Global effects of the DEAD-box RNA helicase DeaD (CsdA) on gene expression over a broad range of temperatures

Summary

In Escherichia coli, activity of the global regulatory RNA binding protein CsrA is antagonized by two noncoding sRNAs, CsrB and CsrC, which sequester it away from its lower affinity mRNA targets. Transcription of csrB/C requires the BarA-UvrY two component signal transduction system, which responds to short chain carboxylates. We show that two DEAD-box RNA helicases, DeaD and SrmB, activate csrB/C expression by different pathways. DeaD facilitates uvrY translation by counteracting the inhibitory effect of long distance basepairing between the uvrY mRNA leader and coding region, while SrmB does not affect UvrY or UvrY-phosphate levels. Contrary to the prevailing notion that these helicases act primarily at low temperatures, DeaD and SrmB activated csrB expression over a wide temperature range. High-throughput sequencing of RNA isolated by crosslinking immunoprecipitation (HITS-CLIP) revealed in vivo interactions of DeaD with 39 mRNAs, including those of uvrY and 9 other regulatory genes. Studies on the expression of several of the identified genes revealed regulatory effects of DeaD in all cases and diverse temperature response patterns. Our findings uncover an expanded regulatory role for DeaD, which is mediated through novel mRNA targets, important global regulators and under physiological conditions that were considered to be incompatible with its function.

Introduction

Post-transcriptional regulation plays a major role in coordinating the gene expression networks that dynamically determine the biological phenotype. Two well studied global regulatory systems serve this purpose in many bacteria: The RNA chaperone Hfq mediates pairing of mRNAs with small noncoding RNAs (sRNAs), while the RNA binding protein CsrA (RsmA) recognizes conserved sequences in the leader and/or N-terminal coding region of mRNAs, resulting in altered mRNA translation and/or stability (Storz et al., 2004, Balasubramanian & Vanderpool, 2013, Romeo et al., 2013). We propose that DeaD and perhaps other DEAD-box RNA helicases also post-transcriptionally regulate bacterial gene expression on a global scale and that the current understanding of the roles of these proteins represents only the `tip of an iceberg'.

With rare exceptions, RNA helicases are ubiquitous in living organisms, where they mediate ATP-dependent effects on transcription and translation, RNA turnover, remodeling of RNA-protein complexes, RNA splicing, and other processing mechanisms that involve structured RNA molecules (Cordin et al., 2006, Owttrim, 2013). Bacteria typically contain several proteins from the largest group of RNA helicases, the DEAD-box helicases, so named for the conserved amino acid motif D-E-A-D (Iost & Dreyfus, 2006). Despite the fact that DEAD-box helicases have been linked to bacterial virulence factors such as the type-three secretion system (T3SS), quorum sensing, and biofilm formation (Yang et al., 2008, Field et al., 2008, Tu Quoc et al., 2007), the extent to which they influence gene expression and their molecular mechanisms remain largely unexplored.

E. coli contains five DEAD-box RNA helicases, which facilitate ribosome biogenesis at low temperature (DbpA, DeaD, RhlE, SrmB) and RNA turnover (RhlB) (Iost & Dreyfus, 2006). RhlB is part of the RNase E-dependent RNA degradosome and assists in the turnover of RNAs containing stable secondary structure (Carpousis, 2007). The remaining helicases play independent roles in 50S ribosomal subunit assembly, particularly at low temperature (Cordin et al., 2006). DbpA and its orthologs are the only bacterial DEAD-box helicases known to bind RNA with sequence specificity (Fuller-Pace et al., 1993, Wang et al., 2006). However, a dbpA mutation has little effect on growth or the ribosome maturation profile, thus its biological function is still uncertain (Iost & Dreyfus, 2006). In contrast, the DEAD-box proteins DeaD or CsdA (cold-shock DEAD-box protein A) and SrmB have strong effects on the ribosome maturation profile and on growth at low temperatures (Iost & Dreyfus, 2006). SrmB and DeaD appear to act at different steps in the production of 50S subunits. While the specific role of RhlE has yet to be determined, it is thought to participate in 50S ribosome biogenesis (Iost & Dreyfus, 2006).

Although RhlB has a well-established role in mRNA turnover as a component of the degradosome, the helicases RhlE, DeaD, and SrmB have also been reported to associate with RNase E (Prud'homme-Genereux et al., 2004, Khemici et al., 2004). These latter interactions occur only at low temperature extremes and their physiological consequences are uncertain. Despite their important roles at low temperatures, a strain lacking all five helicases only exhibits a modest growth defect at 37 °C in rich medium and no growth defect in minimal medium (Jagessar & Jain, 2010). Thus, the bacterial DEAD-box helicases are presently regarded as being primarily important for overcoming the consequences of increased RNA secondary structure stability under low temperature conditions.

While no role beyond ribosome biogenesis is known for SrmB, a few reports have implicated DeaD helicase in the control of mRNA turnover and translation. However, the biological relevance of most of these studies is limited by the analysis of synthetic regulatory targets, DeaD overexpression phenotypes, or the transcription of mRNA targets using T7 bacteriophage RNA polymerase, which possesses an unusually rapid elongation rate (Iost & Dreyfus, 1994, Butland et al., 2007). A recent study demonstrated that translation of the stationary phase and general stress response sigma factor RpoS is controlled by DeaD helicase under low temperature conditions and that DeaD modestly activates in vitro translation of rpoS (Resch et al., 2010). How this regulation occurs, its effect on RpoS-dependent gene expression, and whether translation control is a major theme for DeaD remain to be demonstrated. A role for DeaD in the expression of heat shock genes may be mediated through effects on rpoH mRNA (Jones et al., 1996).

Here, we discovered that two DEAD-box RNA helicases of E. coli influence the carbon storage regulatory (Csr) system. CsrA, an RNA-binding protein at the heart of the Csr system, globally regulates gene expression through effects on mRNA translation and turnover (Romeo et al., 2013). CsrA activity in turn is regulated by the sRNAs CsrB and CsrC, which use molecular mimicry to sequester CsrA away from its mRNA targets (Romeo et al., 2013). The BarA-UvrY two-component signal transduction system (TCS) activates transcription of these sRNAs (Fig. 1A). We found that DeaD and SrmB regulate csrB expression by distinct processes. DeaD activates uvrY mRNA translation and enhances UvrY levels in the cell. SrmB activates csrB expression by an unknown mechanism, which does not involve changes in the levels of UvrY or its active, phosphorylated form (UvrY-P). Contrary to the notion that DeaD and SrmB function principally at low temperatures, we found that they control gene expression over a wide range of temperatures, including temperatures relevant for growth within mammalian hosts. Studies of uvrY and other mRNA targets of DeaD demonstrated that DeaD and perhaps other bacterial DEAD-box helicases serve a much greater role in posttranscriptional regulation than has been realized.

