Flagella are important for bacterial motility as well as for pathogenesis. Synthesis of these structures is energy intensive and, while extensive transcriptional regulation has been described, little is known about the posttranscriptional regulation. Small RNAs (sRNAs) are widespread posttranscriptional regulators, most base pairing with mRNAs to affect their stability and/or translation. Here we describe four UTR-derived sRNAs (UhpU, MotR, FliX and FlgO) whose expression is controlled by the flagella sigma factor σ28 (fliA) in Escherichia coli. Interestingly, the four sRNAs have varied effects on flagellin protein levels, flagella number and cell motility. UhpU, corresponding to the 3’ UTR of a metabolic gene, likely has hundreds of targets including a transcriptional regulator at the top flagella regulatory cascade connecting metabolism and flagella synthesis. Unlike most sRNAs, MotR and FliX base pair within the coding sequences of target mRNAs and uniquely act on ribosomal protein mRNAs connecting ribosome production and flagella synthesis. The study shows how sRNA-mediated regulation can overlay a complex network enabling nuanced control of flagella synthesis.
The full role of small RNAs, often key global regulators of gene expression in connection with environmental adaptation, in flagella regulation has not been determined. This study presents important findings regarding the function of small RNAs in bacterial motility and regulation of flagella biosynthesis regulation and number. The authors combine multiple methods, including RIL-seq, structural probing, RNA blots and reporter assays, to provide convincing biochemical evidence for direct effects of specific small RNAs, whose expression is dependent on flagella sigma 28 (FliA), on transcriptional attenuation of flagella operons via regulation of ribosomal proteins.
Most bacteria are motile and can swim through liquid and semiliquid environments in large part driven by the flagellum. The highly complex bacterial flagellum consists of three major domains: an ion-driven motor, which can provide torque in either direction; a universal joint called the hook-basal body, which transmits motor torque; and a 20 nm thick hollow filament tube composed of the flagellin subunit, which acts as a propeller (reviewed in (Altegoer & Bange, 2015, Nakamura & Minamino, 2019)). The flagellum structure is comprised of many proteins, and the flagellar regulon encompasses more than 50 genes. Since it is estimated that one flagellum constitutes ∼2% of the total protein in the cell, synthesis is a costly process requiring extensive use of ribosomes (reviewed in (Soutourina & Bertin, 2003, Guttenplan & Kearns, 2013)).
To ensure that flagellar components are made in the order in which they are needed, transcription of the genes in the regulon is activated in a sequential manner in Escherichia coli (Kalir et al., 2001) and Salmonella enterica (reviewed in (Chevance & Hughes, 2008)). The genes can be divided into three groups based on their time of activation: early genes, middle genes, and late genes (Fig 1A). The FlhDC transcription regulators, encoded by the two early genes, activate the transcription of the middle genes, which are required for the hook-basal body. FlhDC also activates transcription of fliA, encoding sigma factor σ28 (Fitzgerald et al., 2014). σ28 in turn activates transcription of the late genes responsible for completing the flagellum and the chemotaxis system. σ28 additionally increases expression of several of the middle genes. σ28 activity itself is negatively regulated by the anti-sigma factor, FlgM, which is transported out of the cell when the hook-basal body complex is complete (reviewed in (Smith & Hoover, 2009, Osterman et al., 2015). Given the numerous components required at different times and different stoichiometries during flagellum assembly, various factors can be rate limiting under specific conditions (reviewed in (Chevance & Hughes, 2008)). The dependence of flagella synthesis on FlhDC and σ28 generates a coherent feed-forward loop. In this loop, the first regulator (FlhDC) activates the second regulator (σ28), and they both additively activate their target genes. This results in prolonged flagellar expression, protecting the flagella synthesis from a transient loss of input signal (Kalir et al., 2005).
Given flagella are so costly to produce, synthesis is tightly regulated such that flagellar components are only made when motility is beneficial. Thus, flagellar synthesis is strongly impacted by environmental signals. For instance, flagellar gene expression is decreased in the presence of D-glucose, in high temperatures, high salt, and extreme pH, as well as the presence of DNA gyrase inhibitors (Shi et al., 1993, Adler & Templeton, 1967). The flagellar genes are activated under oxygen-limited conditions (Landini & Zehnder, 2002) and at various stages of infection (reviewed in (Erhardt, 2016)). Consequently, transcription of many genes in the flagellar regulon is regulated in response to a range of environmental signals. For example, transcription of flhDC is controlled by at least 13 transcription factors, each of them active under different conditions (reviewed in (Prüß, 2017)).
While the activation of flagella synthesis has been examined in some detail, there has been less investigation into the termination of synthesis, which we presume is equally important for the conservation of resources. Additionally, while transcriptional regulation of flagella genes has been studied for many years, less is known about the post-transcriptional control of the regulon. Small RNAs (sRNAs) that can originate from many different genetic loci (reviewed in (Adams & Storz, 2020)) are key post-transcriptional regulators in bacteria. They usually regulate their targets in trans via limited base-pairing, affecting translation and/or mRNA stability (reviewed in (Hör et al., 2020)). Many characterized sRNAs are stabilized and their base pairing with targets increased by RNA chaperones, of which the hexameric, ring-shaped Hfq protein has been studied most extensively (reviewed in (Updegrove et al., 2016, Holmqvist & Vogel, 2018)). The only post-transcriptional control by base pairing sRNAs described for the E. coli flagellar regulon thus far is negative regulation of flhDC by ArcZ, OmrA, OmrB and OxyS (De Lay & Gottesman, 2012) and positive regulation of the same mRNA by McaS (Thomason et al., 2012). These sRNAs and a few other sRNAs also were shown to affect motility and biofilm formation (Bak et al., 2015).
Herein we characterize four σ28-dependent sRNAs, which were detected with their targets on Hfq through RIL-seq methodology that captures the sRNA-target interactome (Melamed et al., 2016, Melamed et al., 2020). These sRNAs originate from the untranslated regions (UTRs) of mRNAs, three of which belong to the flagellar regulon. We identified a wide range of targets for the sRNAs, including genes related to flagella and ribosome synthesis and observed that the sRNAs act on some of these targets by unique modes of action. We show that three of these sRNAs regulate flagella number and bacterial motility, possibly imposing temporal control on flagella synthesis and integrating metabolic signals into this complex regulatory network.
σ28-dependent sRNAs are expressed sequentially in log phase cells
Analysis of several different RNA-seq data sets suggested the expression of four σ28-dependent sRNAs in E. coli (Melamed et al., 2016, Fitzgerald et al., 2014, Thomason et al., 2015, Melamed et al., 2020, Bar et al., 2021). These sRNAs originate from the UTRs of protein coding genes (Fig 1B and S1A). UhpU corresponds to the 3’ UTR of uhpT, which encodes a hexose phosphate transporter (Marger & Saier, 1993). UhpU is transcribed from its own promoter inside the coding sequence (CDS) of uhpT (Thomason et al., 2015). The other three σ28-dependent sRNAs correspond to the UTRs of the late genes in the flagellar regulon. MotR originates from the 5’ UTR of motA, which encodes part of the flagellar motor complex. Based on previous transcription start site analysis, the promoter for motR is within the flhC CDS and is also the promoter of the downstream motAB-cheAW operon (Thomason et al., 2015, Fitzgerald et al., 2014). FliX originates from the 3’ UTR of fliC, which encodes flagellin, the core component of the flagellar filament (reviewed in (Thomson et al., 2018)). FlgO originates from the 3’ UTR of flgL, a gene that encodes a junction protein shown to connect the flagella to the hook in S. enterica (Ikeda et al., 1987). The observation that FliX and FlgO levels decline substantially in RNA-seq libraries treated with 5’ phosphate-dependent exonuclease to deplete processed RNAs (Thomason et al., 2015), indicates that both of these sRNAs are processed from their parental mRNAs.
Northern blot analysis confirmed σ28-dependent synthesis of these sRNAs since expression was significantly decreased in a mutant lacking σ28 (ΔfliA) (Fig 1C). Given that most σ28-dependent mRNAs encode flagella components, the σ28 regulation suggests the sRNAs impact flagella synthesis. The northern analysis also showed that the levels of the four σ28-dependent sRNAs are highest in the transition from mid-exponential to stationary phase growth, though there are some differences with UhpU and MotR peaking before FliX and FlgO (Fig 1C and S2). Since flagellar components are expressed at precise times, the difference in the UhpU and MotR peak times compared to the FliX and FlgO peak times hints at different roles for each of these sRNAs. For UhpU, two predominant bands were observed, a long transcript and a shorter transcript processed from UhpU (denoted UhpU-S), which corresponds to the higher peak in the sequencing data (Fig 1B). One prominent band was detected for MotR and for FliX, while a doublet was observed for FlgO. Additional higher bands detected by the MotR probe could be explained by RNA polymerase read through of the MotR terminator into the downstream motAB-cheAW operon, while the additional bands seen for FliX could be explained by alternative processing of the fliC mRNA.
