Protection against deleterious nitrogen compounds: role of σS-dependent small RNAs encoded adjacent to sdiA.

Here, we report the characterization of a set of small, regulatory RNAs (sRNAs) expressed from an Escherichia coli locus we have denoted sdsN located adjacent to the LuxR-homolog gene sdiA Two longer sRNAs, SdsN137 and SdsN178 are transcribed from two σ(S)-dependent promoters but share the same terminator. Low temperature, rich nitrogen sources and the Crl and NarP transcription factors differentially affect the levels of the SdsN transcripts. Whole genome expression analysis after pulse overexpression of SdsN137 and assays of lacZ fusions revealed that the SdsN137 directly represses the synthesis of the nitroreductase NfsA, which catalyzes the reduction of the nitrogroup (NO2) in nitroaromatic compounds and the flavohemoglobin HmpA, which has aerobic nitric oxide (NO) dioxygenase activity. Consistent with this regulation, SdsN137 confers resistance to nitrofurans. In addition, SdsN137 negatively regulates synthesis of NarP. Interestingly, SdsN178 is defective at regulating the above targets due to unusual binding to the Hfq protein, but cleavage leads to a shorter form, SdsN124, able to repress nfsA and hmpA. Published by Oxford University Press on behalf of Nucleic Acids Research 2016. This work is written by (a) US Government employee(s) and is in the public domain in the US.

INTRODUCTION

In order to survive in many different, constantly-changing environments, bacteria have intricate mechanisms to sense environmental cues and increase or decrease the levels of the appropriate proteins and enzymes at the transcriptional and/or post-transcriptional levels. Key post-transcriptional regulators are small RNAs (sRNAs), typically 50 to 300 nucleotides in length, that base pair with mRNAs encoded in trans at a distinct genomic location (1,2). By base pairing at or near the ribosome-binding site, the sRNAs can block translation. These sRNAs can also activate translation when base pairing results in a change in mRNA secondary structure that liberates a ribosome binding site. In addition, base pairing can lead to the recruitment of RNase E either in conjunction with or independent of effects on ribosome binding. Given the limited complementarity to their target mRNAs, the sRNAs in a number of bacteria including Escherichia coli require the RNA chaperone Hfq to stabilize the sRNAs and facilitate sRNA–mRNA duplex formation (3,4).

Many base pairing sRNAs are induced in response to very specific environmental signals and then act to protect the cells and/or make maximal use of limited resources under these conditions. For example, RyhB, whose levels are induced by conditions of iron starvation, represses the synthesis of non-essential iron-storage and iron-utilization proteins (5). Similarly, FnrS RNA, whose levels are induced by oxygen limitation, represses the synthesis of proteins that are not needed under anaerobic conditions (6,7). Three σE-dependent sRNAs, RybB, MicA and MicL, that are all induced by cell envelope stress, repress the synthesis of all abundant outer membrane proteins thus allowing chaperones required for the insertion of new proteins to be redirected to misfolded cell envelope proteins (8).

Expression of a number of sRNAs is highest in stationary phase (9), conditions under which bacteria undergo substantial changes in morphology and physiology to conserve energy and become resistant to starvation and various environmental stresses. The key regulator in stationary phase is σS, encoded by the rpoS gene (10,11). This alternative sigma factor directs the expression of hundreds of genes when cells enter stationary phase or encounter other stresses, alone at some promoters and in conjunction with additional transcription factors at other promoters. The GadY, SraL and SdsR sRNAs have been shown to be σS-dependent sRNAs (12–14). GadY, positively regulates the expression of GadX and GadW, two transcription factors controlling the acid response, by directing cleavage of the gadXW mRNA to give more stable products (13,15). SraL was found to down-regulate the expression of the tig gene, which encodes the chaperone Trigger Factor involved in protein folding (14). SdsR acts as a repressor by base pairing with the coding region of mutS encoding a component of the methyl-directed mismatch repair complex and the 5′UTR of tolC encoding an outer membrane porin in E. coli (16,17). In Salmonella, SdsR also represses synthesis of the major porin OmpD by basepairing with the coding region of the ompD mRNA (12).

Here, we report on another σ-dependent sRNA in E. coli, SdsN, which is induced in stationary phase, particularly when cells are grown at low temperature or with preferred nitrogen sources, and regulates the levels of the nitrate- and nitrite-responsive NarP transcription factor as well as enzymes that metabolize oxidized nitrogen compounds.

MATERIALS AND METHODS

Bacterial strains and plasmids

The strains and plasmids used in this study are given in Supplementary Tables S1 and S2, respectively. E. coli K-12 MG1655 was employed as the wild-type (WT) strain. Phage λ Red-mediated recombineering was used to construct deletion alleles marked by KmR flanked by FRT (FLP recognition target) sites, which were amplified using pKD13 as template for PCR (18). After P1 transduction into MG1655, the KmR cassette was removed by introduction of the FLP expression plasmid pCP20. Other alleles were transduced from previously published strains (19,20). The nfsA-lacZ, hmpA-lacZ and narP-lacZ translational fusions were created as described (21); the first nine codons of nfsA and hmpA, or the first 30 codons of narP as well as the entire 5′-UTR of each gene were PCR amplified and fused to lacZ driven by an arabinose-inducible pBAD promoter. SdsN derivatives were overexpressed from the pBRplac plasmid (22). The sdsN, sdsN and sdsN fragments were PCR amplified from MG1655 chromosomal DNA, digested with EcoRI and AatII, and cloned into the corresponding sites of pBRplac. The mutant derivatives of the lacZ fusions and SdsN-overexpression plasmids were generated by overlapping PCR and introduced into the chromosome or pBRplac as described above. All plasmid and chromosomal constructs were verified by DNA sequencing. Primers used for PCR and sequencing are listed in Supplementary Table S3.

Growth conditions

Unless indicated otherwise, bacterial strains were grown overnight with shaking at 37°C in Luria Broth (LB) or M63 with 0.2% glucose media, both supplemented with standard concentrations of the appropriate antibiotics, and then diluted (OD600 ≈ 0.02–0.03) into the same medium. For the nitrofurazone sensitivity assays, back-diluted cultures were grown to OD600 ≈ 0.3 in LB or for 14 h in M63 glucose. Half of each culture was challenged with 1 or 2 mM nitrofurazone (Sigma) or azomycin (Sigma) for 1 h after which 100 μl of 10−5-dilutions of the LB cultures and 10−6-dilutions of the M63 glucose cultures were plated on LB agar.

