Expression of the DeaD RNA Helicase Is Regulated at Multiple Levels through Its Long mRNA 5′ Untranslated Region

ABSTRACT

DEAD-box proteins (DBPs) are a prominent class of RNA remodeling proteins that alter RNA structure, a process they typically perform through an ATP-dependent RNA helicase activity. Although many DBPs have been characterized at the structural and functional level in detail, much less is known about how they are regulated. We previously showed that the mRNA for the Escherichia coli DeaD DBP contains an unusually long 5′ untranslated region (5′ UTR) of 838 nucleotides (nt) and that it is the primary RNA determinant of DeaD autoregulation. We speculated that such a long and complex 5′ UTR might regulate deaD expression in additional ways. Here, we show that the deaD mRNA 5′ UTR regulates deaD expression at two additional levels, temperature-dependent expression and through a stem-loop structure overlapping the start codon. These results support the hypothesis that a long 5′ UTR can regulate gene expression through multiple mechanisms.

IMPORTANCE The expression of genes is frequently regulated by determinants in the 5′ UTR. Although many different regulatory mechanisms that operate via the 5′ UTR have been described, the functional relevance of genes with long UTRs is less clear. Here, we show that the 838-nt-long 5′ UTR in the deaD mRNA regulates the expression of DeaD at multiple levels. We propose that long UTRs originate to provide precise control of gene expression through multiple regulatory mechanisms, and they are indicators of the importance of their associated gene products for cellular adaptation to different environments.

INTRODUCTION

The expression of genes is regulated in all organisms, typically in response to changes in developmental stage, growth conditions, and/or environmental stresses. At the RNA level, the regulatory mechanisms frequently operate via transcription control, RNA stability, and translational control. Apart from the trans-acting factors involved, the cis-acting determinants that mediate RNA regulation often include sequences within the 3′ and 5′ untranslated regions (UTRs), with the latter more frequently associated with the regulation of gene expression in bacteria. Thus, many examples of genes that are regulated through the 5′ UTR via transcriptional mechanisms, the regulation of RNA stability, and translational control have been described (1–5).

DBPs are present in nearly all organisms and have been found to play important roles in RNA-related processes, including transcription, translation, RNA degradation, RNA export, and splicing (6, 7). Many DBPs have been functionally and structurally characterized, which has provided important insights into the mechanisms of RNA recognition and unwinding (8–10). In addition, DBPs have also been associated with additional RNA remodeling functions, such as the displacement of proteins bound to RNA and ATP-independent RNA annealing (11–13). In contrast, very little is known about how DBPs are regulated, a significant exception being that some bacterial DBPs are overexpressed during cold shock, i.e., when cells are rapidly shifted to low temperatures (14–16).

We recently found that the mRNA encoding DeaD, a DBP from E. coli, contains an unusually long 5′ UTR of 838 nt (4). Based on mutational analysis, we showed that 60 to 90% of transcription originates from the associated promoter under different conditions. To our knowledge, this represents the longest verified primary 5′ UTR identified for any E. coli mRNA. We also showed that DeaD self-regulates, or autoregulates, its production through a combination of Rho-dependent transcription termination and mRNA stability mechanisms (4). A deletion of the 5′ UTR abolished feedback control of a deaD-lacZ reporter fusion, suggesting that the cis-acting determinants that regulate Rho-dependent transcription termination and RNA stability both localize to the 5′ UTR.

Given its considerable size, we speculated that the deaD mRNA 5′ UTR might be involved in regulating deaD expression at additional levels. Here, we show that deaD mRNA levels and deaD-lacZ expression are highly sensitive to growth temperature and that this temperature-dependent expression requires the 5′ UTR. We also found evidence for a stem-loop structure in the 5′ UTR that overlaps the deaD translation initiation codon, and we show that individual residues within this region both positively and negatively regulate deaD expression. These findings indicate that the deaD mRNA 5′ UTR regulates deaD expression at several levels, suggesting that long 5′ UTRs, such as the one found in the deaD mRNA, arise to regulate gene expression through multiple mechanisms.

