Eric C. Lai

545 Life Sciences Addition, University of California, Department of Molecular and Cell Biology, Berkeley, CA 94720-3200, USA

E-mail:  lai@fruitfly.org

Recent years have seen the discovery of myriad new regulatory and catalytic functions for RNA. A series of new studies have now demonstrated that certain RNA sequences can directly sense ambient temperature or any of a variety of small molecule metabolites. Remarkably, these sensors allow associated mRNAs to regulate their own transcription or translation accordingly, without the need for regulatory proteins.

According to traditional genetic dogma, nucleic acids are the blueprint and instructions for a cell, while proteins
carry out enzymatic and regulatory functions. This simple textbook view has been greatly challenged by work on RNA over the past two decades. Although its mainstay will always be as the intermediate in the DNA–RNA–protein trinity, new functions and activities for RNA are regularly being discovered. In extant living organisms, distinct RNA molecules directly perform or mediate enzymatic processes such as RNA cleavage, splicing and translation [1,2] , and non-coding RNAs are also involved in a tremendous variety of gene regulatory mechanisms that operate at both the DNA and mRNA level [3–5] . In the test tube, a staggering number of additional activities of RNA have been selected for in vitro, ranging from nucleotide synthesis, polymerase activity, peptide bond formation and specific recognition of small molecule and protein ligands [6–8]. RNA is truly a renaissance molecule.

A new story has emerged from work performed in the Breaker, Nudler, Yura and Cossart laboratories. In a
series of recent papers, they report that specific RNA sequences can act as environmental sensors of vitamin
cofactors (including vitamins B₁, B₂ and B₁₂) and temperature, which allow them to directly regulate the
transcription or translation of associated mRNAs [9–14] . Amazingly, these RNAs perform both sensing and
regulatory functions without the need for any proteins. If it is true that the simplest solution to a problem is also
the most beautiful, then these examples are surely among the most beautiful strategies for gene regulation.

The machinery responsible for synthesizing and/or importing a wide variety of essential small molecule
metabolites and vitamin cofactors in bacteria is under negative feedback control. When availability of such small molecules is low, the transcription/translation of the relevant genes to allow their synthesis or import is activated. Conversely, in times of plenty, there is no need to maintain high level production or import capability, and the relevant genes are down regulated.

The classic mechanism of negative feedback control is for a transcription factor to be allosterically regulated by
the small molecule in question. For example, a repressor might become functional in the bound state, allowing it
to suppress transcription in the presence of the appropriate molecule. Biotin synthesis is one of many processes
regulated in this manner: birA encodes a bifunctional protein that is not only a biotin–protein ligase, but a
biotinyl-5'AMP-activated transcriptional repressor of the biotin synthesis operons bioA and bioBCDF. BirA
represses biotin synthesis when enough biotin is present [15] . Post-transcriptional initiation mechanisms have
also been described, such as regulation of the B. subtilistrpEDCFBA operon by TRAP. Tryptophan-bound
TRAP can bind the nascent trp mRNA and cause its premature termination at a site in the first gene of the
operon. TRAP also functions by a slightly different mechanism to regulate trpG, where tryptophan-bound
TRAP can interact with the Shine-Dalgarno (SD) sequence and inhibit ribosome binding and translation [16] .

Specific sequences in the 5' untranslated regions (UTRs) of many vitamin-related genes are essential for
negative feedback regulation, and are conserved amongst a wide variety of Gram-positive and Gram-negative
bacteria ( Table 1). For instance, many genes in the vitamin B₁₂ (CN-Cbl) pathway are negatively regulated by
the intracellular concentration of 5'-deoxy-5'adenosyl-cobalamin (Ado-Cbl) [17] . These include genes
involved in Cbl biosynthesis and cellular import; regulation involves a conserved sequence termed the 'B₁₂ box'. Similarly, genes involved in vitamin B₁ (thiamine) synthesis or metabolism are regulated by the 'THI box', which responds to thiamine pyrophosphate (TPP) levels [18] . Finally, many genes involved in vitamin B₂
(riboflavin) synthesis or import contain a so-called 'RFN element', which is sensitive to the biologically active
riboflavin derivatives flavin mononucleotide (FMN) and flavin adenine dinucleotide (FAD) [19] . Mutation of
any of these motifs decreases or eliminates vitamin-induced repression. Moreover, these motifs are always
associated with genes whose function is relevant to the corresponding metabolite, so that they have also been a
useful diagnostic for implicating certain previously uncharacterized genes in vitamin cofactor transport,
biosynthesis or metabolism [19–21] .