RNA extraction

Total RNA was isolated by extraction with hot acid phenol (23) or TRIzol Reagent (Ambion). For the hot acid phenol extraction, RNA was isolated from 750 μl of LB-grown cells or 10 ml of cells grown in M63 glucose, collected and resuspended in 700 μl of M63. For both types of samples, the cells were mixed with 500 μl of acid-phenol-chloroform (Ambion) and 102 μl of cell lysis solution (320 mM sodium acetate, 8% SDS and 16 mM EDTA) and incubated 15 min at 65°C. Supernatants were transferred to a new tube containing pre-heated 500 μl of acid-phenol-chloroform and incubated at 65°C for another 15 min. RNA was precipitated with 700 μl of 100% ethanol. For the TRIzol extraction, total cellular RNA was isolated from ∼5–10 OD600 of cells according to the manufacturer's instructions. RNA was precipitated by combining the ∼0.6 ml of the top aqueous phase with 0.5 ml isopropyl alcohol. For both extraction methods, precipitated RNA pellets were washed with 70% ethanol and resuspended in nuclease-free water. Total RNA concentrations were determined using a NanoDrop (Thermo Scientific).

Northern analysis

Total RNA (10 μg) was separated on a 8% polyacrylamide-7M urea gel in 1X TBE (90 mM Tris-borate 2 mM EDTA). The RNA was transferred to a Zeta-Probe GT blotting membrane (Bio-Rad) at 20V for ∼16 h at 4°C in 0.5X TBE. After transfer, membranes were allowed to dry, UV cross-linked on both sides, and incubated overnight at 45°C in UltraHyb (Ambion) hybridization buffer and oligonucleotides 5′-end-labeled with 32P-ATP with T4 polynucleotide kinase (New England Biolabs). Subsequently, membranes were washed once with 2X SSC (150 mM NaCl 15 mM sodium citrate) 0.1%SDS, incubated 10 min at 45°C in 2X SSC 0.1%SDS, and washed 5X with 0.2X SSC 0.1% SDS. After washing, air-dried membranes were exposed to HyBlot CL film (Denville Scientific) at −80°C.

Primer extension analysis

Primer extension assays were carried out as previously described (24). Briefly, RNA samples (5 μg of total RNA) were incubated with 2 pmol of 5′-32P -end-labeled primer at 80°C and then slow‐cooled to 42°C. After the addition of dNTPs (1 mM each) and AMV reverse transcriptase (10 U, Life Sciences Advanced Technologies Inc.), the reactions were incubated in a 10 μl reaction volume at 42°C for 1 h. Reactions were terminated by adding 10 μl of Gel Loading Buffer II (Ambion). The DNA sequencing ladder was generated using Thermo Sequenase™ Dye Primer Manual Cycle Sequencing Kit (Affymetrix) and an sdsN or sdiA PCR fragment. The cDNA products and sequencing ladder were fractionated on an 8% polyacrylamide urea sequencing gel containing 8 M urea in 1X TBE buffer at 70 W for 70 min. The gel was dried and imaged using a STORM 840 (Amersham Biosciences).

Terminator-5′-phosphate-dependent exonuclease (TDE) digestion

Total RNA (7 μg) extracted from MG1655 cells grown to OD600 ≈ 5 at 25°C in LB or 20 h at 37°C in M63 glucose were placed in two RNase-free 1.5 ml tubes; the sample in one was treated with TDE (Epicentre) while the second sample was incubated with buffer as described (8).

In vitro RNA synthesis

The narP, hmpA, SdsN137 and SdsN178 RNAs were synthesized using Megascript T7 kit (Ambion). SdsN137 and SdsN178 RNA were 5′-end-labeled with 32P by treating the RNA with alkaline phosphatase (New England Biolabs) and then T4 polynucleotide kinase. 32P-labeled RNAs were purified from 8% polyacrylamide-7M urea gels by excising and crushing the bands in RNA elution buffer (0.1 M sodium acetate, 0.1% SDS and 10 mM EDTA). The elution was extracted with an equal volume of phenol:chloroform:IAA (Invitrogen). The RNA was then ethanol precipitated, re-suspended in nuclease free dH2O and quantified using the NanoDrop.

In vitro RNA structure probing

32P-labeled SdsN137 or SdsN178 (∼2 nM) was incubated with purified Hfq (or equal volume of buffer) and 1 μg of yeast RNA (Ambion) in 1X RNA structure buffer (Ambion) in a total volume of 8 μl at 37°C for 15 min. Samples were mixed with RNase T1 (0.02 U, Ambion) or an equal volume of buffer and incubated at 37°C for 6 min. Inactivation/Precipitation Buffer (20 μl, Ambion) was added, and samples were placed at −80°C for ∼30 min. RNA pellets were collected by centrifugation, washed with 100 μl of 70% ethanol, air-dried and dissolved in 7 μl Gel Loading Buffer II. For the hydroxide (OH) ladder, 1 μl of 32P-labeled SdsN137 or SdsN178 in 9 μl Alkaline Hydrolysis Buffer (Ambion) was incubated 5 min at 90°C. For the RNase T1 ladder, 1 μl of 32P-labeled SdsN137 or SdsN178 in 9 μl Sequencing Buffer (Ambion) was denatured by incubating at 95°C for 1 min followed by cooling to 37°C. RNase T1 (0.1 U) was added, and the sample was incubated for 5 min at 37°C. For both ladders, the reactions were stopped by adding 12 μl of Gel Loading Buffer II. Samples (2 μl) were run on a 8% polyacrylamide-7M urea sequencing gel in 1X TBE. The gel was transferred onto Whatman filter paper, dried at 80° for 1 h, and imaged using the STORM 840.

Electrophoretic mobility shift assays (EMSA)

32P-labeled SdsN137 or SdsN178 RNA (4 nM) was incubated at 37°C for 15 min with purified Hfq and 1 μg of yeast RNA in 1X RNA Structure Buffer in a total volume of 10 μl. For some samples, 1 μl 1X RNA Structure Buffer or unlabeled hmpA or narP RNA diluted in 1X RNA Structure Buffer was added and samples were incubated at room temperature for an additional 1 h. After the indicated incubation, 2.5 μl of non-denaturing RNA loading buffer (100 mM NH4Cl, 50 mM NaCl, 50 mM KCl, 20 mM Tris-HCl, pH 8 and 50% glycerol) was added to each sample. Samples were then separated at 100 V on a native 6% polyacrylamide gel (37.5:1 Bis-Acrylamide, National Diagnostics) at 4°C in pre-chilled 0.5X TBE. Gels were imaged using the STORM 840.