RESULTS

DeaD expression is reduced at elevated growth temperatures.

As mentioned above, the expression of several bacterial DBPs, including DeaD, is increased under conditions of cold shock or cold growth. This response likely reflects a need to stimulate DBP synthesis to resolve an increased population of RNAs that have become trapped in misfolded conformations at low temperatures. However, the question of how DBP concentration might be regulated above normal growth temperatures has not been addressed in much detail. Nonetheless, as part of a study to determine the effects of DeaD on global gene expression, it was noticed that expression of FLAG epitope-tagged DeaD was significantly reduced when cells were grown at 42°C (17). This observation is consistent with the hypothesis that some DBPs may become less important at higher temperatures because spontaneous thermal unwinding of their structured RNA targets can replace the helix-unwinding function of the DBPs (18, 19). To characterize this effect in greater detail, a wild-type (WT) strain was grown at 30°C, 37°C, or 42°C, and total RNA preparations made from these cultures were analyzed by Northern blotting (Fig. 1A). We found that the cellular deaD mRNA concentration was moderately affected between 30°C and 37°C but was significantly reduced at 42°C. These results indicate that the effect of temperature on deaD expression is manifested, at least in part, at the RNA level.

FIG 1.

Expression of the deaD mRNA and DeaD protein at different temperatures. (A) A WT strain was grown at 30°C, 37°C, or 42°C to the mid-exponential phase and harvested, and RNA was isolated. Equal amounts of RNA were analyzed by Northern blotting using a radiolabeled probe complementary to part of the deaD coding region. The full-length deaD mRNA is indicated by an arrow. 16S rRNA served as a loading control. Mean values and standard errors are reported based on three measurements. (B) The DM strain was analyzed at different temperatures as described in panel A. Expression levels were normalized to the levels of deaD mRNA in the WT strain grown at 30°C. (C) Protein extracts were prepared from cultures of the DM strain grown at 30°C, 37°C, or 42°C to the mid-exponential phase and analyzed by SDS-PAGE and dye-staining. The DM mutant DeaD protein is indicated by an arrow.

The reduced expression of deaD mRNA could be due to different reasons, including an effect on autoregulation. To determine whether the temperature-dependent effects are dependent on autoregulatory mechanisms, we analyzed the expression of a previously described double mutant (DM), which contains two mutations in conserved domains that inactivate DeaD function, as well as autoregulatory control (4). Like the WT mRNA, the expression of the mutant RNA was found to be reduced with increasing growth temperature (Fig. 1B), indicating that these effects are not likely to be mediated through effects on autoregulation.

To corroborate these effects at the protein level, we examined the effects of temperature on the expression of the DM protein. We had previously shown that the DM protein is expressed at high levels at 37°C, and it can be visualized directly by staining gels after electrophoresis (4). We repeated the protein analysis using cell extracts prepared from the DM strain after growth at 30°C, 37°C, or 42°C. A DM strain-specific band was clearly observable at 30°C and 37°C, but no such band was visible at 42°C (Fig. 1C), indicating that the amount of the DM protein is also significantly reduced at high temperatures, further confirming that temperature-dependent regulation of the protein occurs even when autoregulation is disrupted.

The deaD 5′ UTR mediates the temperature-dependent response.

