RNA binding proteins and sequence elements in untranslated regions

RBPs associate co-transcriptionally on nascent mRNA precursors as soon as these arise in order to mediate early RNA processing events like capping, splicing or polyadenylation [282], [283].

Database and literature mining defined a consensus of ~1,500 human RBPs, which constitute 7.5 % of all protein-coding genes [284]. Two independent studies provided experimental evidence for the functionality of ~800 of these predicted RBPs in HEK and HeLa cells by protein occupancy profiling or interactome capture using UV crosslinking of RNA-protein complexes and subsequent oligo(dT) affinity purification [285], [286].

In general, RBPs are able to bind thousands of transcripts via ~600 structurally distinct RNA-binding domains (RBDs) [284]. The vast majority of RBPs is ubiquitously expressed and only 6 % show tissue-specific expression patterns [284]. Whereas the function of ~ 1/3 of all

51

RBPs is still unknown or insufficiently inferred from homologues, approximately 700 RBPs bind to mRNA and act in mRNA metabolic processes [284]. Most often these mRNA interacting RBPs bind within 3’UTRs of selective mRNAs to direct mRNA-specific translational repression [287]. Translational regulation by protein-RNA interaction within 5’UTRs is uncommon and IRE-regulated mRNAs constitute a well-studied example, which is described in a later chapter (see section 1.3.7) [13].

Even though regulation by RBPs is almost invariably inhibitory, PABP interaction with the poly(A) tail is an exceptional and well-known example of translational stimulation [13].

Human PABP contains four non-identical RNA-recognition motifs (RRMs) at the N-terminus of which the first and the second one are able to bind eIF4G, while the last two RRMs exhibit high affinity for poly(A) sequence [288]–[290]. The interaction of PABP with eIF4G and the poly(A) tail of an mRNA results in a so-called “closed loop” configuration leading to mRNA circularization [288], [289]. In the circularized configuration, the affinity of eIF4F for the mRNA cap increases by an order of magnitude, thus facilitating translation initiation and possibly ribosome recycling [289]. The interaction of PABP with eIF4G tethers the eIF4F complex to the mRNA so that it does not need to be recruited de novo when interaction of eIF4F with the cap gets lost, establishing a significant translational advantage [13].

An alternative prominent example of RBP-mediated translational repression is the msl-2 gene, an important factor in dosage compensation in Drosophila. Translational repression in female flies is achieved by the female-specific RBP SXL, which on the one hand regulates msl-2 splicing and on the other hand inhibits msl-2 translation [291]. The msl-2 transcript contains two poly(U) rich sequence elements in the 5’UTR as well as four poly(U) elements in the 3’UTR and efficient translational repression requires association of SXL to binding sites in both UTRs [291], [292]. When bound to the 3’UTR, SXL prevents recruitment of the 43S complex, while SXL binding to the 5’UTR blocks scanning of those complexes that potentially escaped the control at the 3’UTR [291]. The suppression of 43S complex assembly on msl-2 transcripts further involves the co-repressor UNR, which is recruited to the 3’UTR by SXL where it interacts with poly(A) tail-bound PABP and possibly additional unknown factors [291].

The case of msl-2 regulation is in line with a generic model of translational repression in which a RBP binds specific sequence elements in a transcripts 3’UTR to cooperate with another repressive protein that interacts with the cap structure and possibly additional intermediate proteins that establish a connection between the RBP at the 3’UTR and the cap-binding protein at the 5’UTR, resulting in an inhibitory closed loop formation which prevents translation initiation [283]. The mRNA conformation in a closed loop constitutes a physical framework to

52

account for the fact that RBPs associating with the 3’UTR often control translation initiation events at the 5’UTR [291].

RBP-mediated translational control is usually stimulated by intrinsic or extrinsic signals which require rapid adjustment of gene expression levels to initiate specific cellular response programs. One example group of transcripts, which is collectively regulated in response to nutrient or oxygen deprivation, are mRNAs encoding ribosomal proteins, most translation factors and a few RBPs all containing a TOP motif at their 5’end [278], [293]. A TOP motif consists of a cytosine directly downstream of the m⁷Gppp-cap structure which is followed by an uninterrupted stretch of 4-14 pyrimidines [79]. TOP mRNAs are specifically sensitive to mTOR mediated translational regulation and how this set of mRNAs is selectively repressed by mTOR inhibition in comparison to all other cellular mRNAs that are also translationally controlled by mTOR remained a mystery until recently [79], [269]. The discovery of LARP1 acting as a downstream repressor of mTORC1 specifically controlling TOP mRNA translation and hence the expression of fundamental players of global mRNA translation eventually elucidated this essential control mechanism of cellular homeostasis [278], [294]. Upon phosphorylation by mTOR, LARP1 associates with mTORC1 through RAPTOR and TOP mRNAs get translated under physiological conditions [278]. When mTOR is inhibited and LARP1 is present in a dephosphorylated state, it displays enhanced binding affinity for TOP mRNAs and directs association with the cap and the adjacent TOP motif, which results in translational repression of TOP mRNAs due to LARP1 blocking eIF4E from recognizing the cap and subsequent eIF4F assembly [278], [293].

Although the tight translational regulation of mRNAs is predominantly controlled at the step of translation initiation, RBPs also effect mRNA stability to influence translational rates.

One well-studied example of such a type of gene expression regulation is the interplay of ELAVL1/HuR and TTP with AU-rich elements (ARE) containing mRNAs. AREs are located in the 3’UTR of a mRNA and are frequently assembled in overlapping AUUUA sequence stretches [295]. They are commonly found in mammalian transcripts encoding inflammatory cytokines, oncoproteins and G-coupled receptors and are estimated to occur with a frequency of 8 % in the human transcriptome [295], [296]. HuR is one of several ARE-binding proteins that can interact with thousands of mRNAs, promoting their stability [297], [298]. TTP is another prominent ARE-interacting protein with opposing biological function, promoting the destabilization of mRNAs by recruitment of the CCR4-NOT complex, which leads to mRNA decay [295]. In the case of p21 mRNA, bound HuR protects the transcript from TTP binding but upon ubiquitination, HuR gets displaced by p97 and UBXD8 ATPase complex which results in p21 destabilization [283], [296]. Over 80 % of TTP binding sites in 3’UTRs overlap with HuR,

53

suggesting a wide and complex interplay in the regulation of mRNA stability of thousands of target mRNAs, like it was described for p21 [299].

In summary, throughout their whole lifecycle mRNAs are found in association with RBPs that orchestrate the various processing, modification and localization events of RNA metabolism.

Due to their modular structure made of multiple RNA-binding domains organized in various ways, RBPs achieve high target specificity and affinity to facilitate diverse regulatory functions [282]. They form interconnected networks, which collectively act on processes of posttranscriptional gene expression like mRNA translation discussed here [285]. However, we are still far away from completely understanding the complex interplay of mRNA-RBP interaction networks and it remains elusive how a certain RBP that is able to interact with thousands of mRNAs and that is also able to conduct more than one function, fulfills the correct processing event at the appropriate target transcript at the right time.