# Chapter 80. Bacterial Small RNAs, Hfq, ProQ, Csr/Rsm Systems, and Regulatory Networks

## Scope Note

This chapter explains how bacterial small RNAs and RNA-binding proteins regulate translation, RNA stability, mRNA decay, metabolism, stress responses, motility, biofilms, and virulence. The focus is on trans-encoded and cis-antisense small RNAs, Hfq-mediated base-pairing regulation, ProQ and other FinO-domain RNA chaperones, Csr/Rsm protein sequestration by regulatory RNAs, and the experimental logic used to discover small RNA targets. [Chapter 78](chapter1073.md) covers cis-regulatory leader RNAs and thermosensors, [Chapter 79](chapter1074.md) covers riboswitches, and [Chapter 81](chapter1076.md) extends the discussion to RNA sponges, toxin-antitoxin RNAs, and network buffering motifs.

## Executive Summary

Bacterial small RNAs are short regulatory transcripts, often about 50 to 300 nucleotides long, that change gene expression after transcription has begun. The most familiar examples act by base pairing with mRNAs. A small RNA can mask a Shine-Dalgarno sequence and repress translation, open an inhibitory mRNA structure and activate translation, recruit or block RNases, change the lifetime of an mRNA, or coordinate several targets at once. Other bacterial regulatory RNAs do not primarily use base pairing. CsrB/CsrC-like and RsmY/RsmZ-like RNAs regulate by sequestering CsrA or Rsm proteins through repeated binding motifs. Some processed 3′ UTRs, attenuated leaders, and mRNA fragments can also become regulatory RNAs after they are cut from longer transcripts.

The words "small RNA" describe size and regulatory behavior, not one biogenesis pathway. Trans-encoded small RNAs are transcribed from loci separate from their targets and usually pair imperfectly with multiple mRNAs. Cis-antisense RNAs are transcribed from the opposite strand of the same locus and often pair extensively with one overlapping sense transcript. A 3′ UTR-derived small RNA begins as part of an mRNA and becomes a separate regulator after processing or termination. A dual-function small RNA carries regulatory base-pairing information and a short open reading frame. These categories are useful, but they overlap: an RNA can be processed from a leader, bind Hfq, encode a peptide, and also regulate targets in trans.

Hfq is the best-studied bacterial small RNA chaperone. Hfq is a ring-shaped Sm-like homohexamer with distinct RNA-binding surfaces. U-rich 3′ ends and terminator-proximal sequences commonly contact the proximal face; A-rich or ARN motifs in target RNAs can contact the distal face; and basic residues around the rim help position imperfect duplexes. Hfq does not simply hold RNAs. Hfq changes the kinetics of RNA encounter, protects many small RNAs from degradation, presents seed regions to nascent or unfolded target sites, and can couple base pairing to RNase E and the degradosome. Rodgers et al. (2023) emphasize that sRNAs and Hfq can capture target sites while the target RNA is still being transcribed, which explains why timing and local structure are central to bacterial riboregulation.

ProQ is a second major bacterial RNA-binding protein family, but ProQ biology is not a duplicate of Hfq biology. ProQ proteins contain a FinO-like RNA-binding domain and tend to bind structured RNAs, terminator hairpins, and subsets of mRNAs and small RNAs that only partially overlap with Hfq-associated RNAs. ProQ can stabilize RNAs, influence base pairing, and reshape regulatory networks in species-specific ways. Ghandour et al. (2025) provide a recent example in Vibrio cholerae, where ProQ-associated small RNAs affect motility. Other bacteria use different RNA-binding proteins or RNA decay proteins as regulatory cofactors, so the absence of a canonical Hfq-centered system does not imply the absence of small RNA regulation.

The Csr/Rsm systems illustrate a different design principle. CsrA and RsmA-like proteins bind GGA-containing motifs, often in loops, on mRNAs and regulatory RNAs. On many mRNAs, binding near a ribosome binding site represses translation, while binding at other positions can stabilize transcripts or alter decay. Regulatory RNAs such as CsrB, CsrC, RsmY, and RsmZ contain many CsrA/RsmA binding sites and act as molecular sponges. When these RNAs are abundant, they titrate the protein away from mRNA targets; when these RNAs are degraded, free protein represses or remodels target expression. This sequestration logic makes the Csr/Rsm module a post-transcriptional switchboard for metabolism, motility, biofilm formation, and virulence.

Small RNA networks are powerful because they are fast, economical, and conditional. A single transcription factor can induce an sRNA that immediately changes translation or stability of many preexisting mRNAs. A stress response can combine transcriptional control, sRNA base pairing, RNase E-mediated decay, Csr/Rsm sequestration, and feedback through target mimic RNAs. This architecture explains why bacterial sRNAs are repeatedly linked to envelope stress, iron homeostasis, carbon use, quorum behavior, biofilms, sporulation, and host infection. It also creates interpretation hazards: deletion of an sRNA chaperone changes many RNAs indirectly, overexpressed sRNAs can pair with nonphysiological targets, and ligation-based interactome methods capture proximity that still requires functional validation.

The current consensus is that bacterial small RNAs are not peripheral fine-tuners. They are central nodes in post-transcriptional regulatory circuits. Strong mechanistic claims require evidence that an RNA is expressed in the relevant condition, physically contacts a target or protein partner, changes target expression in the native context, and has a causal pairing site or protein-binding motif. High-throughput methods such as RIL-seq, CLASH, Grad-seq, RNA-seq, proteomics, and interactome profiling have transformed target discovery, but they are best used as entry points into genetics, biochemistry, and physiology rather than as substitutes for them.