Table 1. Natural small molecule-regulated riboswitches.
Available on-line:  http://www.current-biology.com/cgi/content/full/13/7/R285

The locations of these different motifs in transcribed regions suggest that they function at the RNA level. This is bolstered by the observation that, over evolution, different examples of a given motif display covariant
nucleotide substitutions that maintain a predicted RNA secondary structure. In virtually all other cases,
conserved regulatory sequences such as these represent binding sites for either regulatory proteins or small
antisense RNAs. Suspicion has mounted over the years, however, that the motifs in these genes might directly
bind the vitamin derivatives that regulate them [22,23] . This idea has grown in large part from the failure to
identify mutations in any putative negative regulatory factors, either RNA or protein. As a general rule, the only mutants that show a defect in vitamin-cofactor-induced repression and map outside of the cis-regulatory
sequences themselves turn out to affect proteins involved in vitamin import, biosynthesis or salvage [24,25] .
This is consistent with the known role for vitamin derivatives in repression of gene activity, but does not
implicate any candidate repressors. Although a negative result always carries with it a certain amount of doubt,
this result may be more meaningful with bacteria, where saturating mutagenesis screens can be trivially performed.

More compelling, although still indirect, evidence comes from highly simplified assays that recapitulate regulation by vitamin cofactors. For example, Ado-Cbl can directly inhibit ribosome binding to btuB mRNA in an in vitro system containing only these components [23] . In addition, FMN was shown to induce transcriptional termination of rib-leader–lacZ fusion mRNAs in a simple transcriptional assay containing only these components and highly purified RNA polymerase holoenzyme; TPP had similar effects on transcription from THI box-containing tenA leader templates [12] . Data such as these indicate very direct effects of these
metabolites on gene expression, and suggest that other cofactors may not be necessary for them to exert their
regulatory effect.

In vitro selection techniques have allowed researchers to experimentally define RNA molecules, known as
'aptamers', that directly and specifically bind a variety of small molecules, including vitamin cofactors,
nucleotides, amino acids and antibiotics [8] . FMN is actually one of the compounds for which an RNA
aptamer has been defined [26] . So mRNA is certainly capable of functioning as a sensor. Until very recently,
however, the question has remained open as to whether this actually occurs in nature. Several recent reports
from the Breaker and Nudler groups answer this resoundingly in the affirmative [9,10, 12,13] .

Obtaining evidence for vitamin cofactor-binding to RNA by using cross-linking methodology has been
problematic to date. But a variety of other techniques strongly indicate that these conserved RNA motifs are
involved in direct recognition of vitamin cofactors. One strategy involves deduction of free RNA structures by
base hydrolysis or RNAse T1 cleavage in the absence or presence of vitamin cofactors. Internucleotide bonds
are less susceptible to cleavage in structured regions, such as base-paired stems, more so in unstructured
regions or in certain regions of tertiary structure. Using this type of analysis, the Breaker group [10]
demonstrated that Ado-Cbl actively restructures the E. colibtuB leader sequence through the B₁₂ box [10] .
As expected if this were to reflect direct binding, the restructuring was shown to be a saturable phenomenon,
with an apparent K_D of ~300 nM. In addition, btuB RNA was shown to shift the distribution of radiolabelled
Ado-Cbl in an equilibrium dialysis chamber.

These data argue for direct binding of Ado-Cbl by the btuB RNA leader, resulting in allosteric reconformation
of the RNA. Similar tests by Breaker's lab demonstrated that TPP is bound by THI boxes in the E. coli thiM
and thiC leaders, with an apparent K_D of 100–600 nM [9] , and that FMN is bound by RFN elements in the
B. subtilis ribD and ypaA leaders, with an apparent K_D of <10 nM [13] . Interestingly, the affinity of the
RFN element for FMN is over fifty-fold greater than that of an FMN aptamer independently selected in vitro
[27] , indicating that nature evolved a superior product.

Nudler's group [12]  has also demonstrated direct binding of small metabolites by RNA. Their work focused
primarily on binding of FMN to the B. subtilis rib operon. Using an oligo-annealing approach to discriminate
between free and duplexed regions of RNA, they showed that the RFN leader adopts different structures in the
absence or presence of FMN. Binding of FMN by the RFN leader was further inferred through the ability of
RFN-containing RNA to actively quench the fluorescence of FMN in solution.

It is notable that, in all of these cases, a high level of specificity in molecular recognition by RNA was
demonstrable. Ligand discrimination is essential in any biological setting, of course, but it is especially important during feedback regulation of biosynthetic pathways. Sensors must recognize only the biologically active vitamin derivative, and not any of the many structurally related biosynthetic intermediates in each pathway. In vitro selection techniques have previously established such capability for specific RNA aptamers. In one of the more spectacular examples, a particular RNA aptamer was shown to bind theophylline 10,000 times more avidly than the related molecule caffeine, though the latter contains but a single extra methyl group [28] .

