regulation of several ribosomal protein (r-protein) operons caused by the lack of Hfq. Taken together, Hfq appears to have a much greater role in RNA metabolism than previously anticipated involving all the steps of an RNA molecule's "life", from its synthesis to its degradation.
The name and origin of the strains used here are listed in Table 2. A DNA fragment comprising the rpsO regulatory regions (rpsO promoter and translation initiation region, positions from -188 to +58 with respect to A+1 in the initiator ATG) was amplified from the genomic E. coli DNA and inserted into pEMBLΔ 46/BamHI, HindIII in frame with the lacZ coding sequence . Primers used for PCR amplification were: PrpsO-for 5'-
Escherichia coli strains and plasmids used in this work
Strain or plasmid
HfrHlacZ43 relA1 spoT1 thi-1
IBPC633 hfq1::Ω(kanR, Bcl I)
N3431 hfq1::Ω(kanR, Bcl I)
N3433 hfq1::Ω(kanR, Bcl I)
N3433 hfq V43R cycA30::Tn10
N3433 hfq Δ22-294 cycA30::Tn10
IBPC903 hfq Δ22-294 cycA30::Tn10
ENSO rpsB'-'lacZ bearing deletion in
rpsB TIR from -72 to -69
pCL1921 derivative expressing ΔrpsO
pACYC184 derivative expressing rpsO
pGEM3 derivative expressing hfq
pACYC184 derivative expressing hfq
The plasmid pS15 was created by cloning the whole rpsO gene flanked with its own promoter and terminator into BamHI and HindIII sites of pACYC184. The DNA fragment to be cloned was generated by PCR on E. coli genomic DNA using rpsO-for corresponding to the positions (-149) to (-126) relative to the rpsO translation start and bearing the BamHI site; 5'-CAGGGATCC GTCTTGCGATAACAG and rpsO-rev, complementary to the positions from +303 to +325 of the rpsO mRNA and bearing the HindIII site 5'-CCGTAAGCTT GAAAAAAGGGGCC.
Construction of strains LABrpsB208::lacZ carrying chromosomal rpsB'-'lacZ fusion with the rpsB portion comprising the rpsB promoter and TIR (positions from -208 to +41 relatively the initiator ATG codon) was described . Here, we used a derivative of this strain, LABrpsB208::lacZ ΔGGGU bearing a small deletion in the rpsB 5'-untranslated region (from -72 to -69), which abolishes autoregulation . The hfq Δ allele  was transferred into this derivative by P1 transduction to generate hfq+ anf Δhfq variants.
In the case of IBrpsO188::lacZ and LABrpsB208::lacZ, cell cultures were grown at 37°C in LB medium supplemented if necessary with tetracycline (12 μg/ml) or chloramphenicol (35 μg/ml), in the absence of IPTG (lac-promoter is closed, rpsO or rpsB promoters are active). Cells were harvested in exponential phase (OD600≈0.4-0.5). The β-galactosidase activity was measured in clarified cell lysates  and expressed in nmol ONPG (o-nitrophenyl-b-D-galactopiranoside) hydrolysed per minute per milligram of total soluble proteins. Protein concentration in lysates was measured by Bradford assay (Bio-Rad).
Cellular proteins (20 μg) were separated on a 12% SDS-PAGE gel and analyzed with the use of 125I-iodinated A-protein as described in . Polyclonal rabbit antibodies raised against β-galactosidase (ICN Biomedicals, Inc.) and PNPase (kindly provided by A. J. Carpousis) were used.
RNAs were prepared form bacteria grown to an A650 = 0.35-0.4, according to . Templates for the synthesis of rpsO and rpsT RNA probes were obtained by PCR amplification as described . The lacZ template was obtained using TAATACGACTCACTATAGGG ATACTGACGAAAC and GCCGTCGTTTTACAACGTC, the lpp probe using TAATACGACTCACTATAGGG TATTTAGTAGCCATGTTG and GTTCTACTCTGCTGGCAG, and the rpsB probe using TAATACGACTCACTATAGGG TTCACGAAGAACTGGTCG and ATGGCAACTGTTTCCATGCG oligonucleotides as primers. The upstream primer includes the T7 promoter indicated in italics. RNA probes were synthesized by T7 RNA polymerase yielding uniformly labeled RNAs with [α-32P] UTP . 5S rRNA, tRNALeu1 and tRNAMetY were probed with the 5'-labelled oligonucleotides 5'-ACTACCATCGGCGCTACGGC, 5'-CCCCCACGTCCGTAAGGACA and 5'-CGGGTTATGAGCCCGACGA, correspondingly.
Total RNA separation on 1% agarose formaldehyde gel or acrylamide/urea gel and Northern blotting were previously described [34, 35]. RNA levels were quantified using a PhosphoImager, and values were corrected for variations in RNA loading by hybridizing the same blot for the 5S rRNA.
We are grateful to A. Kolb and C. Condon for discussion and reading the manuscript. We are indebted to C. Pennetier who contributed to this work at early stage. We thank M.E. Winkler and A.J. Carpousis for supplying plasmids and antibodies. This work was supported by the Centre National de la Recherche Scientifique (UPR 9073) and Paris VII University (France). IB was supported by University Paris VII as a visiting scientist and by RFBR grant 06-04-48353.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.