Meeting report

Abstract

On June 4–8, 2013, the 3rd Conference on Regulation with RNA in Bacteria took place in Würzburg, Germany. Following two earlier meetings in Berlin and San Juan, this conference has established itself as the primary bi-annual meeting for everyone interested in RNA-based regulations in prokaryotes. The 2013 meeting was organized by Joel Belasco, Susan Gottesman, Franz Narberhaus, and Jörg Vogel. Close to 300 participants from more than 27 countries in Europe, North America, and Asia enjoyed four days of talks and posters on many experimental and biocomputational aspects of prokaryotic RNA biology.

Keywords: :

• Cascade
• Hfq
• RNA stability
• RNA-seq CRISPR
• bioinformatics
• ribonucleoprotein complex
• small RNA

The study of regulatory RNAs in the control of prokaryotic genomes has become a very active and rapidly growing field. New small and large noncoding RNA molecules continue to be discovered at a staggering rate in model bacterial and archaeal organisms as well as in the transcriptomes of microbial communities. A six-year research program on Sensory and Regulatory RNAs in Prokaryotes funded by the German Research Foundation (DFG) has significantly contributed to progress in this field. The intention of this program was to bring together scientists to explore the structural and functional aspects of such riboregulators in diverse prokaryotic system.¹ About 25 research groups were involved in this joint program from 2007 to 2013. This Special Focus Issue of RNA Biology serves as the final report and presents an overview of most of the projects involved in this program.

One major contribution of this RNA priority program was to initiate an international conference series dedicated to prokaryotic riboregulators in order to provide a unique forum for the presentation of cutting-edge advances and the latest perspectives in the areas of discovery, mechanisms and structure of regulatory RNAs in bacteria and archaea. The first conference in 2009 in Berlin fully addressed the diversity of RNA regulators of gene expression and brought together 200 scientists involved in these studies. It also was the first meeting to highlight RNA discovery by emerging high-throughput sequencing technology in prokaryotes.² After four exciting days, it had become clear that the growing community of microbiologists with an interest in RNA would appreciate a regular meeting, ideally every two years, and alternating between Europe and North America. Gisela Storz, Karen Wassarman, and Jörg Vogel volunteered to organize the next meeting in San Juan, Puerto Rico, in March 2011. Around 180 participants enjoyed the science and informal interactions during relaxed strolls on the sandy beaches between the sessions. While small regulatory RNAs (sRNAs) and riboswitches still constituted the majority of the program, CRISPR systems emerged as a new topic, and the meeting also included a podium discussion on the pressing need for better annotations of functional RNAs and an overview of available bioinformatics approaches. No formal meeting report was published but many of the ideas discussed were covered in a review written by the co-organizers the same year.³

The series continued in 2013 at the Institute for Molecular Infection Biology (IMIB) in Würzburg and attracted close to 300 participants from more than 27 countries in Europe, North America, and Asia. Many will remember the impressive high water, which prevented the boat we had booked for Wednesday night from cruising the river Main. In terms of science, the conference lived up to the expectations from the previous meetings. Lectures by international experts, short presentations selected from the abstracts, and a large number of posters provided excellent forums for interactions between the participants. Impressively, the meeting showed how the field has moved forward in only a few years. In the following meeting report, we will briefly outline the fascinating results reported in the oral presentations. Many of these findings have now been published; appropriate references are included. In order to increase the breadth of presentations at this series of meetings, an emphasis was placed on inviting speakers who had not presented at the previous meeting in 2011. Additional members of the community who were not speakers were invited to chair sessions.

The opening session on Tuesday night, chaired by Jörg Vogel, was dedicated to new regulatory mechanisms in two workhorses of bacterial RNA research, Escherichia coli and Salmonella enterica. Gisela Storz reported on a dual function of the multi-cellular adhesive (McaS) sRNA in controlling biofilm formation in E. coli. Together with Poul Valentin-Hansen’s lab, they discovered that the McaS sRNA acts by two different mechanisms: base pairing and protein titration. Their initial studies established that McaS base pairs with mRNAs encoding the master transcription regulators of curli and flagella synthesis.^4,5 More recently, they found that McaS activates synthesis of the exopolysaccharide β-1,6 N-acetyl-D-glucosamine (PGA) by sequestering the global RNA-binding protein CsrA, a negative regulator of pgaA translation.⁶ Thus, McaS has roles in the two major post-transcriptional regulons controlled by the RNA-binding proteins Hfq and CsrA and can be added to the growing list of bi-functional RNAs.⁷

Another new post-transcriptional activation mechanism was the subject of the talk by Kai Papenfort. Trying to understand how the important phosphatase YigL is upregulated as part of the phosphosugar stress response in Salmonella, he discovered that the Hfq-dependent SgrS sRNA captures and stabilizes an intermediate of RNase E-mediated decay of the bicistronic pldB-yigL mRNA.⁸ This stabilization of an mRNA cleavage product may constitute a general route for sRNAs to rapidly activate genes by a mechanism independent of translation initiation, and complements the well-known translational activation by sRNAs through disruption of inhibitory structures around a target’s ribosome binding site. While SgrS was found to act within an operon mRNA, independent work demonstrated that RydC sRNA activated the monocistronic cfa mRNA by suppressing RNase E cleavage in the 5′ UTR.⁹

The last speaker of this session, Eduardo Groisman, focused on gene control in cis by alternative RNA stem-loop structures in 5′ leader regions of Salmonella mRNAs. The leader region of the virulence transcript mgtCBR responds to an increase in cytoplasmic ATP and/or a decrease in proline-charged tRNA^Pro by promoting transcription elongation into the associated coding region. Sensing of these two signals involves the coupling/uncoupling of transcription of the mgtCBR leader region and translation of two short open reading frames located within the leader RNA. The mgtCBR leader enables Salmonella to interpret environmental signals by the effects these have on cytosolic constituents: an increase in cytoplasmic ATP denotes a mildly acidic environment, while a decrease in proline-charged tRNA^Pro indicates hyperosomotic stress.^10,11

Tuesday evening closed with an insightful keynote lecture by Nobel laureate Sidney Altman who presented his view of the RNA-protein world. This talk covered general considerations of how regulatory RNA functions evolved and how bacterial enzymes such as RNase P may be exploited for biotechnological applications.