## Concept Inventory

- **Bacterial small regulatory RNA:** A bacterial RNA, usually shorter than a typical mRNA, that regulates gene expression without needing to encode a full-length protein. Many bacterial small regulatory RNAs are noncoding, but some are dual-function RNAs that also encode a small peptide.
- **Trans-encoded small RNA:** A small RNA transcribed from a locus separate from its target genes. Trans-encoded sRNAs usually pair imperfectly with several targets and often need Hfq, ProQ, or another RNA-binding protein for efficient regulation.
- **Cis-antisense RNA:** An RNA transcribed from the DNA strand opposite an overlapping sense transcript. Because the antisense RNA and target are complementary over a shared genomic interval, cis-antisense RNAs often form long duplexes and frequently affect transcription, translation, or RNA stability at the same locus.
- **3′ UTR-derived small RNA:** A regulatory RNA generated from the 3′ untranslated region of an mRNA or from a terminator-proximal region. These RNAs blur the line between mRNA fragments and independent regulators because the same genomic transcript can encode a protein upstream and a regulatory RNA downstream.
- **Hfq:** A bacterial Sm-like RNA-binding protein that forms a homohexameric ring and facilitates many sRNA-mRNA interactions. Hfq stabilizes numerous sRNAs, accelerates pairing, organizes RNA surfaces, and can connect regulatory duplexes to RNA decay.
- **Seed region:** A short segment of an sRNA that initiates base pairing with a target RNA. The seed often contains accessible single-stranded nucleotides and can determine target specificity.
- **RNase E and degradosome coupling:** A mechanism in many Gram-negative bacteria in which sRNA-mRNA pairing promotes cleavage by RNase E, often within the multiprotein degradosome. Coupling can degrade the target, the sRNA, or both.
- **ProQ:** A FinO-domain RNA-binding protein family that binds structured RNAs and regulates a target set partly distinct from Hfq-bound RNAs. ProQ proteins are widespread but variable, so species-specific evidence matters.
- **Csr/Rsm system:** A protein-centered post-transcriptional regulatory system in which CsrA or RsmA-family proteins bind GGA motifs on mRNAs and regulatory RNAs. Large noncoding RNAs with repeated binding sites sequester the proteins and modulate target regulation.
- **RNA chaperone:** A protein that assists RNA folding, RNA-RNA pairing, or RNA remodeling without being a template. The term does not imply that the protein uses ATP or that it creates one final structure; many RNA chaperones change folding kinetics and binding opportunities.
- **CLASH-style target discovery:** A family of methods that enrich RNA-RNA contacts by ligating physically associated RNA fragments, often after immunoprecipitation of an RNA-binding protein. In bacteria, related strategies include RIL-seq and other ligation-based or affinity-based RNA interactome approaches.

## What to Know Before Reading This Chapter

Bacterial gene expression is often controlled after transcription initiation. A promoter may turn on a transcript, but the final protein output depends on translation initiation, mRNA structure, RNA processing, and decay. A small RNA can therefore regulate a gene without changing promoter activity. If the sRNA blocks the ribosome binding site, protein synthesis falls even if mRNA synthesis is unchanged. If the sRNA recruits RNase E, mRNA abundance falls and translation falls as a consequence. If the sRNA opens an inhibitory structure, translation can rise even though the sRNA is base paired to the mRNA.

Readers should also distinguish imperfect and extensive RNA pairing. A trans-encoded sRNA usually recognizes targets through short, imperfect seed pairing because the sRNA and target did not arise from opposite strands of the same DNA segment. A cis-antisense RNA can pair extensively with its overlapping transcript because the two RNAs are genomically complementary. Extensive duplex formation can trigger RNase III cleavage, block transcription elongation, alter translation, or change RNA stability. Imperfect pairing often depends more strongly on chaperones such as Hfq or ProQ.

Finally, small RNA regulation should be read as network regulation. A bacterial sRNA rarely has only one consequence. The sRNA may directly repress one mRNA, indirectly activate another by repressing a repressor, compete with other sRNAs for Hfq, be sequestered by an RNA sponge, and be degraded together with a highly expressed target. Csr/Rsm systems add another layer because the regulatory RNA controls a protein that controls many mRNAs. These interactions create feed-forward loops, feedback loops, thresholds, and stress-specific priorities.

## Core Mechanisms and Molecular Players

## 80.1. Bacterial small RNA classes and biogenesis

The term bacterial small RNA, often abbreviated sRNA, is a practical name for short regulatory RNAs in bacteria. The name does not define a single structure, function, origin, or protein partner. Some sRNAs are independent transcripts with their own promoters and terminators. Some are antisense RNAs made from the opposite strand of a coding locus. Some are produced by cleavage from mRNAs or untranslated regions. Some are stable regulatory molecules, whereas others are transient decay intermediates that acquire regulatory activity only in a particular condition.

Trans-encoded sRNAs are the best-known class in Hfq-centered Gram-negative model systems. These RNAs are encoded at loci separate from their mRNA targets and usually share only short imperfect complementarity with each target. A trans-encoded sRNA often contains a structured scaffold that binds Hfq plus an exposed seed region that pairs with target mRNAs. Because the pairing is short, one sRNA can regulate many targets, and one mRNA can be regulated by several sRNAs. This modularity explains why trans-encoded sRNAs are useful for stress responses: the cell can transcribe one short RNA and rapidly adjust a regulon of preexisting mRNAs.

Cis-antisense RNAs follow different rules. A cis-antisense RNA is made from the opposite strand of the same genomic region as the target transcript. The overlap can include a 5′ leader, coding sequence, 3′ UTR, or terminator. Because the antisense and sense RNAs are highly complementary, they can form longer duplexes than most trans-encoded pairs. This extensive pairing can block ribosome binding, expose or hide processing sites, promote RNase III cleavage of double-stranded RNA, or influence transcription through convergent transcription and local DNA topology. Wagner and Romby (2015) place cis-antisense RNAs alongside trans-encoded sRNAs and other bacterial regulatory RNAs, while older antisense-focused reviews document the genetic-element examples that shaped this classification. The strongest cis-antisense claims require strand-specific transcript mapping; otherwise a read pileup can be mistaken for the wrong strand or for readthrough transcription.

A third category includes sRNAs derived from longer transcripts. A 3′ UTR can become a stable regulatory RNA after RNase processing or after transcription termination leaves a structured downstream fragment. A 5′ leader or attenuated transcript can also survive as a small RNA. The biological meaning of such RNAs must be judged case by case. A processed fragment may be waste from RNA decay; it may protect an mRNA by forming a structure; or it may act in trans on other targets. The chapter on cis-regulatory elements introduced leader-derived regulatory RNAs; this chapter emphasizes the broader principle that bacterial regulatory RNAs can be born from mRNA architecture rather than from dedicated noncoding genes.

Small RNA biogenesis begins with transcription. Many sRNA promoters respond to alternative sigma factors, two-component systems, metabolite-responsive transcription factors, envelope stress regulators, iron regulators, quorum sensing regulators, or virulence programs. A stress-induced sRNA is therefore often a bridge between transcriptional and post-transcriptional control. For example, a transcription factor can induce an sRNA that represses mRNAs made under a previous condition. The response is fast because the target mRNAs already exist and because the sRNA can change translation or decay before new transcriptional steady states are reached.

RNA 3′ ends and terminators are important for sRNA stability. Many Hfq-dependent sRNAs end in an intrinsic terminator with a stem-loop followed by a U-rich tail. The U-rich tail can bind Hfq, and the stem-loop can help protect the RNA from exonucleases. A small change in terminator structure can therefore change sRNA abundance independently of target pairing. This is one reason why mutational tests must separate seed-region mutations from scaffold and terminator mutations. A seed mutation tests target recognition; a terminator mutation may mainly change RNA stability or Hfq binding.

Processing and degradation enzymes are not merely cleanup machinery. RNase E can cut sRNA targets, trim sRNA precursors, or promote coupled degradation of an sRNA-mRNA duplex. RNase III can cleave extended double-stranded RNA formed by antisense pairing. PNPase, poly(A) polymerase, and other decay factors influence the lifetime of regulatory RNAs and their targets. Dendooven et al. (2021) highlight that PNPase, Hfq, and RNA can form a cooperative carrier complex, reinforcing the idea that decay enzymes and RNA chaperones can participate directly in riboregulation rather than acting only after regulation has occurred.

Not every bacterial group uses the same toolkit. Escherichia coli and Salmonella have made Hfq-dependent trans-encoded sRNAs the canonical examples, but Gram-positive bacteria, mycobacteria, and many pathogens often have different sRNA repertoires, different chaperone requirements, and different RNA decay pathways. Taneja and Dutta (2019) emphasize that mycobacterial sRNA identification is methodologically challenging because annotation, growth state, transcript boundaries, and validation standards all affect the apparent repertoire. The conservative rule is to define the RNA class, biogenesis evidence, and protein partners for the organism being studied rather than importing assumptions from E. coli.

![Figure 80.1. Bacterial small RNA classes and biogenesis](../assets/figures/chapter1075_figure1.png)

**Figure 80.1. Bacterial small RNA classes and biogenesis.** Bacterial small regulatory RNAs arise as independent trans-encoded transcripts, cis-antisense transcripts, processed 5′ or 3′ UTR fragments, leader-derived transcripts, dual-function RNAs, and protein-sequestering RNAs. Each class differs in target range, pairing extent, protein cofactors, and evidence requirements.

## 80.2. Hfq-mediated base pairing and target regulation

Hfq is a bacterial RNA chaperone whose molecular architecture explains much of its regulatory logic. The protein forms a ring-shaped hexamer related to Sm and Lsm proteins. Each Hfq hexamer has a proximal face, a distal face, and a rim. In many Gram-negative bacteria, the proximal face binds U-rich sequences such as sRNA 3′ tails, the distal face binds A-rich motifs found in many mRNAs and some sRNAs, and the rim helps bring short complementary regions into register. This organization allows Hfq to hold two RNAs at once, concentrate them locally, and increase the probability that a seed region finds a target site.

The first causal step in Hfq-mediated regulation is usually sRNA loading and stabilization. A newly transcribed sRNA with a suitable terminator can bind Hfq before nucleases destroy it. Hfq binding can protect the sRNA, keep the seed accessible, and place the sRNA in competition with other Hfq-bound RNAs. This competition matters physiologically. If one sRNA is induced to very high abundance, it can occupy Hfq and change the activity of other sRNAs. Therefore an hfq deletion phenotype is not simply the sum of all direct sRNA-target interactions; it also reflects altered RNA stability, altered decay, and altered allocation of a limiting protein.

The second step is target encounter. An sRNA seed region pairs with a target site on an mRNA. The target site may be in the 5′ untranslated region, the Shine-Dalgarno sequence, the start codon region, an early coding sequence, an internal coding region, or the 3′ UTR. Pairing near the Shine-Dalgarno sequence commonly represses translation by preventing 30S ribosomal subunit access. Pairing just upstream can stabilize an inhibitory structure or recruit RNase E. Pairing within a coding sequence can alter ribosome movement, RNA structure, or protein output, although internal coding-sequence regulation usually requires careful validation because changes in mRNA abundance and translation are easy to confound.

The third step is output. The same physical event, sRNA-mRNA pairing, can produce different outputs depending on target architecture. Repression occurs when pairing blocks ribosome binding or accelerates target degradation. Activation occurs when pairing disrupts an mRNA structure that was hiding the ribosome binding site. DsrA activation of rpoS, supported by biochemical and genetic studies from Vecerek et al. (2010) and Wang et al. (2013), is a classic example in which Hfq and an sRNA remodel the translation-initiation region rather than merely occluding it. Stabilization occurs when pairing blocks an RNase site or creates a protective structure. Target-coupled sRNA decay occurs when the duplex is cleaved and both molecules are lost. This diversity explains why the word "target" is insufficient by itself; each target must be annotated with the site, direction of regulation, molecular output, and evidence.

RNase E and the degradosome provide one major route from pairing to target decay. In many Gram-negative bacteria, RNase E cleaves single-stranded RNA and works within a degradosome that can include helicase, enolase, and PNPase components. An sRNA-Hfq complex can recruit RNase E directly or indirectly, and base pairing can expose a cleavage site or create a conformation favorable for cleavage. After cleavage, exonucleases complete degradation. The consequence can be rapid silencing of an mRNA and turnover of the sRNA. This coupling makes regulatory output sensitive to growth rate, RNA abundance, target transcription, RNase availability, and competing RNA substrates.

Timing is a central principle. Rodgers et al. (2023) showed that small RNAs and Hfq can capture unfolded target sites during transcription. This point resolves a common puzzle: many target sites are predicted to be buried in the mature mRNA structure, yet regulation occurs in vivo. If the sRNA encounters the nascent transcript before the target folds into a stable structure or before ribosomes occupy the site, the regulatory duplex can form at a window of vulnerability. Co-transcriptional pairing means that transcription speed, pausing, ribosome loading, and local folding kinetics all influence whether the sRNA succeeds.

Hfq-dependent regulation is therefore both sequence-specific and context-dependent. Seed pairing provides specificity, but the accessible target structure, Hfq-binding motifs, RNA abundance, cellular condition, and timing determine whether a predicted pair functions. A short perfect seed match may fail if the site is never accessible. A weak-looking interaction may succeed if Hfq positions the RNAs and the target is exposed during transcription. For this reason, computational target prediction is strongest when combined with expression data, Hfq occupancy, RNA-RNA interaction data, and mutational rescue of the seed and target site.

Hfq also participates in networks beyond individual sRNA-mRNA pairs. In some systems, RNAs can sponge an sRNA by presenting a decoy target site; this topic is developed further in [Chapter 81](chapter1076.md). In other systems, an mRNA target can titrate an sRNA away from additional targets, creating target hierarchy. Hfq-associated decay factors can make a strongly expressed target accelerate sRNA turnover, thereby relieving repression of weaker targets. These network effects are not exceptions; they are expected when many RNAs share a finite chaperone and a finite set of degradation enzymes.

![Figure 80.2. Hfq-mediated sRNA targeting and output](../assets/figures/chapter1075_figure2.png)

**Figure 80.2. Hfq-mediated sRNA targeting and output.** Hfq organizes sRNA and mRNA contacts through proximal, distal, and rim surfaces. The same base-pairing logic can block translation initiation, open an inhibitory mRNA structure, recruit RNase E and the degradosome, or trigger coupled turnover of target and sRNA.

## 80.3. ProQ, Csr/Rsm, and alternative RNA chaperone systems

ProQ is often introduced as another bacterial sRNA chaperone, but that shorthand can be misleading. ProQ proteins belong to the FinO-domain family and bind structured RNAs with preferences that differ from Hfq. Liao and Smirnov (2023) frame FinO/ProQ proteins as an evolutionarily diverse family rather than a single Hfq substitute. The founding FinO system was linked to plasmid transfer regulation, and many ProQ proteins retain an affinity for stem-loop and terminator-like structures. In enteric bacteria, ProQ-associated RNAs include mRNAs, antisense RNAs, and small RNAs, with only partial overlap with the Hfq-bound transcriptome. Thus ProQ is best understood as a parallel RNA-binding hub whose targets and mechanisms must be measured in each organism.

Mechanistically, ProQ can stabilize bound RNAs, alter their folding, and facilitate or inhibit RNA-RNA interactions. A structured RNA end can act as a binding handle. Smirnov et al. (2016) used Grad-seq to identify ProQ as a major bacterial sRNA-binding protein, demonstrating how fractionation-based ribonucleoprotein mapping can reveal RNA-binding hubs without starting from a known sRNA target. Once ProQ is bound, the protein may protect the RNA from nucleases, present a pairing region, or alter access by ribosomes and decay factors. The current evidence base suggests that ProQ biology is broad but less unified than Hfq biology. Some ProQ-regulated RNAs behave like classical base-pairing sRNAs, while others appear to be structured mRNAs or antisense RNAs whose stability changes when ProQ is absent.

Ghandour et al. (2025) provide a recent organism-specific example by identifying ProQ-associated small RNAs that control motility in Vibrio cholerae. That study is useful pedagogically because motility is a complex output: a change in motility can reflect flagellar gene expression, chemotaxis, metabolism, biofilm state, stress, or growth rate. A ProQ-associated sRNA can be a direct regulator of a motility mRNA, an indirect regulator upstream of a transcription factor, or a node in a broader physiological network. The mechanistic burden is therefore to connect RNA binding, target regulation, and the motility phenotype.

The Csr/Rsm systems use a different architecture. Romeo et al. (2013) summarize the form and function of these systems as global post-transcriptional regulators built around CsrA/Rsm-family RNA-binding proteins and regulatory RNAs. CsrA in E. coli-like nomenclature and RsmA or RsmE in many pseudomonads and related bacteria are small RNA-binding proteins that usually recognize GGA-containing motifs in single-stranded loop regions. On mRNAs, these motifs often occur near ribosome binding sites. When CsrA or RsmA binds near a Shine-Dalgarno sequence or start codon, the protein can repress translation directly. Binding elsewhere can stabilize or destabilize an mRNA, change RNA structure, or alter access by decay enzymes. The same protein can therefore repress some genes and activate others, depending on binding position and target architecture.

The regulatory RNAs in Csr/Rsm systems are not primarily base-pairing sRNAs. CsrB, CsrC, RsmY, RsmZ, and related RNAs contain repeated GGA motifs that bind many copies of CsrA or RsmA-family proteins. These RNAs behave as protein sponges. When sponge RNAs are abundant, they sequester the protein away from mRNA targets. When sponge RNAs are degraded or transcriptionally reduced, free protein binds target mRNAs. Two-component systems such as BarA/UvrY or GacS/GacA often regulate production of the sponge RNAs, connecting environmental signals to post-transcriptional protein sequestration.

This sequestration design creates threshold behavior. A modest change in CsrB or RsmY abundance can have little effect if CsrA or RsmA is still mostly bound by sponge RNA. Once sponge capacity falls below protein abundance, free protein can rise sharply and regulate many targets. Conversely, strong induction of sponge RNA can rapidly release mRNAs from repression. Suzuki et al. (2006), Vakulskas et al. (2016), and Leng et al. (2016) show that CsrD, RNase E-dependent turnover, CsrA occupancy, and nutrient-linked signaling can tune CsrB/CsrC-like RNA lifetime, adding a decay-controlled timer to the system.

Alternative RNA chaperone and protein-centered systems extend the same logic. Some bacteria use Hfq together with Crc in carbon catabolite repression, where RNAs such as CrcZ or CrcY can modulate protein availability. Some Gram-positive systems rely less on Hfq and more on organism-specific RNA-binding proteins, RNases, or structured RNA elements. Cold-shock proteins, ribosomal proteins, helicases, and degradosome components can all influence regulatory RNA folding or stability. The practical lesson is not that every bacterium has an Hfq-ProQ-Csr triad, but that bacterial post-transcriptional regulation repeatedly uses small RNAs, RNA-binding proteins, decay enzymes, and structured RNA ends to couple physiology to expression.

**Table 80.1. Comparison of bacterial post-transcriptional RNA regulatory systems.** Bacterial post-transcriptional RNA regulation uses several architectures. The appropriate evidence standard depends on whether the regulatory RNA pairs with another RNA, binds and sequesters a protein, arises from a processed transcript, or changes RNA decay.

| System | Typical RNA origin | Major protein partner | Target-recognition logic | Common output | Key caveat |
| --- | --- | --- | --- | --- | --- |
| **Hfq-dependent trans-encoded sRNAs** | Independent sRNA locus with promoter, structured body, and often U-rich 3′ terminator tail | Hfq, often with RNase E, degradosome, or PNPase context | Short imperfect seed pairing to accessible mRNA sites; Hfq surfaces position both RNAs | Translation repression or activation, mRNA decay or stabilization, target-coupled sRNA turnover | Strongest paradigm is enteric Gram-negative systems; hfq phenotypes are highly pleiotropic |
| **Cis-antisense RNAs** | Opposite strand transcript overlapping a sense RNA at the same locus | Often no dedicated chaperone; RNase III or decay factors may act on duplexes | Extensive complementarity across the overlap region | Translation block, RNA processing, RNase III cleavage, stability change, or local transcription effects | Requires strand-specific mapping; overlap and expression do not prove regulation |
| **ProQ-associated RNAs** | Structured sRNAs, antisense RNAs, and mRNAs with hairpins or terminator-like ends | ProQ or another FinO-domain protein | Recognition of structured RNA features; stabilization, remodeling, or pairing support | RNA stabilization, altered translation or decay, species-specific network outputs such as motility | ProQ is not simply Hfq-like; regulons and mechanisms are organism-specific |
| **Csr/Rsm sponge RNAs** | CsrB/CsrC-, RsmY/RsmZ-, or related RNAs with repeated GGA motifs | CsrA, RsmA, RsmE; CsrD and RNase E can tune sponge lifetime | Stoichiometric sequestration of RNA-binding proteins, not primary mRNA base pairing | Threshold control of free Csr/Rsm protein and downstream metabolism, motility, biofilm, or virulence genes | Total RNA or protein abundance may not report the free active pool |
| **Processed UTR-derived sRNAs** | Stable 5′ leader, 3′ UTR, terminator-proximal, or attenuated-transcript fragment | Context-dependent Hfq, ProQ, RNases, or no known partner | Processed fragment acts in trans, protects a parent RNA, or changes decay architecture | Target regulation, RNA stabilization, or condition-specific bridge from mRNA architecture to regulation | May be a degradation product; requires processing, stability, partner, and output evidence |
| **Alternative chaperone-associated RNAs** | Organism-specific sRNAs in Gram-positive, mycobacterial, pathogenic, or carbon-control systems | Crc-Hfq assemblies, cold-shock proteins, helicases, ribosomal proteins, RNases, or unknown proteins | Protein-assisted folding, remodeling, sequestration, or decay control rather than one universal seed rule | Carbon catabolite control, stress adaptation, virulence tuning, or RNA lifetime changes | Do not import E. coli assumptions without organism-specific partner and mechanism evidence |

![Figure 80.3. ProQ and Csr/Rsm alternatives to the canonical Hfq pair](../assets/figures/chapter1075_figure3.png)

**Figure 80.3. ProQ and Csr/Rsm alternatives to the canonical Hfq pair.** ProQ-family proteins bind structured RNAs and shape species-specific RNA regulons, whereas Csr/Rsm systems use regulatory RNAs with repeated protein-binding motifs to sequester CsrA/Rsm-family proteins. These systems regulate expression through RNA stabilization, RNA remodeling, target competition, and protein availability.

> **Box 80.1. Common overinterpretations in bacterial small RNA studies**
>
> - Expression change is not direct regulation.
>  - Protein enrichment is not regulatory output.
>  - A chimeric read is contact evidence, not causality.
>  - Hfq deletion is a global perturbation.
>  - Csr/Rsm sponge RNAs sequester proteins.
>  - ProQ biology is organism-specific.

## 80.4. Network architecture, stress responses, virulence, and metabolism

Small RNA networks are most powerful when cells must change protein output quickly. Transcriptional regulation changes the production of new RNA molecules. Small RNA regulation can act on transcripts that have already been made. This distinction matters during envelope stress, oxidative stress, iron limitation, nutrient shifts, temperature shifts, quorum transitions, and host entry. A bacterium can induce an sRNA, pair it with many existing mRNAs, and redirect translation or decay within minutes.

Many sRNA circuits combine direct and indirect regulation. An sRNA may directly repress a porin mRNA during envelope stress. The same sRNA may indirectly activate a stress response by repressing a repressor. Another sRNA may target the first sRNA's sponge or compete for Hfq. A transcription factor may induce both an sRNA and a protein regulator, creating a coherent feed-forward loop if both arms repress the same target or an incoherent loop if one arm activates and the other represses. Such circuits can filter noise, delay expression, sharpen thresholds, or prioritize targets under limiting chaperone conditions.

![Figure 80.4. Bacterial sRNA regulatory network motifs and timing behavior](../assets/figures/chapter1075_figure4.png)

**Figure 80.4. Bacterial sRNA regulatory network motifs and timing behavior.** Small RNAs create circuit behavior through recurring network motifs. A coherent feed-forward loop can reject a brief input; negative feedback can stabilize output; competition for limiting Hfq can prioritize one sRNA-target route over another; a sponge can impose a concentration threshold; and repression of a repressor can produce delayed target activation. Arrows and repression bars describe regulatory sign, not direct molecular binding in every edge.

Metabolism is a major arena for sRNA control. Papenfort and Storz (2024) emphasize that small RNAs provide insights into bacterial metabolism because they link nutrient signals to transporters, catabolic enzymes, central carbon flux, iron homeostasis, and stress adaptation. An sRNA can down-regulate an importer when a nutrient is absent, repress energetically costly pathways during stress, or coordinate carbon and nitrogen metabolism with envelope status. Csr/Rsm systems are especially important in metabolic transitions because CsrA/Rsm proteins regulate enzymes, transporters, motility factors, and biofilm-associated genes through a shared post-transcriptional pool.

Virulence regulation uses the same molecular tools but with host-associated inputs. Djapgne and Oglesby (2021) review the impact of small RNAs and chaperones on bacterial pathogenicity. Pathogens encounter temperature change, iron restriction, oxidative stress, antimicrobial peptides, immune pressure, mucus, bile, and nutrient limitation. Small RNAs can tune outer membrane proteins, secretion systems, motility, adhesins, toxins, and metabolic programs. Hfq mutants in many pathogens show pleiotropic defects, but those defects must be interpreted carefully because Hfq affects many sRNAs and mRNAs at once.

Recent primary studies in the local bibliography illustrate this diversity. Barros et al. (2025) connect Hfq and small noncoding RNAs to biofilm formation in the fish pathogen Yersinia ruckeri. Fuchs et al. (2023) describe a network of small RNAs regulating sporulation initiation in Clostridioides difficile. Busch et al. (2025) link the Staphylococcus aureus sRNA IsrR to iron limitation and SaeRS activation, connecting nutrient stress and virulence regulation. Ghandour et al. (2025) connect ProQ-associated small RNAs to Vibrio cholerae motility. These examples should not be collapsed into one universal pathway; they show that bacteria repeatedly recruit sRNAs for condition-specific decisions.

Network architecture also explains why phenotypes can be hard to assign. If deletion of one sRNA changes biofilm formation, the direct target might be a biofilm matrix gene, a motility regulator, a metabolic enzyme, an envelope protein, or another regulator. If deletion of hfq changes virulence, the phenotype might reflect loss of many sRNAs, altered mRNA stability, altered protein synthesis, and stress sensitivity. If a CsrB-like RNA changes metabolism, the direct biochemical event is protein sequestration, but the downstream phenotype can involve dozens of mRNAs. Strong interpretation therefore moves from phenotype to RNA expression, partner binding, target site mapping, mutational rescue, and physiological output.

Some sRNA networks have built-in hierarchy. Targets with accessible sites and strong pairing can be regulated at low sRNA abundance. Targets with weaker sites may be regulated only when the sRNA is highly induced. Highly abundant targets can act as sponges that consume the sRNA and protect lower-affinity targets. RNase-coupled target decay can make regulation irreversible on the timescale of a stress response, whereas translational repression without decay may be rapidly reversible. Csr/Rsm systems add stoichiometric hierarchy because each regulatory RNA molecule can bind multiple protein dimers or monomers, depending on the system and occupancy.

The boundary between stress regulation and virulence regulation is often artificial. Host environments impose stresses, and stress responses often control virulence traits. Iron limitation is both a nutritional challenge and a host defense. Envelope stress can result from antimicrobial peptides or secretion system activity. Motility can help colonization but may also trigger immune recognition. Biofilm formation can aid persistence but reduce acute dissemination. Small RNAs are well suited to such tradeoffs because they can tune rather than simply switch expression.

## 80.5. Target discovery, CLASH-style methods, and modeling

Target discovery begins with a simple question: which RNA or protein does the regulatory RNA affect directly? The answer requires separating direct physical contact from indirect network consequences. RNA-seq after deleting or overexpressing an sRNA identifies transcripts whose abundance changes, but abundance changes can be indirect. Proteomics identifies protein-output changes, but translation and protein stability can obscure the RNA-level mechanism. Reporter assays test candidate target regions, but plasmid reporters can miss native chromosomal context. These methods are valuable when interpreted as complementary evidence rather than as a single decisive test.

Computational target prediction usually searches for short complementary regions, accessible structures, conservation, and favorable hybridization energy. Programs can rank candidate sRNA-mRNA pairs, but the biological problem is harder than finding complementarity. Hfq may bind one or both RNAs. The target site may be exposed only during transcription. Ribosomes may compete with the sRNA. RNase E may convert weak pairing into strong repression. A predicted target with no expression in the tested condition cannot be regulated in that condition. Predictions are therefore most useful for prioritizing mutational tests: disrupt the sRNA seed, disrupt the target site, and restore pairing with compensatory changes.

Affinity and gradient methods identify RNA partners of proteins. Hfq or ProQ co-immunoprecipitation followed by sequencing shows RNAs enriched with the protein. CLIP-style crosslinking can improve spatial resolution but introduces crosslinking biases. Grad-seq separates cellular complexes through a glycerol gradient and sequences RNAs across fractions, allowing RNAs to be grouped by protein-complex association; Smirnov et al. (2016) and Gerovac et al. (2021) provide the local ProQ and RNP-mapping anchors, while Chihara et al. (2022) illustrates a related size-exclusion profiling strategy in Escherichia coli. These approaches can identify candidate sRNAs, mRNAs, and ribonucleoprotein classes, but enrichment does not prove regulation. Highly abundant or structured RNAs can be enriched because they bind well, not because they are functionally regulated.

CLASH-style and RIL-seq-like methods try to capture RNA-RNA pairs more directly. The common logic is to stabilize RNA-protein complexes, immunoprecipitate a chaperone such as Hfq, ligate nearby RNA fragments, sequence chimeric reads, and infer interacting pairs from hybrid reads. Melamed et al. (2016) established RIL-seq as a bacterial small RNA-target mapping strategy, and Iosub et al. (2020) adapted Hfq CLASH to reveal nutrient-linked sRNA-target networks. Such methods are powerful because they can reveal unexpected targets, target-site positions, and network structure. They also have artifacts. Ligation efficiency depends on RNA ends and geometry. Crosslinking captures proximity, not necessarily productive regulation. Overabundant RNAs dominate libraries. Some contacts may form after lysis. A chimeric read should therefore be treated as contact evidence that requires functional validation.

**Table 80.2. Evidence ladder for bacterial sRNA target discovery.** No single high-throughput method establishes direct small RNA regulation. Strong target assignments combine co-expression, physical association, site-specific recognition, output measurement, and physiological consequence.

