Do mRNAs act as direct sensors of small molecules to control their expression?

T paper in this issue of PNAS by Miranda-Rios et al. (1) demonstrates the importance of a conserved RNA structure in the regulation of genes involved in thiamin biosynthesis. Thiamin, also known as vitamin B1, is a cofactor for many important enzymes and therefore essential for growth. Bacteria have genes for all the enzymes necessary to synthesize thiamin, but if adequate amounts are present in the environment, they can use those rather than make their own, saving the energy and materials that would otherwise be used for synthesis. Such feedback inhibition, where the product of a pathway can repress the expression of the enzymes in the pathway, is quite common in bacteria. Many of the most interesting results in molecular biology over the past 50 years have come from unraveling the mechanisms of such feedback regulation. Although bacteria, at least many species, have the ability to make all of the complex molecules they need for growth from simple compounds, they can also use environmental sources of those molecules and, in doing so, repress synthesis of the genes required to make them. Such a regulatory response requires, at a minimum, a means of sensing the concentration of the product and a mechanism to control the expression of the relevant genes that depends on that concentration. In the case of thiamin regulation, the data suggest that the mRNA for the synthesizing enzymes may itself serve as the sensor and provide the mechanism for regulation. Although each regulatory feedback loop has its own features, some mechanisms are quite common. Typically, there is a regulatory protein that can sense the level of the product and then bind to the DNA or RNA to affect the expression of the relevant enzymes. For example, the trp repressor of Escherichia coli binds to DNA only if it is first bound by the amino acid tryptophan (2). Higher concentrations of tryptophan increase the probability of the repressor binding to the DNA where it turns off expression of the genes needed to synthesize tryptophan. This is an example of transcriptional control where regulation affects the activity of the promoter and the amount of mRNA made. Regulation by thiamin is shown to be posttranscription initiation (1). The amount of mRNA initiated from the promoter appears unaffected by the presence of thiamin, but the elongation of the mRNA is attenuated by a terminator structure in the middle of the first gene of the operon, thiC. In Bacillus subtilis, the trp operon is regulated by a similar attenuation process (2, 3). When the TRAP protein binds tryptophan, it can then bind to the mRNA and lead to its premature termination, inhibiting the expression of all the genes required for tryptophan synthesis. E. coli also uses an attenuation process as a second level of control, with the ribosome sensing the concentration of tryptophan in the cell (actually the concentration of charged tRNATrp) (2). In thiamin regulation, the attenuating termination structure forms when translation initiation of the thiC gene is blocked (1). When the mRNA is translated by a ribosome, the structure will not form, but thiamin somehow inhibits ribosome initiation, thereby leading to premature termination of the mRNA. Examples of repressing translation initiation are also well known. Several ribosomal proteins are known to bind to their own mRNA and repress their own synthesis if their primary binding target, the ribosomal RNA, is not available (4). More relevant to the case of thiamin is an example where a sensor protein, in the presence of a sufficient effector molecule, blocks translation initiation. The B. subtilis TRAP also performs this function at the trpG gene, where its binding site overlaps the translation initiation region and, at high concentrations of tryptophan, it binds to the mRNA and blocks ribosome binding (3, 5). So regulation by thiamin appears to have several features that have been observed previously, and the critical open question is the mechanism by which thiamin inhibits translation initiation of the thiC gene. The Role of the thi Box. The mRNA for the thiCOGE operon in Rhizobium etli contains a 211-base leader with two features that are likely to be important. One is a hairpin structure that is just 59 of the thiC-initiating AUG and overlapping the ribosome-binding site (RBS), where one expects it would inhibit ribosome binding. The other feature is called the ‘‘thi box,’’ which is a site of 38 bases that is highly conserved in the region 59 to the start of many genes, from many species, that are involved in thiamin biosynthesis. That the thi box is important in the mRNA, rather than the DNA, is demonstrated by comparison of the different occurrences of it. Only about half of the positions are completely conserved, but all of the sites can fold into the same RNA secondary structure because of compensating changes that maintain complementarity. Because this structure is conserved in different thiamin-related operons and different species rather than the RBS structure, it is most likely the one directly involved in the response to thiamin concentration. However, both structures are required for proper regulation. In an mRNA without the thi box, translation of the thiC gene is inhibited, presumably because of the RBS structure. Somehow the thi box relieves the intrinsic inhibitory effect of the RBS structure in the absence of thiamin but allows it to occur in the presence of thiamin. This can be explained by alternative RNA structures, in which the switch between them is modulated by thiamin. But what sets this example apart from other known posttranscriptional regulatory events is the absence of any regulatory protein. This would be simply another interesting example of a sensory-regulatory mechanism, except there is no protein to do the sensing. It is certainly possible that a regulatory protein