The presentations on Wednesday continued the theme of post-transcriptional regulation by sRNAs. The morning session, chaired by Susan Gottesman and Gerhart Wagner, opened with a talk by Pascale Romby on links between riboregulators and the conserved double-strand-specific ribonuclease RNase III in the human pathogen, Staphylococcus aureus. Previous work in her laboratory had established how RNase III and the virulence-associated sRNA RNAIII coordinately repress numerous mRNAs that encode transcriptional regulators of virulence factors, adhesin factors, peptidoglycan hydrolases, and metabolic enzymes. Using deep sequencing of transcripts bound in vivo to wild-type RNase III or a cleavage-deficient variant, many more potential substrates of this nuclease have now been discovered, implicating it in a host of functions including RNA processing, antisense RNA regulation, and degradation of structured RNAs.¹² One new RNase III substrate, a noncoding RNA indirectly linked to quorum sensing, activates biofilm development and represses capsule formation. This sRNA contributes to the complex interactions of S. aureus with the host immune system to moderate invasiveness and to favor chronic infections.¹³

In a set of related studies, Brice Felden described how the collaborative action of RNAIII and the pathogenicity island-encoded sRNA SprD control the synthesis of the host immune evasion protein Sbi in Staphylococcus aureus. These two sRNAs use antisense mechanisms to modulate Sbi expression at the translational level. Distinct RNAIII domains interact with sbi mRNA, including interactions deep in the coding region. Toeprints and in vivo mutational analysis have revealed a novel regulatory module within RNAIII essential for attenuation of sbi translation. The sophisticated translational control of this mRNA by two differentially expressed sRNAs ensures supervision of host immune escape by a major pathogen.¹⁴

Cynthia Sharma gave an overview of her pioneering work on a different group of human pathogens, the Epsilonproteobacteria Helicobacter pylori, the causative agent of gastritis, ulcers and gastric cancer, and Campylobacter jejuni, the most common cause of bacterial gastroenteritis to date. Her lab has employed differential RNA-Seq (dRNA-seq), originally used to map transcriptional start sites and sRNAs in H. pylori,¹⁵ for comparative analysis of multiple C. jejuni strains. These studies revealed strain-specific transcriptional start sites and sRNA repertoires, which may contribute to phenotypic differences among strains and strain-specific gene regulation.¹⁶ In an unrelated investigation, she showed that a trans-acting sRNA from Helicobacter base-pairs directly with a homopolymeric G-repeat in the mRNA leader of a chemotaxis receptor and that variation in the length of this G-repeat influences the outcome of post-transcriptional regulation by the sRNA.¹⁷ This first example from Helicobacter shows that research in Epsilonproteobacteria, which like many bacteria lack a homolog of the RNA-binding protein Hfq, can reveal new twists in sRNA-mediated gene regulation.

Sue Lin-Chao’s talk revisited one of the earliest bacterial regulatory RNAs identified, the antisense transcript RNAI. This plasmid-encoded antisense RNA is an important regulator of replication of ColE1-type plasmids.^18,19 As ColE1-type plasmids are widely used in genetic engineering and biotechnology, the effect of plasmids on bacterial hosts is an important issue. The Lin-Chao lab has begun to investigate the non-canonical function of RNAI as a regulator of genes important for central metabolism in bacterial host cells and the mechanism whereby ColE1-type plasmid DNA impacts its bacterial hosts.

In Rhodobacter sphaeroides, a complex regulatory network controls the expression of photosynthesis and stress response genes to optimize the supply of energy and minimize photooxidative stress. Gabriele Klug reported her recent discovery of a mixed incoherent feed-forward loop in this regulatory network, which comprises the transcription factor PrrA, the sRNA PcrZ, and photosynthesis target genes. Being itself activated by the response regulator PrrA under low oxygen tension, the trans-acting sRNA PcrZ represses mRNAs important for the synthesis of bacteriochlorophyll and pigment binding proteins. PcrZ thereby counteracts the redox-dependent induction of photosynthesis genes, which is mediated by protein regulators. These data highlight photosynthesis as another important physiological process in bacteria that is fine-tuned with the help of sRNAs.^20,21

In the last talk of this morning session, Ruth Schmitz-Streit presented on sRNAs from archaea. Using a combination of genetic, biochemical and computational approaches, her lab has studied the targets and functional role of sRNA₁₆₂ in the methanoarchaeon Methanosarcina mazei. This study revealed that sRNA₁₆₂ interacts with both cis- and trans-encoded target mRNAs via two distinct domains, a mechanism previously unknown in archaea.²² In the case of the trans-encoded target, sRNA₁₆₂ masks the predicted site of ribosome binding, thereby decreasing the synthesis of the encoded transcriptional regulator. In this manner, sRNA₁₆₂ may facilitate a metabolic switch between the carbon sources methanol and methylamines.