Microarray analysis

MG1655 cells harboring pBR, pBR-SdsN137 or pBR-SdsN178 were grown to OD600 ≈ 0.4 in LB at 37°C, then cells were induced with 100 μM isopropyl β-D-1-thiogalactopyranoside (IPTG) for 5 min. Cells were harvested and total RNA was isolated by hot-phenol method. Chromosomal DNA was removed with DNase I treatment before cDNA synthesis and hybridization of cDNA to the Affymetrix E. coli Genome2.0 array was carried according to the instructions in Affymetrix manual.

Immunoblot analysis

Cells collected from 1 ml of culture were resuspended in 80 μl/ OD600 1X SDS PAGE loading buffer containing 5% β-mercaptoethanol, heated for 20 min at 95°C and then loaded onto a 4–15% Tris-Glycine gel (BioRad). Proteins were transferred onto a nitrocellulose membrane (Invitrogen) at 100 V for 1 h at 4°C in 1X Tris-glycine-SDS running buffer (KD Medical) with 20% methanol. Membranes were blocked 1 h at room temperature in 1X PBST (phosphate buffer saline with Tween20, Corning) with 5% milk and then probed with a 1:4000 dilution of α-RpoS antibody (provided by Susan Gottesman, NCI) for ∼16 h at 4°C. After three washes in 1X PBST, membranes were incubated with a 1:5000 dilution of HRP-anti-rabbit IgG (Sigma) for 1 h at room temperature. After three additional washes, membranes were developed using SuperSignal® West Pico Chemiluminescent Substrate (Thermo Scientific) and exposed to HyBlot CL film.

β-galactosidase assays

Three or four individual colonies from the indicated lacZ translational fusions transformed with pBR plasmids were grown overnight at 37°C in LB media containing 100 μg/ml ampicillin, diluted to an OD600 = 0.03 into fresh LB media with 100 μg/ml of ampicillin and 0.2% arabinose (to induce lacZ expression). IPTG (100 μM or 1 mM) was simultaneously added to some of the samples to induce expression from the pBR plasmids. Cells were grown to OD600 ≈ 0.5 or 1.0 and then lysed in 700 μl of Z buffer containing 0.002% SDS and 30 μl chloroform. β-galactosidase activity levels were assayed as described before (6).

RESULTS

Multiple sRNA species are transcribed from the sdiA-tcyN intergenic region

Microarray analysis of RNAs that co-immunoprecipitate with Hfq in E. coli suggested an sRNA bound by Hfq was encoded in the intergenic region between sdiA, the gene for a LuxR-family transcription regulator that serves as a ‘suppressor of the cell division inhibitor’, and tcyN (formerly yecC), the gene for the ATP binding subunit of an L-cystine/L-cysteine ABC transporter (Figure 1A), but no signal was detected by northern analysis under the conditions examined (25). Recent deep sequence analysis of the E. coli transcriptome again showed a small transcript is encoded in the sdiA-tcyN intergenic region (26,27). To obtain further information about the expression of this putative sRNA, we probed the total RNA isolated from E. coli MG1655 grown to different stages in LB rich medium and M63 glucose minimal medium at 37°C (Figure 1B). We now detected multiple transcripts, particularly in stationary phase. The two most prominent bands were ∼140 nt (SdsN137) and ∼180 nt (SdsN178), with the highest levels at OD600 ≈ 5 in LB medium and after 20 h of growth in M63 glucose medium. A number of additional bands were also observed for both samples. Probing with a downstream oligonucleotide showed that the longest transcripts prevalent early in growth are 3′ extensions of the major transcripts (Supplementary Figure S1A). In contrast, the shorter transcripts of ∼120 nt (SdsN124) lack the 5′ end. These smaller RNAs accumulate in late stationary phase, especially in M63 glucose medium.

Figure 1.

SdsN expression increases in stationary phase. (A) Schematic of region encompassing sdsN. The gene encoding SdsN (shaded in black) is located between tcyN (formerly yecC), encoding the ATP binding subunit of an L-cystine/L-cysteine ABC transporter complex and sdiA, encoding the LuxR homolog DNA binding transcription regulator. Transcription of sdsN is opposite both tcyN and sdiA. (B) Expression of SdsN, GadY, SraL and SdsR across growth in LB rich and M63 glucose minimal medium. Total RNA was isolated from wild-type (WT) MG1655 grown in LB or M63 media at 37°C for the indicated OD600 or time points. The RNA (10 μg) was separated on an 8% polyacrylamide-7M urea gel, transferred to a nitrocellulose membrane and probed with a 32P-labelled oligonucleotide specific for each sRNA or 5S as control. σS levels were assayed by immunoblot analysis of cells taken simultaneously.

The 5′ ends of the ∼140 and ∼180 nt transcripts were mapped by primer extension analysis (Supplementary Figure S1B) and examination of transcriptomic promoter mapping data (27). While multiple 5′ ends were detected by both approaches, the promoter mapping data indicate the prominent bands correspond to transcripts of 137 and 178 nt that share the same Rho-independent terminator (Figure 2A). Both forms were not fully digested by terminator exonuclease from RNA extracted from cells grown in LB and M63 glucose media (Supplementary Figure S1C), unlike the complete digestion observed for the processed derivative of SdsR. Thus, SdsN178 and SdsN137 likely have 5′-triphosphates consistent with transcription from two distinct promoters. The 5′ end of the ∼120 nt transcript was also mapped by primer extension analysis, which showed that one form is 124 nt (Supplementary Figure S1B). This band is no longer detected upon terminator exonuclease treatment indicating it is derived by processing (Supplementary Figure S1C).

Figure 2.

SdsN expression is dependent on σS. (A) Sequence of sdsN and sdsN-sdiA promoter region. The sequence that is common to SdsN178 and SdsN137 is shaded in dark grey, while the sequence specific to SdsN178 is shaded in light grey. The mapped transcription start sites of SdsN178 and SdsN137 are indicated with arrows that show the direction of transcription; the corresponding −10 sequences are boxed. Two transcription start sites for the sdiA gene are also indicated with arrows that show the direction of transcription. The −10 elements for both sdiA transcription start sites are denoted with brackets. The 5′ end of SdsN124 is also indicated with a vertical bracket. The sdiA start and tcyN stop codons are italicized, and the sequences corresponding to the stem of the SdsN terminator are underlined. (B) Levels of SdsN in WT, Δcrl, ΔrpoS, and ΔrpoS Δcrl cells. Total RNA was isolated from WT MG1655 and the isogenic Δcrl::kan (GSO760), ΔrpoS::Tn10 (GSO108) and Δcrl::kan ΔrpoS::Tn10 (GSO761) mutants grown 20 h in M63 glucose media (to OD600 ≈ 2.39, 2.40, 2.48 and 2.45, respectively) and analyzed as in Figure 1. (C) Effect of non-preferred and preferred nitrogen sources on SdsN levels. WT MG1655 and ΔnarP mutant (GSO763) cells were grown 20 h in M63 glucose with preferred ammonium (15 mM) or non-preferred arginine (0.2%) as the sole nitrogen source (to OD600 ≈ 3.61 and 2.36, respectively for wild-type cells and OD600 ≈ 3.62 and 2.76 for the ΔnarP mutant cells). RNA was processed for northern analysis as in Figure 1.

