Initiation of mRNA decay in bacteria

Abstract

The instability of messenger RNA is fundamental to the control of gene expression. In bacteria, mRNA degradation generally follows an "all-or-none" pattern. This implies that if control is to be efficient, it must occur at the initiating (and presumably rate-limiting) step of the degradation process. Studies of E. coli and B. subtilis, species separated by 3 billion years of evolution, have revealed the principal and very disparate enzymes involved in this process in the two organisms. The early view that mRNA decay in these two model organisms is radically different has given way to new models that can be resumed by "different enzymes-similar strategies". The recent characterization of key ribonucleases sheds light on an impressive case of convergent evolution that illustrates that the surprisingly similar functions of these totally unrelated enzymes are of general importance to RNA metabolism in bacteria. We now know that the major mRNA decay pathways initiate with an endonucleolytic cleavage in E. coli and B. subtilis and probably in many of the currently known bacteria for which these organisms are considered representative. We will discuss here the different pathways of eubacterial mRNA decay, describe the major players and summarize the events that can precede and/or favor nucleolytic inactivation of a mRNA, notably the role of the 5' end and translation initiation. Finally, we will discuss the role of subcellular compartmentalization of transcription, translation, and the RNA degradation machinery.

Fig. 1

RNase E: domain structure and substrate binding. a Domain composition an RNase E monomer (1,061 aa). The catalytic amino-terminal half (NTH, aa 1–529) contains a large globular domain (aa 1–400), which is a composite of recurrent structural subdomains as shown [63] and a small folded domain (aa 415–529). The C-terminal half (CTH) of the protein is predicted to be unfolded but contains microdomains that mediate interactions with the cytoplasmic membrane (segment A) and other components of the degradosome (the helicase RhlB, enolase, and PNPase). AR1 and AR2 are arginine-rich segments probably involved in RNA binding. b RNase E exists primarily as a tetramer composed of a dimer of dimers [63]. The monomers of the principal dimer shown here are in light or dark grey and are held together by a dimer interface and a cooperatively coordinated Zn⁺² ion (shown in yellow, the Zn-link, aa 400–415, [86]). Interactions between the small domains of the principal dimers stabilize the tetramer (not shown). Each protomer possesses a 5′ P binding pocket (green circle) and an active site (blue rectangle). In the 5′ tethering pathway, the monophosphorylated 5′ end of the mRNA (in violet) binds to the 5′ P binding pocket of one protomer, whereas cleavage occurs in the active site of the other protomer. The direct entry pathway that operates mainly on primary 5′ PPP transcripts is probably the major route for initiating mRNA decay in E. coli, but its efficiency is largely dependent on the conformation of the mRNA that is recognized by the nuclease. Binding of the substrate to only one active site is thought to be less efficient (slow) than, for example, the simultaneous binding of two single-stranded regions of which one might only serve to tether RNase E to the RNA (fast) [152]. However, binding of multiple sites should be very sensitive to ribosome occupancy and also be more demanding in terms of respecting enzyme geometry. Similarly, in the 5′ tethering pathway, RNase E preferentially cleaves sites, when available, in the 5′ UTR and avoids reaching around translating ribosomes [97]

Fig. 2

The architecture of RNases J and Y. a Domains composing B. subtilis RNase J1 (555 aa). The β-CASP domain is inserted into the β-lactamase domain to which the C-terminal domain is attached by a linker. b Comparison of the open and closed ribbon conformations of the T. thermophilus RNase J monomer. The open conformation is shown with colored backbone (in the presence of a 4 nt RNA, colored in red) [118, 119] and the closed free enzyme in gray [58]. The β-lactamase domain of the open conformation (in green) is superposed on that of the free enzyme to show the relative movements (blue arrows) of the β-CASP (in violet), C-terminal (in pink) and linker (in blue) domains. The catalytic Zn⁺² ions in the active site are in yellow. c Close-up of the RNase J catalytic center complexed with an UMP residue. The 5′ terminal phosphate group is coordinated by serine and histidine residues in a phosphate binding pocket that provides a rationale for the enzyme’s requirement for a 5′ P in exonuclease mode [58]. Dotted orange lines indicate ligand-mediated and hydrogen bond interactions. d Slab view showing electrostatic surface predictions of the major RNase J domains (aa 1–447). Positively charged surfaces are shown in blue and negatively charged surfaces in red. The RNA is shown in yellow. The RNA-binding channel and a proposed nucleotide exit tunnel are indicated [118]. e Similar overall shape and electrostatic charge distribution between T. thermophilus RNase J and the catalytic N-terminal half of E. coli RNase E. The active site in both structures is facing upwards. The C-terminal domain of RNase J (aa 465–555) and RNase E (corresponding to the small domain in Fig. 1a, aa 415–529) share the same architecture, a three-stranded β-sheet facing two α-helices as shown. f Domains composing B. subtilis RNase Y (520 aa) include an N-terminal transmembrane domain (aa 1-25), followed by a large region predicted to be disordered (aa ~30–210), an RNA binding KH domain (aa 211–270) and a metal-chelating HD domain (aa 336–429) containing the conserved His/Asp motif required for RNase activity [59, 65, 166, 207]