In the new studies, the B₁₂ box was shown to be specific for cobalamin analogues containing a 5' deoxyadenosyl group, and does not bind the related compounds methylcobalamin or cyanocobalamin [10] .
Similarly, recognition of TPP by THI boxes is 1000 times greater than for either thiamine or thiamine
monophosphate [9] , while the RFN element distinguishes FMN from riboflavin by three orders of magnitude,
and even discriminates between FMN and FAD by at least sixty fold [12,13] . The absolute requirement for
phosphate groups on TPP and FMN in recognition by RNA is a notable achievement, considering that RNA is
itself polyanionic and thus might be expected to have difficulty binding such compounds.

Vitamin cofactor-binding aptamers are embedded within larger regions that confer vitamin cofactor-mediated
gene regulation, sometimes referred to as 'riboswitches' or 'regulons'. Studies to date have shown that RNA
restructuring induced by vitamin cofactor binding typically has one of two general consequences for gene
regulation. In some cases, transcripts become prematurely terminated – also referred to as attenuation – while in other cases, access of the mRNA to the translational machinery is inhibited ( Figure 1).

Figure 1.  Diverse strategies for gene regulation linked to RNA sensors.

Small molecule-regulated riboswitches have been linked to both transcriptional attenuation (A) and translational inhibition (B) mechanisms, while characterized thermosensors use a translational inhibition strategy (C).

(A) Transcriptional termination mechanism. DNA is depicted as a black double helix, the nascent mRNA transcript as a colored line (red is untranslated leader, green is coding sequence), and RNA polymerase as a green oval. (Top) In the absence of the relevant metabolite, part of the terminator (two blue boxes) is bound by an anti-terminator (green box) and is non-functional. Transcription proceeds past the poly-uridine stretch following the terminator sequence and extends into coding sequence. (Bottom) In the presence of the relevant metabolite (red circle), ligand-bound RNA adopts an alternative conformation in which the anti-terminator is bound by an anti-anti-terminator (black box). This allows formation of the structurally distinctive terminator hairpin, which induces release of RNA polymerase following the poly-uridine tract.

(B) Translational inhibition mechanism. The mRNA is depicted with the colored line; untranslated leader in red and the coding region in green. (Top) In the absence of cofactor-binding, the anti-Shine-Dalgarno (anti-SD) ribosome binding sequence is sequestered by an anti-anti-SD sequence; this conformation allows ribosome (double green oval) binding to the SD box and translation to occur. (Bottom) Cofactor binding (red circle) restructures the RNA so that the anti-SD sequence is allowed to pair with the SD box, thus inhibiting translation. Note that variations on both mechanisms exist, as may dual regulatory mechanisms involving both transcriptional attenuation and translational inhibition (see text).

(C) A thermosensor from L. monocytogenes prfA. (Top) A stem structure adjacent to the SD box prevents
translation at lower temperatures. (Bottom) The increase in ambient temperature following host infection melts this structure, allowing the ribosome to access the SD box and translate prfA.

In genes regulated by translational inhibition, these elements are not coupled to a terminator/anti-terminator
system. Instead, the mRNA leader is capable of adopting alternative conformations that include the SD box
and/or AUG translational initiation codon ( Figure 1B). In the absence of ligand, the SD/AUG sites are exposed
and accessible to the ribosome ( Figure 1B, top). But in the presence of the appropriate cofactor, an anti-SD
sequence is allowed to bind the SD sequence, thus denying the ribosome access to the mRNA and preventing
translation ( Figure 1B, bottom). In some cases, the anti-SD sequence is paired with an anti–anti-SD sequence
in unbound mRNA to ensure accessibility to the ribosome. A translational inhibition mechanism has been
proposed for regulation of E. coli btuB and the S. typhimurium cob operon by Ado-Cbl, and for E. coli thiM
by TPP [9,10, 23, 30] .

Bioinformatic searches revealed the presence of B₁₂ boxes, THI boxes and RFN elements in about 80
genomes from a phylogenetically broad range of bacteria [19–21] . Analysis of the type of likely regulatory
system associated with these elements revealed a curious generalization: Gram-positive bacteria are more likely to couple these elements to a terminator/anti-terminator system, while Gram-negative bacteria more typically link these elements to a SD-sequestration mechanism. Exceptions to this tendency occur, some of which weretaken as evidence for horizontal gene transfer between bacterial subtypes. It has also been noticed that operons are more likely to be regulated by a transcriptional termination system, while single genes are more typically regulated by a translational inhibition mechanism. This trend would appear to be metabolically frugal, as it is more efficient to prematurely cease transcription of a long, multigene biosynthetic operon if its products are not needed.

The two generalizations are manifest in the observation that riboflavin synthesis genes in many Gram-positive
organisms, such as B. subtilis, are arranged in operons, while the same genes in a Gram-negative organism such as E. coli are instead scattered about its genome. But neither generalization is to be taken as absolute, and as noted, many exceptions exist. THI boxes are also found in some archaeal and eukaryotic genomes, including those of certain fungi and plants. This indicates that gene regulation by cofactor-binding RNA aptamers was quite an ancient innovation.