To test whether the temperature dependence of deaD expression can be recapitulated in a genetic reporter, we used a previously described deaD-lacZ translational fusion (4) and measured its activity in WT and ΔdeaD strains after growth at different temperatures. As a control for any generalized temperature-dependent changes in gene expression, the activity of endogenous lacZ was measured separately. As was noted at the mRNA level, the expression of the deaD-lacZ fusion was reduced dramatically when the growth temperature was increased to 42°C, in both a WT and a ΔdeaD strain background (Fig. 2A). In comparison, endogenous lacZ activity was only marginally affected over this temperature range, suggesting that the temperature-dependent changes observed for the deaD-lacZ fusion are not due to any general changes in gene expression. Over the course of multiple experiments, we have found deaD-lacZ fusion activity in a WT strain background to vary between 40- and 130-fold over this temperature range on different days and between 10- and 50-fold using LB medium made from components obtained from alternate suppliers. The reasons for these differences are not clear, but nonetheless are consistent with a significant decrease in deaD-lacZ expression with increased temperature.

FIG 2.

Characterization of the temperature-dependent response of deaD-lacZ fusion activity (A) (Top) Schematic description of a deaD-lacZ fusion. The regions specific to the deaD and lacZ coding regions are shown by gray or black rectangles, respectively, and the deaD mRNA 5′ UTR is depicted by a line. (Bottom) β-galactosidase activity of the deaD-lacZ fusion in WT or ΔdeaD strain backgrounds at different temperatures, as well as of endogenous lacZ in a WT background. Mean values and standard errors are based on four measurements. (B) Relative expression of deaD-lacZ activity and deaD-lacZ mRNA levels at 30°C versus 42°C. (C) Plot of relative deaD-lacZ mRNA levels as a function of time after transcription arrest. Squares, data points for RNA isolated from cultures grown at 30°C; triangles, data points for cultures grown at 42°C. (D) β-galactosidase activity of the deaD-lacZ fusion at 30°C and 42°C and of constructs that lack different regions of the 5′ UTR.

To determine whether the decreased expression of the deaD-lacZ fusion can be attributed to changes in RNA levels or on translation, cultures of the deaD-lacZ strain were grown to the mid-exponential phase at either 30°C or 42°C and used to measure β-galactosidase activity. Separately, the cultures were used to make total RNA preparations. In these experiments we found an ∼130-fold decrease in the activity of the deaD-lacZ fusion, whereas the decrease in the levels of the deaD-lacZ mRNA, as determined by using quantitative reverse-transcription PCR (qRT-PCR), was only ∼3-fold (Fig. 2B). These results indicate that the primary determinant of the temperature-dependent response of deaD-lacZ activity is at the translational level. In contrast, in our previous work on autoregulation, we showed that both deaD mRNA levels and deaD-lacZ activity increase by 20- to 30-fold in deaD-mutant strains, indicating a minor role for translation in autoregulation (4), which implies that the mechanisms used by deaD to effect temperature-dependent regulation are very different from those used for autoregulation.

We also treated WT cultures with rifampicin for various periods of time before total RNA preparation and used qRT-PCR to measure deaD-lacZ mRNA degradation rates at 30°C and 42°C. We found that the half-life of the deaD-lacZ mRNA was 2.9 min at 30°C, similar to the 2.7 min half-life previously observed for the full-length deaD mRNA at this temperature (4), whereas the half-life at 42°C was reduced to 1.2 min (Fig. 2C). Thus, most of the temperature-dependent effects on deaD-lacZ mRNA levels can be attributed to changes in RNA stability.