Fig. 3

RNA degradation pathways in E. coli and B. subtilis. Initiation of mRNA decay as defined by the first nucleolytic cleavage can depend on a variety of parameters that render a given mRNA susceptible to the action of an RNase (e.g., translation efficiency, 5′ end conversion, stochastic events, etc., see text). a In E. coli, the major direct entry pathway involves a primary cleavage of the native transcript by RNase E. The upstream fragments are rapidly degraded by 3′ exoribonucleases (RNase II, PNPase, RNase R, and oligoribonuclease for short oligonucleotides). The 5′ monophosphorylated downstream fragment is preferentially recognized by the 5′ sensor of RNase E, which enhances the rate of subsequent cleavages (>20-fold, at least in vitro). This causes a wave of secondary downstream cleavages proceeding in a 5′–3′ direction each generating a 3′-OH upstream fragment that is degraded by 3′ exonucleases. Decay intermediates whether or not protected by 3′ secondary structure can be polyadenylated by poly(A) polymerase, enabling the 3′ exonucleases to re-engage several times if necessary to produce complete degradation (see main text). Polyadenylation can also be observed on full-length transcripts containing the transcription terminator (not shown in the figure) but does not represent a major pathway to initiate mRNA decay (see main text, “The 3′ end: tailing, scavenging and surveillance”). A second pathway of mRNA degradation in E. coli is 5′ end-dependent and starts with pyrophosphate removal by the pyrophosphohydrolase RppH. This tethers RNase E to the 5′ end of the transcript and stimulates downstream cleavage in the same way as described for secondary cleavages above. Refer to the legend of Fig. 1 and text. b In B. subtilis, the pathways initiating mRNA decay are similar to E. coli but the players are different. In the major direct entry pathway, the primary cleavage is affected by RNase Y and to a lesser extent by RNase J1/J2 or another endonuclease. The upstream fragments are degraded mainly by PNPase, in contrast to RNase II in E. coli [–315]. The monophosphorylated downstream cleavage products are degraded 5′–3′ by RNase J1/J2 in exonuclease mode and can proceed to the 3′ end. It is interesting to note that the B. subtilis extracts used to demonstrate the largely phosphorolytic degradation of RNA to mononucleotides [313] most likely did not measure the contribution of the, at the time, unknown hydrolytic RNases J1/J2 to exonucleolytic decay, due to the 5′ triphosphorylated RNA substrate used and the fact that most of the ribosome associated RNase J was probably eliminated during extract preparation [313]. B. subtilis has no poly(A) polymerase but A-rich polynucleotide tails synthesized by an unknown enzyme (indicated by an ANA sequence) are found essentially on degradation intermediates [138]. The question mark indicates that it is not clear whether they contribute to the degradation of 3′ structured fragments. However, 3′ terminal fragments containing the transcription terminator are very resistant to 3′ exonuclease attack. The 5′ exonuclease activity of RNase J is thus very useful to degrade 3′ structured RNA fragments. Similar to E. coli, conversion of the native 5′ PPP to a 5′ P by BsRppH (which prefers a G in second position, see text) and BsRppH-2 (not yet identified, but insensitive to N-terminal sequence, see text) renders the mRNA susceptible to the 5′ exonuclease activity of RNase J. In vitro, RNase J can also cleave endonucleolytically a native transcript close to the 5′ end probably by threading the 5′ PPP through the RNA entry channel and past the 5′ P binding pocket. It is not known whether this “sliding endonuclease” mode plays a significant role in 5′ end conversion in vivo. Similarly, RNase Y activity is stimulated by a 5′ P group in much the same way as RNase E, but to what extent RNase Y competes with RNase J for binding to a monophosphorylated 5′ end in vivo remains to be analyzed (indicated by a question mark)