This page has been archived and is no longer updated
Gene regulation by riboswitches
Author: Maumita Mandal
Keywords
Keywords for this Article
Add keywords to your Content
Save
|
Cancel
Share
|
Cancel
Revoke
|
Cancel
Rate & Certify
Rate Me...
Rate Me
!
Comment
Save
|
Cancel
Flag Inappropriate
The Content is
Objectionable
Explicit
Offensive
Inaccurate
Comment
Flag Content
|
Cancel
Delete Content
Reason
Delete
|
Cancel
Close
Full Screen
"Expression of the numerous genes in a cell must be tai- lored to provide the appropriate levels of RNA and protein production at all times. Gene-control systems face an enormous challenge. They must coordinate numerous tasks that a typical cell carries out on an ever-changing cycle, and they must interpret many dif- ferent chemical and physical signals. Even the simplest, single-celled organisms need to modulate the expres- sion of hundreds of genes in response to a myriad of cellular needs and environmental cues. Gene-control systems, therefore, must have the ability to respond precisely to specific signals, rapidly bring about their intended genetic effect, and have sufficient dynamic character to fine-tune the level of expression for hun- dreds of different genes. It has long been recognized that organisms make extensive use of protein-based control systems to mod- ulate gene expression. The networks of protein sig- nalling and gene-control factors are most complex in eukaryotes, where numerous factors typically work together to influence transcription, translation, mRNA processing/degradation and other mechanisms that control the levels of gene products in a cell 1 .In addi- tion, there is an emerging awareness of the role of RNA factors in gene control, which are of fundamental importance to such processes as developmental timing and other gene-silencing events 2,3 .A variety of microRNAs (miRNAs) and related short-interfering RNAs (siRNAs) function by a series of protein-medi- ated processing events that eventually lead to the pro- duction of ~22-nucleotide-long fragments that form base-paired complexes with their target mRNAs. These interactions cause inactivation of the targeted gene by subsequent nuclease processing or by other non-nucle- olytic mechanisms 4,5 .Although protein factors are required for these gene-silencing mechanisms to oper- ate, they highlight the fact that many organisms rely on RNA molecules for critical regulatory tasks, and sug- gest that additional roles for RNA in gene-control sys- tems might be possible. Recent studies have demonstrated that the involve- ment of RNA in fundamental gene-control processes is indeed more extensive. Numerous mRNAs in prokary- otes carry complex folded domains, which are known as riboswitches 6,7 ,within the non-coding portions of their polynucleotide chains. Each riboswitch directly binds a specific metabolite, without the obligate involvement of a protein factor, and then controls gene expression by harnessing changes in RNA structure to influence transcription elongation, translation initia- tion, or other aspects of the process that leads to pro- tein production 7?12 .Here,we provide an overview of riboswitch mechanisms and we discuss the evidence which indicates that an expanded set of RNA switches might be of fundamental importance to the control of genes in many modern biological systems. Allosteric RNAs and genetic switches Roles for metabolite-binding RNAs in modern cells could not easily have been predicted from the earliest studies on protein enzymes and receptors. Polypeptide chains made of the standard 20 amino acids are an excellent medium for forming complex molecular architectures that are capable of serving as receptors or GENE REGULATION BY RIBOSWITCHES Maumita Mandal and Ronald R. Breaker Riboswitches are complex folded RNA domains that serve as receptors for specific metabolites. These domains are found in the non-coding portions of various mRNAs, where they control gene expression by harnessing allosteric structural changes that are brought about by metabolite binding. New findings indicate that riboswitches are robust genetic elements that are involved in regulating fundamental metabolic processes in many organisms. NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 5 | JUNE 2004 | 451 Department of Molecular, Cellular and Developmental Biology, Yale University, P. O. Box 208103, New Haven, Connecticut 06520-8103, USA. Correspondence to R.R.B. e-mail: ronald.breaker@yale.edu doi:10.1038/nrm1403 REVIEWS � 2004 Nature Publishing Group ALLOSTERIC ENZYME An enzyme that is triggered to alter its function in response to the binding of a target compound at a site that is distal from the active site of the enzyme. RNA WORLD A hypothetical time in early evolution, before the emergence of DNA and proteins, when biological processes were guided entirely by RNA molecules. RIBOZYME A nucleic-acid molecule that folds to form an active site and catalyzes a chemical reaction. APTAMER An RNA domain, either engineered or natural, that forms a precise three-dimensional structure and selectively binds a target molecule. IN VITRO EVOLUTION The use of various separation and amplification techniques that serve to mimic Darwinian evolution and create variants of proteins or nucleic acids that have new or improved functions. UTR (Untranslated region). Stretches of untranslated sequences located upstream and downstream of the coding region of an mRNA. IN-LINE PROBING An RNA-structure probing method that can be used to examine secondary-structure models and to determine whether RNAs undergo substantive structural rearrangements under different incubation conditions. 452 | JUNE 2004 | VOLUME 5 www.nature.com/reviews/molcellbio REVIEWS appended onto catalytic RNAs to create allosteric ribozymes, wherein ligand binding to an aptamer domain triggers a catalytic response in an adjoining ribozyme 24?28 .These findings strongly indicate that RNA molecules have the structural and functional sophistica- tion needed to function as natural genetic switches in ancient or even modern settings. This led a number of researchers to speculate, mostly in unpublished forums, that modern organisms might exploit ligand-binding RNAs for this very role 29?34 . Riboswitch discoveries Beginning with the discovery of the lac repressor, an enormous body of data emerged that described a diver- sity of protein genetic factors that respond to various metabolites and signalling compounds. However, a number of puzzling instances of gene-control phenom- ena had been accumulating in the scientific literature for many years. Particularly intriguing to us was a series of reports that described the unique genetic-control char- acteristics of the btuB gene of Escherichia coli and the cob operon of Salmonella typhimurium.These genes are responsible for maintaining adequate levels of coen- zyme B 12 in the cell either by importing or synthesizing this complex metabolite. It was known for some time that coenzyme B 12 is involved in some form of feedback inhibition to control the expression of these genes, and that the 5?-untrans- lated region (5?-UTR) of each mRNA carries a conserved sequence element that is required for this control 35?37 . However, repeated attempts to identify a protein factor that would convert a chemical signal (increasing coen- zyme-B 12 concentrations) into a genetic response had been unsuccessful 32,38 .Since a protein factor could not be implicated, it seemed likely that a direct interaction between the coenzyme and the mRNA was being made. Various attempts to show that the RNA bound the coenzyme directly, in the absence of proteins, were also unsuccessful 32,36 and this hypothesis remained unproven. We chose to examine the question of direct binding between the metabolite and the btuB RNA of E. coli using several approaches. First, we reasoned that nat- ural metabolite-binding mRNAs would undergo some degree of structural reorganization on ligand binding, much like that which is observed with engi- neered allosteric RNAs. To search for metabolite- induced allosteric changes, we used a method known as IN-LINE PROBING,which takes advantage of the fact that the spontaneous rate of RNA cleavage is highly depen- dent on the local structure at each internucleotide linkage (BOX 2) 39?41 .For example, a 315-nucleotide sec- tion of the btuB 5?-UTR has numerous changes in its pattern of spontaneous cleavage when coenzyme B 12 is present, which indicates that the RNA undergoes a substantial level of allosteric reorganization 6 (FIG. 1). Similar results were found when in-line probing was conducted with 5?-UTR sections from the btuB and cob mRNAs of S. typhimurium 6,42 . Further examination of the RNAs using equilibrium dialysis, mutational analyses and gene-expression studies as biocatalysts that bind to the myriad of compounds in living systems. Furthermore, ALLOSTERIC ENZYMES that change shape and function in response to ligand bind- ing have been known since the 1960s, and one could easily imagine how protein genetic switches might be made 13?15 .DNA and RNA seemed well suited for their roles in storing and transferring genetic information, whereas proteins were viewed as the molecules of bio- chemical action. Therefore, the discovery of the first example of a protein genetic factor (the lac repressor) that responded to metabolites or other macromolecules, also in the early 1960s (REF. 16), fitted nicely with the idea that the division of labour among the biopolymers was clearly defined. There was little need to invoke the existence of riboswitches when allosteric modulation of protein enzymes and protein genetic factors seemed sufficient to eventually satisfy most gene-control needs in even the most complex organisms. However, two questions remained to be addressed before RNA could be dis- missed as a possible metabolite-sensing genetic element. First, how might early life forms have controlled molec- ular functions before the emergence of proteins? It is plausible that life passed through an ?RNA WORLD?, i n which genetic storage and primitive biological processes were thought to be carried out entirely by RNA 17,18 . Support for this theory was delivered by the discovery of RIBOZYMES 19,20 ,which proved that RNA could form com- plex shapes and catalyze reactions of biological impor- tance. If there were complex organisms in the RNA world, and particularly as such organisms began the transition into the ?DNA?RNA?protein world? of mod- ern times, then it is likely that RNA switches would have been necessary to exert control over these biological processes. Second, does RNA even have sufficient structural potential to serve as a highly specific receptor for small organic compounds? Indeed, numerous RNAs or APTAMERS 21?23 (BOX 1) that bind to a diverse collection of targets are being generated in laboratories by using IN VITRO EVOLUTION.These engineered receptors can be Box 1 | The ligand-binding potential of RNA aptamers Numerous RNA aptamers have been created by using in-vitro-evolution methods 21?23 . Typically, pools of relatively short random-sequence nucleic acids are subjected to a selection and amplification process whereby molecules that bind to a particular target are selectively amplified. A large number of aptamers have been created for protein and small-molecule targets 92 ,which indicates that RNA has the structural sophistication that is needed to serve as a sensor element. One of the first demonstrations of the precise molecular-recognition capability of RNA was with a theophylline-dependent aptamer, which discriminates against caffeine by more than 10,000 fold (REF. 93).Caffeine differs from theophylline by only a single methyl group. In several instances, there are both engineered and natural aptamers for the same compound. For example, guanine-binding 94 and FMN-binding 95 aptamers have been isolated using in vitro evolution. In each instance, the engineered aptamers are smaller and carry fewer essential nucleotides and structural elements than their natural counterparts. Furthermore, the engineered aptamers are less able to discriminate against closely related analogues, and have poorer binding affinities for their respective targets. These observations imply that there is a higher demand for specificity and affinity with natural aptamers, which is achieved by the formation of more complex motifs. � 2004 Nature Publishing Group NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 5 | JUNE 2004 | 453 REVIEWS response to changing metabolite concentrations. For any riboswitch candidate, the challenge for researchers is to provide convincing evidence that the RNA serves as the molecular recognition element for the metabolite in the absence of any possible protein factor. In addition, evidence for a mechanism of gene control is needed to justify its classification as a riboswitch. This level of proofofthe existence of TPP- and FMN-dependent riboswitches 7,57,58 was generated almost simultaneously with that for the coenzyme-B 12 riboswitch. Some of the structural and functional characteristics of riboswitches revealed by these studies are outlined below. Anatomy of a riboswitch In prokaryotes, most riboswitches are located in the 5?-UTRs of mRNAs, and are typically composed of two functional domains 10 (FIG. 2).The first structure that is encountered as the nascent mRNA emerges from the RNA polymerase during transcription is the aptamer domain. Like engineered aptamers, each natural aptamer serves as a molecular sensor embedded within the riboswitch, where it selectively recognizes its corre- sponding target molecule within the complex sea of other metabolites. The aptamer takes the place of a conventional protein factor that, otherwise, would serve as the sensory element. Like many protein factors, the aptamer must selectively recognize a metabolite with the appropriate affinity. Unlike protein, with 20 confirmed that direct binding of coenzyme B 12 by mRNAs does occur and that this interaction is required for proper expression. The btuB riboswitch from E. coli binds coenzyme B 12 with an apparent DISSOCIATION CONSTANT (K D ) of ~300 nM, but rejects most other chemical analogues of the coenzyme that were exam- ined 6 .The extraordinary light sensitivity of coenzyme B 12 (it has a half life of seconds in ambient light) 6,43 resulted in technical challenges for these experiments, and might also have caused difficulties in previous attempts to prove direct binding. Nevertheless, coen- zyme-B 12 -dependent riboswitches are now recognized as a widespread genetic-control element in prokaryotes, as bioinformatics approaches have identified more than 200 variants of this element in diverse groups of bacteria 42,44 .Furthermore, the resulting phylogenetic alignments of riboswitch variants are particularly useful when working to establish a model for the secondary structure of these complex folded RNAs. There was also evidence in the literature for the metabolite-dependent control of several other genes, but in each case no protein factors had been identified. The list of metabolites included thiamin pyrophos- phate (TPP) 34,45 , flavin mononucleotide (FMN) 31,46?48 , S-adenosylmethionine (SAM) 49 , guanine 50,51 and lysine 52?56 .In each instance, these early reports have recently proven to correspond to known riboswitches that control the expression of adjoining genes in DISSOCIATION CONSTANT The equilibrium constant for a ligand binding to its receptor, which, in the case of riboswitches, represents the concentration of ligand that is required to convert half of the aptamers that are present in a mixture to their ligand-bound form. Box 2 | RNA analysis by in-line probing The dominant pathway by which RNA degrades involves an ?in-line? nucleophilic attack by the 2? oxygen on the adjacent phosphorus centre. Efficient cleavage takes place only if the attacking 2? oxygen, the phosphorus and the departing 5? oxygen of the phosphodiester linkage approach a linear configuration, hence the term ?in-line? (see figure, part a;red dashed line). The speed at which spontaneous cleavage occurs 40 depends on the local structural context in which each RNA linkage exists 39,41 .Linkages that reside in highly structured regions of a folded RNA (for example, a base-paired helix) typically resist cleavage because the relevant atoms are not held in an in-line configuration. However, if folding contacts do not restrict their structure, linkages occasionally take on in-line geometry through random motions and therefore are subject to an enhanced rate of spontaneous cleavage. In some instances, linkages within parts of a folded RNA molecule that have taken on a tertiary structure can have accelerated spontaneous cleavage because a linkage is fortuitously held in an optimal configuration for a nucleophilic attack 39 .In part b of the figure, a radiolabelled RNA that is incubated in the absence of ligand (left) gives a distinctive pattern of spontaneous-cleavage products on separation by gel electrophoresis. This same RNA construct has an altered pattern of cleavage products (red bracket) when incubated with ligand X, which indicates that ligand binding alters the folding of these nucleotides. ba 5? 32 P 5? 32 P Ligand added X No ligand 3?3? G G C A G U C G A C A C G G C A G U C G A C A C Base O OHO P O O ? O O Base O OHO Base O OO P O ? O ? O O Base O OHO Base O OO HO O Base O OHO P ? OO 3? 2? 5? � 2004 Nature Publishing Group 454 | JUNE 2004 | VOLUME 5 www.nature.com/reviews/molcellbio REVIEWS The only similarity among these expression plat- forms and their varied mechanisms is that they undergo a conformational change in response to ligand binding by the aptamer domain, and that they harness this shape change to influence some part of the gene- expression process. In many cases, these shape changes involve the formation of Watson?Crick base pairs that render the alternative structures mutually exclusive. So, it is expected that some riboswitches will work more like a molecular fuse than a reversible switch. Once a partic- ular folding pathway for the RNA has been established by the presence or absence of the metabolite, extensive base pairing is likely to kinetically trap the riboswitch in the resulting structure, such that the expression plat- form cannot convert to its opposing form without being actively denatured. Gene-control mechanisms of a riboswitch In the case of the coenzyme-B 12 riboswitch, two main mechanisms for gene control are apparent (FIG. 2a).The first mechanism regulates RNA transcription and involves the ligand-dependent formation of an INTRINSIC TERMINATOR stem. Intrinsic terminators are extended stem-loop structures, which are typically followed by a run of six or more U residues, that cause RNA poly- merase to abort transcription before the coding portion of the mRNA has been made 67,68 .When coenzyme B 12 is not present in sufficient quantities, transcription of an mRNA that is associated with the coenzyme-B 12 riboswitch produces a nascent mRNA wherein the aptamer domain remains uncomplexed with ligand. The unbound aptamer domain permits formation of an ?ANTI-TERMINATOR?stem,which precludes formation of the intrinsic terminator stem and thereby permits transcription of the complete mRNA. However, when coenzyme-B 12 concentrations are adequate, the nascent mRNA binds to a coenzyme-B 12 molecule and the allosteric change in structure permits the intrinsic ter- minator stem to form. Transcription termination results and gene expression is prevented because the coding region of the mRNA is not made. Although the only evi- dence for this mechanism with the coenzyme-B 12 riboswitch comes from sequence analysis 44 , direct exper- imental evidence for this mechanism exists for other riboswitch classes 7,57?61 . The second mechanism that is used by the coen- zyme-B 12 riboswitch functions at the level of transla- tion initiation. Just as allosteric changes in aptamer structure can control the formation of intrinsic termi- nator and anti-terminator stems, coenzyme-B 12 bind- ing causes structural changes in full-length mRNAs to control access to the RIBOSOME-BINDING SITE (FIG. 2b). Specifically, it has been shown 35 that ribosomes are not able to form stable complexes with the btuB mRNA of E. coli when coenzyme B 12 is present in an in vitro assay. This observation, coupled with sequence 44 and biochemical 6,42 evidence for coenzyme-dependent alternative folding by the btuB 5?-UTR, supports this model for gene control. In rare instances, it appears that both transcription and translation can be controlled simultaneously. This different amino acids to choose from, RNA has only four different nucleotides with which to construct the appropriate receptor. Furthermore, the metabolite that is bound by each aptamer remains unchanged through- out evolution. So, it is not surprising that both the nucleotide sequence and the secondary structure of each aptamer remain highly conserved. This conserved pattern is therefore used to define the identity of each riboswitch class. The second structure is the EXPRESSION PLATFORM 57 , which is typically located immediately downstream from the aptamer domain, but in many instances the two domains overlap to some extent. The role of the expression platform is to transduce metabolite-binding events into gene-control consequences by allosteric modulation of the structure of the 5?-UTR. The expres- sion platform varies substantially between even closely related organisms, which reflects the diversity in expres- sion-platform function and the diversity of RNA struc- tures that can achieve the intended goal. Riboswitches can modulate gene expression by controlling the effi- ciency of translation initiation 6,57 , the transcription elongation of mRNA 7,58?64 , and presumably even the stability and splicing of mRNA transcripts 65,66 . Therefore, expression platforms that differ substantially in sequence permit riboswitch variants to control expression at a variety of levels. EXPRESSION PLATFORM The part of a riboswitch that interacts with an aptamer to transduce metabolite binding into a change in gene expression. INTRINSIC TERMINATOR A hairpin structure followed by a run of U residues in a nascent RNA transcript that stalls the RNA polymerase and induces transcription termination. ANTI-TERMINATOR A hairpin structure that, on formation, precludes the formation of an intrinsic terminator and thereby permits transcription to proceed. RIBOSOME-BINDING SITE Also known as the Shine?Dalgarno sequence, it is a short stretch of conserved nucleotides that is situated several nucleotides upstream of the start codon in prokaryotic mRNAs. This sequence is recognized by the ribosome during translation initiation. Figure 1 | Structural model and allosteric changes of a coenzyme-B 12 riboswitch. The secondary-structure model for the aptamer domain (nucleotides 1?202) of the coenzyme-B 12 riboswitch that precedes the btuB coding region of Escherichia coli. This aptamer has one of the most complex secondary structures, which is defined by 12 base-paired elements (P1?P12). In- line probing assays have shown that many nucleotides in the loops and bulges are floppy and undergo relatively rapid spontaneous cleavage (yellow) regardless of whether coenzyme B 12 is present. However, nucleotides at nine locations experience a decrease (red) or an increase (green) in the spontaneous cleavage rate. These results indicate that substantial structural reorganization occurs throughout this aptamer when coenzyme B 12 becomes bound. These structural changes are harnessed by the mRNA to control gene expression by one of several possible mechanisms (FIG. 2). The coenzyme-B 12 aptamer has one of the most complex metabolite-binding structures, but other riboswitches with simpler aptamers can also sense their targets precisely and use the resulting structural changes to control gene expression. This figure is adapted from REF. 42. 80 160 U 40 P5 P4 P2 P1 P12 P3 P6 P7 P8 P9 P10 P11 C G C U A CGC G G GC C U A A G G A ACU C CU U UU G C U U U U U C U U U U U U U U CG U U U C C U A GGU C C A A G G G G A A A A A A A A A A U A A U GU Coding region 200 241 180 100 140 AGU A G CG GG G G GC G G G G U GC A A C U G C C G G U G G A G AUGA A C C C UA 120 60 20 CU U A A A GC CCUC C C C GG C C C C G G C C G5? 3? ppp A A C C C C A U A G U A U C G G CCU A A A A G C G U G A G C UGGA U C C A G G ? ? ? ? U U G G AU G U A A � 2004 Nature Publishing Group NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 5 | JUNE 2004 | 455 REVIEWS site. Further experiments are needed to examine whether this combination of mechanisms is indeed exploited by some riboswitches. The diversity of riboswitches Riboswitches and the RNA world. The coenzyme-B 12 riboswitch fits the hypothesis that allosteric RNAs could have emerged early in evolution. Sequence analyses 42,44 would be possible if the transcription-terminator stem is formed by including base pairing with the ribosome- binding site. This combination of mechanisms would allow newly initiated mRNA transcripts to be aborted by the terminator stem, whereas transcripts whose syn- thesis had already passed the point of transcription ter- mination could exploit the same structural change to prevent translation by occluding the ribosome-binding Figure 2 | Coenzyme-B 12 -riboswitch structure and gene-control function. Each riboswitch is composed of an aptamer domain that remains largely conserved and an expression platform that varies more widely during evolution. The coenzyme-B 12 aptamer controls gene expression by two main mechanisms, as dictated by the architecture of the expression platform. a | Transcription control is mediated by the mutually exclusive formation of an anti-terminator stem (left) or an intrinsic terminator stem (right). Ligand binding requires that base pairing be formed between the loop of P5 and a stretch of nucleotides that would otherwise form an anti-terminator. In its simplest form, metabolite binding favours the formation of the base pairs that comprise the intrinsic terminator stem, at the expense of the anti-terminator stem. The involvement of additional structures could provide more complex kinetics of RNA folding and, therefore, more diverse mechanisms for gene control. Red shadings identify the complementary sequences that comprise the terminator and anti-terminator stems, wherein L and R identify the left and right shoulders of the terminator stem. b | Translation control is mediated by the mutually exclusive formation of an anti-anti-RBS (ribosome-binding site) stem and an anti-RBS stem. This mechanism also involves the mutually exclusive formation of base-paired elements; however, pairing to the RBS prevents the ribosome from binding to the mRNA. In the absence of the target metabolite, the anti-anti-RBS yields a structurally unencumbered RBS and the ribosome can bind. AUG designates the translation start codon, and the boxed UUU represents a repetitive U sequence that is required for transcription termination in bacteria of the Bacillus/Clostridium group. P1?P12 represent base-paired elements. Terminator (L) Anti-terminator Terminator (R) UUU AUGRBS 3? T erminator (L) Anti-terminator T erminator (R) UUU 3? 3? 3? Low coenzyme-B 12 concentration: ?ON? High coenzyme-B 12 concentration: ?OFF? Low coenzyme-B 12 concentration: ?ON? High coenzyme-B 12 concentration: ?OFF? 5? Expression platform Aptamer Expression platform Aptamer Anti-anti-RBS Anti-anti-RBS Anti-RBS Anti-RBS AUG AUG RBS RBS a b P5 P4 P3 P2 P1 P12 P6 P7 P8 P9 P10 P11 5? 5? 5? Anti-terminator stem Anti-anti-RBS stem Anti-RBS stem Terminator stem � 2004 Nature Publishing Group 456 | JUNE 2004 | VOLUME 5 www.nature.com/reviews/molcellbio REVIEWS all modern organisms 69?72 .From this viewpoint, it seemed reasonable to speculate that numerous riboswitches for many different compounds could have been present in ancient organisms. Although the even- tual emergence of proteins probably displaced many of these RNA-based gene-control elements, it also seems possible that modern cells might still use some of these motifs to control genes that are relevant to the metabo- lites being sensed. indicate that the coenzyme-B 12 riboswitch is wide- spread in prokaryotes ? as would be expected if its ori- gin were ancient. Furthermore, the chemical structure of coenzyme B 12 carries an adenosyl moiety and another moiety that is strikingly similar to a nucleotide. This RNA-like character of portions of coenzyme B 12 has led to speculation that this compound was impor- tant to organisms of the RNA world, as might be a number of other coenzymes that are essential to nearly Figure 3 | Consensus sequences and structures for known riboswitch classes. For each riboswitch class (a?g), a consensus sequence and structural motif is shown. Nucleotides in red are conserved in at least 90% of the representatives examined, circles reflect the presence of a nucleotide of any base identity, thin lines identify base pairing, and thick lines represent non-conserved regions. Distinct pairing elements (P) are numbered. (d) The adenine riboswitch is a variant of the guanine riboswitch with a critical C-to-U mutation in the aptamer core (U). (g) The glucosamine-6-phosphate (GlcN6P) riboswitch is also a ribozyme that catalyzes a self-cleaving reaction at the site indicated by the arrow. Data supporting the consensus patterns for (a) coenzyme B 12 (one of two main variants of the aptamer) 6,42,44 , (b) thiamine pyrophosphate (TPP) 7,34,65,73 , (c) flavin mononucleotide (FMN) 46 , (d) guanine/adenine 59,74 , (e) S-adenosylmethionine (SAM) 49,60,62 , (f) lysine 64,77 and (g) GlcN6P 78,79 riboswitches have been published elsewhere. G c P4 Y G U C A G U C A R Y R Y R R R Y K K P5 G C A G A R Y R R R P2 C CG G U P6 G A G A Y R C G G U P1 A R R Y C U Y Y P3 R G G G A G C C CC G C G G A A U U Y Y RC GG C G R YGA P2 P3 P4 P5 P1 G A A G A G C G C G R A Y U R CG G U A A U AC G U G GY A G P2a G YP2b A A G G U A GCR G U Y R RY H WW G R A GR f P2 P4 P5 P1 U D G G D R A G A C GA C U A G C G A AU GC GC b P3 P3a a C G A P1 A A C CG G R U R R A G CC G G CG C R G A R Y A R A G CC G G U P3 P4 P6 P7 P8 P2 A U A R U P5 G C G CC C P12 P9 G A A A G CG C G U G Y G P10 P11 e P3 P2 P1 A G G CR Y G G CG C G U N G R C N R U U A R U A A G A G CC G C C C P2a C G Y A G A A A Y G G P4 C A G U Y R U U G C C U C C A G U A A A U C A GG R U A A P1 A C Y A U G U R UR A P2 P3 d 5? 