We next tested whether the effects of temperature on deaD expression depend upon determinants within the 5′ UTR (Fig. 2D). For this purpose, the activity of the WT deaD-lacZ fusion, as well as of a deletion lacking the 5′ UTR sequence from 101 to 820 nt upstream of the initiation codon (Δ101-820), was measured in a WT strain at 30°C and 42°C. The first 100 nt of the deaD 5′ UTR were retained initially because, as we show later, this region contains additional determinants that affect deaD expression. Upon testing, the WT fusion was found to show a >50-fold decrease in activity between 30°C and 42°C, whereas the Δ101-820 construct displayed a much smaller, 7-fold effect over this temperature range. Thus, the deaD mRNA 5′ UTR plays a major role in regulating temperature-dependent DeaD expression. We next investigated whether there are any specific regions within the 5′ UTR that are responsible for regulation and, accordingly, tested three smaller nonoverlapping deletions that span the Δ101-820 deletion. Each of the smaller deletions reduced the extent of regulation, but none had as significant an effect as the Δ101-820 deletion. We conclude that the determinants for temperature-dependent regulation of deaD are distributed over the 5′ UTR, rather than being localized to a smaller region, and that they jointly contribute to regulation. Because the Δ101-820 deletion was still regulated by temperature, albeit to a significantly reduced extent, we made an additional construct that removes an even larger part of the 5′ UTR (Δ17-820). Measurements performed on this construct showed that the effects of temperature on deaD expression were reduced still further, indicating that the region from 17 to 100 nt of the deaD 5′ UTR also contributes to regulation. Because the Δ17-820 construct reduced the >50-fold effect of temperature on expression of the deaD-lacZ fusion to only 2-fold, we conclude that nearly all of the RNA sequences that mediate temperature-dependent regulation of deaD expression are contained within the 5′ UTR.

A stem-loop structure overlaps the deaD mRNA initiation codon.

To gain additional insights into the regulatory functions of the deaD mRNA 5′ UTR, we computationally folded the RNA to identify any potentially structured RNA regions. These analyses suggested that a region of the deaD mRNA that overlaps the start codon might be forming a stem-loop structure. To test this prediction, we applied selective 2′ hydroxyl acylation analysis by primer extension (SHAPE) by using N-methylisatoic anhydride (NMIA) to identify reactive single-stranded residues (20). We found that single- and double-stranded regions, as determined by NMIA treatment, were highly consistent with the structure predicted in silico (Fig. 3A and B).

FIG 3.

Structure of the mRNA region upstream of the initiation codon. (A) Chemical probing analysis of the deaD 5′ UTR. In vitro transcribed deaD mRNA was treated with NMIA, followed by primer extension analysis and gel electrophoresis. A sequencing ladder was loaded in parallel. (B) The structure of a stem-loop element overlapping the initiation codon (shown in bold characters), as predicted using Mfold (38). A putative Shine-Dalgarno (S/D) sequence (GAGG) is indicated. Single-stranded regions that were assigned based on NMIA reactivity are depicted by dots.

Identification of stem mutations that affect deaD-lacZ expression.

To determine whether the stem-loop structure overlapping the start codon affects deaD expression, we introduced a stem-disrupting mutation (G-39C) within the context of a deaD-lacZ gene fusion. We found that lacZ expression was increased by nearly 4-fold, suggesting that the stem inhibits deaD expression (Fig. 4A). To verify that the increased expression of the mutant is due to disruption of the stem, a complementary C-1G mutation was introduced. We found that the lacZ expression was significantly lowered for the G-39C + C-1G double mutant. We note that the C-1G mutation, by itself, decreased expression, which can be attributed to its proximity to the initiation codon. However, within the context of the C-1G mutation, the incorporation of the G-39C mutation reduced expression even further, consistent with the effects of the stem region in inhibiting deaD expression.

FIG 4.

Nucleotide changes in hairpin structure that affect deaD-lacZ expression. (A) Effects of a stem-disrupting mutation (G-39C) and of the compensatory C-1G mutation on the activity of the deaD-lacZ fusion. (B) Identification of mutations that affect deaD-lacZ fusion activity. Mutations that increase expression are shown in green, whereas those that reduce expression are shown in red. (C) β-galactosidase activity of mutants identified through screening. The β-galactosidase activities of the different mutants, as well as their relative activities, are shown. (D) β-galactosidase activity of the U-33C and A-37C mutations and of compensatory mutations that map to the complementary strand. (E) Effect of a Δrnc mutation on deaD-lacZ and endogenous lacZ expression.