Variations on these regulatory mechanisms exist. For example, although B. subtilis ypaA was suggested to be
controlled by FMN through a translational inhibition mechanism [13] , microarray analysis supported an
attenuation mechanism [31] . In fact, it was noticed that, in many species, the ypaA leader can form a
terminator hairpin that also overlaps SD sequences [20] , suggesting that non-terminated ypaA transcripts can
still be regulated at the translational level. E. coli thiC was also shown to be regulated at both the transcriptional attenuation and translational levels [9] , and a similar dual mode of regulation was proposed for regulation of yuaJ by TPP in many species [20] . Other possibilities include direct sequestration of SD boxes by
metabolite binding aptamers (predicted for many THI boxes in actinomycetes, cyanobacteria and
thermoplasmas [20]  as well as for RFN elements of T. thermophilius ribD and A. minutum ypaA [21] )
or combined modulation of SD and translational enhancer availability [30] .

An additional layer of regulatory complexity comes with the observation that the affinity of an aptamer for its
ligand, and thus the quality of regulation, can be dynamic with procession of transcription [9, 12] . Finally,
there is no reason, at least in principle, why RNA sensors of these sorts could not also confer positive gene
regulation. At least one microarray-based study of B. subtilis revealed only vitamin-repressed genes [31] .
This type of experiment would not, however, have detected examples of translational regulation. We therefore
await further studies on regulatory possibilities associated with riboswitches.

The Heat Is On: Activation of Gene Expression by RNA Thermosensors

There are two well-studied examples of temperature-dependent induction of gene expression and/or activity:
during the heat shock response and during pathogenic invasion. The need to control gene induction as a function of temperature in the former setting is self-explanatory; the rationale in the latter setting is also sensible, if not necessarily immediately obvious to the uninitiated. It is most efficient to keep virulence-associated genes transcriptionally silent until the pathogen enters an animal host, and a convenient way of detecting this turns out to be the increase in ambient temperature to ~37°C upon host entry. A heat-inducible multigene transcriptional response in either setting is typically achieved by placing the desired genes under the common control of a transcription factor, the expression or activity of which is then directly subject to thermal regulation. A surprisingly large number of ways of achieving this desired form of regulation have evolved.

Protein folding and activity are well known to be influenced by temperature, a fact that underlies the isolation of
temperature-sensitive mutants. Temperature controls the multimerization status of heat shock transcription factor (HSF), a protein conserved from yeast to humans which is the sole mediator of the heat shock response. HSF normally exists in an inactive monomer state, but acquires DNA-binding activity following its trimerization at temperatures of ~28–37°C [32] . A related mechanism operates in S. typhimurium. In this pathogen, the
conformation of a coiled-coil domain in TlpA is sensitive to ambient temperature and regulates multimerization
and DNA-binding activity [33] . At host temperature, TlpA becomes monomeric and non-functional.
Although the only known transcriptional target of TlpA thus far is itself, conservation of tlpA in the virulence
plasmids of various Salmonella strains has suggested a possible role in infection. A mechanism involving the
status of DNA supercoiling is used in Shigella. Here, the promoter of S. flexneri virF, which encodes an
activator of VirB – which in turn activates several operons of virulence genes – undergoes a structural shift at
37°C that renders it less accessible to the repressor H-NS and therefore transcriptionally competent [34] .

As RNA secondary structure is highly influenced by temperature, RNA could in principle also act as a
thermosensor. This was suggested to be the case for regulation of rhizobial heat shock genes. Several genes,
including those encoding small heat shock proteins and s³², a global regulator of heat shock responsive genes,
were found to contain a conserved sequence in their 5' untranslated regions referred to as the ROSE
(repression of heat shock gene expression) element [35] . From comparative RNA structure predictions and
mutation studies using fusion constructs, it was proposed that the ROSE element engages the SD box and AUG
start codon in a stem structure, rendering them inaccessible. Melting of this structure at heat-shock-inducing
temperatures would then allow translation of these associated mRNAs [36] .

Functional evidence for this type of model was obtained for translational regulation of E. coli rpoH, which
encodes s³². It was already known that s³² is post-translationally regulated by a negative feedback mechanism involving the chaperone subunits DnaK, DnaJ and GrpE [37] . Although they normally assist in protein folding during extreme conditions, they negatively regulate the activity of s³² by making it unstable. During heat shock, titration of chaperones by other misfolded proteins results in s³² stabilization and accumulation. However, work primarily from the Yura lab [38,39]  showed that activation of rpoH translation is itself independent of chaperones. Instead, it involves a region including the 5' coding sequence that has the capacity to sequester the SD and AUG sites; the sequences involved here are different from the ROSE element. The RNA structures predicted by this model were substantiated using chemical probing techniques.