The Wednesday afternoon session, chaired by Hiroji Aiba and Poul Valentin-Hansen, focused on functions of the widespread RNA chaperone Hfq, alone or together with the sRNAs with which is associates. In the first talk, Udo Bläsi proposed an unexpected new function of Pseudomonas aeruginosa Hfq in carbon catabolite repression (CCR), a general response that helps bacteria to quickly adapt to nutrient availability. In Pseudomonas, the catabolite repression control protein Crc was previously reported to be a translational regulator, repressing functions involved in the uptake and utilization of carbon sources. CCR relief was previously attributed to the regulatory RNA CrcZ, which was proposed to trap Crc by mimicking target mRNAs. However, Bläsi showed structural and biochemical data to suggest that Crc is devoid of RNA-binding activity.²³ Instead, Hfq may be the principal post-transcriptional regulator of CCR in this bacterium. Hfq was shown to translationally repress CCR-regulated mRNAs, while the regulatory RNA CrcZ seemed to titrate Hfq, counteracting Hfq-mediated repression. Thus, Bläsi postulated that in addition to being the usual mediator of riboregulation, Hfq itself can be regulated by a noncoding RNA.

The subsequent talk by Eric Massé also put forward a model for direct mRNA repression by Hfq, albeit one in which sRNA acts as a recruitment factor. In the standard model of translational repression by Hfq and sRNA, the sRNA takes the driver’s seat by competing directly with 30S ribosomal subunits, while Hfq is thought to facilitate efficient sRNA-mRNA pairing. However, Massé now showed in vitro and in vivo results suggesting that the Hfq-dependent sRNA Spot42 binds too far upstream of the sdhC start codon to repress translation initiation directly. Instead, Spot42 acts to guide Hfq to an AU-rich mRNA region in the vicinity of the sdhC start codon. Upon binding there, Hfq competes directly with 30S binding. Thus, contrary to the canonical model of sRNA-mediated regulation, here translational repression by Hfq is assisted by an sRNA rather than vice versa.²⁴

Lionello Bossi focused on the mechanism by which the Hfq-dependent sRNA ChiX elicits transcriptional polarity in the chiPQ operon of Salmonella. His lab had previously shown that, in repressing chiP translation, ChiX induces Rho-dependent transcription termination at a site within the chiP region.²⁵ The efficiency of termination at this site was strongly dependent on the elongation regulator NusG in vitro and in vivo. In a comparative study of ChiX-mediated repression in Salmonella and E. coli, they found that NusG is required at terminators that deviate from the optimal Rho-binding architecture. This is a nice example of how the study of regulation by an sRNA can elucidate other regulatory processes, such as transcription termination.

Sarah Woodson addressed the question posed by the observation that although the proximal and distal faces of the Hfq hexamer specifically bind sRNA and mRNA targets, this co-localization alone does not explain how Hfq accelerates the formation and exchange of RNA base pairs. Her lab has used a variety of approaches to show that conserved arginines on the outer rim of the Hfq hexamer are required for the chaperone activity of Hfq. Mutation of these arginines impairs the ability of Hfq to act in sRNA regulation of rpoS translation and prevents annealing of natural sRNAs or unstructured oligonucleotides to mRNA targets without preventing sRNA or mRNA binding to Hfq. Woodson speculated that the arginines on the rim may permit rapid diffusion of RNA strands over the surface of Hfq, and this may enable Hfq to act as a chaperone that delivers sRNAs to structured RNA targets in vivo.²⁶

Cari Vanderpool focused on the physiological consequences of multiple target regulation by the sRNA SgrS, which has a central role in regulating the glucose-phosphate (GP) stress response in E. coli and Salmonella. SgrS promotes recovery from GP stress by base pairing with sugar transporter mRNAs to inhibit production of transporters responsible for uptake of the stress-inducing sugars and with a sugar phosphatase mRNA to increase phosphatase levels, allowing dephosphorylation and subsequent efflux of sugars.⁸ Recent studies revealed that the severity of GP stress and the critical targets of SgrS vary depending on the availability of nutrients. Growth inhibition is most severe when stress is induced in nutrient-poor media, and under these conditions, SgrS must regulate sugar transporter, phosphatase, and other as-yet-undefined target mRNAs to promote recovery. When stress is induced in nutrient-rich media, SgrS regulation of transporter mRNA alone is sufficient to promote full growth recovery. These observations imply that SgrS and perhaps other small RNAs are flexible regulators that can help cells respond to a variety of different stress conditions by controlling multigene regulons.²⁷

A new function of Hfq that may be independent of small RNAs was reported by Nils Schürgers from Annegret Wilde’s lab. Cyanobacteria harbor a special subfamily of Hfq that is structurally conserved but seems to have different RNA-binding preferences than other investigated members of this protein family.²⁸ Inactivation of the Hfq homolog in the cyanobacterium Synechocystis causes an aberrant accumulation of several transcripts and, more importantly, a loss of type IV pili.²⁹ Employing co-immunoprecipitation, the Wilde group has identified PilB1, the secretion ATPase that energizes pilus extension, as an interaction partner of Hfq. Fluorescence microscopy suggests that Hfq is localized at the cell periphery in a PilB1-dependent manner, and this localization was shown to be important for Hfq function in Synechocystis. Mutational analysis suggests that the critical residues of Synechocystis Hfq are on the proximal ‘outer ring’ and not in the structurally conserved proximal RNA binding pocket, arguing that Hfq may promote type IV pilus formation independent of its RNA binding property.³⁰

Continuing the theme of regulatory sRNAs, the Thursday morning session chaired by Shoshy Altuvia and Franz Narberhaus commenced with two talks on the carbon storage regulator/regulator of secondary metabolism (Csr/Rsm) system, in which an sRNA (CsrB/RsmZ) activates translation initiation by sequestering a homodimeric repressor protein (CsrA/RsmE) that would otherwise bind and mask the ribosome binding site of a subset of mRNAs.³¹ The sRNA contains several Shine-Dalgarno-like GGA binding motifs, typically in hairpin loops.