5? 5? 5? 5? 5? 5? (U) A AGCGCCHG UA CG RU RU NN WW P1 R Y GC RY GC G C G RA P2 P2a GACGAGG KN U G P3 G Y CG GC YNA A A A C A A A A N N A N RY NN NN CG G P4 P4a N A N N U N N NN R Y RY Y R R A U G C C G U R U G Cleavage R Y R C A A U g � 2004 Nature Publishing Group NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 5 | JUNE 2004 | 457 REVIEWS riboflavin, which differs from FMN only by the absence of a single phosphate residue. This high level of molecular discrimination, which is observed with all the riboswitch classes, is an essential characteristic if RNA is to perform adequately as a genetic switch. In this case, riboflavin is just one of several close ana- logues of FMN that are present in the cell as a result of biosynthetic processes, or that might be encoun- tered in the environment. If the RNA switch did not reject such compounds, then gene expression might not be accurately coupled to the concentration of the active coenzyme. Guanine and adenine riboswitches. One of the small- est metabolite-binding domains is present in a purine-specific class of riboswitches. The consensus motif is composed of a three-stem junction that is interspersed with stretches of conserved nucleotides in the central bulge and at the tips of the stem-loops (FIG. 3d).In most instances, this motif binds guanine with an apparent K D value of ~5 nM (for example, the xpt?pbuX variant from B. subtilis), whereas ade- nine is completely rejected by the binding pocket 59,74 . These observations are consistent with genetic evi- dence indicating the xpt-pbuX operon is controlled by guanine. However, several examples exist, upstream of adenine-metabolism genes, wherein the aptamer carries a C-to-U mutation within its core 74,75 .Interestingly, these specialized variants of the guanine riboswitch have altered ligand specificity, and now reject guanine and bind adenine with an apparent K D of ~300 nM. S-adenosylmethionine riboswitches. Gram-positive prokaryotes such as B. subtilis are particularly rich in both diversity and repeat occurrences of riboswitches. All known riboswitch classes are present in B. subtilis, whereas more-distantly related organisms such as E. coli carry only a subset of the motifs that are depicted in FIG. 3. SAM-dependent riboswitches 60?62 are the most frequently occurring within Gram-positive organ- isms. Although SAM is used as a coenzyme for methylase enzymes, in certain bacteria this com- pound also seems to be an important genetic signal. For example, B. subtilis has 11 TRANSCRIPTIONAL UNITS (which encode 26 genes) that are controlled by SAM riboswitches 49 . B. anthracis is even more enriched for metabolite-binding RNAs, and has 17 transcriptional units that are controlled by this riboswitch class. The mRNAs in this REGULON encode gene products that are related to sulphur metabolism, including genes that are involved in the biosynthesis of cysteine, methion- ine and SAM 49,76 .This arrangement suggests that SAM serves as a master molecule, the concentration of which is monitored by numerous riboswitch variants in each cell. Specifically, a drop in SAM concentration might be indicative of a general lack of sulphur-con- taining compounds. This is independently detected by each of the mRNAs in the regulon, which results in an increase in the expression of genes that are required for sulphur metabolism. How many modern riboswitches? Recent studies have made it clear that riboswitches are a diverse and wide- spread form of regulation. At the time of writing, there are reports of seven distinct classes of natural aptamers that recognize eight metabolites (FIG. 3).With one excep- tion, differently folded RNAs are required to form dis- tinct binding pockets for different metabolites. The exception is an aptamer class that was first shown to bind guanine, but can swap its molecular specificity to adenine by a single point mutation (see below). In this instance, no substantive change in the molecular architecture of the aptamer is required to swap binding specificity. Therefore, other riboswitch structural classes might also include variants whose purpose is to respond to chemically related targets. The potential for the dis- covery of new riboswitch classes, either as variants of known classes or as entirely new motifs, seems assured. We anticipate that additional riboswitches will be reported as researchers seek to explain other gene-con- trol mysteries. Thiamine pyrophosphate riboswitches. Each of the riboswitch ligands that have been identified so far functions in fundamental metabolic pathways. TPP, which is a target of one of the most widespread riboswitch classes, is commonly involved as a coen- zyme for decarboxylase enzymes. The TPP riboswitch consensus sequence (FIG. 3b), like the coenzyme itself, appears in all three domains of life 65,66,73 .The riboswitch controls both transcrip- tion 58 and translation 7,58 using mechanisms that are similar to those described for the coenzyme-B 12 riboswitch. Interestingly, the TPP aptamer domain from the thiC gene of E. coli has ~1,000-fold tighter binding to TPP (apparent K D of ~100 nM) compared with thiamine, which lacks the pyrophosphate moiety. Therefore, RNAs can form binding pockets that make productive contacts with phosphate moieties. This is surprising, given that the phosphodiester backbone of RNA is exceptionally polyanionic, and so the TPP riboswitch must overcome inherent repulsive effects to accommodate the two additional phosphates of its target molecule. Flavin mononucleotide riboswitches. Many of the characteristics of the TPP riboswitch are shared with another riboswitch class that selectively binds FMN (FIG. 3c),which is another key coenzyme that is required for the function of many redox-active enzymes. The FMN-specific aptamer has been identi- fied in numerous bacteria by database searching 31 , and is one of the most common riboswitches in prokaryotes along with the coenzyme-B 12 - and TPP- specific classes (J. E. Barrick and R.R.B., unpublished observations). Similarly, evidence exists for the con- trol of both transcription and translation by FMN riboswitches that are guided by allosteric structural changes 57,58 .The riboswitch from the ribDEAHT operon of B. subtilis binds FMN with an apparent K D of ~10 nM (REF. 57).Again, this is nearly 1,000-fold better than the apparent K D that was measured for TRANSCRIPTIONAL UNIT An RNA transcript, such as mRNA, that is transcribed separately. In the case of operons, one transcriptional unit can encode several proteins. REGULON A collection of separate genes, the expression of which is controlled as a unit by a specific signalling compound or factor. � 2004 Nature Publishing Group 458 | JUNE 2004 | VOLUME 5 www.nature.com/reviews/molcellbio REVIEWS other compounds that have no chemical similarity to nucleic acids. Perhaps this capability is best manifested by the lysine-dependent class of riboswitches 59,63,64,77 .Although lysine is much smaller than most other target molecules, the consensus sequence and secondary structure for this aptamer is one of the largest known (FIG. 3f).The demands for accurate molecular recognition by this riboswitch are extreme, as there are many compounds in the cell, including high concentrations of similar amino acids, that must be excluded from triggering changes in gene expression. Again, the lysine aptamer has remarkable selectivity for its target molecule, in this case, despite the absence of nucleotide-like moieties. The Lysine riboswitches. All the compounds that are sensed by the riboswitches described above are fundamental in their biochemical importance and, like coenzyme B 12 ,have substantial chemical similarity to RNA. At least in the case of the purine-specific riboswitches, this RNA-like character is exploited by the aptamer for molecular recognition. Guanine and adenine, even when integrated into larger compounds such as coen- zyme B 12 and SAM, provide chemical moieties that can be bound through simple Watson?Crick base-pair- ing 59,74 .But, if primitive, RNA-world organisms were going to evolve ever-greater metabolic sophistication, molecular recognition by aptamers and by ribozymes would have had to expand to accommodate many Figure 4 | Differences in ligand specificity and switch direction for purine-specific riboswitches. a | The guanine-specific riboswitch from the xpt-pbuX operon of B. subtilis 59 . Guanine (G) binding involves base pairing to a cytidine (C) residue of the aptamer and causes formation of an intrinsic terminator stem that turns off gene expression. P1?P3 represent pairing elements that are essential for formation of the aptamer. When G is absent, the anti-terminator stem is formed at the expense of P1. b | The adenine-specific riboswitch from the ydhL gene of B. subtilis 74 . A uridine (U) residue is responsible for base pairing to the ligand. Adenine (A) binding prevents formation of the terminator, thereby causing gene expression to switch on. In this case, the anti- terminator structure is part of the aptamer (stems P1 and P2). In the absence of adenine, some sequences of P1 and P2 now form a part of the intrinsic terminator. Some types of riboswitches probably become kinetically trapped due to the extensive formation of base-paired stems, and are not able to switch freely between ?ON? and ?OFF? states as depicted here. Extensive anti-terminator and terminator base-pairing interactions would be slow to interconvert, and so gene-control decisions would be made by each riboswitch molecule as it commits to a specific folding pathway as established by the presence or absence of bound ligand. Thin black lines represent base pairs, thick black lines represent non-conserved regions and residues that are involved in the formation of terminator or anti-terminator stems are shaded yellow. b a U C A A U G UA U U A AA AU U UG G C C U C A U A GG A U G U AG C U AA UU U A C C A A A C A C U U A G C G G G U A A C G P1 P2 P3 A A U U G U U A A A U UG G C C U C A U A G G A U G U AG C U AA UU U A C C A C A A G G G U C G P2 A U U U A A A U UA C A C G G U CG AU A AU U AU A U U U A A CG AU UUU A C U A C G A G U GC CG UA AU U G A C UG G C GU CG U A U U U U U U U U U U U G U A A C G UU U U G AUAGUGCU G U U G C C A A Anti-terminator stem Anti-terminator stem Terminator stem 100 120 140 160 G CG C G CA UU A C U A G U A A U C G GC U G 20 80 U C A A A AU U UU G C C U C A U A GG G U A G CG C C GU AC G A C C A C U C A U C G G G C G G G U A A 60 U A 5? 5? 5? 5? P1 P2 P3 G A G ?OFF? ?OFF? U A U U U U 140 ?ON? ?ON? C A U A A C C U 20 U C A A A A U U UU G C C U C A U A GG G U A G CG C C GU AC G A 4040 C C A C U C A U C G G G C G G G U A A 60 U A P1 P2 P3 G A U G C U C A UA G C U G GU U G UA GC GC Terminator stem A � 2004 Nature Publishing Group NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 5 | JUNE 2004 | 459 REVIEWS Although OFF switches have been demonstrated by experiment and seem to be the most common 63,64 , sequence data indicate that expression of lysine- degradation genes might be activated by ON switches 77 .These main types of gene-control mecha- nisms are undoubtedly sufficient for most of the reg- ulatory needs in prokaryotes. However, many other mechanisms could be coupled to metabolite binding by RNA, and if these are realized then the level of sophistication of RNA-based gene-control elements would become even closer to that observed for pro- tein factors. An alternative form of gene regulation seems to be used with TPP-dependent riboswitches of eukaryotes. In both fungi and plants, sequence elements that match the TPP aptamer consensus and bind TPP in vitro are affiliated with INTRONS that reside in the 5?- or 3?-UTR of thiamine-metabolism genes 65 .In addition, these ele- ments seem to be essential for splicing events that are necessary for the expression of the gene 66 .Presumably, ligand binding precludes pre-mRNA splicing, which would exemplify a new mechanism whereby high con- centrations of a metabolite downregulate expression of biosynthetic or import genes for that compound. Similar mechanisms can be envisioned for lysine riboswitches or any of the other classes as well, although it not yet known whether eukaryotes carry additional riboswitches. Other RNA switches Many prokaryotic RNA elements that have been known for years also function, in some respects, as genetic switches, including small non-coding RNAs 80,81 .Even eukaryotic organisms are known to use RNA elements that interface with proteins to control gene expression or to guide the subcellular location of RNA. For example, recent work on RNA elements that are embedded within cytokine genes indicates that they form complexes with an RNA- dependent protein kinase, which ultimately leads to downregulation of protein synthesis 82 .Similar systems probably control gene expression by a variety of dis- tinct mechanisms in prokaryotes and eukaryotes, and it has been proposed that modern organisms contain many such elements 83 .But are there distinctions to be made between riboswitches and some of these other RNA elements? These elements cannot simply be classified by whether a protein factor is involved. For example, ther- mosensor RNAs 84,85 and T-box RNAs 76 (BOX 3) do not necessarily rely on protein factors for function, but also lack some of the characteristics of metabolite-binding riboswitches. Similarly, a metabolite-binding riboswitch could bind its target, and then recruit a protein factor as part of its control mechanism. So, it is difficult in some instances to draw clear-cut distinctions between riboswitches and other gene-control processes that involve RNA structures. However, we think that there is sufficient difference between metabolite-binding riboswitches and RNAs that respond to other factors such that separate classifications are warranted. isolated aptamer binds L-lysine with an apparent K D of 1 �M, and discriminates against closely related lysine analogues such as ornithine, cadaverine, 5-hydroxylysine, and even the D-isomer of lysine 64 .These observations indicate that RNA aptamers of sufficient complexity can selectively bind even small and structurally uncon- strained target molecules. Therefore, we should place no preconceived restrictions on the range of targets that riboswitches could possibly recognize. Glucosamine-6-phosphate riboswitch. A very different mechanism for riboswitch action controls the glmS gene of many Gram-positive organisms. The glmS gene encodes an enzyme that produces glucosamine-6- phosphate (GlcN6P). This compound selectively trig- gers a novel SELF-CLEAVING RIBOZYME that is located in the 5?-UTR of the mRNA 78,79 (FIG. 3g).The ribozyme is acti- vated 1,000 fold when incubated with GlcN6P, which reduces its half-life for cleavage from over 4 hours to less than 15 seconds. In engineered reporter constructs, mutations that deactivate the ribozyme lead to dere- pression of gene expression. It is not yet clear whether ribozyme action leads to gene repression by inducing structural alterations in the cleaved UTR, or whether some other processes are involved. Regardless, the inte- gration of ribozymes and aptamers adds new possibili- ties for gene control by riboswitches, just as molecular engineers have created similar RNAs for biosensing applications 24?28 . Diversity of gene-control mechanisms. Like the glmS ribozyme, several of the riboswitch classes described above also showcase the diversity of mechanisms that are available for controlling gene expression. For exam- ple, the adenine-binding variant associated with the ydhL gene of B. subtilis serves as an ?ON? switch. In pre- vious examples that were studied in detail, including the xpt-pbuX motif (FIG. 4a),riboswitches of all classes turn gene expression off in response to ligand binding. This common observation makes sense when it is considered that most control tasks related to metabolism will require downregulation of biosynthetic genes when an adequate level of the desired compound is produced. In the case of ydhL, the function of the gene product seems to be purine transport that expels excess adenine if the need arises 75 .Activation ofydhL expression is achieved by the adenine-specific riboswitch by a variation of the transcription-termination mechanism that was described for the coenzyme-B 12 riboswitch. In this instance, sequences that comprise the right side of the terminator stem are also required to form part of the aptamer core (FIG. 4b).So, adenine binding precludes for- mation of the terminator stem, and gene expression proceeds at a high level 74 .RNA can undoubtedly serve either as an ON switch or an OFF switch. However, genetic necessity does not demand a large number of metabolite-induced ON switches, and so these are expected to be far less common. As with other classes, lysine-dependent riboswitches seem to control gene expression by both transcription- termination 63,64 and translation-initiation mechanisms 77 . SELF-CLEAVING RIBOZYME Five of the nine known natural ribozymes catalyze self-cleavage using an internal phosphoester transfer reaction. INTRON A non-coding segment of mRNA that is removed by splicing processes before translation by ribosomes. ROSE ELEMENT An RNA sequence in certain bacteria that responds to changes in temperature and controls expression of adjacent heat-shock genes. � 2004 Nature Publishing Group 460 | JUNE 2004 | VOLUME 5 www.nature.com/reviews/molcellbio REVIEWS to high concentrations of aminoacylated tRNA trp (rapid translation), then transcription is halted before the main coding regions of the mRNA are transcribed. By con- trast, low concentrations of aminoacylated tRNA trp (slow translation, which is indirectly due to low trypto- phan concentrations) permits the formation of an anti-terminator stem, which prevents formation of the terminator stem and thereby facilitates transcription of the full mRNA for the trp operon. Unlike metabolite-binding riboswitches, the termi- nator and anti-terminator stems form in response to the action of a protein-containing complex. Specifically, the ribosome serves as the sensor element by its innate abil- ity to recognize aminoacylated tRNA trp when the proper codon is present in the mRNA that is being translated. So, molecular recognition is carried out by the ribo- some, and the presence of the tRNA trp target is relayed to RNA polymerase by the selection of one of two folding pathways for the nascent mRNA. Riboswitches are dis- tinct in that they directly serve as the sensor element by recognizing their target metabolite in the absence of proteins. For some riboswitches, this molecular-recog- nition event is coupled to gene control because it estab- lishes a folding pathway for the 5?-UTR, typically using Watson?Crick base-paired structures like those described above. In B. subtilis,formation of an intrinsic terminator stem occurs if an interaction forms between a trypto- phan-binding protein complex (TRAP) and an RNA sequence in the 5?-UTR of the nascent mRNA for the tryptophan operon (BOX 4).The RNA?protein interac- tion, which is induced by high tryptophan concentra- tions, prevents the formation of an anti-terminator stem, and the resulting terminator stem causes tran- scription to abort before the coding regions of the mRNA are made. In this case, TRAP serves as the sepa- rate protein factor that senses the target metabolite. Although the 5?