The Yura lab [14]  subsequently showed that the RNA secondary structure of this region, as deduced from
circular dichroism (CD) spectra, is melted over a heat-inducing range that directly correlates with the
thermoregulatory range in vivo. They were also able to alter the 'thermostat' of this sensor by making various
base substitutions that either increased or decreased the amount of duplexed RNA. Changes in the
thermostability of RNA secondary structures in this region also directly correlated with their thermoregulatory
properties in vivo, with more stable structures responding at higher temperatures, and vice versa. Finally, they
found that the rpoH 5' region inhibits ribosome binding at low temperature – but not at high temperature – in an
in vitro assay containing only mRNA, purified 30S ribosomes and initiator tRNA^Met. These data strongly
support the model that this RNA region directly senses temperature and regulates translation by controlling
access of rpoH transcript to the ribosome.

RNA thermosensors have also been proposed to exist in certain pathogenic bacteria. Nearly a decade ago,
translation of LcrF, a general activator of virulence-related gene expression in Yersinia pestis, was found to be
thermally regulated [40] . An RNA sensor model was proposed based upon structural predictions, but not
subsequently tested experimentally. A new report from the Cossart lab [11]  now provides strong evidence
for an RNA thermosensor that regulates translation of PrfA, a general activator of virulence genes in a
pathogenic variety of Listeria, L. monocytogenes.

The prfA gene is transcribed at both 30°C and 37°C, but it is translated only at the latter temperature. Cossart
and colleagues observed that sequences in the 5' UTR of the prfA mRNA could form an extended hairpin that
includes the SD sequence. Mutations that destabilized this structure reduced or eliminated thermoregulation by
the prfA leader in vivo as well as in a heterologous in vitro translation assay (using E. coli extract, which does
not contain PrfA). Temperature-dependent reconformation of this RNA was indeed demonstrated using gel
mobility assays and chemical probing, and showed that the stem structure surrounding the SD sequence is
melted at 37°C but not at 30°C. Finally, they found that the prfA leader even confers thermal regulation on
heterologous transcripts in living E. coli. Taken together, these data indicate that the prfA mRNA leader is a
thermosensor that directly regulates translation by selectively blocking access of the ribosome at lower, but not
higher, temperatures ( Figure 1C).

Unlike the metabolite-regulated riboswitches, most of these RNA thermoregulators are not broadly distributed,
either within a given genome or among different species. The known RNA thermosensors are all directly
involved in SD/start sequestration, which makes them somewhat less amenable to evolutionary mixing and
matching than the modular riboswitches described above, in which ligand binding capacity and gene regulatory
sequences are separable. In the case of rpoH, involvement of a downstream coding sequence in SD
sequestration further limits its transfer to other functionally distinct genes. Thermoregulation of prfA probably
evolved to fit the specific needs of this pathogen, and this gene is not even present in the genome of the related,
non-pathogenic bacterium L. innocuous [41] . Only the ROSE element is associated with multiple types of
heat-regulated genes in multiple (rhizobial) genomes, although validation of its status as a genuine
thermoregulator will require additional work.

It is human nature to yearn for simpler times, when living required fewer moving parts and gadgets. Perhaps the ultimate extension of this lies in trying to imagine what the very origins of life were. A great deal of evidence from the past two decades has led to the popular idea that an 'RNA world' may have predated DNA/protein-based biological life. The possibility of such an RNA-dominated era is supported by the dual ability of RNA to serve not only as a self-replicating genetic template, but also as a catalyst of a large variety of organic chemical reactions.

The series of new studies reviewed above demonstrates the ability of RNA to act as an environmental sensor,
thus adding a potential layer of regulatory sophistication to the hypothetical RNA world. Indeed, it has been
suggested that these sensors may represent 'molecular fossils' of such a world [9,10] . Notably, the structures
of vitamins B₁, B₂ and B₁₂ are closely related to those of nucleotides, and may have originated during the time
of the RNA world [42] . It seems quite reasonable to hypothesize that they may have originally evolved and
functioned as RNA cofactors or allosteric elements before their eventual adaptation by protein enzymes, where
they are currently essential for the catalysis of a wide variety of chemical reactions in all life.

Additional examples of RNA sensors may yet be found. Many genes involved in methionine and cysteine
biosynthesis contain a so-called 'S box' motif in their 5' leader. Features of S-box-mediated regulation resemble
those of the other vitamin cofactor regulated systems discussed above, including alternative RNA structures
containing a terminator site or a sequestered antiterminator/terminator complex [43,44] . Ongoing work in the
Breaker lab strongly indicates that the S box functions as an S-adenosyl-methionine (SAM)-regulated
riboswitch. Their characterization of the S box from B. subtilisyitJ indicates that it selectively binds SAM (and
not methionine) with an apparent binding constant of approximately 1 nM and mediates transcriptional
termination in the presence of SAM (R. Breaker, personal communication).