| Evidence type | What it shows | What it cannot show alone | Strongest follow-up |
| --- | --- | --- | --- |
| **RNA-seq perturbation** | Transcript abundance changes after sRNA, chaperone, or target perturbation | Direct contact, translation-only effects, and indirect network cascades | Pair with protein-output data and site-specific target tests |
| **Proteomics** | Protein abundance or output changes downstream of an RNA regulator | Whether the effect is translational, post-translational, growth-related, or indirect | Measure matched RNA abundance and test a native target-site reporter |
| **Reporter assay** | Candidate target segment is sufficient for regulation in the reporter context | Native chromosomal structure, expression level, long-range RNA features, or physiology | Mutate the endogenous site and compare with the reporter result |
| **Hfq/ProQ pulldown** | RNA enrichment with a candidate chaperone or RNA-binding hub | Direct RNA-RNA pairing or regulatory consequence | Map the site, perturb the RNA, and measure target output |
| **Grad-seq** | Co-sedimentation of RNAs with ribonucleoprotein fractions or protein complexes | Direct binding, contact geometry, or functional regulation | Validate the partner by IP or CLIP and test genetic dependency |
| **RIL-seq or CLASH-style chimera** | Chaperone-associated RNA-RNA proximity and candidate interaction sites | Productive regulation; ligation, crosslinking, abundance, and lysis biases remain | Disrupt both sites and restore regulation by compensatory rescue |
| **Structure probing** | RNA accessibility or protein/sRNA-induced structural change | Causal regulatory output or the same structure in native cellular timing | Combine with toeprinting, binding assays, and site mutants |
| **Seed and target mutation** | Loss of regulation after disrupting an sRNA seed, mRNA site, or Csr/Rsm GGA motif | Whether loss reflects recognition failure rather than RNA instability or coding changes | Use compensatory rescue while preserving abundance and coding potential |
| **Compensatory rescue** | Restoration of regulation by matched sRNA-target changes or restored binding motifs | Physiological relevance under native expression and stress conditions | Install native-context alleles and measure RNA, protein, and phenotype |
| **Endogenous physiological validation** | Molecular regulation affects stress, metabolism, motility, biofilm, sporulation, or virulence context | The direct molecular route when site and contact evidence are absent | Reconcile phenotype with expression, contact, mutation, and rescue evidence |

RNA interactome profiling and proximity approaches extend discovery beyond Hfq. Liu et al. (2023) provides a processed 3′ UTR-derived sRNA and RNA-interactome example that supports the broader evidence logic used here: proximity data are strongest when paired with genetic perturbation. If an sRNA-target chimera is observed, deletion of the sRNA should change the target, mutation of the pairing site should alter regulation, compensatory rescue should restore regulation, and the phenotype should be consistent with the target's known function.

Modeling helps interpret sRNA networks because abundance, pairing, competition, and decay are quantitative. A minimal model might include sRNA synthesis, target synthesis, Hfq binding, duplex formation, translation, RNase cleavage, and sRNA recycling or co-degradation. Such a model can predict threshold behavior, target hierarchy, sponge effects, and response times. For Csr/Rsm systems, models must account for protein abundance, multiple binding sites per sponge RNA, affinity differences among sites, mRNA target competition, and regulated sponge degradation. The goal is not to make regulation look mathematical for its own sake; the goal is to identify which measurements are needed to distinguish mechanisms.

Validation usually requires a ladder of evidence. First, show that the sRNA and target are expressed in the same condition and cell type. Second, show physical association by Hfq or ProQ pulldown, RNA-RNA ligation, or direct binding. Third, show that changing the sRNA changes target RNA or protein output. Fourth, map the site and test it with mutations. Fifth, rescue the interaction by compensatory base-pair restoration or by restoring Csr/Rsm binding motifs. Sixth, connect the molecular effect to physiology. The strongest studies avoid overexpression artifacts and test endogenous or near-endogenous expression whenever possible.

## Experimental Foundations and Evidence

Northern blotting remains valuable because it shows RNA size, processing, and condition-dependent abundance. A short RNA-seq read count can suggest that an sRNA exists, but a northern blot can distinguish a discrete sRNA from a smear of degradation products or from a fragment of a longer transcript. Strand-specific RNA-seq and differential RNA-seq help map transcription start sites and distinguish primary transcripts from processed RNAs. Termination mapping and long-read approaches can clarify whether a 3′ UTR-derived sRNA is independently transcribed or produced from an mRNA.

Reporter assays test regulatory output. A translational fusion preserves the target 5′ UTR and early coding region upstream of a reporter and is appropriate for testing ribosome binding site occlusion or activation. A transcriptional fusion tests promoter or transcriptional effects and is not sufficient for most sRNA translation mechanisms. Reporters should include the native target site and enough surrounding sequence to preserve RNA structure. A short target oligonucleotide inserted into a reporter can prove that the site is sufficient for regulation, but it may not prove that the full endogenous mRNA is regulated in the same way.

Mutational analysis is the strongest causal tool. For base-pairing sRNAs, mutate the sRNA seed so that regulation is lost, mutate the target site so that regulation is lost, and combine compensatory mutations that restore pairing and restore regulation. If only the sRNA mutation is tested, loss of regulation might reflect reduced sRNA stability or Hfq binding. If only the target mutation is tested, the mutation might alter mRNA structure or translation independently of sRNA pairing. Compensatory rescue is especially important when the predicted pairing site is short.

Biochemical assays add physical mechanism. Electrophoretic mobility shift assays can test RNA-protein binding, but the concentrations and RNA constructs must be biologically plausible. In-line probing, RNase footprinting, or chemical probing can test whether Hfq, ProQ, CsrA, or an sRNA changes RNA structure. Toeprinting can show whether an sRNA blocks ribosome initiation. Pulse-chase RNA stability assays can show whether regulation changes target half-life. Proteomics and ribosome profiling can distinguish mRNA abundance changes from translation changes, provided RNA abundance is measured in parallel.

Chaperone deletion experiments are useful but broad. An hfq mutant can destabilize many sRNAs, alter mRNA decay, change stress physiology, and affect translation. A proQ mutant can change structured RNA stability and indirect regulatory networks. A csrA mutant or csrB overexpression strain can perturb a large regulon. These perturbations are best used to identify candidate pathways and dependencies; direct target claims need site-specific tests.

## Biological Contexts Across Bacterial Systems

In enteric bacteria, Hfq-dependent sRNAs are deeply integrated into stress and nutrient responses. Many examples follow a common pattern: a transcriptional regulator detects the condition, induces an sRNA, and the sRNA post-transcriptionally represses or activates targets that would be costly or harmful under the new condition. The small size of the sRNA makes synthesis inexpensive, and the post-transcriptional output makes the response fast. However, the specific targets differ across species and strains, so conserved sRNA names do not guarantee identical regulons.

In Vibrio species, sRNAs, Hfq, ProQ, and quorum-associated circuits intersect with motility, biofilm formation, virulence, and environmental transitions. Ghandour et al. (2025) highlight ProQ-associated RNAs in Vibrio cholerae motility, while the broader literature links sRNAs to quorum-regulated behaviors in vibrios. The important teaching point is that motility and biofilm phenotypes can be downstream of several RNA-regulatory layers rather than direct readouts of one target.