We also monitored the induction of the sdsN promoter by integrating the lacZ reporter gene downstream of both the sdsN and sdsN promoters. Expression of this P fusion was low in exponential phase and increased in stationary phase, particularly in cells grown at lower temperatures (Supplementary Figure S2A).

Stationary phase induction of SdsN is dependent on σS

The stationary phase induction of SdsN suggested regulation by σS. To investigate this possibility, we examined the SdsN levels in an rpoS deletion strain. Northern analysis showed that the levels of all SdsN transcripts are significantly lower in an ΔrpoS mutant strain in both LB (Supplementary Figure S2B) and M63 glucose medium (Figure 2B), indicating SdsN is indeed a σS-dependent sRNA. Consistent with the northern analysis, expression from the P fusion also was decreased in the rpoS mutant strain (Supplementary Figure S2A).

We also compared the levels of SdsN with the levels of the other known σS−dependent sRNAs (12–14). Interestingly, the expression patterns are not identical, and do not necessarily correlate with the highest levels of σS (Figure 1B). In LB, the levels of SdsN, GadY and SraL begin to increase at OD600 ≈ 3 and are highest at OD600 ≈ 5, while the levels of SdsR increase and peak at later time points. In M63 glucose medium, the peak of GadY expression is much earlier than the peaks for SdsN, SraL and SdsR. Furthermore, there is limited sequence similarity among the sdsN, gadY, sdsR andsraL promoters (Supplementary Figure S2C). These observations suggest additional factors may modulate the transcription of SdsN as well as the other σS-dependent sRNAs or σS is acting indirectly, possibly by modulating sRNA stability.

σS-dependent induction of SdsN is impacted by nitrogen availability

Given the partial match to the consensus σS promoter (Supplementary Figure S2C) and sdsN-lacZ induction at low temperature (Supplementary Figure S2A), we considered the possibility that expression might be influenced by the Crl assembly factor, which stimulates σS binding to RNA polymerase and thus enhances expression from σS-dependent promoters when Crl levels are elevated under conditions such as low temperature (28–30). Consistent with an influence of Crl, the levels of SdsN178 and SdsN137 are somewhat lower in a crl deletion strain grown in both LB (Supplementary Figure S2B) and M63 glucose (Figure 2B) medium.

Since Crl levels are modulated by nitrogen, with reduced Crl synthesis under conditions of nitrogen limitation (30), we examined SdsN expression in cells grown in M63 glucose medium with rich and poor nitrogen sources. Growth on the preferred nitrogen source ammonium was associated with higher SdsN levels than growth on the poor nitrogen source arginine (Figure 2C). There are a number of transcription factors that regulate gene expression in response to nitrogen availability. One of these is NarP, the response regulator of the NarP-NarQ two-component system, which mediates the nitrate/nitrite responsive transcriptional regulation of anaerobic respiration (31). Since we found narP to be a target of SdsN (see below), we considered the possibility of a feedback loop and examined SdsN levels in a ΔnarP background (Figure 2C). We did not observe a significant difference between the two strains grown with ammonium, but interestingly, the levels of SdsN178 and SdsN137 were elevated, and the level of SdsN124 was reduced in the ΔnarP strain grown with arginine.

Given the proximity between sdsN and sdiA and reports of SdiA autoregulation (32), we also wondered whether the LuxR-type SdiA transcription factor impacts SdsN levels. Primer extension analysis (Supplementary Figure S3A) and dRNA-seq data (27) revealed two sdiA transcription start sites. These are both absent in an ΔsdiA background but differ from the previously reported starts (33), which we still detect in the ΔsdiA mutant and thus are likely to be artifacts (Supplementary Figure S3B). Both of the corresponding promoters overlap the sdsN promoters (Figure 2A). We examined the levels of SdsN in two ΔsdiA backgrounds but did not find a significant difference in SdsN expression under the conditions tested (Supplementary Figure S3C).

SdsN137 and SdsN178 bind the RNA chaperone Hfq with different affinities

Energetically favorable secondary structures of SdsN137 (Figure 3A) and SdsN178 (Supplementary Figure S4) were predicted using mfold (34). These predictions are supported by in vitro RNase T1 cleavage (Figure 3B). SdsN137 and SdsN178 contain the same two stem-loop structures, with the 3′ stem-loop forming the Rho-independent terminator. The 5′ AU-rich region present only on SdsN178 appears to be largely single stranded.

Figure 3.

Both SdsN137 and SdsN178 bind Hfq. (A) Structure of SdsN137 predicted by the mfold algorithm (34). Sites of RNase T1 cleavage as determined from Figure 3B are indicated by solid black dots with the size of the dots proportional to the amount of cleavage. Arrows indicate enhanced (↑) or repressed (↓) cleavage when RNA is pre-mixed with the indicated amounts of purified Hfq hexamer. The 5′ end of SdsN124 is indicated with a vertical bracket. Regions involved in base pairing with narP at 5′ end and with nfsA and hmpA between two stem-loops are shaded. (B) Probing of SdsN137 and SdsN178 structures. Purified 32P end-labeled SdsN137 and SdsN178 were incubated with RNase T1 (Ambion) and, for indicated samples, 100 or 200 nM purified Hfq protein and run on an 8% polyacrylamide-7M urea sequencing gel. RNase T1 and OH ladders are shown. Sites of RNase T1 cleavage are indicated as in Figure 3A. (C) SdsN co-immunoprecipitation with Hfq. Cell extracts were prepared from WT MG1655 grown in LB to early stationary phase (OD600 ≈ 5) and subject to immunoprecipitation with α-Hfq or preimmune serum. Northern analysis was carried out on the immunoprecipitated samples (1 μg RNA loaded) as well as on total RNA (10 μg RNA loaded) isolated from WT MG1655 and the isogenic Δhfq::cat-sacB mutant (GSO748). (D) SdsN137 and SdsN178 both bind Hfq in vitro. Purified 32P end-labeled SdsN137 and SdsN178 (4 nM) were incubated with the indicated amounts of purified Hfq hexamer for 15 min at 37°C. Samples were run on a native 6% polyacrylamide gel. Unbound, complex I and complex II are indicated by U, I and II, respectively. The intensities for the bands in each lane were determined using Multi Gauge Imaging Software V3.0. Kd values were computed by plotting the appearance of protein–RNA complexes as a function of Hfq6 concentration using the saturation-binding feature of GraphPad Software.