Fig. 1.

DEAD-box RNA helicases DeaD and SrmB affect csrB-lacZ expression. (A) A model for the Csr regulatory circuitry that includes DeaD and SrmB functions. A broken line depicts an unidentified mechanism. Short-chain fatty acids (SCFA). (B) Transposon insertions in the DEAD box RNA helicase genes deaD and srmB affect csrB-lacZ expression in the screening strain CR1-93. Transposon insertion mutant strains barA∷kan, uvrY∷kan, deaD∷kan or srmB∷kan were assayed for β-galactosidase activity in mid-exponential growth phase. Mutant phenotypes were complemented with DeaD (pDeaD), SrmB (pSrmB), or control (pBR322) expression plasmids. (C) Expression of a csrB-lacZ transcriptional fusion in the indicated MG1655 derivatives. Open and solid symbols depict growth and β-galactosidase activity, respectively.

Results

DeaD and SrmB activate csrB expression

We began this study with a transposon mutagenesis screen for novel regulators of csrB expression. While our laboratory and others have used mutagenesis screens successfully for this purpose, a complication of these screens has been that the effects of regulatory mutations affecting csrB expression tend to be counteracted by a negative feedback loop through the BarA/UvrY pathway (Fig. S1 and Suzuki et al., 2002). To improve the sensitivity of our genetic screen, we eliminated the feedback loop by replacing the csrB promoter region of the natural chromosomal locus, with the constitutive lacUV5 promoter. In this strain background (CR1-93), mutations that affect csrB-lacZ (located at att_λ) expression, e.g. in the BarA/UvrY pathway, no longer affect CsrB RNA levels (Fig. S1). Among the mutations identified were insertions in the deaD and srmB genes, which encode DEAD-box RNA helicases (Fig. 1B). Transposon insertions in either deaD or srmB result in a 6-fold decrease in csrB-lacZ expression. These phenotypes were complemented by ectopic expression of the deaD and srmB genes, respectively (Fig. 1B). Deletions of srmB and deaD in the wild-type strain MG1655 greatly reduced the expression of a csrB-lacZ transcriptional fusion in the exponential phase of growth and modestly in the stationary phase (Fig. 1C). Thus, DeaD and SrmB also affected csrB expression in a strain with an intact BarA/UvrY feedback loop. Consistent with this observation, deletion of deaD or srmB dramatically reduced CsrB RNA in the early to mid-exponential phase of growth and modestly in the stationary phase (Fig. 2A). Deletion of both deaD and srmB resulted in undetectable CsrB levels until the culture reached the late exponential phase of growth (Fig. 2A).

Fig. 2.

DeaD and SrmB affect CsrB RNA throughout the growth curve. (A) Northern blot of CsrB levels in strain MG1655 (WT) containing ΔdeaD, ΔsrmB, or ΔdeaD ΔsrmB mutations. (B) Northern blot for CsrB RNA in DEAD-box RNA helicase deletion mutants. All mutant strains in this panel contain deletion mutations of the indicated gene(s).

Previous reports have suggested that DeaD, SrmB, and perhaps other DEAD box proteins in E. coli may have overlapping or redundant activities (Iost & Dreyfus, 1994). To test this hypothesis, CsrB was measured in a set of all DEAD-box helicase single and double deletion mutants. CsrB levels were greatly affected only in the strains with deaD or srmB deletions, although modest effects were observed for the dbpA rhlB and rhlB rhlE double mutant strains (Fig. 2B and S2). The deaD and srmB deletions also decreased CsrC levels, albeit modestly (Fig. S2 and data not shown). These findings correlate with previous reports showing that expression of csrB is more highly dependent on UvrY than that of csrC (Weilbacher et al., 2003), and hinted that DeaD and SrmB function through the BarA-UvrY TCS.

DeaD and SrmB regulate UvrY-dependent CsrB expression

An epistasis experiment confirmed that neither helicase affected CsrB levels in the uvrY mutant strain (Fig. 3A). We then tested for the effects of helicase deletions on proteins that are known to be involved in csrB/C expression (Figs 1A and 3B). None of the helicases affected CsrA, BarA_FLAG, or RpoB (control) levels, and only the deaD deletion decreased UvrY_FLAG levels (Fig. 3B). No further reduction in UvrY_FLAG levels resulted upon pairing ΔdeaD with other helicase deletions (Fig. 3B). Note that the recombinant BarA_FLAG and UvrY_FLAG fully complemented the defects caused by a barA or uvrY deletion on CsrB RNA levels, respectively, implying that they function normally within the Csr regulatory circuitry (Fig. S3). Although Hfq is frequently involved in posttranscriptional regulation (Lalaouna et al., 2013, Storz et al., 2004) and is a regulator of csrB expression (Suzuki et al., 2006), DeaD and SrmB still strongly affected csrB expression in an hfq deletion strain (data not shown).

Fig. 3.

Effects of DEAD-box RNA helicases on Csr system components. (A) Northern blot of CsrB levels in wild type (WT), ΔuvrY (uvrY), ΔuvrY ΔdeaD (uvrY deaD), ΔuvrY ΔsrmB (uvrY srmB), or ΔuvrY ΔdeaD ΔsrmB (uvrY deaD srmB) mutant strains. The hyphen (−) indicates RNA collected from a ΔcsrB strain. (B) Western blot depicting effects of DEAD-box RNA helicase deletions from mid-exponential phase cultures on the proteins that influence csrB expression, and the RpoB control. The hyphen (−) depicts proteins from the untagged strains (upper images) or the csrA∷kan mutant (lower images). (C) Effects of ΔbarA, ΔdeaD and ΔsrmB on the in vivo percentage of phosphorylated UvrY_FLAG. Exponential phase extracts were fractionated in the presence of Phos-tag™ reagent and analyzed by Western blotting. Exposure of UvrY and UvrY-P images were individually optimized for quantitation. UvrY_FLAG protein levels relative to those in the wild type strain (WT) and the percentage of UvrY_FLAG-P vs. UvrY_FLAG total (% phosphorylation) are given. The hyphen (−) depicts the untagged strain.