We also examined the levels of the four sRNAs in minimal media (M63) supplemented with different carbon sources (Fig S1B). Generally, the sRNAs levels in minimal medium are comparable to their levels in rich media (LB) except medium with glucose 6-phosphate (G6P), where the levels of UhpU-S are elevated while the levels of full-length UhpU transcript and the other σ28-dependent sRNAs are decreased. These observations suggest an alternative means for UhpU-S generation from the uhpT mRNA known to be induced by G6P (Verhamme et al., 2001). We also observe more FliX products, particularly for cells grown in minimal medium with ribose or galactose.
The predicted structures for the four σ28-dependent sRNAs (Fig S1C), with strong stem-loops at the 3’ ends, are consistent with the structures of known Hfq-binding sRNAs and the association with Hfq observed in the RIL-seq data (Melamed et al., 2016). To confirm Hfq binding, we probed RNA that co-immunoprecipitated with Hfq (Fig 1D). Strong enrichment and fewer background bands were observed for all of the sRNAs; ∼260 nt and ∼60 nt bands for UhpU and UhpU-S, respectively, a ∼95 nt band for MotR, a ∼200 nt band for FliX and a doublet of ∼75 nt bands for FlgO. For FliX, we also detected a second ∼100 nt FliX band (denoted FliX-S; Fig S1A) that corresponds to the 3’ peak in the sequencing data (Fig 1B) and includes one of the REP sequences downstream of fliC.
σ28-dependent sRNAs impact flagella number and bacterial motility
To begin to decipher the roles of the four σ28-dependent sRNAs, we constructed plasmids for overexpression of the sRNAs (Fig S3A). Given that it was challenging to obtain constructs constitutively overexpressing UhpU because all clones had mutations, this sRNA could only be expressed from a plasmid with a IPTG-inducible Plac promoter (Guo et al., 2014). The other sRNAs were expressed from a plasmid with the constitutive PLlacO-1 promoter (Urban & Vogel, 2007). We also obtained a plasmid constitutively overexpressing MotR*, a more abundant derivative of MotR identified by chance (TGC at positions 6-8 mutated to GAG; Fig S1A).
We tested the effects of overexpressing the sRNAs on flagellar synthesis by determining the number of flagella by electron microscopy and on bacterial motility by assaying the spread of cells on 0.3% agar plates. The WT E. coli strain used throughout the paper is highly motile due to an IS1 insertion in the crl gene, thus eliminating expression of a protein that promotes σS binding to the RNA polymerase core enzyme (Typas et al., 2007), and resulting in higher expression of the flagellar regulatory cascade (Pesavento et al., 2008). However, we also assayed a less motile strain with the restored crl+ gene for UhpU and MotR effects on motility, given that no effects were observed with the highly motile crl- strain.
Intriguingly, overexpression of the individual sRNAs had different consequences. UhpU overexpression caused a slight increase in flagella number (Fig 2A) and a marked increase in motility (Fig 2B). Overexpression of MotR, particularly MotR*, led to a dramatic increase in the flagella number (Fig 2C and S4A) and MotR but not MotR* had a slight effect on motility (Fig 2D and S4B). It has been suggested that the run/tumble behavior of bacteria, which affect their swimming, is only weakly dependent on number of flagella (Mears et al., 2014), possibly explaining these somewhat contradictory effects on flagella number compared to motility. In contrast to UhpU and MotR, FliX overexpression led to a reduction in the number of flagella (Fig 2E), an effect that was even more pronounced in a strain overexpressing FliX-S (Fig S4C). Overexpression of FliX-S but not FliX also reduced bacterial motility (Fig 2F and S4D). While FliX-S overexpression seems to lead to aflagellated bacteria, we hypothesize that the sRNA is delaying but not eliminating flagella gene expression, explaining why the bacteria are still moderately motile. Some motility phenotypes can be explained by differences in growth rate, but we do not think that this is the case for MotR and FliX as we observed only slight effects on growth upon MotR, MotR*, FliX and FliX-S overexpression (Fig S3B). FlgO overexpression did not result in a detectable change in flagella number or motility (Fig 2G and 2H). Together, these results show that the σ28-dependent sRNAs have a range of effects on the number of flagella and motility, with UhpU and MotR, which are expressed first, increasing both phenotypes and FliX, which is expressed later, decreasing both. Given that MotR* and FliX-S have stronger effects for some phenotypes, these derivatives were included in some subsequent assays.
σ28-dependent sRNAs have wide range of potential targets based on RIL-seq analysis
To understand the phenotypes associated with overexpression of the σ28-dependent sRNAs expression, we took advantage of the sRNA-target interactome data obtained by RIL-seq (Melamed et al., 2020, Melamed et al., 2016). We analyzed RIL-seq data generated from 18 samples representing six different growth conditions (Table S1). The sRNAs differ significantly in their target sets (Fig S5). In general, UhpU is a hub with hundreds of RIL-seq targets. Its target set comprises a wide range of genes, including multiple genes that have roles in flagella synthesis and carbon metabolism. MotR and FliX were associated with fewer targets, but uniquely, both sets were enriched for genes encoding ribosomal proteins. Interestingly, the fliC gene encoding flagellin was present in the target sets for UhpU, MotR, and FliX. Although FlgO is one of the most strongly enriched sRNAs upon Hfq purification (ranked fourth in Melamed et al., 2020), it had the smallest set of targets, and almost none of the targets were found in more than two conditions, hinting FlgO might not act as a conventional Hfq-dependent base-pairing sRNA. Unlike for most characterized sRNA targets, the RIL-seq signal for the sRNA interactions with fliC and the ribosomal protein genes is internal to the CDSs (Table S1). Before turning our attention to these unique targets, we first examined the UhpU interaction with a canonical target.
UhpU represses expression of the LrhA transcriptional repressor of flhDC
We were intrigued to find that the mRNA encoding the transcription factor LrhA, which represses flhDC transcription, was among the top RIL-seq interactors for UhpU (Table S1). The signals that activate this LysR-type transcription factor (Lehnen et al., 2002), are not known, but the lrhA mRNA has an unusually long 5’ UTR (Fig 3A), a feature that has been found to correlate with post-transcriptional regulation (reviewed in (Adams & Storz, 2020)). Chimeras between UhpU and lrhA were detected in the majority of the conditions examined, and the predicted base pairing between UhpU and the lrhA 5’-UTR (Fig 3B) corresponds to the seed sequence suggested for UhpU (Melamed et al., 2016).
To test the effects of UhpU on this target, we fused the 5’ UTR of lrhA, which includes the region of the RIL-seq lrhA-UhpU chimeras and the predicted base-pairing region, to a lacZ reporter (Mandin & Gottesman, 2009). UhpU overexpression reduced expression of the chromosomally-encoded PBAD-lrhA-lacZ reporter (Fig 3C). A single nucleotide mutation in the base pairing region of uhpU (uhpU-M1) eliminated UhpU repression of lrhA-lacZ, while a complementary mutation introduced into the chromosomal lrhA-lacZ fusion (lrhA-M1) restored the repression providing direct evidence for UhpU base pairing to lrhA leading to repression. Down-regulation of LrhA by UhpU, which is expected to lead to increased FlhDC levels, is in accord with the positive impact of UhpU on motility (Fig 2). To test this model, we monitored the effect of UhpU on bacterial motility in a lrhA deletion strain compared to a WT strain (Fig 3D). Motility was increased in the WT background as expected. In contrast, while the ΔlrhA strain was more motile, likely due to flhDC de-repression, motility was unaltered by UhpU overexpression indicating that significant UhpU effects on motility are mediated by LrhA.
Interestingly, the RIL-seq data also suggested that lrhA directly interacts with other sRNAs such as RprA, ArcZ and McaS (Fig 3A). Regions of predicted base pairing overlap known seed regions for these sRNAs (Fig 3E). In translational reporter assays using the lrhA-lacZ fusion, both RprA and ArcZ reduced expression, while McaS, despite having the most chimeras, had no effect (Fig 3F). Possibly the McaS-lrhA interaction has other regulatory consequences such as McaS inhibition.
UhpU, MotR and FliX modulate flagellin levels
The high number of chimeras between UhpU, MotR or FliX with the fliC mRNA encoding flagellin were striking, particularly between the 3’ end of fliC corresponding to FliX (blue) and the 5’ end of fliC (red) (Fig 4A). As mentioned above, it was also noteworthy that most of the chimeras were internal to the fliC CDS. When we examined the consequences of overexpressing UhpU, MotR, MotR*, FliX or FliX-S on the levels of the flagellin protein, we observed somewhat increased levels of flagellin, both as cytosolic monomers (Fig 4B) and de-polymerized flagella (Fig S6A) with UhpU and MotR* overexpression and reduced levels with FliX or FliX-S overexpression. These differences are reflected in increased levels of the fliC mRNA with overexpression of UhpU, particularly in a crl+ background, or MotR or MotR*, particularly at OD600 ∼0.2 (Fig 4C and S6B). In contrast, fliC mRNA levels decreased with FliX and FliX-S overexpression (Fig 4C and S6B). The impacts of the sRNAs on flagellin protein and fliC mRNA levels are consistent with the increased flagella number and/or motility upon UhpU or MotR overexpression and decreased flagella number upon FliX overexpression.
We predicted base pairing between the three sRNAs and sequences overlapping the RIL-seq peaks internal to the fliC CDS (Fig 4D) and encompassing seed sequences suggested for the sRNAs (Melamed et al., 2016). To test for UhpU, MotR and FliX base pairing with these predicted sequences, we carried out in vitro footprinting with labeled fragments of the fliC mRNA (Fig S7). Upon cleavage with RNase III and lead, we observed changes in the regions predicted to be involved in base pairing (thick brackets) that were dependent on the WT RNAs but not with derivatives carrying mutations in the regions predicted to be involved in base pairing. We also observed Hfq dependent changes (thin bracket) in the region from ∼+40 to +66 from the fliC AUG, which is enriched for ARN sequences (AAA, AAT, AAC, AAG, AAC), known to be important for mRNA binding to the distal face of Hfq binding (reviewed in (Updegrove et al., 2016)). Additionally, we noted that both MotR and the MotR-M1 mutant RNAs led to increased cleavage (asterisks) at some positions and additional protection at other regions (dotted bracket) suggesting a second region of MotR base pairing with fliC as well as MotR induced structure changes. In general, the differences in cleavage by RNase III (preference for double-stranded RNA) and lead (preference for single-stranded RNA), indicate the fliC sequence from ∼+40 to ∼+170 is more structured than the surrounding regions. These differences in secondary structure could be the reasons for positive regulation by UhpU and MotR and negative regulation by FliX but also complicate analysis using standard reporter fusions with compensatory mutations.
MotR and FliX modulate the S10 operon
Given that genes encoding ribosomal proteins were almost exclusively found in chimeras with MotR and FliX and no other sRNAs, we investigated MotR and FliX regulation of these genes. Several of the top interactions for MotR and FliX in the RIL-seq data mapped to the essential S10 operon, again within the CDSs (Table S1, Fig 5A, 5F and S5). The co-transcriptional regulation of the S10 operon has been studied extensively (Zengel & Lindahl, 1996, Zengel et al., 2002, Zengel & Lindahl, 1992). The leader sequence upstream of the first gene rpsJ encoding S10 is bound by the ribosomal protein L4, encoded by the third gene in the operon (rplD), causing transcription termination, thus modulating the levels of all the ribosomal proteins in the operon in response to the levels of unincorporated L4. L4 binding also has been shown to specifically inhibit translation of rpsJ, an effect that can be genetically distinguished from the L4 effect on transcription termination (Freedman et al., 1987).
To test for MotR regulation of rpsJ expression, we fused the S10 leader and part of the rpsJ CDS, including the position of the rpsJ-MotR chimeras (Fig 5A) and the region of predicted base-pairing (Fig 5B), to a GFP reporter (Corcoran et al., 2012, Urban & Vogel, 2009). MotR overexpression elevated the levels of the rpsJ-gfp fusion, and MotR* enhanced this effect (Fig 5C). Positive regulation of S10 expression by MotR and MotR* was similarly observed by immunoblot analysis of an N-terminal FLAG-tagged S10 protein encoded along with the S10 leader behind the heterologous promoter on a pBAD plasmid (Fig 5D). A mutation in the MotR seed sequence (MotR-M1 and MotR*-M1, Fig S1A) eliminated the up-regulation of the FLAG-tagged S10 (Fig 5D) and the MotR effect on motility (Fig S8A). To examine base pairing between MotR and the sequences internal to the rpsJ CDS, we carried out in vitro structure probing in the presence of Hfq (Fig S9A and S9B). The RNase and lead cleavage assays supported the position of the predicted base-pairing between MotR and rpsJ mRNA, indicating MotR binds to rpsJ at ∼+150 nt in its CDS. Again, we detected Hfq binding (thin bracket), here to the attenuator hairpin in the S10 leader sequence (Fig S8B), which has three ARN sequences (AGG, AGU and AAC). The M1 mutation eliminated binding in the predicted region of pairing but a complementary mutation in the corresponding region of rpsJ mRNA did not restore MotR binding (Fig 5E). We suggest that, as for the MotR target region of fliC, MotR binds to more than one site, the MotR target region of rpsJ is highly structured, and MotR and Hfq binding might all lead to conformational changes that compound the interpretation of the mutations.
Nevertheless, to further define the determinants needed for MotR-mediated up regulation, we generated a series of rpsJ-gfp fusions to include only the first seven amino acids of S10 removing the MotR base pairing site, to remove the S10 leader sequence, to remove stem D required for L4-mediated regulation, or to remove the attenuator hairpin stem E (Fig S8B). MotR-dependent regulation was eliminated for each of these constructs suggesting that S10 leader sequence is needed along with the MotR binding site internal to rpsJ CDS for MotR-dependent regulation (Fig S8B). To test if Hfq binding to rpsJ is critical for the activation, we repeated the GFP reporter assay in an HfqY25D mutant defective for binding ARN sequences on the distal face of the protein (Zhang et al., 2013a). Supporting a role for Hfq, MotR, which is present at the same levels in an Hfq WT and in the HfqY25D mutant (Fig S8C), no longer upregulates rpsJ-gfp in the distal face mutant background (Fig S8B). Collectively, our results are consistent with MotR base pairing internal to rpsJ affecting Hfq binding to the S10 leader sequence, which in turn results in increased rpsJ translation.
Based on the RIL-seq data, FliX interacts with multiple regions in the S10 operon mRNA, all internal to CDSs (Fig 5F). The predicted base-pairing regions (Fig 5G) align with the highest peaks of chimeras in the RIL-seq data and overlap with the seed sequence suggested for FliX (Melamed et al., 2016). We tested the effects of FliX on expression from this operon by constructing gfp fusions to regions of rplC, rpsQ, and rpsS-rplV. In all cases, overproduction of FliX or FliX-S led to a reduction in the expression of these fusions (Fig 5H). To test for a direct interaction between FliX and the rpsS mRNA, we again carried out structure probing (Fig 9C and S9D). The regions that were changed in rpsS and FliX in the in vitro footprinting aligned with the predicted binding region between the two RNAs. Introduction of M1 mutation to FliX (Fig S1A) eliminated binding to rpsS mRNA whereas introduction of a complementary mutation to rpsS mRNA restored FliX binding (Fig 5I). We hypothesize that FliX downregulation of the rplC, rpsQ, and rpsS-rplV fusions as well as the fliC mRNA is due to sRNA-directed mRNA degradation. Further experiments are needed to test this model, but in vivo primer extension assays carried out for RNA isolated from in mid-log phase cells (OD600 ∼0.6) showed an increase in 5’ ends in proximity to the binding site on rpsS mRNA in FliX or FliX-S overexpressing strains (Fig S10).
Increased S10 levels correlate with increased readthrough of the flagellar operons
We wondered whether the positive regulation of rpsJ by MotR could impact flagella synthesis. The S10 protein encoded by rpsJ has two roles in the cell. It is incorporated into the 30S ribosome subunit but also forms a transcription anti-termination complex with NusB (Lüttgen et al., 2002, Luo et al., 2008, Baniulyte et al., 2017). We evaluated the importance of each of the two S10 roles to flagella number by electron microscopy. First, we overexpressed a S10 mutant (S10Δloop) that is missing the ribosome binding loop but is still active in anti-termination (Luo et al., 2008) from an inducible plasmid and analyzed the number of flagella per cell. Cells carrying S10Δloop plasmid had higher number of flagella like cells overexpressing MotR* (Fig 6A). We noted that overexpression of wild type S10 from the plasmid used for overexpression of S10Δloop did not lead to an increase in flagella number (Fig 6A), though presumably MotR is normally increasing flagella number by impacting the levels of the WT protein. Possibly, only a specific concentration of S10 relative to other ribosome proteins increases the S10 role as an anti-terminator. Since rpsJ is essential and cannot be deleted, we also examined the effect of MotR* overexpression in a ΔnusB strain that cannot form the S10-NusB anti-termination complex. In this background, the stimulatory effect of MotR* on flagella number was eliminated (Fig 6B) as is also observed for S10Δloop overexpression in the ΔnusB background (Fig S11A). Based on these observations, we hypothesize that increased S10 levels upon MotR overexpression leads to increased anti-termination of some of the long flagella operons.
To directly test this anti-termination hypothesis, we carried out RT-qPCR analysis in WT and ΔnusB backgrounds to examine the effects of MotR and MotR* overexpression on genes in the motAB-cheAW and tar-tap-cheRBYZ operons. For both operons, the mRNA levels of the tested genes were increased in WT upon MotR and MotR* overexpression (Fig 6C and Fig S11B). This increase was not observed for the non-flagellar control gene cadB. While the levels of the flagellar mRNAs in ΔnusB background were lower than in the WT, MotR and MotR* no longer induced these genes. Together these observations are consistent with the proposal that increased levels of non-ribosome associated S10 leads to increased levels of the S10-NusB anti-termination complex associated with RNA polymerase-σ28 and increased antitermination of the long operons encoding flagellar proteins. It is also conceivable that even a slight upregulation of the S10 operon, as well as the S6 operon, given a significant number of MotR-rpsF chimeras (Table S1), along with antitermination of rrn operons, could lead to more active ribosomes, which are needed for flagellar protein synthesis. On the other hand, a negative effect of FliX on ribosomal components, which could reduce the number of active ribosomes, would be consistent with the repressive role of this sRNA.
