Proteins are not the only regulators of metabolite synthesis — some RNA molecules do it too. These RNAs lack chemical diversity, so how do we explain the variety of their respective substrates?
The traditional view of RNA as a passive messenger in the transfer of genetic information has long been abandoned. Yet every few years we are still surprised by the discovery of another function that RNA performs. Some bacterial messenger RNAs directly regulate the expression of proteins involved in the syn-thesis of certain metabolic products. This regulation responds to changing levels of the metabolites in the cell, a type of feedback mechanism that was previously known to be mediated only by proteins. These RNAs — known as riboswitches — elegantly couple metabolite recognition with gene regulation in the absence of protein helpers. In this issue, Serganov et al. (page 1167)1 and Montange and Batey (page 1172)2 describe the structures of two distinct classes of riboswitch, providing remarkable insight into RNA metabolite recognition and the control mechanism for gene expression.
Riboswitches are not rare — they account for about 2–3% of genetic control in bacteria3, and are also found in archaea, fungi and plants4. These regulatory RNA domains are typically located in regions of mRNA that directly precede the protein-coding sequence. They use a simple modular architecture — a metabolite-sensing domain recognizes the substrate, and another region, known as the gene-expression signal, regulates protein production in response to substrate binding. When the substrate binds to the metabolite-sensing domain, a structural reorganization of the RNA occurs that unveils (or sometimes masks) the gene-expression signal. In many cases, the end result is suppression of the production of enzymes that are responsible for the biosynthesis of the detected metabolite (Fig. 1).
When riboswitches were first discovered, two questions immediately arose: how is the metabolite recognized, and how does binding of the substrate trigger genetic regulation? Preliminary answers were provided by the structures of guanine- and adenine-sensing riboswitches5,6 — guanine and adenine are organic bases collectively known as purines. The most astonishing finding was the extra-ordinary selectivity achieved by these RNAs: for instance, the guanine-sensing riboswitch binds to guanine with 100,000-fold greater affinity than it does to adenine, a remarkable feat considering the close chemical similarity of the two purine substrates. Amazingly, the strict selectivity is completely reversed by the substitution of only a single RNA base in the binding site. This exquisite molecular recognition is accomplished by completely encapsulating the substrate in the riboswitch, surrounding the substrate with specific RNA contacts.
The two structures presented in this issue1,2 provide even more remarkable examples of RNA molecular recognition. Serganov et al.1 describe a riboswitch that senses thiamine pyrophosphate (TPP, an enzyme cofactor derived from vitamin B1) in bacteria. The structure of a TPP-sensing riboswitch in plants has also been reported elsewhere7. Montange and Batey2 study a riboswitch that binds S-adenosylmethionine (SAM, a cofactor that donates a single carbon unit and that is required for many enzymes). Both TPP and SAM are chemically and structurally more complex than purines. Notably, TPP carries two negatively charged phosphate groups. In contrast, molecules that bind to RNA are typically decorated with positively charged groups that provide favourable electrostatic inter-actions with the negatively charged RNA.
The TPP-sensing structures1,7 show that RNA recruits positively charged metal ions to mediate otherwise unfavourable electrostatic interactions, resembling the use of metal ions to glue together other RNA structures8. Just like guanine or adenine in the purine-sensing riboswitches, TPP and SAM become deeply buried in the RNA structures, embedded between interacting helices. The similar mode of substrate recognition observed in diverse riboswitch classes suggests a common functional mechanism, whereby conformational adaptability enables the RNA to encapsulate the substrate.
Highly specific substrate recognition by proteins exploits an arsenal of 21 amino acids that vary significantly in charge, size and polarity. By contrast, RNA has only four components at its disposal — adenine, guanine, cytosine and uracil — that are all similar in size and chemistry. The riboswitches described in this issue1,2 demonstrate how RNA overcomes its limited chemical diversity by adaptively folding into remarkably complex architectures to create highly specific binding pockets. The overall architectures created by these two riboswitches are surprisingly similar and resemble small RNA enzymes (ribozymes).
The close structural similarity of riboswitches from bacteria1 and plants7 emphasizes how difficult it may be for a single RNA molecule to combine substrate recognition with genetic control. This similarity has fuelled interest in riboswitches as attractive targets for antibacterial drugs, particularly because there is an apparent absence of such RNA systems in humans. Indeed, a TPP analogue (pyrithiamine pyrophosphate, PTPP) shows antimicrobial activity by blocking the binding site of the TPP-sensing riboswitch, shutting down thiamine metabolism in bacteria.
Unfortunately, these RNAs may be surprisingly proficient at developing drug resistance. Microbes develop resistance to PTPP by allowing mutations to emerge within the riboswitch, despite the need to preserve the RNA's structural architecture. A single mutation is sufficient to override the inhibitory effects of the drug, rescuing thiamine synthesis1,7. It seems that the structural flexibility that allows a riboswitch to alter its conformation upon substrate binding may also facilitate the emergence of resistance to drugs targeted against it.
Since their discovery a few years ago, numerous riboswitches have been found that specifically recognize substrates as diverse as nucleotides, amino acids, vitamins and co-enzymes9. These RNAs are evolutionarily ancient, and are widespread in bacteria, plants and fungi. Studies of aptamers — RNAs selected experimentally to recognize small molecules — demonstrate that RNA can discriminate between closely related structures at least as well as antibodies can10. It is now clear that natural selection can also produce RNAs with extraordinary substrate specificity. Does this mean that small-molecule drugs could be found that bind to these RNAs and inhibit their function? Would bacteria and fungi become resistant to such drugs, or do riboswitches represent a chink in microbial armour?