-UTR carries the necessary control ele- ments (terminator and anti-terminator stems), its over- all RNA structure is made simpler because proteins are involved in recognition of the appropriate target com- pound and of the mRNA sequence. Metabolite-binding riboswitches take a dominant role in forming receptor structures that dock their respective metabolites. This is evident on examination of the architectures of riboswitches, which all carry complex folded structures that vary little through evolu- tion because their metabolite ligands are static targets that do not change in chemical composition. Most RNA motifs that are bound by proteins are substantially smaller, and show rapid evolutionary changes in sequence and structure when compared to metabolite- dependent riboswitches 79 .This is due to the fact that the RNA or protein factor in the complex also varies through evolutionary time. The complexity of these other RNA switches should not be underestimated however. Another structural fea- ture of the nascent trp mRNA described above is a third stem-loop, which causes TRANSCRIPTIONAL PAUSING.This pausing, which is overcome once the ribosome begins to translate the leader peptide, seems to be important for These distinctions can be highlighted best by exam- ining gene-control processes in prokaryotes that have some characteristics of riboswitches. For example, clas- sic translation-mediated transcription-attenuation mechanisms (BOX 4) are controlled by differentially folded secondary structures 86 .Control of the trypto- phan operons in E. coli and B. subtilis are perhaps the most extensively studied examples of this form of gene regulation 87 .In E. coli, the speed of ribosome movement while translating a short LEADER PEPTIDE from the nascent mRNA chain for tryptophan biosynthetic genes dictates whether an intrinsic terminator stem 67,68 will form. The coding region for the leader (trpL) is located in what would otherwise be the untranslated region of the trp operon mRNA, and instructs the ribosome to incorpo- rate several tryptophan residues. If the terminator stem forms in the 5?-UTR of the nascent mRNA in response AMINOACYL-tRNA SYNTHETASE An enzyme that recognizes a specific tRNA and selectively loads each with its cognate amino acid. LEADER PEPTIDE A peptide that is encoded upstream of a larger open reading frame, the translation of which is used as a sensor for adequate levels of a particular aminoacylated tRNA. TRAP A complex formed by the trp RNA-binding attenuation protein from B. subtilis. TRAP binds tryptophan and serves as a protein factor for regulating the trp operon. Box 3 | Thermosensors and T-boxes RNA thermosensors and T-box RNAs are gene-control elements that share some characteristics with metabolite-binding riboswitches. The 5?-untranslated region (5?-UTR) of the prfA gene from Listeria monocytogenes carries a hairpin structure that occludes the ribosome-binding site (RBS; see figure, part a). A shift in temperature from 30�C to 37�C induces a fivefold increase in pfrA expression 84 ,presumably by melting local base pairs and permitting access by the ribosome. Mutational analysis indicates that the stability of the hairpin structure is crucial and, therefore, the RNA alone might serve as a molecular thermometer. This form of control might be widespread, as a similar system seems to function in the ROSE ELEMENT in various rhizobial species 85 and in the untranslated region of the E. coli rpoH gene 96 . T-box RNAs (see figure, part b) are normally found in the 5? UTR of genes that encode AMINOACYL-tRNA SYNTHETASES or related amino-acid-biosynthesis genes of Gram- positive organisms. Each folds into a structure that selectively recognizes a specific tRNA 76 .Recognition is guided largely through the formation of Watson?Crick base pairing between the T-box RNA and the anticodon loop of the target tRNA. Gene expression is activated in the presence of tRNAs that are not aminoacylated, thereby boosting the expression of genes that are needed to maintain an adequate pool of charged tRNAs. The T-box associated with the tyrS gene of B. subtilis (shown in the figure) functions via a transcription-termination mechanism. The T-box RNA rejects its matched tRNA when the tRNA is already aminoacylated. This permits formation of the terminator stem and subsequently downregulates gene expression (not shown). But, if the tRNA is not aminoacylated (as depicted in the figure), it is bound by the T-box and this permits expression of the adjoining synthetase gene by formation of an anti-terminator structure. ab G U A C A U U C GG AC U A tRNA T-box RNA Coding region G G G G G RBS GUA 5? 5? 5? Coding region Heat Terminator Te r m inator Anti-terminator stem � 2004 Nature Publishing Group NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 5 | JUNE 2004 | 461 REVIEWS Conclusions The studies that have been conducted so far prove that riboswitches directly bind metabolites in the absence of proteins. The experiments used to assess the mechanism of riboswitch gene control do contain proteins such as RNA polymerase or RNase H. However, these added proteins are usually highly purified, and, therefore, are not likely to lead investi- gators to a false conclusion by harbouring protein fac- tors that would be essential for a more conventional gene-control mechanism. Neither of these points rule out the possibility that metabolite binding by the RNA recruits or displaces some protein factor that is critical for controlling gene expression inside cells. What is clear from the current data set is that there is no need to invoke the existence of proteins, as each riboswitch seems to have all the components that are necessary to bind its target and influence gene control without the aid of other factors. The diversity of natural aptamers is in accordance with the finding by molecular engineers that RNA has a robust ability to selectively bind to a variety of organic compounds. Given the versatility of RNA, particularly those molecules that are large in size, it is likely that almost any compound could serve as a target for an RNA aptamer. Therefore, the question regarding the extent of participation by riboswitches in modern genetic control might be one of evolutionary persis- tence as opposed to capability. How many riboswitches emerged early in evolution and were able to survive in the face of stiff competition from protein factors? Perhaps clues to this mystery can be derived by looking at the results of the long-standing competition between ribozymes and protein enzymes. It seems that the only ribozymes to survive were those that were difficult to replace by small evolutionary changes (for example, the ribosome, RNASE P and self- splicing ribozymes), or which have functions that are particularly well suited to their task (such as self-cleav- ing ribozymes). Protein factors might be able to provide even more precise molecular recognition capabilities or more versatile control mechanisms than those shown by riboswitches. However, the relative simplicity of mRNAs conducting their own metabolic surveillance and making the necessary changes to gene expression in the absence of protein factors might still be most effi- cient in many circumstances. We suspect that many such circumstances exist and that new riboswitches will continue to be discov- ered as new bioinformatics and biochemical strate- gies are applied. For example, comparative sequence analysis has been used to identify six new RNA motifs with structural characteristics that are indicative of riboswitch function 79 .However,this approach tends to favour the identification of RNA elements that are widely conserved and can miss RNA elements that are important for gene-control mechanisms in a few closely related species. Other bioinformatics approaches 88?90 and biochemical strategies, such as the cloning of non-coding RNAs 91 ,provide complementary routes to the identification of novel RNA motifs. coordinating transcription and translation timing. In each case, the mRNA structures that are differentially formed are a result of the action of other complexes (either RNA polymerase or the ribosome), which deter- mine the folding pathway of the mRNA. For pausing, anti-termination and termination, the structures formed by the nascent mRNA are usually established by Watson?Crick base pairing. Therefore, the nucleotide sequences of these elements show considerable variabil- ity from organism to organism. By contrast, riboswitches rely heavily on critical tertiary contacts that tend to require more-extensive sequence conservation compared with other RNA genetic elements. TRANSCRIPTIONAL PAUSING The temporary stalling of RNA polymerase during transcription that is typically caused by hairpin structures or other sequence elements within the nascent mRNA. RNase P A ribonucleoprotein-enzyme complex wherein the RNA component serves as a ribozyme that processes precursor RNAs such as pre-tRNA transcripts. Box 4 | Transcription attenuation and protein-mediated RNA switches Classical mechanisms of bacterial gene control that involve mRNA structures require the ribosome (transcription attenuation) or protein factors. Transcription attenuation of the E. coli trp operon makes use of the process of translation to determine the levels of tryptophan in the cell. A sufficient supply of tryptophan allows the cell to produce a level of tryptophan-charged tRNA such that the ribosome rapidly transcribes a tryptophan-rich leader peptide (trpL). The ribosome sequesters anti- terminator sequences and, therefore, transcription of the remaining mRNA chain is terminated (as shown in the figure, part a). If tryptophan-charged tRNAs are rare, then the ribosome transcribes the leader more slowly and permits formation of the anti-terminator stem (not shown). The result of this is transcription of the full-length mRNA and expression of the operon, which encodes genes that are needed to boost the tryptophan concentration. Transcription-termination control in the B. subtilis trp operon makes use of an 11-subunit protein complex, known as TRAP, wherein each member of the complex binds tryptophan. At high concentrations, tryptophan binds and triggers changes in the protein structure that permit the complex to bind the nascent trp mRNA. If bound, sequences of the anti-terminator are sequestered and transcription is terminated (see figure, part b). When tryptophan concentrations are low, the protein complex cannot bind the nascent mRNA, and anti-terminator-stem formation assures that transcription of the complete mRNA proceeds (not shown). Ribosome 5? 