Frederic Allain’s group set out to understand the mechanism by which mRNAs and sRNAs compete for RsmE binding and the molecular basis for the wide range of protein-RNA binding affinities that are observed. In Pseudomonas fluorescens, they showed that several RsmE protein dimers assemble sequentially, specifically, and cooperatively onto the sRNA RsmZ. The solution structure of RsmZ bound to five RsmE proteins support models in which the first RsmE protein binds RsmZ by conformational selection and binding of the third RsmE protein dimer changes RsmZ from an RNase E-sensitive form to a protected form. These findings illustrate the molecular basis of translation activation by the sRNA RsmZ but also suggest how the RsmE proteins could ultimately be released from its RNA antagonist.^33,34

The Csr system in Yersinia pseudotuberculosis utilizes two sRNAs, CsrB and CsrC, to control the expression of two major virulence regulators.³⁴ Petra Dersch reported that upregulation of CsrB and CsrC leads to expression of RovA-dependent colonization factors during the early steps of infection. The Csr system also participates in the control of LcrF-dependent later-stage virulence factors, including a type III secretion system and Yop effectors important for evasion of the innate host defense. Translation of the lcrF transcripts is prevented by a FourU-type RNA thermometer in the 5′-UTR, which forms a structure comprising two stem-loops outside the host.³⁵ At 37 °C and upon host cell contact, LcrF translation is strongly induced. Thermal induction is mediated by melting of base pairs in the RNA thermometer, whereas activation of LcrF synthesis by host cell contact requires the global RNA-binding regulator CsrA.

Igor Ruiz de los Mozos introduced a novel concept of translational control. He reported evidence that around one-third of the mapped mRNAs of the major human pathogen Staphylococcus aureus have 3′-UTRs longer than 100 nucleotides and thus potential regulatory functions. For example, base pairing between the 3′-UTR and the Shine-Dalgarno region of icaR mRNA, which encodes the repressor of the main exopolysaccharidic compound of the S. aureus biofilm matrix, interferes with translation initiation and generates a double-stranded substrate for RNase III. Deletion or substitution of a motif (UCCCCUG) within the icaR 3′-UTR was sufficient to abolish this interaction and resulted in the accumulation of IcaR repressor and inhibition of biofilm development. These findings provide an example of a new post-transcriptional regulatory mechanism in which bacterial gene expression is modulated through the interaction of a 3′-UTR with the 5′-UTR of the same mRNA.³⁶

Noting that all regulatory RNAs known to act on translation target the mRNA, Norbert Polacek raised the fundamental question whether there are sRNAs that bind directly to the ribosome and regulate its translational activity. To address this question, his group analyzed the sRNA interactomes of ribosomes of prokaryotic and unicellular eukaryotic model organisms, revealing thousands of putative ribosome-associated noncoding RNAs.^37-39 In the archaeon Haloferax volcanii a tRNA-derived fragment (tRF) was found to target the small ribosomal subunit close to the decoding center upon alkaline stress in vitro and in vivo. As a consequence, this tRNA fragment reduces protein synthesis by interfering with crucial steps of translation such as mRNA binding and peptide bond synthesis. This tRF from H. volcanii was shown to interact and to function similarly on E. coli ribosomes, thus arguing for a conserved target site and mode of action.

Francesca Short from George Salmond’s group showed how an sRNA acts as an antitoxin to counteract a bacterial toxin. In the prototypic Type III toxin-antitoxin (TA) system, toxIN_Pa of Pectobacterium atrosepticum, a toxic ribonuclease (ToxN_Pa) is held in check by an RNA antitoxin (ToxI_Pa). ToxN_Pa cleaves the repetitive ToxI_Pa precursor RNA into individual pseudoknot units and stays bound to the products in an inactive, heterohexameric complex that has been examined by crystallography.⁴⁰ The ToxI_Pa RNA is a completely self-contained inhibitor that can counteract the RNase activity of ToxN_Pa in vitro without assistance from cellular factors or an exogenous energy source, and this inhibition is linked to the spontaneous self-assembly of the trimeric complex.⁴¹ ToxI inhibition is selective, as cross-inhibition was not observed with a second ToxIN system from Bacillus thuringiensis. In addition, ToxN_Pa and ToxN_Bt both have sequence-specific RNase activity that dictates the correct processing of their respective antitoxins. The results show how a small, processed RNA can act as a highly selective and robust inhibitor of a potentially lethal protein.

RNA chaperones were the subject of Eliora Ron’s presentation. Several members of the cold shock family can modulate RNA stability. Two of them, CspC and CspE, have significant roles during stress conditions such as the stationary/stress response and the heat shock response, where they are involved in modulating the stability of the various stress gene transcripts. These two RNA chaperones are very similar, yet they differ in their function and regulation. The cellular levels of CspC decrease during heat shock and increase during stationary/stress response, resulting in a decrease or increase (respectively) in stability of the stress-specific transcripts. It therefore appears that CspC levels modulate transcript stability upon exposure to environmental stress while CspE–whose levels are stable–acts as a “housekeeping RNA chaperone” under general stress conditions.^42,43

Following lunch and poster presentations, the afternoon session on regulating transcription machinery was chaired by Sabine Brantl. The first two talks dealt with transcriptional control by 6S RNA that binds to the housekeeping form of RNA polymerase holoenzyme (EcEσ⁷⁰ in E. coli, BsEσ^A in B. subtilis) and downregulates transcription at a subset of promoters.