Now that these precedents exist, the search will be on for additional classes of RNA riboswitches that detect the cellular availability of other metabolites and regulate gene expression. But what other capabilities exist? For example, although protein sensors of pH, redox state and metal ion concentration are known, do equivalent
RNA sensors exist? And even if such sensors are not to be found in extant life forms, can such sensors be
selected for in vitro? Work in this exciting field will surely further inform us of the potential complexity of an 'RNA world'.

These RNA sensors and switches may find practical application in the present day world as well. One can
imagine exploiting RNA thermoregulation to create heat-inducible transgenes. Although heat-shock-inducible
promoters have long been used for this purpose, expression of inherently heat-inducible transcripts with the
binary Gal4/UAS system [45]  would now allow both spatial and temporal control in ectopic expression
studies. Other riboswitches could also be adapted to create small molecule-regulated transgenes, which may
allow researchers to manipulate expression of any individual construct within a battery of simultaneously
introduced experimental constructs. Previously identified aptamers have been placed within 5' UTRs and shown
to inhibit gene expression upon introduction of the appropriate ligand [46]  by a simple occlusion mechanism.
A little structure-informed engineering may conversely permit small molecules to activate gene activity as well.

These and other RNA aptamers may also be exploited as biosensors. Pioneer studies have shown that
self-cleaving ribozymes can be placed under allosteric control by various small molecules, which can then be
used to analyze the composition of chemical and biological mixtures [47] . Although this technology is little
beyond the 'proof of principle' stage at present, identification of additional and more affine aptamers may
increase the sophistication and practical usage of this concept [48] . Finally, these sensors may also find utility
as targets of antibacterial compounds. It may be possible to design chemical mimics of their cognate ligands that would constitutively repress associated gene activity. Such compounds could potentially be quite powerful, as they would efficiently inhibit bacterial growth by simultaneously repressing multiple components in a given
biosynthetic/metabolic/transport pathway. At the same time, such compounds might be likely to have relatively
low toxicity, as they would be designed to target RNA, not protein. I look forward to the realization of these
tools, and if recent years are any indication, expect to continue to be amazed by new capabilities and
applications for RNA that will undoubtedly emerge in the future.

Acknowledgments

Special thanks to Ron Breaker for sharing many unpublished observations and for useful discussions on this
topic. Jennifer Doudna, Pascale Cossart, Jorgen Johansson, Mikhail Gelfand and Evgeny Nudler also
contributed ideas and critiques. Finally, I acknowledge the gracious support of Gerald Rubin and the Damon
Runyon Cancer Research Foundation, DRG 1632.

[1] Joyce G.F. (2002) The antiquity of RNA-based evolution. Nature, 418:214-221.

[2] Doudna J.A. and Cech T.R. (2002) The chemical repertoire of natural ribozymes. Nature, 418:222-228.

[3] Szymanski M. and Barciszewski J. (2002) Beyond the proteome: non-coding regulatory RNAs. Genome
    Biol., 3

[4] Dernburg A.F. and Karpen G.H. (2002) A chromosome RNAissance. Cell, 111:159-162.

[5] Hannon G.J. (2002) RNA interference. Nature, 418:244-251.

[6] Bittker J.A., Phillips K.J. and Liu D.R. (2002) Recent advances in the in vitro evolution of nucleic acids.
    Curr. Opin. Chem. Biol., 6:367-374.

[7] Jaschke A. (2001) Artificial ribozymes and deoxyribozymes. Curr. Opin. Struct. Biol., 11:321-326.

[8] Hermann T. and Patel D.J. (2000) Adaptive recognition by nucleic acid aptamers. Science, 287:820-825.

[9] Winkler W., Nahvi A. and Breaker R.R. (2002) Thiamine derivatives bind messenger RNAs directly to
    regulate bacterial gene expression. Nature, 419:952-956.

[10] Nahvi A., Sudarsan N., Ebert M., Zou X., Brown K. and Breaker R.R. (2002) Genetic control by a
    metabolite binding mRNA. Chem. Biol., 9:1043-1049.

[11] Johansson J., Mandin P., Renzoni A., Chiaruttini C., Springer M. and Cossart P. (2002) An RNA
    thermosensor controls expression of virulence genes in Listeria monocytogenes Cell, 110:551-561.

[12] Mironov A., Gusarov I., Rafikov R., Lopez L., Shatalin K., Kreneva R.A., Perumov D.A. and Nudler E.
    (2002) Sensing small molecules by nascent RNA: a mechanism to control transcription in bacteria. Cell,
    111:747-756.

[13] Winkler W., Cohen-Chalamish S. and Breaker R.R. (2002) An mRNA structure that controls gene
    expression by binding FMN. Proc. Natl. Acad. Sci. U.S.A., 99:15908-15913.

[14] Morita M.T., Tanaka Y., Kodama T.S., Kyogoku Y., Yanagi H. and Yura T. (1999) Translational
    induction of heat shock transcription factor sigma32: evidence for a built-in RNA thermosensor. Genes Dev.,
    13:655-665.