5? Genes of trp operon DNA RNAP Terminator stem TRAP complex Anti-terminator b a Genes of trp operon trpL DNA Terminator stem Polypeptide RNAP Anti-terminator � 2004 Nature Publishing Group 462 | JUNE 2004 | VOLUME 5 www.nature.com/reviews/molcellbio REVIEWS 1. Ptashne, M. & Gann, A. Genes and Signals (Cold Spring Harbor Laboratory Press, New York, 2002). 2. Hannon, G. J. RNA interference. Nature 418, 244?251 (2002). 3. Dykxhoorn, D. M., Novina, C. D. & Sharp, P. A. Killing the messenger: short RNAs that silence gene expression. Nature Rev. Mol. Cell Biol. 4, 457?467 (2003). 4. McManus, M. T. & Sharp, P. A. Gene silencing in mammals by small interfering RNAs. Nature Rev. Genetics 3, 737?747 (2002). 5. Carrington, J. C. & Ambros, V. Role of microRNAs in plant and animal development. Science 301, 336?338 (2003). 6. Nahvi, A. et al. Genetic control by a metabolite binding mRNA. Chem. Biol. 9, 1043?1049 (2002). The first demonstrations that mRNAs bind metabolites directly in the absence of proteins are described in this paper and in reference 7. 7. Winkler, W., Nahvi, A. & Breaker, R. R. Thiamine derivatives bind messenger RNAs directly to regulate bacterial gene expression. Nature 419, 952?956 (2002). This paper and references 32, 57 and 58 report evidence for the existence of two main forms of riboswitch gene control: transcription termination and translation initiation. 8. Lai, E. C. RNA sensors and riboswitches: self-regulating messages. Curr. Biol. 13, R285?R291 (2003). 9. M�ller, S. Another face of RNA: metabolite-induced ?riboswitching? for regulation of gene expression. Chembiochem 4, 817?819 (2003). 10. Winkler, W. C. & Breaker, R. R. Genetic control by metabolite-binding riboswitches. Chembiochem 4, 1024?1032 (2003). 11. Nudler, E. & Mironov, A. S. The riboswitch control of bacterial metabolism. Trends Biochem. Sci. 29, 11?17 (2004). 12. Vitreschak, A. G., Rodionov, D. A., Mironov, A. A. & Gelfand, M. S. Riboswitches: the oldest mechanism for the regulation of gene expression? Trends Genet. 20, 44?50 (2004). 13. Monod, J. & Jacob, F. General conclusions: teleonic mechanisms in cellular metabolism, growth, and differentiation. Cold Spring Harbor Symp. Quant. Biol. 26, 389?401 (1961). 14. Monod, J., Changeux, J.-P. & Jacob, F. Allosteric proteins and cellular control systems. J. Mol. Biol. 6, 306?329 (1963). 15. Kuganov, B. I. Allosteric enzymes. (John Wiley & Sons Ltd., New York, 1978). 16. Jacob, F. & Monod, J. Genetic regulatory mechanisms in the synthesis of proteins. J. Mol. Biol. 3, 318?356 (1961). 17. Gilbert, W. The RNA world. Nature 319, 618 (1986). 18. Joyce, G. F. The antiquity of RNA-based evolution. Nature 418, 214?221 (2002). 19. Kruger, K., Grabowski, P. J., Zaug, A. J., Sands, J., Gottschling, D. E. & Cech, T. R. Self-splicing RNA: autoexcision and autocyclization of the ribosomal RNA intervening sequence of Tetrahymena. Cell 31, 147?157 (1982). 20. Guerrier-Takada, C., Gardiner, K., Marsh, T., Pace, N. & Altman, S. The RNA moiety of ribonuclease P is the catalytic subunit of the enzyme. Cell 35, 849?857 (1983). 21. Gold, L., Polisky, B., Uhlenbeck O. & Yarus, M. Diversity of oligonucleotide functions. Annu. Rev. Biochem. 64, 763?797 (1995). 22. Osborne, S. E. & Ellington, A. D. Nucleic acid selection and the challenge of combinatorial chemistry. Chem. Rev. 97, 349?370 (1997). 23. Hermann, T. & Patel, D. J. Adaptive recognition by nucleic acid aptamers. Science 287, 820?825 (2000). 24. Tang, J. & Breaker, R. R. Rational design of allosteric ribozymes. Chem. Biol. 4, 453?459 (1997). First demonstration that engineered RNAs can function as allosteric molecular switches and respond to small metabolites. 25. Soukup, G. A. & Breaker, R. R. Engineering precision RNA molecular switches. Proc. Natl Acad. Sci. USA 96, 3584?3589 (1999). 26. Seetharaman, S., Zivarts, M., Sudarsan, N. & Breaker, R. R. Immobilized RNA switches for the analysis of complex chemical and biological mixtures. Nature Biotechnol. 19, 336?341 (2001). 27. Breaker, R. R. Engineered allosteric ribozymes as biosensor components. Curr. Opin. Biotechnol. 13, 31?39 (2002). 28. Silverman, S. K. Rube Goldberg goes (ribo)nuclear? Molecular switches and sensors made from RNA. RNA 9, 377?383 (2003). 29. Gold, L., Brown, D., He, Y.-Y., Shtatland, T., Singer, B. S. & Wu, Y. From oligonucleotide shapes to genomic SELEX: novel biological regulatory loops. Proc. Natl Acad. Sci. USA 94, 59?64 (1997). Although unpublished speculation that riboswitches might exist had been ongoing for several years, this is an early publication that briefly mentions the possibility. 30. Gold, L., Singer, B., He, Y.-Y. & Brody, E. SELEX and the evolution of genomes. Curr. Opin. Genet. Dev. 7, 848?851 (1997). 31. Gelfand, M. S., Mironov, A. A., Jomantas, J., Kozlov, Y. I. & Perumov, D. A. A conserved RNA structure element involved in the regulation of bacterial riboflavin synthesis genes. Trends Genet. 15, 439?442 (1999). This publication, as well as reference 49, used sequence comparisons to make the first secondary- structure models for genetic-control elements that have since proven to be riboswitches. 32. Nou, X. & Kadner, R. J. Adenosylcobalamin inhibits ribosome binding to btuB RNA. Proc. Natl Acad. Sci. USA 97, 7190?7195 (2000). 33. Stormo, G. D. & Ji, Y. Do mRNAs act as direct sensors of small molecules to control their expression? Proc. Natl Acad. Sci. USA 98, 9465?9467 (2001). 34. Miranda-Rios, J., Navarro, M., & Sober�n, M. A conserved RNA structure (thi box) is involved in regulation of thiamin biosynthetic gene expression in bacteria. Proc. Natl Acad. Sci. USA 98, 9736?9741 (2001). 35. Lundrigan, M. D., K�ster, W. & Kadner, R. J. Transcribed sequences of the Escherichia coli btuB gene control its expression and regulation by vitamin B 12 . Proc. Natl Acad. Sci. USA 88, 1479?1483 (1991). 36. Ravnum, S. & Andersson, D. I. Vitamin B 12 repression of the btuB gene in Salmonella typhimurium is mediated via a translational control which requires leader and coding sequences. Mol. Microbiol. 23, 35?42 (1997). 37. Richter-Dahlfors, A. A., Ravnum, S. & Andersson, D. I. Vitamin B 12 repression of the cob operon in Salmonella typhimurium: translational control of the cbiA gene. Mol. Microbiol. 13, 541?553 (1994). 38. Ravnum, S. & Andersson, D. I. An adenosyl-cobalamin (coenzyme-B 12 )-repressed translational enhancer in the cob mRNA of Salmonella typhimurium. Mol. Microbiol. 39, 1585?1594 (2001). 39. Soukup, G. A. & Breaker, R. R. Relationship between internucleotide linkage geometry and the stability of RNA. RNA 5, 1308?1325 (1999). 40. Li, Y. & Breaker, R. R. Kinetics of RNA degradation by specific base catalysis of transesterification involving the 2?-hydroxyl group. J. Am. Chem. Soc. 121, 5364?5372 (1999). 41. Soukup, G. A., DeRose, E. C., Koizumi, M. & Breaker, R. R. Generating new ligand-binding RNAs by affinity maturation and disintegration of allosteric ribozymes. RNA 7, 524?536 (2001). 42. Nahvi, A., Barrick, J. E. & Breaker, R. R. Coenzyme B 12 riboswitches are widespread genetic control elements in prokaryotes. Nucleic Acids Res. 32, 143?150 (2004). 43. Bond, C. M., Lees, K. A. & Enever, R. P. Photolytic decomposition of three cobalamins. A quantitative study. J. Pharm. Pharmacol. 24 (Suppl.), 143P (1972). 44. Vitreschak, A. G., Rodionov, D. A., Mironov, A. A. & Gelfand, M. S. Regulation of the vitamin B 12 metabolism and transport in bacteria by a conserved RNA structural element. RNA 9, 1084?1097 (2003). 45. Webb, E., Claas, K. & Downs, D. thiBPQ encodes an ABC transporter required for transport of thiamine and thiamine pyrophosphate in Salmonella typhimurium. J. Biol. Chem. 273, 8946?8950 (1996). 46. Mironov, V. N., Perumov, D. A., Kraev, A. S., Stepanov, A. I. & Skryabin, K. G. Unusual structure in the regulation region of the Bacillus subtilis riboflavin biosynthesis operon. Mol. Biol. 24, 256?261 (1990) (in Russian). 47. Kreneva, R. A. & Perumov, D. A. Genetic mapping of regulatory mutations of Bacillus subtilis riboflavin operon. Mol. Gen. Genet. 222, 467?469 (1990). 48. Kil, Y. V., Mironov, V. N., Gorishin, I. Y., Kreneva, R. A. & Perumov, D. A. Riboflavin operon of Bacillus subtilis: unusual symmetric arrangement of the regulatory region. Mol. Gen. Genet. 233, 483?486 (1992). 49. Grundy, F. J. & Henkin, T. M. The S box regulon: a new global transcription termination control system for methionine and cysteine biosynthesis genes in Gram-positive bacteria. Mol. Microbiol. 30, 737?749 (1998). 50. Ebbole, D. J. & Zalkin, H. Cloning and characterization of a 12-gene cluster from Bacillus subtilis encoding nine enzymes for de novo purine nucleotide synthesis. J. Biol. Chem. 262, 8274?8287 (1987). 51. Christiansen, L. C., Schou, S., Nygaard, P. & Saxild, H. H. Xanthine metabolism in Bacillus subtilis: characterization of the xpt-pbuX operon and evidence for purine- and nitrogen- controlled expression of genes involved in xanthine salvage and catabolism. J. Bacteriol. 179, 2540?2550 (1997). 