Roland Hartmann put his focus on the structure and function of the second 6S RNA paralog in B. subtilis, 6S-2 RNA.⁴⁴ Comparative studies of 6S-1 and 6S-2 RNA suggest that both bind to BsEσ^A with similar affinities, serve as templates for product RNA (pRNA) synthesis, and inhibit transcription from genomic promoters. However, pRNAs of equal length dissociate faster from 6S-2 than 6S-1 RNA because 6S-2 pRNAs have a higher A/U content. This may make the escape of RNA polymerase from the idling cycle of abortive pRNA transcription less efficient in the case of 6S-2 RNA. It is unclear whether 6S-2 RNA can dissociate from RNAP by pRNA synthesis in vivo. The possibility that 6S-1 RNA my displace 6S-2 RNA from RNAP in the absence of 6S-2 pRNA synthesis could not be experimentally supported.

Karen Wassarman reported that the identity of the initiating nucleotide (iNTP) determines the efficiency of pRNA synthesis: iGTP (e.g., for Bs6S-1) is much more efficient than iATP (e.g., for Bs6S-2). In contrast, EcEσ⁷⁰ does not have a preference for the initiating nucleotide in pRNA synthesis.⁴⁵ Her group also found that cells lacking 6S-1 RNA sporulate earlier than wild type cells or cells lacking 6S-2 RNA, indicating a role of 6S-1 RNA for appropriate timing of sporulation.⁴⁶ Several pathways known to be important for the timing of sporulation are not required for this 6S-1-dependent phenotype. Therefore, they hypothesize that B. subtilis cells lacking 6S-1 RNA utilize nutrients differently, leading to an earlier reduction of the nutrients in their environment, and thus early sporulation. These results are reminiscent of the role of E. coli 6S RNA in survival during long-term nutrient deprivation.

Thursday session closed with a session chaired by John van der Oost and featuring three presentations covering aspects of prokaryotic protection against foreign nucleic acids, including the CRISPR-Cas systems, which are small RNA-based immune systems that control invasion by viruses and plasmids in bacteria and archaea. Multiple variations of the CRISPR-Cas pathway are found among prokaryotes, each mediated by largely distinct components and mechanisms that we are only beginning to delineate.

In the first talk, Michael Terns discussed RNA and DNA targeting of CRISPR-Cas systems in the hyperthermophile archaeon Pyrococcus furiosus. Distributed throughout the genome of this organism are seven transcriptionally active CRISPR loci that together encode a total of ~200 CRISPR RNAs (crRNAs) and 27 Cas proteins in 3 distinct pathways. The Cas6 protein “dices” large precursor CRISPR transcripts to generate the individual invader-targeting crRNAs. The crRNAs are integrated into three distinct crRNA-Cas protein complexes that likely function as immune effector complexes. The complex formed by the Cmr (type IIIB) system proteins cleaves complementary target RNAs and can be programmed to cleave novel RNA targets in an RNAi-like manner. The other two CRISPR-Cas systems in P. furiosus, Csa (type IA) and Cst (type IB), target invaders at the DNA level. Altogether, the analysis of P. furiosus has uncovered an arsenal of multiple RNA-guided mechanisms to resist diverse invaders.^47,48

Ekaterina Semenova from Konstantin Severinov’s lab reported a new concept in CRISPR-Cas biology: primed spacer acquisition. They have developed experimental systems to study CRISPR-mediated immunity (also called ‘CRISPR interference’) and spacer acquisition in Escherichia coli cells during bacteriophage M13 infection or plasmid transformation.⁴⁹ Using these systems, they showed that efficient spacer acquisition in E. coli requires the prior existence of a CRISPR spacer matching the foreign genome, a phenomenon referred to as ‘priming’. Point mutations in target DNA prevent CRISPR-mediated immunity; however, the interaction of CRISPR RNA with the mutated target leads to very robust acquisition of additional spacers from this DNA, thus greatly stimulating the immunity of the host.⁵⁰

In the closing talk on Thursday, Daan Swarts from the Van der Oost group presented their investigation of the Argonaute homolog of Thermus thermophilus.^51,52 Argonautes have been iconic proteins in the rapidly expanding field of RNA biology and have emerged as key effectors in virtually all eukaryotic small RNA-mediated gene silencing pathways. Central to their activities is their association with small guide RNAs for recognition and in some cases also cleavage of cellular transcripts. In other words, the dogma has been that Argonautes are RNA-binding proteins that use RNA guides to target cellular RNA. By contrast, the new work in Thermus thermophilus revealed that the Argonaute of this organism uses small guide DNAs to cleave double-strand DNA. Physiologically, this bacterial Argonaute was shown to restrict the invasion and propagation of plasmids, suggesting that it may function as a general surveillance system to protect its host bacterium from parasitic DNA. Intriguingly, since bacterial Argonautes likely predated their eukaryotic counterparts in evolution, genome defense on the DNA level may constitute an ancient function of Argonautes prior to their recruitment to RNA-based gene silencing.⁵³