To gain further insights into the determinants within this structure that regulate deaD expression, we created a library of single base pair mutants that map to one side of the stem spanning the region from 25 to 44 nt upstream of the initiation codon in the deaD-lacZ fusion construct. Phenotypic screening of the mutants resulted in the identification of several additional mutations that increased deaD-lacZ expression by a factor of 2-fold or more, as well as those that decreased expression by a similar degree (Fig. 4B and C). With one exception, mutations that increased deaD-lacZ expression mapped to a region from 36 to 40 nt upstream of the initiation codon within the lower stem region, which is defined as a region containing nine uninterrupted base pairs. Thus, it appears that deaD expression is negatively regulated mainly by the lower stem.

To further test the prediction that disruption of base-pairing is responsible for the increased expression of the mutants isolated by screening, we introduced complementary mutations in the context of U-33C and A-37C mutants. In each case, the addition of the complementary mutation decreased deaD-lacZ expression, confirming that disruption of base-pairing is primarily responsible for the effects of the mutations that increase deaD-lacZ expression (Fig. 4D).

Finally, we noted that the structure of the stem-loop region bears some resemblance to RNA targets that can be cleaved by the double-strand-specific RNase, RNase III (21). The typical substrates of RNase III contain a relatively long stem-loop structure interrupted by bulged nucleotides or internal loops, like the one observed in the deaD mRNA, suggesting that RNase III might also have a role in regulating deaD expression. To test this possibility, we measured deaD-lacZ activity in strains containing a WT or a Δrnc deletion allele (Fig. 4E). We also tested the activity of endogenous lacZ in strains containing WT or Δrnc alleles as a control for any generalized changes in gene expression. We found that the expression of the deaD-lacZ fusion was only minimally affected by rnc deletion and thereby conclude that the stem-loop structure in the deaD 5′ UTR is not a target for RNase III.

Analysis of stem mutations that increase deaD-lacZ expression.

To verify that the mutations that increase deaD-lacZ expression do so because they disrupt secondary structures that inhibit translation, we repeated lacZ assays on a subset of the mutations, in conjunction with RNA isolation followed by qRT-PCR. A plot of the relative lacZ expression of the mutants versus changes in RNA levels showed that the RNA levels were affected to a significantly lower extent than the increases in lacZ expression (Fig. 5A), which suggests that the effects of the mutations are primarily at the translational level rather than through changes in deaD-lacZ mRNA levels. The modest increases in RNA levels observed for some of the mutants may be due to the protective effects of translating ribosomes that block access to degradative RNases (22, 23).

FIG 5.

Characterization of mutants that increase deaD-lacZ expression. (A) Effects of mutations that increase deaD-lacZ activity on deaD-lacZ mRNA levels. The data points for the WT strain and the mutants tested are indicated. (B) Effects of growth temperature on the expression of WT and mutant deaD-lacZ fusion constructs.

Because deaD-lacZ expression was significantly regulated in a temperature-dependent manner (Fig. 2), and stem-loop stability is also expected to be temperature-dependent, we also tested the effect of stem-disrupting mutations on deaD-lacZ expression at 30°C and 42°C (Fig. 5B). We found that the extent of temperature-dependent regulation was reduced by up to 3-fold for most of the mutants, but nevertheless, each construct still exhibited a high degree of regulation. These findings are consistent with a comparison of the Δ101-820 and Δ17-820 constructs (Fig. 2D), which indicated that the region encompassing the stem-loop might only be regulating the temperature-dependent response 3-fold, with the major regulatory determinants located more than 100 nt upstream of the initiation codon.

Analysis of stem mutations that decrease deaD-lacZ expression.

Apart from mutations that increase deaD expression, we additionally found several mutants that map to the upper stem and reduced deaD expression by 2- to 10-fold (Fig. 4C). Single nucleotide mutations that reduce expression are commonly found within or near the translation initiation codon or S/D sequence, but the location of these mutants, from 25 to 35 nt upstream of the initiation codon, indicated the presence of unusual determinants within the 5′ UTR that regulate deaD expression. To determine whether the mutations affect deaD-lacZ translation or mRNA levels, we measured deaD-lacZ activity for a subset of the mutants and isolated total RNA in parallel to determine deaD-lacZ mRNA levels. We found that changes in mRNA levels were significantly smaller than their effects on deaD-lacZ mRNA activity (Fig. 6A), indicating that this set of mutants exerts its effects on deaD-lacZ expression primarily at the translation level, with possible secondary effects on deaD-lacZ mRNA abundance due to reduced levels of translating ribosomes.