52. Vold, B., Szulmajster, J. & Carbone, A. Regulation of dihydrodipicolinate synthase and aspartate kinase in Bacillus subtilis. J. Bacteriol. 121, 970?974 (1975). 53. Lu, Y., Chen, N. Y. & Paulus, H. Identification of aecA mutations in Bacillus subtilis as nucleotide substitutions in the untranslated leader region of the aspartokinase II operon. J. Gen. Microbiol. 137, 1135?1141 (1991). This paper defines the precise locations of mutations that confer resistance to the toxic effects of a lysine analogue. The resistant phenotype, first described nearly 20 years earlier, is now known to be caused by mutations in a lysine-specific riboswitch. 54. Kochhar, S. & Paulus, H. Lysine-induced premature transcription termination in the lysC operon of Bacillus subtilis. Microbiol. 142, 1635?1639 (1996). 55. Patte, J. C. in Escherichia coli and Salmonella: Cellular and Molecular Biology Vol. 1 (eds Neidhardt, F. C. et al.) 528?541 (American Society for Microbiology Press, Washington DC, 1996). 56. Patte, J.-C., Akrim, M. & M�jean, V. The leader sequence of the Escherichia coli lysC gene is involved in the regulation of LysC synthesis. FEMS Microbiol. Lett. 169, 165?170 (1998). 57. Winkler, W. C., Cohen-Chalamish, S. & Breaker, R. R. An mRNA structure that controls gene expression by binding FMN. Proc. Natl Acad. USA 99, 15908?15913 (2002). 58. Mironov, A. S. et al. Sensing small molecules by nascent RNA: a mechanism to control transcription in bacteria. Cell 111, 747?756 (2002). 59. Mandal, M., Boese, B., Barrick, J. E., Winkler, W. C. & Breaker, R. R. Riboswitches control fundamental biochemical pathways in Bacillus subtilis and other bacteria. Cell 113, 577?586 (2003). 60. McDaniel, B. A. M., Grundy, F. J., Artsimovitch, I. & Henkin, T. M. Transcription termination control of the S box system: direct measurement of S-adenosylmethionine by the leader RNA. Proc. Natl Acad. Sci. USA 100, 3083?3088 (2003). 61. Epshtein, V., Mironov, A. S. & Nudler, E. The riboswitch- mediated control of sulfur metabolism in bacteria. Proc. Natl Acad. Sci. USA 100, 5052?5056 (2003). 62. Winkler, W. C., Nahvi, A., Sudarsan, N., Barrick, J. E. & Breaker, R. R. An mRNA structure that controls gene expression by binding S-adenosylmethionine. Nature Struct. Biol. 10, 701?707 (2003). 63. Grundy, F. J., Lehman, S. C. & Henkin, T. M. The L box regulon: lysine sensing by leader RNAs of bacterial lysine biosynthesis genes. Proc. Natl Acad. Sci. USA 100, 12057?12062 (2003). 64. Sudarsan, N., Wickiser, J. K., Nakamura, S., Ebert, M. S. & Breaker, R. R. An mRNA structure in bacteria that controls gene expression by binding lysine. Genes Dev. 17, 2688?2697 (2003). 65. Sudarsan, N., Barrick, J. E. & Breaker, R. R. Metabolite- binding RNA domains are present in the genes of eukaryotes. RNA 9, 644?647 (2003). Demonstration that some eukaryotic mRNAs carry metabolite-binding domains that are probably components of riboswitches. 66. Kubodera, T. et al. Thiamine-regulated gene expression of Aspergillus oryzae thiA requires splicing of the intron containing a riboswitch-like domain in the 5?-UTR. FEBS Lett. 555, 516?520 (2003). 67. Gusarov, I. & Nudler, E. The mechanism of intrinsic transcription termination. Mol. Cell 3, 495?504 (1999). 68. Yarnell, W. S. & Roberts, J. W. (1999) Mechanism of intrinsic transcription termination and antitermination. Science 284, 611?615 (1999). 69. White, H. B. III Coenzymes as fossils of an earlier metabolic state. J. Mol. Evol. 7, 101?104 (1976). 70. White, H. B. III in The Pyridine Nucleotide Coenzymes. 1?17 (Academic Press, New York, 1982). 71. Benner, S. A., Ellington, A. D. & Tauer, A. Modern metabolism as a palimpsest of the RNA world. Proc. Natl Acad. Sci. USA 86, 7054?7058 (1989). 72. Jeffares, D. C., Poole, A. M. & Penny, D. Relics from the RNA world. J. Mol. Evol. 46, 18?36 (1998). 73. Rodionov, D. A., Vitreschak, A. G., Mironov, A. A. & Gelfand, M. S. Comparative genomics of thiamin biosynthesis in prokaryotes. New genes and regulatory mechanisms. J. Biol. Chem. 276, 5093?5100 (2002). 74. Mandal, M. & Breaker, R. R. Adenine riboswitches and gene activation by disruption of a transcription terminator. Nature Struct. Mol. Biol. 11, 29?35 (2004). Demonstration of a riboswitch that activates gene expression. 75. Johansen, L. E., Nygaard, P., Lassen, C., Agers�, Y. & Saxild, H. H. Definition of a second Bacillus subtilis pur regulon comprising the pur and xpt-pbuX operons plus pbuG, nupG (yxjA), and pbuE (ydhL). J. Bacteriol. 185, 5200?5209 (2003). � 2004 Nature Publishing Group NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 5 | JUNE 2004 | 463 REVIEWS 76. Grundy, F. J. & Henkin, T. M. The T box and S box transcription termination control systems. Frontiers Biosci. 8, D20?D31 (2003). 77. Rodionov, D. A., Vitreschak, A. G., Mironov, A. A. & Gelfand, M. S. Regulation of lysine biosynthesis and transport genes in bacteria: yet another RNA riboswitch? Nucleic Acids Res. 31, 6748?6757 (2003). 78. Winkler, W. C., Nahvi, A., Roth, A., Collins, J. A. & Breaker, R. R. Control of gene expression by a natural metabolite- responsive ribozyme. Nature 428, 281?286 (2004). A riboswitch that uses the action of a ribozyme to control gene expression. 79. Barrick, J. E. et al. New motifs suggest an expanded scope for riboswitches in bacterial genetic control. Proc. Natl Acad. Sci. USA 101, 6421?6426 (2004). 80. Gottesman, S. et al. Small RNA regulators of translation: mechanisms of action and approaches for identifying new small RNAs. Cold Spring Harbor Symp. Quant. Biol. 66, 353?362 (2001). 81. Mass�, E., Majdalani, N. & Gottesman, S. Regulatory roles for small RNAs in bacteria. Curr. Opin. Microbiol. 6, 120?124 (2003). 82. Ben-Asouli, Y., Pel-Or, Y., Shir, A. & Kaempfer, R. Human interferon-? mRNA autoregulates its translation through a pseudoknot that activates the interferon-inducible protein kinase PKR. Cell 108, 221?232 (2002). 83. Gold, L., Brodey, E., Heilig, J. & Singer, B. One, two, infinity: genomes filled with aptamers. Chem. Biol. 9, 1259?1264 (2002). 84. Johansson, J. et al. An RNA thermosensor controls expression of virulence genes in Listeria monocytogenes. Cell 110, 551?561 (2002). This paper and references 85 and 96 show that domains within certain mRNAs serve as thermosensing genetic switches. 85. Chowdhury, S., Ragaz, C., Kreuger, E. & Narberhaus, F. Temperature-controlled structural alterations of an RNA thermometer. J. Biol. Chem. 278, 47915?47921 (2003). 86. Henkin, T. M. & Yanofsky, C. Regulation by transcription attenuation in bacteria: how RNA provides instructions for transcription termination/antitermination decisions. Bioessays 24, 700?707 (2002). 87. Yanofsky, C. Using studies on tryptophan metabolism to answer basic biological questions. J. Biol. Chem. 278, 10859?10878 (2003). 88. Wassarman, K. M., Repoila, F., Rosenow, C., Storz, G. & Gottesman, S. Identification of novel small RNAs using comparative genomics and microarrays. Genes Dev. 15, 1637?1651 (2001). 89. Lyubetskaya, E. V., Leont?ev, L. A., Gelfand, M. S. & Lyubetsky, V. A. Search for alternative RNA secondary structures regulating expression of bacterial genes. Mol. Biol. 37, 707?715 (2003) (translated from Russian). 90. Klein, R. J. & Eddy, S. R. RSEARCH: Finding homologs of single structured RNA sequences. BMC Bioinformatics 4, 44 (2003). 91. Vogel, J. et al. RNomics in Escherichia coli detects new sRNA species and indicates parallel transcriptional output in bacteria. Nucleic Acids Res. 31, 6435?6443 (2003). The isolation of numerous small non-coding RNAs in bacteria offers the possibility that a rich diversity of RNAs might participate in various cellular functions. 92. Lee, J. F., Hesselberth, J. R., Meyers, L. A. & Ellington, A. D. Aptamer database. Nucleic Acids Res. 32, D95?D100 (2004). 93. Jenison, R. D., Gill, S. C., Pardi, A. & Polisky, B. High- resolution molecular discrimination by RNA. Science 263, 1425?1429 (1994). An early demonstration of the molecular recognition power of RNA aptamers. 94. Kiga, D., Futamura, Y., Sakamoto, K. & Yokoyama, S. An RNA aptamer to the xanthine/guanine base with a distinctive mode of purine recognition. Nucleic Acids Res. 26, 1755?1760 (1998). 95. Burgstaller, P. & Famulok, M. Isolation of RNA aptamers for biological cofactors by in vitro selection. Angew. Chem. Int. Ed. Engl. 33, 1084?1087 (1994). 96. Morita, M. T. et al. Translational induction of heat shock transcription factor ? 32 : evidence for a built-in RNA thermosensor. Genes Dev. 13, 655?665 (1999). Acknowledgements We thank members of the Breaker laboratory for helpful discus- sions and comments on the manuscript. Competing interests statement The authors declare that they have no competing financial interests. Online links DATABASES The following terms in this article are linked online to: Entrez: http://www.ncbi.nlm.nih.gov/entrez btuB | glmS | lac repressor | prfA | pbuX | rpoH | tyrS | ydhL | xpt Access to this links box is available online. � 2004 Nature Publishing Group "
Add Content to Group
|
Bookmark
|
Keywords
|
Flag Inappropriate
share
Close
Digg
Facebook
MySpace
Google+
Comments
Close
Please Post Your Comment
*
The Comment you have entered exceeds the maximum length.
Submit
|
Cancel
*
Required
Comments
Please Post Your Comment
No comments yet.
Save Note
Note
View
Public
Private
Friends & Groups
Friends
Groups
Save
|
Cancel
|
Delete
Please provide your notes.
Next
|
Prev
|
Close
|
Edit
|
Delete
Genetics
Gene Inheritance and Transmission
Gene Expression and Regulation
Nucleic Acid Structure and Function
Chromosomes and Cytogenetics
Evolutionary Genetics
Population and Quantitative Genetics
Genomics
Genes and Disease
Genetics and Society
Cell Biology
Cell Origins and Metabolism
Proteins and Gene Expression
Subcellular Compartments
Cell Communication
Cell Cycle and Cell Division
Scientific Communication
Career Planning
Loading ...
Scitable Chat
Register
|
Sign In
Visual Browse
Close
Comments
CloseComments
Please Post Your Comment