Quick or quality? How mRNA escapes nuclear quality control during stress

ABSTRACT

Understanding the mechanisms for mRNA production under normal conditions and in response to cytotoxic stresses has been subject of numerous studies for several decades. The shutdown of canonical mRNA transcription, export and translation is required to have enough free resources for the immediate production of heat shock proteins that act as chaperones to sustain cellular processes. In recent work we uncovered a simple mechanism, in which the export block of regular mRNAs and a fast export of heat shock mRNAs is achieved by deactivation of the nuclear mRNA quality control mediated by the guard proteins. In this point of view we combine long known data with recently gathered information that support this novel model, in which cells omit quality control of stress responsive transcripts to ensure survival.

Keywords:

• Heat stress
• cellular stress
• mRNA quality control
• nuclear export

Efficient protein production relies on the coordination and correctness of gene expression, transcript processing and translation, which are crucial prerequisites for every cell's growth. In particular, the nuclear regulation of pre-mRNA transcription, maturation, and mRNA export are tightly connected and monitored. Recent work in the model organism Saccharomyces cerevisiae provided novel insights into mRNA quality control and gene expression during stress.

Pre-mRNA maturation

Under normal growth conditions a newly transcribed mRNA undergoes several processing steps before it is exported into the cytoplasm.^1-3 The major platform for regulation of mRNA maturation is provided by the C-terminal domain (CTD) of Rpb1, the largest catalytically active component of the RNA polymerase II (RNAP II).⁴ Changes in the phosphorylation pattern of the CTD's highly conserved YSPTSPS-heptads from the transcription start to the end allow recruitment of proteins at the correct time point.^5,6 The first of these steps is the capping reaction, which is performed after approximately 20 nucleotides of the mRNA are synthesized by RNAP II.⁷ At this point the m⁷G-cap is added at the 5′ end of the transcript by the Cet1/Ceg1 capping machinery⁸ and the methyltransferase Abd1⁹ to protect it from 5′−3′ degradation.^10,11 Efficiency of this process is supported by a free 5′-end and the immediate recruitment of the capping complex to the CTD, which is at that time highly phosphorylated at Ser5.^8,12-14 Once the m⁷G-cap is synthesized, the cap-binding complex (CBC) can bind and is later involved in nuclear export.¹⁵ Co-transcriptional recruitment of the spliceosome follows capping as it requires the presence of the CBC¹⁶ and results in excision of introns from the pre-mRNA.¹⁷ Splicing is accompanied by binding of the THO/TREX complex to the spliceosome¹⁸ and the RNAP II CTD.^19,20 The THO complex (Hpr1, Tho2, Mft1, Thp2) forms the TREX (transcription-coupled export) complex by recruitment of Yra1 and Sub2. It supports transcription elongation and links it to the later export of the mRNA.^19,21,22 Another complex, which recently was shown to connect transcription with export of the matured mRNA is TREX-2, composed of Sac3, Thp1, Sem1, Sus1 and Cdc31.²³ TREX-2 together with the SAGA complex are involved in docking transcribing genes to the NPC in a process called “gene gating” that allows preferential export of these mRNAs.²⁴

To finalize mRNA transcription, the 3′ processing machinery (CF1A, CF1B, CPF) is recruited to the polyadenylation site and after cleavage of the mRNA a 70-90 nucleotide long poly(A) tail is added by the poly(A) polymerase Pap1.^25,26

Nuclear mRNA quality control and the guard proteins

During the last years the idea emerged that these maturation processes not only rely on the enzymes and cooperating factors that carry out each processing step, but that the essential steps of mRNA maturation are controlled for their correctness and linked to the recruitment of shuttling adaptor proteins. As such they interact with the mRNA and the export receptor heterodimer Mex67-Mtr2 (TAP-p15 in humans) and accompany the transcript to the cytoplasm. In the bakers yeast S. cerevisiae the most prominent RNA-binding proteins that shuttle with an mRNA from the nucleus to the cytoplasm are the serine/arginine (SR)-rich proteins Npl3, Gbp2 and Hrb1 and the poly(A)-binding protein Nab2. In their function to release the matured mRNA from the nucleus to the cytoplasm by recruitment of Mex67, we will name them from now on the guard proteins. Npl3 is one of the first proteins that contacts a newly emerging mRNA as it interacts with the RNAP II²⁷ as well as with the CBC²⁸ arguing for a very early recruitment. Following that, Npl3 supports efficient splicing by interacting with the early spliceosome.²⁹ Finally, Npl3 recruits Mex67-Mtr2 signaling export competence.³⁰ Correct splicing is subsequently controlled by the guard proteins Gbp2 and Hrb1 that interact with the late spliceosome.³¹ These two proteins are loaded co-transcriptionally by the THO/TREX complex,^32,33 and recruit the export receptor Mex67 in case the mRNA is processed properly. Errors in this step on the other hand result in acquiring the degradation machinery.³¹ This nuclear removal of faulty RNAs relies on the TRAMP (Trf4/5, Air1/2, Mtr4) complex that marks these RNAs with a short oligo(A) tail for subsequent degradation by the nuclear exosome, which in contrast to its cytoplasmic counterpart contains Rrp6.^34-36 It was suggested earlier that preventing export of immature transcripts might be the result of kinetic competitions between factors that facilitate export and others involved in retention and degradation.^37,38 Thus, the more an ribonucleoparticle (RNP) differs from being perfectly covered with export receptors, the less likely it is exported³⁹ and rather marked by TRAMP to be degraded by the Rrp6-containing exosome. The proper coverage of large RNA-containing molecules with export receptors is also required for the export of ribosomal subunits, as missing export factors result in the accumulation of subunits in the nucleus.^40-46