MotR and FliX have opposing effects on the expression of middle and late flagella genes
In a parallel line of experimentation, we examined the impact of overexpressing MotR* and FliX on the transcriptome by RNA-seq analysis (Table S2). The transcripts whose levels increased most with MotR* overexpression compared to the vector control (Fig 7A) corresponded predominantly to late genes and, to a lesser extent, middle genes, of the flagellar regulon. Of the 332 genes whose expression increased significantly (FDR = 0.05) by MotR* overexpression, 40 are reduced significantly (FDR = 0.05) in a strain lacking σ28 (ΔfliA) (Fitzgerald et al., 2014) (Fig S12A). Additionally, the sequence motif found for the promoters of the transcription units for which expression increased the most (FDR = 0.05 and ≥2 fold) upon MotR* overproduction (Fig S12B) is nearly identical to a σ28 recognition motif (Fitzgerald et al., 2014, Shi et al., 2020). In contrast, transcripts for flagellar genes were reduced by FliX overexpression (Fig 7B). Specifically, 28 of 149 genes for which the expression is reduced significantly (FDR = 0.05) are middle or late genes of the flagellar regulon (Fitzgerald et al., 2014). We note that we did not observe differential levels of the S10 operon transcript in the RNA-seq analysis (Fig 7B and Table S2), however the total RNA for the RNA-seq experiment was isolated from cells earlier in growth (OD600 ∼0.2).
The effects of MotR, MotR*, FliX and FliX-S on flagella gene expression were further examined by monitoring fluorescence from gfp fused to the promoters of flgB, a representative class 2 promoter, and fliL, a representative combined class 2 and 3 promoter (Zaslaver et al., 2006). MotR and MotR* overexpression increased the activity of the two promoters, while FliX and FliX-S overexpression led to a reduction of their activity (Fig 7C, 7D, S13A and S13B). The levels of C-terminally SPA-tagged FlgJ, also encoded by a class 2 gene, similarly increased across growth upon MotR* overexpression, particularly early in growth, and decreased upon FliX-S overexpression (Fig S13C and S13D). The data suggest that in addition to the effect on antitermination (Fig 6), MotR more broadly effects transcription initiation at flagellar genes though we do not know the mechanism. In general, these results are coherent with a positive effect of MotR on flagella synthesis and a negative effect of FliX.
MotR increases and FliX decreases flagella synthesis
To examine the impact of chromosomally encoded MotR and FliX on flagella synthesis and the flagellar regulon, we introduced the three-nucleotide M1 substitutions in the seed sequences of motR and fliX (MotR-M1 and FliX-M1, Fig S1A) at their endogenous chromosomal positions, avoiding the disruption of the nearby genes. MotR-M1 RNA levels were comparable to WT MotR levels (Fig S14A). The prominent ∼200 nt FliX band was reduced for FliX-M1, while other FliX processing products were affected less (Fig S14B).
We first examined the flagella number and motility for these strains. The motR-M1 chromosomal mutation was associated with a moderate reduction in flagella number at two time points (OD600 ∼ 0.6 and 2.0) (Fig 8A), while slightly higher numbers of flagella were observed for the fliX-M1 strain at the later time point (OD600 ∼ 2.0) (Fig 8B). In motility assays carried out as in Fig 2, we found reduced motility of the motR-M1 strain compared to WT but no change was observed for fliX-M1 strain (Fig S14C and S14D). We also compared the motility of the motR-M1 and fliX-M1 strains to WT strains by mixing strains transformed with plasmids expressing either GFP or mCherry. WT strains expressing GFP were mixed with motR-M1 or fliX-M1 cells expressing mCherry or vice versa, and their motility was compared on 0.3% agar. For both combinations of WT and motR-M1, the fluorescent signal produced by the WT strain was more extensive than the fluorescent signal generated by motR-M1 mutant outside of the site of inoculation (Fig 8C). Thus, in two independent assays, the motR-M1 mutant exhibits reduced motility compared to the WT strain, while no significant difference was observed between WT and fliX-M1 (Fig 8D).
We also assessed the effects of the chromosomal mutations on the flgB-gfp and fliL-gfp fusions (Fig 8) as well as FlgJ-SPA levels (Fig S14). The motR-M1 mutant showed reduced activity of the two promoters (Fig 8E), in line with the increased activity of the promoters that was observed upon MotR overexpression (Fig 7C). The fliX-M1 mutant showed similar activity of the two promoters in comparison to WT (Fig 8F). In the western analysis in the motR-M1 strain in comparison to its parental WT, a delayed initiation of FlgJ-SPA synthesis in the mutant was observed (Fig S14E). In contrast, FlgJ-SPA levels in the fliX-M1 strain increased compared to the parental WT strain (Fig S14F).
While negative effects of the motR-M1 mutation on flagella synthesis, motility and flagellar gene expression were easily observed, positive effects of the fliX-M1 mutation were not detected in all assays. However, for both sRNAs the mutation phenotype is opposite the overexpression phenotype as expected. Collectively these observations indicate that MotR, expressed earlier in growth, increases flagella synthesis by positively regulating the middle and the late genes, while FliX, whose levels peak later, decreases flagella synthesis by downregulating the flagellar regulon, mainly the middle and the late genes. Thus, MotR and FliX, along with UhpU, add another layer of regulation to the flagellar regulon (Fig 8G).
In this study, we describe four E. coli sRNAs whose expression is dependent on σ28. We found three of these sRNAs affect flagella number and bacterial motility. Although previous studies showed that base pairing sRNAs act on the flhDC mRNA (Thomason et al., 2012, De Lay & Gottesman, 2012), our results revealed that the effect of sRNAs on flagellar synthesis is far more pervasive. Intriguingly, two of the σ28-dependent sRNAs show opposite effects. MotR, expressed earlier in growth, increases expression of flagellar and ribosomal proteins along with flagella number, while FliX, expressed later in growth, decreases expression of these proteins and flagella number. Thus, the two sRNAs, respectively, might be considered an accelerator and a decelerator for flagellar synthesis.
Non-canonical mechanisms of sRNA action
Most commonly, sRNAs base pair with the 5’ UTRs of mRNA targets or at the very beginning of the CDS, primarily affecting the ribosome binding or mRNA stability. However, MotR and FliX bind in the middle or even close to the ends of their target CDSs in the fliC gene and S10 operon. For both fliC and the S10 operon, the consequences of MotR and FliX overexpression are different. MotR leads to higher levels of fliC and the S10 protein, whereas FliX leads to lower levels of fliC and three genes in the S10 operon. We suggest that the positive and negative regulatory effects of MotR and FliX, respectively, occur by the same mechanisms on the fliC and S10 transcripts, with MotR changing the conformation of the RNAs and FliX leading to increased cleavage. However, these suggested mechanisms needed to be investigated further in future experiments. It is also noteworthy that, based on RIL-seq data, more examples of this type of interaction remain to be characterized.
Given that our study made extensive use of RIL-seq data, it provides an opportunity to evaluate these data. While RIL-seq provides a comprehensive map of RNA-RNA interactions that take place on Hfq under a specific condition, some caution about the interpretation is warranted as the interactions represent multiple types of relationships between two RNAs. As was found by a recent study (Faigenbaum-Romm et al., 2020), we suggest that if an interaction is highly abundant and discovered under multiple conditions, the sRNA is more likely to have a regulatory impact on the target mRNA though the mechanisms may be unknown. We noticed that the spread of the RIL-seq signal varies significantly between targets. One possible explanation for multiple peaks and a broad distribution is more than one base pairing site for the sRNA on the mRNA, but this hypothesis requires further investigation. We predict additional studies of sRNA-target pairs with different types of RIL-seq signals will give further insights into the steps and outcomes of base pairing.
The most studied and conserved sRNA-binding protein in gram-negative bacteria is Hfq. However, there are other sRNA-binding proteins (reviewed in (Melamed, 2020)). Among these is ProQ, which was shown to have overlapping, complementary, and competing roles with Hfq in E. coli (Melamed et al., 2020). Interestingly, ProQ was found to affect motility and chemotaxis in S. enterica (Westermann et al., 2019). In the absence of ProQ, the target sets for the σ28-dependent sRNAs on Hfq were changed significantly in E. coli (Table S5 in (Melamed et al., 2020)) suggesting that competition between Hfq and ProQ for binding RNAs likely also influences this regulatory circuit. In this context, it is worth noting that FlgO, the fourth σ28-dependent sRNA, which originates from the 3’UTR of the flgL and strongly binds Hfq, does not have many targets. Possibly, FlgO has a role in titrating Hfq from other sRNAs or proteins, or in the recruitment of other proteins to a complex with Hfq. Interestingly, while the overall sequence of flgO is conserved in other bacterial species (Fig S15), the nucleotides in one of the single stranded loops (Fig S1) differ in S. typhimurium, possibly suggesting unique regulatory mechanisms in different bacteria.