FIG 6.

Characterization of mutants that decrease deaD-lacZ expression. (A) Effects of mutations that decrease deaD-lacZ activity on deaD-lacZ mRNA levels. (B) Effects of mutations that reduce deaD-lacZ expression and of compensatory mutations.

We also tested the possibility that the effects of the expression-lowering mutations might be related to the stability of the upper stem, and accordingly, for several of these mutations, we created mutations that could be expected to restore base pairing. For three of the combinations, which involve the A-35U, U-33A, and A-25U mutants, increased expression of the deaD-lacZ fusion was not observed, suggesting that the upper stem likely does not play a role in mediating the effects of mutations that downregulate deaD expression (Fig. 6B). For the fourth mutation, G-31A, the addition of the complementary C-12U mutation did increase expression, but since the C-12U mutation, by itself, did so as well, it is possible that the effects of the compensatory base-pair change reflect the combined effects of the individual mutations rather than synergistic effects caused by base pairing. We conclude that the mutations that decrease deaD expression most likely do so through direct effects on sequence rather than through changes in RNA structure.

DISCUSSION

Our earlier studies to characterize the determinants of deaD expression resulted in the surprising finding that the deaD mRNA contains an unusually long 5′ UTR of 838 nt. We showed that this 5′ UTR is required to mediate the effects of Rho-dependent transcription termination and RNA degradation on autoregulatory control, which help to maintain the expression of DeaD at a relatively low level under normal growth conditions. Based on the size of the 5′ UTR, we hypothesized that it might regulate deaD expression at additional levels. Here, we describe two ways the 5′ UTR does so.

First, we found that the levels of both deaD mRNA and protein are relatively similar when cells are grown at 30°C or 37°C, but there is a sharp decrease in their expression at 42°C. These observations suggested the presence of regulatory mechanisms that reduce DeaD expression at the higher temperature, possibly due to a reduced requirement for this factor under these conditions. These findings are consistent with our previous observations that the ribosome assembly and rRNA processing defects of multiply deleted helicase strains are attenuated when temperature is increased (24). In this context, it is notable that the decrease in deaD or deaD-lacZ mRNA levels from 30°C to 42°C ranged from 3- to 6-fold (Fig. 1A and 2B), whereas the corresponding decrease in deaD-lacZ activity in different experiments was found to range from 40- to over 100-fold (Fig. 2), which indicates a significant role for translational regulation. Consistent with this interpretation, a recent high-throughput study identified deaD as one of the genes found to display reduced translation efficiency upon heat shock (25).

The proposed mode of regulation is conceptually similar to one previously described for temperature-sensing RNA sequences, also referred to as “RNA thermometers,” some of which are found in genes involved in the response to high-temperature stress (26, 27). However, unlike most of the known RNA thermometers that stimulate gene expression when temperature is increased—for example, by unwinding RNA secondary structures that normally inhibit translation—the response for deaD is in the opposite direction. We propose that the deaD mRNA contains a novel RNA thermometer that dials down DeaD synthesis at elevated temperatures in response to a decreased requirement under such conditions. The sequences responsible for regulation appear to be included within the deaD 5′ UTR, as a large deletion in the UTR almost completely abolished the regulatory response (Fig. 2D). Efforts to further delimit the responsible regulatory elements to a smaller region were unsuccessful, however, as we found that even though each of three smaller deletions reduced the temperature dependent response, none had as great an effect as the larger deletion. One interesting finding from the analysis of deletion constructs is that most of the temperature-dependent response can be attributed to sequences that are more than 100 nt upstream of the initiation codon (Fig. 2D). To explain these observations, we suggest that the upstream sequences adopt local but metastable structures that are kinetically formed during transcription, which normally do not affect translation. When the temperature is increased, these structures become disrupted and form inhibitory long-range interactions with regions that are important for deaD translation. Thus, the mechanism by which the deaD mRNA 5′ UTR exerts its regulatory effects may be fundamentally different from many of the 5′ UTRs that have been characterized in detail so far, whose regulatory effects on gene expression have been found to be generally attributable to modular RNA elements.