In Clostridioides difficile, Fuchs et al. (2023) show that a network of small RNAs regulates sporulation initiation. Sporulation is a developmental decision, not a simple stress marker. A small RNA network can tune the timing and threshold of entry into a durable cell state. This example also warns against defining sRNAs only by Gram-negative Hfq paradigms; Gram-positive regulatory RNAs can be central even when their chaperone dependencies differ.

In Staphylococcus aureus, Busch et al. (2025) connect an sRNA to iron limitation and SaeRS activation. Iron limitation is a biologically meaningful host signal because vertebrate hosts restrict iron availability during infection. An sRNA that links iron limitation to virulence regulator activation can couple metabolism and host adaptation. The direct molecular details should be evaluated from the primary study, but the network principle is broadly relevant.

In mycobacteria, sRNA identification and validation require special care. Taneja and Dutta (2019) discuss challenges in finding and interpreting mycobacterial sRNAs. Mycobacteria differ from enteric bacteria in cell envelope physiology, growth rates, RNA decay context, and annotation maturity. An RNA detected under one growth condition may be absent under another. Slow growth and stress adaptation can make RNA stability and processing especially important variables.

## Technology, Computational, Clinical, and Engineering Links

Small RNA biology has practical consequences for antimicrobial research, pathogen surveillance, synthetic biology, and biotechnology. In pathogens, sRNAs can control virulence traits, antibiotic tolerance-associated physiology, biofilm formation, and host adaptation. This does not make every sRNA a drug target. A useful antimicrobial target must be accessible, conserved enough for the intended spectrum, important under infection conditions, and difficult for the bacterium to bypass. Hfq and CsrA-like proteins are pleiotropic, so targeting them could impose strong pressure but may also select rapid compensatory changes.

Synthetic biology uses bacterial sRNA principles to design post-transcriptional control devices. An engineered trans-acting RNA can repress a chosen mRNA by targeting the ribosome binding site. A synthetic sponge can sequester an sRNA. A CsrA-like scaffold can be adapted as a protein-sequestration module. The design challenge is the same as in natural systems: expression level, target accessibility, chaperone availability, RNA stability, and off-target pairing all influence output. Design should therefore include negative controls, orthogonal seed sequences, endogenous abundance estimates, and off-target transcriptome checks.

Computational modeling has improved because high-throughput data provide RNA abundance, target candidates, protein association, and sometimes direct RNA-RNA contacts. Still, models are only as good as their assumptions. A model that ignores Hfq competition may fail when many sRNAs are induced. A model that treats all CsrA binding sites as identical may miss site hierarchy. A model that uses steady-state RNA-seq may miss rapid transient pairing during transcription. Good models state what they include, what they omit, and what measurements would falsify them.

Clinical and ecological interpretation should remain cautious. Loss of an sRNA can attenuate virulence in a model host, but the effect may be indirect through growth, stress resistance, motility, or metabolism. An sRNA that matters in one strain may be absent or rewired in another. A biofilm phenotype in vitro may not predict persistence in a host or environment. The strongest translational claims connect molecular mechanism to organismal phenotype under conditions that resemble the relevant ecological or host context.

## Recent Consensus

The current consensus is that bacterial small RNAs are core components of regulatory networks rather than rare exceptions. Hfq-dependent sRNAs are a mature paradigm in many Gram-negative bacteria, and recent work has shifted attention from lists of targets to timing, target hierarchy, RNA structure, and network architecture. ProQ-associated RNAs show that major RNA-binding hubs can exist beside Hfq and can regulate physiologically important outputs such as motility. Csr/Rsm systems show that regulatory RNAs can control gene expression through protein sequestration rather than through base pairing.

There is also consensus that the boundaries among RNA classes are porous. A 3′ UTR can become a regulatory RNA. A leader transcript can become a trans-acting RNA. A nominally noncoding RNA can encode a small peptide. A target mRNA can become an RNA sponge. A protein-binding regulatory RNA can indirectly change many base-pairing interactions by changing physiology. Classification is useful only when it remains tied to mechanism.

Finally, evidence standards have become stricter. Differential expression alone is not direct regulation. Protein binding alone is not direct regulation. RNA-RNA ligation alone is not direct regulation. A convincing target assignment links expression, physical association, site-specific recognition, regulatory output, and physiological consequence. The field increasingly uses high-throughput methods for discovery and targeted genetics or biochemistry for causality.

## Open Questions, Controversies, Deprecated Models, and Common Misconceptions

Open questions:

- How many detected sRNAs are functional regulators? Deep sequencing can reveal thousands of short transcripts, but some are processing intermediates, degradation products, or condition-specific fragments with no separate function. Function should be assigned only when there is evidence for regulated production, stability, partner interaction, target output, or phenotype.
- What determines how ProQ-family mechanisms vary across bacteria? ProQ binds structured RNAs in several systems, but target repertoires and outputs differ. Some ProQ effects may reflect RNA stabilization more than active pairing. Some bacteria have FinO-domain proteins with specialized roles rather than broad ProQ-like hubs. Comparative claims should therefore be organism-qualified.
- Csr/Rsm quantitative behavior in natural environments? Laboratory models often measure one sponge RNA, one protein, and one target. In cells, multiple sponge RNAs, many mRNA targets, regulated decay, growth-rate changes, and environmental inputs can create complex stoichiometry. Predicting free CsrA or RsmA activity requires measuring more than total protein or total sponge RNA.

Common misconceptions:

- "All bacterial sRNAs work by Hfq-dependent base pairing." Many do, especially in enteric model systems, but cis-antisense RNAs, ProQ-associated RNAs, Csr/Rsm sponge RNAs, CRISPR RNAs, toxin-antitoxin RNAs, and processed UTR RNAs use different mechanisms or cofactors.
- "An hfq mutant phenotype identifies the target of one sRNA." Hfq affects many RNAs, many targets, and RNA decay pathways. Hfq dependency is a starting point for mapping, not a target assignment.
- "A chimeric read in RIL-seq or CLASH proves regulation." A chimera supports physical proximity or contact under the assay conditions. Functional regulation requires output changes and site-specific validation.
- "CsrB and RsmY are base-pairing sRNAs." Their primary function is protein sequestration through repeated binding motifs, not Watson-Crick pairing to mRNA targets.
- "Small RNA regulation is always repression." Base-pairing sRNAs and RNA-binding proteins can repress, activate, stabilize, destabilize, or tune expression depending on target architecture.