The first session on Friday morning was chaired by Harald Putzer and focused on control of RNA stability. Joel Belasco kicked off this session with a talk on the specificity of RNA degradation by the pyrophosphate-removing hydrolase RppH. Bacterial RNAs often decay by a 5′-end-dependent mechanism in which degradation is triggered by conversion of the 5′ terminus from a triphosphate to a monophosphate by RppH. In E. coli, this modification creates better substrates for the endonuclease RNase E, whose cleavage activity is enhanced when the RNA 5′ end is monophosphorylated; in Bacillus subtilis, it enables RNase J to degrade the transcript exonucleolytically. His lab has characterized the substrate specificity of E. coli RppH (EcRppH) and B. subtilis RppH (BsRppH). Both require at least two unpaired nucleotides at the RNA 5′ end, explaining the stabilizing influence of 5′-terminal base pairing on mRNA lifetimes in vivo. However, whereas EcRppH tolerates a variety of 5′-terminal sequences, BsRppH requires guanylate at the second position of its substrates. Interestingly, B. subtilis appears to contain a second, as yet unidentified RNA pyrophosphohydrolase whose sequence specificity is more relaxed than that of BsRppH.⁵⁴

Ciarán Condon continued the discussion of pathways for mRNA and regulatory RNA turnover in B. subtilis, presenting the crystal structure of BsRppH bound to a 5′ triphosphorylated dinucleotide.⁵⁵ The structure shows that BsRppH specifically recognizes a guanine base at the second position of its substrates; it can thus act only on a subset of cellular RNAs (at most 14%), consistent with the idea that B. subtilis has at least one other enzyme capable of deprotecting primary transcripts. In the second part of his talk, he focused on endonucleolytic initiation of mRNA decay in B. subtilis, whose ribonucleolytic machinery is distinct from that of E. coli. Analyzing B. subtilis cells depleted for RNase Y, III, or J1 allowed them to determine that RNase Y and J1 are involved in the degradation of over half of B. subtilis mRNAs and regulatory RNAs. In addition, their work revealed why RNase III is essential in this species: it represses the synthesis of toxins TxpA and YonT, which are expressed by the resident prophages Skin and SPβ, respectively.⁵⁶

Hilde de Reuse presented her findings on a minimal RNA-degradosome in H. pylori, whose repertoire of ribonucleases is both a restricted and unexpected. Among these, the essential 5′ exo- and endoribonuclease RNase J was shown to be involved in the regulation of the urease operon. Tandem affinity purification suggested that RNase J forms an RNA degradosome with the sole DExD-box RNA helicase of H. pylori, RhpA, and the functionality of these interactions was validated with purified components in vitro. Interestingly, this RNA degradosome co-localizes with translating ribosomes and not with individual ribosomal subunits,⁵⁷ suggesting that translation and mRNA degradation are coupled.

The next two talks discussed the roles of RNases in the turnover of small regulatory RNAs in bacteria. Cecilia Arraiano reported that, in E. coli cells lacking Hfq, polynucleotide phosphorylase (PNPase) is the main enzyme involved in the degradation of many small RNAs, especially during stationary phase.⁵⁸ These results contributed to the understanding of the turnover of small RNAs which do not bind to Hfq. She also described evidence that, in Salmonella, RNase III cleaves duplexes that the sRNA MicA forms with its targets.^59,60

Boris Görke reported the discovery of a novel adaptor protein for RNase E-mediated decay of the conserved GlmY and GlmZ sRNAs in Eschericha coli. GlmY and GlmZ are homologous sRNAs that have been known to act jointly by a feedback mechanism to control the synthesis of glucosamine-6-phosphate synthase GlmS, a key enzyme required for synthesis of the cell envelope.⁶¹ Only GlmZ is able to activate glmS mRNA by base-pairing. GlmZ is subject to endonucleolytic processing by RNase E, which removes the base-pairing nucleotides and inactivates the sRNA. GlmY, which accumulates when glucosamine-6-phosphate is scarce, acts indirectly by counteracting cleavage of GlmZ. New work has identified the protein RapZ as a factor essential for GlmZ processing. RapZ recruits RNase E to GlmZ by binding to both the central stem loop of GlmZ and the catalytic domain of RNase E. GlmY, which is similar to GlmZ in sequence and structure, can inhibit this process by sequestering RapZ. Therefore, GlmY is a decoy sRNA that acts by molecular mimicry to impede the targeted decay of GlmZ.⁶²

The RNA stability session finished with a talk by Murray Deutscher discussing the dual roles of the tmRNA-SmpB system in RNase R instability and trans-translation. RNase R is a very short-lived protein in exponential-phase E. coli cells due to acetylation of Lys 544. The acetylation stimulates binding of tmRNA-SmpB to the C-terminal region of the protein, which in turn stabilizes binding of proteases to the N-terminal region, thereby initiating proteolysis. tmRNA-SmpB also is required to bind RNase R to ribosomes. Deutscher showed that binding to ribosomes completely stabilizes RNase R. Binding to ribosomes can be disrupted if RNase R is not acetylated, if either tmRNA or SmpB is absent, or if a mutant form of tmRNA that cannot bind ribosomes is present. Only the free form of this nuclease is subject to proteolysis, and free RNase R is extremely deleterious to cells. When RNase R is unable to bind to ribosomes, RNA degradation increases as much as 20-fold, explaining why cells sequester the enzyme on ribosomes and rapidly turn over the free protein.^63,64

The afternoon session on RNA deep sequencing and bioinformatics, which was jointly chaired by Eric Westhof and Rolf Backofen, opened with a talk by Rotem Sorek on how comparative bacterial transcriptomics has revealed conserved and diverged antisense RNAs. Antisense RNAs, overlapping protein-coding genes in cis, are abundant in the transcriptomes of bacteria and archaea, but their functional roles are currently unclear. Specifically, it has been suggested that much of the antisense RNA observed in bacteria represents transcriptional noise. A comparative transcriptomics approach uncovered conserved antisense RNAs that are likely to be functional. Application of this approach to the genus Listeria exposed a conserved subset of long antisense transcripts, which were termed 'excludons'⁶⁵. The excludon is a genomic locus encoding an unusually long antisense RNA that spans divergent genes or operons having related or opposing functions. These antisense RNAs can inhibit the expression of one operon while functioning as an mRNA for the adjacent operon, and hence may act as regulatory switches for fine tuning. Another topic presented was the discovery of archaeal mRNAs with 5-methylcytosine base modifications, which might have a post-transcriptional regulatory function.⁶⁶

Brian Tjaden introduced two computational methods for the analysis of the large and growing transcriptome data sets generated by deep-sequencing technology. The first, Rockhopper, is a comprehensive and user-friendly system for computational analysis of bacterial RNA-seq data.⁶⁷ Rockhopper supports the various stages of analysis of bacterial RNA-seq data, including aligning sequencing reads to a bacterial genome, normalizing data from different experiments, assembling transcripts, discovering novel regulatory RNAs, quantifying transcript abundance, and testing for differential expression between experiments. The second system, TargetRNA2, which builds on the popular TargetRNA web server, predicts targets of regulatory RNA action and, optionally, can incorporate RNA-seq data to more accurately identify mRNA targets of trans-acting regulatory RNAs.⁶⁸ TargetRNA2 dramatically reduces the rate of false-positive predictions when compared with other approaches for predicting targets of regulatory RNA action.

Even larger data sets are acquired when gene expression changes of both partners in host-pathogen interactions are examined by RNA sequencing. Alex Westermann introduced the ‘Dual RNA-seq’ approach⁶⁹ and demonstrated its feasibility by using a model of human cell lines and the facultative intracellular model pathogen Salmonella Typhimurium. Application of this technique led to the detection of all major (coding and noncoding) transcript classes of an infected cell. In the pathogen, one of the most highly induced genes upon host cell entry was a previously uncharacterized small noncoding RNA. Classical genetics and biochemical assays placed this sRNA at the center of a complex regulatory network of virulence gene expression, and Dual RNA-seq revealed the effects of this sRNA on the host immune response. Dual RNA-seq may become an important technique to identify roles of both coding and noncoding RNAs in host-pathogen interactions.⁷⁰

Modern high-throughput sequencing technology enables sRNA discovery at an ever faster rate. Determining their cellular targets and regulatory functions, however, remains a challenging and time-consuming task. Since most sRNAs bind to mRNA targets by means of base-pairing interactions, targets can be predicted by using computational methods based on thermodynamics. Ivo Hofacker presented a newly developed algorithm to model the efficiency of translation initiation in the presence of sRNAs.⁷⁰ This algorithm correctly identifies most known instances where sRNA binding results in upregulation of gene expression. Filtering candidate sRNA targets by their predicted effect on translation yield significantly improves the specificity of sRNA target prediction and thus can reduce the burden of experimental validation.

Jens Georg showed that bioinformatic target prediction for conserved sRNAs can be significantly improved by a comparative approach. The web tool CopraRNA⁷¹ is based on the assumption that an sRNA is evolutionarily passed from a common ancestor to daughter species together with a core target set. CopraRNA statistically combines predictions for a conserved sRNA from multiple organisms, resulting in enhanced specificity and sensitivity compared with existing bioinformatics target prediction tools. The false-positive rate is further reduced by a post-processing of the prediction results by functional enrichment and network analysis. In a benchmark test with 18 well-investigated enterobacterial sRNAs, CopraRNA was found superior to existing prediction algorithms and competitive with target prediction by sRNA pulse expression microarray experiments. CopraRNA precisely predicted the interaction regions in sRNAs and the known biological functions of several sRNAs. Furthermore, it predicted mRNAs that are targeted by multiple sRNAs and thus may constitute previously undetected regulatory hubs.

The evolutionary patterns of regulatory sRNAs and their targets was the topic of a presentation by Asaf Peer from Hanah Margalit’s group. The sRNAs of Escherichia coli appeared late in evolution, compared with cis-regulatory RNA elements such as riboswitches and attenuators, and the evolution of sRNAs in the Enterobacteriales was greatly accelerated when they split from the rest of the Gammaproteobacteria. By comparing the evolution of each sRNA with that of its mRNA binding sites, they found several examples of binding sites that appeared before or at about the same time as the sRNA that binds there, explaining the selective pressure to maintain these sRNAs in the genome.⁷²

The final session of the meeting, which on Saturday morning was chaired by Wade Winkler and Jens Wöhnert, was devoted to riboswitches and RNA structure. In bacteria, many regulatory circuits involve evolutionarily conserved mRNA regions, termed riboswitches, that specifically interact with various cellular metabolites and control expression of metabolite-associated genes.⁷³

To explain how riboswitches achieve their exquisite ligand selectivity and trigger a genetic response, Alexander Serganov described the three-dimensional structure of the riboswitch specific for coenzyme adenosylcobalamin, a derivative of the vitamin B₁₂.⁷⁴ This riboswitch adopts intricate three-dimensional architecture composed of a ligand-bound multihelical junction stabilized by long-distance tertiary interactions. The structure explains previous genetic and biochemical results and reveals how this riboswitch achieves selective, high-affinity binding of its cognate ligand. He also reported the X-ray crystal structure of E. coli RppH bearing a trinucleotide ligand. This RNA pyrophosphohydrolase utilizes a bipartite RNA-recognition mechanism to discriminate against mononucleotides and has a binding cleft for the second RNA nucleotide that is structurally distinct from that of B. subtilis RppH, likely explaining the greater promiscuity of the E. coli enzyme.