A second role for the 5′ UTR was revealed through the identification of a hairpin motif that includes the initiation codon and neighboring residues important for translation. Through mutagenesis and screening, we identified several mutations, primarily in the lower stem, which increased deaD-lacZ expression, and we showed that compensatory mutations within a subset of these mutations restore the reduced levels of deaD-lacZ expression. Thus, the lower stem appears to function as a classical structural motif that negatively regulates gene expression. Unexpectedly, we also found several mutations in the upper stem region that dampened deaD-lacZ expression. We tested the hypothesis that the effect of these mutations might be linked to the structure of the upper stem but found that compensatory mutations did not generally restore expression to normal levels, which suggests that the effect of the mutations are due to changes in sequence rather than structure. As was true with temperature-dependent regulation, the effects of both the mutants that decrease deaD expression, as well as those that decrease it, were found to be primarily at the level of translation (Fig. 5A and 6A), in sharp contrast to a very limited role, if any, of translation in mediating the autoregulatory response (4). Thus, it appears that the deaD 5′ UTR has evolved to utilize a variety of different mechanisms to regulate its expression. It should be noted that the region encompassing the stem-loop region is included within a deaD-internal open reading frame (ORF) for yrbN, which encodes a 26-amino acid protein (28, 29), and consequently several of the mutations we introduced within the stem-loop region would also be expected to alter the ORF sequence. Given that short ORFs can affect the expression of downstream genes (30–32), it remains to be determined whether the mutations that map to the stem-loop region confer any effects on deaD expression via changes in the yrbN ORF.

In conclusion, the current evidence indicates that the deaD 5′ UTR affects deaD expression at multiple levels, which include autoregulatory and temperature-dependent responses, as well as through a stem-loop motif overlapping the initiation codon. Based on these findings, it cannot be ruled out that there are yet additional regulatory roles for the deaD 5′ UTR which remain to be identified. Even for the three levels identified so far, there appear to be further intricacies involved. Thus, autoregulation is manifested through a combination of transcription elongation and mRNA stability control; temperature-dependent regulation appears to occur at both the mRNA and the translational level; and individual nucleotides within the stem-loop region have either a positive or negative effect on deaD expression. Based on the complex layers of regulation exerted by the deaD 5′ UTR, we suspect its functional evolution has been driven by a need to provide exquisite control of deaD expression under a variety of conditions, which in turn suggests that regulated expression of deaD is critical for cells to adapt to different environments. We predict that the presence of long 5′ UTRs in other mRNAs will be diagnostic of a need to provide precise expression control of genes that are important for cellular function, in part by using long-range interactions, and these will likely turn out to be associated with multiple layers of regulation in order to achieve that goal.

MATERIALS AND METHODS

Strains.

Strains were derived from MG1655*, a derivative of MG1655 that contains a point mutation in rph that restores the reading frame of its prematurely terminated gene product (33). A strain containing a deaD-lacZ fusion on the chromosome has been described (4). Briefly, this fusion contains the deaD promoter, the entire 5′ UTR, and the first 33 codons of the deaD coding sequence translationally fused to the ninth codon of lacZ and integrated as part of a λ phage lysogen at the chromosomal att site. Strains in which rnc was inactivated were made by transduction of an rnc-14::Tn10 mutation into the recipient strains (34).