Because uvrY is required for regulation of CsrB by both DeaD and SrmB, we further examined the effects of deaD and srmB mutation on UvrY-P levels (Fig. 3C). While deletion of the sensor kinase (barA) strongly decreased UvrY-P fractional levels (from 12% to 1% of total UvrY), minimal or no effects on fractional UvrY phosphorylation occurred with deletions of deaD, srmB, or both genes. Thus, the effect of SrmB on csrB transcription appears to involve unknown factor(s). Because DeaD and SrmB activate CsrB expression through different mechanisms, each is worthy of investigation. However, because SrmB effects most likely involve unknown factor(s), the remainder of this study focuses on the role of DeaD.

Effects of DeaD on CsrA regulon members

To determine if DeaD affects the expression of genes within the CsrA regulon, we used two gene fusions that are repressed by CsrA, pgaA′-′lacZ and PlacUV5-nhaR′-′lacZ (Wang et al., 2005, Pannuri et al., 2012). Deletion of deaD decreased the expression of both fusions (Fig. S4A and B). In addition, negative effects of DeaD were observed in the absence of CsrB/C (Fig. S4C and D), suggestive of complex regulatory circuitry, with more than one input from DeaD into the expression of these genes. Thus, we calculated the effects of DeaD that are transmitted specifically through CsrB and CsrC by correcting for expression in the ΔcsrB ΔcsrC genetic background (Fig. S4E and F). The results confirmed strong downstream activation of pgaA and nhaR genes by DeaD via its effects on CsrB/C.

Effects of DeaD require the uvrY untranslated mRNA leader and proximal coding region

While UvrY_FLAG levels were regulated by DeaD (Fig. 3B), a translational fusion containing the promoter region, untranslated leader, and first 12 codons (uvrY'-'lacZ) was not affected (data not shown). Furthermore, DeaD did not affect UvrY_FLAG protein stability (data not shown). To assess the genetic requirements for regulation, we constructed and analyzed uvrY_FLAG fusion derivatives that replaced the promoter or promoter and leader with those from araB (Fig. 4A and B). This experiment revealed that regulation by DeaD requires the uvrY noncoding leader but not the promoter region.

Fig. 4.

Activation of uvrY expression by DeaD depends on base pairing between the uvrY mRNA leader and coding segments. (A) Gene fusions used in this study: (i) native uvrY regulatory region and coding sequence containing a FLAG® tag, (ii) araB promoter and mRNA leader fused to the uvrY_FLAG coding sequence, (iii) araB promoter fused at the transcription start site to the uvrY leader and uvrY_FLAG coding sequence, and (iv) lacUV5 promoter fused at the transcription start site to the uvrY mRNA leader and N (12, 15, 20, 22, or 51) uvrY codons fused to lacZ. (B) Effects of ΔdeaD on expression of uvrY_FLAG constructs (panel A. i, ii, iii). Strains were grown in the presence of arabinose, and UvrY_FLAG was detected by Western blotting. (C) Effect of ΔdeaD on expression of P_lacUV5-uvrY'-`lacZ reporter fusions (panel A, iv). (D) Structure prediction for the uvrY leader and minimal uvrY coding sequence that responds to DeaD. (E) Effects of DeaD on chromosomal P<sub>lacUV5</sub>-uvrY'-`lacZ fusions containing single and compensatory base changes (panel D). The values depict DeaD-dependent activation (β-galactosidase expression in deaD⁺ / ΔdeaD strains). The dashed horizontal line shows activation of the wild type (WT) fusion. Fig. S5 shows raw data.

To test the idea that regulation requires sequence from the uvrY coding region that was missing in the original uvrY'-`lacZ translational fusion which ended at codon 12 (data not shown), we measured the expression of reporter fusions containing the leader and 12, 15, 20, 22, or 51 codons of uvrY in wild type and ΔdeaD strains (Fig. 4A and C). The expression of fusions with 22 or 51 codons was decreased by the deaD deletion, 2.8 and 5 fold, respectively, while fusions with ≤ 20 codons of uvrY were unaffected (Fig. 4C). To predict whether the segments of uvrY mRNA required for regulation form secondary structure that might serve as a target site for DeaD helicase we used mfold analysis (Zuker, 2003). Indeed, the most stable structure exhibited extensive pairing between nucleotides 13–29 of the leader and nucleotides 92–111 (codons 15–21) of the coding region (Fig. 4D).

To test for the possible role of uvrY mRNA secondary structure in regulation by DeaD, a series of single mutations was constructed in the most extensively paired region, along with compensatory mutations that were predicted to restore pairing (Fig. 4D). Point mutations in nucleotides 25–28 or their pairing partners 96–93 eliminated DeaD helicase activation (Fig. 4D and E), while compensatory mutations (T25A-A96T, C26G-G95C, C27G-G94C, and C28G-G93C) fully or partially restored regulation (Fig. 4D and E). In contrast, mutations in nucleotides flanking this core region either had modest or no effects on DeaD activation (T23A, A24T, T97A, and A92T) or were not activated by DeaD when pairing was restored (T29A-A92T). These flanking nucleotides might not pair, may not be required for DeaD function if critical bases of the core region are paired or may have more complex roles in regulation, perhaps involving sequence requirements (Figs 4D and E and S5). The results indicate that uvrY expression is inhibited by secondary structure formed between the uvrY mRNA leader and coding region and that DeaD helicase counteracts this inhibition.

Because elongating ribosomes possess intrinsic helicase activity, and can bypass RNA secondary structure (Qu et al., 2011), we also suspected that inhibitory base pairing between the uvrY coding segment and leader may be favored by suboptimal translation initiation. Consistent with this hypothesis, uvrY translation uses a suboptimal TTG start codon (Maillet et al., 2007). Mutation of this TTG to an optimal ATG codon increased expression and decreased the effect of DeaD from 4-fold to 2.5-fold (Fig. S6). Thus, suboptimal translation initiation accentuates the regulatory effect of DeaD. Mfold analysis predicted pairing between the Shine-Dalgarno (SD) sequence and an adjacent anti-SD sequence, which should also reduce translation (Fig. 4D). Though we tested point mutations in the anti-SD sequence, interpreting their effects on regulation by DeaD was complicated because all of the mutations were predicted to cause RNA structural rearrangements and compensatory mutations in the SD would have directly altered ribosome binding (Fig. S7).