[15] Cronan J.E. Jr. (1989) The E. colibio operon: transcriptional repression by an essential protein
    modification enzyme. Cell, 58:427-429.

[16] Babitzke P. and Gollnick P. (2001) Posttranscription initiation control of tryptophan metabolism in
    Bacillus subtilis by the trp RNA-binding attenuation protein (TRAP), anti-TRAP and RNA structure. J.
    Bacteriol., 183:5795-5802.

[17] Richter-Dahlfors A.A. and Andersson D.I. (1992) Cobalamin (vitamin B₁₂) repression of the cob operon in Salmonella typhimurium requires sequences within the leader and the first translated open reading frame.
    Mol. Microbiol., 6:743-749.

[18] Miranda-Rios J., Navarro M. and Soberon M. (2001) A conserved RNA structure (thi box) is involved in
    regulation of thiamin biosynthetic gene expression in bacteria. Proc. Natl. Acad. Sci. U.S.A., 98:9736-9741.

[19] Gelfand M.S., Mironov A.A., Jomantas J., Kozlov Y.I. and Perumov D.A. (1999) A conserved RNA
    structure element involved in the regulation of bacterial riboflavin synthesis genes. Trends Genet., 15:439-442.

[20] Rodionov D.A., Vitreschak A.G., Mironov A.A. and Gelfand M.S. (2002) Comparative genomics of
    thiamin biosynthesis in procaryotes: new genes and regulatory mechanisms. J. Biol. Chem., 277:48949-48959.

[21] Vitreschak A.G., Rodionov D.A., Mironov A.A. and Gelfand M.S. (2002) Regulation of riboflavin
    biosynthesis and transport genes in bacteria by transcriptional and translational attenuation. Nucleic Acids        Res., 30:3141-3151.

[22] Stormo G.D. and Ji Y. (2001) Do mRNAs act as direct sensors of small molecules to control their
    expression? Proc. Natl. Acad. Sci. U.S.A., 98:9465-9467.

[23] Nou X. and Kadner R.J. (2000) Adenosylcobalamin inhibits ribosome binding to btuB RNA. Proc. Natl.
    Acad. Sci. U.S.A., 97:7190-7195.

[24] Begley T.P., Downs D.M., Ealick S.E., McLafferty F.W., Van Loon A.P., Taylor S., Campobasso N.,
    Chiu H.J., Kinsland C. and Reddick J.J. et al. (1999) Thiamin biosynthesis in prokaryotes. Arch. Microbiol.,
    171:293-300.

[25] Perkins J. and Pero J. (2002) Biosynthesis of riboflavin, biotin, folic acid and cobalamin. In: Sonenshein
    A., Hoch J. and Losick R. (Eds.) Bacillus subtilis and its closest relatives: from genes to cells. (pp.
    271-286) Washington: ASM Press

[26] Burgstaller P. and Famulok M. (1994) Isolation of RNA aptamers for biological cofactors by in vitro
    selection. Angew. Chem. Int. Ed. Engl., 33:1084-1087.

[27] Fan P., Suri A.K., Fiala R., Live D. and Patel D.J. (1996) Molecular recognition in the FMN-RNA
    aptamer complex. J. Mol. Biol., 258:480-500.

[28] Jenison R.D., Gill S.C., Pardi A. and Polisky B. (1994) High-resolution molecular discrimination by RNA. Science, 263:1425-1429.

[29] Gusarov I. and Nudler E. (1999) The mechanism of intrinsic transcription termination. Mol. Cell,
    3:495-504.

[30] Ravnum S. and Andersson D.I. (2001) An adenosyl-cobalamin (coenzyme-B₁₂)-repressed translational
    enhancer in the cob mRNA of Salmonella typhimurium Mol. Microbiol., 39:1585-1594.

[31] Lee J.M., Zhang S., Saha S., Santa Anna S., Jiang C. and Perkins J. (2001) RNA expression analysis
    using an antisense Bacillus subtilis genome array. J. Bacteriol., 183:7371-7380.

[32] Zhong M., Orosz A. and Wu C. (1998) Direct sensing of heat and oxidation by Drosophila heat shock
    transcription factor. Mol. Cell, 2:101-108.

[33] Hurme R., Berndt K., Normark S. and Rhen M. (1997) A proteinaceous gene regulatory thermometer in Salmonella Cell, 90:55-64.

[34] Falconi M., Colonna B., Prosseda G., Micheli G. and Gualerzi C.O. (1998) Thermoregulation of Shigella and Escherichia coli EIEC pathogenicity. A temperature-dependent structural transition of DNA modulates accessibility of virF promoter to transcriptional repressor H-NS. EMBO J., 17:7033-7043.