Harald Schwalbe reported on an interesting new twist to RNA-based regulation. He presented insights into the coupling of metabolite binding or base-pair melting with allosteric structural transitions in riboswitch RNA. In the commonly accepted model, riboswitches function by a ligand-dependent conformational switch between two mutually exclusive states. However, many riboswitch properties cannot be explained by a pure two-state mechanism. Indeed, the regulatory mechanism of the adenine-sensing riboswitch from the Vibrio vulnificus add gene involves three distinct stable conformations.⁷⁵ The temperature and Mg²⁺ dependence of the population ratios of the three conformations and the kinetics of their interconversion at nucleotide resolution were determined. The observed temperature dependence of a pre-equilibrium involving two structurally distinct ligand-free conformations of the add riboswitch conferred efficient regulation over a physiologically relevant temperature range. Such robust switching is a key requirement for gene regulation in bacteria that have to adapt to environments with varying temperatures. This adenine-sensing riboswitch represents the first example of a temperature-compensated regulatory RNA element.

Peter Stadler described his collaboration with Mario Mörl on the de novo design of a synthetic riboswitch that regulates transcription termination. They developed an in silico pipeline for the rational design of synthetic riboswitches that regulate gene expression at the transcriptional level. Using the well-characterized theophylline aptamer as sensor, they designed actuator components comprising RNA sequences that can fold into intrinsic terminator structures. Several of the designed constructs were effective at controlling gene expression in a ligand-dependent manner in E. coli, demonstrating that it is possible to engineer riboswitches that regulate transcription.⁷⁶

J.R. Mellin from Pascale Cossart’s lab described a riboswitch-regulated antisense RNA. In Listeria monocytogenes, a vitamin B₁₂-binding (B₁₂) riboswitch was identified, downstream and antisense to the adjacent gene, pocR, suggesting that it might regulate pocR in a non-classical manner. The B₁₂ riboswitch is transcribed as part of a non-coding antisense RNA named AspocR. Evidence from other species indicates that PocR is a transcription factor that is activated by 1,2-propanediol and subsequently activates expression of the pdu genes, which mediate propanediol catabolism. As enzymes involved in propanediol catabolism require B₁₂ as a cofactor, it was hypothesized that the Listeria B₁₂ riboswitch might be involved in pocR regulation. In the presence of B₁₂, the riboswitch was found to induce premature termination of aspocR transcription, resulting in a short transcript. In the absence of B₁₂, aspocR is transcribed to produce a long antisense RNA that inhibits pocR expression. Regulation by AspocR ensures that pocR, and consequently the pdu genes, are maximally expressed only when both propanediol and B₁₂ are present. AspocR can inhibit pocR expression in trans, suggesting that it acts through a base-pairing interaction with pocR mRNA. These findings extend the classical definition of riboswitches from elements governing solely the expression of mRNAs to a wider role that also encompasses non-coding RNAs.⁷⁷

Laurene Bastet from the groups of Daniel Lafontaine and Eric Massé described a dual-acting riboswitch from Escherichia coli that, in addition to modulating translation initiation, also is directly involved in controlling the initiation of mRNA decay. Upon lysine binding, the lysC riboswitch adopts a conformation that both inhibits translation initiation and exposes RNase E cleavage sites located in the riboswitch expression platform. However, in the absence of lysine, the riboswitch folds into an alternative conformation that simultaneously allows translation initiation and sequesters the RNase E cleavage sites. Thus, translation initiation and mRNA decay can be modulated independently by using the same conformational switch. Such dual inhibition is in contrast to other riboswitches, such as the thiamin pyrophosphate-sensing thiM riboswitch, which triggers mRNA decay only as a secondary consequence of translation inhibition. Riboswitch control of RNase E cleavage is a mechanism by which metabolite sensing can be used to regulate gene expression in response to environmental changes.⁷⁸

In the final presentation of the meeting, Danielle Garsin, speaking on behalf of her group and her collaborator Wade Winkler, introduced protein sequestration as a new mechanism of riboswitch action. Adenosylcobalamin (AdoCbl) is required for the metabolism of ethanolamine (EA) in Enterococcus faecalis, both as an enzyme co-factor and for the induction of the EA utilization (eut) genes. It acts through an AdoCbl-binding riboswitch to induce the eut genes, but the mechanism of control is incompletely understood. Gene expression also requires EA to activate a two-component system composed of the sensor kinase EutW and its cognate response regulator EutV by phosphorylation.⁷⁹ Active EutV is an antiterminator that binds nascent transcripts by recognizing a dual-hairpin ligand and preventing terminator formation.⁸⁰ They found that the AdoCbl-binding riboswitch is part of a small, trans-acting RNA. In the absence of AdoCbl, the sRNA imprisons EutV in an inactive complex. When AdoCbl is present, its binding to the riboswitch prevents complex formation between EutV and the sRNA, and EutV is free to induce gene expression.

The 45 oral presentations were complemented by 170 posters of exceptionally high quality. Lively discussions after the talks and at the posters were a clear indication that this is a dynamic research field that fascinates and attracts many junior scientists. Social events included dinner aboard a riverboat on Wednesday night and a conference dinner in the wine cellars of the UNESCO World Heritage site ‘Residenz Würzburg’ on Friday night.

Thanks to the continued enthusiasm for a regular, bi-annual meeting on this topic, Gisela Storz, Cari Vanderpool, John van der Oost, and Gerhart Wagner have graciously volunteered to serve as organizers for the next such conference. A place and date have meanwhile been chosen: we will meet again in 2015 on July 9–12, this time in Punta Cana in the Dominican Republic. We look forward to seeing everyone again in 2015!

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

This conference was supported by the German Research Foundation (DFG, grant NA 240/6–2).

10.4161/rna.29533