While one report suggested the RNA encoded in the sdiA-tcyN intergenic region was bound by Hfq (25), another report showed the transcript was not stabilized by Hfq (26). To resolve this discrepancy, we subjected extracts from stationary phase cells to immunoprecipitation with either Hfq antibody or preimmune serum. Total RNA from WT and hfq mutant cells as well as immunoprecipitated RNA were probed for SdsN. The different forms of SdsN were all absent in the hfq mutant strain and all were enriched by immunoprecipitation with Hfq. However, we observed stronger enrichment for SdsN178 than for SdsN137 or SdsN124 (Figure 3C). The enhanced Hfq binding to SdsN178 relative to SdsN137 in vivo was also observed in an EMSA in vitro binding assay in which increasing amounts of purified Hfq were added to radiolabeled SdsN137 or SdsN178 in the presence of an excess of competitor RNA (Figure 3D). The addition of Hfq to both sRNAs resulted in two shifts in mobility (complex I and II), similar to what is observed upon Hfq binding to other sRNAs (35). The dissociation constant (Kd) for RNA bound to Hfq in complex I, was calculated to be ∼80 nM for SdsN137 and ∼28 nM for SdsN178, comparable to the 25 nM Kd observed for DsrA and RprA using the same assay (35). Together these results show Hfq binds SdsN178 with higher affinity than SdsN137 in vivo and in vitro. Structure probing carried out after incubating SdsN138 and SdsN178 with Hfq (Figure 3B), showed one region of protection (G96 G97 G99) for SdsN138, and two regions for SdsN178 (same region between two stem-loops and another in the 5′ extension).

SdsN137 targets mRNAs encoding proteins involved in the metabolism of and response to nitrogen compounds

Given SdsN137 and SdsN178 association with Hfq, we postulated the sRNAs were acting to modulate expression by basepairing with target mRNAs. To identify possible targets, we examined the genome-wide changes in RNA levels upon short-term overproduction of either SdsN137 or SdsN178. Exponentially-growing MG1655 cells carrying the pBR vector control, pBR-SdsN137 or pBR-SdsN178 were induced with 100 μM IPTG for 5 min after which total RNA was isolated and analyzed on microarrays (Supplementary Table S4). Table 1 lists genes showing greater than two-fold changes upon SdsN137 overexpression in the two independent experiments.

Table 1.

Transcripts changed more than 2-fold upon SdsN137 overexpression

Gene	Function	Array 1	Array 2
*nfsA	Oxygen-insensitive nitroreductase	7.7↓	2.7↓
*ybjC	Predicted oxidase co-transcribed with nfsA	2.4↓	2.6↓
*hmpA	NO dioxygenase, flavohemoglobin	2.9↓	2.3↓
*narP	Nitrate/nitrite response regulator	2.6↓	2.7↓
cyaY	Frataxin homolog with role in iron-sulfur cluster assembly	2.5↓	2.6↓
*uraA	High-affinity uracil/protein symport system	2.6↓	2.2↓
*inaA	Induced by low pH	2.1↓	2.2↓
rluA	23S rRNA and tRNA pseudouridine synthase	2.0↓	2.0↓
fliC	Flagellin	2.1↑	2.6↑
ymfT	Predicted DNA-binding transcription regulator, e14 phage	2.5↑	3.2↑

Arrows indicate genes repressed (↓) or induced (↑) upon SdsN137 overexpression. mRNAs for genes marked with an astericks (*) are bound by Hfq (20). Annotation of function is from Ecocyc.org (52).

SdsN137 overexpression led to the repression of three mRNAs encoding proteins associated with the metabolism of and response to nitrogen compounds: nfsA encodes an oxygen-insensitive NADPH-dependent nitroreductase that catalyzes the reduction of a nitrogroup (NO2) in nitroaromatic compounds to an amino group (NH2) (36), and hmpA encodes flavohemoglobin with nitric oxide (NO) dioxygenase activity under aerobic conditions and reductase activity under anaerobic conditions (37). ybjC, which encodes a predicted oxidoreductase, is co-transcribed with nfsA, indicating that the products of these two genes may have related physiological functions. A third mRNA whose levels are repressed by SdsN137 encodes the response regulator NarP (31).

The other transcripts whose levels were changed by SdsN137 overproduction did not fall into a single category (cyaY, uraA, inaA and rluA were down-regulated and fliC and ymfT were up-regulated). We generated lacZ fusions to cyaY, uraA and inaA as described for nfsA, hmpA and narP below. However, we did not detect expression from the inaA fusion, and the cyaY and uraA fusions showed no regulation by SdsN137 (data not shown).

Interestingly, SdsN178 overproduction did not affect any of the genes modulated by SdsN137 (data not shown). In fact, only one mRNA, casA encoding a subunit of CRISPR-associated complex, showed a 2-fold increase in levels upon SdsN178 overproduction in the two experiments.

SdsN137 base pairs directly with nfsA, hmpA and narP mRNAs

The ybjC-nfsA, hmpA and narP mRNAs all co-immunoprecipitate with Hfq (20) and are predicted to have complementarity with SdsN137 (Figure 4A, C and E) by the IntaRNA base pairing prediction program (38,39). For nfsA and hmpA, the predicted pairing is with the internal region of SdsN137 shown to be single stranded by structure probing of SdsN137 (Figure 3A and B). For narP, the predicted pairing is at the base of the 5′ stem, though this region is likely to be partially single stranded, as is the case for the FnrS 5′ stem (6). To test whether SdsN137 directly base pairs with these targets, we constructed chromosomal lacZ fusions with the 5′ UTR as well as first nine codons of nfsA and hmpA and the first 30 codons of narP fused to the tenth codon of lacZ, all transcribed from the heterologous arabinose-inducible PBAD promoter (21). As shown in Figure 4, all of the target-lacZ fusions were repressed 1.6- to 3-fold by SdsN137 overexpression compared to the vector control. Mutations were introduced into the two regions of SdsN137 predicted to pair with the nfsA, hmpA and narP mRNAs. Consistent with the predictions, the nfsA-lacZ and hmpA-lacZ fusions were not repressed by SdsN137-2 (Figure 4B and D), while narP-lacZ was not repressed by SdsN137-1 (Figure 4F) but was repressed by SdsN137-2 (data not shown). However, repression was regained for fusions carrying compensatory mutations, nfsA-2-lacZ, hmpA-2-lacZ and narP-1-lacZ, that restore base pairing with the SdsN137 mutants (Figure 4B, D and F). These results confirm direct base pairing between SdsN137 and the hmpA, nfsA and narP mRNAs, and show that SdsN137 utilizes at least two different regions to regulate targets.