[35] Narberhaus F., Kaser R., Nocker A. and Hennecke H. (1998) A novel DNA element that controls
    bacterial heat shock gene expression. Mol. Microbiol., 28:315-323.

[36] Nocker A., Hausherr T., Balsiger S., Krstulovic N.P., Hennecke H. and Narberhaus F. (2001) A
    mRNA-based thermosensor controls expression of rhizobial heat shock genes. Nucleic Acids Res.,
    29:4800-4807.

[37] Hurme R. and Rhen M. (1998) Temperature sensing in bacterial gene regulation–what it all boils down to. Mol. Microbiol., 30:1-6.

[38] Nagai H., Yuzawa H. and Yura T. (1991) Interplay of two cis-acting mRNA regions in translational
    control of sigma 32 synthesis during the heat shock response of Escherichia coli. Proc. Natl. Acad. Sci.
    U.S.A., 88:10515-10519.

[39] Morita M., Kanemori M., Yanagi H. and Yura T. (1999) Heat-induced synthesis of sigma 32 in
    Escherichia coli: structural and functional dissection of rpoH mRNA secondary structure. J. Bacteriol.,
    181:401-410.

[40] Hoe N.P. and Goguen J.D. (1993) Temperature sensing in Yersinia pestis: translation of the LcrF
    activator protein is thermally regulated. J. Bacteriol., 175:7901-7909.

[41] Glaser P., Frangeul L., Buchrieser C., Rusniok C., Amend A., Baquero F., Berche P., Bloecker H.,
    Brandt P. and Chakraborty T. et al. (2001) Comparative genomics of Listeria species. Science,
    294:849-852.

[42] White H.B.I. (1976) Coenzymes as fossils of an earlier metabolic state. J. Mol. Evol., 7:101-104.

[43] Grundy F.J. and Henkin T.M. (1998) 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.

[44] Murphy B.A., Grundy F.J. and Henkin T.M. (2002) Prediction of gene function in methylthioadenosine
    recycling from regulatory signals. J. Bacteriol, 184:2314-2318.

[45] Brand A.H. and Perrimon N. (1993) Targeted gene expression as a means of altering cell fates and
    generating dominant phenotypes. Development, 118:401-415.

[46] Werstuck G. and Green M.R. (1998) Controlling gene expression in living cells through small
    molecule-RNA interactions. Science, 282:296-298.

[47] Seetharaman S., Zivarts M., Sudarsan N. and Breaker R.R. (2001) Immobilized RNA switches for the
    analysis of complex chemical and biological mixtures. Nat. Biotechnol., 19:336-341.

[48] Breaker R.R. (2002) Engineered allosteric ribozymes as biosensor components. Curr. Opin. Biotechnol.,
    13:31-39. 

8. Hovsepian JA, and Frenster JH, "RNA-Induced Melting of DNA during Selective Gene Transcription".

7. Storz G, "An Expanding Universe of Noncoding RNAs".

6. Eddy SR, "Non-Coding RNA Genes and the Modern RNA World".

5. Frenster JH, "Activation of DNA Transcription within Repressed Chromatin".

4. Frenster JH, "Selective Gene De-Repression by De-Repressor RNA".

3. Frenster JH, "Selective Control of DNA Helix Openings during Gene Regulation".

2. Frenster JH, Nakatsu SL, and Masek MA, "Ultrastructural Probes of DNA Templates within Human Bone Marrow and Lymph Node Cells".

1. Frenster JH, "Mechanisms of Repression and De-Repression within Interphase Chromatin".

A. DeCarvalho S, "Effect of RNA from Normal Human Marrow on Leukaemic Marrow In-Vivo".

8. Saha S, Ansari AZ, Jarell KA, and Ptashne M, "RNA Sequences that Work as Transcriptional Activating Regions".

7. Lanz RB, Razani B, Goldberg AD, and O'Malley BW, "Distinct RNA Motifs are Important for Coactivation of Steroid Hormone Receptors by Steroid Receptor RNA Activator (SRA)".

6. Kwek KY , Murphy S , Furger A , Thomas B , O'Gorman W , Kimura H , Proudfoot NJ  and Akoulitchev A, "U1 snRNA Associates with TFIIH and Regulates Transcriptional Initiation".

5. Frenster JH, and Hovsepian JA, "RNA Feedback Mechanisms during Eukaryotic Gene Regulation".

4. Frenster JH, "Yeast RNA Re-Programming of Already-Active Mammalian Chromatin".

3. Frenster JH, "Uni-Polar Clustering of Lymphocyte DNA Templates Toward Neoplastic Target Cells Within Hodgkin's Disease Lymph Nodes".

2. Frenster JH, "Oncogenes as Molecular Targets within Active Chromatin".

1. Frenster JH, "Nuclear RNA Species Activate DNA Transcription within Chromatin".

1. Frenster JH, "Ultrastructural Probes of Active DNA Sites, and the RNA Activators of DNA".