Nanos gigantum humeris insidentes
—Bernard of Chartre, XIIth Century
Directly translating from the Latin, the quoted phrase reads "dwarfs ride on the shoulders of giants." This metaphor and its many derivatives have been used over the centuries to mean that new knowledge is discovered by building on previous discoveries. Most of the time, that is the perception we have when we think of progress in science. But every now and then, a new finding seems to come out of nowhere, greatly challenging prior ideas. Such might be the case on how we came to gain new knowledge regarding the structural organization of the basic processes of transcription and translation in Escherichia coli.
Let me start with a quick recount of old knowledge. For several decades – going back to the early days of molecular biology – the accepted view was that transcription and translation in E. coli were coupled, meaning that as soon as an RNA polymerase began to produce a nascent mRNA, ribosomes bind it and initiate translation. Perhaps the most striking early evidence in support of this coupling was provided by the electron micrographs that Miller and colleagues published in 1970. These images are remarkable and seemed to withstand the test of time. So much so that Elio chose one of them for the first "Pictures Considered" post of STC in 2013. I find the image so beautiful in its simplicity and power that I show it here once again (Figure 1). Indirect evidence for coupling had accumulated throughout the 1960s (including the first description of polysomes by Elio) but this picture came to cement in people's minds the general concept. It all made perfectly sense. E. coli because it had no nuclear membrane – like all Bacteria and Archaea – made this coupling not only possible but almost unavoidable; it was an easy concept to accept. The coupling came to be seen as a defining feature that distinguished members of the domains Bacteria and Archaes from the Eukarya, as very nicely described in 2014 by Nanne Nanninga in another STC post. Transcription-translation coupling was also deemed most appropriate to achieve fast growth rates and it allowed for the evolution of mechanisms whereby transcription was regulated by the state of the ribosomes following close by.
Many features of E. coli gene expression studied extensively in the 1970s fit perfectly in the context of coupled transcription and translation. Let me briefly remind you of the details of my two personal favorites: attenuation and polarity. In attenuation, because of the coupling, the RNA polymerase gets information regarding the state of the translating ribosome and either terminates transcription or continues to transcribe genes downstream (Figure 2). This is because a ribosome loads onto the nascent mRNA and initiates translation of a leader peptide so that it closely trails the RNA polymerase. In the case of attenuation control of amino acid biosynthetic operons, the leader peptide gene is rich in codons for the cognate amino acid. When there are plentiful tRNAs charged with that amino acid, translation of the leader peptide proceeds with the ribosome reaching the termination codon. By doing so, the ribosome masks a portion of the mRNA and the subsequent bases of the mRNA fold into a stem loop structure that tells the RNA polymerase to terminate transcription. However, if the amount of charged tRNAs for the cognate amino acid is low, the ribosome stalls at the corresponding codons. This results in an alternative mRNA structure, an antiterminator, that precludes the formation of the terminator. As a consequence, the RNA polymerase continues transcribing and thus the downstream genes are expressed. It is clear that attenuation absolutely depends on the coupling of transcription and translation.
Transcriptional polarity refers to the phenomenon where non-sense mutations (where translation termination codons replacea sense codon) in genes located in the promoter proximal end of operons cause decreased expression of genes located distal to the promoter. The molecular mechanism of polarity is conceptually similar to attenuation. In wild-type operons, translating ribosomes follow the RNA polymerase very closely and thus naked mRNA is not exposed. But, when a non-sense mutation causes the early release of the translating ribosome, the mRNA is exposed and the termination factor Rho can bind at normally occluded sites, resulting in early transcription termination. Here again, it is clear that this interpretation of polarity depends on the coupling of transcription and translation.