DeaD activates UvrY translation

To determine whether DeaD directly activates uvrY expression we purified a recombinant protein (DeaD_His) that is fully functional in vivo (Fig. S8), and tested its effects in a defined transcription-translation system (Ohashi et al., 2010). Plasmid templates for the reactions contained the deaD-independent fusion lacUV5-uvrY12'-`lacZ or the deaD-dependent fusion lacUV5-uvrY51'-`lacZ. DeaD_HIS caused up to 3-fold activation of uvrY51'-`LacZ, but did not affect uvrY12'-`lacZ fusion or the β-lactamase (bla) gene of these plasmids, which served as an internal control (Fig. 5A). Because DeaD_HIS did not affect uvrY51'-`lacZ mRNA synthesis in these reactions (Fig. 5A), it must have stimulated translation. We were unable to demonstrate activation by DeaD using purified mRNA substrates (data not shown). Thus, ongoing transcription may be important for this regulation, perhaps because of a requirement for proper mRNA folding during synthesis of the nascent transcript.

Fig. 5.

DeaD activates uvrY translation and mRNA accumulation. (A) Coupled transcription-translation reactions. The purified DeaD_HIS protein was added (0.125, 0.25, 0.5, 1.0, and 0 μM, respectively) to reactions with uvrY12'-`lacZ or uvrY51'-`lacZ expressing plasmids. Top 4 images: [³⁵S] methionine-labeled proteins. Lower 2 images: Northern blot of uvrY'-`lacZ mRNA synthesized in reactions lacking [³⁵S] methionine. (B) Effect of DeaD on uvrY mRNA (top) and protein (bottom) from mid-exponential growth at 30 °C and 37 °C were analyzed by Northern and Western blotting, respectively. Strain identities: ΔuvrY (uvrY), uvrY_FLAG (WT), and ΔdeaD uvrY_FLAG (deaD). (C) Effects of DeaD on native uvrY mRNA levels and decay. Wild type (WT) (•) and ΔdeaD (□) strains were grown to mid-exponential phase at 37 °C and rifampicin was added. RNA was isolated at various times after rifampicin addition (at Time = 0 sec), and mRNA was analyzed by q-RT-PCR, and plotted relative to mRNA at 0 sec. Steady state mRNA relative levels (Time = 0) are also shown. The values depict the averages from three independent experiments, and error bars represent the standard errors of the means. Error bars not shown were below resolution. (D) Mutations in the uvrY mRNA translation initiation region affect uvrY mRNA levels. Stains expressing chromosomal P_araB-uvrY5'UTR-uvrY_FLAG fusions with the wild type (TTG) or mutant (TGA) initiation codon or ΔRBS (GGAG to CCAC) were grown in the presence of arabinose, and uvrY_FLAG RNA was analyzed by Northern blotting. The uvrY_FLAG mRNA was the only signal detected (mRNA from the native uvrY locus was below detection).

Effect of DeaD on uvrY mRNA levels in vivo

To test whether DeaD affects uvrY mRNA levels in vivo we monitored uvrY_FLAG mRNA by Northern blotting and native uvrY mRNA by q-RT-PCR (Fig. 5B and C). Both transcripts were decreased in the ΔdeaD mutant. Changes in translation often alter mRNA stability, presumably because translating ribosomes protect mRNA from ribonuclease cleavage (Deneke et al., 2013). However, ΔdeaD did not affect the stability of the mature uvrY mRNA (Figs 5C and S9). Because there was no evidence that DeaD affects uvrY transcription initiation (Fig. 4 and data not shown), we suspected that DeaD-dependent translation protects the nascent uvrY mRNA from decay or premature termination. To further test the role of translation in this process, we changed the TTG start codon to a TGA stop codon or mutated the SD sequence (GGAG to CCAC) of a chromosomal P_araB-uvrY5'UTR-uvrY_FLAG fusion. Both changes resulted in undetectable levels of UvrY_FLAG protein (Fig. S6) and uvrY mRNA (Fig. 5D), indicating that translation is necessary for uvrY mRNA accumulation. Altogether, our findings suggest a model in which the primary regulatory effect of DeaD is on uvrY translation, with secondary effects on RNA accumulation.

Phylogenetic conservation of uvrY mRNA

The above features of uvrY mRNA that are involved in regulation by DeaD are highly conserved among most, but not all other Enterobacteriaceae (Fig. S10). This observation suggested that a common mechanism for DeaD-dependent activation of uvrY translation may operate in many members of this important bacterial family. As a proof of principle, we tested the effect of a deaD helicase gene deletion on the expression of the chromosomally-encoded UvrY_FLAG homolog, SirA_FLAG, of the important mammalian pathogen Salmonella enterica serovar Typhimurium strain 14028S (Fig. S10H). This experiment confirmed that expression of this response regulator also depends upon the deaD helicase gene in this bacterium.

Helicase-dependent activation over a broad temperature range

DeaD helicase is known as a cold shock protein, which acts at low temperatures (Jones et al., 1996). In addition, CsrB/C RNAs relieve CsrA-mediated repression of genes for biofilm formation and other stress resistance mechanisms (Romeo et al., 2013). Thus, we wondered if DeaD might serve within a pathway that permits E. coli to recognize that it has exited the warm host environment and respond by inducing the expression of stress resistance genes. To determine if DeaD activates uvrY expression in response to a decrease in the growth temperature, we first monitored the steady state levels of DeaDFLAG, and UvrYFLAG in deaD deletion and wild type strain backgrounds over a range of temperatures. DeaD_FLAG levels were highest at low temperatures in the exponential phase of growth, and were very low during the stationary phase (Fig. 6A). In contrast, UvrY_FLAG levels were largely unaffected by the growth temperature (Fig. 6A).

Fig. 6.

DeaD influences gene expression over a broad range of temperatures. (A) Western blotting to examine the effect of temperature on levels of DeaD_FLAG and the UvrY_FLAG protein in ΔdeaD mutant and wild type strains. Cultures were harvested at early, middle, and late-exponential phase, and the stationary phase of growth. (B) UvrY and CsrB levels during (1 hr) cold shock in wild type and ΔdeaD backgrounds. Strains carrying the indicated FLAG fusions were grown to mid-exponential phase (0.4 OD₆₀₀) in LB at 37 °C. Cultures (10 ml) were placed in a 15 °C water bath and samples were isolated at the indicated time points after temperature shift. Samples were subjected to Western blotting (DeaD_FLAG, CspA_FLAG, UvrY_FLAG) or northern blotting (CsrB). CspA served as a positive control for the cold shock procedure (Goldstein et al., 1990).