Conservation of σ28-dependent sRNAs
We were surprised to find so many σ28-dependent Hfq-binding sRNAs and wondered about their phylogenic distribution. A previous study assessing the conservation of the motR and uhpU promoters showed that, while the motR promoter is well conserved across proteobacteria species, the uhpU promoter was not, implying different evolutionary pressures (Fitzgerald et al., 2018). Interestingly, however, a sRNA named RsaG, which originates from the 3’ UTR of uhpT and also is induced by glucose-6-phosphate, was found in the Gram-positive bacterium, Staphylococcus aureus (Bronesky et al., 2019). Although there is no sequence similarity between UhpU and RsaG, and RsaG has not been reported to regulate flagella synthesis, the independent evolution of regulatory sRNAs at the 3’ UTRs of uhpT in two disparate bacterial species is intriguing.
The σ28-dependent sRNAs themselves are conserved among some of the Enterobacteriaceae (Fig S15) and thus may play a role in pathogenicity. During the preparation of this manuscript, two studies describing the application of RIL-seq to S. enterica and Enteropathogenic E. coli Hfq were published (Mizrahi et al., 2021, Matera et al., 2022). The RIL-seq analyses were carried out for cells grown under conditions that do not favor flagellar gene expression, but UhpU, MotR, FliX and FlgO were detected, confirming their synthesis and association with Hfq in pathogenic bacteria. It is likely that still other sRNA regulators of the flagellar regulon remain to be characterized. In S. enterica, a leader RNA originating from the mgtCBR virulence operon was shown to affect the synthesis of one of the two flagellin genes that exist in this bacterium, impacting virulence and motility (Choi et al., 2017). In neonatal meningitis-causing E. coli, a sRNA that is missing from the E. coli MG1655 strain used in our study was shown to reduce fliC mRNA levels (Sun et al., 2022).
Roles of σ28-dependent sRNAs
The UhpU RIL-seq target set includes many flagellar regulon genes and some transcription regulators of the flagellar regulon, such as LrhA (Lehnen et al., 2002), hinting at a mechanism by which UhpU can affect flagella number and bacterial motility. However, since UhpU can also be derived from the uhpT mRNA (Fig S1B) and is predicted to have many targets that participate in carbon and nutrient metabolism (Table S1), we suggest this sRNA may play a broader role in linking carbon metabolism with flagella synthesis and motility.
MotR and FliX each have more limited target sets in the RIL-seq data but may comprise a unique regulatory toggle. The transcription of the two sRNA is dependent on the same sigma factor, and they base pair in the CDS of targets in the same operons, but base pairing results in opposing regulation. MotR, which is transcribed from within flhC at the top of the flagellar regulon, reaches its highest levels earlier than FliX and increases the flagella synthesis, while FliX, which is cleaved from the mRNA required to make the last protein needed to complete the flagellum, reaches its highest levels later in growth and appears to decrease flagella synthesis.
It is not yet clear how MotR and FliX base pairing with only a few targets can have such pervasive effects on flagellar gene expression and flagella number, but we suggest multiple mechanisms may be involved. One possibility is that the levels of flagellin encoded by fliC, up and down regulated by MotR and FliX, respectively, could be part of an autoregulatory loop that impacts the transcription of flhDC or other middle or late flagellar gene promoters. The increased and decreased levels of ribosomal proteins brought about by MotR and FliX regulation of the S10 operon also could impact the levels of available ribosomes, where even slight changes could have consequences given the high ribosome cost of flagella synthesis. Finally, we hypothesize that elevated levels of the S10 protein, due to the regulation by MotR, could, in conjunction with NusB, lead to increased antitermination of long flagellar operons.
Based on our hypothesis that the MotR-mediated increase in S10 levels leads to increased antitermination, we speculate that MotR activation of S10 expression could serve an autoregulatory role. Early in growth, transcription initiating from the σ28-dependent promoter in flhC terminates at the 5’ of motA generating MotR. As MotR levels increase, there is a concomitant increase in S10 levels, which could promote readthrough of the motR terminator leading to decreased MotR levels and increased full-length motRAB-cheAW mRNA.
In general, the σ28-dependent sRNAs add a new layer of regulation to the flagellar regulon and reinforce the fact that flagella synthesis is exquisitely regulated. The regulon will continue to serve as a model of a temporal and environmentally-controlled regulatory network with contributions from both transcription factors and regulatory RNAs.
Materials and Methods
Bacterial strains and growth conditions
E. coli MG1655 (GSO982 or GSO983) or MC4100 (GSO614) strains served as the WT strains in this study. All other bacterial strains studied here are listed in the Table S3 along with plasmids and oligonucleotides used . E. coli K-12 MG1655 genome was used as template to amplify mRNAs and sRNAs to be cloned into their respective sites. Unless indicated otherwise, all strains were grown with shaking at 250 rpm at 37°C in LB rich medium. Ampicillin (100 µg/ml), chloramphenicol (25 µg/ml), kanamycin (30 µg/ml), arabinose (0.2%) and IPTG (1 mM) were added where appropriate. Unless indicated otherwise, overnight cultures were diluted to an OD600 = 0.05 and grown for the indicated times or to the desired optical densities.
fliA::kan, fliCX::kan, and fliX-M1:kan strains were constructed by amplifying the kanR sequence from pKD4 (Datsenko & Wanner, 2000) using oligonucleotides listed in Table S3 and recombining (Datsenko & Wanner, 2000) the product into the chromosome of strain NM400 (kind gift of Nadim Majdalani). flgJ was SPA-tagged by amplifying the SPA sequence adjacent to kanR sequence from pJL148 (Zeghouf et al., 2004) using oligonucleotides listed in Table S3 and recombining (Datsenko & Wanner, 2000) the product into the chromosome of strain NM400. motR-M1 strain was constructed using the scar-free system, FRUIT (Stringer et al., 2012) as previously described. Briefly, thyA was deleted from MG1655 (crl-) (GSO983) strain by PCR amplification of ΔthyA from AMD061 (Stringer et al., 2012) followed by recombination using pKD46 (Datsenko & Wanner, 2000). Next, thyA was inserted back to the genome next to the site of mutation and selection was made by growth on minimal media lacking thymine. The motR-M1 mutation was introduced while simultaneously removing thyA. The selection for colonies missing thyA was carried out using minimal medium M9 plates supplied with 0.4% glucose, 0.2% casamino acids, 20 µg/ml trimethoprim, and 100 µg/ml thymine. flgM::kan, lrhA::kan, and nusB::kan deletion strains were obtained from other groups (Baba et al., 2006) as referenced in Table S3. All deletions and mutations were confirmed by sequencing and then transferred to new backgrounds by P1 transduction. Where indicted, kanR was removed from the chromosome using plasmid pCP20 (Cherepanov & Wackernagel, 1995). Construction of strains carrying chromosomal lacZ fusions was carried out using PM1205 as previously described (Mandin & Gottesman, 2009). In brief, the lrhA fragment was amplified using KAPA Hifi (Fisher Scientific) using oligonucleotides SM079 and SM080 (Table S3) and transformed into PM1205 with a series of selective screens on minimal media plates supplemented with sucrose, LB, LB supplemented with chloramphenicol, and LB supplemented with tetracycline. Mutagenesis of lrhA-lacZ fusion was achieved by recombineering an lrhA-M1 sequence instead of the WT lrhA sequence, using gBlock listed in Table S3.
Descriptions of plasmids used in this study are in Table S3. Construction of the constitutive overexpression plasmids was done according to (Urban and Vogel, 2009) using pZE12-luc. The IPTG-inducible UhpU overexpression plasmid was constructed using a pBRplac derivative harboring kanR, pMSG14 (Guo et al., 2014). The uhpU sequence, starting from its second nt, was amplified by PCR using oligonucleotides TU558 and TU561 (Table S3), digested with AatII and HindIII and cloned into pMSG14 digested with the same restriction enzymes. 3xFLAG-rpsJ was expressed from pBAD33 (Guzman et al., 1995). The S10 leader and rpsJ sequence along with the 3xFLAG sequence was PCR amplified using oligonucleotides SM533 and SM435, digested with KpnI and HindIII and cloned into pBAD33 digested with the same restriction enzymes.
Construction of GFP-fusion plasmids was carried out principally as described in (Urban and Vogel, 2009), using the pXG10-SF or pXG30-SF (Corcoran et al., 2012). Briefly, regions of target genes, mainly regions captured in the chimeric fragments, were PCR amplified, digested with Mph1103I and NheI and cloned into pXG10-SF or pXG30-SF digested with the same restriction enzymes. The full list of oligonucleotides used in this study can be found in Table S3. Mutagenesis of the different plasmids was achieved using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent). All plasmids were freshly transformed into the appropriate strains before each of the experiments.