Figure 4.

SdsN137 base pairs directly with nfsA, hmpA and narP. (A) Predicted base pairing between SdsN and nfsA. The bases mutated in pBR-SdsN137 and nfsA are indicated. (B) β-galactosidase activity of PM1205-derived strains with nfsA-lacZ (GSO751) or nfsA-2-lacZ (GSO754) carrying pBR control vector, pBR-SdsN137 or pBR-SdsN137 mutant derivative. (C) Predicted base pairing between SdsN and hmpA. The bases mutated in pBR-SdsN137 and hmpA are indicated. (D) β-galactosidase activity of PM1205-derived strains with hmpA-lacZ (GSO752) or hmpA-2-lacZ (GSO755) carrying pBR control vector, pBR-SdsN137 or pBR-SdsN137 mutant derivative. (E) Predicted base pairing between SdsN and narP. The bases mutated in pBR-SdsN137 and narP are indicated. (F) β-galactosidase activity of PM1205-derived strains with narP-lacZ (GSO753) or narP-1-lacZ (GSO756) carrying pBR control vector, pBR-SdsN137 or pBR-SdsN137 mutant derivative. For (A), (C) and (E), base pairing was predicted by IntaRNA (38,39), and numbering for mRNA is relative to start codon. For (B), (D) and (F), the levels of β-galactosidase activity of the different fusions were assayed after ∼1 h of induction with 0.2% arabinose and either 100 μM IPTG (grey bars) or no IPTG (black bars). The average values from three independent assays are shown with error bars corresponding to the standard deviation of those values.

SdsN137 protects against nitrofurans

Previous studies have shown that the NfsA nitroreductase sensitizes E. coli cells to nitrofurans, nitrogen compounds that contain one or more nitrogen groups on a nitroaromatic backbone, and that cells lacking NfsA are resistant to these antibiotics (40,41). To test whether SdsN regulation of nfsA affects the sensitivity to these compounds, we treated LB-grown cells with 1 mM of the nitrofurans nitrofurazone or azomycin for 1 h. Overexpression of SdsN137 led to more resistance to these antibiotics than the vector control, SdsN178 or SdsN137-2, the mutant unable to regulate nfsA-lacZ (Figure 5A and Supplementary Figure S5A). SdsN137 did not confer further resistance to nitrofurazone in a ΔnfsA background (Supplementary Figure S5B).

Figure 5.

SdsN confers resistance to nitrofurazone. (A) WT MG1655 cells carrying the pBR control vector, pBR-SdsN137, pBR-SdsN178 or pBR-SdsN137-2 were grown in LB supplemented with 100 μM IPTG and 100 μg/ml amp to OD600 ≈ 0.3. Nitrofurazone was added to a final concentration of 1 mM to each culture. After 1 h at 37°C, cultures were serial diluted and plated on LB plates. The number of colonies is given below each plate. (B) WT MG1655 and the corresponding ΔsdsN strain (GSO762) were grown in M63 glucose medium for 14 h. Nitrofurazone was added to a final concentration of 2 mM. After 1 h at 37°C, cultures were again serial diluted and plated on LB plates. The data shown are for assays carried out on three separate days, each time for three independent cultures. Statistical significance (*) was calculated using an unpaired, one-tailed T-test with both data sets (Pvalue = 0.048).

We also examined the sensitivity of WT and ΔsdsN mutant cells grown 14 h in M63 glucose medium, when SdsN levels are high, to a 1 h-exposure to 2 mM nitrofurazone. Consistent with chromosomally-expressed SdsN contributing to resistance against oxidized nitrogen compounds, we always observed fewer colonies for the ΔsdsN mutant strain than the WT parent strain after treatment with nitrofurazone (Figure 5B).

5′ end of SdsN178 inhibits regulation

We noted that none of the mRNAs showing regulation by SdsN137 in the microarray experiments were affected by SdsN178 overexpression. Since the two regions of SdsN137 involved in base pairing with nfsA, hmpA and narP are contained in SdsN178, we tested whether SdsN178 regulates the corresponding lacZ fusions. As seen in Figure 6A, SdsN178 was less effective at repressing all three fusions, particularly narP-lacZ.

Figure 6.

5′ A- and U-rich sequences block SdsN178 function. (A) β-galactosidase activity of PM1205-derived strains with narP-lacZ, nfsA-lacZ or hmpA-lacZ (GSO753, GSO751 and GSO752, respectively) carrying pBR control vector, pBR-SdsN137 or pBR-SdsN178. (B) β-galactosidase activity of PM1205-derived strain with narP-lacZ (GSO753) carrying pBR control vector, pBR-SdsN137, pBR-SdsN178 or plasmids with 5′ truncations of the SdsN178. (C) β-galactosidase activity of PM1205-derived strain with narP-lacZ (GSO753) carrying pBR control vector, pBR-SdsN137, pBR-SdsN178, pBR-SdsN140, pBR-SdsN140-3 (AUU to CGG mutant) and pBR-SdsN140-4 (AUU to GGG mutant). (D) β-galactosidase activity of PM1205-derived strain with narP-lacZ (GSO753) carrying pBR control vector, pBR-SdsN137, pBR-SdsN178, pBR-RyhB or pBR-SdsN5′-RyhB (chimeric RyhB carrying 5′ region of SdsN). For (A), (B), (C) and (D), the levels of β-galactosidase activity of the different fusions were assayed after ∼1 h of induction with 0.2% arabinose and either 100 μM IPTG (grey bars) or no IPTG (black bars). The average values from three independent assays are shown with error bars corresponding to the standard deviation of those values. (E) EMSA of 4 nM of 32P end-labeled SdsN137 with 100 nM Hfq hexamer and the indicated amounts (nM) of unlabeled hmpA or narP RNA. Ternary complexes are indicated with asterisks. (F) EMSA of 4 nM 32P end-labeled SdsN178 with 100 nM Hfq hexamer and the indicated amounts (nM) of unlabeled hmpA or narP RNA.