Mutagenesis and screening.

A strain containing a tetA-sacB cassette (35) was obtained from Donald Court (National Cancer Institute), DNA from which was used to amplify DNA encoding the tetA-sacB genes using primers with flanking sequences that map from 201 to 240 nt upstream of the deaD coding region and from 110 to 149 nt downstream of the lacZ coding region, respectively. The amplified PCR product was electroporated into a strain containing a deaD-lacZ fusion on the chromosome (4) and pSIM6 to express λ red function (36). Tetracycline-resistant lacZ⁻ colonies were selected, yielding strain CJ2208, which lacks 200 nt of the deaD 5′ UTR and 110 nt of the lacZ coding region. Independently, site-specific mutants mapping to the stem-loop region in the deaD mRNA upstream of the initiation codon were introduced into a plasmid that contains the deaD-lacZ fusion using the Q5 site-directed mutagenesis kit (New England Biolabs). Successful constructs were amplified using primers with the sequences 5′-CGCGTAACGATTTTTCGCAAGCG-3′ and 5′-CGACGACAGTATCGGCCTC-3′, and the product was electroporated into strain CJ2208. Cells containing a replacement of the tet-sacA cassette with the electroporated sequences and a restored deaD-lacZ fusion were selected by growing the cells on plates containing fusaric acid and sucrose, as described (35). The incorporation of the desired mutations was confirmed by Sanger sequencing (Genewiz). A mixed pool of mutants was made similarly using an equimolar mix of oligonucleotides that contained mutations at each of 20 positions spanning nt 25 to 44 upstream of the deaD initiation codon, which was followed by restreaking individual colonies growing on fusaric acid-sucrose plates onto LB agar plates containing X-gal and phenotypic screening to identify mutants displaying changes in lacZ activity.

Protein analysis.

Cell extracts were prepared and analyzed as described (4). Cultures for lacZ assays were grown in 1.5-mL volume, and β-galactosidase assays were performed on 3 to 4 biological replicates as described previously (37), using LB medium made from powder supplied by VWR International. The activity of endogenous lacZ was measured on MG1655* grown in the presence of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside). Chloramphenicol, to 100 μg/mL final concentration, was added to cultures grown at 42°C at the time of harvest to minimize any further translation during the subsequent handling steps that were performed at lower temperatures.

RNA analysis.

Northern blot and qRT-PCR analyses of deaD or deaD-lacZ mRNA were performed as described previously (4). Each qRT-PCR assay was performed four times. For deaD-lacZ mRNA decay measurements, rifampicin was added to 200 μg/mL for various periods of time before cultures were harvested for RNA isolation. For SHAPE analysis, the region of the deaD mRNA from 838 bp upstream of the initiation codon to 47 bp downstream was amplified using primers with the sequences 5′-AATTTAATACGACTCACTATAGGGTCTGAGGGCATTAGCGCG-3′ and 5′-GGAGCCTTCAGGCCCAG-3′, with a T7 promoter sequence appended to the upstream primer. Then, 300 ng of the DNA product was used for RNA synthesis using a HiScribe kit (New England Biolabs); 400 ng of the transcribed RNA and a 5′-end radiolabeled primer with the sequence 5′-GGAGCCTTCAGGCCCAG-3′ were heated to 65°C in reverse-transcription buffer (50 mM Tris-HCl at pH 8.3, 75 mM KCl, and 3 mM MgCl₂) and slowly cooled to 35°C. A one-tenth volume of dimethyl sulfoxide (DMSO) or 130 mM NMIA were added to 200 ng each of the folded RNAs, and the reaction mixtures were incubated at 37°C for 30 min. Primer extension was performed on the treated RNAs by adding deoxynucleoside triphosphates (dNTPs) to 0.5 mM and Moloney murine leukemia virus (MMLV) reverse transcriptase, and the primer extension products were resolved on a 10% polyacrylamide-8 M urea gel.