Cells corresponding to the equivalent of 10-20 OD600 were collected, washed once with 1X PBS, and frozen in liquid nitrogen. RNA was extracted according to the standard TRIzol protocol (Thermo Fisher Scientific) as described previously (Melamed et al., 2020). At the last step, RNA was resuspended in 20-50 µl of DEPC water and quantified using a NanoDrop (Thermo Fisher Scientific).
RNA coimmunoprecipitation (Co-IP) assay
RNAs that co-IP using polyclonal antibodies to Hfq were isolated as described (Zhang et al., 2002) with the following modifications. MG1655 (GSO983) was grown to OD600 ∼0.6 and ∼1.0 in LB medium. Cells corresponding to the equivalent of 20 OD600 were collected, and cell lysates were prepared by vortexing with 212-300 µm glass beads (Sigma-Aldrich) in a final volume of 1 ml of lysis buffer (20 mM Tris-HCl/pH 8.0, 150 mM KCl, 1 mM MgCl2, 1 mM DTT). Co-IPs were carried out using 100 ml of α-Hfq, 120 mg of protein A-Sepharose beads (GE Healthcare), and 950 µl of cell lysate. Co-IP RNA was isolated from protein A-Sepharose beads by extraction with phenol: chloroform:isoamyl alcohol (25:24:1), followed by ethanol precipitation. Total RNA was isolated from 50 ml of cell lysate by TRIzol (Thermo Fisher Scientific) extraction followed by chloroform extraction and isopropanol precipitation. Total and co-IP RNA samples were resuspended in 15 µl of DEPC H2O, and 5 µg total RNA and 0.5 µg co-IP RNA were subjected to northern analysis as described below.
Northern blot analysis
For smaller RNAs, total RNA (5 μg) was separated on a denaturing 8% polyacrylamide urea gel containing 6 M urea (1:4 mix of Ureagel Complete to Ureagel-8 (National Diagnostics) with 0.08% ammonium persulfate) in 1X TBE buffer at 300V for 90 min. The RNA was transferred to a Zeta-Probe GT membrane (Bio-Rad) at 20 V for 16 h in 0.5X TBE. For longer RNAs, total RNA (10 μg) were fractionated on formaldehyde-MOPS agarose gels as previously described (Adams et al., 2017). Briefly, RNA was denatured in 3.7% formaldehyde (Fisher), 1X MOPS (20 mM MOPS, 5 mM NaOAc, 1 mM EDTA, pH 7.0) and 1X RNA loading dye (Thermo Fisher Scientific) for 10 min at 70°C and incubated on ice. The RNA was loaded onto a 2% NuSieve 3:1 agarose (Lonza), 1X MOPS, 2% formaldehyde gel and separated at 125-150V at 4°C for 1-2 h and then transferred to a Zeta-Probe GT membrane (Bio-Rad) via capillary action overnight (Streit et al., 2009). For both types of blots, the RNA was crosslinked to the membranes by UV irradiation. RiboRuler High Range and Low Range RNA ladders (Thermo Fisher Scientific) were marked by UV-shadowing. Oligonucleotide probes (listed in Table S3) for the different RNAs were labelled with 0.3 mCi of [γ-32P] ATP (Perkin Elmer) by incubating with 10 U of T4 polynucleotide kinase (New England Biolabs) at 37°C for 1 h.
Primer extension assay
Primer extension analysis was performed using an oligonucleotide (listed in Table S3) specific to the rpsS as described (Zhang et al., 1998). RNA samples (5 µg of total RNA) were incubated with 2 pmol of 5-32P-end-labeled oligonucleotide 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. The reactions were terminated by adding 10 μl of Stop Loading Buffer. The cDNA products then were fractionated on 8% polyacrylamide urea gels containing 8 M urea in 1X TBE buffer at 70 W for 70 min.
Total RNA was isolated from cultures grown to OD600∼0.2 and RNA concentrations were determined using a NanoDrop. Samples were treated with DNase using TURBO DNA-free™ Kit (Thermo Fisher Scientific). DNA-free RNA was used for cDNA synthesis using iScript cDNA Synthesis Kit (Bio-Rad) and cDNA concentrations were measured by Qubit fluorimeter (Invitrogen). Equal amounts of cDNA were loaded into 96-well plate and cDNA was quantified by CFX Connect Real-Time system (Bio-Rad) using iTaq Univer SYBR Green mix (Bio-Rad) according to manufacturer instructions. Specific oligonucleotide primers were designed for each gene and the expression was normalized using ssrA levels. Serial dilutions of E. coli genomic DNA in known concentrations were used to generate a standard curve. Starting quantity of cDNA samples were determined based on the standard curve and normalization was done using the starting quantities of ssrA. Cq was measured in duplicate or triplicate for each biological sample. CFX maestro analysis software was also used for conducting the analysis.
RNA structure probing
GeneBlock fragments carrying the motR, fliX, rpsJ or rpsS CDS (IDT) were used as DNA templates for in vitro transcription with MEGAshortscript T7 High Yield Transcription Kit (Invitrogen). The transcripts were dephosphorylated with calf intestinal alkaline phosphatase (CIP, New England Biolabs) and then radioactively labeled at 5’ end with [γ-32P] ATP (Perkin Elmer) and T4 kinase (Invitrogen), and purified on an 8% polyacrylamide/7M urea gel and eluted in buffer containing 20 mM Tris-HCl/pH 7.5, 0.5 M NaOAc, 10 mM EDTA and 1% SDS at 4°C for overnight, followed by ethanol precipitation. The RNA concentration was determined by measuring the OD260 on Nanodrop.
For all the structural probing assays, 0.2 pmole of the labeled transcript, 2 pmole of unlabeled transcript and 1 µg of yeast RNA with or without 2 pmole (hexameric concentration) of purified Hfq were mixed in 10 µl of 1x Structural Buffer in Ambion RNase T1 Kit (Invitrogen). The reactions were incubated at 37°C for 10 min, followed by treatment at 37°C with 0.02 U RNase T1 for 10 min, 1.3 U RNase III for 1.5 min, or 50 µmole lead acetate for 10 min, whereupon 20 µl Inactivation Buffer and 1 µl Glycoblue were added. The RNAs were precipitated and resuspended in 10 µl Loading Buffer, and analyzed on a 8% polyacrylamide/7 M urea gel run in 1x TBE. RNase T1 and alkali digestion ladders of the end-labeled transcripts were used as molecular size markers.
Translational reporter assay
The GFP reporter assays were carried out essentially as described (Melamed et al., 2016). Overnight cultures were grown in 2 ml of LB media supplemented with the appropriate antibiotics at 37°C with constant shaking at 250 rpm. Cells were then diluted to OD600∼0.05 in 1 ml of fresh LB medium supplemented with the appropriate antibiotics in 96-well plate and grown at 37°C with constant shaking at 250 rpm for 3 h. Cells were pelleted and resuspended in filtered 1 X PBS. Fluorescence was measured using the BD LSRFortessa or Beckman Coulter Cytoflex flow cytometer. The level of regulation was calculated by subtracting the auto-fluorescence and then calculating the ratio between the fluorescence signal of a strain carrying the sRNA over-expressing plasmid and the signal of a strain carrying the control plasmid. Three biological repeats were prepared for every sample.
The β-galactosidase assays were carried out as described (Miller, 1992). Overnight cultures grown as for the GFP reporter assays were diluted 1:100 into 5 ml of fresh LB with antibiotic and 0.2% arabinose and grown at 37°C with constant shaking at 250 rpm until OD600 ∼0.7. IPTG (1 mM) was added to cells harboring inducible sRNAs plasmids. After β-galactosidase activity was measured, the Miller units were calculated from the following formula:
Transcriptional reporter assays
Overnight cultures harboring flgB-gfp and fliL-gfp fusions (Zaslaver et al., 2006) were grown as described for the translation reporter assays and then diluted to OD600∼0.05 in 150 µl of fresh LB medium supplemented with the appropriate antibiotics in a transparent bottom 96-well plate. Bacterial growth and promoter activity were monitored for 330 min at 37°C using OD600 and GFP fluorescent measurements, respectively, using a Synergy H1 plate reader (Agilent).
Bacteria were grown to the desired OD600, and the cells in 0.5 ml – 4 ml of culture were collected. Cell lysates were prepared by resuspending cell pellets with SDS-loading buffer normalized to the cell density, and samples were then heated for 10 min at 95°C. Protein samples were subjected to a 4%-15% polyacrylamide SDS gel electrophoresis followed by electrotransfer to a nitrocellulose membrane (Fisher Scientific). The membrane was blocked with 3% milk, probed with anti-flagellin antibodies (1/10,000) (Abcam) and then with anti-rabbit secondary antibody (1/10,000) or with ANTI-FLAG M2-Peroxidase (HRP) (1/1,000), (Sigma-Aldrich). Signals were visualized by the ECL system (Biorad).