To elucidate what sequences present in SdsN178 reduce regulatory function, we constructed a number of deletion, mutant and chimeric derivatives of SdsN178. The 5′ end sequence of SdsN178 is extremely AU rich, and previous studies have shown AU-rich sequences are preferentially bound by Hfq and cleaved by RNase E (4). We first generated a series of 5′-truncations of SdsN178 and assayed narP-lacZ activity upon overexpression of these derivatives (Figure 6B). All of the truncations were less effective than SdsN137 at repressing the narP-lacZ fusion, including the shortest truncation SdsN140, which is only three nucleotides longer than SdsN137. We further mutated the three AU residues at the 5′ end of SdsN140 by changing the residues to GC-rich sequences (AUU to CGG or GGG). With these changes, both of the mutants repress narP-lacZ as well as SdsN137 (Figure 6C). It is possible some of the defects in regulation are due to somewhat lower levels of the mutant derivatives (Supplementary Figures S6A and S6B), but the sRNAs expressed from the plasmids are in vast excess. We also fused the 5′ end of SdsN178 (nt 1–41) to the 5′ ends of the well-characterized RyhB and FnrS sRNAs, which both repress sodB. The addition of the SdsN sequence to both sRNAs caused a reduction in sRNA-mediated repression of a sodB-lacZ translational fusion despite similar sRNA levels (Supplementary Figure S6C and S6D). Interestingly, we found RyhB also represses the narP-lacZ fusion (Figure 6D). This regulation likely is via direct base pairing, given extensive predicted pairing (Supplementary Figure S6E) and high expression of the narP-lacZ fusion in a Δhfq mutant, consistent with repression by multiple sRNAs (Supplementary Figure S6F). As for SdsN178, the chimeric form of RyhB did not repress the narP-lacZ fusion (Figure 6D). The appended SdsN sequence could alter the structures of base pairing regions of RyhB and FnrS, but because the SdsN 5′ sequence reduces the activities of both sRNAs and the retention of base pairing has been surprisingly robust in other chimeric sRNAs (24), we do not think an altered structure fully explains the reduced repression.

To further dissect the lack of target regulation by SdsN178, we examined SdsN137 and SdsN178 base pairing with target mRNAs in vitro. Radiolabeled SdsN137 and SdsN178 were mixed with Hfq at concentrations to ensure most of the sRNAs were bound by Hfq. The addition of increasing amounts of unlabeled hmpA or narP mRNA fragments to the SdsN137-Hfq complex led to ternary complexes formed between Hfq, SdsN137 and hmpA or narP (Figure 6E). In contrast, no ternary complex was observed when the mRNA fragments were added to SdsN178-Hfq, even at the highest concentration of unlabeled RNA (Figure 6F).

Cleavage leads to SdsN124 capable of regulating hmpA and nfsA

For cells grown to late stationary phase in M63 glucose medium, the levels of a third transcript, SdsN124, were high (Figure 1B). This derivative is derived from processing given the transcript is eliminated when cells are treated with terminator exonuclease (Supplementary Figure S1C). Since a band corresponding to SdsN124 can be detected when SdsN178 or SdsN137 are overexpressed from a plasmid (Figure 7A), we suggest that this derivative can be derived from either of the longer sRNAs. To test whether SdsN124 can serve as a regulator, we again assayed the lacZ fusions (Figure 7B). As expected, given that SdsN124 lacks the region for base pairing with narP, this fusion was not affected by SdsN124 overexpression. In contrast, SdsN124 repressed nfsA-lacZ and hmpA-lacZ as well as or slightly better than SdsN137.

Figure 7.

Cleavage activates SdsN178. (A) Levels of SdsN178, SdsN137 and SdsN124 expressed from pBR plasmid. Total RNA was isolated from the GSO752 cultures assayed in Figure 7B. (B) β-galactosidase activity of PM1205-derived strains with narP-lacZ, nfsA-lacZ or hmpA-lacZ (GSO753, GSO751 and GSO752, respectively) carrying pBR control vector, pBR-SdsN137, pBR-SdsN178 or pBR-SdsN124. The levels of β-galactosidase activity of the different fusions were assayed after ∼3 h of induction with 0.2% arabinose and 1 mM IPTG. The average fold difference (for four independent assays) relative to the corresponding pBR vector control samples are shown with error bars corresponding to the standard deviation of the differences. (C) Effect of mutations affecting, rim (R16A), proximal (Q8A) and distal (Y25D, K31A) faces of Hfq. WT cells and strains expressing the indicated mutant Hfq derivatives from the chromosome (20) (DJS2317, DJS2318, DJS2319, DJS2294 and DJS2321) were grown 20 h in M63 glucose. For (A) and (C), RNA was processed for northern analysis as in Figure 1.

Previous studies have shown that while all sRNAs bind to the proximal face of the Hfq hexamer, interactions with other Hfq surfaces can differ (24). Most sRNAs (Class I) also bind the rim; these contact mRNAs on the distal face and are thus stabilized by mutations of distal face residues. A subset of sRNAs (Class II) binds the distal face; these contact mRNAs on the rim and are stabilized by mutations of rim residues. Given that sRNA binding is reflected in their levels in strains expressing proximal (Q8A), rim (R16A) and distal (Y25D, K31A) face mutants of Hfq, we examined SdsN178, SdsN137, SdsN124 levels in these backgrounds (Figure 7C). The low levels of SdsN124 in the Q8A and R16A mutants together with the elevated levels in the Y25D and K31A mutant, suggest that SdsN124 behaves like a typical Class I binding the proximal and rim surfaces, leaving the distal face free to bind the target mRNA. In contrast, SdsN137 and SdsN178 are present at WT levels in the rim and distal mutants and are only partially decreased in the Q8A mutant, consistent with multiple contact sites including the distal face.

DISCUSSION

The set of SdsN RNAs can be added to a growing list of σS-dependent sRNAs that have been characterized for E. coli. In addition to σS, we found that the Crl and NarP transcription factors impact SdsN expression. Crl is not a DNA binding protein and is not known whether σS and NarP directly bind to the sdsN promoters. There is no signal for σS in the sdsN region in a recent chromatin immunoprecipitation-sequencing experiment (42), but there also did not appear to be binding to the gadY and sraL promoters. We were intrigued to find that multiple transcripts are expressed from the sdiA-tcyN intergenic region. Two forms, SdsN137 and SdsN178, are transcribed from two different σS-dependent promoters but share a terminator. These forms can be processed at the 5′ end to generate SdsN124, which accumulates maximally after 20 h growth in M63 glucose media. The relative levels of these three main transcripts vary depending on the growth condition in ways we do not yet fully understand. SdsN124, SdsN137 and SdsN178 all bind Hfq but only SdsN124 and SdsN137 were found to be effective regulators.

