A predicted bidirectional transcriptional terminator for both cueO and the convergent gene gcd is found in the predicted 42 nt 3' UTR for cueO.
The predicted bidirectional terminator for cueO has a delta G of -15.8 using mfold; it has 7/9 GC pairs in the stem, a tetraloop, 2 trailing Us and 4/5 leading As. Although this is currently the predicted cueO 3' UTR in EcoGene, it is not supported by the experimental data of Conway et al., who find cueO mRNA 3' ends both before and after this predicted terminator (Conway, 2014). The first Conway-observed 3' end is at position 25 of this 42 nt 3' UTR and is preceded by a small 3 GC pair stem with a tetraloop (mfold delta G = -3.2) with no trailing Us that is predicted to be a pause site but not a factor-independent terminator. Although of course this Conway-observed 3' end could indeed be a terminator, it also seems possible that long pauses might generate some 3' ends during RNA extraction or at a low level in vivo. Consistent with this possibility, many Conway-observed 3' ends occur in REP sequences that were not previously found to be terminators but that have GC-rich stem-loops that can act as pause sites, although this issue was not discussed in Conway et al. In fact the second 3' end past the EcoGene-annotated 3' UTR is within RIP10's IHF binding sequence, not previously known to act as a terminator. Although Conway et al. state that this cueO mRNA 3' end in RIP10 is preceded by a predicted factor-independent terminator as confirmed by TransTerHP, mfold does not predict a stable secondary structure in this region. RIPs are unusual REP elements that contain IHF binding sites and may be involved in chromosome organization. It is possible that RIPs are previously undiscovered terminators in vivo, possibly a previously undiscovered type of factor-dependent terminator mediated by IHF. However the RIP IHFs are highly conserved and thus one might expect them all to act as terminators, which does not seem to be the case. Alternatively, perhaps RIPs are long pause sites or perhaps the bound IHF (or possibly IHF and DNA gyrase) leads to cleavage during RNA extraction. See the REP TopicPage for more information and references to REP and RIP functional characterizations.
Nonetheless, it seems as if Conway et al. should have observed a 3' end corresponding to this EcoGene-predicted bidirectional terminator if it is functional, but they did not. Therefore this 3 'UTR might be shortened or lengthened in the future, possibly with or without verification of the Conway-observed high throughput 3' ends. With two trailing Us versus 4/5 leading As, the EcoGene-predicted bidirectional terminator should disassociate better in the counterclockwise direction. Indeed, as noted in the gcd 3' UTR TopicPage, Conway et al. do observe a convergent transcript's 3' end at this terminator (Conway, 2014). However, the Conway-observed 3' end is in the tetraloop, which seems very unlikely. Conway et al. state that the 30 nts just prior to their observed 3' end contain a predicted factor-independent terminator as confirmed by TransTermHP, but this claim could not be validated using mfold, which predicts no secondary structures predicting a terminator, in contrast to the excellent mfold prediction of a bidirectional terminator. Perhaps the loop was cut by an RNase during mRNA extraction prior to the RNA-Seq step, although it seems unlikely this would have been cut to completion. Until this issue is resolved by validation of the Conway-observed highthroughput 3' end, the gcd mRNA annotated in EcoGene will extend to the end of the EcoGene-predicted bidirectional terminator.
Convergent transcripts are abundant and present special problems for transcription termiantion as some of the convergent genes actually overlap and have no intergenic region and thus must terminate in the antisense RNA of the convergent gene's codng region. Indeed Conway et al. found that 276 convergent mRNAs terminate well within the converging gene, creating an average overlap (dsRNA) of 286 nts; however they also found 94 converging gene pairs, like cueO-gcd, that do not read through into each other (Conway, 2014). The cueO-gcd intergenic region has two 3' ends for cueO and three for gcd. Although they may not all be termination events (perhaps some long pauses or an RNase III cleavage), the presence of RIP10, a predicted bidirectinal terminator, and a possible RNase III cleavage site may indicate that antisense mRNA for these genes is deleterious to cell growth or survival and has been selected against. Another consideration is the fact that convergent transcription can introduced positive supercoils that may need to be actively rewound to negative supercoils by DNA gyrase. REP elements have been found to bind DNA gyrase so possibly RIP10 serves this purpose. The IHF site might stabilize the REP-gyrase complex, which might also be responsible for the generation of 3' ends from pause sites in this region when disrupted. cueO is predicted to be a foreign gene and gcd is predicted to be a native gene; perhaps when cueO was transmitted to E. coli the sudden juxaposition of these converging transcripts was problematic, leading to the subsequent insertion of RIP10 or the other intergenes to block the deleterious antisense mRNAs from inpinging upon mRNA function or stability.
Bibliography (1 total) : Review Only   Up
Conway T, Creecy JP, Maddox SM, Grissom JE, Conkle TL, Shadid TM, Teramoto J, San Miguel P, Shimada T, Ishihama A, Mori H, Wanner BL (2014) Unprecedented high-resolution view of bacterial operon architecture revealed by RNA sequencing. MBio 5:e01442-14