The last processing step that results in decoration of the mRNA with the guard proteins is the formation of the 3′ end and synthesis of the poly(A) tail. Here the poly(A)-binding protein Nab2, together with its mainly cytoplasmic homolog Pab1,^47,48 controls length and quality of the 3′ tail – a process that is antagonized by the exosome component Rrp6.^49,50 A fine-tuned interplay between factors that facilitate export and degradation is crucial for cellular functionality. All binding events of the guard proteins suggested to represent quality control checkpoint marks that allow export in case of proper processing or else retain the wrong transcripts and recruit the degradation machinery.

This serial decoration of the maturing mRNA with the guard proteins and the dependence of their recruitment on the previous step suggested a potential function as control factors that are able to finally allow export by recruitment of the Mex67-export receptor. However, this was not conclusively shown until lately, when an in vivo “leakage assay” was developed in which the retention of false mRNAs was visibly alleviated upon elimination of the guard proteins, Npl3, Gbp2, Hrb1 and Nab2.^31,51 Covering the pre-mRNA with the guard proteins might prevent an immediate and independent contact of Mex67 with the premature mRNA, which would seem to be an elegant retention mechanism.

These data together allow describing a model, in which maturing mRNAs bind the guard proteins to prevent Mex67 association until the transcripts are fully processed (Fig. 1). Completed maturation is signaled by Mex67 recruitment and results in their cytoplasmic transfer. Although it seems likely that Npl3 is involved in controlling correctness of 5′ capping, Gbp2 and Hrb1 ensure accurate splicing and Nab2 monitors proper 3′ tailing, it is well possible that the guard proteins have additional functions and are not only loaded once.^52,53 Also the existence of other, yet missing guard proteins is conceivable, because a smooth transit of the highly charged mRNA through the hydrophobic interior of the nuclear pore complex (NPC) will probably be more effective in the presence of more Mex67 molecules that shield the charged mRNA and facilitate a smoother passage through the hydrophobic interior of the NPC.

Figure 1. Guard proteins control mRNA maturation and nuclear export. Every maturation step results in association of guard proteins that recruit the export receptor Mex67-Mtr2 to the correct mRNA and support degradation by the TRAMP/exosome pathway for faulty transcripts. Finally, Mlp1 controls proper Mex67 decoration of the guard proteins to allow nuclear export.

Transcripts that do not show the necessary Mex67-decoration of the guard proteins are not only detained because they lack export factors, but also by a last quality control check at the NPC. At the nuclear basket, Mlp1 and the highly homologous nucleoporin Mlp2 are the last nuclear factors involved in retaining erroneous transcripts.^54,55 As a final gatekeeper, Mlp1 was shown to interact with the guard proteins,^31,55,56 and a model is conceivable that Mlp1 and Mlp2 might control the Mex67-guard-protein interactions, before letting them pass³¹ and retains the mRNA, if no or insufficient Mex67 is bound to the guard proteins.

This elaborated system ensures efficient translation of correct mRNAs and prevents the translation machinery to deal with an overwhelming number of suboptimal or even defective transcripts. In fact, cytoplasmic non- or wrongly processed mRNAs that are translated into missfunctional proteins threaten homeostasis and have in general a detrimental effect on cellular growth.⁵⁷ There is growing evidence that not only in yeast but also in mammals mRNA adaptor proteins act as guards and are tightly linked to correctness and thus quality of mRNA expression.⁵⁸ In the same way as a lack of quality control is harmful to the cell, an overexpression of the guard proteins has the same adverse effect on cellular fitness in yeast, as it possibly even leads to a retention of correct mRNAs.⁵¹ In fact, defects or overexpression of human homologues of the guard proteins (ZC3H14, as a human Nab2 ortholog), or the SR-proteins SRSF1, SRSF3 and SRSF7, which also accompany the mRNAs to the cytoplasm and interact with TAP-p15,⁵⁹ are known to cause several diseases, including cancer and neurodegenerative diseases, cardiovascular diseases or conditions like neuronal dysfunction^60-65 and (http://www.cbioportal.org/).

