Oxidative damage or other chemical modifications during the stress response can have severe consequences for RNA half-lives or translation fidelity (Nunomura et al., 2017; Simms et al., 2014). Assuming that stress-induced mRNPs can protect RNAs from harmful conditions through a chaperoning effect, it is surprising that RNA imaging studies show that generally only between 1% and 10% of cellular RNAs localize to PBs and SGs (Sheinberger and Shav-Tal, 2017; Stöhr et al., 2006). Although some exceptions exist, these findings were generally confirmed by the recent partial purifications and transcriptomic analysis of PBs and SGs (Hubstenberger et al., 2017; Khong et al., 2017).
Aizer et al. used the MS2-MCP mRNA imaging system in living human cells and showed that mRNAs accumulate inside of PBs during amino acid starvation (Aizer et al., 2014). Using a FRAP approach they show that during the ongoing stress response mRNA exchange kinetics with the surrounding cytosol are slow and that a large immobile fraction exists. mRNAs cleared gradually from PBs after stress is over. Although the majority of mRNA molecules in their experiments did not localize to PBs, the authors argue that the release of mRNAs from PBs during the relief from stress indicates a storage function for PBs. Currently, no published live cell data is available which clearly shows that mRNA localization to PBs or SGs significantly increases RNA half-life compared to their unbound counterparts. In Chapter 2, translation and decay RNA imaging experiments are presented which call a protective role of PBs and SGs into question.
58
1.5 mRNP granules and disease
Through local enrichment of biomolecules stress-induced mRNPs are thought to influence the cellular biochemistry in two ways. First, the recruitment of catalytically active molecules into mRNP complexes results in a high local concentration. As a consequence, reaction equilibria are driven towards bound states, that can specifically enhance or block a reaction. Second, mRNP complexes can reduce molecular interactions in the cytosol through sequestration and physical separation of two binding partners. Experimental evidence exists for both models which are not mutually exclusive, but highly depend on the recruited molecules and the physiological situation (Protter and Parker, 2016). Connected to the above-described conceptual physiological roles, stress-induced mRNPs also seem to play roles in various human diseases. Here, the focus will mainly be on SGs rather than PBs since their relatively well-studied link with mRNA translation has led to large body of evidence connecting this granule type with altered cell physiology.
SGs in neurological diseases and cancer
Recently, SGs-like structures have been related to human neurodegenerative disorders defined by the presence of toxic insoluble protein aggregates. This link is strongest for amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD), where several disease-causing mutations also influence the dynamics of SGs. Mechanistic insights have not been obtained in great detail, but disturbed phase separation induced by LCDs of the proteins FUS and TDP-43 are increasingly in the focus (Haeusler et al., 2016). In addition, translation regulation can be severely disturbed the ALS/FTD context (see section 1.2.2 on RAN translation). SGs also frequently occur inside of solid tumors, presumably induced by nutrient starvation or hypoxia. In addition, several different types of eIF2α and eIF4F targeting chemotherapeutic agents have been shown to induce SGs (Anderson et al., 2015). What might seem an unimportant secondary effect could cause a severe resistance to cancer therapy. Chemotherapeutic drugs can induce apoptosis through the stress-activated p38 and JNK/MAPK (SAPK) pathways. Importantly, Arimoto et al. show that SGs negatively regulate the SAPK apoptotic response. Mechanistically, the signaling protein RACK1 becomes sequestered inside of SGs and cannot fulfill its SAPK-activating function anymore. As a result, apoptosis induction is inhibited (Arimoto et al., 2008). A similar discovery was recently made in tumor cells bearing a KRAS mutation. Here, the signaling prostaglandin molecule PGJ2 is produced in excess by the mutant cells, disrupts eIF4F complex formation and induces SGs, resulting in increased and unwanted tumor fitness (Grabocka and Bar-Sagi,
59 2016). On the other hand, several chemotherapeutic reagents have been shown to induce eIF2α phosphorylation leading specifically to the stimulation of the immune system, which might help to counteract tumor formation (Bezu et al., 2018). Considering the large amount of correlative disease data on cancer and SGs (Anderson et al., 2015), it is surprising that relatively few published studies exist trying to identify unbiased or at least multidimensional ways to target SG integrity. The discovery of the molecule ISRIB is a notable exception and highlights the enormous scientific and therapeutic potential when existing chemical compound libraries are combined with a very specific stress response relevant readout (see also section 1.2.3 and Chapter 3).