WT (GSO983) cells harboring pBR*, pBR*-UhpU, pZE, pZE-MotR, pZE-MotR*, pZE-FliX or pZE-FliX-S were grown with shaking at 180 rpm in 5 ml of LB at 37°C to OD600 ∼ 1.0. Cell pellets collected by centrifugation were suspended in 5 ml of PBS and then heated at 65°C for 5 min, followed by centrifugation to obtain the cell pellets and supernatants, which contained the cytoplasmic flagellin molecules and depolymerized flagellin monomers, respectively. The cell pellets were resuspended in the SDS-loading buffer, normalized to the cell density. Proteins in the supernatants were precipitated by 10% trichloroacetic acid, resuspended in Tris/SDS loading buffer and heated at 95°C for 10 min.
Overnight cultures were diluted in fresh medium and grown with shaking at 180 rpm, at 37°C to mid-log phase (OD600∼0.6-0.8) unless indicated otherwise. Cells were collected by centrifugation at 1,000 rpm for 20 min, and pellet was resuspended in 300 µl of saline. Next, 3 µl of bacterial suspension were placed on a freshly glow-discharged carbon covered electron microspical (EM) support grid (EMS, Hatfield, PA) for 5 min. The grid was washed twice with distilled water and stained for 1 min with 0.75% aqueous solution of uranyl formate, pH 4.5. The grids were imaged in Thermo Fisher Scientific (Hillsboro, OR) FEI Tecnai 20 electron microscope operated at 120 kV. The images were recorded using AMT (Woburn, MA) XR81 CCD camera. Flagella were counted for 20-40 cells in each sample as indicated in the Fig legends. Each analysis was repeated a minimum of three times.
Overnight cultures (∼1 µl) were spotted onto 0.3% soft agar plates by touching the agar softly with the tip and ejecting the culture. Plates were incubated right-side up at 30°C above a beaker filled with water for 9-24 h. Plates were made with the appropriate antibiotics and with 1 mM IPTG. The plates were imaged using Bio-Rad imager (using Colorimetric settings) and the diameter of the bacterial culture was calculated using the ImageJ software. Two technical repeats and three biological repeats were carried out for each strain. For motility competition assays cells were first transformed with pCON1.proC-GFP or pCON1.proC-mCherry plasmids (Cooper et al., 2017), resulting in a GFP or an mCherry signal, respectively. In each case, equal numbers of bacterial cells based on OD600 of each overnight culture for one strain expressing a green fluorescence signal and a second strain expressing a red fluorescent signal were mixed before spotting them onto 0.3% soft agar plate and the plates were incubated as described above. Images were taken using Bio-Rad imager with the following settings: Colorimetric (1-2 sec) for bright field, Cy2 for GFP (auto optimal exposure), Cy3 for mCherry (auto optimal exposure). Images were merged using Image Lab (Bio-Rad).
Overnight cultures were diluted in fresh LB medium and grown to early-log phase (OD600∼0.2). RNA was extracted using the standard TRIzol protocol (Thermo Fisher Scientific) as described above. Total RNA libraries were constructed using the RNAtag-Seq protocol with a few modifications to allow capture of short RNA fragments as previously described (Melamed et al., 2018). The libraries were sequenced by paired-end sequencing using the HiSeq 2500 system (Illumina) at Molecular Genomics Core, Eunice Kennedy Shriver National Institute of Child Health and Human Development. RNA-seq data processing follows the same procedures as RIL-seq data analysis for QC analysis, adaptor removal, and alignment with the Python RILSeq package (Melamed et al., 2018). The raw fastq records were demultiplexed with python script index_splitter.py (https://github.com/asafpr/RNAseq_scripts/blob/master/index_splitter.py) followed by adapter removal with cutadpt software (version 3.4). The trimmed fastq reads were mapped to E. Coli genome (ecoli-k12-MG1655-NC_000913-3) with Python RILSeq package (version 0.74, https://github.com/asafpr/RILseq). Deeptools software (version 3.5.1) was used to generate bigwig file for coverage visualization. Read counts were obtained with featureCounts tool of Subread software (version 2.0.3) and a customized annotation file based on EcoCyc version 20.0 (Keseler et al., 2013) with manual addition of sRNAs and small proteins from (Hör et al., 2020). Differential expression analyses were conducted with R DESeq2 package (Love et al., 2014) and default normalization. Deferentially expressed genes were extracted with the parameter of ‘independentFiltering=FALSE’. The sequencing data reported in this paper have been deposited in GEO under accession number GSE1774487.
Determination of sequence motifs and base-pairing predictions
Common binding motifs were searched with MEME software (Bailey et al., 2009). Genes that were induced the most by MotR* overexpression in RNA-seq (Table S2) (FDR = 0.05 and ≥2 fold) were extracted from the data and divided to transcription units based on EcoCyc version 20.0 (Keseler et al., 2013). For each transcription unit, genomic sequence was extracted using coordinates for the start codon of the first gene in the transcription unit and 250 nt upstream to the gene. For sRNAs, genomic sequence was extracted using coordinates for the transcription start site and 250 nt upstream to the gene. For outputs, motif length was restricted to 28.
Functional annotation analysis
Functional annotation analysis of sRNAs targets was carried out using the Database for Annotation, Visualization and Integrated Discovery (DAVID) (Huang da et al., 2009). Gene names served as the input list in each case. Targets that were present in at least three RIL-seq conditions in Table S1 were included in the analysis.
Circos plots follow the procedures of R RCircos Package (Zhang et al., 2013b). Link lines are used to label the statistically significant chimeric fragments (S-chimeras as defined in (Melamed et al., 2016)). RIL-seq from six different growth conditions was analyzed and S-chimeras present in at least four of the six conditions are included in the plots.
Data from RIL-seq experiment 1 from Melamed at al., 2020 extracted for unified S-chimeras files for the different sRNAs were mapped based on the first nt of each read in the chimera. BED files are generated with Python RILSeq package (Melamed et al., 2018) and viewed using the UCSC genome browser (Kent et al., 2002). For previously annotated RNA in GTF file, BED files are directly generated with command of generate_BED_file_of_endpoints.py and EcoCyc ID. For genes annotated in the current study, significant chimeras which involve the relevant gene are first extracted from significant interaction file, then chimeric reads involving the S-chimeras are extracted from chimeric read file. To be a qualified chimeric read, RNA1 start position of the read must overlap with the genomic range of RNA1 in S-chimera and RNA2 start position of the read must overlap with the genomic range of RNA2 in S-chimera. Finally, the read list for genes annotated in the current study is supplied to generate_BED_file_of_endpoints.py command to generate BED file.
We thank M. Gottesman for plasmids expressing wild type and rpsJ mutants, O. Steele-Mortimer for plasmids constitutively expressing GFP or mCherry, and D. Court for the S10 antibody. We thank J. Wade for sharing the sequences used to generate the σ28 binding motif and J. Wade and G. Baniulyte for advice on the FRUIT method. We thank the NICHD Molecular Genomics Core, particularly Tianwei Li, for all the library sequencing. We also appreciate the help of A. Peer with the sRNA conservation analysis. We are grateful to the Storz and S. Gottesman labs for all of the helpful discussions and thank the Storz lab, S. Gottesman and J. Wade for their comments on the manuscript. This work was supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant number ZIA HD001608-31) and by the Israel Science Foundation (Grants 826/22 and 2859/22).
S.M. and G.S. conceived of the project. S.M., A.Z., J.M., and A.S. designed, performed, and analyzed the experiments. M.J. performed all electron microscopy. H.Z. performed all the computational analyses under the supervision of S.M. S.M., A.Z. and G.S. prepared the figures and wrote the manuscript. G.S. supervised the project.
Conflict of interest
The authors declare no competing interests.