The stress induced mRNA export block

In situations that are stressful to the cell, such as elevated temperatures or high salt concentrations, a cellular survival response – the heat shock response- is initiated, in which heat shock (HS) proteins are produced to protect other cellular proteins from denaturing. To establish a fast response to survive the hazardous condition, expression, maturation, nuclear export and translation of normal “housekeeping” mRNAs need to be blocked and the export of stress mRNAs has to be of utmost priority.⁶⁶ That this in fact happens and that mRNAs are blocked in the nucleus upon 42°C heat stress, while HS mRNAs are exported⁶⁷ and splicing is disrupted^68,69 was already described over 20 years ago. Additionally other cellular stress reactions are known, as Npl3 dissociates from mRNAs and is dispensable for HS mRNA export,^70,71 and Gbp2 aggregates reversibly upon stress.⁷² Furthermore, nuclear accumulation of mRNA requires activity of the MAPK kinase Slt2, which is also responsible for the phosphorylation of Nab2 under stress.⁷³ Stress furthermore leads to formation of foci that contain the RNA binding proteins Nab2 and Yra1 as well as the NPC gatekeeper Mlp1.⁷³

However, a systematic analysis of stress in the light of nuclear quality control and nuclear mRNA export was not done until recently. In a systematic approach we examined the behavior of the guard proteins and Mex67 and found that the export block of mature or maturing housekeeping mRNAs appears to be facilitated by a global dissociation of the guard proteins in complex with the exporter Mex67⁵¹ (Fig. 2). Even though it is most likely that structural changes or post-translational modifications are responsible for the dissociation of all guard proteins, details still need to be elucidated. Amazingly, inactivation of the guard proteins upon stress not only stops regular mRNA export, but also provides a sophisticated system to prevent the association of guard proteins with newly transcribed HS mRNAs. Thus, production of HS mRNAs is accomplished without any of the guard proteins.⁵¹ Certainly the greatest benefit of expressing stress mRNAs without the help of guard proteins is that in this way a time-consuming quality control is omitted and a response to stress is promptly initiated. If formation of nuclear guard protein-containing foci is needed to allow evasion of quality control, or if they form because these proteins are not required and present in excess at that time, remains elusive to date. However, the deactivation of the guard protein-mediated quality control is of particular importance, as the degradation machinery is functional during heat stress and degrades HS RNAs under certain conditions, e.g. when these transcripts are artificially trapped in the nucleus.^74,75 Thus, by preventing the guard proteins from binding to HS mRNAs, their potential quality control-mediated degradation is circumvented.

Figure 2. Stress conditions lead to dissociation of Mex67-complexes and a subsequent mRNA export block. Upon stress housekeeping mRNAs are no longer transcribed and Mex67 is dissociated with the guard proteins from mature mRNAs. The recruitment of the guard proteins is prevented by mechanisms, including phosphorylation, aggregation and foci-formation of guard proteins and the NPC gatekeeper Mlp1.

The preferential nuclear export of HS mRNAs

The expression of stress responsive transcripts utilizes a slimmed down pathway. As guard proteins are dispensable, export is very likely promoted by direct binding of the exporter Mex67 to the HS mRNA.⁵¹ Not only is Mex67 among the few proteins absolutely essential for the export of stress transcripts,⁷⁶ but further it was shown to be able to bind, via its loop domain, directly to any kind of RNA including stress mRNAs.^46,51 Furthermore, binding of Mex67 to either the guard protein Npl3 or RNA is mutually exclusive,⁵¹ arguing for a domain in Mex67 that can either bind to guard proteins or directly to mRNA. This points toward a new view in mRNA export that rather proposes a mechanism in which Mex67 under normal conditions is constantly prevented from binding to the maturing mRNA, by the co-transcriptional decoration of the immature transcript with the guard proteins. Direct binding of the export receptor Mex67-Mtr2 to certain transcripts was also shown for its homolog in higher eukaryotes TAP-p15 where it binds to constitutive transport elements (CTEs) that can be found in viral RNAs as well the TAP pre-mRNA itself^77-79 potentially allowing these transcripts to circumvent quality control and being immediately exported.