Maintenance of UvrY_FLAG levels was highly dependent on DeaD in the exponential phase at all temperatures, and only modestly dependent on DeaD in the stationary phase of growth. The basis of stationary phase-specific, DeaD-independent UvrY_FLAG accumulation was not further studied. We also observed that UvrY_FLAG and CsrB levels did not respond to temperature shift during a 1 hr cold shock experiment (Fig. 6B). In summary, DeaD does not regulate uvrY expression in response to the ambient temperature, but is present in sufficient cellular concentrations in the exponential phase of growth to activate expression over a wide range of temperatures.

Finally, the effects of SrmB on CsrB RNA levels were also found to be unresponsive to the ambient temperature (Fig. S11). While the pathway necessary for regulation by SrmB is unknown, these results revealed that SrmB helicase also acts over a wide range of temperatures.

DeaD affects the expression of diverse genes

Because of the unexpected temperature response of uvrY to DeaD, we wondered how DeaD might affect the temperature response patterns of other genes. To identify the RNA targets of DeaD helicase we used HITS-CLIP analysis with strains expressing DeaD_FLAG or a similar SrmB_FLAG protein (Figs 7 and S12). SrmB_FLAG served as a control to distinguish RNAs that interact specifically with DeaD_FLAG vs. RNAs that may be nonspecifically detected or that interact with both helicases. Altogether, 39 transcripts were identified at a threshold of >5-fold excess in the DeaD_FLAG vs. SrmB_FLAG experiment, including uvrY mRNA (Figs 7 and S13).

Fig. 7.

Effects of DeaD helicase on the expression of genes identified by HITS-CLIP analysis. Gene fusions introduced a FLAG^® epitope at the carboxy-terminus of each protein product and were integrated at the native chromosomal loci. Wild-type or ΔdeaD (Δd) strains were grown to mid-exponential phase at the indicated temperatures for Western blotting using anti-FLAG® or anti-RpoS antibodies. RNA sequence abundance (RNAseq), collected from crosslinking with DeaD_FLAG or SrmB_FLAG, is shown from the start of transcription to the end of each open reading frame. Chevrons show gene polarity.

We next examined the effects of DeaD on the steady state levels of 7 proteins from mRNAs detected specifically by DeaD_FLAG and 3 proteins from mRNAs detected equally by DeaD_FLAG and SrmB_FLAG (Fig. 7). The strains were cultured at 30, 37, and 42 °C. All of the proteins expressed from DeaD-specific mRNAs responded to the deaD deletion (Fig. 7). UvrY, SdiA, MntR were dependent on DeaD at all temperatures. TatE and IaaA were primarily or exclusively dependent on DeaD at 42 °C. IbpA_FLAG showed increased levels in the deaD mutant strain at 30 and 37 °C, but not at 42 °C. As previously shown, RpoS levels were dependent on DeaD at 30 °C (Resch et al., 2010). However, they were increased at 42 °C in the deaD mutant strain. DeaD did not affect the proteins whose mRNAs were equally detected by DeaD_FLAG and SrmB_FLAG.

Discussion

This study advances our understanding of the role of the DEAD-box RNA helicase DeaD in bacterial gene expression in a number of ways. First, we have shown that DeaD mediates regulation of the Csr system and have determined the underlying circuitry (Figs 1 and 2 and 3). Specifically, DeaD stimulated the translation of UvrY, the response regulator that activates transcription of the sRNAs CsrB and CsrC, which sequester and antagonize CsrA. The CsrA protein interacts with hundreds of mRNAs, allowing it to coordinate the expression of genes critical for major shifts in physiology, e.g. exponential vs. stationary phase, planktonic vs. biofilm growth, chronic vs. acute infective states of pathogens, as well as quorum sensing, motility and stress resistance (Romeo et al., 2013, Goodman et al., 2004). DeaD strongly regulated the expression of genes of the CsrA regulon through its effects on CsrB/C RNAs (Fig. S4). However, deletion of csrB and csrC genes did not completely eliminate the effects of DeaD on these genes (Fig. S4), suggesting that the circuitry of this system may be complex. Consistent with this idea, pgaA expression responds to several regulators, which might serve as intermediate targets of DeaD regulation (Goller et al., 2006, Thomason et al., 2012, Jorgensen et al., 2013), although none that were identified by our RNA cross-linking studies.

Second, HITS-CLIP analysis identified 39 different candidate mRNA targets of DeaD, including mRNAs of 10 regulatory proteins (Figs 7 and S13). Of these, only RpoS has been previously reported to be regulated by DeaD, and only at low temperatures (Resch et al., 2010). This minimal list of candidate mRNA targets is likely to be incomplete because: i) a stringent threshold (5-fold above the SrmB control) may have eliminated authentic targets, ii) a single growth condition is not likely to support transcription of all DeaD target RNAs, iii) some of the mRNAs identified by both DeaD and SrmB may interact specifically with each helicase. Studies of 7 presumptive targets of DeaD confirmed regulation by DeaD in all cases (Fig. 7). We infer that DeaD helicase influences gene expression by interacting with at least a few dozen mRNAs and indirectly affecting the expression of many more genes, which are controlled by CsrA and perhaps RpoS and other regulators.

Third, this study establishes a new paradigm for the function of DeaD with respect to the growth temperature. Based on the literature, we expected that DeaD would only affect gene expression at low temperature (Iost & Dreyfus, 2006, Resch et al., 2010, Jones et al., 1996, Butland et al., 2007). In fact, its regulatory effects were seen over a broad temperature range, and depended on the genetic target of interest (Fig. 7). Therefore, we propose that regulatory responses mediated by DeaD are governed primarily by mRNA specific properties, including mRNA structure as well as the sRNAs, proteins, or other factors that interact specifically with a given mRNA. At high temperatures, DeaD is present in concentrations sufficient for it to facilitate posttranscriptional regulation of many mRNA targets. At low temperatures DeaD accumulates (Fig. 6). However, much of the accumulated DeaD protein may be dedicated to the biogenesis of 50S ribosomal subunits, because DeaD is predominantly bound to ribosomes under this condition (Iost & Dreyfus, 2006, Jones et al., 1996). We recognize that changes in the levels or activity of DeaD may drive regulation in certain cases, and that deaD expression also responds to variations in osmolarity, pH, and polyamines (Owttrim, 2013, Yohannes et al., 2005, Stancik et al., 2002). The consequences of these effects for posttranscriptional regulation remain to be determined. Finally, SrmB regulated CsrB at temperatures ranging from 24 to 42 °C (Fig. S11) implying that this helicase also functions in vivo over a wide range of temperatures.