- 1.In vivo expression technology and 5’ end mapping of the Borrelia burgdorferi transcriptome identify novel RNAs expressed during mammalian infectionNucleic Acids Res 45:775–792
- 2.Prevalence of small base-pairing RNAs derived from diverse genomic lociBiochim. Biophys. Acta Gene Regul. Mech 1863
- 3.The effect of environmental conditions on the motility of Escherichia coliJ. Gen. Microbiol 46:175–184
- 4.Undiscovered regions on the molecular landscape of flagellar assemblyCurr. Opin. Microbiol 28:98–105
- 5.Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collectionMol. Syst. Biol 2
- 6.MEME SUITE: tools for motif discovery and searchingNucleic Acids Res 37:W202–W208
- 7.Identification of novel sRNAs involved in biofilm formation, motility, and fimbriae formation in Escherichia coliSci. Rep 5
- 8.Identification of regulatory targets for the bacterial Nus factor complexNat. Commun 8
- 9.Prediction of novel bacterial small RNAs from RIL-Seq RNA-RNA interaction dataFront. Microbiol 12
- 10.A multifaceted small RNA modulates gene expression upon glucose limitation in Staphylococcus aureusEMBO J 38
- 11.Gene disruption in Escherichia coli: TcR and KmR cassettes with the option of Flp-catalyzed excision of the antibiotic-resistance determinantGene 158:9–14
- 12.Coordinating assembly of a bacterial macromolecular machineNat. Rev. Microbiol 6:455–465
- 13.A trans-acting leader RNA from a Salmonella virulence geneProc. Natl. Acad. Sci. USA 114:10232–11023
- 14.Predictable, tunable protein production in Salmonella for studying host-pathogen interactionsFront. Cell Infect. Microbiol 7
- 15.Superfolder GFP reporters validate diverse new mRNA targets of the classic porin regulator, MicF RNAMol. Microbiol 84:428–445
- 16.One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR productsProc. Natl. Acad. Sci. USA 97:6640–6645
- 17.A complex network of small non-coding RNAs regulate motility in Escherichia coliMol. Microbiol 86:524–538
- 18.Strategies to block bacterial pathogenesis by interference with motility and chemotaxisCurr. Top. Microbiol. Immunol 398:185–205
- 19.Hierarchy in Hfq chaperon occupancy of small RNA targets plays a major role in their regulationCell Rep 30:3127–3138
- 20.Comprehensive mapping of the Escherichia coli flagellar regulatory networkPLoS Genet 10
- 21.The evolutionary impact of intragenic FliA promoters in proteobacteriaMol. Microbiol 108:361–378
- 22.Autogenous control of the S10 ribosomal protein operon of Escherichia coli: genetic dissection of transcriptional and posttranscriptional regulationProc. Natl. Acad. Sci. USA 84:6516–6520
- 23.MicL, a new σE-dependent sRNA, combats envelope stress by repressing synthesis of Lpp, the major outer membrane lipoproteinGenes Dev 28:1620–1634
- 24.Regulation of flagellar motility during biofilm formationFEMS Microbiol. Rev 37:849–871
- 25.Tight regulation, modulation, and high-level expression by vectors containing the arabinose PBAD promoterJ. Bacteriol 177:4121–4130
- 26.RNA-binding proteins in bacteriaNat. Rev. Microbiol 16:601–615
- 27.Trans-acting small RNAs and their effects on gene expression in Escherichia coli and Salmonella entericaEcoSal Plus 9https://doi.org/10.1128/ecosalplus.ESP-0030-2019
- 28.Systematic and integrative analysis of large gene lists using DAVID bioinformatics resourcesNat. Protoc 4:44–57
- 29.Localization and stoichiometry of hook-associated proteins within Salmonella typhimurium flagellaJ. Bacteriol 169:1168–1173
- 30.A coherent feed-forward loop with a SUM input function prolongs flagella expression in Escherichia coliMol Syst Biol 1
- 31.Ordering genes in a flagella pathway by analysis of expression kinetics from living bacteriaScience 292:2080–2083
- 32.The human genome browser at UCSCGenome Res 12:996–1006
- 33.TargetRNA2: identifying targets of small regulatory RNAs in bacteriaNucleic Acids Res 42:W124–W129
- 34.EcoCyc: fusing model organism databases with systems biologyNucleic Acids Res 41:D605–D612
- 35.The global regulatory hns gene negatively affects adhesion to solid surfaces by anaerobically grown Escherichia coli by modulating expression of flagellar genes and lipopolysaccharide productionJ. Bacteriol 184:1522–1529
- 36.LrhA as a new transcriptional key regulator of flagella, motility and chemotaxis genes in Escherichia coliMol. Microbiol 45:521–532
- 37.Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2Genome Biol 15
- 38.Structural and functional analysis of the E. coli NusB-S10 transcription antitermination complexMol. Cell 32:791–802
- 39.Transcriptional regulation by antitermination. Interaction of RNA with NusB protein and NusB/NusE protein complex of Escherichia coliJ. Mol. Biol 316:875–885
- 40.A genetic approach for finding small RNAs regulators of genes of interest identifies RybC as regulating the DpiA/DpiB two-component systemMol. Microbiol 72:551–565
- 41.IntaRNA 2.0: enhanced and customizable prediction of RNA-RNA interactionsNucleic Acids Res 45:W435–W439
- 42.A major superfamily of transmembrane facilitators that catalyse uniport, symport and antiportTrends Biochem. Sci 18:13–20
- 43.Global RNA interactome of Salmonella discovers a 5’ UTR sponge for the MicF small RNA that connects membrane permeability to transport capacityMol. Cell 82:629–644
- 44.Escherichia coli swimming is robust against variations in flagellar numbereLife 3
- 45.New sequencing methodologies reveal interplay between multiple RNA-binding proteins and their RNAsCurr. Genet 66:713–717
- 46.RNA-RNA interactomes of ProQ and Hfq reveal overlapping and competing rolesMol. Cell 77:411–425
- 47.Mapping the small RNA interactome in bacteria using RIL-seqNat. Protoc 13:1–33
- 48.Global mapping of small RNA-target interactions in bacteriaMol. Cell 63:884–897
- 49.A short course in bacterial genetics : a laboratory manual and handbook for Escherichia coli and related bacteria
- 50.The impact of Hfq-mediated sRNA-mRNA interactome on the virulence of enteropathogenic Escherichia coliSci. Adv 7
- 51.Flagella-driven motility of bacteriaBiomolecules 9
- 52.Regulation of flagellar gene expression in bacteriaBiochemistry (Mosc 80:1447–1456
- 53.Inverse regulatory coordination of motility and curli-mediated adhesion in Escherichia coliGenes Dev 22:2434–2446
- 54.Involvement of two-component signaling on bacterial motility and biofilm developmentJ. Bacteriol 199:e00259–217
- 55.Mechanism of adverse conditions causing lack of flagella in Escherichia coliJ. Bacteriol 175:2236–2240
- 56.Structural basis of bacterial σ28-mediated transcription reveals roles of the RNA polymerase zinc-binding domainEMBO J 39
- 57.Deciphering bacterial flagellar gene regulatory networks in the genomic eraAdv. Appl. Microbiol 67:257–295
- 58.Regulation cascade of flagellar expression in Gram-negative bacteriaFEMS Microbiol. Rev 27:505–523
- 59.Northern blot analysis for detection and quantification of RNA in pancreatic cancer cells and tissuesNat. Protoc 4:37–43
- 60.FRUIT, a scar-free system for targeted chromosomal mutagenesis, epitope tagging, and promoter replacement in Escherichia coli and Salmonella entericaPLoS One 7
- 61.Bacteria reduce flagellin synthesis to evade microglia-astrocyte-driven immunity in the brainCell Rep 40
- 62.Global transcriptional start site mapping using differential RNA sequencing reveals novel antisense RNAs in Escherichia coliJ. Bacteriol 197:18–28
- 63.A small RNA that regulates motility and biofilm formation in response to changes in nutrient availability in Escherichia coliMol. Microbiol 84:17–35
- 64.Bacterial flagellins: does size matter?Trends Microbiol 26:575–581
- 65.Stationary phase reorganisation of the Escherichia coli transcription machinery by Crl protein, a fine-tuner of sigmas activity and levelsEMBO J 26:1569–1578
- 66.Hfq: the flexible RNA matchmakerCurr. Opin. Microbiol 30:133–138
- 67.Translational control and target recognition by Escherichia coli small RNAs in vivoNucleic Acids Res 35:1018–1037
- 68.A green fluorescent protein (GFP)-based plasmid system to study post-transcriptional control of gene expression in vivoMethods Mol. Biol 540:301–319
- 69.Glucose-6-phosphate-dependent phosphoryl flow through the Uhp two-component regulatory systemMicrobiology 147:3345–3352
- 70.The major RNA-binding protein ProQ impacts virulence gene expression in Salmonella enterica serovar TyphimuriummBio 10:e02504–2518
- 71.A comprehensive library of fluorescent transcriptional reporters for Escherichia coliNat. Methods 3:623–628
- 72.Sequential Peptide Affinity (SPA) system for the identification of mammalian and bacterial protein complexesJ. Proteome Res 3:463–468
- 73.Ribosomal protein L4 and transcription factor NusA have separable roles in mediating terminating of transcription within the leader of the S10 operon of Escherichia coliGenes Dev 6:2655–2662
- 74.A hairpin structure upstream of the terminator hairpin required for ribosomal protein L4-mediated attenuation control of the S10 operon of Escherichia coliJ. Bacteriol 178:2383–2387
- 75.Surprising flexibility of leader RNA determinants for r-protein L4-mediated transcription termination in the Escherichia coil S10 operonRNA Biol 8:572–578
- 76.The OxyS regulatory RNA represses rpoS translation and binds the Hfq (HF-I) proteinEMBO J 17:6061–6068
- 77.Mutations in interaction surfaces differentially impact E. coli Hfq association with small RNAs and their mRNA targetsJ. Mol. Biol 425:3678–3697
- 78.The Sm-like Hfq protein increases OxyS RNA interaction with target mRNAsMol. Cell 9:11–22
- 79.RCircos: an R package for Circos 2D track plotsBMC Bioinformatics 14