This instant binding of Mex67 to newly transcribed HS mRNAs seems to happen already at the site of transcription as the exporter interacts with the RNAP II component Rpb1 and the transcription factor Hsf1 when heat stress conditions are applied⁵¹ (Fig. 3). The transcription factor Hsf1 is the master factor of heat stress gene induction as it regulates their expression by binding to the heat-shock-element (HSE) in the gene's promoter.⁸⁰

Figure 3. Export of HS RNAs. Direct co-transcriptional recruitment of Mex67 to HS mRNAs allows omission of quality control and fast HS mRNA export.

Overall, the expression of stress responsive transcripts appears to rely on other requirements than normal gene expression. Thus, transcriptional activation by Hsf1 does not require the presence of general transcriptional activators like TFIIA,⁸¹ Taf9 (a component of TFIID and SAGA)^82,83 or the TFIIH kinase Kin28.^84,85 Even the CTD of RNAP II, which usually mediates most regulatory processes is dispensable upon stress.⁸⁵ On the contrary, Hsf1 on its own is able to recruit the Mediator complex⁸⁶ that tightly regulates stress gene expression.⁸⁷ Whether “gene gating” supports HS mRNA export and involves for instance TREX-2 is currently unknown, but might be an attractive way of supporting fast nuclear export.

Being expressed under the control of Hsf1 appears to change the fate of an mRNA dramatically. Thus, housekeeping transcripts can be converted into heat stress responsive mRNAs that are not quality controlled by only inserting an HSE in their promoter.⁵¹ Therefore, the HSE resembles an express ticket to the cytoplasm. Furthermore, it allows a preferential translation, as glucose starvation was shown to induce an Hsf1-controlled transcription and translation priority for heat shock proteins, while mRNAs controlled by other stress transcription factors (Msn2/4) are indeed produced, but directly stored in stress granules for later use.⁸⁸ Bypassing quality control appears to be a drastic course of action when facing cellular stress, but the gain of time in responding to a threat potentially exceeds the drawbacks of translating a couple of faulty proteins - at least for a limited time. Moreover, the overall cellular mRNA export block relieves the competing situation for the many mRNAs at the translation machinery, as in comparison only a small amount of HS RNAs needs to be translated so that it is not necessary to be very efficient in producing correct proteins. Consistently, cells can survive a period of severe heat stress without guard proteins and quality control factors, but not without the general exporter Mex67 or heat shock proteins, as shown in growth analyses.⁵¹

Not only changes in the nucleus contribute to efficient HS mRNA expression, but also cytoplasmic processes adapt to ensure a fast heat shock protein production. In response to stress the general translation is inhibited and polysomes can no longer be detected.^89,90 Depending on the type of stress, initiation of translation is inhibited either by disturbing formation of the closed loop structure in an Eap1/Caf20-dependent way, or by titrating GTP-bound eIF2 out of the ternary complex.^91,92 All cellular responses in the end lead to the preferential translation of HS mRNA.

Concluding remarks and open questions

In summary, the mechanisms, by which normal and HS mRNA expression is performed are significantly different, as the guard protein mediated quality control is deactivated during stress. The export receptor heterodimer Mex67-Mtr2 alone is sufficient to ensure a fast transport to the cytoplasm. The important role of the guard proteins lies in normal mRNA transport by regulation and control of maturation and the stepwise approval through Mex67 recruitment. Future studies have to address how the guard proteins detect defects and initiate degradation of false transcripts. It might be a combination of the timely detection of the successively associating complexes and a competition with the degradation machinery. Also post-translational modifications may play a role. De-phosphorylation of Npl3 was indeed shown to lead to the recruitment of Mex67,³⁰ but what triggers de-phosphorylation is unknown. Furthermore, it is unclear if there are additional guard proteins and how exactly Mlp1 at the NPC detects the proper Mex67-coverage of the guards. Many questions are also unanswered for the particular situation during cellular stress, such as what triggers the dissociation or aggregation of the guard proteins and which complexes and subcomplexes that contribute to a smooth mRNA production under normal conditions (like THO, TREX, TREX-2 or SAGA) are also involved in HS mRNA transcription. Finally, it remains to be understood how HS mRNA translation differs from normal translation. Comparing the molecular basics of gene expression under normal and challenging conditions may help to unravel fundamental principles in science and get a better understanding of diseases based on defects in mRNA surveillance factors.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgment

We thank the Krebber laboratory for comments on the manuscript.

Funding

This work was funded by grants of the Deutsche Forschungsgemeinschaft (DFG) and the SFB860 awarded to H.K.