We provided evidence that DeaD regulation of uvrY expression requires long distance pairing between the mRNA leader and coding sequence in E. coli (Fig. 4) and perhaps most Enterobacteriaceae (Fig. S10). Our working model for this regulation (Figs 4 and 8) is that ribosome binding to uvrY mRNA is governed by the structural consequences of the long distance base pairing: strand separation by DeaD facilitates ribosome binding to the untranslated leader, unwinding of the inhibitory stem-loop structure formed by the SD and anti-SD sequences, and translation. While the biological function of this regulation is unclear at present, it seemingly may be used, along with the stringent response (Edwards et al., 2011), to govern CsrB/C expression according to the overall status of translation. Consistent with this idea, enhancing uvrY mRNA translation by mutating the TTG initiation codon to ATG decreased its requirement for DeaD (Fig. S6). Viewed in this way, the function of DeaD would be to increase the relative level of uvrY expression, and therefore CsrB/C expression, as cellular conditions for translation deteriorate. The purified DeaD protein activated uvrY translation in coupled transcription-translation reactions, although it did not activate translation of the purified mRNA (Fig. 5A and data not shown), suggesting that DeaD participates in a dynamic expression mechanism that was not recapitulated by the in vitro folded mRNA. The possibility that the purified uvrY mRNA was not folded properly to permit productive interaction with DeaD diminished our confidence to interpret molecular features of our model using conventional RNA footprinting and ribosome toeprinting experiments (Qu et al., 2011).

Fig. 8.

A model for the role of DeaD in uvrY expression. We propose that upon its transcription, the nascent uvrY mRNA adopts a secondary structure between the untranslated leader and the proximal coding segment, which is relatively inaccessible to translation and is a target for DeaD RNA helicase. Remodeling of this paired region by DeaD facilitates ribosome binding to the mRNA and translation. In the absence of translation, the full-length uvrY mRNA is undetectable and most likely cannot be synthesized. Because the ribosome possess helicase activity, once translation has begun on an mRNA molecule DeaD helicase may no longer be required. Residual uvrY expression in the deaD mutant may arise due to occasional ribosome binding to mRNA before the inhibitory secondary structure has formed, low efficiency binding to mRNA that has already folded, or both. Upon synthesis and phosphorylation (Figs 1A and 3C) the UvrY protein can activate transcription of the CsrB/C sRNAs.

Our in vitro transcription-translation experiments demonstrated that DeaD helicase activated uvrY translation without affecting mRNA synthesis, suggesting that the primary regulatory effect of DeaD is to facilitate translation. Nevertheless, DeaD was needed for normal accumulation of uvrY mRNA in the cell (Fig. 5). Interestingly, DeaD did not affect the stability of the mature mRNA (Fig. 5). It is possible that DeaD enhances the stability of the nascent transcript against nuclease attack due to its effect on translation, permitting the full-length mRNA to be formed. Alternatively, reduced uvrY mRNA levels in the deaD mutant might result from premature transcription termination. The in vitro reactions lacked the ribonucleases and transcription termination factors that would be required for either mechanism. Thus, additional studies are required to determine how DeaD affects uvrY mRNA accumulation in vivo.

These studies greatly expand our understanding of the roles that DeaD helicase plays in bacterial gene expression, and hint that it participates in diverse post-transcriptional regulatory mechanisms. DeaD activates uvrY translation over a broad temperature range and may not require the participation of specialized trans-acting factors in this role. In contrast, DeaD activates rpoS expression at low temperatures (Fig. 7), and may require Hfq and the sRNA DsrA for this function (Resch et al., 2010). While translational regulation involving RNA structures that influence the formation of an inhibitory SD and anti-SD stem loop is not uncommon (Waters & Storz, 2009, Kortmann & Narberhaus, 2012), inhibitory structures formed between the transcript leader and coding segment, as seen in uvrY regulation, appear to be unusual. Similar long-range base pairing occurs in some bacterial RNA thermometers, where temperature-dependent changes in RNA structure affect translation efficiency and/or RNA stability (Kortmann & Narberhaus, 2012); yet unlike RNA thermometers, the uvrY translation mechanism is not temperature-dependent (Fig. 6). Interestingly, DeaD was essential for temperature regulation of the ibpA RNA thermometer in vivo (Fig. 7). The present model for ibpA expression posits that mRNA secondary structural changes alone are sufficient to permit an increase in temperature to activate its translation (Kortmann & Narberhaus, 2012). While this may be true in vitro, the finding that ibpA was fully expressed at low temperature in the ΔdeaD mutant (Fig. 7) suggests that a more complex regulatory mechanism exists in the cell. The presence of numerous candidate DeaD target RNAs, without an obvious overriding pattern of regulation by DeaD with respect to the growth temperature, also hints that DeaD mediates diverse mRNA transactions. The molecular details of these roles and their impact on regulatory circuitry, bacterial physiology, and systems biology represent important challenges for study.

Experimental Procedures

Bacterial strains and culture conditions

Bacterial strains used in this study are described in Table S1. Escherichia coli and Salmonella strains were maintained on LB medium containing the following antibiotics as necessary: ampicillin (100 μg ml⁻¹), tetracycline (15 μg ml⁻¹), kanamycin (50 μg ml⁻¹), and chloramphenicol (25 μg ml⁻¹). Bacteria were routinely grown using the following protocol unless otherwise indicated: LB medium (2 ml) was inoculated with bacterial strains from frozen glycerol stocks and cultures were grown with shaking at 37 °C overnight. Stationary phase cultures were used to inoculate LB medium and growth was monitored (OD₆₀₀). Cultures were grown with shaking (250 rpm) at 37 °C, and samples were taken for RNA/protein extraction or gene expression analysis at early exponential phase (OD₆₀₀ 0.1), mid-exponential phase (OD₆₀₀ 0.4), late exponential phase (OD₆₀₀ 0.8), transition to stationary phase (OD₆₀₀ 1.2), or stationary phase (OD₆₀₀ 3–5, 24 hrs) unless otherwise indicated.