So much for old knowledge, now comes the surprise. It's February 2011 and a paper, led by Keren Nevo-Dinur from the group of Orna Amster-Choder (Hebrew University of Jerusalem), bears the unexpected title "Translation-Independent Localization of mRNA in E. coli." The abstract goes on to state: "In contrast to the view that transcription and translation are coupled in bacteria, our results show that, subsequent to their synthesis, certain mRNAs are capable of migrating to particular domains in the cell where their future protein products are required." What had they done to allow such a bold claim? Their studies were both elegant and thorough. At the heart of their approach was the clever tagging of the mRNA of two genes whose final protein products had two different cellular addresses: lacY, encoding the lactose permease – an integral membrane protein – and cat, encoding the soluble (that is, cytoplasmic) chloramphenicol acetyl transferase. How were the mRNAs tagged? Using a cute trick. Six repeats of a sequence which, when transcribed into RNA, becomes bound by the MS2 phage coat protein are first cloned into the gene of interest. Then, a fusion of the MS2 phage coat protein and the green fluorescence protein (MS2-GFP) is expressed in the cell. Visualizing MS2-GFP fluorescence thus identifies the location of the mRNA. The results were striking: the lacY mRNA localized to the membrane all around the cell periphery while the cat mRNA localized in the cytoplasm in a helix-like pattern (Figure 3). The authors carried out all the required and expected controls to ensure that these results were not artifactual (including following the fate of other mRNAs, for example, one encoding the membrane-bound sugar permease BglF). Then they went a step further. To determine if the mRNA localization was a property inherent to the RNA and not the protein product, they uncoupled transcription and translation through the use of protein synthesis inhibitors. Sure enough, the specific localization of the different mRNAs was not affected by protein synthesis inhibitors! In addition, when they uncoupled transcription and translation by introducing a stretch of rare leucine codons at the beginning of the bglF gene (creating a "translational blockage", they still saw the mRNA localizing to the membrane. All in all, the authors made a very compelling case that untranslated mRNAs can localize to distinct cellular addresses in keeping with where their protein product is to function. With this series of experiments they appeared to have put a major dent on a "defining" feature of cells lacking a nuclear membrane, the Bacteria and Archaea: the coupling of transcription and translation.
Remember, these results were published in 2011. Yet, here at STC (and elsewhere) there was still a strong sense that in Bacteria and Archaea, transcription and translation were coupled. Perhaps the slow acceptance of this new Weltanschauung (worldview) was due both to the strength of the evidence favoring coupling and the fact that only a few genes were analyzed in the 2011 publication. But now, just last week, a paper by Kannaiah et al., also from the Amster-Choder group, presents a global analysis of the spatiotemporal organization of the E. coli transcriptome. To carry out the global analysis they developed a technique that fractionates the cell contents and sequences the mRNA present (they call the approach "Rloc-seq"). The results are impressive, they detect wholesale specific localization of mRNA – at least half of it independent of translation – whereby the mRNA placement correlates with the final destination of the protein product (Figure 4).
In all fairness, to some investigators the lack of coupling was not all that unexpected. As far back as 2000, there were reports that there was clear segregation between the ribosomes and the RNA polymerase within cells. It just seemed that there was no room for most of the ribosomes to be near the nucleoid, where the vast majority of the RNA polymerase could be found.
Do these results completely overturn the prior notion of coupled transcription and translation in cells without a nucleus? Personally, I do not think so. Perhaps that is due to the fact that in this aging mind I carry too much intellectual baggage. Still, I do feel that the evidence for coupling is extremely strong; I hope I argued that case strongly enough at the outset. I am extremely impressed by the papers presenting the results of mRNA localization. But I was surprised that neither of these two key papers offers a discussion of these new results in light of what is known about such processes as attenuation and polarity, which obviously depend on close coupling between transcription and translation. In my own thinking, I look at that compelling electron micrograph of Miller et al. (Figure 1) and see the scale bar. Even a single gene transcript is likely to be around 0.5 microns in length. That's pretty long to have the RNA polymerase still transcribing while the 5' end of the mRNA, loaded with ribosomes, can be reaching a distant cellular address. I think that specific mRNA localization and transcription-translation coupling are not mutually exclusive. But I would like to hear from the experts in the field. I sense that even though these results appear to shatter prior claims, the new knowledge still stems – at least in part – from "riding on the shoulders of giants."