Fancy calling a thermometer “cool"? An RNA Thermometer (RNAT) is, technically speaking, similar to a Breguet thermometer − not named after the famous French aviation pioneer, but after his great-great-grandfather Abraham-Louis (1747–1823) − in that it reacts to changing temperatures with thermal expansion or contraction of a strip: a bimetalic strip in Breguet's device and partially double‑stranded RNA in a "RNA Thermometer".
As you know from the "cloverleaf" secondary (2D) and the intricate, capital-L shaped tertiary structure (3D) of transfer RNAs (tRNAs), RNAs are never just a linear string of ribonucleotides but fold into a various structures, often within a single molecule: duplexes, hairpins, internal loops, junctions and single-stranded regions (masterpieces were the elucidation of the complete E. coli 16S rRNA structure by Harry Noller & Carl Woese in 1981, and the "80% version" by Carola Glotz & Richard Brimacombe, published the previous year.) To make things a tad more complicated, RNA folding and hydrogen-bonding of complementary bases can and often does involve non-canonical base-pairing. For the "thermometer" function, RNA duplex regions are especially important because they react sensitively by stabilizing at lower or "melting" at higher temperatures (modulated by the presence of divalent cations as Mg2+, to not forget this.) Thus, different 2D and 3D structures in a single RNA molecule are successively formed or disrupted over a temperature range and, in a messenger RNA (mRNA), translation signals as start codon or ribosome-binding site (RBS, or Shine-Dalgarno sequence) hidden or exposed.
A bit of history, that of the discovery of the first RNA thermometer 30 years ago. And nobody who's familiar with the development of bacterial genetics − or has experienced it in personam − will be surprised that phage Lambda was involved. When entering an E. coli host cell, λ DNA can either "go lytic" and kill its host immediately, or integrate into the host chromosome for an indeterminate period of time (see here in STC for a brief introduction to lysogeny by Merry.) The cIII gene product of phage λ participates in the regulation of the lysogenic pathway. Amos Oppenheim and his coworkers found several mutations in the cIII gene, two of which were located in the 5'-untranslated region of the mRNA and did not change the amino acid sequence (à propos Women in STEM: I should mention that Amos' co-authors included Shoshy Altuvia. She later became a professor at the Hebrew University of Jerusalem, Israel, like her mentor.)
Since one of the mutations in the 5'-untranslated region (5'-UTR) of the cIII transcript, tor862, led to high level of expression while the other, 'mutant 12', only to low level, Altuvia et al. (1989) assumed that alternative mRNA structures determine the rate of cIII translation. According to their model the cIII mRNA exists at equilibrium in two alternative structures, A and B (Figure 1). In structure B the mRNA is efficiently translated, whereas in the alternative structure, A, it is unavailable for ribosome binding. The mutation tor862 leading to overexpression of cIII locks the mRNA in structure B, whereas 'mutant 12' leading to a low level of expression lock the RNA in structure A. They tested and validated their model by various experiments, including footprinting with specific RNases and ribosome-binding assays, but here I focus on their hybridization experiments. They hybridized matching (radiolabeled) oligonucleotides to three predicted single-stranded regions, performed primer extension reactions (with deoxyribonucleotides), digested the products with RNase H (specific for RNA:DNA hybrids), and separated the reaction products on gels. Figure 1 shows that oligonucleotide "3" (complementary to the Shine-Dalgarno sequence) binds efficiently to the RNA of tor862 (structure B) but does bind to the RNA of 'mutant 12' (structure A). The oligonucleotide complementary to the loop of structure A binds most efficiently to the mRNA of 'mutant 12' (structure A), whereas the oligonucleotide complementary to the loop of structure B shows the strongest signal with RNA of mutant tor862 (structure B). Wild-type mRNA displays an intermediate level of binding to all three oligonucleotides. Clearly, cIII expression is controlled at the level of translation through regulation of the ratio between the two alternative structures. Wild-type cIII mRNA did not show this "intermediate behavior" but a clear shift from the less heat-stable structure B (ΔG0=−18.1 kcal/mol) to the slightly more heat-stable structure A (ΔG0=−19.6 kcal/mol) when Altuvia et al. performed RNase T1 footprinting at 37°C and 42°C. Voilá, their predicted and experimentally proven structure made sense to explain − in part, and in somewhat shortened fashion, because with λ everything is a bit more tricky − the temperature-dependent switch to the lytic pathway of phage λ. Such temperature-dependent hiding or exposing of translation signals by changing 2D/3D structures in the 5'-untranslated region of an mRNA is what makes an "RNA thermometer".
The term "RNA thermometer" was coined by Gisela Storz in a paper summarizing her efforts and those of others before to elucidate the intricate post-transcriptional regulation of expression of the E. coli rpoH gene encoding the heat-shock sigma factor, σ32. Isn't it cool (again, sorry!): translation of a heat shock protein, which is a key regulator for the expression of a whole range of heat shock proteins, is itself regulated by temperature, that is, by a thermometer in the 5'-UTR of its mRNA. We now know of several more examples across various bacterial families, which Franz Narberhaus characterizes in a recent review: "Typical RNA thermometers control translation initiation of heat shock or virulence genes by forming a secondary structure that traps the ribosome binding site (RBS). An increase in temperature to 37°C (virulence genes) or higher (heat shock genes) destabilizes the structure, liberates the RBS and permits formation of the translation initiation complex." It is not without a certain irony that of all things the first proven "RNA thermometer" works the other way round: structure A in the λ cIII mRNA is stable at 42°C and prevents its translation.