Plasmid construction

Primer names, sequences, and relevant restriction sites are identified in Tables S2 and S3. The DeaD and SrmB expression plasmids pDeaD and pSrmB were created by PCR amplification of each gene locus (500 bp upstream of the start codon to the stop codon) and cloning the resulting fragments into pBR322. The DeaD_HIS (NTD) expression and purification plasmid pDeaD_HIS was created by PCR amplification of the deaD ORF, and cloning the resulting fragment into pCOLADuet™-1. The plasmid pDeaD_HIS-2, used for complementation of a deaD deletion strain with DeaD_HIS, was created by PCR amplification of the native deaD promoter (500 bp upstream of the start codon to transcription start) and the deaD_HIS ORF from pDeaD_HIS, and cloning the resulting fragments into pBR322. The plasmid p1VR146a, which encodes the P_uvrY-uvrY5'UTR-uvrY_FLAG construct (Fig. 4A “i”), was created by PCR amplification of the entire uvrY locus including the promoter region (−200 bp relative to transcription start), uvrY transcript leader and coding sequence, the carboxy-terminal FLAG epitope tag, and cloning the resulting fragment into the CRIM plasmid pAH125 (Haldimann & Wanner, 2001). The p1VR146b plasmid, which encodes the P_araB-araB5'UTR-uvrY_FLAG construct (Fig. 4A “ii”), was created by PCR amplification of the araC-P_araB promoter and transcript leader cassette (−1250 bp relative to P_araB transcription start), and the uvrY coding sequence and carboxy-terminal FLAG epitope tag, and cloning the resulting fragments into pAH125. The p1VR153 plasmid, which encodes the P_araB-uvrY5'UTR-uvrY_FLAG construct (Fig. 4A “iii”), was created by PCR amplification of the araC-P_araB promoter, and the uvrY transcript leader, uvrY coding sequence, and CTD FLAG epitope tag, and cloning the resulting fragments into pAH125. The uvrY leader fusion plasmid p1VR166 and its derivatives were created by PCR amplification of the uvrY transcript leader (−47 bp relative to start codon) and various portions of coding sequence, and cloning the resulting fragments into pUV5 (Edwards et al., 2011). Integration of the CRIM plasmid pAH125 derivatives into the chromosome at the λatt site was accomplished using previously published procedures (Haldimann & Wanner, 2001). Point mutations in the p1VR153 and p1VR166 plasmids were introduced by megaprimer PCR as described previously (Vakulskas et al., 2009). The amino-terminal FLAG® epitope-tagged deaD_FLAG and srmB_FLAG genes were introduced at the native chromosomal loci using the pKOVDeaD_FLAG and pKOVSrmB_FLAG plasmids, respectively; using published chromosomal replacement protocols (Link et al., 1997). These plasmids were constructed by PCR amplification of 1000 bp of DNA immediately flanking the deaD or srmB start codons using primers that incorporate the FLAG® epitope, and cloning the resulting fragments into pKOV. The p2VR83 plasmid, which encodes uvrY_FLAG expressed from its native regulatory elements (Fig. S9), was created by PCR amplification of uvrY_FLAG (from p1VR146a) and the rpoC terminator, and cloning the resulting fragments into pBR322.

β-Galactosidase Assays

β-galactosidase assays were performed as described (Edwards et al., 2011). The reported values are the averages from three independent experiments, and error bars represent the standard errors of the means. β-galactosidase activity was measured from bacterial cultures grown to mid-exponential phase unless otherwise indicated.

Construction of E. coli chromosomal deletions and FLAG® fusion proteins

E. coli and Salmonella gene deletions and carboxy-terminal FLAG®-tagged constructs were introduced by the phage λ Red recombinase method, as described (Datsenko & Wanner, 2000, Uzzau et al., 2001).

Transposon mutagenesis and screening

Strain CR1-93 was transduced with the previously described csrB-lacZ transcriptional reporter (Gudapaty et al., 2001), and a transposon mutagenesis screen was performed in the resulting strain using the EZ-Tn5^TM <KAN-2>Tnp Transposome Kit (Epicentre) as per manufacturer's specifications. Briefly, strain CR1-93 containing the integrated csrB-lacZ fusion was grown to early exponential phase (0.3 OD₆₀₀), electroporated with purified EZ-tn5-kan transposomes, and plated to a density of 3000 CFU per 150 mm plate containing LB, kanamycin (50 μg ml⁻¹), and 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal, 40 μg ml⁻¹). We screened 20,000 CFU (95% coverage) and isolated 52 mutants that were white or lighter blue than the parent strain, including 3 deaD and 2 srmB mutants.

Northern Blotting

E. coli cultures were grown as indicated and total cellular RNA was isolated using the RNeasy mini kit (Qiagen) according to the manufacturer's instructions. Total cellular RNA or RNA from in vitro transcription-translation reactions was separated on denaturing 5% acrylamide gels containing 7 M urea (CsrB) or 2% formaldehyde agarose gels (UvrY mRNA) and transferred to a positively charged nylon membrane (Roche Diagnostics) by electro-blotting using the Mini Trans-Blot Cell (Bio-Rad). RNA was fixed to the membrane by UV cross-linking. Blots were hybridized overnight at 68 °C with a DIG-labeled antisense CsrB RNA probe and developed using the DIG Northern Starter kit (Roche Diagnostics) according to the manufacturer's instructions. The antisense CsrB RNA and uvrY coding region probes were transcribed in vitro using the DIG Northern Starter kit (Roche Diagnostics) from a PCR product, according to the manufacturer's instructions. Blots were imaged using the ChemiDoc XRS+ system (Bio-Rad) and RNAs were quantified using Quantity One image analysis software (Bio-Rad). The rRNAs (16S and 23S) served as loading controls, and were detected by methylene blue staining.

Western Blotting

Bacterial cells were harvested by centrifugation, immediately mixed with Laemmli sample buffer, and lysed by sonication and boiling. Samples (10 μg protein) were subjected to SDS-PAGE. Proteins were transferred to 0.2-μm polyvinylidene difluoride membranes by electroblotting using the GENIE® electrophoretic transfer apparatus (Idea Scientific Company) according to the manufacturer's instructions. FLAG epitope-tagged proteins were detected using the anti-FLAG® M2 monoclonal antibody (Sigma), CsrA was detected using previously described polyclonal antibodies (Gudapaty et al., 2001), RpoS was detected using anti-RpoS monoclonal antibodies (Neoclone), and RpoB was detected using anti-RpoB monoclonal antibodies (Neoclone). Western blots were developed using horseradish peroxidase-linked secondary antibodies and the SuperSignal® West Femto Chemiluminescent Substrate (Pierce).

Analysis of UvrY-P levels in vivo

Relative UvrY-P levels were analyzed as described previously, with minor modifications (Boulanger et al., 2013). Briefly, cells were grown to mid-exponential growth phase (OD₆₀₀-0.4) at which time 0.375 ml cells were collected by centrifugation (16,900 × g for 1 min) and immediately flash-frozen in liquid nitrogen. Modified Laemmli sample buffer (0.16 M Tris, pH 9, 4% SDS, 20% glycerol, and 10% 2-mercaptoethanol) was then added (60 μl) to each cell pellet, samples were vortexed vigorously for 30 sec, and immediately placed on ice. Samples were fractionated on SDS-PAGE gels (1.0 mm Protean 3 [Bio-Rad] minigels) of the following composition: 7.5% separating gel (7.5% [29:1] acrylamide:bis-acrylamide, 357 mM Bis-tris, pH 6.8, 100 μM Zn(NO₃)₂, and 50 μM Phos-tag^TM) and 4% stacking gel (4% [29:1] acrylamide:bis-acrylamide, 357 mM Bis-tris, pH 6.8). Electrophoresis was allowed to proceed at constant 150 volts at 4 °C for 70 min using modified MOPS running buffer (100 mM Tris, 100 mM MOPS, 0.5% SDS, and 5 mM NaHSO₃). Gels were subsequently washed in Western transfer buffer (25 mM Tris, 192 mM Glycine, 20% methanol, and 0.1% SDS) containing 1 mM EDTA for 10 min at room temperature. Gels were then washed with Western transfer buffer without EDTA for 20 min. Protein were transferred at 4 °C overnight to PVDF membranes using the Mini Trans-Blot (Bio-Rad) at 30 volts constant. Western blotting was performed as described above.

RNA structure prediction and amino acid sequence alignment

RNA structure predictions were created with the mfold web server (http://mfold.rna.albany.edu/?q=mfold) using default folding parameters. Amino acid sequence alignments were created with ClustalW2 (http://www.ebi.ac.uk/Tools/msa/clustalw2/) using FASTA format, and the default protein submission parameters.

Purification of DeaD_HIS

E. coli Tuner (DE3) cells carrying pDeaD_HIS were grown at 30 °C in LB containing ampicillin to an OD₆₀₀ of 0.5, at which time isopropyl-β-D-thiogalactopyranoside (1 mM) was added and the culture was incubated for an additional 2 hr at 30 °C. Bacteria were harvested by centrifugation and suspended in 20 mM Tris (pH 8.0), 500 mM NaCl, and 20 mM imidazole supplemented with a protease inhibitor cocktail (Roche). Cells were disrupted using a French pressure cell, and the DeaD_HIS-containing lysates were cleared by centrifugation (20,000 g, 15 min, 4 °C), subjected to Ni²⁺ affinity chromatography as previously described (Resch et al., 2010). Peak protein fractions were pooled and dialysed against DeaD storage buffer [20 mM Tris (pH 8.0), 250 mM KCl, 1 mM dithiothreitol and 50% glycerol]. Aliquots of purified DeaD_HIS were flash frozen in liquid nitrogen and stored at −80 °C. Protein concentrations were determined using the bicinchoninic acid assay (Pierce Biotechnology) with bovine serum albumin as the protein standard.

Coupled transcription-translation reactions

The PURExpress^® In Vitro Protein Synthesis Kit (New England Biolabs) was used for transcription-translation reactions according to the manufacturers instructions. DNA template consisted of the pUV5-uvrY12'-`lacZ or pUV5-uvrY51'-`lacZ leader fusion plasmid, which was purified using the Plasmid Midiprep Kit (Qiagen) followed by phenol/chloroform extraction and ethanol precipitation. The PURExpress^® reactions were assembled on ice in the following order: 7.5 μl Buffer B was added to a pre-chilled RNase-free 1.5 ml microcentrifuge tube containing 10 μl Buffer A, 2 μl (5 μg total) DNA template, 37 pmol [³⁵S]-methinonine [1000 mCi mmol⁻¹] (Perkin Elmer), and DeaD_HIS (as necessary). 1 U of σ⁷⁰-saturated E. coli RNA polymerase holoenzyme (Epicentre) was added and the reactions were gently mixed on ice. The reactions were started by incubation at 37 °C, and allowed to proceed for 2 hrs. Reactions were terminated by the addition of an equal volume of 2x Laemmli sample buffer and boiling at 100 °C for 5 min. Reactions were then subjected to SDS-PAGE. The resulting gels were incubated in fixative [30% methanol and 10% acetic acid] with gentle shaking, washed in 10% glycerol, heat-dried onto chromatography paper, and analyzed using a phosphorimager.

RNA decay analysis

Wild type and ΔdeaD mutant strains were grown in LB at 37 °C to mid-exponential phase (OD₆₀₀-0.6) and rifampicin was added to a final concentration of 200 μg ml⁻¹. At various time points after rifampicin addition, 0.5 ml of the bacterial culture was added to 2 volumes of RNAprotect Bacteria Reagent (Qiagen) and incubated for 5 min at room temperature. Total cellular RNA was subsequently purified with the RNeasy Mini Kit (Qiagen) and contaminating DNA was digested with Turbo DNase (Ambion). uvrY mRNA was detected by q-RT-PCR (Fig. 5C) or Northern blotting (Figs 5D and S9).

Quantitative RT-PCR

Quantitative RT-PCR (q-RT-PCR) was carried out in an iCycler™ (Bio-Rad Laboratories) thermocycler using the iScript™ One-Step RT-PCR Kit with SYBR^® Green (Bio-Rad), according to the manufacturer's instructions. Briefly, 5 μl of the extracted sample RNA was added to a 25 μl reaction containing 1X SYBR^® Green RT-PCR Reaction Mix, 300 nM of each primer, and 100 ng of total RNA template. The thermal cycling conditions consisted of a 10 min reverse transcription step at 50°C, 5 min of iScript Reverse transcriptase inactivation at 95°C, followed by 45 cycles of PCR at 95°C denaturation for 10 sec, 60°C of annealing, extension, and detection for 30 sec. Following amplification, melting curves were used to verify the specificity of the PCR product according to its T_m. Melting curve analysis consisted of a denaturation step at 95°C for 1 min, lowered to 55°C for 1 min, and followed by 80 cycles of incubation in which the temperature is increased to 95°C at a rate of 0.5°C/10 sec/cycle. The T_m of each specific PCR product was analyzed using iCycler iQ optical system software version 3.1 (Bio-Rad).

HITS-CLIP

DeaD_FLAG was crosslinked to its targets in vivo with formaldehyde treatment, and the resulting complexes were immunoprecipitated. Crosslinks were reversed with brief heat treatment, and cDNA libraries were prepared from immunoprecipitated RNA and sequenced using the HiSeq 2000 platform (Illumina). Additional details of the procedures and analyses are presented in the SI Experimental Procedures.