σS-dependent sRNAs share similar and differing features

There are some interesting similarities and differences among the σS-dependent sRNAs. SraL and SdsR are conserved in a range of enteric bacteria, while SdsN and GadY are restricted to E. coli and Shigella. SdsN, GadY and SdsR all strongly bind Hfq, while SraL does not. In addition, though synthesis of GadY, SraL, SdsR and SdsN is σS-dependent, they show different patterns of accumulation, with SdsR generally expressed later than the other sRNAs in stationary phase (Figure 1B). This observation suggests that additional factors impact the transcription or stability of the σS-dependent sRNAs as we found for SdsN. Little is known about what other factors might modulate GadY, SraL and SdsR levels, but it is interesting to note that overlapping transcripts are encoded on the strand opposite each of these sRNAs; sdiA mRNA for SdsN, gadXW mRNA for GadY, soxR mRNA for SraL and RyeA sRNA for SdsR (27).

The mRNA targets of the σS-dependent sRNAs do not show any functional overlap. GadY increases the synthesis of acid response transcription factors and thus may protect against low pH in stationary phase (13). The contribution of the SraL block of the translation of the Trigger Factor chaperone to stationary survival is less clear (14). SdsR represses expression of the MutS component of methyl-directed mismatch repair, possibly enhancing mutagenesis when cells are starved (16). SdsN represses genes involved in the metabolism of oxidized nitrogen compounds. By repressing the synthesis of NfsA, SdsN provides resistance to nitrofurans (Figure 5 and Supplementary Figure S5). The physiological role of SdsN repression of hmpA is less clear. The flavohemoglobin eliminates NO under aerobic conditions, but we did not observe any difference in NO sensitivity in the ΔsdsN mutant (data not shown). However the enzyme has also been shown to have other activities (43), which may be detrimental under conditions of SdsN induction.

sRNAs are induced by various nitrogen compounds

Only a few sRNAs whose expression is responsive to the levels of nitrogen compounds have been characterized in bacteria thus far (44–46). One example is RoxS from Staphylococcus aureus and Bacillus subtilis, whose transcription is induced by the presence of NO by the ResDE two-component transduction system (44). B. subtilis cells lacking RoxS have increased expression of genes whose functions are related to oxidative stress and oxidation-reduction reactions. Among these, the ppnKB mRNA encoding a NAD+/NADH kinase was confirmed to be a direct target of RoxS. However, apart from a few genes involved in the TCA cycle, there was little overlap in RoxS targets between S. aureus and B. subtilis. Another example is NrsZ from Pseudomonas aeruginosa, whose transcription is induced by nitrogen limitation by the NtrBC two-component transduction system (45). The only confirmed target of NrsZ, rhlAB encodes enzymes for the synthesis of rhamnolipid biosurfactants required for swarming motility.

Multiple forms of an sRNA allow for additional levels of regulation

SdsN178 showed only partial regulation of nfsA and hmpA and almost no regulation of narP (Figure 6A), despite stronger binding to Hfq (Figure 3C and D). The secondary structure of SdsN178 is similar to SdsN137 indicating the base-pairing sites should be accessible. However, the 5′ end of SdsN178 is extremely AU-rich. In vitro, Hfq protects the 5′ region present on SdsN178 from RNase T1 cleavage (Figure 3B); thus this region is an Hfq binding site not present on SdsN137. However, we found SdsN137 was better able to form the Hfq-sRNA-mRNA target ternary complex (Figure 6E and F) required for base pairing (47). This difference in the two forms can be explained by the 5′ end of SdsN178 competing with the mRNA for the same surface of Hfq.

SdsN124 represses nfsA and hmpA substantially better than SdsN178 and slightly better than SdsN137 (Figure 7B). The stronger regulation is not due to higher abundance since the levels of SdsN124 are lower than SdsN137 and equal to SdsN178 (Figure 7A). Thus, cleavage of SdsN178 and SdsN137 to generate SdsN124 probably improves base pairing with hmpA and nfsA through altered secondary structure or differences in the interaction with Hfq (Figure 7C). Cleavage also leaves a 5′-monophosphate, which may promote RNase E cleavage of the target, as was shown for MicC (48).

A number of other sRNAs have been found to be activated by post-transcriptional cleavage of the 5′ end. The first 20 nt of B. subtilis RoxS are removed by RNase Y generating a truncated version, which was found to be an efficient regulator in vivo and form a more extended duplex with the ppnKB mRNA in vitro, possibly due to disruption of the 5′ stem (44). However, the amount of this cleavage product relative to full-length is quite low under conditions tested. The two forms of E. coli RprA have different stabilities; the unprocessed form is subject to rapid degradation by RNase E, while the shorter cleaved form is more stable. Like SdsN, the relative levels of the two isoforms vary during growth. Intriguingly, each isoform contains seed regions responsible for regulation of different sets of mRNA targets (49). A final example is E. coli ArcZ (50). Again the primary transcript (ArcZ121) is processed at the 5′ end to give a shorter form (ArcZ56). The 5′ region of ArcZ121 contains AU-rich sequences that are absent in ArcZ56. As seen for SdsN, ArcZ121 binds Hfq more tightly than ArcZ56, in vivo, but ArcZ56 forms a more stable ternary complex with Hfq and its mRNA target in vitro. In addition, the levels of the processed form are greatly elevated in strains expressing distal-face Hfq mutants (20), as we also observe for the cleaved derivatives of SdsN as well as GadY (Figure 7C). Given that ArcZ56 shows greater regulation of its target than ArcZ121, it is hypothesized that initial ArcZ expression produces a functionally inert sRNA, but is activated upon cleavage that removes the extra Hfq binding site at the 5′ end (50). We suggest that similar activation is occurring with SdsN178. However, while ArcZ121 is almost entirely processed under most conditions tested, SdsN124 levels are only high in the late stationary phase in minimal media with a rich nitrogen source (Figures 1B and 2C).

We speculate additional characterization of SdsN178, SdsN137 and SdsN124, as well as the processed derivatives of other sRNAs, will reveal further roles for these alternative forms. For sRNAs with multiple base pairing regions, alternative transcription and processing can lead to the regulation of different sets of target mRNAs under slightly different environmental conditions. Expression of sRNAs that bind Hfq in an inert state can prevent Hfq binding to competitor sRNAs and perhaps poise the sRNA to serve as a regulator upon cleavage in response to a specific input. These additional levels of regulation, together with tight transcriptional regulation, competition for limiting Hfq and ‘sponge RNAs’ described previously, illustrate the exquisite control of sRNA regulators (2,51).