SGs in viral infections
While excess SG formation might be harmful for cancer patients, some evidence points towards that boosting SG presence might be effective against viral infections. SG targeting and inhibition by viruses during their life cycles has been documented extensively (McCormick and Khaperskyy, 2017). Since the discovery that dsRNA causes activation of PKR and the induction of the ISR including translational repression, viruses have been studied in the context of SGs. Considering the large variations in viral structure, genome organization and replication strategies, it is surprising that all virus classes have been shown to be able to alter SG dynamics. Interestingly, viral SG suppression often occurs downstream of PKR and stress-induced translation arrest and strongly suggests that SGs have antiviral properties (McCormick, Nat Rev Immu, 2017). The most prominent role for SGs during viral infections could be the block of viral gene expression through translation inhibition, although not necessarily in a localized manner. Several viruses block PKR activation to prevent their detection. For example, Zika virus inhibits eIF2α-dependent SG assembly upstream eIF2 (Amorim et al., 2017) and also piconavirus has been shown to regulate SG formation via its protease 2A to specifically enhance the translation of its own mRNAs (Yang et al., 2018). Further, SGs have been shown to sequester antiviral factors which might make them a preferred target for viruses (McCormick and Khaperskyy, 2017). The viral block of SG formation can be surprisingly robust. HIV-1 Gag blocks SG assembly irrespective of eIF2α phosphorylation (SA & pateamine A were tested) and even when SG assembly is forced by overexpression of G3BP1 or TIAR (Valiente-Echeverría et al., 2014). Interestingly, cells can form anti viral granules (AVGs) upon viral infection that resemble SGs, but are not identical to them. AVGs are for example positive for the SG marker proteins TIA1 and G3BP1, but do not contain 40S ribosomal subunits and are cycloheximide resistant (Rozelle et al., 2014). Whether AVGs are the effect of an arms race between host cells and viruses, battling for SG stability, is currently unclear. In line with such a theory are findings by Ruggieri et al. The researches find that SG presence in human cells can oscillate upon
60 infection with dsRNA (Ruggieri et al., 2012). Potentially, this represents a mechanism for cells to minimize opportunities for viruses to downregulate SGs. Several translation-targeting antiviral approaches involving eIF4A helicase inhibitors exist. However, understanding the anti-viral role of SGs independently of translation might help to use the active and forced induction of SGs as an antiviral therapy without the need to disturb translation with all of its side effects for the host. Taken together, the clearly demonstrated activities of viruses to block SG formation or to promote their disassembly are one of the strongest overall indications that SGs have important functions for cellular homeostasis during the stress response. Which functions exactly, remains an open question.
61
1.6 Functional mRNA dynamics during stress are unknown – an
experimental approach
Reasearch on stress-induced mRNP complexes has come a long way since the discovery was made that translation repression is linked to SG formation (Kedersha, JCB, 1999). Knowledge about the transcriptomic and proteomic composition of PBs and SGs has dramatically increased (Hubstenberger et al., 2017; Jain et al., 2016; Khong et al., 2017; Markmiller et al., 2018; Namkoong et al., 2018). Also the dynamics and architecture of proteins within PBs and SGs have been characterized in detail recently (Niewidok et al., 2018; Wheeler et al., 2016). Further, the macroscopic interaction of PBs and SGs is well documented (Decker and Parker, 2012; Stoecklin and Kedersha, 2013). Despite these recent advances, two important aspects of PB and SG biology have not been assessed in detail so far.
mRNA dynamics relative to PBs and SGs are unknown
Direct observations and quantification of mRNA interactions with stress-induced mRNPs at high resolution in living cells have not been performed. As a result, it is only incompletely understood during which phases of the stress response mRNAs enter PBs and SGs. Further, it is not known whether subpopulation of the same mRNA species interact differently with granules, i.e. what is the fraction of granule-bound mRNA compared to their unbound counterparts during the stress response. In addition, it is not clear whether these dynamics are differing between transcripts of different genes and which cis- and trans-acting elements could be responsible for their recruitment. It has also not been demonstrated at high resolution to what extend mRNA interactions differ between PBs and SGs. RNA regulation inside and outside of PBs and SGs has not been quantified
It is unlikely that all mRNA regulation occurs in a granule-dependent manner, while the intriguing clustering of mRNA-binding proteins inside of granules points towards some localized regulation. Due to the lack of high-resolution insights into the localization pattern of mRNAs relative to PBs and SGs, it has not been possible to assesses in detail to what extend both structures contribute to mRNA biology.
Experimental requirements for the study of localized mRNA regulation
The reason for the lack of direct evidence for localized mRNA regulation during the stress response seems to be mainly of a technical nature. Requirements for the study of localized mRNA biology are
62 complex. Such experiments involve the sequential performance of imaging, quantification (detection, tracking and colocalization), and functional assessment of RNAs relative to their localization in real-time. To understand mRNA regulation relative to stress-induced mRNPs in a high-resolution and mRNA- centered manner, the following experimental requirements are obligatory: