Structural and Functional Studies of Mrna Stability Regulators

Research Collection

Doctoral Thesis

Structural and Functional Studies of mRNA Stability Regulators

Author(s): Ripin, Nina

Publication Date: 2018-11

Permanent Link: https://doi.org/10.3929/ethz-b-000303696

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library

DISS. ETH NO. 25327

Structural and functional studies of mRNA stability regulators

A thesis submitted to attain the degree of

DOCTOR OF SCIENCES of ETH ZÜRICH

(Dr. sc. ETH Zürich)

presented by

NINA RIPIN

Diplom-Biochemikerin, Goethe University, Frankfurt, Germany

Born on 06.08.1986

citizen of Germany

accepted on the recommendation of

Prof. Dr. Frédéric Allain

Prof. Dr. Stefanie Jonas

Prof. Dr. Michael Sattler

Prof. Dr. Witold Filipowicz

2018

“Success consists of going from failure to failure without loss of enthusiasm.”

Winston Churchill

Summary

Posttranscriptional gene regulation (PTGR) is the process by which every step of the life cycle of an mRNA following transcription – maturation, transport, translation, subcellular localization and decay - is tightly regulated. This is accomplished by a complex network of multiple RNA binding proteins (RNPs) binding to several specific mRNA elements. Such cis-acting elements are or can be found within the 5’ cap, the 5’ untranslated region (UTR), the open reading frame (ORF), the 3’UTR and the poly(A) tail at the 3’ end of the mRNA. Adenylate-uridylate-rich elements (AU-rich elements; AREs) are heavily investigated regulatory cis- acting elements within 3’untranslated regions (3’UTRs). These are found in short-lived mRNAs and function as a signal for rapid degradation. AREs are present in 5-8% of human genes involved in the regulation of many essential cellular processes, such as stress response, cell cycle regulation and apoptosis and must therefore be tightly regulated. In the cytoplasm, trans-acting ARE binding proteins regulate the transport localization, stability and translation of these mRNAs. One of these factors is the embryonic lethal abnormal visual (ELAV)/ Human antigen R (HuR) protein. It increases the stability and/ or the translation of many important cellular mRNAs. Another cis-activing element is the poly(A) tail of mRNA, which protects the mRNAs from degradation. These are bound by multifunctional poly(A)-binding proteins (PABPs), which play a central role in translation initiation, translation termination and mRNA decay.

In this study, we have investigated the mRNA stability regulators HuR and PABPC1, both containing multiple RNA recognition motifs (RRMs). Excitingly, some single RRMs have several functions due to the presence of additional binding interfaces that allow them to bind both RNA and other factors. We characterized the C-terminal RRM of HuR, which is hypothesized to be involved in RNA binding, homo-dimerization and protein-protein interacting. We show the first 1.9- Å-resolution crystal structure of HuR RRM3 bound to several short ARE-motifs. Our structure reveals the presence of the homodimer. The combination of several biophysical techniques validate the homo-dimerization and promiscuous RNA binding in solution. Additionally, the binding of the canonical AUUUA pentameric motifs, found in the majority of AREs, is possible by the recognition of two registers. Excitingly, RRM3 homo-dimerization increases the affinity for RNA, highlighting the cooperativity between the two binding surfaces. Moreover, despite the known stabilizing role of HuR, we provide evidence that RRM3 counteracts this effect in a Huh7 cell - based ARE reporter assay containing multiple AUUUA motifs. Finally, we investigated the mechanism of the cytoplasmic PABP RRM1 in binding to poly(A) and to the anti-proliferative B- cell translocation gene (BTG2) protein. BTG2 recruits the CCR4-associated factor 1 (CAF1), a

subunit of CCR4-NOT deadenylase complex, to induce deadenylation of mRNAs. We show that PABPC1 RRM1 uses its α1 to bind BTG2 while simultaneously binding the poly(A) RNA. This interaction seems to orient the poly(A) 3’end such that it is close to the CAF1 enzymatic pocket.

Our findings provide new details of the HuR RRM3-RNA recognition and homo-dimerization as well as the PABPC1 RRM1-poly(A)-BTG2 binding to recruit CAF1 and thus highlight the diversity of RRMs.

Zusammenfassung

Posttranskriptionelle Genregulation (PTGR) ist ein Prozess nach der Transkription, bei dem jeder Schritt des Lebenszyklus einer mRNA - Prozessierung, Transport, Translation, subzelluläre Lokalisierung und Abbau - streng reguliert wird. Dieser wird durch ein komplexes Netzwerk von RNA-bindenden Proteinen (RBP) erreicht, die mehrere spezifische mRNA-Elemente binden. Solche cis-wirkenden Elemente sind oder liegen innerhalb der 5'-Kappe, der 5'-untranslatierten Region (UTR), des offenen Leserahmens (ORF), des 3'UTR und des Poly(A) -Schwanzes am 3'- Ende. Adenylat-Uridylat-reiche Elemente (AU-reiche Elemente; AREs) sind stark untersuchte regulatorische cis-wirkende Elemente innerhalb von den 3'-untranslatierten Regionen (3'UTRs). Diese finden sich in kurzlebigen mRNAs und fungieren als ein Signal für den schnellen mRNA Abbau. AREs sind in 5-8% der menschlichen Gene vorhanden, die an der Regulation vieler essenzieller zellulärer Prozesse beteiligt sind, wie Stressreaktion, Zellzyklusregulation und Apoptose und müssen daher streng überwacht werden. Im Zytoplasma, trans-wirkende ARE- bindende Proteine steuern die Lokalisierung, Stabilität und Translation dieser mRNAs. Einer dieser Proteine ist das „embryonic lethal abnormal visual (ELAV)/ human antigen R (HuR)“ Protein. Es erhöht die Stabilität und/ oder Translation vieler dieser zellulärer mRNAs. Ein weiteres cis-wirkendes Element ist der Poly(A) -Schwanz von mRNAs, der die mRNAs vor Abbau schützt. Dieser wird durch multifunktionelle Poly(A) -bindenden Proteine (PABPs) gebunden, die eine zentrale Rolle bei der Translationsinitiation, Translationstermination und dem mRNA-Zerfall spielen.

Das Ziel dieser Doktorarbeit ist die Untersuchung der mRNA bindenden Proteine HuR und PABPC1. Diese enthalten mehrere RNA-bindinge Domänen/ „RNA recognition motifs, (RRMs)“. Interessanterweise haben einige dieser RRMs zahlreiche Funktionen aufgrund der Anwesenheit von mehreren Bindungsstellen. Diese ermöglichen sowohl RNA- als auch Protein-Bindung. Wir charakterisierten die C-terminale RRM von HuR, von welcher angenommen wird, dass sie an RNA-Bindung, Homodimerisierung und Protein-Protein-Wechselwirkung beteiligt ist. Wir zeigen die erste 1,9-Å-Kristallstruktur von HuR RRM3, die an mehrere kurze ARE-Motive gebunden ist. Unsere Struktur zeigt die Anwesenheit eines Homodimers. Durch die Kombination mehrerer biophysikalischer Metoden validieren wir die Homodimerisierung und die RNA-Bindung in Lösung. Darüber hinaus ist die Bindung der kanonischen AUUUA-Pentamer-Motive, die in den meisten AREs gefunden werden, durch die Erkennung von zwei Bindungsregistern möglich. Die RRM3-Homodimerisierung erhöht die Affinität für RNA und verdeutlicht die Kooperativität zwischen den beiden Bindungsoberflächen. Schließlich, trotz der bekannten stabilisierenden

Rolle von HuR, liefern wir Beweise, dass RRM3 diesem Effekt in einem auf Huh7-Zellen-basierten ARE-Reporter-Test entgegenwirkt. Darüber hinaus untersuchten wir den Mechanismus der zytoplasmatischen PABP-RRM1 bei der Bindung an poly(A) und an das anti-proliferative B-Zell- Translokationsgen (BTG2) -Protein. BTG2 rekrutiert den CCR4-assoziierten Faktor 1 (CAF1), eine Untereinheit des CCR4-NOT-Deadenylase-Komplexes, um die Deadenylierung von mRNAs zu induzieren. Wir zeigen, dass PABPC1 RRM1 die α1 verwendet, um BTG2 zu binden, während es gleichzeitig die poly(A) RNA bindet. Diese Wechselwirkung scheint das Poly(A) 3'-Ende so auszurichten, dass es nahe der CAF1-enzymatischen aktivem Zentrums liegt.

Unsere Ergebnisse liefern interessante Informationen über die HuR-RRM3-RNA-Erkennung und Homodimerisierung sowie die PABPC1-RRM1-BTG2-Bindung, um CAF1 zu rekrutieren und verdeutlichen damit die Diversität von RRMs.

Acknowledgment

My time during the doctorate was the toughest challenge I ever faced. I grew as a scientist but also as a person. For that, a special thanks to Fred. You took me in under special circumstances and for that, I am very grateful. The time in your group was and still is one of the best and valuable experiences I had in my life. Thank you for all your support. I would like to acknowledge Prof. Stefanie Jonas, Prof. Michael Sattler and Prof. Witold Filipowicz for joining as co-referees for my thesis. Michael, I really enjoyed our discussions and your challenging questions in Parpan. Another very big thanks goes to Malgosia. You have been always there for me. Your wisdom is highly precious and I learned a lot from you. Fred D, thanks a lot for teaching me how to set up and the basics of NMR experiments and especially for correcting my English whenever it was needed and thanks to Julien for introducing me to ITC, your support on the HuR project and all the discussions we had. Nana and Irene, it was fun working with you. One cannot imagine better colleagues and friends in the lab. Ahmed, you are always there to help and support others. Thank you for being such a good listener. Yaro, thanks for the fun working evenings and late discussions. Thea, Gerry, Fred D, Alvar and Simon, thank you for keeping our NMR spectrometers running. Stefanie, many thanks for all your advice on the cell culture experiments and more. I wish I had met you earlier. You would have saved me from all the struggle I faced in the cell lab. To all my office mates over the years: first L24, Johannes, Georg, Kyle, Nana, Irene and Esteban, thanks a lot for all the discussions and the fun time. Then, L12 office: Thanks to Laurent, Tebbe, Leonidas and Elisabeth for the great and fun atmosphere and your patience about my nagging during my thesis-writing period. Elisabeth, you do an amazing job with the AKTA! Cristina thanks for trying to motivate me going to Yoga. Thanks to all former and present colleagues Grégoire, Christine, Dominik, Sébastien, Antoine and our newest lab members Xing and Pengzhi, as well as all the members of the Gossert and Jonas group. It was and is always a pleasure to come to work and one of the reasons is the amazing atmosphere because of all you guys. Big thanks to Roddy and Naomi, two great students I supervised. I enjoyed and had a lot of fun working with you. I learned a lot from you too! Thank you Isabelle for your amazing help with all the administration, and Iwona and Gabi for keeping the institute running.

Additionally, I thank all my collaborators, namely Fabienne Mauxion (Institut de Génétique et de Biologie Moléculaire et Cellulaire, France) for the collaboration on the BTG2 project. Jiří Šponer and Miroslav Krepl (Institute of Biophysics of the Czech Academy of Sciences) for molecular dynamics data, Robert Schneider and Abhilash Gadi (NYU School of Medicine, USA) for validating some of our luciferase assay data, Nicole Meisner-Kober,

Alexandra Hinniger and Michael Faller (Novartis, Basel) and especially Alexandra and Michael for teaching me X-ray crystallography and always being there for me. Additionally, thanks to further colleagues during my time at Novartis, César and Sascha (CPC) and Katrin, Julian, Wolf, Lena, Dominik, Anja and Cornelia (DMP) and my flatmates from the Basler Murbi WG, Cedi, Lena and Christof.

Thanks to the ladies from the FSSB and the ladies from the NCCR peer mentoring group for all the valuable discussions and advice on the scientific environment. Thanks to all my colleagues and friends from the scientific staff associations AMB and AVETH, especially Shady, Betty, Michaela, Tanja, Markus, Rebekka, Elisa, Florian, Arik, Martin, Alok, Jenna, Michael and Alina. Also a big thanks to Antonio, Wilfred and Maryvonne. We were a great team and I learned a lot from you. It was fun working with all of you and we had some – politically - exciting years! Let’s see what comes next! Lastly, thanks to my climbing partners Florian, Daniel and Bettina and especially my friends Lori for always being there for me.

Abschließend möchte ich allen meinen Cousins, Cousinen, Tanten und Onkeln in Deutschland und Russland für all die Unterstützung und Motivation danken. Vor allem meinen Eltern Tatjana und Julius und meiner Schwester Kristine. Ohne eure Geduld, Vertrauen und Unterstützung wäre ich niemals so weit gekommen oder hätte so viel erreicht. Und nochmal: Ich habe mich nie vor der Hausarbeit gedrückt, ich habe wirklich gearbeitet! NMR Protein- „Assignments“ ist kein Computerspiel! Darüber hinaus möchte ich meiner Großmutter Nina danken. Sie ist letztes Jahr verstorben. Sie war eine wundervolle Person, stark, selbstbewusst und immer für ihre Familie da. Danke für alles, was du für mich getan hast.

Table of Contents 1. Introduction ...... 1

1.1 mRNA biology ...... 2

1.1.1 Post-transcriptional gene regulation ...... 2

1.1.2 Translation and mRNA turnover ...... 3

1.1.3 mRNA decay regulation by cis-acting AU-rich elements...... 4

1.1.4 Subcellular localization and phase separation ...... 5

1.1.5 The RNA recognition motif (RRM) – a multifunctional binding scaffold ...... 6

1.1.6 ELAV protein family and HuR ...... 8

1.1.7 Interplay of PABP and BTG2/ Tob family in mRNA decay ...... 10

1.1.8 mRNA stability regulators and human disease ...... 12

1.2 Investigation of protein-protein and protein-RNA interactions ...... 13

1.2.1 Structural studies of macromolecules ...... 13

1.2.2 Structure determination by X-ray Crystallography ...... 14

1.2.3 Characterization of protein-RNA and protein-protein complexes by NMR Spectroscopy ...... 18

1.2.4 Prediction of protein complexes by HADDOCK ...... 23

1.2.5 ITC to study protein-RNA and protein-protein interactions ...... 25

1.2.6 Dual Luciferase Reporter Assay ...... 27

1.3 Research Objectives ...... 30

2. Molecular basis for AU-rich element recognition and dimerization by the HuR C- terminal RRM ...... 31

2.1 Abstract ...... 32

2.2 Introduction ...... 33

2.3 Results ...... 35

2.4 Discussion ...... 54

2.5 Materials and Methods ...... 58

2.6 Acknowledgements ...... 62

3. Structural basis of the PABPC1 RRM1-BTG2 interaction to recruit CAF1 deadenylase ...... 63

3.1 Abstract ...... 64

3.2 Introduction ...... 65

3.3 Results ...... 68

3.4 Outlook ...... 85

3.5 Discussion ...... 86

3.6 Materials and Methods ...... 89

3.7 Acknowledgements ...... 92

4. Concluding remarks ...... 93

4.1 RNA recognition motifs: boring? Not at all! ...... 94

4.2 The multitasking RRM3 of HuR ...... 94

4.3 The multitasking RRM1 of PABPC1 binds poly(A) and BTG2 to induce deadenylation ...... 97

4.4 Towards understanding mRNA stability regulators ...... 99

5. Appendix ...... 101

5.1 A2. Supplementary Tables Chapter 2 ...... 102

5.2 A2. Supplementary Figures Chapter 2 ...... 110

5.3 A2. Supplementary Materials and Methods Chapter 2 ...... 117

5.4 A3. Supplementary Tables Chapter 3 ...... 120

5.5 A3. Supplementary Figures Chapter 3 ...... 127

6. References ...... 129

7. Curriculum Vitae ...... 142

Abbreviations

ARE (AU)-rich element (Adenylate-uridylate-rich element)

Cryo-EM Cryo-Electron Microscopy dsRBD double-stranded RNA binding domain

EPR Electron Paramagnetic Resonance

FL Firefly luciferase

HADDOCK High Ambiguity Driven protein-protein Docking

HEK293 human embryonic kidney cells 293 hetNOE heteronuclar nuclear Overhauser effect

HSQC Heteronuclear Single Quantum Coherence spectroscopy

Huh7 Human hepatocarcinoma cell line

IDR Intrinsic disordered region

ITC Isothermal titration calorimetry

LCD Low complexity domains

MR Molecular Replacement miRNA microRNA miRISC miRNA-induced silencing complex mRNA messenger RNA mRNP messenger ribonucleoprotein particles

MW molecular weight nt nucleotides

NLS nuclear localization signal

NMR Nuclear magnetic resonance

NOESY Nuclear Overhauser Enhancement Spectroscopy

ORF Open reading frame p pocket

PTGR Posttranscriptional gene regulation

PB p-bodies/ processing bodies

PBS Phosphate buffered saline ppm parts per million

PRE Paramagnetic relaxation enhancement

RBD RNA binding domain

RBP RNA binding protein

RDC Residual dipolar coupling,

RISC RNA-induced silencing complex

RL Renilla luciferase

RNA Ribonucleic acid

RNP Ribonuleoprotein

RRM RNA recognition motif

RT Room temperature

SANS Small-angle neutron scattering

SAXS Small-angle X-ray scattering

SG Stress granules

SR Serine-arginine-rich protein

TNF Tumor necrosis factor

TOCSY Total Correlation Spectroscopy

UTR Untranslated region

WT Wild type

Less frequent abbreviations are defined upon their first use in the text

1. Introduction

1 Introduction

1.1 mRNA biology

All the genetic information is stored in our DNA, which is tightly packed and protected in the cell nucleus. To execute genetic instructions, the DNA sequence is transcribed into a complementary pre-messenger RNA (pre-mRNA). The pre-mRNA is processed through multiple steps in the nucleus and the mature mRNA is transported to the cytoplasm to fulfill its functions (Figure 1.1). From the very beginning, multiple factors, including RNA binding proteins (RBPs), bind mRNAs and regulate them throughout their entire life cycle, a process called posttranscriptional gene regulation (PTGR).

Figure 1. 1. mRNPs and post-transcriptional regulation. RBPs are trans-acting factors that bind cis-acting elements within the mRNAs RBPs regulate alternative mRNA splicing, maturation, transport, subcellular location, lifetime, and translation. Adapted from (García-Mauriño et al., 2017).

1.1.1 Post-transcriptional gene regulation

After the transcription within the nucleus, every step in the life cycle of an mRNA is tightly regulated to produce the mature mRNA for translation in the cytoplasm. During these nuclear processing events, a 7-methylguanosine cap is added at the 5’end, introns are spliced and the 3’end is poly-adenylated (Figure 1.1). The mature mRNA is composed of a 5’ cap, the 5’ untranslated region (UTR), the open reading frame (ORF), the 3’UTR and the poly(A) tail at the 3’ end (Figure 1.2). These mRNAs are exported to the cytoplasm, where they are translated to proteins, stored in cytoplasmic bodies or targeted for degradation (Figure 1.1). All these steps are regulated by a dynamic interaction with multiple RBPs and formation of mRNA-protein complexes (RNPs).

Introduction

Figure 1. 2. Composition of mRNAs. mRNA elements and interacting proteins form mRNPs. Translation initiation factors, such as eIF4E and eIF4G interact with the 5’UTR-cap structure and factors including PABP interact with the 3’-poly(A) tail. Trans-binding regulatory factors recognize specific cis-elements, e.g. AREs within the 3’UTRs of certain mRNAs. Adapted from (Rissland, 2017).

1.1.2 Translation and mRNA turnover

The mRNA cap and poly(A) tail protect mRNAs from degradation. They are involved in two major processes affecting all mRNAs: translation and decay (Rissland, 2017). Many RBPs recognize these two structural elements and link both pathways. In the cytoplasm, the translation initiation factor eIF4E binds and protects the 5’ cap structure (Figure 1.2). The cytoplasmic poly(A)-binding protein 1 (PABP) on the contrary binds the poly(A) tail and the translation initiation factor eIF4G, which is bound to eIF4E, leading to the closed-loop structure (Wells et al., 1998). This facilitates translational initiation but also regulates mRNA stabilization/ degradation. PABP interacts with members of the degradation machinery (Webster et al., 2018; Yi et al., 2018) or proteins that recruit the degradation machinery (Ezzeddine et al., 2007; Stupfler et al., 2016). The anti-proliferative B-cell translocation gene (BTG)/ transducer of ERBB2 (Tob) family members bind PABP and thereby recruit the CCR4- associated factor 1 (CAF1; also known as CNOT7), a subunit of CCR4-NOT deadenylase complex (Ezzeddine et al., 2007; Stupfler et al., 2016). Shortening of poly(A) tails is the first major step that triggers mRNA decay. This is catalyzed by various different deadenylases, such as PAN2-PAN3 and the CCR4-NOT complex. When the poly(A) tail reaches a certain length, degradation of the mRNA is mediated by the removal of the 5’-end cap by DCP1/2, followed by the 5’ exonuclease XRN1, which degrades the mRNA in the 5’ to 3’ direction (Clark et al., 2009; Heck and Wilusz, 2018; Wahle and Winkler, 2013). Alternatively, a second pathway of mRNA decay following deadenylation is the 3’ to 5’ degradation of some mRNAs. These are degraded by the exosome from the 3’-end. Afterwards, the scavenger decapping complex removes the 7-methyl-guanosine cap (Figure 1.3) (Clark et al., 2009; Heck and Wilusz, 2018; Wahle and Winkler, 2013).

3 Introduction

Figure 1. 3. Simplified illustration of the mRNA ARE degradation pathway. Cis acting AREs within the 3’UTR provides binding sites for various ARE binding proteins, such as HuR, AUF-1, TTP or KSRP, which regulate mRNA turnover. The 3’ to 5’ decay pathway (above the mRNA) is comprised of the deadenylase complex PAN2-PAN3 or the CCR4/NOT complex, followed by the 3’ degradation of the mRNA body by the exosome and 5’ end processing by the scavenger decapping complex (DCPS). The major 5’ to 3’ decay pathway (below the mRNA) involves the same initial deadenylation step. Afterwards the decapping complex (DCP1-DCP2) removes the 5’ cap and the nuclease XRN1 degrades the mRNA body. Adapted from (Clark et al., 2009).

Destabilizing elements such as adenylate/ uridylate (AU)-rich elements in the 3’UTRs (Barreau et al., 2005; Chen and Shyu, 1995), elements in protein coding regions (Chang et al., 2004; Grosset et al., 2000), premature stop codons (Chen and Shyu, 2003) and miRNA binding sites (Behm-Ansmant et al., 2006; Wu et al., 2006) are all inducing deadenylation and mRNA decay.

1.1.3 mRNA decay regulation by cis-acting AU-rich elements

(AU)-rich elements (AREs) are regulatory cis-acting elements found in 5-8% of human genes (Bakheet et al., 2006). AREs function as a signal for rapid degradation and are found mainly in the 3’ UTRs of short-lived mRNAs (Bakheet et al., 2006). mRNAs containing AREs are involved in the regulation of many important cellular processes such as stress response, cell cycle regulation, inflammation, immune cell activation, apoptosis and carcinogenesis (Bakheet et al., 2006). They range in size from 50 to 150 nucleotides and are classified by the

4 Introduction presence of the AUUUA motif repeats. Class I and II contains several AUUUA motifs while class III is U-rich (Benjamin and Moroni, 2007). Six well known RBPs regulate the transport, stability and translation of ARE containing mRNAs: Tristetrapolin (TTP), AU-binding Factor 1 (AUF-1), KH-type splicing regulatory protein (KSRP), Human antigen R (HuR), T-cell intracellular antigen 1 (TIA-1) and TIA-1-related protein (TIAR). While TTP, AUF-1 and KSRP mainly functions in mRNA degradation and TIA-1 and TIAR in silencing of translation, HuR, stabilizes mRNAs and/or upregulates translation. However, to some small extent, opposite functions for all proteins were reported for various mRNAs (García-Mauriño et al., 2017). Most of these proteins are able to shuttle between the nucleus and cytoplasm and compete for the same RNA binding sites; therefore, they regulate similar mRNA targets (Figure 1.3).

1.1.4 Subcellular localization and phase separation

Next to translation or decay, mRNAs can be stored in subcellular compartments, such as stress granules (SGs) or processing bodies (p-bodies or PBs) (Figure 1.1). These are membrane-less granules composed of proteins, nucleic acids and other molecular factors, which form under various stimuli. Both SGs and PBs form in response to stress, where the translation of mRNAs is strongly repressed (Uversky, 2017). The hallmarks of proteins enriched in such granules, are the presence of RNA-binding domains or motifs and intrinsically disordered regions (IDRs), sequences that lack a defined 3D structure (“disordered”) (Calabretta and Richard, 2015). A subset of IDRs are defined by >100 residue long regions, called low complexity domains (LCDs), that are composed of repeating amino acids with low overall diversity. In vitro, such IDRs drive phase transitions. These sequences are rich in uncharged polar side chains (glutamine, asparagine, glycine, serine, proline), charged amino acids (arginine, lysine, glutamic acid, aspartic acid), or aromatic residues (phenylalanine and tyrosine) and mediate interactions by electrostatic, dipole–dipole, pi–pi, cation–pi, hydrophobic, and hydrogen bonding interactions (Boeynaems et al., 2018). Most of the proteins involved in translation and decay, such as eIF4B, eIF4G, eIF4E, TTP, TIA-1, XRN1, PABP, TIAR and HuR can be found in such granules (Uversky, 2017). In addition, RNA alone is also able to form phase transition in vitro. Recent studies showed that RNAs which contain repetitive sequences are able to form phase separation through intermolecular base-pairing (Jain and Vale, 2017).

In the past years, multiple discoveries created a completely new field in cell biology, that focuses on understanding how these membrane less organelles form, their composition and their effect on biological function and disease (Boeynaems et al., 2018).

5 Introduction

1.1.5 The RNA recognition motif (RRM) – a multifunctional binding scaffold

RBPs are diverse and vary in their structure and function. They represent 7,5% of all protein-coding genes in humans (Gerstberger et al., 2014). Many RBPs contain characteristic domains, which bind to single stranded or double stranded RNA: RNA recognition motif (RRM) (also known as RBD (RNA binding domain) or RNP (ribonucleoprotein domain)), zinc fingers domains (ZnF), the K-Homology (KH), cold shock domains (CSD) and double stranded RNA- binding domain (dsRBD)(Lunde et al., 2007). These domains can specifically bind RNA by hydrogen bonding, electrostatic interactions and hydrophobic and aromatic stacking interactions with the nucleobases. Non-sequence-specific contacts are mediated by the sugar- phosphate backbone.

The RNA recognition motif (RRM) is the most abundant RNA binding domain and is present in 0.5-1% of human genes (Venter et al., 2001). Proteins containing RRMs are involved in all steps of PTGR. The RRM is comprised of around 90 amino acids, which fold into a

β1α1β2β3α2β4 topology. Four β-strands are packed against two α-helices (Figure 1.4). Within the β-sheet surface, three aromatic side-chains are often located in the conserved RNP1 (β3- strand) and RNP2 (β1-strand). RRMs are able to recognize two to eight nucleotides. Two RNA bases stack on the aromatic ring within the β1 (RNP2, position 2) and in the β3 (RNP1, position 5). A third aromatic ring, which is located in β3 (RNP1, position 3), is often inserted between the two RNA sugar rings (Cléry et al., 2008). Excitingly, despite the conserved RRM fold and similar binding surfaces, proteins show differences in RNA recognition. These deviations from the canonical RNA binding mode are possible due to the N- and C-terminal extensions, loops and interdomain linkers or binding to other factors which affect the number of bound nucleotides and RNA specificity. Some RRMs do not contain the canonical aromatic residues within their β-sheet surface and adopted different RNA binding strategies. The RRMs of the heterogeneous nuclear ribonucleoprotein (hnRNP) F, an alternative splicing and polyadenylation regulator, recognize RNA G-tracts by a β-hairpin, the β1-α1 loop and the β2- β3 loop. Such RRMs are called quasi RRMs (qRRMs) (Dominguez and Allain, 2006). Another alternative-splicing regulators, the SR protein SRSF1, also contains a RRM lacking the conserved RNP (Clery et al., 2013). The structure of this so-called pseudo-RRM (ΨRRMs) reveals the involvement of the patch where α1 packs against β2.

RRM containing proteins use a wide set of additional mechanisms to modulate RNA recognition. RNA can be recognized by unique contacts between amino acids and specific nucleotides. This leads to a precise binding on a certain position within the target mRNA. However, RRMs can also recognize degenerate or repetitive RNA sequence motifs, such as

6 Introduction

Figure 1. 4. Canonical RRM β-sheet-RNA interaction. (A) Example of an RRM structure in complex with RNA (hnRNP A1 RRM2 in complex with single stranded telomeric DNA). (B) Schematic representation of the four- stranded β-sheet surface with the main conserved RNP1 and RNP2 aromatic residues indicated in green. RNP1 and RNP2 consensus sequences of RRMs are shown (X is for any amino acid). Figure from (Cléry et al., 2008).

poly-pyrimidine or poly-adenosine tracts (Banerjee et al., 2003; Deo et al., 1999; Mackereth et al., 2011). On such targets, RRMs are able to recognize multiple binding registers, which enhances the overall binding affinity. This was shown for the the polyU binding protein hnRNP C and the pre-mRNA splicing regulator U2AF65 (Cieniková et al., 2014; Mackereth et al., 2011). Dynamic binding, by multiple register binding or sliding along the RNA, is an additional mechanism of RRMs to fine-tune their affinity.

Other variability in RNA recognition of proteins comes from the existence of multiple copies of RRMs or a combination of different domains (Figure 1.5). The domains can be in tandem or separated by short linkers or longer unstructured regions. Tandem RRMs show higher specificity and affinity compared to separate RRMs. The RRMs can interact with each other, sometimes involving their interdomain linkers, to create an extended RNA binding surface or a deep cleft for the interaction with the RNA, as shown for the ARE binding proteins Sex-lethal, HuD and HuR RRM12 (Handa et al., 1999; H. Wang et al., 2013; Wang and Tanaka Hall, 2001). Many RRM containing proteins are multifunctional. Aside from RNA binding they also participate in other macromolecular assemblies. Structural studies reveal that protein recognition by RRMs is very diverse. Interactions can form between two RRMs, between an RNA binding RRM and a non-RRM domain and between RRMs that do not bind RNA and another protein (Cléry et al., 2008; Muto and Yokoyama, 2012). RRM-RRM interaction can induce RNA looping, as shown for the polypyrimidine tract binding protein (PTB) RRM3-

7 Introduction

Figure 1. 5. RRM organization of key proteins involved in PTGR. RRMs are shown in yellow, Gly-rich sequences in green and the MLLE domain in PABP in grey.

RRM4, hnRNPA1 RRM12 and hnRNPL RRM34 (Barraud and Allain, 2013; Beusch et al., 2017; Oberstrass et al., 2005; Vitali et al., 2006; Zhang et al., 2013). RRM-protein interactions can cooperatively affect RNA affinity. This is observed for PABP RRM2. When PABP RRM12 is in complex with an eIF4G fragment, it binds 10 fold higher to poly(A) RNA than without (Safaee et al., 2012).

RNA recognition and affinity can thus be influenced by additional protein-protein interactions. These diverse binding modes help us to understand the basis for RRM-RNA or RRM-protein recognition. However, new modes of interaction are still being discovered. This highlights the versatility of the RRMs and that more investigations are needed to understand the potential code for RRM-RNA/ protein recognition (Cléry et al., 2008).

1.1.6 ELAV protein family and HuR

The embryonic lethal abnormal visual (ELAV)/ Human antigen (Hu) protein family consists of three members found in neurons, HuB (Hel-N1), HuC and HuD and one ubiquitously expressed member HuR (HuA) (Antic and Keene, 1997; Fan and Steitz, 1998b; King et al., 1994; Ma et al., 1996). Out of this family of proteins, HuR is heavily under investigation. It is a pivotal regular of ARE containing mRNAs, which play a role in essential biological processes such as including cell growth, differentiation, apoptosis, signal transduction, hematopoiesis and metabolism. HuR acts by stabilizing a large number of transcripts such as cyclin A, cyclin B1, p21, p53, tumor necrosis factor alpha (TNF-α), interleukin-3 (IL-3) and vascular endothelial growth factor (VEGF) (Dean et al., 2001; Levy et al., 1998; Ming et al., 2001; W. Wang et al., 2000; Wengong Wang et al., 2000; Zou et al., 2006). Moreover, HuR promotes translation as

8 Introduction reported for glucose transporter 1 (GLUT1),and cationic amino acid transporter 1 (CAT1), prothymosin alpha (ProTα) (Gantt et al., 2006; Lal et al., 2005; Yaman et al., 2002) It is also known to destabilize a small number of mRNAs and/or to suppress their translation (Cammas et al., 2014; Kim et al., 2009; Leandersson et al., 2006; Meng et al., 2005). HuR binds rather U-rich than AU-rich targets (Lebedeva et al., 2011; López de Silanes et al., 2004). HuR is mostly localized in the nucleus but undergoes cytoplasmic translocation under various cellular and stress conditions (J. Wang et al., 2013). HuR localizes in stress granules when cells are stressed by heat shock or arsenate treatment, (Gallouzi et al., 2000; Yoon et al., 2013). HuR and other Hu family members are composed of three highly conserved canonical RNA recognition motifs (RRMs). The first two RRMs (RRM12) are in tandem and are separated from the C-terminal RRM (RRM3) by a ~50-residue unstructured basic region (hinge region). A nucleocytoplasmic shuttling element within the hinge region is responsible for the translocation between the nucleus and the cytoplasm (Fan and Steitz, 1998a). RRM12 is mainly responsible for ARE-binding (Chen et al., 2002), while the exact function of RRM3 is still not fully characterized.

So far, only the crystal structures of the free HuR RRM12, crystal structures of HuR and HuD RRM12 bound to RNA and NMR structures of HuC RRM1 and HuC RRM2 are solved (Inoue et al., 2000; H. Wang et al., 2013; Wang and Tanaka Hall, 2001). In the free form, HuR RRM12 has an open conformation with no inter-domain contacts (Figure 1.6A). (H. Wang et al., 2013). The crystal structure of HuR RRM12 in complex with AUUUUUAUUUU shows a closed shape mediated by hydrogen bonds between the two domains and the inter domain linker. These interdomain interactions are also present in HuD RRM12, when bound to RNA. (Wang and Tanaka Hall, 2001). Both RRMs create a deep cleft for the RNA. RRM1 binds to five nucleotides U5-U8 and U10 while the inter-domain linker interacts with U9 and RRM2 binds to U3-U4 (Figure 1.6B) (Wang and Tanaka Hall, 2001).

RRM3 and the hinge region are involved in homo-dimerization (Scheiba et al., 2014; Toba and White, 2008), protein-protein interactions (Brennan et al., 2000; Cho et al., 2010), cooperative binding of multiple HuRs to long AREs and counteracting miRNA mediated repression to promote miRISC release from target mRNAs (Kundu et al., 2012; Mukherjee et al., 2016). Due to the insolubility and instability of the RRM3 domain in vitro, structural studies has remained challenging. A recent NMR study revealed that RRM3 dimerizes through helix α1, which is located opposite to the RNA binding interface (Scheiba et al., 2014), highlighting a new RRM-RRM interaction mode. Despite the provided structural model of the free RRM3 and two potential RRM3 dimer models in that study, the lack of atomic resolution structures of the free and RNA-bound forms prevents a complete understanding of HuR RRM3 dimerization and RNA recognition.

9 Introduction

Figure 1. 6. RRM orientation of free HuR RRM12 and in complex with HuR Cartoon representation of (A) the free HuR RRM12 shows an open confirmation (pdb code 4EGL). (B) 5’-AUUUUUAUUUU-3’ binding induces a closed conformation of HuR RRM12, creating a deep cleft for the RNA (pdb code 4ED5).

1.1.7 Interplay of PABP and BTG2/ Tob family in mRNA decay

The poly(A)-binding protein (PABP) family plays a role in both translation and mRNA stability by binding to the mRNA 3’ poly(A) tails. In addition to the canonical cytoplasmic PABP (PABPC1), there are four other PABP genes in humans. They have a similar domain architecture but differ in their expression patterns. Three of the cytoplasmic PABPs consist of four RNA recognition motifs (RRMs) followed by an extended C-terminus, while one PABP protein lacks the C-terminal region. There is also a nuclear PABP (PABPN1) comprised of only one RRM, flanked by an acidic N-terminus and an arginine-rich C-terminal domain (Mangus et al., 2003). The cytoplasmic PABP C-terminus contains a conserved MLLE domain, also known as poly(A)-binding protein C-terminal domain (PABC) (Kozlov et al., 2001), which recognizes PABP-interacting motif 2 (PAM2) found in a wide set of proteins to recruit them to the poly(A) tails (Kozlov et al., 2001; Lim et al., 2006; Okochi et al., 2005). Two crystal structures of the tandem PABP RRM12 with poly(A) RNA reveal that the two tandem RRMs contact each other to create an extended β-sheet surface and bind a single-stranded RNA motif (Figure 1.7A) (Deo et al., 1999; Safaee et al., 2012). The adenines are recognized by multiple contacts with the sugar-phosphate backbone and the ribose moieties (Deo et al.,

10 Introduction

1999). Further structural studies show that RRM2 α1 and β4 form hydrophobic interactions, hydrogen bonds and salt

Figure 1. 7. RRM domain orientations of PABPC1 RRM12. Cartoon representation of (A) the poly(A) bound PABPC1 RRM12, showing an extended β-sheet surface (pdb code 1CVJ) and (B) PABPC1 RRM12 bound to poly(A) and the eIF4G fragment (pdb code 4F02). RRMs are colored in grey, eiF4G in blue. RNA is shown as stick representation in yellow. (C). Polar contacts (red dashed lines) (left) and hydrophobic interactions (right) contribute to eIF4G-RRM2 binding

bridges with eIF4G (Figure 1.7B, C) (Safaee et al., 2012). PABP protects poly(A) tails from deadenylation. However, interaction with members of the anti-proliferative B-cell translocation gene (BTG)/ transducer of ERBB2 (Tob) family, recruits the CCR4-associated factor 1 (CAF1), a subunit of CCR4-NOT deadenylase complex, which induces poly(A) tail shortening (Ezzeddine et al., 2007; Stupfler et al., 2016). In detail, Tob contains a PAM2 motif, which binds the PABPC1 C-terminal MLLE domain (Ezzeddine et al., 2007). Interestingly, BTG2 is lacking such a PAM2 motif. In fact, the interaction is mediated by the BTG2 APRO domain and the PABPC1 RRM1 (Stupfler et al., 2016), suggesting a novel mode of RRM-domain interaction.

11 Introduction

1.1.8 mRNA stability regulators and human disease

All post-transcriptional steps in gene expression need to be tightly regulated. Aberrations, such as gene mutations, differential abundance of mRNAs or proteins, or changes in protein behavior could lead to undesirable pathologic effects. RNPs, among them ARE binding proteins are fundamental players, which control the stability and translation of ARE containing mRNAs. ARE is a signal for rapid degradation located in short-lived mRNA 3’UTRs. Such mRNAs code for proteins, which are involved in all essential biological processes, including cell growth, differentiation, apoptosis, signal transduction, hematopoiesis and metabolism (Khabar, 2005). After the mRNAs fulfill their functions, they need to be degraded. Prolonged stabilization of such ARE mRNAs causes continuous responses. In case of HuR, elevated levels increases the upregulation of mRNAs which cause tumor growth and disease progression in various cancer types, e.g. breast-, colon-, ovarian-, prostate-, pancreatic- and oral cancer (Kotta-Loizou et al., 2016; Srikantan and Gorospe, 2012; J. Wang et al., 2013). Additionally, HuR expression levels correlate with viral infections, cardiovascular diseases, neurological pathologies and muscular disorders (Di Marco et al., 2005; Farooq et al., 2009; Figueroa et al., 2003; Li et al., 2009; Misquitta et al., 2001; Sokoloski et al., 2010; Van Der Giessen et al., 2003). Consequently, understanding of the structure and function of mRNPs will help us to develop new biomarker for disease prognosis and new therapeutic drug targets.

12 Introduction

1.2 Investigation of protein-protein and protein-RNA interactions

1.2.1 Structural studies of macromolecules

Structural studies of biomolecules are necessary to understand their functions. Atomic models of enzymes or macromolecular machines help us elucidate their mechanism of action and their architecture. Biomolecular structures enable us to make targeted modification and engineering or structure based drug design to generate new drugs.

Multiple methods can be used to derive atomic resolution structures of biomolecules. Nuclear Magnetic Resonance (NMR) Spectroscopy, for example can be used to study molecules in solution. This method is based on nuclear spins response to magnetic fields and provide structural information as well as give insights into the dynamics of a system. However, NMR is limited to the size of a molecule and starts to become challenging beyond 30 kDa (Ikeya et al., 2018). To determine structures of larger molecules, NMR can be combined with other methods such as Electron Paramagnetic Resonance (EPR) Spectroscopy (Duss et al., 2015). There, longer distance restrains are derived which can be used for structure calculation. NMR is widely used for structure determination of RRMs or RRM-RNA complexes, to investigate their dynamics and for binding studies to RNA (Dominguez et al., 2011).

Another technique to determine structures is X-ray crystallography, where one measures the X-ray diffraction pattern of a crystalline sample. To be able to resolve a structure, this method requires that biomolecules form diffracting crystals. The derived atomic structures represent the packed state of the molecule in the crystal. This provides us with an instantaneous view of the biomolecule. In some cases this might lead to faulty interpretations of biomolecular function, for example protein dimers, which are only formed due to crystal packing but do not exist in solution or an overall different domain arrangement as in solution (Mackereth and Sattler, 2012). Proteins which are too flexible or are present in multiple conformations do not form crystals and shorter flexible regions or sidechains are not observable.

Recently, Single Particle Cryo-Electron Microscopy (Cryo-EM) has become a favored method for studying large assemblies at atomic resolution. One of the advantages of Cryo-EM is that large and complex structures can be determined, which cannot be crystallized for X-ray crystallography or are too large for NMR spectroscopy. Samples are directly imaged in vitrified solutions. Structures starting from around 64 kDa can thus be studied with resolutions around 3-4 Å (Murata and Wolf, 2018)

13 Introduction

Small-angle X-ray scattering (SAXS) and small-angle neutron scattering (SANS) are additional methods to gain insights into the structure of biomolecules, however, with lower resolution. Small-angle scattering of X-rays or neutrons prove information about the overall shape, relative position of the binding partners and binding stoichiometries. The advantage of SAXS and SANS is that the crystallization of the sample is not needed and that both methods can be combined with high-resolution structural information obtained from X-ray crystallography, NMR or Electron Microscopy (EM) to tackle complex multi-subunit complexes (Trewhella, 2016).

In the following studies, we used X-ray crystallography for protein-RNA complex structure determination and the docking program HADDOCK (High Ambiguity Driven protein-protein Docking) to generate models of a protein-protein-complex. NMR and Isothermal titration calorimetry (ITC) helped us to validate and characterize protein-protein and protein-RNA binding in solution, while a cellular luciferase assay highlighted the biological relevance in Huh7 cells. Thus, these methods are described in more detail.

1.2.2 Structure determination by X-ray Crystallography

To obtain a structure by X-ray crystallography, molecules need to form crystals. Usually, different crystallization conditions are tested. X-ray diffraction pattern of the crystals is based on the crystal packing of the molecules. If the diffraction pattern indicates a good resolution, data is recorded and processed. After determining the electron density, the structure can be obtained and validated (Figure 1.8).

1.2.2.1 Crystallization screening

For crystallization often a high quality, pure and homogeneous sample is needed. Crystals form from a supersaturated solution. When molecules change from a solution state into a solid state, they can either become amorphous and precipitate or ordered and form a crystal. To induce saturation and crystal formation, a precipitant is added to the solution. Over time, the total drop volume decreases and the evaporation/condensation from the reservoir reaches an equilibrium. Thereby, both protein and precipitant concentration increase and reach a critical concentration, at which the protein goes out of solution and crystallizes. Nowadays, crystal screens are performed at high throughput in 96-well plates using commercially available suites. Common screening suites are JCSG+, PACT, PEGS, AmSO4 and Classics (Hampton

14 Introduction or Qiagen) which cover different pH, precipitants (salts, polyethylene glycol, organic solvents, etc) and additives. Additionally, sample concentration and temperature affect the

Figure 1. 8. Work flow for an X-ray crystallographic structure determination. After crystal growth, diffraction is measured and the data processed to derive an electron density map and build a model.

crystallization process. Pipette robots set up plates by mixing the screening suite solution (precipitants) and protein samples in either hanging drop or sitting drop format (Chayen and Saridakis, 2008). A good starting point is to screen multiple conditions such as 3-4 different screening suites at two temperatures (4°C and 20°C) and two different protein concentrations. Crystal growth can be observed after a few days but also after a few weeks or months.

1.2.2.2 Data collection and Processing

X-ray diffraction pattern of crystals are measured at an in-house X-ray source/ diffractometer, if available, or at a Synchrotron light source. Before, crystal can be cryo- protected, so that the solvent around them is vitrified and prevents ice formation, which would lead to loss of diffraction. Various cryo-protectants (glycerol, ethylene glycol, sucrose, etc.) should be tested. Then, crystals are frozen in liquid nitrogen. Cryo-cooling helps preventing the crystals from radiation damage. Additionally, to avoid radiation damage, crystals are measured under a nitrogen gas stream.

Crystals are shot by X-rays and the diffraction is recorded when it hits the detector. X-rays are scattered by the electrons at a certain angle. These angles are derived by considering the diffraction to be reflections from parallel planes of atoms in a crystal. The hkl Miller indices

15 Introduction define these parallel planes. Thus, any reflection (spot on the screen) in the diffraction pattern is characterized by its index (hkl) and the reflection intensity (I). Crystals are rotated to obtain a complete data set (Garman and Schneider, 1997). The degree of rotation needed dependents on the internal symmetry of the crystal. The symmetry is described by a crystal lattice, which like a coordinate system defines the position of the atoms within a molecule. The lattice is comprised by a set of repetitive unit cells. The unit cells, described by the dimensions a, b, c (in Å) and angles α, β, γ (in °), is the scattering unit of the crystal. Therefore, the unit cell is derived directly from the diffraction pattern. The molecules within one unit cell are related by crystal symmetry. The asymmetric unit is the smallest unit of the cell, which cannot be transferred into another unit cell by symmetry elements (translation, rotation). From such an asymmetric unit, the entire crystal lattice can be built by applying symmetry elements. A combination of symmetry elements are referred to as space group. For each crystal, the unit cell dimensions and the space group needs to be determined (Smyth and Martin, 2000).

The position of the diffraction spots depend on the size of the unit cell as well as the position of the molecules in the unit cell. Thus, each spot corresponds to distances between molecules in the crystal. The size and shape of the molecules is encoded in the phases and intensities of the diffraction spots. To get the three dimensional electron density maps, the structure factor F and the Miller indices h, k, l of each reflection must be determined. Software such as HKL-2000 (Otwinowski and Minor, 1997) or XDS (Kabsch, 1993) determines all reflections by comparing the background with high intensity spots. Based on the position of all reflections, the unit cell dimensions of the crystal can be obtained (which is called “indexing”). The signal intensity and hkl values are obtained by integrating all reflections. Subsequently “merging and scaling” are performed; first, all peaks that appear in more than one image are identified (merging) and they are scaled such that they have consistent intensity (scaling). The structure factor contains information about the amplitude and the phase of a wave. Both of which are needed to generate an electron density map by Fourier Transformation (FT) (Smyth and Martin, 2000). The square of the amplitude is proportional to the signal intensity (which has been measured in the diffraction image), but the phase cannot be measured directly, leading to what is called the “phase problem”.

Several parameters enable us to judge the quality of the processed data: Resolution, Rmerge (accuracy), I/σ(Ι) (signal-to-noise ratio), redundancy and completeness. At high resolution (in Å), structural elements are better visible and a more precise model can be build. Rmerge (or Rsym) determines the accuracy of the data set. It is derived from differences in intensity between symmetry-related and unique reflections that should have identical intensities. Overall Rsym of 5% are very good and of more than 20% indicate severe problems of the data. I/σ(I) is the signal-to-noise ratio where I is the intensity of a unique reflection and

16 Introduction

σ(I) the deviation/error. The redundancy of the data indicates how many times a unique reflection has been measured. The completeness of the data shows the difference of unique reflections (not symmetry related) and the theoretical number for a given unit cell and a space group. The overall completeness should be 95-100% (Otwinowski and Minor, 1997; Wlodawer et al., 2008) .

1.2.2.3 Phase Determination

The structure factor contains information about the amplitude and the phase. To generate the electron density map, both is needed. However, only the amplitude can be determined directly. The phase can be determined by various methods. In Multiple Isomorphous Replacement (MIR), crystals are soaked or co-crystalized with a heavy-metal compound (mercury, platinum, uranium, etc.). The induced strong scattering from such an heavy atom allows to determine their location due to a change in intensities. Soaking with at least two different heavy-metal compound are required for a reliable phase determination. In Multiple Wavelength Anomalous Dispersion (MAD) different wavelengths around the absorption edge of a certain atom (anomalous scatterer) are used. The differences in the resulting diffraction patterns can then be used to reconstruct the phase. Besides soaking a heavy metal, incorporation of selenium is often chosen. A protein is expressed in presence of the amino acid seleno-methionine, where this residue is incorporated at the position of the methionines (Taylor, 2003).). If a homologue structure is available, the phase can be determined by Molecular Replacement (MR). A related structure gives the orientation and position of the molecules within the unit cell. This is used to estimate an initial phase which helps generate the necessary electron density map (Taylor, 2003).

1.2.2.4 Model building and Structure quality

To build the structure into an the electron density map, the software COOT (Emsley and Cowtan, 2004) is used. Software such as REFMAC5 (Vagin et al., 2004) or Buster (Smart et al., 2012) can be used to refine the structure and check the quality of the structure by calculating the structure factors R and Rfree after each manual change. The R-factor indicates how much the calculated structure factors from the model (Fcalc) and the observed structure factors (Fobs) deviate. The parameter Rfree is determined analogously to normal R-factor but excludes a random amount of reflections. Rfree is an important validation parameter and indicates over-fitting of the experimental data. Both values should decrease if refinement is

17 Introduction proceeding sensibly. Desirable R values are between 10 and 30%, which depends on the resolution. Rfree should not deviate from the R factor more than 7% (Wlodawer et al., 2008).

Occupancy is an additional parameter included in the refinement. Occupancy of an atom indicates the fraction of molecules in the crystals, where this atom occupies the position. If the position of all atoms is identical, then the occupancy is 1. In case of two conformations, for example due to a sidechain being 50% in one conformation and 50% in the other conformation, the occupancy is 0.5. Lower occupancy is observed for certain key nucleotides in chapter 2. Very dynamic protein regions are not visible in the electron density map. They occupy multiple positions which are averaged out in the electron density maps (Wlodawer et al., 2008). Therefore, loops or long sidechains that are too flexible and not stabilized by crystal contacts are often not visible and therefore missing in crystal structures. Moreover, the root- mean-square deviation (RMSD) indicate how much the model differ from geometrical parameters. Bond lengths RMSDs are expected to be around 0.02 Å. Additionally, further deviations of stereo-chemical parameters need to be controlled, such as the peptide planarity. The Ramachandran plot indicates outlier of φ/ψ torsion angles of the polypeptide backbone. 98% should lie in the allowed region (Wlodawer et al., 2008). Final structures are uploaded to the protein data bank (PDB). Papers showing crystal structures include the statistics of the data (resolution, Rsym, completeness, I/σ(Ι), etc. for the overall set and for the highest resolution shell) and the refinement parameters (R/Rfree, RMSDs, Ramachandran outliers, etc) to judge the quality of the data.

1.2.3 Characterization of protein-RNA and protein-protein complexes by NMR Spectroscopy

NMR Spectroscopy can provide a wide set of information about structural, mechanistic, thermodynamic and kinetic aspects of a biomolecular interaction (Waudby et al., 2016). NMR is based on the behavior of nuclei with spin ½ in a magnetic field. In biomolecules, 1H and 31P are naturally abundant and as they are spin ½ nuclei, they can be measured using NMR. On the other hand, the majority of naturally occurring 14N and 12C atoms have spin 1 or 0, respectively and isotope labelling methods are needed, to increase the fraction of 13C and 15N with spin ½. Isotopic labeling of proteins can be done by using a recombinant expression system and growing cells, for example E. coli, in a minimal medium supplemented with 15N-

13 NH4Cl and C glucose as the only nitrogen and carbon sources. In addition, deuterated proteins can be obtained by growing E. coli in a medium containing D2O instead of H2O. To produce labelled RNA, 15N and 13C labeled nucleotides are used during in vitro transcription.

18 Introduction

Active nuclei behave like magnets, align with the magnetic field and start to precesse around the magnetic field with a frequency

ʋ0= ω0/2π, called the Larmor frequency. The sum of the spins is called bulk magnetization. Radio frequency pulses can manipulate the orientation of the nuclear spins. The angular velocity ω0 of the precession depends on the static magnetic field B0 and is proportional to the gyromagnetic ratio γ:

ω0=γB0.

γ is an intrinsic property of a nuclei. Due to the dependence of the magnetic field B0, NMR frequencies are difficult to compare when measured at different spectrometers. Therefore, these frequencies are referenced with a specific reference compound to give the chemical shift:

6 δ=10 * (ʋspin- ʋref)/ ʋref, expressed in parts per million (ppm) (Keeler, J. Understanding NMR Spectroscopy/2nd ed. 2010). The chemical shift depends on the chemical environment. Within a protein, every amino acid atom has a different chemical environment and a characteristic chemical shift. When the atom experience a change in chemical environment, for example, when the amino acid is involved in ligand binding or the protein unfolds, the chemical shift changes.

Two types of spin-spin interactions, scalar coupling and dipolar coupling, are observable by NMR. The scalar coupling, also called J coupling, is indirect and mediated through a chemical bond while the dipolar coupling is based on the direct interaction of the two dipoles. The later depends on the distance between the two spins and their orientation relative to the magnetic field. Due to the orientation dependence, dipolar coupling cannot be observed in solution. There, spins are always in fast motion, adopting different orientations, such that the dipolar coupling is averaged out.

1.2.3.1 The chemical exchange

Protein-RNA or protein-protein interactions can be monitored by various NMR titration experiments. Unlabeled ligand or protein is titrated into a labelled protein in smaller steps until saturation is reached. Usually, 1H-15N Heteronuclear Single Quantum Coherence spectroscopy (HSQC) experiments are recorded. The 1H-15N HSQC spectrum is a ‘fingerprint’ of a protein showing one peak for each amino-acid NH group, where the position of the signal

19 Introduction represents its local chemical environment. To determine which signal corresponds to which amino acid in the protein, a protein backbone assignment needs to be done (1.2.3.3).

Upon ligand or protein binding, the local chemical environment and thus the position of the peak changes (Waudby et al., 2016). The appearance and position of the signal depends on

Figure 1. 9. Exchange regimes. Different exchange regimes of a protein-ligand interaction indicated on a 1D 1H spectra (A) and 2D 1H-15N HSQC spectra (B). Simulated spectra for a protein-ligand interaction showing line shapes under different exchange regimes. Adapted from Waudby et al, 2016.

the exchange rate kex between the free and bound conformation, relative to their frequency difference, Δω. When the exchange rate is smaller than the difference in frequency (kex ≪ Δω), two signals are observed, one corresponding to the free and one to the bound form. With increasing ligand concentrations, the intensity decreases for the free form and increases for the bound form. This exchange regime is called “slow exchange” (Figure 1.9, top). When the exchange rate is larger than the difference in frequency (kex ≫ Δω), only one signal is observed at the average frequency of both bound and free form. With increasing ligand concentrations, a progressive change in peak position is observed. This exchange regime is called “fast exchange” (Figure 1.9, bottom). If the exchange rate is close to the difference in frequency

(kex ≈ Δω), a more complex behavior is observed. The chemical shifts are broadened due to exchange and result in lower or invisible signal intensities. This is called “intermediate exchange” (Figure 1.9, middle) (Waudby et al., 2016). If intermediate exchange is observed, it is possible to shift the exchange regime by a change in temperature (increasing or decreasing the exchange rate) or magnetic field (change in resonance frequency). When the protein-RNA interaction is in fast exchange, the change of the signal position can be easily followed (Williamson, 2013). To quantify the chemical shift and map the binding surface, the combined chemical shift difference (ΔCS) between the free and bound state is generated according to:

20 Introduction

ΔCS= ( + ( ∗ ))

The 15N chemical shift needs to be scaled by α to consider the smaller gyromagnetic ratio of 15N. In case of slow or intermediate exchange, when it is unclear where the signal moved, the protein backbone needs to be reassigned.

Similarly, the appearance and position of the peaks depend on the exchange rate kex between different conformations. Thus, after saturation of the complex, signals might still be invisible due to exchange broadening of the resonances caused by two different states (Dominguez et al., 2011) such as the equilibrium between HuR RRM3 monomer and dimer and multiple register binding in chapter 2 or potential multiple conformations for the PABC1 RRM1-BTG2 complex in chapter 3.

To observe protein-RNA interactions on the RNA side, 1H-1H-Total Correlation Spectroscopy (TOCSY) spectra can be recorded (Dominguez et al., 2011). Cross-peaks in the TOCSY result from a correlation between the H5 and H6 resonances of uracil and cytidine nucleotides. The position depends on the local chemical environment, similar to the position of H-N correlations for each backbone amide of a protein in a 1H-15N HSQC spectrum. In the TOCSY of the complex between a RRM and for example AUUAU, three H5-H6 cross peaks are expected (one for each uridine). If a nucleotide is involved in binding, the position of the signal changes due to the different chemical environment. Interestingly, poly-nucleotide rich stretches provide multiple registers for an RNA binding protein. RRMs are typically comprised of two to eight binding pockets to bind RNA (Cléry et al., 2008). A poly-nucleotide binding protein cannot discriminate between individual nucleotides in a poly-nucleotide rich sequence. For poly(U) binding proteins, that are able to bind multiple registers (Cieniková et al., 2014; Mackereth et al., 2011), the binding is thought to be dynamic, comprised of binding and unbinding events which might result in a constant movement along the sequence. This results in an increase in affinity. TOCSY experiments can provide information about such dynamic binding. For example, a 5 nucleotide poly(U) sequence in a complex with an RRM, where five signals are expected, can show less cross-peaks. The others are exchange-broadened and therefore small or not visible, due to conformational chemical exchange, potentially due to multiple register binding or sliding.

1.2.3.2 Relaxation

Relaxation, where spins return to equilibria after being excited by a radio frequency pulse, is another process that influences the chemical shift. Spin relaxation provides dynamic

21 Introduction information of a system. There, we distinguish two types, the longitudinal and the transverse relaxation, with corresponding time constants T1 and T2. Relaxation experiments provide information about the dynamics of the system. T1 causes loss of the signal. The line width of the signal is determined by T2. When relaxation is very fast, NMR lines get broad and the signal becomes difficult or impossible to measure. T2 relaxation is highly dependent of the molecular motion, which is described by the correlation time τc (assuming a spherical shape).

15 τc is calculated from the ratio of N longitudinal (T1) and transverse (T2) relaxation times of each residue:

τc = 6∗ −7 ʋ

τc is the average time needed for a molecule to rotate one radian. Assuming that the molecule of interest has a spherical shape, one can estimate its approximate molecular weight (MW) (Rossi et al., 2010). With an increase in size, proteins tumble more slowly which leads to increased relaxation rates and broader lines. Higher temperature (increase in tumbling) and deuteration of a molecule have to be used for larger complexes. In deuterated proteins, the relaxation time of heteronuclear signals decreases, which leads to narrow linewidth and an increase in resolution and sensitivity.

The nuclear Overhauser effect (NOE) is another relaxation mechanism. It is based on dipole-dipole relaxation and the transfer of magnetization from a population of one type of nuclei to another (cross relaxation). The NOE is dependent on the orientation of the spins in the magnetic field. Thus, internal and global motions influence the NOE. 15N heteronuclar NOE (hetNOE) experiments can provide information about the flexibility of the protein backbone. For rigid residues, main-chain amides display NOE value around 0.8. Fast motions lead to a decrease of the NOE. Therefore, amides located in dynamic regions have an NOE smaller than 0.5 (Kay et al., 1989).

1.2.3.3 Resonance assignment and structure determination

To identify which signal in the 1H-15N HSQC spectrum corresponds to which amino acid in the protein, multiple 3D NMR experiments are recorded. There, backbone and sidechain resonances are correlated through scalar coupling, which provide information about connectivity of the protein spin system (Dominguez et al., 2011). The standard 3D experiment HNCACB correlate each backbone NH with the Cα and Cβ chemical shifts of its own and the preceding residue. Thus, strong Cα and Cβ for each residue and weak Cα and Cβ signals for the preceding residue are visible. HNCA correlate each backbone NH with the Cα chemical shifts of its own and the preceding residue. Further experiments provide information about

22 Introduction different atom correlations. Already the backbone chemical shifts N, HN, Cα and Cβ provide information about the secondary structure. The program TALOS+ uses these chemical shifts to predict the backbone torsion angles Φ and Ψ and generates a secondary structure plot (Shen et al., 2009).

Structure determination of a protein-RNA or protein-protein complex by NMR requires that resonances for most of the amino acids can be found and assigned. The distances between protons are obtained from Nuclear Overhauser Enhancement Spectroscopy (NOESY) experiments, which are based on the NOE. The NOE takes place through space and not through a chemical bond, enabling determination of intra- and intermolecular distances, which are needed for structure calculation. The distances (r) between the protons are obtained from the NOE signal intensities (I):

I ∝ r-6

All resonances from the free form should also be visible in the bound form. In addition, a sufficient number of intermolecular NOEs need to be observed to solve for structure of a complex.

1.2.4 Prediction of protein complexes by HADDOCK

High-resolution structural information about biomolecular complexes is usually acquired by X-ray crystallography, NMR spectroscopy or Cryo-EM. However, many challenges, such as large systems, presence of flexible or unstructured regions, small amounts or solubility of the complex, can make structure determination challenging. Over the past years, several methods were developed to model biomolecular complexes based on the known structures of its constituents (Smith and Sternberg, 2002). The docking program HADDOCK (High Ambiguity Driven protein-protein Docking) uses biochemical and/or biophysical interaction data such as NMR chemical shift mapping, RDCs, NOEs, mutagenesis experiments, etc (Dominguez et al., 2003; Van Zundert et al., 2016). HADDOCK has been applied already to dock a large variety of systems, including protein-protein, protein-nucleic acids and protein- small molecules (Karaca and Bonvin, 2013). A considerable number of experimental structures of complexes calculated using HADDOCK have been deposited into the Protein Data Bank (PDB) (De Vries et al., 2010). A dedicated web server (http://haddock.science.uu.nl/services/HADDOCK2.2) is available on-line. In chapter 3, this HADDOCK webserver - the Easy interface- was applied. We aimed to determine the PABPC1 RRM12-BTG2 complex, as the structure determination by NMR and X-ray crystallography

23 Introduction proved to be challenging. The PDB files of the free proteins and the residues obtained by NMR chemical shift mapping were used for the docking. These interacting residues are incorporated

Figure 1. 10. HADDOCK clustering and examples. (A) Flow chart of the HADDOCK clustering. (B) Examples of a HADDOCK output of the E2A (pdb code 1F3G) and HPr (pdb code1HDN) clusters (top) and the structures displayed by pymol (bottom). In cluster 2, one of the proteins is rotated around 90°. The E2A-HPr docking results can be viewed at http://haddock.science.uu.nl/services/HADDOCK2.2/Files/E2A-HPR.

as ambiguous interaction restraints (AIRs) to drive the docking. During HADDOCK, one protein is fixed in space and the second one is rotated and translated around the first one (precise docking protocol described in (Dominguez et al., 2003). The final structures are clustered based on the backbone root-mean-square deviations at the interface (iRMSD) (between 0.8 and 2 Å). A cluster is defined as an ensemble of a few conformations displaying a user defined iRMSD (e.g. smaller than 1.0 Å). The resulting clusters are analyzed and ranked according to their HADDOCK score, which is the sum of electrostatic interactions, van der Waals repulsion, average buried surface area, etc. (Figure 1.10A). The Z-score indicates how many standard deviations from the average this cluster is located in terms of score (the more negative the better) (De Vries et al., 2010; Dominguez et al., 2003). As the search is

24 Introduction performed through the entire conformational space of the complex geometry, the output could result in multiple solutions. Figure 1.10B shows a HADDOCK output for the proteins E2A (pdb code 1F3G) and HPr (pdb code 1HDN) and the visualization of two clusters with the program Pymol (The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC.). Both proteins, E2A and HPr are involved in the E. coli sugar phosphotransferase system and are shown on the HADDOCK webserver as example. The docking was driven by NMR chemical shift mapping data. Both clusters show a similar HADDOCK score, however, one of the proteins is rotated around 90°. Therefore, predicted HADDOCK models should be validate by additional experimental techniques.

1.2.5 ITC to study protein-RNA and protein-protein interactions

ITC is a powerful and widely used method to study molecular interactions. ITC provides thermodynamic information of molecular interactions, affinity and stoichiometry of a biomolecular complex. Especially for protein RNA interactions, where the effects of protein mutations or different RNA sequences can be tested. The instrument has a reference cell and a reaction cell, which are connected by a thermoelectric device that determines the temperature differences between the two cells. This is achieved by measuring the power differences needed to keep the temperature at a predetermined value (Figure 1.11A). One binding partner (e.g. protein) is transferred into the sample cell while the second binding partner (e.g. RNA or other ligand) is loaded into the syringe. When the ligand is injected into the sample cell, the ITC machine measures the heat that is released or absorbed in an exothermic or an endothermic reaction respectively. The injections are repeated at a certain time interval. With time, heat changes should decrease due to saturation of the system (Figure 1.11B). Towards the end of the titration only heats of dilution are measured. Data is analyzed with the Origin® software using a fitting model to determine the reaction stoichiometry (N), binding affinity constant (Ka), enthalpy (∆H), and entropy (∆S). While N, Kd (inverse of the affinity constant Ka) and ∆H are derived from the integrated heat changes (Figure 1.11C), the molar standard free energy is derived using:

ΔG = -RT ln Ka Thus, ∆S is determined by:

TΔS =∆H - ∆G The enthalpy component of a system is defined by its polar interactions, while hydrophobic interactions contribute to entropic components. Therefore, ITC thermodynamic parameters

25 Introduction provide various insights about a binding mechanism. ITC can also be used to study dissociations of homodimers. A dimeric protein is titrated into a buffer, which leads to

Figure 1. 11 ITC setup, titration curve and analysis. (A) ITC instrument with major features noted in the figure; adapted from (Freyer and Lewis, 2008). (B) Power vs. time; each peak represents an injection (from a RBM38-

DNA measurement). (C) Integrated representation of the data. Determination of the Kd, N and ΔH is indicated in the figure.

dissociation into monomers. Dissociation constant Kd, ∆H and ∆S are determined by a fit using a dissociation model (Zhang et al., 2011). Three important parameters affect the binding isotherm. The c value defines the shape of the binding isotherm and is determined by the binding constant Ka, the concentration of the sample before the start of the measurement Mtot and the stoichiometry value N:

c=Ka*Mtot*N

Higher c-values (>1000) lead to very tight binding and result in good determination of N and

∆H but makes it impossible to determine Ka. Lower c-values (<10) on the other hand lead to flatter curves, where neither of the parameters (N, K, ∆H) can be determined correctly (Figure 1.12 (Turnbull and Daranas, 2003).

Another important aspect to take into consideration is the heat of dilution. Every injection causes heat changes due to interactions but also due to the injection of the ligand solution into the buffer, summing up to the total heat of a reaction. Heat of dilution arises due to friction forces of the ligand injected into the cell but also due to slight temperature changes. Additional buffer mismatches can cause strong heats of dilution. It is therefore important to use the identical buffer in both the sample cell solution and syringe solution to prevent buffer mismatches. Heat of dilution and slight buffer mismatches are only critical when the heat of a reaction is too low to be distinguished from the dilution heat. Such effects can be subtracted

26 Introduction from the main data, by performing additional experiments where either the ligand is titrated into buffer or buffer titrated into sample cell with the binding partner.

Figure 1. 12. c-value determines the shape of the ITC curve. Unfavorable c-values are <10 and >1000. (Turnbull and Daranas, 2003).

Finally, binding heats depend strongly on the temperature. Low heat values during a reaction can come from low sample cell concentrations but also from working near temperatures at which the heat of the reaction is zero (∆H=0) (as was probably observed in chapter 3). There, the signal-to-noise ratio is poor and leads ill defined binding isotherms. It is not possible to distinguish the heat of dilution and the heat of reaction. Many systems have a ∆H of zero between 20 and 30°C. Varying the temperature results in a change of the binding enthalpy and an increase in heat changes. However, variations of 0- 10°C only lead to low heat changes. For systems showing no heat changes at 25°C, temperatures at <15°C or >35°C can be tested (ITC Expert User’s Manual). However a stable protein sample at these temperature ranges is a prerequisite.

1.2.6 Dual Luciferase Reporter Assay

To validate protein-RNA interactions in cells, one can choose multiple techniques, based on the questions one wishes to answer (Schlundt et al., 2017). In case of a protein playing a role in mRNA stability and/ or translation regulation, a luciferase reporter assay can be used (Doller et al., 2008; Léveillé et al., 2011). Standard laboratory cell lines, such as the human embryonic kidney cells 293 (HEK 293), HeLa or hepatocarcinoma cell line (Huh7) can be easily transfected and thus can be used for such an assay. Various methods can be used to transfect eukaryotic cells with DNA and should be tested to optimize the conditions that

27 Introduction

Figure 1. 13 Dual Luciferase Reporter Assay. (A) Schematic representation of the construct used, transfection, harvest, read out and analysis. (B) Bioluminescent reactions catalyzed by firefly and Renilla luciferases. Adapted from Dual-Luciferase® Reporter Assay System, Technical Manuel.

guarantee the highest transfection efficiency. Commonly, cationic lipids such us the lipofectamine reagent (Invitrogen; used in chapter 2) are used. These form a cationic lipid- DNA complex, which is able to pass the cell membrane and deposit the DNA inside the cell. Transfection efficiency can be controlled by transfecting a vector coding for a GFP-labelled protein. Celular fluorescence intensity directly correlates with transfection efficiency. Additionally, GFP fluorescence can be used to determine the peak of protein over expression and to decide whether cells can be harvested after 24 or 48 h. For the luciferase assay, a plasmid coding for the protein of interest (for example Flag tagged HuR, which is cloned in the

28 Introduction pcDNA vector, as described in chapter 2) is transiently co-transfected with a plasmid harboring the genes for Renilla luciferase (Renilla reniformis). Additionally, firefly luciferase (Photinus pyralis) is encoded on the same plasmid (psiCHECK2 vector; Promega). As two plasmids are co-transfected, different ratios of both plasmids can be tested to get an optimal readout. One of the luciferase enzymes is coupled to the RNA of interest, the other serves as an internal control to normalize for differences caused by viability of cells, transfection efficiency or lysis (Figure 1.13A). If Renilla is used as the primary reporter gene, the gene of interest, for example a 3’UTR sequence comprised of AREs (as the 60 nucleotide COX-2 sequence in chapter 2), which leads to rapid degradation of the mRNA target, is cloned downstream of the Renilla translational stop codon. By co-transfecting the reporter with a protein expected to bind and stabilize the reporter, an increase in Renilla protein expression and thus Renilla activity is expected (Figure 1.13A). The internal firefly control reporter should not be affected. Both firefly and Renilla luciferase have different enzyme structures and use different substrates. This makes it possible to discriminate between both luciferases. The firefly luciferase uses beetle luciferin as a substrate, which is oxidized and emits photons. The Renilla luciferase catalyzes a reaction by using coelenterazine as a substrate (Figure 1.13B). Usually, 24 or 48 h after transfection, the dual-luciferase assay is performed. First, cells are lysed and separated from the debris by centrifugation. The assay is usually done in 24 or 96 well plates with 3-4 replicates per condition. In each single well, the supernatant is mixed with the luciferase substrate for firefly enzyme and the firefly luminescence measured. Then, luciferase substrate for Renilla is added into the same well, which is quenching the firefly activity and activating the Renilla activity instead. The fold change is calculated by dividing the Renilla activity by the firefly activity (Dual-Luciferase® Reporter Assay System, Technical Manuel). All cell-based experiments needs to be repeated at least 3 times independently. The firefly luciferase enzyme indicates whether the same well-to-well transfection efficiencies of the luciferase plasmid are achieved. The transfection efficiently of the plasmid with the protein of interest or the expression levels of the protein of interest needs to be additionally controlled by a western blot. Especially when comparing significant effects of protein mutants, one need to assure that these effects are not due to reduced mutant expression levels.

29 Introduction

1.3 Research Objectives

Outline of the thesis

The goal of this research is to understand the structural and functional basis of key RNA binding proteins playing a role in mRNA stability and decay. We aimed to elucidate the mechanism of multifunctional RRMs, which are able to bind proteins as well as RNA by two different binding surfaces and how both interacting factors influence each other. Here, we focus on the RBPs HuR and PABPC1, both composed of canonical RRMs.

Chapter 2: Molecular basis for AU-rich element recognition and dimerization by the HuR C-terminal RRM

HuR/ ELAVL1 is a major regulator of cellular ARE containing mRNAs. While RRM1 and 2 are structurally characterized and assigned to be critical for ARE-binding, little is known about the RRM3 function. In chapter 2, we present a 1.9-Å-resolution crystal structure of RRM3 bound to several short ARE-motifs. In combination with other biophysical methods and cell culture assays, we found that recognition of the canonical AUUUA pentameric motifs is possible by binding to two registers and that RRM3 homo-dimerization increases RNA affinity, which is functionally relevant for HuR. Finally, despite the role of HuR in stabilization of ARE-containing RNAs, we found that RRM3 counteracts this effect in a cell-based ARE reporter assay containing multiple AUUUA motifs.

Chapter 3: Structural basis of the PABPC1 RRM1-BTG2 interaction to recruit CAF1 deadenylase

Deadenylation is a key step of mRNA decay in eukaryotic cells. BTG2, which play a central role in cell cycle regulation, differentiation and cancer, recruits CAF1 deadenylase to mRNA poly(A) tails by interacting with PABPC1 and cause poly(A) tail shortening. In chapter 3, we show that the BTG2 APRO boxC region recognizes mainly the helix α1 of PABPC1 RRM1. In addition, we show evidence that BTG2 seems not to interfere with poly(A) binding, suggesting a stable ternary complex formation comprised of BTG2-RRM1-poly(A). In combination with our modeling approach, our data provide the first structural information of how BTG2 recognizes and directly binds PABPC1 and support a model how BTG2 bridges CAF1 to PABPC1 bound poly(A)-RNA.

2. Molecular basis for AU-rich element recognition and dimerization by the HuR C-terminal RRM

Nina Ripin1,2, Julien Boudet1, Malgorzata M Duszczyk1, Alexandra Hinniger2 , Michael Faller2, Miroslav Krepl3,4, Abhilash Gadi5, Robert J. Schneider5, Jiří Šponer3,4, Nicole C. Meisner- Kober2, Frédéric H.-T. Allain1

This work is currently submitted and under review

1 Institute of Molecular Biology and Biophysics, Department of Biology, ETH Zürich, 8093 Zürich, Switzerland 2 Novartis Institutes for Biomedical Research, Novartis Campus, 4002 Basel, Switzerland 3 Institute of Biophysics of the Czech Academy of Sciences, Kralovopolska 135, 612 65 Brno, Czech Republic 4 Department of Physical Chemistry, Faculty of Science, Regional Centre of Advanced Technologies and Materials, Palacky University Olomouc, 17.listopadu 12, 771 46 Olomouc, Czech Republic 5 Department of Microbiology, New York University School of Medicine, New York, New York 10016, USA

Author contributions

N.R., N.M.K. and F.H.-T.A. initiated and designed the project. N.R. prepared the GB1-tagged and Flag-tagged constructs including several dimerization- and RNA binding inactive mutants, purified all proteins, solved the crystal structure, performed and analyzed NMR experiments, performed, processed and analyzed ITC experiments, cloning of cell culture constructs, cell culture experiments in Huh7 cells and did western blots. A.H. collected crystal diffraction data at the SLS synchrotron, A.H. and M.F. processed them with XDS and assisted with molecular replacement, model building and initial refinements. J.B. designed, set up, performed and analyzed ITC experiments to determine Kddimer, processed and analyzed 6mers/c-fos RNAs ITC data. M.K. and J.S. designed, set up and analyzed the MD simulations. R.J.S. and A.G. set up and performed cell culture experiments in C3H/10T1/2 cells and did western blots. N.R., F.H.-T.A. and M.D. wrote the manuscript. N.M.K. provided the HuR pTXB1 construct. All authors edited and approved the manuscript.

HuR RRM3 dimerization and AU-rich element recognition

2.1 Abstract

HuR is a key regulator of cellular mRNAs containing adenylate-uridylate-rich elements (AU- rich elements; AREs). AREs are a major class of cis-elements within the 3’ untranslated regions (3’-UTR) targeting these mRNAs for rapid degradation. HuR contains three RNA recognition motifs (RRMs); a tandem RRM1 and 2, followed by a flexible linker and a C- terminal RRM3. While RRM1 and 2 are structurally characterized, little is known about RRM3. Here we present a 1.9-Å-resolution crystal structure of RRM3 bound to several short ARE- motifs. This structure together with other biophysical methods and cell culture assays allowed us to describe the mechanism of RRM3 ARE recognition and dimerization. We found that recognition of the canonical AUUUA pentameric motifs is possible by binding to two registers and that RRM3 homo-dimerization increases RNA affinity, which is functionally relevant for HuR. Finally, although HuR stabilizes ARE-containing RNAs, we found that RRM3 counteracts this effect in a cell-based ARE reporter assay containing multiple AUUUA motifs, possibly by competing with RRM12 RNA binding.

32 HuR RRM3 dimerization and AU-rich element recognition

2.2 Introduction

Adenylate-uridylate-rich elements (AU-rich elements; AREs) are regulatory cis-acting elements within untranslated regions (UTRs) of short-lived mRNAs which function as a signal for rapid degradation. AREs are present in 5-8% of human genes involved in the regulation of many important cellular processes such as stress response, cell cycle regulation, inflammation, immune cell activation, apoptosis, carcinogenesis (Bakheet et al., 2006) and must therefore be tightly regulated. There are three classes of AREs. Class I contains several AUUUA motifs dispersed over the 3’UTR, class II multiple, overlapping copies of AUUUA and class III is U-rich (Benjamin and Moroni, 2007). Multiple RNA binding proteins (RBPs) regulate the transport, stability and translation of ARE containing mRNAs. An important class of such RBPs is the embryonic lethal abnormal visual like/ human antigen (ELAVL/Hu) protein family. In mammals, the Hu family is comprised of four highly conserved members. HuB/Hel-N1, HuC and HuD are expressed in neurons whereas HuR (also known as ELAV-like protein 1 (ELAVL1) or HuA) is ubiquitously expressed (Antic and Keene, 1997; Fan and Steitz, 1998b; King et al., 1994; Ma et al., 1996). HuR positively regulates the stability and translation of ARE containing targets (J. Wang et al., 2013). It is also known to destabilize a small number of mRNAs and/or to suppress their translation (Cammas et al., 2014; Kim et al., 2009; Leandersson et al., 2006; Meng et al., 2005). cDNA arrays and PAR-CLIP experiments identified a rather U-rich than AU-rich motif as HuR binding sequence within thousands of targets (Lebedeva et al., 2011; López de Silanes et al., 2004), consistent with previous biochemical studies defining a consensus motif of NNUUNNUUU (Meisner et al., 2004). HuR is predominantly localized in the nucleus but undergoes cytoplasmic translocation under various cellular and stress conditions (J. Wang et al., 2013), where it stabilizes its targets and promotes translation. In the nucleus, HuR was reported to play a role in splicing and polyadenylation (Chen et al., 2007; Dutertre et al., 2014; Izquierdo, 2010; Lebedeva et al., 2011; Mukherjee et al., 2011; Zhu et al., 2007). Cellular localization of HuR as well as HuR-RNA binding are in addition regulated by various posttranslational modifications, including phosphorylation (Abdelmohsen et al., 2007; Hyeon et al., 2008; Kim et al., 2008), methylation (Li et al., 2002) and caspase-mediated cleavage during apoptosis (Mazroui et al., 2008). Moreover, HuR counteracts miRNA mediated repression upon stress to protect mRNA from degradation (Bhattacharyya et al., 2006). Furthermore it has been reported that HuR competes or cooperates with miRNAs in regulation of mRNA stability and translation (Srikantan et al., 2012). Mice lacking HuR show higher rates of embryonic lethality (Katsanou et al., 2009). Conditional HuR knock-out animals revealed the essential role of HuR in organ development and tissue homeostasis (Zucal et al., 2015). Elevated HuR expression levels are associated with altered physiological functions, e.g.

33 HuR RRM3 dimerization and AU-rich element recognition

promoting viral infections, cardiovascular diseases, neurological pathologies and muscular disorders (Di Marco et al., 2005; Farooq et al., 2009; Figueroa et al., 2003; Li et al., 2009; Misquitta et al., 2001; Sokoloski et al., 2010; Van Der Giessen et al., 2003). Predominantly, HuR overexpression increases upregulation of cancer associated transcripts, correlating with tumor growth and disease progression in various cancer types, e.g. breast-, colon-, ovarian-, prostate-, pancreatic- and oral cancer (Kotta-Loizou et al., 2016; Srikantan and Gorospe, 2012; J. Wang et al., 2013). Moreover, HuR is involved in the stabilization of transcripts encoding drug-resistant proteins, which leads to drug resistance in cancer (Filippova et al., 2011; Hsia et al., 2013). Consequently, HuR emerges as a potential biomarker and therapeutic drug target, and deciphering its molecular function and details of its RNA target recognition will be directly relevant for informed progression. HuR is composed of three highly conserved canonical RNA recognition motifs (RRMs). These domains adopt a βαββαβ topology, where a four-stranded antiparallel β-sheet is packed against two α-helices (Maris et al., 2005). The tandem RRM1 and 2 (RRM12) is separated from the C-terminal RRM (RRM3) by a ~50- residue basic region (hinge region) bearing the nucleocytoplasmic shuttling element (Fan and Steitz, 1998a). RRM12 is suggested to be critical for ARE-binding by HuR (Chen et al., 2002), whereas the exact function of RRM3 is not fully understood. Previously, RRM3 was predicted to bind poly-A tails of mRNAs (as shown for HuC and HuD (Abe et al., 1996; Gao and Keene, 1996; Ma et al., 1997). RRM3 and the hinge region have been identified to be involved in protein-protein interactions (Brennan et al., 2000) and in homo-dimerization (Scheiba et al., 2014; Toba and White, 2008). Involvement of RRM3 in RNA binding and mRNA stability is controversial (Chen et al., 2002; Fan and Steitz, 1998b). Gel-based assays demonstrated that the hinge region and RRM3 play a role in cooperative oligomerization on long AREs (Fialcowitz-White et al., 2007). The hinge region as well as RRM3 are important for counteracting miRNA mediated repression and promoting miRISC release from target mRNAs (Kundu et al., 2012). Precisely, HuR binds the miRNA miR-122 and replaces Argonaute 2 (Ago 2) while the truncated version of HuR lacking RRM3 resulted in weaker miR-122 binding (Mukherjee et al., 2016). The crystal structures of the free HuR RRM12, HuR RRM12 in complex with an 11 nucleotide fragment of the 3’UTR c-fos ARE (H. Wang et al., 2013) and HuD RRM12 in complex with 11 nucleotide fragments of c-fos and TNF-α AREs (Wang and Tanaka Hall, 2001) were solved previously. Solution structures of the free HuC RRM1 and RRM2 have been solved separately and ARE recognition of tandem RRM12 has been characterized by NMR (Inoue et al., 2000). Structural studies of the hinge region and the RRM3 domain within the whole Hu family have proven to be challenging, due to insolubility in recombinant expression systems and instability of the domains in vitro. Recently it was shown by NMR that dimerization of RRM3 is mediated through α1, which is located opposite to the RNA binding interface and that RRM3 prefers 5’UUUUU-3’ over 5’AUUUA-3’ motifs (Scheiba

34 HuR RRM3 dimerization and AU-rich element recognition

et al., 2014). Although, a structure of the free RRM3 and two potential RRM3 dimer models were described in that study, the lack of an atomic scale analysis of the free and RNA-bound forms prevented a complete understanding of HuR RRM3 function. In this context, a mechanistic description of HuR RRM3 dimerization and RRM3-RNA recognition is needed. To understand the function of HuR RRM3 in ARE regulation, we determined a 1.9 Å crystal structure of a RRM3 homo-dimer in complex with short ARE motifs derived from the c- fos 3’UTR. This revealed the first molecular insights into RNA binding by HuR RRM3 and HuR dimerization. We tested the importance of RRM3 RNA binding and dimerization for ARE mRNA regulation by combining biophysical measurements in solution and functional studies using mutant HuR expression constructs in a cell-based ARE reporter assay .

2.3 Results

Both HuR RRM12 as well as RRM3 bind ARE motifs

To understand the molecular basis of HuR RRM3 in gene regulation, we analyzed the RNA interactions of HuR RRM3 (residue 241-326, Figure 2.1A) using NMR chemical shift mapping and isothermal titration calorimetry (ITC). Titrations with c-fos ARE 11-mer (5’- AUUUUUAUUUU-3’, Figure 2.1B) and 15N-labeled RRM3 or 15N-labeled RRM12 revealed chemical shift variations mostly in intermediate exchange regime for RRM3 and slow to intermediate exchange regime for RRM12 on the NMR time scale (Figure 2.1C,D left). This is compatible with a low μM affinity of this RNA motif. Binding affinity measurements by ITC of c-fos ARE 11mer for RRM3 and RRM12 confirmed the NMR results with Kd values of 0.65 μM and 0.14 μM, respectively (Figure 2.1C,D right). The affinity of RRM12 is consistent with the previously reported Kd of 0.20 μM (H. Wang et al., 2013). Interestingly, for RRM3, an ITC N value of 0.5 suggests that two RRM3 molecules bind one c-fos ARE 11mer. RRM3 was also predicted to be involved in binding of poly-A tails of mRNAs, as shown for HuC and HuD (Abe et al., 1996; Gao and Keene, 1996; Ma et al., 1997) and to possess an ATP-binding pocket, comprised of a metal-ion-coordinating DxD motif, which mediates 3- terminal adenosyl modification of non-polyadenylated RNA substrates (Meisner et al., 2009).

1 15 15 H- N HSQC titrations of N-labeled RRM3 with EDTA, MgCl2 or ATP showed no significant chemical shift perturbations (Supplementary Figure A2.1 A), indicating no binding to RRM3 under these conditions. Titrating an 11 nucleotide A-rich RNA motif revealed smaller chemical shift perturbations at higher protein to RNA ratio (Supplementary Figure A2.1 B) compared to c-fos ARE 11mer, indicating a weaker binding to poly-A than to the ARE motif. These findings imply that RRM3 preferentially binds A/U containing RNAs and that both RRM12 as well as

35 HuR RRM3 dimerization and AU-rich element recognition

Figure 2. 1. All three HuR RRM domains are involved in ARE binding. (A) Schematic representation of HuR domain organization, with the RRM12 and RRM3 construct length indicated. RRM3 amino acid sequence including the secondary structure is highlighted. (B) Schematic representation of c-fos ARE with the 11mer motif indicates which was used for the in vitro experiments. (C) 1H-15N HSQC titrations (left) and ITC binding curves (right) of RRM12 with c-fos ARE 11mer. (D) 1H-15N HSQC titrations (left) and ITC binding curves (right) of RRM3 with c-fos ARE 11mer. 1H-15N HSQC spectra of free RRM3 or RRM12 (blue) overlaid with protein:RNA molar ratio of 1:0.25 (red), 1:0.5 (yellow), 1:1 (green) and 1:1.5 (cyan). Negative peaks corresponding to the amides of arginine side chains in the free and the RNA bound form in green. Examples of different exchange regimes are indicated by a box and shown in a close-up view on the right. ITC curves are shown from one measurement. Measurement was repeated three and two times for RRM3 and RRM12 respectively. Top: ITC raw data as a function of time, bottom: heat release as a function of molar RNA:protein ratio. Red line represents the fit of the Origin 7 Software with a one binding site model. Errors indicate s.d. of two or three measurements.

36 HuR RRM3 dimerization and AU-rich element recognition

RRM3 are able to bind ARE motifs in a similar affinity range. This suggests that RRM3 might play a more relevant role in direct interaction with ARE containing mRNAs than previously expected.

Crystal structure shows the first molecular insights into RRM3 ARE recognition

To further understand the structural basis for ARE recognition by RRM3, we crystallized RRM3 in complex with the c-fos ARE 11-mer. Crystals diffracted up to 1.9 Å resolution and the crystal structure was determined by molecular replacement using the structure of HuR RRM1 (PDB code 3HI9) (Figure 2.2, Supplementary Table A2.1). The crystals belong to C 121 space group and contain four proteins and four RNA molecules per asymmetric unit (Figure 2.2A). We refer to each RRM in the asymmetric unit with chain labels A-D and each RNA molecules with the chain labels E-H (Figure 2.2C). Each RRM adopts the expected βαββαβ fold, where four antiparallel β-strands are packed against two α-helices (Figure 2.2A,B). All four RRM molecules in the asymmetric unit adopt the same conformation, except that some electron density is missing for the β2-β3 loop for chains B and D (Supplementary Figure A2.2 A,B). However, the RNA electron densities of chains E-H, to which we refer binding pockets according to their position, reveal variable numbers of bound nucleotides (Figure 2.2C, Supplementary Figure A2.2 A). Out of 11 nucleotides of the c-fos ARE 11mer, electron density for RNA is visible for only 3, 4 or 5 nucleotides bound to the RRM. Our crystal structure shows binding of RRM3 to pure single-stranded U-rich motifs (F and H), as previously suggested (Scheiba et al., 2014), as well as binding to AUUUU and UUUA (E and G) (Figure 2.2). These sequences are found in the c-fos ARE 11mer. The remaining unbound nucleotides of the 11mer are either too flexible to be observed or might have been degraded during the crystallization process. The electron density found in all four RRM3-RNA structures clearly indicates that only uracils can be bound in pockets 2 and 3. For the nucleotide in chain E and G in pocket 1 and 4, the electron density could not perfectly accommodate an uracil or an adenine (Supplementary Figure A2.2 A), which is explained by the occupancy of both adenines and uridines in these pockets. For our final structure (Figure 2.2A), adenines were placed in pocket 1 of chain E and in pocket 4 of chain G considering that in pocket 1 of chain G and in pocket 4 of chain E, F, H we have placed already uracils. See materials and methods section for a detailed description of the final structure determination. Overall, the structure revealed that RRM3 has a binding preference for uracil in pocket 2 and 3 and for both uracil and adenine in pocket 1 and 4.

Another difference was observed by overlaying all four RNA chains (Supplementary

Figure A2.2 B, right). Chain E shows the presence of an additional fifth pocket, containing U5,

37 HuR RRM3 dimerization and AU-rich element recognition

Figure 2. 2. The crystal structure of HuR RRM3 in complex with RNA shows homo-dimers and canonical RNA recognition. (A) Cartoon representation of the four RRM-RNA molecules (RRMs chain A to D and RNA chain E-H) in one asymmetric unit. Chains A:B (green:cyan) and C:D (green:cyan) present the same dimer interface. RNA is shown in stick representation and the heavy atoms are indicated in yellow (carbon of uridines), magenta (carbon of adenines), red (oxygen), blue (nitrogen) and orange (phosphorus). (B) RRM3 of chain A and C is shown as cartoon representation. RNP sidechains and RNA are shown in stick representation. Color scheme as in A. Nucleotide pockets are abbreviated with p. (C) A schematic representation of the RNP arrangement is colored in grey, uracils and adenines in yellow or magenta respectively and aromatic residues of the RNP in green (chain A/C) or cyan (chain B/D). Protein and RNA chains, pocket number (abbreviated with p. #) and the nucleotides are shown at the top. (D) Protein-RNA contacts shown as sticks for each chain. Chains are indicated at the top. Nucleotide pockets (abbreviated with p.) are indicated below. Left and right panels show interactions distinct to each chain in pockets 1 and 4 and the central panel shows pocket 2 and 3 interactions representative of all four chains. Color scheme of the RNA as in A. Hydrogen bonds are shown by a red dashed line.

38 HuR RRM3 dimerization and AU-rich element recognition

which does not exhibit any stacking contacts or hydrogen bonds to the protein. U5 is pulled away from the RRM due to interactions with a symmetry related molecule (Supplementary Figure A2.2 D). Such conformation is unlikely to be present in solution and therefore not biologically relevant.

Three to four RNA nucleotides are bound on the surface of RRM3 β-sheet using the conserved nucleic acid-binding ribonucleoprotein motifs RNP1 and RNP2, and amino acids within the β2-β3 and β1-α1 loops (Figure 2.2B,C). In pockets 2 and 3, U2 and U3 form stacking interactions with RNP2 Y249 and F247 (both β1), respectively. In addition, U2 is sequence- specifically recognized by forming two hydrogen-bonds with the Q316 side-chain (β4), while

U3 forms multiple hydrogen-bonds. One is with the main-chain amide of T321, one with the

T321 side chain and two are formed by water molecules bridging S318 and U3. Finally, the

K320 side-chain forms two hydrogen-bond contacts with the 2’- and 3’-oxygens of U3

(Figure 2.2D) and the phosphate oxygen of U4/A4. The S318 side chain was found to be phosphorylated by PKCδ influencing the HuR function (Doller et al., 2011, 2008). This is in agreement with our structure and reduced RNA binding affinity previously reported for the phosphomimetic RRM3 S318D (Scheiba et al., 2014). Unlike pockets 2 and 3, which consistently bind uracils, pockets 1 and 4 are able to accommodate both adenine and uracil.

A1/U1 contacts are mainly mediated by hydrophobic interactions with the aliphatic side chain of K285 (Figure 2.2D). However, adenine at this position exhibits stronger stacking interactions due to its larger aromatic ring. Interestingly, HuR RRM1 (pdb 4ED5) and HuR RRM3 both show similar interactions with an adenine in this pocket (Supplementary Figure A2.2 C). Stacking with L61 in RRM1 is equivalent to K285 in RRM3 and Q29 in RRM1 which forms two sequence-specific hydrogen bonds with the adenine, is conserved in RRM3 as

Q253. In addition, a water mediates interactions between A1/U1 and T281. An uracil at this position is only able to form one hydrogen bond to Q253. Thus, the structure clearly explains how both A and U can be accommodated in pocket 1. The uridine in pocket 4 reveals imperfect stacking with F289 (β3), but also potential water-mediated hydrogen bonds with N272 (β2) and T291 (β3) in chain B and D (Figure 2.2D). Remarkably, the adenine at the same position in chain C, still stacks imperfectly with F289 but is able to form two direct hydrogen bonds with N272 (β2) and C245 (β1) (Figure 2.2D). The observation of both A and U stably bound in pocket 1 and 4 within the different RRM molecules suggests multiple RNA binding modes relative to the bound sequences (AUUUU, UUU, UUUA). This led us to propose the (A/U)UU(A/U) as the recognition sequence for RRM3.

39 HuR RRM3 dimerization and AU-rich element recognition

Molecular dynamics simulations of show large-scale dynamics of adenine and uracil binding in pocket one 1 and four 4

Considering the degenerate specificity revealed by the structure of HuR RRM3 bound to RNA, we performed molecular dynamics (MD) simulations to evaluate the stability of the four binding pockets. The simulation revealed stable binding for the uracils in pocket 2 and 3. In contrast, nucleotides in pockets 1 and 4 (which can be both uracil and adenine) were found highly dynamic, forming many different transient interactions within the protein with both direct and water-mediated hydrogen-bonds. In all simulations, a nucleotide located at the binding pocket 1 was continuously fluctuating with the exception of an adenine base adopting a syn conformation. When uracil is bound in this pocket, mainly two different binding modes were observed. In one, the protein/RNA hydrogen-bonds were formed with the K285, K313, and

Q253 side-chains. In another, a water was mediating an interaction between U1 imino and the carbonyl oxygen of C284 (Supplementary Figure A2.2 E, left). The uracil also occasionally interacted with the N250 side-chain or backbone carbonyl of L251 (not shown). When an adenine is bound in pocket 1, the nucleotide interaction with the protein was dependent on the orientation of the base. A base in the anti conformation was highly dynamic, similarly to the uracil. The anti-adenine formed interactions with the Q253 side-chain and a water-mediated interaction with C284 carbonyl oxygen. Occasional water-mediated interaction with the carbonyl of L251 was also observed. Interestingly, when adenine was present in a syn conformation, it was more stably bound with hydrogen-bonds formed with L251, C284, and K313 main-chains. Occasionally, the direct hydrogen-bond with C284 became temporarily water-mediated. Lastly, an interaction between K285 carbonyl oxygen and the 2’ hydroxyl group of both uracil and adenine nucleotides was observed in all simulations (Supplementary Figure A2.2 E).

We then checked if we could observe a syn conformation in solution by NMR which would be obvious with a strong H8 to H1’ NOE. In the NOESY spectra of RRM3 in complex with AUUAUU, we could detect a syn conformation (Supplementary Figure A2.2 G). Yet, when we tried to build a syn-adenine conformation in the electron density, which differed to the syn-adenine seen in MD, its proposed protein/RNA interactions were unstable in MD simulations and the adenine flipped immediately to anti or to the syn conformation as seen in the MD (Supplementary Figure A2.2 E). Therefore, an anti-adenine was placed in pocket 1 in our final crystal structure.

In binding pocket 4, depending on whether uracil or adenine was present, different protein residues would become involved in binding. The simulations showed more structurally stable binding of the adenine as it formed greater number of direct hydrogen-bonds with the

40 HuR RRM3 dimerization and AU-rich element recognition

protein than the uracil. However, water-mediated interactions around the uracil were compensating for this difference. When simulations were performed with an adenine at this pocket, direct hydrogen-bonds were formed with the side-chains of both N272 and C245 as described previously but also with the side-chains of K320 and K274 (Supplementary Figure A2.2 F, right). However, the fairly large dynamics in binding pocket 4, caused these interactions to constantly fluctuate in simulations and they would rarely all co-exist at the same time. For instance, the K320 side-chain would sometimes interact with the RNA sugar- phosphate backbone between binding pockets 3 and 4 instead of with A4 N7. The K274 and N272 side-chains would sometimes temporarily flip away, forming other protein/protein interactions or interacting with the solvent. Lastly, the entire adenine base flipped temporarily away numerous times in the course of simulations, breaking even the stacking interaction with F289. The noticeably large dynamics of the adenine bound in pocket 4 was still smaller compared to the uracil. Simulations with uracil bound at this pocket, showed only the K320 and K274 forming direct hydrogen-bond interactions, supplemented by water-mediated hydrogen-bonds to the backbone of N272. Since the K320 would sometimes also interact with the RNA sugar-phosphate backbone instead of the base, the uracil at this position was often directly recognized only by the K274 (Supplementary Figure A2.2 E, left).

In conclusion, the MD simulations showed a highly dynamic RNA recognition and hydrogen bond network in pocket 1 and 4. This is in perfect agreement with our crystal structure where a degenerate specificity is detected. Furthermore, MD simulations revealed that adenine in pocket 1 is able to bind in both anti or syn conformations and that a syn- adenine would be better stabilized by several direct interactions with the protein.

Solution studies of HuR RRM3-ARE recognition

To investigate further the (A/U)UU(A/U) recognition sequence suggested by the crystal structure, we studied the binding of RRM3 to several 5- and 6-nucleotide (nt) long U-rich RNA molecules using NMR titrations (Figure 2.3) and ITC (Table 2.1). RRM3 binding to 5’-AUUUU- 3’ or 5’-AUUAU-3’ led to different chemical shift perturbations compared to binding to 5’- UUUUU-3’ (Figure 2.3A, Supplementary Figure A2.3). K285 as well as G286 show larger 1H-15N perturbations upon binding to 5’-AUUUU-3’, whereas G288, F289, T291 resonances exhibit larger perturbations for 5’-AUUAU-3’ (Figure 2.3B,C). All these residues are located in pocket 1 or pocket 4 (Figure 2.3B), in agreement with the crystal structure and MD simulations where both pockets can accommodate an adenine. Placing the A in the middle as in 5’-UUAUU-3’ led to a similar chemical shifts pattern as 5’-AUUAU-3’, indicating the

41 HuR RRM3 dimerization and AU-rich element recognition

preference for A in pocket 4. An A at the end as in 5’-UUUUA-3’ shows the least differences compared to 5’-UUUUU-3’ (Supplementary Figure A2.3). With ITC titrations, we obtained

Kds for 5’-UUUUUU-3’, 5’-AUUUUU-3’, 5’-UUUAUU-3’ or 5’-AUUAUU-3’ of 3.58 ± 0.21 μM, 4.03 ± 0.25 μM, 5.60 ± 0.40 μM and 5.75 ± 0.27 μM, respectively (Table 2.1, Supplementary Figure A2.4 A).

Table 2.1: ITC analysis of RRM3 affinities and binding stoichiometries for various 6 nucleotide long ARE motifs. Errors indicate s.d. of at least three measurements (two for UUUAUU).

RNA N (sites) Kd (μM) ΔH (kcal/mol) ΔS (cal/mol/deg) UUUUUU 0.99±0.03 3.58 ± 0.21 -27.02 ±1.59 -65.4± 5.5 AUUUUU 0.98 ±0.03 4.03 ± 0.25 -30.67 ±3.86 -78.67± 13.3 UAUUUA 1.02 ±0.08 4.83 ± 0.17 -27.7 ±2.1 -68.50± 7.2 UUUAUU 1.02 ± 0.07 5.60 ± 0.40 -17.50 ± 0.50 -35.4 ± 1.8 AUUAUU 1.05 ± 0.06 5.75 ± 0.27 -11.07 ± 2.37 -13.3 ± 7.9

The slight increase in affinity for the first two oligonucleotides could be explained by the fact that RRM3 can bind these oligonucleotides at more registers than in the last two, as previously seen for several other RRMs (Cieniková et al., 2014; Mackereth et al., 2011). To validate this hypothesis, we recorded 1H-1H-TOCSY experiments of RRM3 in complex with various 5mer and 6mer oligonucleotides (Figure 2.3D, Supplementary Figure A2.4 B). In the 5’-UUUUU- 3’/RRM3 TOCSY spectrum five H5-H6 cross-peaks would be expected (one for each uridine) if the RRM binds a single register, but only one sharp and one broad correlation are observable (Supplementary Figure A2.4 B). Broadening or absence of a cross-peak result from chemical exchange that could be due to multiple register binding. In the 5’-AUUUU-3’/RRM3 complex, exchange is still present as only three peaks are visible whereas three sharp signals are observable for the complex with 5’-AUUAU-3’ indicating a single binding register. The same observation can be seen for the 6mer oligonucleotides (Figure 2.3D). Only 5’-AUUAUU-3’ shows the four expected signals (two being overlapped) indicative of a single register. Overall, these solution studies confirm the (A/U)UU(A/U) consensus sequence seen in the crystal structure.

42 HuR RRM3 dimerization and AU-rich element recognition

Figure 2. 3. RRM3 recognizes various motifs in solution and binds multiple registers. (A) Overlay of 1H-15N HSQC spectra of free RRM3 (back), in complex with UUUUU (blue), AUUUU (red) or AUUAU (yellow) at a 1:2.5 molar ratio of RRM3:RNA. Negative peaks corresponding to the amides of arginine side chains in the free and the RNA bound form in green. (B) Close-up view of the 1H-15N HSQC spectrum for key residues showing different chemical shift perturbations for AUUUU or AUUAU compared to UUUUU (RRM3:RNA ratios: 1:0 (blue), 1:0.25 (red), 1:0.5 (yellow), 1:1 (green), 1:1.5 (cyan), 1:2.5 (magenta). (C) Residues with chemical shift differences larger for AUUUU and AUUAU then for UUUUU are represented as sticks and highlighted in green on chain A. Pocket 1 and 4 are indicated in purple. (D) Comparison of H5–H6 correlations in 2D TOCSY spectra at 750 MHz of RRM3 in complex with various 6mer RNAs (blue: 5’-UUUUUU-3’, orange: 5’-AUUAUU-3’, green: 5’-UAUUUA-3’). (E) Close-up view of the 1H-15N HSQC spectrum for key residues showing a different exchange regime for AUUAUU compared to UUUUU or UAUUUA. RRM3:RNA ratios: 1:0 (blue), 1:0.25 (red), 1:0.5 (yellow), 1:1 (green), 1:1.5 (cyan), 1:2.5 (magenta).

Binding to the prototypical AUUUA ARE element

The 5’-UAUUUA-3’ sequence which is the prototypical element enriched in class I and II AREs does not fit the consensus for RRM3 since three uracils separate the two adenines. Yet, if one would consider binding to AUUUA, we would predict binding to two registers, AUUU and UUUA, in exchange. In such binding mode, the affinity would increase compared to a single binding mode (AUUA). In addition, the protein would be bound at a specific locus,

43 HuR RRM3 dimerization and AU-rich element recognition

compared to a long poly-U sequence, where the multiple registers would lead to a constant movement along the sequence (Figure 2.7A). We therefore investigated RRM3 binding to 5’- UAUUUA-3’ by NMR and ITC. The TOCSY revealed only two cross-peaks confirming the exchange between the two registers, AUUU and UUUA (Figure 2.3D). In addition, the Kd measured by ITC is between the poly-U and AUUA-containing oligo (Table 2.1, Supplementary Figure A2.4 A) and the NMR titration experiments reveal a different exchange regime for AUUAUU than for UUUUUU and UAUUUA for several residues (Figure 2.3E), indicating a weaker binding of AUUAUU. This unusual dynamical mode of recognition which is composed of sequence-specific binding registers in exchange could explain why the AUUUA pentamers are found in many mRNAs regulated by HuR.

The crystal structure reveals a dimerization interface in RRM3

Two potential dimerization interfaces are visible within one asymmetric unit (Figure 2.2A); one between chain A and B, mediated by several hydrogen bonds and hydrophobic interactions and a second one, mainly hydrophobic, formed between chain B and C (Figure 2.4A). Interestingly, the dimer between chain A and B is stabilized by stacking interactions of a conserved tryptophan side chain W261 (Figure 2.4A, Supplementary Figure A2.5 A). The surface area buried within this dimer is ∼347 Å calculated with PISA (Krissinel and Henrick, 2007). The involvement of the W261 residue in dimerization was reported previously by NMR and analytical ultracentrifugation studies (Scheiba et al., 2014). A model generated in this study, showed two potential orientations of the domains with respect to each other but did not provide an in-depth molecular mechanism of dimerization. Our crystal structure clearly highlights the involvement of the conserved interface in dimerization. Interactions are mainly mediated by hydrophobic residues within α1. Especially the aromatic rings of W261 are stacking between the two monomers. In addition, backbone atoms of G265/P266 and V270 of the first monomer form hydrogen bonds with V270 and G265/P266 of the second monomer respectively (Figure 2.4A). The dimer between chain B and C does not occur in solution (Figure 2.4C).

RRM3 dimerizes in solution via the conserved W261 in a concentration dependent manner

To characterize the mechanism of dimerization in detail, we performed titrations by NMR and ITC. 1H-15N HSQC spectra of RRM3 were recorded over a range of concentrations, showing chemical shift perturbations for several residues with increasing protein concentration

44 HuR RRM3 dimerization and AU-rich element recognition

Figure 2. 4. RRM3 dimerizes in solution via the conserved Trp261 in a concentration dependent manner. (A) Cartoon representation of the RRM3 homo-dimer interface between RRM chain A and B (protein backbone in grey). Amino acids involved in hydrophobic interactions are colored in green (chain A) and cyan (chain B). The conserved W261 involved in stacking and other amino acids involved in hydrogen bonds are shown as stick representation. Hydrogen bonds are represented as dashed lines. (B) Top: Overlay of close-up views of 1H-15N HSQC spectra for several residues at various RRM3 concentrations (350- blue, 113- red, 40- yellow, 20- green, 5 μM cyan). Bottom: Overlay of 1H-15N HSQC spectra sections for the same residues at various W261E concentrations (320, 131, 55, 27, 14 μM). (C) Comparison of chemical-shift perturbations of RRM3 and W261E at two concentrations, 350 vs 20 μM (red) and 320 vs 14 μM (black), respectively. The combined chemical-shift perturbations (calculated using ∆δ = [(δHN)2+δN/6.51)2]1/2). The blue asterisk indicates amino acids located within the dimerization α1 of which signals are missing in the 1H-15N HSQC of RRM3. The side chain Nε of W261 was also included in the analysis and is indicated by the arrow. The green and black horizontal lines represents the s.d. of all chemical shift differences for RRM3 and W261E, respectively. (D) Cartoon representation of the RRM3 dimer

45 HuR RRM3 dimerization and AU-rich element recognition

pair (chains A and B). Residues missing in the 1H-15N HSQC of RRM3 spectrum are highlighted in blue and residues showing ∆δ > 0.3 ppm are in red. (E) Rotational correlation times τc obtained from non-overlapping amide resonances at wildtype RRM3 concentrations of 300 μM (green) and 80uM (white) and for the W216E mutant at 311 μM (black). (F) ITC dilution profiles of wildtype (red) and W261E (black) into buffer. Measurements were repeated twice for each construct at two different concentrations; Errors indicate s.d. of two measurement. Top: ITC raw data as a function of time, bottom: heat release. Red line represents the fit, obtained with a simple dimer dissociation model provided with the ITC-adapted Origin 7 software. In order to superimpose data on the same plot, integrated enthalpy variations (heats of dilution) corresponding to the W261E variant have been centered to 1 kcal/mol.

(Figure 2.4B, Supplementary Figure A2.5 B). This observation suggests that RRM3 dimerizes in a concentration dependent manner. The chemical shift differences are mainly localized within α1 and β2, consistent with the location of the hydrophobic dimerization interface observed in the crystal structure, but changes are additionally seen in the residues of the RNP1 motif within β3 (Figure 2.4C,D), suggesting that dimerization and RNA binding could be allosterically coupled. In addition, residues 260-270 were missing in the 1H-15N HSQC spectra, possibly due to exchange broadening of the resonances caused by the equilibrium between monomer and dimer. Mutating the conserved W261 to E261 abolished the concentration-dependent chemical shift changes of RRM3 (Figure 2.4B,C, Supplementary Figure A2.5 B). A comparison of the 1H-15N HSQC spectra of wild type and the W261E mutant showed a similar pattern indicating that the mutant preserved the fold of RRM3 (Supplementary Figure A2.5 B,C). Consistent with this no significant differences in 13Cα shift deviations from random coil values were observed when comparing wild type RRM3 and W261E, indicating that the secondary structure is unchanged by the mutation (Supplementary Figure A2.5 C). Strikingly, the resonances of residues 260-270 could be detected in the W261E mutant consistent with the absence of dimer-monomer exchange.

To directly probe the oligomerization state, we carried out T1 and T2 NMR relaxation experiments at various protein concentrations and estimated the overall rotational correlation time τc of RRM3, from which the approximate MW can be deduced (Figure 2.4E,

15 Supplementary Figure A2.5 D). τc was calculated from the ratio of N longitudinal (T1) and transverse (T2) relaxation times of each residue. Indeed, at high concentrations of about 300

μM, we observed a mean τc of 13.59 ± 0.76 ns, which corresponds to a ~22.65 ± 1.27 kDa dimer (Figure 2.4E), assuming a spherical particle at 298K (Rossi et al., 2010). In contrast,

W261E has a mean τc of 7.31 ± 0.45 ns, under the same conditions, which corresponds to a monomer with a molecular weight of 12.18 ± 0.75 kDa. Interestingly, the wild type RRM3 showed a reduction of τc to 9.05 ± 0.94 ns (~15.08 ± 1.57 kDa) at 80 μM indicating that even at this concentration RRM3 is still in equilibrium between monomer and dimer. T1 and T2 NMR relaxation experiments of HuR RRM12, resulted in a mean τc of 12.80 ± 0.92 ns for residues

46 HuR RRM3 dimerization and AU-rich element recognition

20-185 (Supplementary Figure A2.5 D). Residues 2-19 are flexible and were not included in the τc determination. This correlation time, which is close to the value obtained for RRM3 at high concentration, suggests that RRM1 and RRM2 tumble as a single particle. This indicates that the RRM12 particles are mainly in monomeric state under our reducing conditions. At the beginning of our studies, we identified that ITC titrations of RRM3 into RNA showed two events, one endothermic and one exothermic, the latter event representing the RNA binding. The endothermic event was explained by the presence of dimers at high concentrations and diluting them during the titration process induces dissociation into monomers. We then performed ITC experiments to quantify the dissociation constant for dimer/monomer equilibrium of wild-type RRM3. A highly concentrated RRM3 sample was injected into buffer. Measurements revealed a dimerization constant of 31.7 μM ± 0.47 μM

(Figure 2.4F), which is two-fold lower than the Kddimer obtained by analytical ultracentrifugation (Scheiba et al., 2014). The difference can be explained by different experimental conditions, such as buffers, especially the salt dependence for dimer formation, described previously (Scheiba et al., 2014). Together, our data suggest that RRM3 indeed forms dimers in solution in a concentration dependent manner with a Kddimer of 31.7 μM ± 0.47 μM. Mutation of the conserved W261 abolishes the dimerization. Furthermore, dimerization of free RRM3 shows chemical shift changes on the RNA binding surface, suggesting the possibility of an interplay between dimerization and RNA binding.

Mutation of RRM3 W261 affects RNA binding

Previous studies reported that both the Drosophila ELAV protein and HuR dimerize through short amino-acid patterns within RRM3, which are conserved in the Hu family, including the tryptophan within the α1 (Scheiba et al., 2014; Toba and White, 2008). Drosophila ELAV-ELAV interaction requires the presence of RNA (Toba and White, 2008). HuD-HuD interaction is reduced after RNase treatment (Kasashima et al., 2002). HuR RRM3 is required for cooperative assembly of HuR oligomers on RNA (Fialcowitz-White et al., 2007). All these findings lead to the possibility that dimerization might affect RNA binding affinity.

The c-fos ARE 11mer sequence, which was used for crystallization, is significantly longer than the usual RNA length bound by one RRM (2-8 nucleotides per RRM (Auweter et al., 2006)). In addition, our ITC experiments showed a 1:2 RNA:protein stoichiometry (Figure 2.1D).Therefore we investigated if the c-fos ARE 11mer could be bound by a dimer or two separated RRMs. Looking at our crystal structure, the distance between the U5 of chain

47 HuR RRM3 dimerization and AU-rich element recognition

E and the U2 of chain F is of 42 Å. This is too long to accommodate the three nucleotide spacer between to two chains if one RRM would bind the first 5 nucleotides and the second one the last 3 nucleotides of the c-fos ARE (with an extended single-stranded RNA, the distance between two phosphate atoms is around 7 Å). However, due to the crystal packing, the orientation of the monomers within the crystal structure might slightly differ from a dimer in solution. In addition, a dimer in solution might be more dynamic and flexible to rearrange upon RNA binding. To test whether RNA binding has an effect on dimerization of RRM3, we performed ITC experiments with W261E mutant and the 11mer c-fos ARE motif. Interestingly, the affinity is decreased 2-fold for the mutant compared to the RRM3 WT (Kd of 1.4 μM compared to 0.65 μM for WT) (Figure 2.5A, right). ITC experiments with the shorter 5’- UUUUUU-3’ motifs on the other hand showed no change in affinity between the two proteins (Figure 2.5A, left). This demonstrates that the ability to dimerize increases the RNA binding affinity. We performed further ITC experiments with the two proteins and DNA analogs of increasing size (from 6 to 21 nt). Similar to U6, T6 showed no difference in Kd for RRM3 or W261E (Table 2.2). With a length equivalent to c-fos (T11), RRM3 and W261E show no difference in Kd. This indicates that the sequence and not only the nucleotide length is a factor affecting dimerization. Only at a length of 12 nucleotides or above, an increase in affinity can be detected for the RRM3 WT compared to the W261E mutant RRM. This coincides as well with the N-value decreasing to 0.5 (indicating a 1:2 stoichiometry RNA:protein) for the WT, while for the mutant the N-value decreases only to 0.68 when binding T12 (Table 2.2, Figure 2.5C). For T14 or for T6C10T6, the affinity of the RRM3 WT is almost 3 times higher than for the mutant. The Kd for this last DNA sequence of 0.26 μM approaches the value measured for RRM12 ans c-fos (Figure 2.1C) suggesting that an RRM3 dimer and RRM12 might have a comparable affinity for AREs. NMR titrations confirmed qualitatively the data obtained by ITC. Binding of c-fos by the WT RRM3 showed a fast to intermediate exchange regime while binding to short oligos for both proteins or for the mutant W261E to c-fos shows mostly fast exchange regime (Figure 2.5B). Overall, we show evidence that the ability for RRM3 to dimerize increases RNA binding affinity for RNAs above a length of 11 to 12 nucleotides. At this lenght, the affinity increase could originate from an increase in local concentration when one of the monomers is bound to the RNA. The dimer might not persist since the RNA is too short to span the two RRM binding surfaces. However, when the RNA is longer (over 14nt), the dimerization can be maintained and the two RRM surfaces be bound by the same RNA (Figure 2.7B). The apparent gain of affinity is rather small (3-fold maximum) which is in agreement with the high

RRM3 dimerization Kd of 31.7 μM, indicating a weak affinity to dimerize.

48 HuR RRM3 dimerization and AU-rich element recognition

Figure 2. 5. Dimerization and RNA binding of RRM3 directly affect each other. (A) Comparison of ITC profiles of RRM3 or W261E titrated with UUUUU or c-fos. Measurements were repeated at least two times. Errors indicate s.d. of at least two measurements. Top: ITC raw data as a function of time, bottom: heat release. Red line represents the fit, obtained by the Origin 7 Software with a one binding site model. (B) Close-up view of the 1H-15N HSQC spectrum for multiple residues in the RNA binding interface and the comparison of the exchange regime in binding to AUUUU and c-fos of RRM3 vs W261E. Protein:RNA ratios: 1:0 (blue), 1:0.25 (red), 1:0.5 (yellow), 1:1

(green), 1:1.5 (cyan), 1:2.5 (magenta). (C) Ka in μM (thick line, right axis) and n-value (thin line, left axis) measured by isothermal titration calorimetry (ITC) for RRM3 (red) and W261E (black), Errors indicate s.d. of two or three measurements.

49 HuR RRM3 dimerization and AU-rich element recognition

Table 2.2. ITC analysis of RRM3 and W261E for various T-rich DNA motifs. All data were collected at 298K with 35 times 8 μl injections of DNA in 20 mM Na2HPO4 pH 7, 100 mM NaCl and 20mM β-mercaptoethanol into ~10 μM RRM3 in the same buffer. Origin 7 software was used to fit the binding isotherms. Errors indicate s.d. of two measurements

Dimerization and RNA binding of RRM3 regulate ARE containing reporter in a destabilizing manner in living cells

The high dimerization Kd of 31.8 μM ± 0.2 μM and the weaker affinity of RRM3 for RNA compared to RRM12 raised the question of whether these interactions are functionally important in vivo. To address this, we compared the efficiency of the wild type HuR and HuR mutants defective in either RRM3 dimerization, RRM3 RNA binding or both in expression regulation of an ARE containing reporter mRNA (Figure 2.6). Wild type or mutant Flag-HuR expression constructs as well as a dual luciferase reporter bearing the first 60 nucleotides of the COX-2 3’UTR in the 3’UTR of the Renilla CDS (Figure 2.6A,B). were co-transfected in Huh7 cells or in C3H/10T1/2 cells (multipotent cells isolated from C3H mouse embryo). Renilla luciferase measurements were normalized to the firefly luciferase signal within each sample. We observed a significant increase in luciferase activity when transiently overexpressing HuR (Figure 2.6C-E) compared to the empty control vector, similarly to previous reports for different ARE containing targets (eg.(Fan and Steitz, 1998b)). The dimerization inactive mutant (W261E) caused a statistically significant increase in luciferase activity in Huh7 and C3H/10T1/2 cells (Figure 2.6C,E). This is surprising since one would have rather expected

50 HuR RRM3 dimerization and AU-rich element recognition

Figure 2. 6. Dimerization and RNA binding of HuR RRM3 negatively contribute to ARE reporter regulation in cells. (A) Schematic description of the pSICHECK2 dual luciferase reporter constructs, containing the first 60 nucleotides of the COX-2 3’UTR used in this study, fused to Renilla luciferase (RL). Firefly luciferase (FL) is the internal control for transfection efficiency. (B) Schematic description of HuR and the position of the W261E, FY and RR mutations. (C) Dual luciferase reporter assay for evaluating the effect of HuR mutants in upregulation of the Cox-2 ARE reporter. HuR and mutants were co-expressed with the luciferase reporter containing Cox2-1-60 in Huh7 cells. (D) Dual luciferase reporter assay for evaluating the effect of HuR mutants on different reporter constructs. HuR and mutants were co-expressed with either the empty luciferase reporter (psiCHECK-2), luciferase reporter harboring Cox2-1-60 (blue) or Cox2 1-60 AtoUmut (green) in Huh7 cells. (E) Dual luciferase reporter assay for evaluating the effect of HuR mutants in Cox-2 ARE reporter regulation. HuR and mutants were co-expressed with the luciferase reporter containing Cox2-1-60 and a Scr-RNA in C3H/10T1/2 cells, multipotent cells isolated from C3H mouse embryo cells.

The RL/FL luminescence was normalized to the value of mock transfection. The mean values ± sd from at least three independent experiments are shown. P values were determined by the Student’s t test (Two-Sample Assuming Equal Variances). *P<0.05, **P<0.01, ***P<0.001 that inhibition of HuR dimerization would lower HuR’s affinity for the ARE motif and therefore have reduced ability to stabilize COX-2 ARE reporter mRNA. The increase in luciferase activity was further seen both, in presence of endogenous HuR (Figure 2.6E) as well as after siRNA knock down of endogenous HuR before the overexpression of FLAG-tagged HuR in C3H/10T1/2 cells (Supplementary Figure A2.6 F). Interestingly, the luciferase activity in Huh7 cells was also elevated when overexpressing HuR with the RRM3 containing RNP

51 HuR RRM3 dimerization and AU-rich element recognition

mutations F247A-Y249A (abbreviated with FY; Figure 2.6C). RRM3 FY mutant is still folded but defective in RRM3-RNA binding (Supplementary Figure A2.6 A).The triple HuR W261E FY mutant, deficient in both dimerization and RNA binding, showed no changes in Renilla upregulation compared to the RRM3-RNA-binding inactive HuR FY mutant (Figure 2.6C). This indicates that RRM3 dimerization depends on RNA binding. If RRM3 cannot bind to the RNA, dimerization plays a minimal role or cannot even take place. The effect of HuR FY seems to be cell line dependent, as the same experiments in C3H/10T1/2 cells showed no effect of HuR FY (Figure 2.6E). To test if this increase in luciferase activity was RRM12 dependent, we also introduced mutations in RRM12 of HuR to reduce the RNA binding affinity. Two mutations, R97A and R136A (abbreviated with RR, Figure 2.6B), which were shown in a previous study to decrease RNA binding without impacting the stability of the tandem RRM12 (H. Wang et al., 2013), were additionally introduced in our HuR FY double or HuR FY W261E triple mutant. All HuR RR mutant showed as expected a decrease in luciferase activity compared to HuR wildtype (Figure 2.6C, Supplementary Figure A2.6 C), which is explained by the reduced affinity of the RR mutants to RNA. The elevated luciferase activity of the dimerization inactive HuR W261E and the HuR FY which is incapable of RNA binding, was abolished when including the RR mutations (Figure 2.6C). This indicates that effect seen for RRM3 dimerization and RNA binding depends on RNA binding by RRM12 of HuR. HuR knock down as well as expression levels of all mutants were characterized by western blotting (Supplementary Figure A2.6 D,E,G). These data suggest that the mRNA stabilization induced by HuR RRM12 RNA binding can be counteracted by RRM3 RNA binding and dimerization. These results could be rationalized by a competition between RRM3 and HuR RRM12 or other AU-rich binding proteins for binding to the AUUUA repeats in the COX-2 ARE sequence (Figure 2.7C,D). The destabilizing effect mediated by RRM3 is not observed anymore when using a luciferase reporter where all adenines of the COX-2 3’UTR are mutated to uridines in agreement with a sequence-specific recognition of AUUUA (Figure 2.6D). An empty luciferase reporter without the ARE sequence, similar to the COX 2 3’UTR mutant reporter, showed a less than 1.5 fold increase in luciferase activity upon HuR overexpression compared to the luciferase reporter with the COX-2 ARE (2.5 fold). RRM3 mutants show also no effect on such empty luciferase reporter (Figure 2.6D). Overall, these data suggest that both the dimerization and the AUUUA sequence- specific binding of RRM3 do not positively contribute to the activity of HuR in ARE upregulation. Rather, RRM3 might auto regulate the protein function by limiting productive complex formation with mRNA harboring AUUUA motifs in their ARE while not affecting those containing only U-rich sequences.

52 HuR RRM3 dimerization and AU-rich element recognition

Figure 2. 7. Model for AUUUA recognition, RNA binding induced dimerization and the destabilizing function observed in Huh7 cells. (A) RRM3 binds multiple registers on U-rich regions which leads to a dynamic movement along RNA sequences. Binding to AUUUA, which has two registers, results in localization of RRM3. (B) Binding to long RNAs, comprising two motifs separated by a long linker, induces dimerization. This leads to an increase in affinity. (C) RNA binding and dimerization of RRM3 functions in a destabilizing manner. HuR RRM3 may directly compete with HuR RRM12 or other RBPs for overlapping cognate AU-rich RNA motifs. (D) Other proteins could directly interact with the RRM3 dimer and to counteract mRNA stabilization or induce ARE mRNA down-regulation. On the other hand, a RRM3 monomer might interact with proteins playing a role in translation.

53 HuR RRM3 dimerization and AU-rich element recognition

2.4 Discussion

ARE recognition by HuR C-terminal RRM

Despite controversies on whether RRM3 might be involved in direct binding of AREs and mRNA regulation (Chen et al., 2002; Fan and Steitz, 1998b), we show that HuR RRM3 binds a c-fos ARE 11mer RNA with a Kd in the low μM range in vitro (Figure 2.1D). This demonstrates that all three HuR RRM domains are able to bind AREs, indicating that RRM3 might have a potential contribution in direct interaction with mRNA targets. RRM3 is also able to bind poly-A but with a much lower affinity (Supplementary Figure A2.1 B). This does not support a primary role of RRM3 in binding to poly-A tails of mRNAs (Abe et al., 1996; Gao and

Keene, 1996; Ma et al., 1997). Binding studies with other ligands, i.e. EDTA, MgCl2 or ATP showed no significant chemical shift perturbations (Supplementary Figure A2.1 A), indicating no binding of these ligands to RRM3 under our experimental conditions. However, we cannot exclude that in the context of the full-length protein as well as in presence of cellular co-factors a physiologically relevant direct interaction of RRM3 with the poly-A tail or other factors might be possible.

While structural investigation of the full-length HuR or its relatives HuB, HuC and HuD remains challenging due to their poor stability and solubility, a study of the separate RRMs enables us to obtain functional insights. The crystal structures of the free HuR RRM12 and HuR RRM12 in complex with an 11 nucleotide ARE from c-fos reveal a conformational change upon RNA binding, a preference for pyrimidine-rich sequences and a low degree of sequence specificity (H. Wang et al., 2013). HuR RRM3 was previously reported to bind poly-U (Scheiba et al., 2014). In addition, cDNA arrays and PAR-CLIP experiments identified U-rich rather than AU-rich motif for HuR binding in thousands of targets (Lebedeva et al., 2011; López de Silanes et al., 2004). While building our model of RRM3 in complex with c-fos ARE and taken into account this information, we realized that placing uracils in some electron density pockets proved challenging and we soon realized that pockets one and four showed occupancy for both adenine and uracils. Combining NMR, ITC and MD simulations, we could validate that RRM3 can recognize sequence-specifically both nucleotides in these pockets (Figure 2.3, Supplementary Figure A2.3). RRM3 can therefore recognize multiple sequences, UUUUU, AUUUU, UUUAU, AUUUA or AUUAU with comparable affinities but experiences a slight increase in affinity for U-rich sequences due to multiple register binding or avidity (Table 2.1, Figure 2.3). A similar lack of sequence specificity was reported for RRM12 of HuR and HuD (H. Wang et al., 2013; Wang and Tanaka Hall, 2001). This could enable Hu proteins to bind motifs from all three classes of AREs and explains how this versatile protein family can regulate thousands of targets found in pre-mRNAs, mature mRNAs, miRNAs and long

54 HuR RRM3 dimerization and AU-rich element recognition

noncoding RNAs (lncRNAs) (Lebedeva et al., 2011; Legnini et al., 2014; López de Silanes et al., 2004; Mukherjee et al., 2016, 2011; Young et al., 2012). Another related protein family, CELF (CUGBP, ELAV-Like family), is characterized by the same domain architecture, two N- terminal tandem RRMs, a long flexible region followed by the C-terminal RRM domain. Like the Hu family, CELF proteins also show a degenerated consensus sequence (Dasgupta and Ladd, 2012).

When encountering poly-U sequences, RRM3 binding might be highly dynamic with a constant binding and unbinding, likely sliding along U-rich sequences within long 3’UTRs. We found that the possibility to accommodate adenines in pocket one and four reduces or even prevents multiple register binding when bound to AUUA for example. This results in RRM3 being locked and localized to a precise position within the 3’UTRs. Yet, loss of multiple registers leads also to a decrease in affinity, making the RRM3 more prone to be displaced by other AU-rich binding proteins. When binding the AUUUA pentamer motifs, which is enriched in class I and II AREs, we predict two binding registers in exchange, AUUU and UUUA, on the basis of our structure. Our data confirmed the register change and the increase in affinity compared to AUUAU. In binding AUUUA motifs, RRM3 localizes at these motifs and still binds with substantial affinity due to the two-register binding. Thus, we provide an original hypothesis on why AUUUA is a prototypical motif in ARE and how RRM3 recognizes it (Figure 2.7A). This could be a tuning mechanism for HuR to find and bind its target sequences on the 3’UTRs. This is reminiscent of what we previously observed for the polyU binding protein hnRNP C in terms of structure and affinity enhancement (Cieniková et al., 2014).

Effect of HuR homodimerization on RNA binding

In addition to new insights into HuR RRM3-ARE recognition our structure provides an additional example of the various strategies by RRM-containing proteins to regulate gene expression. HuR RRM3 is a multi-functional domain containing an RNA binding platform but also a protein interaction surface to further tune its functions. Concentration dependent dimerization of RRM3 is mediated primarily by W261 within α1 which is conserved in all Hu proteins, suggesting that an analogous dimerization mechanism via RRM3 is likely to exist for

HuB, HuC and HuD. Considering the high dimerization Kd of 30 μM measured for RRM3 in vitro (Figure 2.4F), we questioned whether RRM3 of HuR would be able to dimerize under physiological concentrations. It should be considered though that in our in vitro studies characterization of the dimerization was performed in the context of the isolated RRM3 domain, while the full-length HuR dimerization affinity might be different due to the additional hinge region. Furthermore, we could show that the capacity of RRM3 to dimerize increases the RNA binding affinity to oligonucleotide sequences above 12 nucleotides in length. In turn,

55 HuR RRM3 dimerization and AU-rich element recognition

RNA binding also affects HuR dimerization. ARE elements are mostly 30-100 nucleotides long and typically in the context of 3’UTRs harboring additional HuR binding sites scattered across the sequence. Upon oligomerization of HuR along the 3’UTR, as reported before (Fialcowitz- White et al., 2007), the local HuR concentration is increased, which would certainly shift the equilibrium towards dimerization. It has been shown that during chicken neurogenesis, HuR exhibited a higher expression in the ventricular zone of the spinal cord to maintain proliferation of neuronal precursor cells (Wakamatsu and Weston, 1997). After proliferation, HuR levels are reduced during differentiation but raise again during maturation of neurons. With such changes in concentration, RRM3-mediated dimerization could affect the association and dissociation of HuRs and influence the regulation of the different processes to fine-tune regulatory activity. Under certain stress signals, HuR is highly accumulated in stress granules (Bhattacharyya et al., 2006; Gallouzi et al., 2001). In cancerous cells or tumors HuR concentration is also elevated. Given the micro molar affinity of RRM3 dimers, such concentration changes could become relevant to directly affect RRM3-mediated HuR dimerization and thereby affect its function in ARE-based gene regulation.

Many RRM-containing proteins contain multiple RRMs to fine-tune their specificity and affinity. Tandem RRMs can interact with each other, sometimes involving their interdomain linkers, to create either extended RNA binding surfaces or deep clefts for interaction with RNA, as shown for Sex-lethal, HuD and HuR RRM12 (Handa et al., 1999; H. Wang et al., 2013; Wang and Tanaka Hall, 2001). Other RRMs interacts through their helices at the opposite of the RNA binding surface, leading to conformations inducing looping of bound RNAs. This is exemplified by the RRM34 of PTB, RRM12 of hnRNPA1 or hnRNPL RRM34 (Barraud and Allain, 2013; Beusch et al., 2017; Oberstrass et al., 2005; Vitali et al., 2006; Zhang et al., 2013). Interestingly, while PTB interacts with its RRM4 α2 and RRM3 α1, hnRNPA1 RRM1 uses its β4 strands to bind α2 of RRM2. Homodimerization via α1 as in HuR RRM3 results also in the β-sheets surfaces being in opposite directions. Dimer formation could therefore induce RNA looping and bring together distant motifs within one mRNA target, like in PTB or it could bring close together different mRNAs, which could induce liquid-liquid phase transition and formation of cellular granules (Protter and Parker, 2016). In fact, members of the RNA binding protein with multiple splicing (RBPMS) family are also able to form homodimers, via residues in the α1 and adjacent loop regions (Sagnol et al., 2014). Yet, the dimer interface of the RBPMS family is formed by multiple salt bridges, additional hydrophobic contacts and hydrogen bonds, resulting in a stable and permanent dimer (Sagnol et al., 2014; Soufari and Mackereth, 2016; Teplova et al., 2016), whereas HuR RRM3 exhibits a concentration dependent equilibrium between monomer and dimer. Interaction studies of RBPMS family members with an RNA containing two binding sites and

56 HuR RRM3 dimerization and AU-rich element recognition

a sufficiently long linker to overcome the distance, lead to a two- to nine-fold increase in affinity compared to a short RNA containing a single binding site (Soufari and Mackereth, 2016). This is in good agreement with our measurements under conditions where the HuR RRM3 is not quantitatively in the dimer state (Figure 2.5, Table 2.2). In addition, dimerization mutations of RBPMS result in 2.5-4 fold lower binding affinities to RNA comprising two binding sites (Teplova et al., 2016), similarly to what we measured for HuR RRM3 for binding c-fos (Figure 2.5) or other long DNA sequences (Table 2.2).

HuR RRM3 RNA recognition and homo-dimerization affects ARE stabilization

We could show quite unexpectedly that in Huh7 cells, overexpression of HuR mutants which are inactive in RRM3 dimerization or RNA binding, stabilize a reporter construct harboring multiple AUUUA motifs better than the HuR WT (Figure 2.6). These findings may suggest that the ARE binding capability of HuR RRM3 might exert an auto-inhibitory role by competing with HuR RRM12 for ARE binding and thereby limiting its role in upregulation of the COX-2 ARE reporter. The COX-2 sequence used contains seven AUUUA motifs, which are also known targets of HuR RRM12 and mutating all adenines abolished the positive effect of the RRM3 mutants (Figure 2.6D). It is thus conceivable that mRNAs containing different classes of AREs could be differently fine-tuned by such an auto-inhibitory function of RRM3. Alternatively, the negative effect of RRM3 could be due to protein-protein interactions mediated by this domain via interaction with other proteins, possibly in dependence of ARE binding and/or dimerization. Interestingly, the HuR RRM3/ KH-type splicing regulatory protein (KSRP) complex destabilizes nucleophosmin (NPM) mRNA to regulate muscle fiber formation (Cammas et al., 2014). Depletion of HuR by siRNAs resulted in elevation of NPM mRNA. During muscle cell differentiation HuR is cleaved into two products, HuR-CP1 (24 kDa) and HuR-CP2 (8 kDa), this latter containing the C-terminal RRM3. HuR-CP2 associates with KSRP and recruits two ribonucleases to induce destabilization of the NPM mRNA. The destabilizing effect observed in our luciferase assays, could also act via a similar mechanism, the recruitment of ribonucleases via KSRP-RRM3-dimer binding (Figure 2.7D). In addition, RBPMS2, a RBPMS paralog which forms a stable dimer, is able to interact with the translation elongation factor eEF2. This interaction is abolished in the dimerization inactive mutant (Sagnol et al., 2014). Similarly, RRM3 monomers could also influence binding to other proteins which play a role in translation (Figure 2.7D). Future studies are needed to address whether dimerization and RNA binding dependent RRM3 interactions with other proteins plays a role in HuR ARE interaction to up or down-regulate gene expression.

57 HuR RRM3 dimerization and AU-rich element recognition

Since elevated HuR levels are associated with several chronic and deadly diseases, understanding its molecular mechanism is of high importance to develop new therapeutic strategies to target and downregulate it. Our model of an auto-inhibitory role of HuR RRM3 opens new perspectives on how to control HuR function by therapeutic agents.

2.5 Materials and Methods

Cloning and site-directed mutagenesis

Full-length HuR (UniProtKB Q15717) in the expression vectors pTXB1 (IMPACT -CN system, New England Biolabs) and the first two RRM domains of HuR (2–189) in a modified Novagen pET28a vector were described before (Benoit et al., 2010; Meisner et al., 2004). The sequence containing the C-terminal RRM domain (residue 241-326, RRM3) was amplified by PCR from the pTXB1-HuR plasmid introducing NcoI and EcoRI restriction sites and subcloned into the pETGB-1a vector (EMBL, kind gift from Drs Arie Geerlof and Gunter Stier). To generate mammalian expression constructs, full-length HuR was sub cloned between the XhoI and HindIII cleavage sites into pcDNA3.1+ (Invitrogen) with an inserted Flag-tag between NheI and HindIII. Mutations were introduced into pETGB-1a-HuR, pETGB-1a-RRM3, pcDNA3.1+Flag-HuR according to the QuickChange Site-Directed Mutagenesis protocol from Stratagene. To generate an ARE containing reporter construct, the first 60 nucleotides of the murine COX-2 3’UTR were amplified by PCR using pGL3-COX2 (a gift from Prof. Aubrey R. Morrison, Washington University School of Medicine) and 5'- CTAGGCGACTCGAGGATCGCCGTGTAATTCTA-3' and 5'- TGGCCGGCGGCCGCTATCATGTCTGCTCGAAG-3' forward and reverse primers, respectively to introduce XhoI and NotI sites. PCR products were ligated into the dual reporter plasmid psiCHECK-2 (Promega). The COX-2 AtoU mutant, where all adenines are mutated to uracils, was synthesed by General Biosystems, Inc.

Expression and Purification of recombinant proteins

Recombinant human 6-His-tagged –GB1-fusion proteins (RRM3, RRM3 W261E, RRM3 FY) were over expressed in BL21(DE3) cells (Novagen) and purified by immobilized metal affinity chromatography. His-tag was removed by TEV digestion. The protein was further purified by an additional mobilized metal affinity chromatography to remove the 6-his-GB1-tag, followed

58 HuR RRM3 dimerization and AU-rich element recognition

by size exclusion chromatography. A detailed expression and purification protocol can be found in the Materials and Methods section within the Appendix.

RNA and DNA synthesis

The c-fos-ARE 11mer (5’-AUUUUUAUUUU-3’) was provided by Jürg Hunziger (Novartis, Basel) and synthesized as described in (Masliah et al., 2018).

Short single-stranded RNAs (5’-UUUUU-, -AUUUU-, -UUAUU-, -AUUAU-, -UUUUA-, - UUUUUU-, -AUUUUU-, -UUUAUU-, -AUUAUU-3’) were purchased from Thermo Scientific and deprotected as described in the manufacturer’s instructions, lyophilized and dissolved in NMR or ITC buffer.

Single-stranded DNAs were purchased from Mycrosynth and dissolved in ITC buffer.

Sample preparation and Crystallization

The protein was concentrated to 7 mg/ml in 20 mM Tris (pH8), 100 mM NaCl, 10% (w/v) Glycerol, 1mM TCEP. RRM3:RNA complex was prepared by adding 1.2 molar excess c-fos- ARE RNA. Crystals were obtained by sitting drop vapor diffusion at 298K, in which 200 nl protein:RNA complex was mixed with 200 nl precipitant and suspended over 80 μl precipitant. Crystals were obtained in the 2 M Ammonium sulfate, 0.1 M Bis-Tris well (IndexHT screen (Hampton research)), and flash frozen in liquid nitrogen (without cryoprotection).

Data Collection and Structure Determination

The crystals were measured under a N2 cryo stream at the PXII-X10SA beamline (equipped with the PILATUS detector) of the SLS synchrotron in Villigen. The wavelength of data collection was 1 Å. Data were indexed, integrated and scaled using XDS (Kabsch, 1993). The structure was solved by molecular replacement using a modified version of the HuR RRM1 structure (PDB code 3HI9) in PHASER (McCoy et al., 2007). The structure was refined using PHENIX (Adams et al., 2002). In the final model, all residues were in the favored regions (99%) of the Ramachandran plot with 0.3% outliers, generated by PHENIX. The value of 0.3% for outliers is due to the same amino acid in all four chains being outside the preferred or allowed region, leading to an actual outlier value of less than 0.1%. COOT (Emsley and Cowtan, 2004) was used for manual inspection and model building. RNA was built in manually. See Materials and Methods section within the Appendix for the detailed structure determination.

59 HuR RRM3 dimerization and AU-rich element recognition

Electron density map was calculated with Phenix FFT map coefficient. Figures of the sigma 2Fo − Fc electron density maps were contoured at 1.2 sigma level (in Pymol). Figures of the structures were prepared using the program PyMol version 1.4.1 (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC).

NMR spectroscopy

All NMR spectroscopy measurements were done in 20 mM Na2HPO4 (pH7), 100 mM

NaCl,1 mM DTT, 10% D2O at 298K using Buker AVIII-500 MHz, AVIII-600 MHz and AVIII-700 MHz (equipped with cryoprobes) and AVIII-750 MHz.

For the 1H, 15N and 13C assignments of the protein backbone of RRM12 2D 1H,15N-HSQC and 3D HNCO, HN(CA)CO, HNCA, HN(CO)CA, HNCACB, CBCA(CO)NH and 3D 15N-NOESY- HSQC (mixing time of 100 ms) spectra were recorded. Chemical shift assignments of RRM12 are in Supplementary Table A2.2. 2D 1H,15N-HSQC and 3D HNCA, HNCACB, CBCA(CO)NH and 3D 15N-NOESY-HSQC (mixing time of 100 ms) were recorded for RRM3 and 2D 1H,15N- HSQC, 3D HNCA, HNCACB for W261E. Chemical shift assignments of RRM3 and W261E are in Supplementary Table A2.3.

For RNA titrations, various RNA sequences were titrated into 15N labelled RRM3 at about 100 μM. The combined chemical shift difference was calculated according to (δH2+(δN/6.51)2)1/2, were δH and δN were the difference in the 1H and 15N chemical shifts respectively.

1 1 2D H- H-TOCSY experiments (mixing time of 50 ms in H2O and 60 ms in D2O) of protein RNA complexes were recorded to observe the H5-H6 cross-peaks. 2D NOESY was measured of a

15 1:1 N RRM3: AUUAUU in D2O (mixing time of 150 ms) to determine the syn or anti confirmation of the adenine in pocket 1.

NMR data were processed with Topspin 3.1 (Bruker) and the analysis performed using Sparky (T. D. Goddard and D. G. Kneller, SPARKY 3, University of California San Francisco)

T1 and T2 relaxation experiments were carried out on the AVIII-750 MHz at 298 K at various concentrations of 15N-labelled or 13C-15N-labelled RRM3, 0.3 mM 15N-labelled W261 and

15 0.4 mM N-labelled RRM12. Longitudinal relaxation time T1 were determined from a pseudo- 3D experiment with twelve delay times ranging from 12-3000 ms (Bruker Pulse sequence

1 15 hsqct1etf3gpsi3d.2). Transverse relaxation times T2 were obtained from H- N correlation spectra with ten different delay times (17-271 ms) (Bruker Pulse sequence hsqct2etf3gpsi3d).

15 15 N T1, N T2 and the corresponding standard deviation (SD) for each residue were determined by fitting the peak heights from each spectrum to a decaying exponential using

60 HuR RRM3 dimerization and AU-rich element recognition

Sparky (rh command) (Lee et al., 2015). Correlation times (τc) for each residue were calculated from the T1/T2 ratio (Gryk et al., 1998; Rossi et al., 2010). Molecular weights were roughly estimated by τc/0.6, assuming a spherical particle at 298K (Rossi et al., 2010).

Camcoil was used for Cα random coil chemical shifts calculations (De Simone et al., 2009).

Isothermal titration calorimetry

Protein and RNA/DNA oligonucleotides were dialyzed or dissolved respectively in 20 mM

Na2HPO4 (pH7), 100 mM NaCl and 10 mM β-mercaptoethanol. ITC measurements were performed on a VP-ITC instrument (Microcal) at 25°C. To estimate the affinities for various DNA and RNA motifs, DNA/RNA was titrated into protein solution by 35 injections of 8 μl every 300s (at 307 rpm). At least two replicates were measured. To estimate the dimerization constant, 8 μl of 35 sequential injections were used with the concentrated protein diluted into buffer. Raw titration data were integrated and analyzed according a one binding site model provided in the Origin7.0 software. We used the simple dimer dissociation model to determine

Kddimer and ΔHdimer. Reported errors correspond to the standard deviation of at least two replicates.

Luciferase Assay and western blot

Huh7 hepatoma cells (a generous gift from Prof. Dr. Wilhelm Krek, ETH Zürich) were cultured in Dulbbeccos’s modified Eagle’s medium (DMEM)+GlutaMAX (ThermoFischer Scientific) supplemented with 10% fetal bovine serum, 1% penicillin/ streptomycin at 37°C and 5% CO2. All plasmids were transiently transfected with Lipofectamine 2000 (Invitrogen) as described by the manufacturer’s instructions. 1 μl: 1 μg Lipofectamine: DNA ratio was used. 0.1 μg psiCHECK-2 -COX2-1-60 was co-transfected with 1 μg pcDNA or pcDNA-HuR/HuR mutants per well in a 24 well plate, each well seeded with 0.5*105 cells the night before. Cells were harvested after 48h by adding passive lysis buffer (Promega). Lysate was cleared by centrifugation at 13,000 rpm for 10 min at 4˚C. Renilla and firefly activities were measured using the Dual-Luciferase Assay Kit (Promega) as described in the manufacturer’s instructions. Renilla measurements were normalized to the firefly luciferase signal within each sample. Protein concentration of the same lysates was determined using BCA assay. Western blot was performed as described in Materials and Methods section within the Appendix. In addition, luciferase assay was performed in C3H/10T1/2 cells, with and without silencing of endogenous HuR. For more details, see Materials and Methods section within the Appendix.

61 HuR RRM3 dimerization and AU-rich element recognition

Molecular Dynamics Simulations

The X-ray structure of the HuR RRM3 in complex with c-fos-ARE was used as the starting structure of all molecular dynamics (MD) simulations. Each simulated system contained only single, monomeric copy of the complex. Different chains from the X-ray structure were utilized to obtain starting structures where either uracil or adenine were positioned in binding pockets 1 and 4, respectively. Thus the simulated systems included HuR RRM3 complexed with 5’- UUUA-3’ (chain C/G), 5’-AUUUU-3’ (chain A/E), or 5’-UUUUU-3’ (obtained by superimposing chains A/E and C/G). The underlined nucleotides are those located in binding pockets 1 and 4. We conducted multiple simulations of each system and all simulations were run for the

minimal length of one microsecond. We have used ff12SB (Maier et al., 2015) and bsc0χOL3 (Pérez et al., 2007; Zgarbová et al., 2011) force fields to describe protein and RNA, respectively. The simulations were performed in explicit-solvent conditions, (Berendsen et al., 1987; Joung and Cheatham, 2008) using the AMBER16. Standard equilibration and simulation protocol for protein/RNA complexes (Krepl et al., 2015) was used and the HBfix potential function (Kührová et al., 2016) was applied in all simulations to improve structural stability of the protein/RNA interface. See Materials and Methods section within the Appendix for more details.

Data availability

The structure has been deposited in the Protein Data bank (http://www.rcsb.org) under the accession number 6GC5.

2.6 Acknowledgements

The authors thank Stier G. and Geerlof A. (EMBL) for providing the petGB1-1a plasmid, Prof. Dr. Wilhelm Krek (ETH Zürich) for the Huh7 hepatoma cells, Prof. Aubrey R. Morrison (Washington University School of Medicine) for the pGL3-COX2 plasmid and Dr. Jürg Hunziger (Novartis) for the c-fos-ARE RNA. In addition, we thank the MX-group at Swiss Light Source Villigen for their excellent support in data collection and F.F Damberger (ETH Zürich) for help with the setup and analysis of the NMR experiments and proof reading the manuscript. This work was supported by the EU FP7 ITN project RNPnet (contract number 289007) and NCCR RNA and Disease. M.K. and J.S. acknowledges support by the Czech Science Foundation [grant number P305/12/G034] and by the project LO1305 of the Ministry of Education, Youth and Sports of the Czech Republic under the National Sustainability Programme I

3. Structural basis of the PABPC1 RRM1-BTG2 interaction to recruit CAF1 deadenylase

Contributions:

Nina Ripin, Fabienne Mauxion (Institut de Génétique et de Biologie Moléculaire et Cellulaire, 67404 Illkirch, France) and Frédéric H-T Allain, designed the project. F.M. cloned the GB1- tagged constructs and N.R. purified all proteins, performed and analyzed NMR, ITC and SEC experiments, set up crystal screens, performed and analyzed the HADDOCK modeling.

PABPC1 RRM1-BTG2 interaction

3.1 Abstract

Deadenylation is a key step of mRNA decay in eukaryotic cells. BTG/Tob protein family members, which play a central role in cell cycle regulation, differentiation and cancer, recruit CAF1 deadenylase to mRNA poly(A) tails by interacting PABPC1 and cause poly(A) tail shortening. Compared to Tob1 and 2, which contain a PABP-interacting (PAM) motif at their C-termini, other BTG factors lack this motif. The interaction between BTG2 and PABPC1 is mediated by the highly conserved APRO domain of BTG2 and the two RNA recognition motifs of PABPC1 (RRM12). Moreover, the BTG2 APRO domain – PABPC1 RRM1 interaction was able to activate CAF1 deadenylase in vitro without the presence of other CCR4–NOT complex subunits and to reduce proliferation of U2OS cells. Here, our NMR chemical shift mapping experiments reveal that the boxC region of BTG2 APRO domain recognizes the first helix, second β-stand and the following loop of PABPC1 RRM1. Interestingly, BTG2 seems not to interfere with poly(A) binding, suggesting a stable ternary complex formation comprised of BTG2-RRM1-poly(A). In combination with our modeling approach, our data provide the first structural information about how BTG2 recognizes and directly binds PABPC1 to recruit CAF1.

PABPC1 RRM1-BTG2 interaction

3.2 Introduction

mRNA decay is a crucial part of post-transcriptional gene regulation in eukaryotic cells. After the synthesis of the pre-messenger RNA (pre-mRNA), an 7’methyl-G cap is added to the 5’end and a poly(A) tail to the 3’end, to protect mRNAs from exonucleases. The poly(A) tails of mRNAs are bound by multiple poly(A)-binding proteins (PABPs). PABPs interact with eukaryotic initiation factor eIF4F cap binding complex consisting of the cap-binding protein eIF4E, which is connected to an RNA helicase eIF4A by the bridging protein eIF4G. According to the “closed loop” model, PABP binds to both mRNA ends (Wells et al., 1998), to facilitate translational initiation or to regulate mRNA stabilization/degradation, depending on the type of interaction. Various destabilizing sequences, such as adenylate/ uridylate (AU)-rich elements in the 3’ untranslated regions (UTRs) (Barreau et al., 2005; Chen and Shyu, 1995), elements in protein coding regions (Chang et al., 2004; Grosset et al., 2000), nonsense-containing mRNAs (Chen and Shyu, 2003) and miRNA binding sites (Behm-Ansmant et al., 2006; Wu et al., 2006) can induce deadenylation and regulate mRNA turnover. Shortening of poly(A) tails is the first major step that triggers mRNA decay in eukaryotic pathways. First, the PAN2-PAN3 complex shortens the poly(A) tail to around 100 nucleotides. Then, the CCR4-NOT complex degrades the poly(A) tail until a few nucleotides remain. After the DCP2-mediated removal of the 5’ end cap, the mRNA body is degraded by the 5’ exonuclease Xrn1 or at the mRNA 3’- end by the exosome (reviewed in (Heck and Wilusz, 2018; Wahle and Winkler, 2013).

The multifunctional PABPs play a central role in translation initiation, translation termination and mRNA decay (reviewed in (Kühn and Wahle, 2004; Mangus et al., 2003). In human cells, genes encode a single nuclear PABP (PABPN1) and four cytoplasmic PABPs (PABPC1, PABPC3, iPABP and x-linked PABPC5). The cytoplasmic PABPs consist of four RNA recognition motifs (RRMs) followed by an extended C-terminus (Figure 3.1A). With the exception of PABPC5, PABPs contain a conserved MLLE domain, also known as poly(A)- binding protein C-terminal domain (PABC) (Kozlov et al., 2001), within the C-terminal segment following RRM4. While the RRMs bind the 3’ poly(A) tail and the 5’ translation complex eIF4F, the MLLE domain is found to interact with various other regulatory proteins, recognizing their PABP-interacting motif 2 (PAM2) and recruiting them to the poly(A) mRNAs (Kozlov et al., 2001; Lim et al., 2006; Okochi et al., 2005).

Two crystal structures of the tandem PABP RRM1 and RRM2 (RRM12) with polyadenylate RNA are solved (Deo et al., 1999; Safaee et al., 2012). They reveal that the highly conserved canonical RNA recognition motifs adopt a βαββαβ topology, where a four- stranded antiparallel β-sheet is packed against two α-helices. The two tandem RRMs contact each other to create an extended β-sheet surface and bind a single-stranded RNA motif.

65 PABPC1 RRM1-BTG2 interaction

Figure 3. 1. PABPC1 domain boundaries used in this study. (A) Schematic representation of PABPC1 domain organization, with the four RRMs and the MLLE domain highlighted. RRM1, RRM2 and RRM12 constructs including their boundaries used in this study are presented below. (B) RRM1 and RRM2 amino acid sequence. The secondary structure is shown above. (C) 1H-15N HSQC spectra of free PABPC1 RRM12 (blue) is overlaid with free RRM1 (red). Aliased negative peaks in pink correspond to the amides of arginine side chains were recorded in 20 mM Na2HPO4 pH 7, 100 mM NaCl and 1 mM DTT at 600 MHz and 298K.

Adenines are recognized by conserved residues within the RNP motifs of the two RRMs. Multiple contacts are mediated by the sugar-phosphate backbone and the ribose moieties (Deo et al., 1999). In addition, structural studies show that PABP-eIF4G interaction is mediated by RRM2 and is comprised of residues within α1 and β4, which form hydrophobic interactions, hydrogen bonds and salt bridges (Safaee et al., 2012). Among PABPC1 binding partners, members of the anti-proliferative B-cell translocation gene (BTG)/ transducer of ERBB2 (Tob) family members were found to interact with the CCR4-associated factor 1 (CAF1), also known as CNOT7, a subunit of CCR4-NOT deadenylase complex. Therefore, PABPC1 is suggested to be involved in deadenylation (Ezzeddine et al., 2007).

The BTG/Tob family members function negatively in cell cycle regulation, predominantly by inducing the G0 or G1 arrest. Therefore, overexpression of these factors results in inhibition of cell proliferation (Ikematsu et al., 1999; Matsuda et al., 2001; Rouault et al., 1992).

66 PABPC1 RRM1-BTG2 interaction

Figure 3. 2. BTG2 domain boundaries used in this study. (A) Schematic representation of BTG2 full-length domain organization, with the BTG2 APRO domain highlighted in orange. APRO domain amino acid sequence including the secondary structure is highlighted The conserved Box A, B and C are indicated in light blue, green and magenta, respectively. (B) Cartoon representation of the structure of human BTG2 (pdb code 3DJU) colored in orange, except for boxA, boxB and boxC which are highlighted in light blue, green and magenta, respectively. (C) Cartoon representation of the CAF1-Tob1(APRO) (PDB code 2D5R) overlaid with the BTG2(APRO) (PDB code 3E9V) structure. CAF1 is colored in green, Tob1(APRO) in grey and BTG2(APRO) in orange. CAF1, Tob1 and BTG2 key residues involved in the interaction are shown as sticks.

Several studies revealed a reduction in BTG/Tob expression levels in cancers (Boiko et al., 2006; Kawakubo et al., 2004; Struckmann et al., 2004; Yoshida et al., 2003). Consequently, these factors might emerge as a potential biomarker for prognosis in cancer patients. In humans, the BTG2/Tob family is comprised of six conserved members and can be grouped in three classes based on amino acid sequence similarity (Matsuda et al., 2001). While BTG1 and BTG2 are highly conserved, BTG3 and BTG4 differ more in their amino acid composition. Even further apart are the two Tob proteins (Tob1 and Tob2), which contain the longest C- terminus, including a Poly(A)-binding-protein-interacting Motif 2 (PAM2) (Ezzeddine et al., 2007; Okochi et al., 2005). All BTG/Tob members are composed of the conserved, N-terminal antiproliferative (APRO) or BTG domain (Matsuda et al., 2001), which is a protein–protein

67 PABPC1 RRM1-BTG2 interaction

interaction module. This domain is found to regulate DNA binding of transcription factors and mRNA turnover by binding to CAF1 (Berthet et al., 2002; Prévôt et al., 2000; Rouault et al., 1998). This APRO domain contains two short and conserved sequences, called box A and box B, which are separated by a stretch of non-conserved residues (Matsuda et al., 2001). Compared with other family members, BTG1 and BTG2 have an additional boxC region (Figure 3.2A,B). The crystal structure of the human and mouse BTG2 APRO domain shows a globular fold and adopts a αααβααβββ topology (Yang et al., 2008). Box A is composed of the end of α2, a connecting loop, a short α-helix α3 and β1. Box B consist of two antiparallel β-strands (β2, β3) and boxC is composed of β4 and the C-terminal loop (Figure 3.2B). Superimposition of the Tob-CAF1 complex crystal structure (Horiuchi et al., 2009) with BTG2 reveals a highly conserved binding mode (Figure 3.2C). The same residues of Tob1 as well as BTG2 lie within the Box A and B region to bind CAF1.

Recently it was shown that expression of BTG2 caused mRNA poly(A) tail shortening by directly interacting with PABPC1 RRM1. Interestingly, the BTG2 APRO domain – PABPC1 RRM1 interaction was sufficient to activate CAF1 deadenylase in vitro and interaction of BTG2 and PABPC1 lead to a reduction in proliferation of U2OS cells. Moreover, mutating the boxC motif abolishes both the binding of the BTG2 APRO to PABPC1 and deadenylation (Stupfler et al., 2016).

To understand the structural basis of PABPC1-BTG2 recognition, we performed NMR titration experiments to identify the protein regions responsible for complex formation. We then used the experimental data from NMR as input for the program HADDOCK (de Vries et al., 2007; Van Zundert et al., 2016), to generate models defining the relative orientation of the domains and allowing us to propose a hypothesis for how CAF1 is recruited to the poly(A) for deadenylation.

3.3 Results

Identification of interface residues in PABPC1 RRM1 involved in BTG2 binding

To characterize the structural basis of the interaction between BTG2 and PABPC1, we analyzed the binding of the PABPC1 RRMs to BTG2 using NMR chemical shift mapping. The close match of RRM1 resonances in 1H-15N-HSQC spectra of 15N labelled PABPC1 RRM12 (residues 1-190, Figure 3.1A,B) and 15N labelled RRM1 alone (residues 1-99, Figure 3.1A,B) shows that RRM1 tumbles independently of RRM2. Because RRM1 and RRM2 do not interact with each other (Figure 3.1C) they can be studied in isolation. Backbone assignments of the

15 PABPC1 RRM1 and for BTG2 were obtained from two-dimensional (2D) N-HSQC and three- dimensional (3D) HNCA, HNCACB, HN(CA)CO, CBCA(CO)NH and HNCO (Supplementary

68 PABPC1 RRM1-BTG2 interaction

Figure 3. 3. Interaction of PABPC1 RRM1 with BTG2 full-length and APRO domain. (A) Overlay of 1H-15N HSQC spectra of free RRM1 (blue) and in complex with increasing amounts of BTG2 full length (1:1 red; 1:2 orange). Key residues are indicated by a box and the close-up view shown on the right. Aliased negative peaks in pink correspond to the amides of arginine side chains. Titrations were recorded in 20 mM Na2HPO4 pH 7, 100 mM NaCl and 1 mM DTT at 600 MHz and 298K. Arrows show the direction of the chemical shift change. (B) The combined chemical-shift perturbations (calculated using ∆δ = [(δHN)2+δN/6.51)2]1/2) of RRM1 upon binding to BTG2 full-length (1:2.5 ratio; blue) and BTG2(APRO) (1:6 ratio; red) in 20 mM Na2HPO4 pH 7, 100 mM NaCl, 1 mM DTT or 20 mM MES pH 5.5, 100 mM NaCl, 1 mM DTT respectively. The secondary structure is represented at the top and the region with high chemical shift changes (in α1 and β2) is highlighted in orange. Key residues are labelled. The red and blue horizontal lines represent two times the s.d.of all chemical shift differences. Peaks which disappear at 1:0.25 or 0.5 and did not return at higher ratios are shown by bars with >0.18 ppm.

Table A3.1, A3.2). Titrations of 15N labelled RRM1 and unlabeled full length BTG2 (residues 1-158, Figure 3.2A) or the BTG2 APRO domain, here referred to as BTG(APRO) (residues 1-126, Figure 3.2A), revealed chemical shift variations mostly in intermediate to fast exchange regime for RRM1 on the NMR time scale (Figure 3.3A), suggesting a Kd in µM range. Multiple

69 PABPC1 RRM1-BTG2 interaction

signals disappear at 1:0.25 RRM1:BTG2 ratio and do not re-appear at higher ratios, possibly due to exchange broadening of the resonances caused by the multiple states. In addition, the increase in size above 25 kDa contributes to broadening of the signals. Both, the unfavorable exchange regime and the increase in size cause a reduction in spectra quality. Spectra with broad but still observable signals are only obtained at high number of scans when the signal to noise ratio is improved. In fact, some broad and therefore weak signals, which are getting more visible upon increase of number of scans, show fast exchange. However, for other signals which disappear, no change is observed upon increase in number of scans indicating intermediate exchange for those residues. To exclude that BTG2-RRM1 form soluble aggregates, which could contribute to line broadened signals, NMR samples were analyzed by analytical size exclusion chromatography. No signal was detected in the void fractions or at higher size, indicating a complex free of aggregates (Supplementary Figure A3.1 A).

To determine the affinity of BTG2 for RRM1 and RRM12, we performed isothermal titration calorimetry (ITC) where RRM1 or RRM12 was titrated into BTG2(APRO) under various conditions (Supplementary Figure A3.1 B, Supplementary Table A3.3). Binding could not be observed, probably due to a combination of the very weak binding affinity and the heat of binding being near zero. Variations in sample cell concentrations, buffer (pH) or a decrease in temperature did not improve the ITC signal.

Mapping of the chemical shift perturbations reveal the involvement of RRM1 α1, β2 (Figure3.3B, area highlighted in yellow) and the loop following β2. Interestingly, residues within α1 such as T23, M26 and Y28 show the strongest shifts and multiple residues in α1 disappear already at 1:0.25 RRM1:BTG2 ratio. This major changes in chemical environment indicate direct interactions or conformational changes of this element. In addition, chemical shift changes appear for residues in β1 and β3. As the β-sheet of the RRM form the canonical RNA binding surface, we wondered whether BTG2 binds to RRM1 at the same interface as poly(A). To investigate this, we analyzed the binding of RRM1 to 11 nucleotide long poly(A)

(poly(A)11) by NMR (Figure 3.4). Due to the nucleotide length, the RNA is rather optimal for PABPC1 RRM12. We still used RRM1 to avoid an increase in size and broadening of the signals. In comparison to BTG2(APRO) binding, the chemical shift perturbations upon poly(A) addition showed the involvement of the canonical RNP composed of β1, β3 and β4, including the β2-β3 loop and the C-terminal linker (linker between RRM1 and RRM2) (Figure 3.4B,C), consistent with previous NMR studies (Deo et al., 1999; Safaee et al., 2012). α1 is not

70 PABPC1 RRM1-BTG2 interaction

Figure 3. 4. Comparison of PABPC1 RRM1 RNA recognition with BTG2(APRO) binding. (A) Overlay of 1H-

15 N HSQC spectra of free RRM1 (blue) and RRM1 in complex with poly(A)11 RNA (1:4 red). Aliased negative peaks in pink correspond to the amides of arginine side chains. Titrations were recorded in 20 mM MES pH 5.5, 100 mM NaCl, 1 mM DTT at 600 MHz and 298K. (B) The combined chemical-shift perturbations (calculated using ∆δ = N 2 2 1/2 [(δH ) +δN/6.51) ] ) of RRM1 in the presence of molar excess of poly(A)11 are shown. The secondary structure is represented at the top and the α1/ β2 region with the largest chemical shift changes observed upon binding BTG2(APRO) (Figure 3.3B) is highlighted in orange. The red horizontal line represents the 2x s.d. of all chemical shift differences. Peaks which disappear at 1:0.25 or 0.5 are shown by bars with >0.27 ppm (dashed line). (C)

Comparison of chemical-shift changes upon poly(A)11 (left) or BTG2(APRO)(right) binding. Cartoon representation of RRM1: poly(A)11 (left) and free RRM1 (right) colored in grey; modified from PABP RRM12 structure (pdb code 1CVJ). RNA is shown in stick representation and the carbon atoms are indicated in yellow, oxygen in red, nitrogen in blue and phosphorus in orange. Chemical-shift perturbations larger than 2x sd are colored in red while signals which disappear at 1:0.25/0.5 are shown in blue.

71 PABPC1 RRM1-BTG2 interaction

affected by poly(A) (Figure 3.4B, area highlighted in yellow, Figure 3.4C), suggesting that whereas α1 is important for recognition of BTG2 it is not involved in poly(A) binding. However, the loop following β2, shows chemical shift perturbations upon binding to both, poly(A) and BTG2(APRO) (Figure 3.4C), suggesting that poly(A) and BTG2 binding sites may either overlap or could be allosterically coupled.

Overall, our NMR studies confirm the binding of RRM1 with BTG2(APRO) and predict α1 and β2 of RRM1 as the interaction surfaces.

Poly(A) does not prevent BTG2 to bind PABPC1 RRM1

PABPC1 binds the 3’-poly(A) tails of mRNAs. The two available crystal structures of PABPC1 RRM12 with poly(A) are almost identical, but show two different conformations of the 3’-end adenine. In one, it is located within RRM1 either close to β3-β2, stacking on top of the preceding adenine (Safaee et al., 2012) or in the second, it is near β2 and the following loop (Deo et al., 1999) (second conformation is shown in Figure 3.4C). As the 3’-end lies within the RRM1 β2, which shows chemical shift perturbations upon BTG2(APRO) binding, this raises the question whether the adenine has a negative effect on BTG2 binding. To determine whether BTG2 competes for binding and causes displacement or rearrangement of RNA from RRM1, BTG2(APRO) was titrated into a sample with 15N-labeled RRM1 in complex with poly(A) and monitored by NMR (Figure 3.5A). No significant differences are observed compared to BTG2(APRO) binding to RRM1 only. This observation indicates that poly(A) does not influence BTG2 binding to RRM1.

To validate this finding, we performed ITC experiments to quantify the affinity of

PABPC1 RRM12 for poly(A)11 in presence or absence of BTG2(APRO). We obtain Kds for the binding of poly(A)11 to RRM12 of 0.22 ± 0.05 μM (Figure 3.5B), which is in the similar range as observed before (Safaee et al., 2012). Similarly, poly(A)11 binds RRM12 in complex with

BTG2(APRO), at both a 1:1 and a 1:6 RRM12:BTG2(APRO) ratio, with a comparable Kd of 0.17 ± 0.06 μM, indicating no effect of BTG2(APRO) on RRM1-poly(A) affinity. However, our ITC experiments show a 1:2 RNA:protein stoichiometry (Figure 3.5B), which disagrees with the previous results, where one RRM12 was bound to poly(A)11 (Safaee et al., 2012). This difference might be due to the use of a different reducing agent. TCEP is a stronger reducing agent and keeps RRM12 in the reduced form. In our conditions, 10 mM β-mercaptoethanol might not be enough to reduce RRM12.

72 PABPC1 RRM1-BTG2 interaction

Figure 3. 5. Comparison of BTG binding to PABPC1 RRM1 or RRM1 in complex with poly(A). (A) Top: Overlay of 1H-15N HSQC close-up view spectra of free RRM1 (blue) and RRM1 in complex with increasing amounts of BTG2(APRO) (1:0.5 red; 1:2 orange). Bottom: Overlay of 1H-15N HSQC close-up view spectra of RRM1 in complex with poly(A) (1:2.5 ratio; blue) and in complex with increasing amounts of BTG2(APRO) (1:0.5 red; 1:2 orange).

Titrations were recorded in 20 mM Na2HPO4 pH 7, 100 mM NaCl, 1 mM DTT at 600 MHz and 303K. (B) Comparison of ITC profiles of RRM12 (left) and RRM12 in complex with BTG2(APRO) (right) titrated with poly(A)11. Errors indicate s.d. of two measurements. Top: ITC raw data as a function of time, bottom: integrated heat release. Red line represents the fit, obtained by the Origin 7 Software with a one binding site model.

Together, our data suggest that BTG2(APRO) binds to the single RRM1 or RRM1- poly(A) complex in the same fashion and that poly(A) in the RRM1-poly(A)11 or RRM12- poly(A)11 complex does not hinder BTG2(APRO) binding to RRM1.

BTG2 APRO domain binds PABPC1 RRM1 with higher affinity than RRM2

To identify the RRM12 binding site of BTG2, we carried out NMR titrations of 15N labelled BTG2(APRO) and unlabeled PABPC1 RRM1 or RRM2 using NMR chemical shift mapping (Figure 3.6). The chemical shift perturbations of 15N labelled BTG2(APRO) upon RRM1 addition are mostly in intermediate to fast exchange regime on the NMR time scale (Figure 3.6A). Upon RRM2 binding, only a few chemical shifts show fast exchange regime, indicating a very weak or unspecific binding of RRM2 (Figure 3.6B,C). Similar to the titrations with 15N- labeled RRM1, multiple signals of the 15N-labeled BTG(APRO) disappear at 1:0.25 or 1:0.5

73 PABPC1 RRM1-BTG2 interaction

Figure 3. 6. Binding of RRM1 vs RRM2 to BTG2(APRO). (A) Overlay of 1H-15N HSQC spectra of free BTG2(APRO) (blue) and in complex with increasing amounts of RRM1 (1:0.25 red; 1:0.5 orange; 1:1 green; 1:2 cyan; 1;4 magenta). The number of scans (NS) is 2, otherwise indicated if different. Key residues are indicated by a box and the close-up view shown on the right. (B) Overlay of 1H-15N HSQC spectra of free BTG2(APRO) (blue) and in complex with increasing amounts of RRM2 (color code as in A) are illustrated. Same key residues as in A, which show an effect for RRM1 but not for RRM2, are shown on the right. (C) Positive effects upon RRM2. Comparison of overlays of 1H-15N HSQC spectra close-up views of residues indicated by an arrow in A (top; upon RRM1 addition) and B (bottom; upon RRM2 addition) (color code as in A). Aliased negative peaks in pink correspond to the amides of arginine side chains. Titrations were recorded in 20 mM Na2HPO4 pH 7, 100 mM NaCl and 1 mM DTT at 600 MHz and 298K. In the close-up view, arrows show the direction of the chemical shift change.

74 PABPC1 RRM1-BTG2 interaction

RRM1:BTG2(APRO) ratio and do not re-appear at higher ratios (Figure 3.6A,C), indicating a change of the chemical environment of these residues. Therefore, we assume that these corresponding amino acids are involved in direct contacting RRM1.

27% of BTG2(APRO) backbone residues (including seven prolines) are missing in the backbone assignment (Figure 3.7A,B, Supplementary Table A3.2), potentially due to exchange causing line broadening. In fact, the 1H-15N HSQCs spectrum shows multiple smaller double peaks, indicating multiple conformations of the BTG2(APRO) under our experimental conditions. Therefore, it is unclear whether these regions are involved in binding. This includes the beginning of α1, the loop between α2 and α3, some residues of α4 and most residues in α5. However, all β-strands are assigned, including the important boxC region (Stupfler et al., 2016). Therefore, to verify the involvement of boxC and identify additional residues important for binding to RRM1 or RRM2, we analyzed the combined chemical shift plots of 15N labelled BTG2(APRO) with RRM1 or RRM2 (Figure 3.7A). Residues with chemical shift perturbations larger than 2x standard deviation or residues disappearing at 1:0.25/ 1:0.5 are highlighted on the published human BTG structure (pdb code 3DJU, (Yang et al., 2008), (Figure 3.7B, left).

This data confirm the involvement of boxC (Figure 3.7A, area highlighted in pink, Figure 3.7B, left), the end of α1 and additionally, show chemical shift changes in the entire β- sheet surface. In comparison, addition of RRM2 shows only small chemical shift changes mainly in the boxC region, potentially due to unspecific binding of conserved residues within RRM1 or RRM2.

Mutation of the APRO boxC region of BTG2 preserves the fold but abolishes binding to PABPC1 RRM1

The BTG2(APRO) boxC mutant was shown to still interact with CAF1 but lost the ability to bind PABP (Stupfler et al., 2016). 1H-15N HSQCs were recorded to test the effect of the boxC mutation, D116K+ S118P+ I119V+ C120K, here abbreviated with BTG2(APRO) boxC mut, on the BTG2(APRO) fold. The corresponding BTG2 residues are mutated to the ones found in Tob1 APRO domain (Figure 3.8A), which alone is unable to bind RRM12. Despite the four residues being mutated, the folding of the BTG2 APRO domain was not affected (Figure 3.8B). Chemical shift changes were only observed in the corresponding boxC region and the neighboring β3. NMR titrations of 15N-labeled BTG2(APRO) boxC mut and unlabeled RRM1 or RRM2, showed no significant changes (Figure 3.8D,E) and as expected, no binding was seen when performing titrations of 15N-labeled RRM1 with BTG2(APRO) boxC mut

75 PABPC1 RRM1-BTG2 interaction

Figure 3. 7. Identification of interface residues in BTG2 APRO domain. (A) Combined chemical-shift perturbations of BTG2(APRO) upon binding to RRM1 (black) or RRM2 (blue). The secondary structure is shown at the top and the boxC region is highlighted in pink. The black or blue horizontal line represents the 2x s.d. of all chemical shift differences. Peaks which disappear at 1:0.25 or 1:0.5 and did not return at higher ratios are indicated by bars with >0.16 ppm (dashed line). (B) Cartoon representation of the human BTG2 structure (pdb code 3DJU). Residues that show significant chemical shift changes (above the 2x sd) or disappear upon addition of RRM1 (left) or RRM2 (right) are shown in blue or red respectively. Undetectable or unassignable backbone residues are colored in orange.

(Figure 3.8F). Interestingly, purification of BTG2(APRO) over a size exclusion column shows a main BTG2(APRO) peak, but also a shoulder at higher molecular weight (Supplementary Figure A3.2A). This shoulder is not observed for the BTG2(APRO) boxC mut. We wondered whether BTG2 is able to form homo-dimers possibly bridged by intermolecular disulfide bonds.

To test this, we carried out T1 and T2 NMR relaxation experiments. τc was calculated from the ratio of 15N longitudinal (T1) and transverse (T2) relaxation times of each residue. For

BTG2(APRO) we observed a mean τc of 12.68 ± 02.15ns, which corresponds to a 21.12 ± 3.58 kDa (Supplementary Figure 3.2B), assuming a spherical particle at 298K (Rossi et al.,

2010). In contrast, the BTG2(APRO) boxC mut has a mean τc of 10.15 ± 1.92 ns, under the same conditions, which corresponds to a molecular weight of 16.9 ± 3.20 kDa. This small

76 PABPC1 RRM1-BTG2 interaction

decrease in size excludes the possibility of BTG2(APRO) forming homo-dimers mediated by a surface, where both domains tumble depend of each other. It rather suggests that BTG2 forms disulfide bonds, where both domains tumble independent of each other, which only lead to the small increase in τc. The mutation of C120 in the boxC mut eliminates disulfide bond formation. Interestingly, the mutations seem to induce flexibility into the boxC region, as the two residues G117, S118 show a decrease in τc. Whether the following residues become more flexible cannot be determined due to the missing backbone assignment for this region of the boxC mut.

These findings validate the boxC region as being the main binding surface. Its mutation still preserves the fold but abolishes binding. In addition, we found that C120 in the boxC region is prone to form disulfide bonds even under high reducing conditions (1-5 mM DTT).

BTG2 and PABPC1 RRM12 complex HADDOCK model

Previous crystallization attempts of the BTG2-RRM12 complex resulted in crystal growth but crystals only contained either BTG2 or RRM12 molecules. New attempts with thioredoxin, which can be used in a fusion construct to drive crystallization of linked proteins (Corsini et al., 2008), was tested as a Trx-BTG2 fusion protein. Trx-tagged BTG2 in complex with RRM12 did not form any crystals, probably due to the unsuitable length of the linker between Trx and BTG2. Only the BTG2(APRO) boxC mutant formed crystals in the NMR tube after storage at 4°C (Supplementary Figure A3.3). These crystals were not measured.

In our NMR studies of the BTG2(APRO)-PABPC1 interaction, signals were broadened either due to exchange on the intermediate NMR timescale or due to the increased molecular weight of the complex, which prevented the acquisition of high quality NMR data required for NMR structure determination. In addition, BTG2 is unstable at high concentrations and increased temperatures.

Therefore, the chemical shift data was used to determine a model of the BTG2- PABPC1 RRM1 complex using the program HADDOCK and the known structures of BTG2 (pdb 3DJU) and PABPC1 RRM12 (pdb 1CVJ). Models were calculated by using various combinations of the pdb 1CVJ. As an input model, we tried out several combinations that lead to similar results. We used either RRM12 in complex with RNA, free RRM12 or free RRM1. Same for pdb 3DJU, where either BTG2 alone was used as an input, or in complex with CAF1 (derived from the superimposition with Tob1-CAF1 complex (pdb 2D5R) (Table 3.1). We assume that the amino acids, disappearing in our NMR titrations at 1:0.25/ 1:0.5 protein:protein ratio, are in or closest to the binding interface. We used these BTG2 and RRM1

77 PABPC1 RRM1-BTG2 interaction

Figure 3. 8. Mutation of the BTG2(APRO) boxC region preserves the fold but abolish binding to PABPC1 RRM1. (A) The sequence alignment of BTG1-4 and Tob1 APRO domains. The boxC region is indicated by a pink box and the residues mutated in the BTG2(APRO) boxC mutant are shown below. (B) Overlay of 1H-15N HSQC spectra of BTG2(APRO) (blue) and the boxC mutant (red). The residues S110 and Y111, which shift can be followed, are indicated by an arrow. (C) Cartoon representation of the human BTG2 structure (pdb code 3DJU). Residues S110 and Y111, where chemical shift changes can be followed, are shown in light red. Most of the residues in the boxC region of the mutant appear at a new position and assignments are missing (red). Unassigned

78 PABPC1 RRM1-BTG2 interaction

residues of BTG2 are colored in orange. (D) Overlay of 1H-15N HSQC spectra of BTG2(APRO) boxC mut (blue) and BTG2(APRO) boxC mut in complex with RRM1 (1:2, red). (E) As above but overlay of BTG2(APRO) boxC mut in complex with RRM2. A close-up view is shown for the few chemical shift changes. (F) Overlay of 1H-15N HSQC spectra of RRM1 (blue) and in complex with increasing amounts of BTG2(APRO) boxC mutant (1:1 red; 1:2 orange;

1:4 green). NMR experiments were recorded in 20 mM Na2HPO4 pH 7, 100 mM NaCl and 1 mM DTT at 600 MHz and 298K. Aliased negative peaks in pink correspond to the amides of arginine side chains.

residues as input for HADDOCK (Table 3.1). The best models are based on the best (most negative) HADDOCK score (below -100; which is the sum of electrostatic interactions, van der Waals repulsion, average buried surface area, etc), highest cluster size, lowest RMSD (from the overall lowest-energy structure) and most negative Z-Score (lower then -1; indicates how many standard deviations from the average this cluster is located), (Table 3.2) reveal three major conformations (Figure 3.9). These three conformations were the most abundant in all HADDOCK calculations. The first two conformations (model 1 and 2) are similar and only differ in the orientation of the BTG2 β4 and RRM1 β2 with respect to each other (Figure 3.9 A,B). In these two conformations, the BTG2 and PABPC1 RRM1 β-strands form an extended β- sheet surface. Interestingly, RRM1 α1 is inserted into the BTG2 α1-β4 groove. In the third conformation (model 3), the interaction is mostly mediated by the BTG2 β2, β3-β4-loop and β2-β3 loop of PABPC1 RRM1.

Table 3. 1 Input parameters for HADDOCK calculations

Domains BTG2 residues PABPC1 RRM1 residues input Model BTG2+RRM1 T26, R112, I113, G114, E24, A25, L27, E29, K30 1 G117, I119, V121, L122 Model BTG2+RRM1 L23, T26, R112, I113, G114, E24, A25, L27, E29, K30, 2 G117, I119, V121, L122 R41, V42, C43, R44, D45

Model BTG2- L23, T26, R112, I113, G114, E24, A25, L27, E29, K30, 3 CAF1+RRM12- G117, I119, V121, L122 R41, V42, C43, R44, D45 poly(A)

79 PABPC1 RRM1-BTG2 interaction

Table 3. 2. HADDOCK scores of the best clusters.

HADDOCK Cluster RMSD Z- score size Score Model 1 -104.8 +/- 3.3 100 2.0 +/- 1.3 -2.4

Model 2 -114.0 +/- 2.4 167 1.3 +/- 0.8 -1.7

Model 3 -112.4 +/- 2.1 66 1.0 +/- 0.7 -1.8

BTG2 directly interacts with CAF1, a subunit of the CCR4–NOT complex. Previous immunoprecipitation experiments showed that the BTG2 APRO domain precipitates CNOT7, one of the two CAF1 paralogues. In addition, PABPC1 is co-precipitated by BTG2 APRO domain, highlighting strong binding of BTG2 to either CAF1 or PABPC1 (Stupfler et al., 2016). The combination of our NMR and ITC studies revealed no difference between BTG2 binding to PABPC1 RRM1 or binding to poly(A)11:RRM1. Similarly, poly(A)11 binding to PABPC1 RRM12 or RRM12-BTG2 did not differ. Therefore, we assume that binding of RRM1 to poly(A) does not hinder its ability to simultaneously bind BTG2. Under the assumption that BTG2- CAF1 binding to PABPC1 RRM12-poly(A) does not lead to dissociation of any of these binding partners, we were able to exclude some additional models proposed by HADDOCK with lower scores (Supplementary Table A3.4). Superimposition of these models and the Tob1:CAF1 complex (pdb 2D5R) or RRM12-poly(A) complex (pdb 1CVJ) indicates an unfavorable conformation of RRM12. Either, RRM12 would clash with CAF1 (Supplementary Figure A3.4 A), or the poly(A) 3’-end appears on the back site of BTG2 with increased distance towards the catalytic pocket of CAF1 (Supplementary Figure 3.4 B).

In Model 1, BTG2(APRO) and PABPC1 RRM1 β-strands interact and form an extended β-sheet surface, mediated by hydrogen bonds (Figure 3.10A). The BTG2 β3-β4 loop within the boxC region seems not to be involved in binding. C120 sidechain is buried between the BTG2 β2 and the RRM1 α1 (Figure 3.10A, right). PABPC1 RRM1 R41 and R44 sidechains form a few hydrogen bonds (Figure 3.10A, right). Model 2 is quite similar to model 1. Model 3 adopt a different orientation, where the BTG2 boxC β3-β4 loop interacts with the PABPC1 RRM1 β2, α1 and adjacent loop regions (Figure 3.10B). The positioning of the PABPC1 RRM1 α1 is different. BTG2 C120 sidechain forms a hydrogen bond to the backbone amide of PABPC1 RRM1 E24 (Figure 3.10B, right). The PABPC1 RRM1 R41 and R49 sidechains form multiple hydrogen bonds to the sidechains of BTG2 E99, D116 and backbone carbonyls of E115, G117 respectively (Figure 3.10B, right).

80 PABPC1 RRM1-BTG2 interaction

Figure 3. 9. BTG2-RRM12 HADDOCK models. (A) Cartoon representation of model 1. BTG2 and RRM1 interaction is mediated by the parallel β-strand interface of BTG2 β4 and RRM1 β2. RRM1 α1 is inserted into the BTG2 α1 β4 groove. BTG2 is colored in orange and PABPC1 RRM12 in blue. (B) Cartoon representation of model 2. Orientation similar to the one in A, but both BTG2 β4 and RRM1 β2 strands are more parallel to each other. Color code as in A. (C) Cartoon representation of model 3. BTG2 and RRM1 interaction is mediated by the BTG2 β3-β4 loop and RRM1 β2-β3 loop. Color code as in A.

81 PABPC1 RRM1-BTG2 interaction

Figure 3. 10. Comparison of the hydrogen bond networks in model 1 and 3. (A) In model 1, BTG2 and PABPC1 RRM1 β-strands form an extended β-sheet surface. There, hydrogen bonds are mediated by backbone atoms. PABPC1 RRM1 α1 is inserted into the BTG2 β4-α1 groove. Residues in the binding interface are colored in orange (BTG2), cyan (PABPC1 RRM1) and shown as sticks. A few residues, which are forming hydrogen bonds, are highlighted on the right. (B) Model 3 adopt a different orientation, where the BTG2 boxC region interacts with the PABPC1 RRM1 β2-α1 and adjacent loop regions. The positioning of the RRM1 α1 is different. Color code as in A. (C) Superimposition of BTG2 and Tob1 (derived from the CAF1-Tob1(APRO) structure (PDB code 2D5R)). Color code as in A, Tob1 is shown in blue.

82 PABPC1 RRM1-BTG2 interaction

Interestingly, the superimposition of BTG2(APRO) with Tob1(APRO) highlights the importance of the BTG2 boxC residues, which are not present in Tob1. This might explain why Tob1(APRO) is not able to bind PABPC1 RRM1 (Figure 3.10C). Tob1 K110 corresponds to BTG2 C120. The long aliphatic side chain in Tob1 might lead to steric clashes and prevent PABPC1 RRM1 binding. Additionally, PABPC1 R41 sidechain forms multiple hydrogen bonds to E99 as well as to D116 sidechains. BTG2 E99 corresponds to Tob1 D89, which might be too short to form similar hydrogen bonds. Additionally, BTG2 D166 corresponds to Tob1 K106, which could even lead to repulsion of the positively changed PABPC1 RRM1 R41 sidechain.

Overall, our HADDOCK calculations propose two major orientations how BTG2 might bind to PABPC1 RRM12. If and which of these models is physiological relevant, needs to be further validated.

BTG2 bridges PABPC1 RRM1 bound to poly(A) to facilitate the transfer of poly(A) to the active center of CAF1 deadenylase

Both our models indicate how BTG2 serves as a platform to bridge PABP and CAF1 to induce deadenylation. However, poly(A), which is bound by RRM12, need to reach the catalytic center of CAF1. BTG2 and Tob1 APRO domains adopt the same fold and contain a conserved binding interface for CAF1 (Figure 3.2C) (Yang et al., 2008). Therefore we superimposed the Tob1-CAF1 complex and RRM12-poly(A) with our BTG2-RRM12 HADDOCK models to obtain quaternary CAF1-BTG2-RRM12-poly(A) models (Figure 3.11). In all models, the 3’-end of poly(A) is on the same side as the active center of CAF1. However, model 3 shows a shorter distance between the 3’-end of the poly(A) and the catalytic site (Figure 3.11B,D). One possibility could be that BTG2 E99 and D116, which create a negative charged patch, close to the RNA phosphate, lead to destabilization and repulsion of the 3’end. However, R89 of RRM12 grabs the RNA and prevents the RNA from displacement and directs it towards the CAF1 catalytic pocket (Figure 3.11D, close-up view).

Thus, our models provide a hypothesis how RRM12-poly(A) are orientated by BTG2 to position the poly(A) 3’-end close to the CAF1 active center. The question remains how poly(A) is transported into CAF1 active site assuming CAF1 is stably bound to BTG2. The mechanism of this dynamic movements, need to be further investigated.

83 PABPC1 RRM1-BTG2 interaction

Figure 3. 11. BTG2-RRM12 HADDOCK models superimposed with poly(A) and CAF1. (A) Cartoon representation of CAF1-BTG2-RRM12-poly(A) model 1. 3’-end of poly(A) is close to the BTG2 C-terminus. CAF1 is colored in green, BTG2 in orange, RRM12 in blue and poly(A) is shown as sticks in yellow. Residues of CAF1 active center are shown as sticks in magenta. (B) Cartoon representation of CAF1-BTG2-RRM12-poly(A) model 3. 3’-end of poly(A) is close to the BTG2 β3-β4 loop and closer to the active site of CAF1 than in model 1. Color code as above. (C) The electrostatic surface potentials of CAF1-BTG2-RRM12-poly(A) model1. Poly(A) 3’-end and CAF1 catalytic site are indicated by an arrow. poly(A) is shown as sticks in yellow. (D) The electrostatic surface potentials of CAF1-BTG2-RRM12-poly(A) model2. Poly(A) 3’-end and CAF1 catalytic site are indicated by an arrow. poly(A) is shown as sticks in yellow. Below is a close-up view of the poly(A) 3’-end and CAF1 catalytic center. BTG2 E99 and D116 create a negative patch close to the RNA phosphate while RRM12 R89 prevents the RNA from dissociation and directs the 3’-end RNA towards the catalytic site.

84 PABPC1 RRM1-BTG2 interaction

3.4 Outlook

Investigation of the >25 kDa BTG2-RRM1 complex by NMR

Previous and new attempts to obtain the structure of the BTG2-RRM12 complex by crystallography were unsuccessful. In addition, our NMR approach turned out to be challenging due to a combination of multiple aspects. The signals upon interaction are in intermediate exchange on the chemical shift time scale and the increase in size above 25 kDa leads to severe broadening of the signals. Both leads to decrease in spectra quality. In addition, the instability of BTG2 at high concentrations and higher temperatures made it impossible to obtain high quality NMR data.

At the beginning of this study, multiple conditions were tested to improve the spectra quality of the complex (temperature, salt concentrations and pH (data not shown)). However, these tests were recorded at low number of scans. A different exchange regime was potentially missed due to the overall poor spectra quality caused by the protein size. Eliminating the broadening of the signals due to the size can be achieved by using deuterated proteins. The deuterated nonexchangeable protons reduce relaxation which leads to sharper signals. This will enhance resolution and sensitivity of both triple resonance experiments to obtain backbone assignments and correlation experiments resulting in more complete chemical shift mapping data.

In addition, tests with poly(A) were only done on the 11 nucleotide long poly(A), which is the optimal binding length for two domains as for RRM12 (Deo et al., 1999; Safaee et al., 2012). However, it is unfavorable for RRM1, due to two binding sites, multiple registers and the additional increase in size. Thus, NMR experiments for the ternary complex should be repeated with a shorter poly(A). Moreover, we found that C120 is able to form a disulfide bond, linking two BTG2 molecules. On one hand, this leads again to an increase in size, on the other hand, a BTG2 dimer linked by a disulfide bridge might have a different confirmation than the free BTG2. This would explain the double peaks seen in the NMR experiments for BTG2. These BTG2 dimers might be inactive in binding and decrease the concentration of active BTG2s in the sample. Therefore, mutation of C120 would improve the spectra quality due to the decrease in size and clearer spectra due to the elimination of double peaks. As C120 lies within the conserved boxC region, it is important to test that mutating this cysteine does not affect its physiological function. Lastly, while we were able to assign all backbone resonances of PABPC1 RRM1, 27% of BTG2(APRO) backbone amide signals are still missing. This could be due to intrinsic dynamics of these regions that lead to peak broadening. As these residues

85 PABPC1 RRM1-BTG2 interaction

were missing in the NMR spectra, we could not provide any information whether these amino acids play a role in RRM1 binding.

HADDOCK model validation

Our modeling approach propose two major orientations of BTG2-RRM12. Which of these models is relevant, needs to be further validated. Additional distance restrains should be collected to validate the models in vitro, for example by EPR or by paramagnetic relaxation enhancement (PRE). Mutagenesis of the binding interface can provide additional information in vitro and in cell culture experiments

Mutating BTG2 C120 to a C120K point mutant will help us elucidating the mechanism of interaction. The mutation to a smaller residue as in C120S will prevent disulfide bond formation but keep its ability to form a potential hydrogen bond, as seen in model 3 (Figure 3.10B). Mutating C120 to an alanine to abolish hydrogen bonds could help validate our models, as in model 1, C120 is not involved in any hydrogen bonds. Both our HADDOCK models show that the mutation C120K will lead to steric clashes of the long K120 sidechain with α1 or other residues from RRM1. This could be an explanation how the mutation abolishes binding to RRM1 (Figure 3.10C). The residues, S118, I119 are only forming hydrogen bonds between backbone atoms while D116 is not involved in binding based on model 1 (Figure 3.10A). If model 1 is correct, mutating these residues will not have an effect on RRM1 binding. In Model 3, D116 and S118 sidechains do form direct hydrogen bonds to residues of RRM1. Therefore, mutating D116 and S118 into single and double point mutants might provide additional evidence to exclude one of the two models.

In addition, SAXS and SANS experiments of the BTG2-RRM12-poly(A)11, CAF1-

BTG2-RRM12-poly(A)11 complex can be measured to validate the ternary and quaternary complex and to position of the domains with respect to each other.

3.5 Discussion

In this study, we examined the structural basis for BTG2 APRO domain interaction with PABPC1 RRMs. The chemical shift mapping revealed the binding interface of PABPC1 RRM1 and BTG2 APRO domain. This allowed us to generate models by HADDOCK and propose a mechanism how both proteins interact with each other and recruit CAF1 deadenylase to poly(A) RNAs.

86 PABPC1 RRM1-BTG2 interaction

New RRM-domain interaction

Interestingly, our study provides an additional example on the versatility of RRM domains. Proteins containing RRMs use various strategies to regulate gene expression. RNA can be recognized by a canonical RRM interface with aromatic side-chains located in the conserved RNP1 (in β3) and RNP2 (in β1). Non-canonical RRMs bind RNA by other regions outside the RNPs. In addition, RRM-RRM interaction or other domain-RRM interactions can lead to looping of RNAs, an increase in RNA binding affinity but also result in competition with the RNA by using the same interface (reviewed in (Cléry et al., 2008). PABPC1 RRMs are multi-functional domains. RRM12 contain a canonical RNA binding surface, but also protein interaction surfaces to further regulate its functions. To increase expression of polyadenylated mRNAs, PABP RRM2 binds the eukaryotic initiation factors eIF4G. Interestingly, RNA binding increase the affinity of eIF4G for PABP. Here we show that BTG2(APRO) recognition is mediated primarily by residues within α1 of RRM1. Binding of BTG2 has no effect on poly(A) affinity (Figure 3.5). Similar binding strategies, mediated by α1, were observed for other RRM containing proteins. PTB RRM2 is able to bind both, RNA and its co-repressor Raver1. There, PTB RRM2 forms a groove, composed of primarily α1 and additional residues from the extended loop region linking α2 with β4, opposite of the RNA binding surface (Rideau et al., 2006). HuR / ELAVL1 RRM3 provides another interesting example. The RRM3 is able to dimerize in a concentration depend manner mediated by a conserved W261 within helix α1. Dimerization of RRM3 has no effect on the affinity of HuR for short ARE RNA motifs. Only a long enough RNA linker between the two RRM3 binding sites causes an increase in affinity (Chapter 2).

Bridging role of BTG2

Our structural analysis indicate how BTG2 serves as a platform to link PAPBC1 to

CAF1. poly(A)11 shows no difference in affinity when bound to RRM12 alone or RRM12 in complex with BTG2 (Figure 3.5). However, it is unknown yet whether the BTG2-CAF1 complex acts in a cooperative manner to bind RRM12-poly(A). We validate that the boxC region of BTG2 APRO domain is the main binding surface for PABPC1 RRM12. Mutation of the boxC region still preserves the fold but abolishes binding to RRM12 (Figure 3.8). We cannot exclude the possibility that the interaction of both binding partners is dynamic and adopts multiple confirmations. Thus, it could be that both our models capture two potential confirmations of BTG2(APRO)-PABPC1 RRM1.

87 PABPC1 RRM1-BTG2 interaction

CAF1-BTG2-RRM12-poly(A) model

Here, our modeling approach propose two models for BTG2-RRM12 complex (Figure 3.9). Both models show BTG2 binding to α1-β2 of PABPC1 with residues within the boxC region and thereby positioning the 3’-end of the poly(A) RNA in a close proximity to the CAF1 active site. In the second model, the distance of poly(A) 3’-end and the catalytic pocket is shorter than in the first model (Figure 3.11). Interestingly, both models show a negative patch close to the poly(A) 3’-end which might lead to destabilization of this nucleotide by repulsion of the negatively charged phosphate backbone. This could cause a decrease in affinity to this binding site. However, the RRM12 R89 sidechain grabs the RNA like an arm and prevents the nucleotide from a complete unbinding (Figure 3.11C,D). In the 2nd model, R89 sidechain might even guide the poly(A) 3’-end towards CAF1 active site (Figure 3.11D, close up view).

Figure 3. 12. Model of recruitment of deadenylases onto mRNAs by BTG2. BTG2 (orange) recruits CAF1 (green) and other deadenylases (CCR4 in violet) to the 3’-end of the poly(A) by binding to RRM1.

Our models provide a good hypothesis on how BTG2 stimulates mRNA deadenylation and reduces cell proliferation by recruiting CAF1 deadenylase through binding to the first RRM domain of PABPC1 and positioning the poly(A) 3’-end in the proximity to the CAF1 active site (Figure 3.11, Figure 3.12). However, the question remains how poly(A) is transported into the CAF1 catalytic pocket. It is possible that additional binding partners might help driving the dynamic process of poly(A) translocation. On the other side, it has been shown that BTG2 and PABPC1 stimulate CAF1 deadenylase activity in vitro without additional factors (Stupfler et al., 2016). Another hypothesis which could explain the poly(A) translocation, is based on multiple register binding of polynucleotide binding proteins. The PABP RNA recognition is mostly mediated by the RNA sugar-phosphate backbone (Deo et al., 1999). The binding mode might be highly dynamic with a constant binding and unbinding or sliding along the poly(A) RNA. By doing so, the 3’-end might get closer to CAF1 until the first adenine can be specifically recognized by CAF1 and pulled into the catalytic pocket. BTG2 could not only be a bridge

88 PABPC1 RRM1-BTG2 interaction

between CAF1 and RRM12 but also destabilize the 3’end poly(A) RNA and thereby induce the sliding mechanism. Additional conformational changes can contribute to that process. Multiple register binding was reported for several pyrimidine-rich binding proteins such as U2AF65, hnRNPC or HuR (Cieniková et al., 2014; Mackereth et al., 2011),(Chapter 2). Knowing whether PABPs use a similar mode to recognize poly(A) tails would be of high interest to understand the dynamics of CAF1-BTG2-RRM12-poly(A)11 complex.

3.6 Materials and Methods

Constructs

Plasmids containing BTG/Tob members and their mutants as well as PABPC1, PABPC1 RRM1, -RRM2 and -RRM12 used in this study were described before (Stupfler et al., 2016). BTG2, PABPC1 RRM1 and PABPC1 RRM12 were sub-cloned into the pETGB-1a- and/or petTrx-vector a kind gift from Dr. Arie Geerlof and Gunter Stier, EMBL.

Expression and Purification of recombinant proteins

Recombinant human 6-His-tagged PABPC1 RRM2 and 6-His-tagged –GB1-fusion proteins (PABPC1 RRM1, -RRM12 and BTG2) were over expressed in BL21(DE3 ) Codon plus (RIL)

15 cells (Novagen). Cells were grown in LB rich or M9 minimal media supplemented with NH4Cl

15 13 or NH4Cl and C-glucose at 37°C until OD600 reached 0.6-0.8, temperature was reduced to 20°C and the expression induced with 0.5 mM isopropyl β-D-thiogalactoside (IPTG). Cells were harvested by centrifugation at 4°C, 20 min at 6000rpm after 19- 22 h. The cell pellet was either stored at -20°C or lysed with a microfluidizer at 75 PSI (0.52 MPa) in 20 mM Tris- HCl

(pH8), 800 mM NaCl, 2 mM MgSO4, 5 mM CaCl2, 10 mM imidazole, 10 mM β- mercaptoethanol, 0.25 mg/mL lysozyme, 10 µg/mL DNase, one complete EDTA free protease inhibitor tabled (Roche) in a total volume of 50 ml. Lysate was cleared by centrifugation at 4°C, 40 min and 17000 rpm and was applied onto a 5 ml HiTrap Chelating HP column (GE Healthcare) charged with nickel and purified on an AKTA prime system (washed with 20mM Tris- HCl pH8, 800 mM NaCl, 10 mM imodazole and 10 mM β-mercaptoethanol and eluted with 50% the same buffer but 500 mM imidazole). Pooled fractions were dialyzed against 3 L buffer (20 mM Na2HPO4 (pH7), 100 mM NaCl, 1 mM EDTA, 1 mM DTT and without imidazole) in the presence of an in house made TEV protease (1:50 ratio mg-TEV: mg-RRM3) at RT overnight to remove the 6-his-GB1-tag. Additional dialysis against 2 L of the same buffer but

89 PABPC1 RRM1-BTG2 interaction

no EDTA was performed for 4h. The protein was further purified by reloading the sample on the HiTrap Chelating HP column to remove the 6-his-GB1-tag, overnight dialysis as above, followed by size exclusion chromatography using the Superdex75 prep grade size exclusion column (GE Healthcare) in 20 mM Na2HPO4 (pH7), 100mM NaCl, 1mM DTT. The protein was stored in size exclusion/NMR buffer at -80C.

RNA synthesis

The poly(A)11 was provided by Jürg Hunziger (Novartis, Basel) and synthesized as described in (Masliah et al., 2018).

NMR spectroscopy

Most NMR spectroscopy measurements were done in 20 mM Na2HPO4 (pH7), 100 mM

NaCl,1 mM DTT, 10% D2O at 298K using Buker AVIII-500 MHz, AVIII-600 MHz and AVIII-700

MHz (equipped with cryoprobes). Backbone experiments and T1 and T2 relaxation experiments of BTG were additionally preformed in 20 mM MES (pH 5.5), 50 mM NaCl,1 mM DTT, 10%

D2O.

For the 1H, 15N and 13C assignments of the protein backbone of PABPC1 RRM1 and BTG2 2D 1H,15N-HSQC and 3D HNCO, HN(CA)CO, HNCA, HN(CO)CA, HNCACB, CBCA(CO)NH and 3D 15N-NOESY-HSQC (mixing time of 100 ms) spectra were recorded.

For titration experiments (protein-RNA or protein-protein), RNA or unlabeled proteins were titrated into 15N labelled proteins at 100-400 μM. The combined chemical shift difference was calculated according to (δH2+(δN/6.51)2)1/2, were δH and δN were the difference in the 1H and 15N chemical shifts respectively.

NMR data were processed with Topspin 3.1 (Bruker) and the analysis performed using Sparky (T. D. Goddard and D. G. Kneller, SPARKY 3, University of California San Francisco)

15 T1 and T2 relaxation experiments were carried out on the AVIII-750 MHz at 298 K of N- labelled BTG2(APRO) or BTG2(APRO) boxC mutant. Longitudinal relaxation time T1 were determined from a pseudo-3D experiment with twelve delay times ranging from 12- 3000 ms

(Bruker Pulse sequence hsqct1etf3gpsi3d.2). Transverse relaxation times T2 were obtained from 1H-15N correlation spectra with ten different delay times (17- 271 ms) (Bruker Pulse

15 15 sequence hsqct2etf3gpsi3d). N T1, N T2 and the corresponding standard deviation (SD) for each residue were determined by fitting the peak heights from each spectrum to a decaying exponential using Sparky (rh command) (Lee et al., 2015). Correlation times (τc) for each residue were calculated from the T1/T2 ratio (Gryk et al., 1998; Rossi et al., 2010). Molecular

90 PABPC1 RRM1-BTG2 interaction

weights were roughly estimated by τc/0.6, assuming a spherical particle at 298K (Rossi et al., 2010).

Isothermal titration calorimetry

Protein and RNA oligonucleotides were dialyzed or dissolved respectively in 20 mM Na2HPO4 (pH7), 100 mM NaCl and 10 mM β-mercaptoethanol or 20 mM MES (pH5.5.), 50 mM NaCl and 10 mM β-mercaptoethanol. ITC measurements were performed on a VP-ITC instrument (Microcal) mostly at 25°C. RRM12/RRM1-BTG2 titration experiments were also recorded at

20 and 15°C. To estimate the RRM12-poly(A) affinity, poly(A)11 was titrated into protein solution by 35 injections of 8 μl every 300s (at 307 rpm). Reported errors correspond to the standard deviation of at least two replicates.

Crystallization

1:1 Trx-BTG2:RRM12 was concentrated to 7 mg/ml in 20 mM Tris (pH8), 100 mM NaCl, 10% (w/v) Glycerol, 1 mM TCEP. Screens were set up by sitting drop vapor diffusion at 298K, in which 200 nl protein:RNA complex was mixed with 200 nl precipitant and suspended over 60 μl precipitant.

Analytical size exclusion

Analytical size exclusion was performed at 25°C in 20 mM Na2HPO4 (pH7), 100 mM NaCl, 1 mM DTT using a Superdex 75 10/30 GL column at a flow rate of 0.6 ml/min. Gel filtration standard contained Thyroglobulin (bovine) 670 kDa, γ-globulin (bovine) 158 kDa, Ovalbumin

(chicken) 44 kDa, Myoglobin (horse) 17 kDa and Vitamin B12 1.35 kDa (BioRad, Cat#: 151- 1901)

HADDOCK Modeling

The BTG2 (pdb code 3DJU) and PABPC1 RRM12 (pdb code 1CVJ) structures were used for docking calculations using HADDOCK2.2 Web Server; the easy interface (de Vries et al., 2007; Van Zundert et al., 2016). Models were calculated by using various combinations of the pdb 1CVJ. Either RRM12 in complex with RNA, single RRM12 or RRM1. Same for pdb 3DJU, where either BTG2 alone was used as an input, or in complex with CAF1 (derived from the superimposition with Tob1-CAF1 complex (pdb 2D5R) (Table 3.1). Residues used as input for HADDOCK are summarized in Table 3.1. The statistics of the top clusters, which are the most reliable according to HADDOCK and their Z-score, which indicates how many standard

91 PABPC1 RRM1-BTG2 interaction

deviations from the average this cluster is located in terms of score (the more negative the better), is shown in Table 3.2.

3.7 Acknowledgements

The authors thank Stier G. and Geerlof A. (EMBL) for providing the petGB1-1a plasmid and

Dr. Jürg Hunziger (Novartis) for the poly(A)11 RNA. We also thank F.F Damberger (ETH Zürich) for proof reading the manuscript. This work was supported by NCCR RNA and Disease.

4. Concluding remarks

Concluding remarks

4.1 RNA recognition motifs: boring? Not at all!

RNA binding proteins play a major role in PTGR. The RRM is the most abundant and very diverse protein domain (Gerstberger et al., 2014). Structural analyses of the free and RNA or protein bound forms highlight the extreme diversity in structure and function. The presence of N- and C-terminal linkers and tandem RRMs can extend the RNA binding surface and influence RNA specificity and affinity. Another level of variation is due to the presence of an additional binding surface within one RRM (Cléry et al., 2008; Muto and Yokoyama, 2012). The β-sheet surface binds the RNA, while structural elements on the opposite site are able to form protein-protein interactions with other RRMs or other domains. With an increasing amount of structural information, we are just starting to understand how the presence of such multifunctional domains influence RBP function and further studies are needed to obtain a full picture of their role in the cell.

In this thesis, we investigated two different proteins containing multitasking RRMs. HuR RRM3 is able to bind ARE motifs and form homodimers, which in turn affect RNA binding affinity. PABPC1 RRM1 simultaneously binds poly(A)-RNA and the BTG2 protein, which recruits the CAF1 deadenylase. Both RRMs use their hydrophobic α1-helix on the opposite site of the β-sheet surface. Also PTB RRM2 binds RNA as well as its co-repressor Raver1 in a groove composed of α1 and the α2/β4 loop (Rideau et al., 2006). These findings highlight the diversity of RRMs. By using a similar mechanism of simultaneous RNA and protein binding, a variety of different functions can be fulfilled.

4.2 The multitasking RRM3 of HuR

In this work, we presented the first crystal structure of HuR RRM3 in complex with c-fos ARE motifs, which suggested the (A/U)UU(A/U) as the preferred recognition sequence. We validated the binding in solution with additional biochemical and biophysical experiments. Various short ARE motifs (UUUUU, AUUUU, UUUAU, AUUUA or AUUAU) are recognized similarly by the RRM3. The differences in affinity arise from multiple register binding. In fact, the pentameric AUUUA motif found in many ARE mRNAs is recognized by a dynamic binding to two registers. RRM3 binds and exchanges between AUUU and UUUA, which leads to an increase in affinity and localization of the RRM at this motif. Dynamic binding of RNPs can play critical roles in the regulation of RNA binding (Schlundt et al., 2017). A similar dynamic interaction was observed for the polyU binding protein hnRNP C, U2AF and the ARE binding proteins TIA-1 and TIAR (Cieniková et al., 2014; Kim et al., 2011; Mackereth et al., 2011;

94 Concluding remarks

Wang et al., 2014). Overall, multiple register binding could be a mechanism how RNPs discriminate between different mRNAs. We investigated the role of multiple register binding with shorter RNA motifs, where only one RRM3 could bind. We still do not fully understand the interplay between RRM3 dimerization and multiple register binding. Additionally, it would be interesting to investigate how multiple register binding of both RRM3 and RRM12, contribute to HuR function and whether multiple register binding is involved in mRNA discrimination or competition with other ARE binding proteins.

Moreover, our crystal structure revealed the mechanism of RRM3 dimerization, mediated by the conserved W261 in α1. This resulted in the β-sheet surfaces being on opposite sides of the dimerization interface. Dimer formation could therefore induce RNA looping. Furthermore, dimerization of HuR RRM3 increased the affinity for longer sequences comprised of two RRM3 binding sites separated by longer linkers. There, RRM3 approached comparable affinities to RRM12. Similar findings were reported for members of the RNA binding protein with multiple splicing (RBPMS) family. The RRMs of those family members are also able to form homodimers, via residues in α1 and adjacent loop regions (Sagnol et al., 2014; Soufari and Mackereth, 2016; Teplova et al., 2016). RNA looping is also reported for tandem RRMs such as RRM34 of PTB, RRM12 of hnRNPA1 and hnRNPL RRM34 (Barraud and Allain, 2013; Beusch et al., 2017; Oberstrass et al., 2005; Vitali et al., 2006; Zhang et al., 2013). Therefore, dimerization of single RRMs could resemble the function of tandem RRMs to increase binding specificity and affinity and can be another strategy for target discrimination.

Solution experiments showed that HuR RRM3 dimerizes in a concentration dependent manner. The high RRM3 dimerization Kd of 30 μM in vitro raised the question if dimerization is physiological. Interestingly, there are indeed many possibilities in cells how HuR dimerization can be induced by local or spatial concentration changes. Spatial HuR concentration can differ during cell differentiation or proliferation, as was shown in the chicken ventricular zone of the spinal cord (Wakamatsu and Weston, 1997). HuR concentration is increased to maintain proliferation of neuronal precursor cells, while HuR concentration is increased during differentiation. They then raise again during maturation of neurons. Under various different stress signals, HuR translocates into stress granules, leading to the accumulation of HuRs (Bhattacharyya et al., 2006; Gallouzi et al., 2001). Lastly, HuR concentration is highly elevated in cancerous cells and tumors. Indeed, HuR is found to form aggregates in clinical brain tumor samples (Filippova et al., 2017).Therefore, RRM3-mediated HuR dimerization might function in cell areas with increased local concentrations, in membrane-less compartments such as stress granules or upon tumorigenesis. Generally, this indicates that protein dimerization can be a complex process, cell specific and dependent on the intracellular milieu (like redox state, pH, ionic strength and co-factors).

95 Concluding remarks

Finally, we highlighted that RRM3 dimerization and RNA binding are physiologically relevant. Despite the characterized role of HuR in translation upregulation, our cell based reporter assay indicated that RRM3 has an opposite, destabilizing role on AUUUA containing, but not on U-rich targets. This might be due to the competition with HuR RRM12 or other ARE binding proteins, but also due to protein-protein interactions. Interestingly, during muscle differentiation, HuR RRM3 is able to bind KH-type splicing regulatory protein (KSRP) to destabilize nucleophosmin (NPM) mRNA by recruiting two ribonucleases (Cammas et al., 2014). The dimeric but not monomeric RBPMS2, a RBPMS paralog, interacts with the translation elongation factor eEF2 (Sagnol et al., 2014). Similarly, HuR RRM3 monomers or dimers could differently influence binding to proteins which play a role in translation or decay. To fully understand how the regulation is achieved, it would be important to test if HuR RRM3 acts at the mRNA stability level, at the translation level and weather HuR RRM3 directly interacts with the translation or decay machinery components or other proteins that recruit the translation or decay machineries. Thus, additional qPCR experiments to measure mRNA levels and immunoprecipitation assays with HuR or HuR W261E could help to elucidate the involvement on mRNA stability or identify pottential binding partners.

Our study answered multiple key questions regarding the function of the HuR RRMs. The tandem RRM12 are separated from RRM3 by a ~ 60 residue long flexible hinge region. This hinge region contains the NLS, responsible for the nucleocytoplasmatic shuttling (Fan and Steitz, 1998a). It was shown before that the hinge region supports RRM3 in protein-protein interactions (Brennan et al., 2000), dimerization (Filippova et al., 2017; Toba and White, 2008), cooperative binding of multiple HuR molecules on long AREs (Fialcowitz-White et al., 2007) and counteracting miRNA mediated repression to promote miRISC release from target mRNAs (Kundu et al., 2012). Moreover, the hinge region contains multiple sites for posttranscriptional modifications (PTMs), such as phosphorylation, methylation and a caspase cleavage site (Grammatikakis et al., 2017). PTMs probably also contribute to the recognition of RNA, dimerization and binding to other proteins. Still, how exactly the hinge region functions in RNA binding and influences protein function is not known. One hypothesis could be that the hinge region contributes to low affinity hydrophobic interactions to help the accumulation and cooperative binding of multiple HuR molecules along ARE sequences as was shown before (Fialcowitz-White et al., 2007). Moreover such low affinity hydrophobic interactions could support the formation of larger protein assemblies in stress granules or other membrane-less granules, as reported for various proteins (Boeynaems et al., 2018; Protter and Parker, 2016) and in protein multimerization in cancer and tumor cells (Filippova et al., 2017). Therefore, the function of the hinge region, especially its influence on the RRMs would be of high interest to investigate.

96 Concluding remarks

4.3 The multitasking RRM1 of PABPC1 binds poly(A) and BTG2 to induce deadenylation

In our work, we showed that PABPC1 RRM1-BTG2(APRO) binding is primarily mediated by residues within PABPC1 RRM1 α1 and the α1-β2 loop on the opposite side of the RNA binding surface. PABPC1 binds other family members of the BTG/Tob family. However both Tob1 and Tob2 APRO domains contain different residues in the corresponding BoxC region and do not interact with PABPC1 RRM1 (Stupfler et al., 2016). Instead, Tob1/2 contain a C- terminal PAM2 domain that interacts with the C-terminal PABPC1 MLLE domain (Lim et al., 2006). While BTG2-PABPC1 binding brings the CAF1 deadenylase in a close proximity to the RNA, in the Tob-PABPC1 complex, CAF1 is further away and need an additional mechanism how to find the RNA. Therefore, the mechanism how Tob1/2 and BTG2 recruit CAF1 to the poly(A) tail, mediate CAF1 access to the poly(A) 3’end and induce deadenylation is different.

Additionally, PABPC1 is able to recruit a profusion of proteins, mostly by the PABP C- terminal MLLE domain (Kozlov et al., 2010, 2001). Only a few binding partners are known to directly bind the RRMs. The PABPC1 binds an eIF4G fragment with its RRM2 (Safaee et al., 2012). The translational regulators PABP-interacting protein 1 and 2 (Paip1 and Paip2) (Roy et al., 2002) bind PABPC1 RRM12 as well as the C-terminus. PABPC1 RRM2 is also able to recognize a motif within the C-terminus of another PABPC1 molecule to induce multimerization along poly(A) tails (Sawazaki et al., 2018). Our study contributes with another example how PABPC1 RRMs recognize other proteins and display versatility. This could be a widely used mechanism of poly-RRM containing proteins to fine tune their function and regulate assembly of macromolecular complexes. To our knowledge, it is not known yet whether RRM3 and RRM4 recruit other proteins. It would be interesting to determine if RRM34 contribute to the regulation by RRM12.

Our data indicated that the free or poly(A) RNA bound PABPC1 RRM1 does not show a difference in binding BTG2. Also RRM12 binds poly(A) RNA with the same affinity in the free form or when bound to BTG2, indicating no signs of cooperativity. This is in contrast to the RRM12-poly(A) RNA-eIF4G complex, where Poly(A) RNA enhances the interaction of RRM12 and the eIF4G fragment through inter-domain allostery (Safaee et al., 2012). Interestingly, the allostery arises from the flexible linkers that connect the RRM domains. Without the RNA, these linkers are flexible, adopt different orientations and show an overall compact conformation of RRM12. Upon RNA binding, the linkers form α helices and induce a more extended confirmation, so that the poly(A) RNA is able to bind the extended RRM12 β-sheet surfaces. This seems not to affect RRM1-BTG2 binding. The interdomain allostery or the lack

97 Concluding remarks of allostery exhibited by PABP could be a general mechanism how RNA binding domains regulate the recruitment and assembly of other proteins.

Our HADDOCK modeling approach suggested two different models for the BTG2-RRM12 interaction. Both models show BTG2 boxC region binding to PABPC1 RRM1 and thereby positioning the 3’-end of the poly(A) RNA in a close proximity to the CAF1 active site. Despite the assumption of a stable ternary BTG2-PABPC1 RRM1-poly(A) RNA complex, we cannot exclude the possibility that the PABPC1 RRM1 binding to BTG2 is dynamic and adopts multiple confirmations. It could be that our two models resemble two potential confirmations. Our models provide a hypothesis how both proteins are oriented towards each other. However, both models need to be further validated. PRE or EPR can be used to derive additional distances. SAXS experiments can be performed to validate the positioning of the domains with respect to each other. Lastly, mutagenesis will identify key residues involved in binding.

Despite our analysis, which showed the positioning of the poly(A) 3’end close to the CAF1 catalytic pocket, the question remains how poly(A) is transported into the CAF1 catalytic center. One hypothesis could be the contribution of additional binding partners which drive the translocation process. On the other side, dynamic movement of the poly(A) RNA within RRM12 due to multiple register binding or sliding might be sufficient for CAF1 to bind the closest nucleotide and “pull” the RNA into the enzymatic pocket. Additional investigations should be performed to investigate the dynamic nature of these binding partners. It would be interesting to determine, if conformational changes contribute to that process.

Finally, the superimposition of our HADDOCK models with Tob1-CAF1 and PABPC1 RRM12-poly(A) revealed no clashes of CAF1 and RRM12. Interestingly, the structure of the CCR4-NOT-CAF1 complex shows that CCR4 and NOT bind CAF1 at different sites leaving the catalytic center accessible (Basquin et al., 2012). This leaves enough space for BTG2- PABPC1 RRM12 to bind to CAF1 without inducing steric hindrance. Therefore, our models are in a good agreement and could explain how BTG2 links PABPC1-polyA RNA to the multisubunit CCR4-NOT deadenylase complex and induce deadenylation. Additional investigations need to be performed to determine whether there is a crosstalk between eIF4G bound to RRM2 and CCR4-NOT deadenylase complex. Moreover, it is still unknown whether the BTG2-CAF1 complex acts cooperatively to bind RRM12-poly(A), if the binding induces a conformational rearrangement or even a release of binding partners. Overall, further investigation of these binding partners, whether they are in competition or allosterically coupled, will shed light on key steps in in mRNA stability, translation and decay.

98 Concluding remarks

4.4 Towards understanding mRNA stability regulators

One of our goals is to understand the dynamic and complex networks of mRNPs and how these combinatorial interactions integrate to control. For that, we need to continue investigating several aspects such as protein and mRNA concentrations and their spatial distributions in cells, affinity and specificity to RNAs and binding partners, the contribution of flexible protein linkers to structure and function, structure of RNAs and how they affect the binding of proteins. Therefore, our structural and functional work of the mRNA stability regulators HuR and PABPC1 provide some new key pieces in the large and complex mRNP puzzle.

5. Appendix

101 Appendix

5.1 A2. Supplementary Tables Chapter 2

Table A2. 1. X-Ray data collection and refinement statistics. Statistics for the highest-resolution shell are shown in parentheses.

#1 Wavelength (Ã…) 1 Resolution range (Ã…) 77.95 - 1.9 (1.968 - 1.9) Space group C 1 2 1 Unit cell 151.287 40.258 106.496 90 132.95 90 Total reflections 120688 (10196) Unique reflections 35681 (3148) Multiplicity 3.4 (3.2) Completeness (%) 95.10 (81.19) Mean I/sigma(I) 15.23 (2.52) Wilson B-factor 31.76 R-merge 0.04556 (0.6683) R-meas 0.05447 CC1/2 0.999 (0.854) CC* 1 (0.96) R-work 0.2007 (0.3761) R-free 0.2312 (0.4525) Number of non-hydrogen atoms 2839 macromolecules 2648 water 191 Protein residues 326 RMS(bonds) 0.011 RMS(angles) 1.23 Ramachandran favored (%) 98 Ramachandran outliers (%) 0.66 Clashscore 3.5 Average B-factor 39.9 macromolecules 39.9 solvent 46.60

102 Appendix

Table A2. 2. Protein chemical shift assignments of HuR RRM12. Measurements were performed in 20 mM

Na2HPO4 (pH7), 100 mM NaCl, 1 mM DTT, 298K. CA, CB, CO, HN and N chemical shifts are in ppm.

CA CB CO HN N

residue Asn3 53.4 38.88 175.5 8.552 120.4 Gly4 45.36 - 173.9 8.431 109.1 Tyr5 58.13 38.63 175.9 7.976 120 Glu6 56.77 30.18 176 8.31 122 Asp7 54.52 40.95 176.1 8.132 120.7 His8 56.33 29.93 175.2 8.114 119 Met9 55.75 32.67 175.2 8.092 120.7 Ala10 52.72 19.22 177.8 8.076 124.3 Glu11 57 30.24 176.4 8.282 119.7 Asp12 54.33 41.09 176.4 8.266 120.6 Cys13 58.68 27.71 174.9 8.196 119.4 Arg14 56.71 30.57 176.9 8.234 122.5 Gly15 45.36 - 173.8 8.282 108.9 Asp16 54.28 41.08 173.8 8.194 120.3 Ile17 61.44 38.43 176.7 7.857 119.7 Gly18 45.5 - 174.3 8.41 111.3 Arg19 56.78 31.51 176.3 7.999 120.1 Thr20 61.98 69.85 174.2 8.095 108.8 Asn21 53.78 39.94 175 7.796 120.4 Leu22 53.56 44.36 176.3 9.659 127.1 Ile23 58.93 40.99 173.2 9.147 121.1 Val24 59.09 33.69 173.8 8.624 128.1 Asn25 51.57 42.95 173.4 9.11 123.6 Tyr26 57.12 34.53 174.2 8.313 114.5 Leu27 52.54 41.48 175.5 7.862 114.1

Pro28

Gln29 Asn30 52.36 37.76 175.5 8.194 112.3 Met31 56.56 33.86 176.6 7.21 121.6 Thr32 60.44 71.68 175.3 8.537 118.6 Gln33 60.24 29.61 177.2 9.164 121.5 Asp34 57.4 40.55 179 8.331 118 Glu35 59.04 30.39 179.5 7.611 121.4 Leu36 58.52 41.63 177.6 8.148 123 Arg37 60.58 29.69 179.2 8.263 118.1 Ser38 62.09 - 179.3 8.479 116 Leu39 58.06 41.59 179.1 8.02 122.7 Phe40 61.79 37.8 179.9 7.861 116.7 Ser41 60.87 62.83 174.5 8.92 121.4 Ser42 60.77 62.97 174.9 7.279 115.7 Ile43 60.24 37.62 175.5 7.057 119.9 Gly44 44.6 - 171 7.103 130.9

103 Appendix

Glu45 57.96 29.87 176.6 8.177 117.1 Val46 62.16 33.78 175.7 8.9 129.1 Glu47 58.74 29.93 176.3 9.162 130.8 Ser48 57.67 64.97 171.6 7.568 109.1 Ala49 50.95 20.77 174.8 8.07 123 Lys50 54.36 36.15 173.9 8.911 122.5 Leu51 53.86 44.89 175.6 8.668 126.6 Ile52 59.5 33.97 174.3 8.16 127.4 Arg53 53.82 33.66 175.8 8.023 124.7 Asp54 54.41 43.19 177.5 8.516 121.9 Lys55 58.34 32.77 176.7 8.672 125.6 Val56 64.52 32.54 177 8.026 119.2 Ala57 52.45 20.24 179.3 8.847 122.2 Gly58 46.09 20.12 174 7.656 131.4 His59 55.13 31.21 175 7.411 117.5 Ser60 58.35 64.28 176.4 8.797 114.2 Leu61 54.64 41.3 - 8.991 125.5 Gly62 46.14 - 172 9.148 107.8 Tyr63 54.93 40.66 172.2 7.268 113.1 Gly64 45.03 - 169.9 8.546 131.1 Phe65 56.02 42.73 175.7 8.708 116.2 Val66 61.34 35.64 172.4 8.587 122.8 Asn67 51.27 40.34 - 8.745 125.3 Tyr68 60.58 40.67 174.6 8.547 125.9 Val69 66.44 31.76 176.9 7.738 118.3 Thr70 59.52 71.56 - 8.94 109.7 Ala71 54.6 18.15 180.6 8.752 126.9 Lys72 58.75 31.85 179.7 8.349 118.7 Asp73 56.88 39.64 176.4 7.291 119.9 Ala74 55.39 18.02 178.4 6.968 122.2 Glu75 59.39 29.86 178.6 7.664 116.8 Arg76 59.26 29.98 179.4 7.664 119.2 Ala77 56.01 19.16 180 8.481 124.8 Ile78 66.34 38.13 178 7.778 119.2 Asn79 55.88 38.75 176.9 7.643 114.6 Thr80 64.96 70.44 175 8.092 112.1 Leu81 55.14 42.46 177 8.587 119.2 Asn82 56.51 39.36 176.9 7.382 114.9 Gly83 45.65 - 173.3 8.448 116.5 Leu84 55.95 43.24 175.4 7.404 123.5 Arg85 54.91 29.95 175.6 8.161 125.9 Leu86 53.55 44.67 175.1 8.577 128.5 Gln87 57.75 26.7 175.9 9.028 120.9 Ser88 59.46 63.33 - 8.236 112.7 Lys89 54.2 34.3 175.5 8.103 122.2 Thr90 61.3 69.42 175.1 8.05 117.5

104 Appendix

Ile91 61.81 39.04 175.5 8.496 121.3 Lys92 54.05 36.88 174.9 8.291 123.5 Val93 61.33 34.58 173.6 8.724 126.8 Ser94 56.54 65.75 172.9 8.526 118.9 Tyr95 60.87 38.39 176.9 8.384 119.7 Ala96 52.61 19.48 177.4 8.703 125.2 Arg97 53.78 30.24 174 8.756 121.1

Pro98 Ser99 58.89 63.6 174.9 8.416 116.5 Ser100 - 63.37 174.5 8.264 117 Glu101 56.74 30.22 176.4 8.146 122.1 Val102 62.86 32.38 176.1 7.917 121.1 Ile103 61.22 38.2 176 7.961 123.6 Lys104 55.96 32.98 175.9 8.192 124.4 Asp105 54.37 41.84 175.9 7.883 121.9 Ala106 50.82 18.91 175.1 8.41 124.5 Asn107 52.85 40.59 - 8.001 117.9 Leu108 53.55 45.56 - 9.555 125 Tyr109 56.69 41 173.3 9.141 122.6 Ile110 58.29 40.87 174 8.398 128.7 Ser111 56.41 66.09 173.6 8.69 117.7 Gly112 45.32 - 174.8 8.088 108.9 Leu113 53.34 40.96 174.8 8.394 120.1

Pro114 Arg115 57.81 29.07 175.1 9.07 124.1 Thr116 61.2 68.84 175.7 7.127 130.1 Met117 57.24 33.81 176.9 7.093 119.8 Thr118 60.12 71.95 175 9.242 116.7 Gln119 60.17 27.91 177.1 8.945 120.1 Lys120 58.79 32.12 177.9 8.046 118.9 Asp121 57.51 40.91 179.5 7.4 118.5 Val122 67.2 31.55 177.8 8.18 121.5 Glu123 60.62 28.71 179.1 8.325 120.3 Asp124 57.5 40.22 178.9 8.731 119.2 Met125 58.83 33.26 177.1 7.767 119.7 Phe126 60.59 40.4 176.8 7.909 113.6 Ser127 61.7 62.9 176 8.449 117.7 Arg128 57.2 28.95 176.7 7.084 117.6 Phe129 59.02 39.19 174.5 7.354 116.7 Gly130 44.93 - 171 7.365 130.3 Arg131 56.24 31.11 174.9 8.22 119.8 Ile132 61.64 38.13 176.7 8.494 127.7 Ile133 61.07 37.43 176.3 8.928 127.8 Asn134 53.31 42.22 172.2 7.346 114.4 Ser135 57.52 66.84 172.9 8.171 115.3 Arg136 56.01 34.02 174.3 8.539 117.9

105 Appendix

Val137 61.5 33.69 175.8 8.618 124 Leu138 55.29 41.62 175.8 8.568 128.3 Val139 59.77 34.93 176 8.258 118.6 Asp140 53.79 42.27 177.2 8.61 125.3 Gln141 58.23 28.57 177 9.115 126.8 Thr142 65.03 68.77 176.1 8.627 113.5 Thr143 61.89 70.94 176.4 7.776 109.6 Gly144 45.62 - 173.6 8.448 110.8 Leu145 53.84 43.07 176.9 7.718 119.9 Ser146 58 64.15 176.2 8.547 115 Arg147 56.11 31.16 177.3 8.839 123.3 Gly148 45.97 - 171.4 9.529 109.7 Val149 58.75 35.37 173.8 6.403 115.6 Ala150 49.58 24.01 173.7 9 126.4 Phe151 55.65 42.27 176.2 8.522 116.3 Ile152 59.72 42.24 172.8 8.135 120 Arg153 54.6 32.99 175.8 8.41 127.2 Phe154 58.54 42.12 174.6 8.308 126.7 Asp155 56.8 43.61 176.2 8.697 120.2 Lys156 53.91 36 177.2 8.332 113.2 Arg157 59.97 41.31 177.8 9.025 125.5 Ser158 61.43 - - 8.576 112.6 Glu159 58.27 29.21 177 6.825 124 Ala160 54.5 18.46 178.9 6.705 119.5 Glu161 59.69 29.66 179.2 8.545 115.2 Glu162 59.07 29.02 178.2 7.404 120.2 Ala163 54.44 18.61 179.3 7.227 123 Ile164 65.92 38.62 178.3 7.59 117.6 Thr165 66.16 68.73 176.8 8.012 114.7 Ser166 61.64 64.18 175.2 7.887 113.5 Phe167 56.93 41.78 176.9 8.039 116.9 Asn168 57.14 38.78 176.8 8.612 119 Gly169 45.7 - 173.5 9.291 118.1 His170 56.35 31.37 173.8 7.973 121.5 Lys171 52.79 32.68 173 7.944 126.4

Pro172

Pro173

Gly174

Ser175

Ser176

Glu177

Pro178 Ile179 60.55 39.39 176.4 8.497 118.4 Thr180 61.51 70.6 173.1 8.744 121.2 Val181 61.22 34.78 173.4 8.624 124.7 Lys182 54.02 36.25 174.7 8.658 122.6

106 Appendix

Phe183 60.88 38.92 177.1 8.522 119.7 Ala184 52.25 19.74 176.9 8.768 125.3 Ala185 52.15 19.48 176.9 8.514 122.9 Asn186 51.17 38.9 176.9 8.302 118.7

107 Appendix

Table A2. 3. Protein chemical shift assignments of HuR RRM3 (left) and HuR W261E (right). Measurements were performed in 20 mM Na2HPO4 (pH7), 100 mM NaCl, 1 mM DTT, 298K. CA, CB, HN and N chemical shifts are in ppm.

RRM3 W261E

CA CB HN N CA CB HN N

residue residue

S242 58.23 63.88 61.96 - S242 - - 7.85 127.3

G243 44.72 - 7.844 109.6 G243 45.34 - 8.388 110.5

W244 56.56 30.52 8.907 122.5 W244 58.22 63.9 - -

C245 59.03 28.09 8.919 124.1 C245 44.75 - 7.877 109.5

I246 61.2 40.39 9.745 132.4 I246 56.76 30.47 8.903 122.4

F247 55.78 41.75 9.423 127.7 F247 59.26 28.04 9.043 124.4

I248 60.54 41.17 8.218 125.5 I248 61.38 40.47 9.861 132.4

Y249 56.37 41.93 9.007 124.4 Y249 55.84 41.78 9.521 127.7

N250 54.03 36.54 7.495 117.2 N250 60.63 41.34 8.372 125.6

L251 54.41 44.33 7.14 113.6 L251 56.38 41.82 9.087 124.5

G252 44.16 - 8.337 107.8 G252 54.19 36.55 7.534 117.4

Q253 58.23 28.94 8.45 116.3 Q253 54.31 44.34 7.227 113.5

D254 53.1 40.32 8.502 116.7 D254 44.09 - 8.454 108

A255 53.54 20.28 7.096 123.3 A255 58.22 28.89 8.524 116.2

D256 52.1 43.93 7 116.5 D256 53.21 40.34 8.6 116.6

E257 61.75 28.27 9.575 119 E257 53.63 20.27 7.164 123.2

G258 47.78 - 8.488 106.4 G258 52.23 43.81 6.931 116

I259 62.65 37.6 7.578 120.4 I259 61.96 28.22 9.821 119.7

L260 55.15 - 9.366 119.5 L260 47.2 - 8.369 107.4

W261 55.68 - 8.456 119.9 E261 63.17 37.61 7.807 122.1

Q262 Q262 58.05 41.49 7.219 119.1

M263 M263 61.02 29.61 8.672 120.1

F264 F264 58.33 28.78 8.043 117.6

G265 G265 58.38 34.55 7.842 117.1

P266 P266 61.25 40.42 8.064 113.9

F267 F267 47.97 - 8.167 107.6

G268 G268 64.81 31.78 - -

A269 A269 59.73 39.7 14.4 112.7

V270 V270 44.39 - 7.514 105.2

T271 61.41 69.6 8.337 118.2 T271 53.7 18.76 7.779 120.3

N272 53.93 41.98 7.214 118.2 N272 60.94 34.31 8.197 124.5

V273 60.76 38.68 7.815 119.7 V273 61.37 69.64 8.856 116.9

K274 55.34 36.33 8.645 124.3 K274 54.2 41.37 7.615 118

V275 61.55 33.45 8.535 123.1 V275 61.49 34.53 8.142 120.5

I276 59.43 33.87 7.807 126.5 I276 55.44 36.27 8.858 125.3

R277 54.42 33.57 8.57 127.1 R277 61.62 33.4 8.572 123.2

D278 53.86 43.08 8.397 121.8 D278 59.55 34.02 7.868 126.5

F279 60.14 39 17.67 105.8 F279 54.42 33.58 8.624 127.2

N280 55.38 39.15 8.61 115.9 N280 54.1 43.08 8.422 121.7

T281 61.66 71 8.101 107.9 T281 60.1 39.17 8.869 126.9

108 Appendix

N282 54.67 38.58 8.299 117.2 N282 55.44 39.17 8.66 115.9

K283 54.7 33.54 7.656 117 K283 61.65 71 8.134 107.9

C284 59.2 27.05 8.811 121.7 C284 54.64 38.6 8.343 117.2

K285 57.12 33.22 9.013 128.6 K285 54.6 33.57 7.696 116.9

G286 44.05 - 8.409 107.2 G286 59.32 27 8.897 121.7

F287 54.42 42.32 6.727 115.6 F287 57.17 33.27 9.121 128.7

G288 44.99 - 8.54 105.7 G288 44.05 - 8.483 107.2

F289 56.08 43.66 8.887 114.6 F289 54.49 42.22 6.828 115.6

V290 60.22 36.04 8.838 120.9 V290 45.14 - 8.658 105.8

T291 61.37 69.75 8.543 121.3 T291 56.2 43.7 9.057 114.8

M292 54.97 37.59 8.756 124.4 M292 60.34 35.92 8.922 119.2

T293 66.18 69.82 7.2 117.4 T293 61.63 69.72 8.746 121

N294 52.74 39.96 9.875 120.1 N294 54.94 37.5 8.916 124.9

Y295 62.56 38.13 9.057 128.4 Y295 65.89 69.89 7.592 116.6

E296 59.79 28.7 9.034 115.5 E296 52.67 40.33 9.411 119.5

E297 58.9 29.07 6.985 119.5 E297 62.54 38.14 9.062 128.2

A298 54.38 18.02 7.905 122.8 A298 59.86 28.63 9.015 116

A299 54.81 17.25 8.669 118.3 A299 58.96 29.04 7.177 119.6

M300 57.8 31.77 7.469 119.2 M300 54.42 18.2 7.932 122.5

A301 54.7 17.79 7.921 125.5 A301 54.76 17.13 8.74 118.4

I302 65.82 38.13 7.993 117.3 I302 57.82 32.17 7.596 119.1

A303 54.48 18.02 7.573 118.9 A303 54.69 18.02 8.02 125.2

S304 60.87 64.56 7.43 110.8 S304 65.84 38.2 8.098 117.4

L305 55.15 43.12 8.045 118.9 L305 54.45 18.04 7.614 118.8

N306 56.43 38.02 8.018 117.5 N306 60.98 64.52 7.46 110.8

G307 45.43 - 8.515 117.4 G307 55.17 43.13 8.138 118.8

Y308 58.83 40 7.896 123.4 Y308 56.38 38.06 8.095 117.5

R309 55.2 30 7.508 127.3 R309 45.45 - 8.627 117.5

L310 53.06 43.5 8.476 130.2 L310 58.8 40.16 7.98 123.5

G311 47.3 - 8.902 118.2 G311 55.17 30 7.561 127.3

D312 53.88 40.8 8.65 125.4 D312 53.07 43.54 8.535 130.2

K313 54.35 35.07 7.644 119.7 K313 47.24 - 8.987 118.3

I314 59.73 37.18 8.052 120.8 I314 53.93 40.85 8.727 125.2

L315 55.65 42.9 9.164 128.6 L315 54.33 35.11 7.704 119.7

Q316 54.23 30.21 8.109 121.7 Q316 59.74 37.13 8.101 120.7

V317 61 34.89 8.768 126 V317 55.74 42.94 9.225 128.6

S318 57.46 65.89 8.812 118.2 S318 54.29 30.3 8.196 121.7

F319 60.27 39.72 8.77 120.2 F319 61.02 35 8.85 125.8

K320 56.98 33.61 8.952 123.5 K320 57.44 65.87 8.918 118.1

T321 61.41 70.32 8.328 117.4 T321 60.3 39.64 8.849 120.2

N322

K323

S324

H325

K326 57.73 - 7.766 127.3 K326 - - 7.856 127.3

109 Appendix

5.2 A2. Supplementary Figures Chapter 2

Figure A2. 1. HuR RRM3 binds c-fos ARE with a higher affinity than polyA. (A) 1H-15N HSQC spectra of free

RRM3 (blue) overlaid with 6 equivalents (eq.) EDTA (red), 12 eq. MgCl2 (yellow), 6 eq. ATP (green), 12 eq. MgCl2 (cyan),and 12 eq. ATP (magenta). (B) 1H-15N HSQC spectra of free RRM3 (blue) overlaid with 1:1.5 RRM3:c-fos (cyan) and 1:9 RRM3:polyA (magenta). Negative peaks corresponding to the amides of arginine side chains in the free and the RNA bound form in green. Examples of different exchange regimes are indicated by a box and shown in a close-up view on the right.

110 Appendix

Figure A2. 2. The crystal structure of HuR RRM3 in complex with RNA contains 4 RRM-RNA molecules in one asymmetric unit. (A) Cartoon representation of the RRM-RNA molecules (RRMs chain A to D and RNA chain E-H) with chains A/C in green and B/D in cyan. RNA is shown in stick representation and the heavy atoms are indicated in yellow (carbon of uridines), magenta (carbon of adenines), red (oxygen), blue (nitrogen) and orange (phosphorus). Top panel: Electron density (sigma of 1.2) for all nucleotides shows a 5’-AUUUU-3’, 5’-UUU-3’ and 5’-UUUAU 3’ occupancy in chain A, B/D and C respectively. Bottom panel: RNA chains, pocket number (abbreviated with p. #) and the nucleotides are shown. (B) Left: Overlay of all protein chains; chains A/C in green and B:D in cyan. Right: Overlay of all RNA chains reveals differences in chain E and F/G/H due to the additional uracil in chain E. U5 interacts with another symmetry related molecule, which pulls it away from the chain A. The nucleotide in pocket 4 has therefore a different conformation, while nucleotides in pocket 2 and 3 overlay well. The RRM is shown in grey and the four RNA chains in different colors. (C) Superposition of RRM3 chain A/E and HuR RRM1 in complex with c-fos ARE 11mer (pdb: 4ED5) showing the binding pocket of the adenine at this position is conserved within this RRMs. RRM3:adenine is shown in green: magenta and RRM1:adenine in light pink. (D) The additional uracil in chain E interacts with a G252 from a symmetry related molecule. The interaction is stabilized by multiple water molecules. (E) Snapshots from molecular dynamics (MD) simulations show a network of protein/RNA

111 Appendix

H-bond interactions formed by uracil (left) or adenine in syn or anti conformation (right) in the first binding pocket. Due to fast thermal sampling of molecular vibrations, the MD snapshots may show nonplanar instantaneous nucleobase geometries. This is seen for the syn- and anti-adenine in pocket 1. (F) Snapshots from molecular dynamics (MD) simulations show a network of protein/RNA H-bond interactions formed by uracil (left) or adenine (right) in the fourth binding pocket. (G) 2D NOESY spectrum measured at 700 MHz of a 1:1 15N RRM3:AUUAUU complex in D2O. Area including the H5-H6 cross-peaks is shown in a close-up view on the right. H5-H6 cross- peaks are highlighted by a box and a strong H8 to H1’ NOE corresponding to a syn-adenine by an arrow.

112 Appendix

Figure A2. 3. RRM3 binds the (A/U)UU(A/U) motif in solution. Top: Residues with chemical shift differences larger for AUUUU and AUUAU.then for UUUUU are highlighted in green on chain A. Pocket 1 and 4 is indicated by a violet circle. The loop between β2-β3 is indicated by a green circle.Bottom: Comparison of chemical-shift perturbations of free RRM3 and in complex with various RNA motifs at a 1:2.5 molar ratio of RRM3:RNA. The combined chemical-shift perturbations were calculated using ∆δ = [(δHN)2+δN/6.51)2]1/2. In each panel, the U- rich motif is compared to various other motifs. Residues in pocket 1 and 4 or in the β2-β3 loop are highlighted in violet and green respectively.

113 Appendix

Figure A2. 4. RRM3 binds the (A/U)UU(A/U) motif in solution. (A) ITC measurements of RRM3 titrated with UUUUU, AUUUU, UUUAUU or AUUAU. Measurements were repeated at least two times. Top: ITC raw data as a function of time, bottom: heat release as a function of molar RNA:protein ratio. Red line represents the fit, obtained by the ITC-adapted Origin 7 Software with a one binding site model. Thermodynamic parameters are summarized in table 2. (B) Comparison of H5–H6 correlations in 2D TOCSY spectra at 750 MHz of RRM3 in complex with various 5mer RNAs (blue: 5’-UUUUU-3’, red: 5’-AUUUU-3’ orange: 5’-AUUAU-3’).

114 Appendix

Figure A2. 5. RRM3 dimerizes in solution via the conserved Trp261 in a concentration dependent manner. (A) The sequence alignment of RRM3 of HuR, HuB, HuC, HuD and Drosophila ELAV shows, that the RNP1 and RNP2, highlighted by the black box, and the dimerization interface (green box) are conserved within the Hu family members. The secondary structure is detailed above. (B) Left: Overlay of 1H-15N HSQC spectra of RRM3 at various concentrations (350, 113, 40, 20, 5 μM). Right: Overlay of 1H-15N HSQC spectra of W261E at various concentrations (320, 131, 55, 27, 14 μM) shows chemical shift perturbations for RRM3 but no changes for W261E. Residues showing chemical shift perturbations for RRM3 but not for W261E are highlighted by a black box. (C) 13 Overlay of the Cα chemical shifts ∆δ ( Cαexperimental- Cαrandom coil) is consistent with the secondary structure observed in the crystal structure indicated above in the diagram for both RRM3 and W261E. The blue starts indicate amino acids located within the dimerization helix α1 that are missing in the 1H-15N HSQC of RRM3. (D) Rotational correlation times τc obtained from non-overlapping amide resonances at RRM3 concentrations of 391 μM (blue), 300 μM (green), 160 μM (red) and 80 μM (white) and W216E at 311 μM (black), in comparison with RRM12 at 311 μM (yellow).

115 Appendix

Figure A2. 6. Interdependence of RRM3 dimerization and RNA binding in ARE reporter regulation. (A) Overlay of 1H-15N HSQC spectra of RRM3 FY and RRM3 FY in complex with c-fos ARE 11mer. The RRM3 FY mutant is folded and inactive to bind the RNA. (B) Comparison of the first 60 nucleotides of the COX-2 3’UTR and the AtoU mut sequence, where all adenines are mutated uracils. (C) Dual luciferase reporter assay for evaluating the effect of HuR RR mutant in Cox-2 ARE regulation. WT HuR and the RR mutant were co-expressed with the luciferase reporter containing Cox2-1-60 in Huh7 cells. (D) Western blot analysis for protein expression levels of all constructs in Huh7 cells. (E) Western blot analysis to evaluate the HuR expression levels upon by si-RNA knockdown in C3H/10T1/2 cells. (F) Dual luciferase reporter assay for evaluating the effect of HuR mutants in Cox- 2 ARE reporter regulation upon knockdown of endogenous HuR. After Si-HuR transfection, WT HuR and mutants were co-expressed with the luciferase reporter containing Cox2-1-60 in C3H/10T1/2 cells, multipotent cells isolated from C3H mouse embryo cells. The RL/FL luminescence was normalized to the value of mock transfection. The mean values ± sd from at least three independent experiments are shown. P values were determined by the Student’s t-test (Two-Sample Assuming Equal Variances). *P<0.05, **P<0.01, ***P<0.001. (G) Western blot analysis to evaluate the protein expression levels of all transfected HuR constructs with (Si-HuR) and without (Scr) knockdown of endogenous HuR.

116 Appendix

5.3 A2. Supplementary Materials and Methods Chapter 2

Expression and Purification of recombinant proteins

Recombinant human 6-His-tagged –GB1-fusion proteins (RRM3, RRM3 W261E, RRM3 FY) were over expressed in BL21(DE3) cells (Novagen). Cells were grown in LB rich or M9

15 15 13 minimal media supplemented with NH4Cl or NH4Cl and C-glucose at 37°C until OD600 reached 0.6-0.8, temperature was reduced to 20°C and the expression induced with 0.5 mM isopropyl β-D-thiogalactoside (IPTG). Cells were harvested by centrifugation at 4°C, 20 min at 6000rpm after 19- 22 h. Cell pellet was either stored at -20°C or lysed with a microfluidizer at 75 PSI (0.52 MPa) in 20 mM Tris- HCl (pH8), 800 mM NaCl, 2 mM MgSO4, 5 mM CaCl2, 10 mM imidazole, 10 mM β-mercaptoethanol, 0.25 mg/mL lysozyme, 10 µg/mL DNase, one complete EDTA free protease inhibitor tabled (Roche) in a total volume of 50 ml. Lysate was cleared by centrifugation at 4°C, 40 min and 17000 rpm and was applied onto a 5 ml HiTrap Chelating HP column (GE Healthcare) charged with nickel and purified on a AKTA prime system (washed with 20mM Tris- HCl pH8, 800 mM NaCl, 10 mM imodazole and 10 mM β- mercaptoethanol and eluted with 50% same buffer but 500 mM imidazole. Pooled fractions were dialysed against 3 L buffer (20 mM Na2HPO4 (pH7), 100 mM NaCl, 1 mM EDTA, 1 mM DTT and without imidazole) in the presence of an in house made TEV protease (1:50 ratio mg-TEV: mg-RRM3) at RT overnight to remove the 6-his-GB1-tag. Additional dialysis against 2 L of same buffer but no EDTA was performed for 4h. The protein was further purified by reloading the sample on the HiTrap Chelating HP column to remove the 6-his-GB1-tag, overnight dialysis as above, followed by size exclusion chromatography using the Superdex75 prep grade size exclusion column (GE Healthcare) in 20 mM Na2HPO4 (pH7), 100mM NaCl, 1mM DTT. The protein was stored in size exclusion/NMR buffer at -80C.

Crystal Structure Determination

While the RNA electron density clearly indicates uridines bound in pockets 2 and 3 in all four chains, the electron density for the nucleotide in chain E and G in pocket 1 (β2-β3 loop) and 4 (β3) is not well defined. Therefore, it was unclear whether to build an adenine or uridine at this position. Refinements were performed of all possible combinations: adenine or uridine in pocket 1 or 4, including anti and syn conformations. For all, the difference density map showed always a negative density in pocket 1 and 4, which is explained by the occupancy of both adenines and uridines in these pockets. Therefore, the occupancy factor was changed to 0.5 for the

117 Appendix nucleotides in pocket 1 and 4, including the side chain Q253 in pocket 1 during the last rounds of refinement. All structures with multiple combinations show only little differences in R-values. For the final structure, based on the slightly better R values and quality of the electron density, shown as 2Fo−Fc density contoured at a sigma level of 1.2 (Supplementary Figure A2.2 A), the conserved pocket 1 (to HuR RRM1) and the MD simulation, an adenine was placed at pocket 1 in chain E and pocket 4 in chain G.

Molecular Dynamics Simulations

In all simulations of the HuR RRM3 protein/RNA complex, a stabilizing HBfix potential of 1 kcal/mol was applied in the 3-4 Å donor-acceptor range to the U2(O4)/Q316(NE2), U2(N3)/Q316(OE1), and U3(O2)/T321(N) H-bonds. The U2 and U3 indicate uracils bound in binding pockets 2 and 3, respectively. Due to their large dynamics, the HBfix was not applied for the nucleotides in binding pockets 1 and 4. HBfix is a gentle short-range structure-specific tuning of the force field designed as a compensation for polarization effects that contribute to H-bonding but are neglected by the force field. A more detailed justification of this approach can be found in (Šponer et al., 2018, 2017).

Western blots & Luciferase Assay in C3H/10T1/2 cells

After 12% SDS-PAGE, proteins were transferred to a PVDF transfer membrane. The membrane was blocked in TBST- 5% milk blotting powder for 1 h at RT. Afterwards, the membrane was incubate with mouse Monoclonal ANTI-FLAG M2 antibody (Sigma #F3165, 1:2500 in TBST- 5% milk blotting powder) or mouse Monoclonal Anti- α-Tubulin antibody (Sigma #041M4798, 1:20000 in TBST- 5% milk blotting powder) overnight at 4°C. After secondary antibody incubation with Anti-Mouse IgG (whole molecule)-Peroxidase antibody (Sigma #A9044, in 1:50000 in TBST- 5% milk blotting powder) for 1h at RT, western was detected by using the Amersham ECL Prime Western Blotting Detection Reagent (GE Healthcare).

C3H/10T1/2 cells, multipotent cells isolated from C3H mouse embryo cells, were cultured in DMEM with L-glutamine (Corning), 10% FBS, 1% penicillin/ streptomycin (Life Technologies), 1 mM sodium pyruvate (Thermo Scientific) at 37°C and 5% CO2. On day one, 12-well plates were seeded with 2*104 cells/ml. On the next day, 25 nM of scrambled-RNA (control) or si- HuR RNA (5’-AUGUGAAAGUGAUUCGUGAtt-3’, manufactured by Ambion Life

118 Appendix

Technologies) was transfected using TransIT-siQUEST (Mirus), as described by the manufacturer’s instructions. After 36 h, cells were transfected with 0.1 μg psiCHECK-2 -COX2- 1-60 and 0.5 μg pcDNA or pcDNA-HuR/HuR mutants using the TransIT-LT1 (Mirus). For the luciferase assay, cells were harvested after 24h and collected in RNase free conditions, washed with sterile PBS and resuspended in 50ml Passive Lysis buffer per million cells. Renilla luciferase signal was generated and recorded by employing Dual Luciferase® Reported Assay System (Promega) and SpectraMax® L Luminometer (Molecular Devices) according to manufacturer instructions. Renilla signal was normalized to the firefly signal. For western blots, cells were lysed in RIPA lysis buffer (1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 50 mM Tris, pH 7.4, 2 mM Na3VO4, 25 mM β- glycerophosphate, 15 mM NaF, and complete protease inhibitor mix; Roche) and lysate cleared by centrifugation at 13,000 rpm for 10 min at 4˚C. Lysate protein concentration was determined for each sample using BCA assay. Equal amounts of lysate resolved by 10% SDS- PAGE and transferred to a PVDF transfer membrane. Membrane was blocked in 5% BSA in TBS-T at 4˚C. Primary antibodies were incubated at 4˚C overnight. Secondary ECL antibodies (GE Healthcare) were incubated for 1h at RT in 5% Milk in TBS-T. Protein was imaged using chemiluminescence method and Genemate autoradiography film. The following antibody dilutions were used: rabbit anti-ELAVL1 (Cell Signaling #12582, 1:1000), rabbit anti-β-tubulin (Cell Signaling #2146, 1:1000), rabbit anti-Flag (Cell Signaling #2368, 1:1000).

119 Appendix

5.4 A3. Supplementary Tables Chapter 3

Table A3. 1. Protein chemical shift assignments of PABPC1 RRM1 in 20 mM MES (pH 5.5), 50 mM NaCl, 1 mM DTT, 298K. CA, CB, CO, HN and N chemical shifts are in ppm.

CA CB CO HN N

residue A5 50.49 18.24 175.2 8.067 126.6

P6 S7 57.94 64 173.4 8.169 115.3 Y8 55.61 38.49 173.8 8.095 122.3

P9 M10 55.77 32.81 174.9 8.199 119.4 A11 51.31 20.02 175 8.442 123.4 S12 - 64.98 173.6 8.045 114.7 L13 53.34 43.98 175.5 9.116 125.4 Y14 56.75 39.94 173.6 9.084 124.7 V15 60.84 32.73 174.1 8.554 128.3 G16 43.23 - 172.7 9.192 112.4 D17 55.07 38.83 175.4 8.341 114 L18 53.76 43.2 177.9 8.104 114.6 H19 56.19 31.83 175.4 9.315 125.9

D21 54.5 40.32 176.8 11.03 120.5 V22 64.58 30.73 176.4 7.857 119.8 T23 59.64 72.5 174.9 7.511 118.3 E24 62.32 28.25 178.2 9.953 121.1 A25 55.51 18.12 180.9 8.417 120.2 M26 58.97 33.6 179.7 7.55 118 L27 57.64 41.9 178.3 7.86 119.6 Y28 63.33 38.02 178.8 9.045 122.8 E29 59.41 29.43 177.8 7.883 119.6 K30 57.58 32.16 176.7 7.355 116.1 F31 61.26 39.94 176.2 8.377 111.9 S32 62.82 - 174.2 8.264 117.1

P33 A34 53.71 18.31 176.6 7.986 116.7 G35 44.56 - 169.9 7.565 105

P36 I37 60.56 40.7 176.4 8.615 126.1 L38 56.48 43.3 177.1 9.065 127.9 S39 57.94 64.73 171.6 7.559 109.6 L40 60.82 42.18 174.1 8.192 119.8 R41 54.81 33.87 175.2 8.651 125.7

120 Appendix

V42 62.62 31.86 176 8.869 127 C43 60.61 27.24 173.4 8.445 129 R44 54.43 34.27 175.5 8.216 122.4 D45 54.53 43.6 178.2 8.503 122.3 M46 57.81 32.29 176.6 8.933 125.9 L47 62.89 37.48 178.3 8.535 118.5 T48 62.79 70.64 176.3 8.722 109.6 R49 57.62 26.66 174.8 7.868 113.7 R50 56.06 31.05 176.3 7.768 119.9 S51 58.45 64.27 176.1 8.881 116.4 L52 54.5 42.15 177.9 9.106 126.3 G53 45.48 - 172.4 9.008 106.7 Y54 54.97 40.13 171.9 6.902 112.6 A55 49.62 25.83 174.1 8.685 120.4 Y56 56.74 40.97 175.6 8.406 115 V57 61.57 33.62 173.4 8.96 122.8 N58 52.21 39.06 174.7 8.688 125.3 F59 59.52 39.97 174.6 9.003 124.6 Q60 59.28 29.63 176.5 8.084 118.5 Q61 51.81 29.46 175.6 9.575 115.8

P62 A63 54.93 18.55 180 8.658 117.2 D64 56.16 39.83 176.6 7.139 119.5 A65 54.36 18.11 178.4 6.811 122.1 E66 59.26 29.13 177.3 8.261 116.7 R67 59.32 29.62 179.9 7.498 118.8 A68 55.87 18.29 178.2 7.951 124.2 L69 59.13 41.73 178.3 7.926 119.3 D70 56.31 40.97 178.3 8.282 115 T71 63.09 71.68 176.4 7.892 106.5 M72 55.77 33.15 175.3 8.191 119.3 N73 54.52 37.31 176.6 7.206 117.8 F74 59.41 36.94 174.7 8.047 118.8 D75 54.82 40.25 175.8 7.935 120.9 V76 62.92 32.18 176.8 8.311 119.3 I77 61.41 39.99 176 8.244 126.3 K78 56.71 29.87 176 9.572 129.5 G79 45.45 - 174 8.008 100.9 K80 52.79 34.55 174.1 7.713 121.8

P81 V82 60.91 32.85 174.3 8.803 123.1 R83 54.2 32.3 174.8 7.667 122.9 I84 - 39.85 174.8 8.73 125.2

121 Appendix

M85 54.23 37.18 174.5 9.205 123.3 W86 58.66 29.97 177.6 8.588 121.7 S87 59.15 63.77 174.7 8.686 117.1 N88 55.87 29.5 175.4 8.543 121.8 R89 56.09 30.97 175.4 8.195 121.3 D90 52 41.45 175.3 8.201 122.9

P91 S92 59.66 63.36 174.8 8.487 115.2 L93 55.3 41.92 177.2 7.714 122.5 R94 56.13 30.59 176.2 7.973 121.2 K95 56.27 32.96 176.5 8.267 122.5 S96 58.38 63.9 174.9 8.327 116.8 G97 45.36 - - 8.421 110.9 V98 62 32.79 175.6 7.943 118.3 G99 46.08 - - 8.043 118.3

122 Appendix

Table A3. 2. Protein chemical shift assignments of BTG2(APRO). Measurements were performed in

20 mM Na2HPO4 (pH7), 100 mM NaCl, 1 mM DTT, 298K. CA, CB, CO, HN and N chemical shifts are in ppm.

CA CB CO HN N

residue

H3 G4 45.42 - 174.2 8.52 110.1 K5 56.49 33.12 177.3 8.416 121.1 G6 45.63 - 174.3 8.607 110.2 T7 61.56 69.57 174.4 8.021 113 D8 54.7 41.06 - 8.357 123.7

L10

P11

E12

I13

A14 A15 54.77 18.04 180.7 7.876 121.4 A16 55.78 18.92 178.9 7.988 122.1 V17 66.9 31.51 179.2 9.016 117.4 G18 47.11 - 176.4 8.779 109.9 F19 60.32 39.92 177.7 7.756 123 L20 58.3 42.22 178.7 8.084 119.9 S21 62.95 - 177.3 8.86 113.1 S22 61.73 62.64 176.8 7.935 117.9 L23 57.79 41.58 179.4 7.476 123.1 L24 57.8 41.87 179 7.237 117.9 R25 59.59 30.81 178.9 8.242 119.6 T26 65.38 69.54 175.9 7.953 111.6 R27 56.65 30.36 177.7 7.896 118 G28 - - 173.9 7.63 105.8 C29 58.41 27.59 173.9 7.779 115.5 V30 62.04 32.93 175.2 7.88 121.5 S31 57.94 64.51 174.5 8.33 122.3 E32 59.42 28.81 179.3 8.993 121.6 Q33 59.67 27.97 178.9 8.598 118.3 R34 59.4 30.87 178.9 7.534 119.6 L35 57.9 41.19 179.4 8.551 120.5 K36 59.75 32.18 180 8.202 121.5 V37 66.3 31.54 178.5 7.396 121 F38 61.51 39.61 175.6 8.482 120.4 S39 61.73 62.9 176.7 8.92 114.5 G40 47.32 - 176.1 7.892 108.2

123 Appendix

A41 54.43 18.57 181.1 8.24 124 L42 57.33 40.38 177.7 8.779 121.4 Q43 59.85 27.87 179.1 8.161 119.1 E44 59.32 29.44 178.3 7.676 120.3 A45 54.9 18.91 181.6 8.171 121.5 L46 57.75 41.12 177.3 8.511 118.3 T47 67.31 68.43 176.1 7.633 115.6 E48 58.58 29.08 177.6 7.798 119.1 H49 57.41 28.9 177.6 7.355 114.8 Y50 56.2 44.1 176.3 7.766 118.1 K51 51.93 41.91 173.5 7.395 117.4

H52

H53

W54

F55

P56

E57

K58

P59 S60 59.51 62.99 177.3 7.823 105.5 K61 57.95 32.27 178.7 8.478 127.9 G62 46.55 - 174.3 9.697 119.2 S63 61.9 63.26 176.1 7.95 116.5 G64 46.92 - 175.7 9.001 108.3 Y65 60.85 38.13 177.1 7.637 123.1 R66 58.24 32.2 176.1 7.96 116.4 C67 59.47 27.56 173.9 7.311 117.4 L68 60.12 39.27 - 9.272 130.5 R69 54.7 32.78 175.5 8.761 126.4 I70 61.16 39.38 174.6 8.965 124.3 N71 52.9 36.93 174.4 8.067 126.2

H72

K73

M74

D75

P76

I77 I78 65.25 36.17 178.1 8.299 122.3 S79 62.71 62.94 176.7 8.161 112.1 R80 59.49 30.08 179.5 7.648 121.7 V81 66.25 30.76 178.3 8.233 120.3 A82 55.32 19.56 179.9 8.869 120.9 S83 61.49 62.79 177.9 7.904 113.2

124 Appendix

Q84 58.38 28.42 177.9 7.769 121.2 I85 60.61 38.33 176.1 7.383 108.5 G86 46.55 - 174.3 7.664 109.1 L87 53.58 43.53 175.6 7.651 120.4 S88 56.66 65.31 174.3 8.583 120.1 Q89 61.61 25.89 - 9.142 119.8

P90

Q91

L92

H93

Q94

L95

L96

P97 S98 59.67 63.74 174.2 8.175 115.5 E99 56.21 30.46 175.5 7.678 118.3 L100 54.75 45.53 175.3 8.554 125.3 T101 62.65 70.65 171.3 8.729 124 L102 54.11 45.67 173.5 8.94 129.5 W103 59.9 29.3 175.1 9.53 127.8 V104 61.72 31.2 172.1 7.905 124.4 D105 51.23 41.59 172.3 8.103 124

P106

Y107 E108 57.77 33 173.3 8.377 121.7 V109 60.81 34.35 175.6 9.126 127.3 S110 57.59 67.3 171.7 9.49 123.3 Y111 56.35 42.99 172.3 9.284 116.5 R112 55.73 35.79 175.3 9.075 118.8 I113 61.31 38.64 176.5 9.637 132 G114 44.32 - 174.4 8.414 115.7 E115 58.86 29.82 177 8.874 121.4 D116 52.73 41.01 176.9 8.392 115 G117 45.45 - 173.6 7.379 107.2 S118 58.41 63.98 174.2 8.405 115.9 I119 61.54 37.44 176.6 8.577 124.3 C120 57.49 28.96 173.6 9.023 129.1 V121 63.73 32.57 176.4 8.758 125.8 L122 55.32 45.35 176.4 9.115 128.6 Y123 57.76 42.18 172.4 7.95 120.8 E124 54.66 31.99 174 7.765 126 E125 55.96 31.22 174.4 8.053 124.6 A126 53.76 20.21 182.5 8.045 101.7

125 Appendix

Table A3. 3. ITC conditions tested to analyses RRM12- BTG2 interaction. BTG2 is in the sample cell and PABP variants in the syringe.

Constructs [BTG2] µM [PABP] µM [T] °C Buffer

BTG2+RRM12 10.5 200 25 20 mM Na2HPO4 pH 7, 100 mM NaCl, 1 mM DTT

BTG2+RRM12 13 329 20 20 mM Na2HPO4 pH 7, 100 mM NaCl, 1 mM DTT

BTG2+RRM12 29 314 14.5 20 mM Na2HPO4 pH 7, 100 mM NaCl, 1 mM DTT BTG2+RRM1 8.9 409 19 20 mM MES pH 5.5, 100 mM NaCl, 1 mM DTT

Table A3. 4. HADDOCK scores of two additional Models (4+5) with moderate HADDOCK scores. Models are shown in Supplementary Figure A.3.4

HADDOCK Cluster RMSD Z- score size Score Model 4 -79.0 +/- 8.6 10 8.3 +/- 0.4 -0.5

Model 5 -96.7 +/- 5.8 73 6.7 +/- 0.2 -0.4

126 Appendix

5.5 A3. Supplementary Figures Chapter 3

Figure A3. 1. Analytical SEC and ITC of RRM1- BTG2 complex. (A) Chromatograms from analytical SEC in 20 mM Na2HPO4 pH 7, 100 mM NaCl, 5 mM DTT. Overlay of the NMR sample (230 µM BTG2 in complex with RRM1; ratio 1:10) with a gel filtration standard (Bio-Rad). (B) ITC profile of RRM1 titrated into BTG2. Top: ITC raw data as a function of time, bottom: integrated heat release.

Figure A3. 2. Comparison of BTG2(APRO) and BTG2(APRO) boxC mut in analytical SEC and T1 and T2 relaxation experiments. (A) Chromatograms from preparative SEC in 20 mM Na2HPO4 pH 7, 100 mM NaCl, 1 mM DTT. Overlay of BTG2(APRO) (blue) and BTG2(APRO) boxC mut (black). (B) Rotational correlation times τc obtained from non-overlapping amide resonances of BTG2(APRO) (blue) and BTG2(APRO) boxC mut (black).

127 Appendix

Figure A3. 1. BTG2 APRO boxC mut crystals obtained in the NMR tube after sample was stored at 4°C

Figure A3. 2. Additional BTG2-RRM12 HADDOCK models. (A) Model 4. Binding of RRM1 will to a clash with CAF1. (B) Model 5. The 3’-end poly(A) appears in the back site of RRM1, further away from the catalytic pocket of CAF1 then in model 1.

128

6. References

Abdelmohsen, K., Pullmann, R., Lal, A., Kim, H.H., Galban, S., Yang, X., Blethrow, J.D., Walker, M., Shubert, J., Gillespie, D. a, Furneaux, H., Gorospe, M., 2007. Phosphorylation of HuR by Chk2 regulates SIRT1 expression. Mol. Cell 25, 543–57. https://doi.org/10.1016/j.molcel.2007.01.011 Abe, R., Sakashita, E., Yamamoto, K., Sakamoto, H., 1996. Two different RNA binding activities for the AU-rich element and the poly(A) sequence of the mouse neuronal protein mHuC. Nucleic Acids Res. 24, 4895–4901. https://doi.org/10.1093/nar/24.24.4895 Adams, P.D., Grosse-Kunstleve, R.W., Hung, L.W., Ioerger, T.R., McCoy, A.J., Moriarty, N.W., Read, R.J., Sacchettini, J.C., Sauter, N.K., Terwilliger, T.C., 2002. PHENIX: Building new software for automated crystallographic structure determination. Acta Crystallogr. Sect. D Biol. Crystallogr. 58, 1948–1954. https://doi.org/10.1107/S0907444902016657 Antic, D., Keene, J.D., 1997. Embryonic Lethal Abnormal Visual RNA-Binding Proteins Involved in Growth , Differentiation , and Posttranscriptional Gene Expression. Am. J. Hum. Genet. 61, 273–278. https://doi.org/10.1086/514866 Auweter, S.D., Oberstrass, F.C., Allain, F.H.-T., 2006. Sequence-specific binding of single- stranded RNA: is there a code for recognition? Nucleic Acids Res. 34, 4943–59. https://doi.org/10.1093/nar/gkl620 Bakheet, T., Williams, B.R.G., Khabar, K.S.A., 2006. ARED 3.0: the large and diverse AU-rich transcriptome. Nucleic Acids Res. 34, D111-4. https://doi.org/10.1093/nar/gkj052 Banerjee, H., Rahn, A., Davis, W., Singh, R., 2003. Sex lethal and U2 small nuclear ribonucleoprotein auxiliary factor (U2AF65) recognize polypyrimidine tracts using multiple modes of binding. RNA 9, 88–99. https://doi.org/10.1261/rna.2131603 Barraud, P., Allain, F.H.T., 2013. Solution structure of the two RNA recognition motifs of hnRNP A1 using segmental isotope labeling: How the relative orientation between RRMs influences the nucleic acid binding topology. J. Biomol. NMR 55, 119–138. https://doi.org/10.1007/s10858-012-9696-4 Barreau, C., Paillard, L., Osborne, H.B., 2005. AU-rich elements and associated factors: Are there unifying principles? Nucleic Acids Res. 33, 7138–7150. https://doi.org/10.1093/nar/gki1012 Basquin, J., Roudko, V. V., Rode, M., Basquin, C., Séraphin, B., Conti, E., 2012. Architecture of the nuclease module of the yeast ccr4-Not complex: The not1-caf1-ccr4 interaction. Mol. Cell 48, 207–218. https://doi.org/10.1016/j.molcel.2012.08.014 Behm-Ansmant, I., Rehwinkel, J., Doerks, T., Stark, A., Bork, P., Izaurralde, E., 2006. mRNA degradation by miRNAs and GW182 requires both CCR4 : NOT deadenylase and DCP1 : DCP2 decapping complexes. Genes Dev. 20, 1885–1898 ST–MRNA degradation by miRNAs and GW1. https://doi.org/10.1101/gad.1424106 Benjamin, D., Moroni, C., 2007. mRNA stability and cancer: an emerging link? Expert Opin. Biol. Ther. 7, 1515–29. https://doi.org/10.1517/14712598.7.10.1515 Benoit, R.M., Meisner, N.-C., Kallen, J., Graff, P., Hemmig, R., Cèbe, R., Ostermeier, C., Widmer, H., Auer, M., 2010. The x-ray crystal structure of the first RNA recognition motif and site- directed mutagenesis suggest a possible HuR redox sensing mechanism. J. Mol. Biol. 397, 1231–44. https://doi.org/10.1016/j.jmb.2010.02.043 Berendsen, H.J.C., Grigera, J.R., Straatsma, T.P., 1987. The missing term in effective pair potentials. J. Phys. Chem. 91, 6269–6271. https://doi.org/10.1021/j100308a038

References

Berthet, C., Guéhenneux, F., Revol, V., Samarut, C., Lukaszewicz, A., Dehay, C., Dumontet, C., Magaud, J.-P., Rouault, J.-P., 2002. Interaction of PRMT1 with BTG/TOB proteins in cell signalling: molecular analysis and functional aspects. Genes Cells 7, 29–39. Beusch, I., Barraud, P., Moursy, A., Cléry, A., Allain, F.H.T., 2017. Tandem hnRNP A1 RNA recognition motifs act in concert to repress the splicing of survival motor neuron exon 7. Elife 6, 1–35. https://doi.org/10.7554/eLife.25736 Bhattacharyya, S.N., Habermacher, R., Martine, U., Closs, E.I., Filipowicz, W., 2006. Relief of microRNA-mediated translational repression in human cells subjected to stress. Cell 125, 1111–24. https://doi.org/10.1016/j.cell.2006.04.031 Boeynaems, S., Alberti, S., Fawzi, N.L., Mittag, T., Polymenidou, M., Rousseau, F., Schymkowitz, J., Shorter, J., Wolozin, B., Van Den Bosch, L., Tompa, P., Fuxreiter, M., 2018. Protein Phase Separation: A New Phase in Cell Biology. Trends Cell Biol. 28, 420–435. https://doi.org/10.1016/j.tcb.2018.02.004 Boiko, A.D., Porteous, S., Razorenova, O. V., Krivokrysenko, V.I., Williams, B.R., Gudkov, A. V., 2006. A systematic search for downstream mediators of tumor suppressor function of p53 reveals a major role of BTG2 in suppression of Ras-induced transformation. Genes Dev. 20, 236–252. https://doi.org/10.1101/gad.1372606 Brennan, C.M., Gallouzi, I.E., Steitz, J. a, 2000. Protein ligands to HuR modulate its interaction with target mRNAs in vivo. J. Cell Biol. 151, 1–14. Calabretta, S., Richard, S., 2015. Emerging Roles of Disordered Sequences in RNA-Binding Proteins. Trends Biochem. Sci. 40, 662–672. https://doi.org/10.1016/j.tibs.2015.08.012 Cammas, A., Sanchez, B.J., Lian, X.J., Dormoy-Raclet, V., van der Giessen, K., de Silanes, I.L., Ma, J., Wilusz, C., Richardson, J., Gorospe, M., Millevoi, S., Giovarelli, M., Gherzi, R., Di Marco, S., Gallouzi, I.-E., 2014. Destabilization of Nucleophosmin mRNA by the HuR/KSRP complex is required for muscle fiber formation. Nat. Commun. 10.1038/ncomms5190. https://doi.org/10.1038/ncomms5190 Chang, T.C., Yamashita, A., Chen, C.Y.A., Yamashita, Y., Zhu, W., Durdan, S., Kahvejian, A., Sonenberg, N., Shyu, A. Bin, 2004. UNR, a new partner of poly(A)-binding protein, plays a key role in translationally coupled mRNA turnover mediated by the c-fos major coding-region determinant. Genes Dev. 18, 2010–2023. https://doi.org/10.1101/gad.1219104 Chayen, N.E., Saridakis, E., 2008. Protein crystallization: From purified protein to diffraction-quality crystal. Nat. Methods 5, 147–153. https://doi.org/10.1038/nmeth.f.203 Chen, C.-Y. a, Shyu, A.-B., 2003. Rapid deadenylation triggered by a nonsense codon precedes decay of the RNA body in a mammalian cytoplasmic nonsense-mediated decay pathway. Mol. Cell. Biol. 23, 4805–4813. https://doi.org/10.1128/MCB.23.14.4805-4813.2003 Chen, C.-Y.A., Xu, N., Shyu, A.-B., 2002. Highly selective actions of HuR in antagonizing AU-rich element-mediated mRNA destabilization. Mol. Cell. Biol. 22, 7268–78. https://doi.org/10.1128/MCB.22.20.7268-7278.2002 Chen, C.Y., Shyu, A.B., 1995. AU-rich elements: characterization and importance in mRNA degradation. Trends Biochem. Sci. 20, 465–70. Chen, Y.I.G., Moore, R.E., Ge, H.Y., Young, M.K., Lee, T.D., Stevens, S.W., 2007. Proteomic analysis of in vivo-assembled pre-mRNA splicing complexes expands the catalog of participating factors. Nucleic Acids Res. 35, 3928–3944. https://doi.org/10.1093/nar/gkm347 Cho, S.J., Zhang, J., Chen, X., 2010. RNPC1 modulates the RNA-binding activity of, and cooperates with, HuR to regulate p21 mRNA stability. Nucleic Acids Res. 38, 2256–67. https://doi.org/10.1093/nar/gkp1229 Cieniková, Z., Damberger, F.F., Hall, J., Allain, F.H.T., Maris, C., 2014. Structural and Mechanistic

130

References

Insights into Poly(uridine) Tract Recognition by the hnRNP C RNA Recognition Motif. J. Am. Chem. Soc. 136, 14536–14544. https://doi.org/10.1021/ja507690d Clark, A., Dean, J., Tudor, C., Saklatvala, J., 2009. Post-transcriptional gene regulation by MAP kinases via AU-rich elements. Front. Biosci. 14, 847–871. https://doi.org/10.2741/3282 Cléry, A., Blatter, M., Allain, F.H.-T., 2008. RNA recognition motifs: boring? Not quite. Curr. Opin. Struct. Biol. 18, 290–8. https://doi.org/10.1016/j.sbi.2008.04.002 Clery, A., Sinha, R., Anczukow, O., Corrionero, A., Moursy, A., Daubner, G.M., Valcarcel, J., Krainer, A.R., Allain, F.H.-T., 2013. Isolated pseudo-RNA-recognition motifs of SR proteins can regulate splicing using a noncanonical mode of RNA recognition. Proc. Natl. Acad. Sci. 110, E2802–E2811. https://doi.org/10.1073/pnas.1303445110 Corsini, L., Hothorn, M., Scheffzek, K., Sattler, M., Stier, G., 2008. Thioredoxin as a fusion tag for carrier-driven crystallization. Protein Sci. 17, 2070–2079. https://doi.org/10.1110/ps.037564.108 Dasgupta, T., Ladd, A.N., 2012. The importance of CELF control: Molecular and biological roles of the CUG-BP, Elav-like family of RNA-binding proteins. Wiley Interdiscip. Rev. RNA 3, 104– 121. https://doi.org/10.1002/wrna.107 De Simone, A., Cavalli, A., Hsu, S.T.D., Vranken, W., Vendruscolo, M., 2009. Accurate random coil chemical shifts from an analysis of loop regions in native states of proteins. J. Am. Chem. Soc. 131, 16332–16333. https://doi.org/10.1021/ja904937a de Vries, S.J., van Dijk, A.D.J., Krzeminski, M., van Dijk, M., Thureau, A., Hsu, V., Wassenaar, T., Bonvin, A.M.J.J., 2007. HADDOCK versus HADDOCK: New features and performance of HADDOCK2.0 on the CAPRI targets. Proteins Struct. Funct. Bioinforma. 69, 726–733. https://doi.org/10.1002/prot.21723 De Vries, S.J., Van Dijk, M., Bonvin, A.M.J.J., 2010. The HADDOCK web server for data-driven biomolecular docking. Nat. Protoc. 5, 883–897. https://doi.org/10.1038/nprot.2010.32 Dean, J.L., Wait, R., Mahtani, K.R., Sully, G., Clark, A.R., Saklatvala, J., 2001. The 3’ untranslated region of tumor necrosis factor alpha mRNA is a target of the mRNA-stabilizing factor HuR. Mol. Cell. Biol. 21, 721–30. https://doi.org/10.1128/MCB.21.3.721-730.2001 Deo, R.C., Bonanno, J.B., Sonenberg, N., Burley, S.K., 1999. Recognition of polyadenylate RNA by the poly(A)-binding protein. Cell 98, 835–45. Di Marco, S., Mazroui, R., Dallaire, P., Chittur, S., Tenenbaum, S.A., Radzioch, D., Marette, A., Gallouzi, I., 2005. NF-kappa B-mediated MyoD decay during muscle wasting requires nitric oxide synthase mRNA stabilization, HuR protein, and nitric oxide release. Mol. Cell. Biol. 25, 6533–45. https://doi.org/10.1128/MCB.25.15.6533-6545.2005 Doller, A., Akool, E.-S., Huwiler, A., Müller, R., Radeke, H.H., Pfeilschifter, J., Eberhardt, W., 2008. Posttranslational modification of the AU-rich element binding protein HuR by protein kinase Cdelta elicits angiotensin II-induced stabilization and nuclear export of cyclooxygenase 2 mRNA. Mol. Cell. Biol. 28, 2608–25. https://doi.org/10.1128/MCB.01530-07 Doller, A., Winkler, C., Azrilian, I., Schulz, S., Hartmann, S., Pfeilschifter, J., Eberhardt, W., 2011. High-constitutive HuR phosphorylation at Ser 318 by PKCδ propagates tumor relevant functions in colon carcinoma cells. Carcinogenesis 32, 676–685. https://doi.org/10.1093/carcin/bgr024 Dominguez, C., Allain, F.H.T., 2006. NMR structure of the three quasi RNA recognition motifs (qRRMs) of human hnRNP F and interaction studies with Bcl-x G-tract RNA: A novel mode of RNA recognition. Nucleic Acids Res. 34, 3634–3645. https://doi.org/10.1093/nar/gkl488 Dominguez, C., Boelens, R., Bonvin, A.M.J.J., 2003. HADDOCK: A protein-protein docking approach based on biochemical or biophysical information. J. Am. Chem. Soc. 125, 1731–

131

References

1737. https://doi.org/10.1021/ja026939x Dominguez, C., Schubert, M., Duss, O., Ravindranathan, S., Allain, F.H.-T., 2011. Structure determination and dynamics of protein-RNA complexes by NMR spectroscopy. Prog. Nucl. Magn. Reson. Spectrosc. 58, 1–61. https://doi.org/10.1016/j.pnmrs.2010.10.001 Duss, O., Yulikov, M., Allain, F.H.T., Jeschke, G., 2015. Combining NMR and EPR to Determine Structures of Large RNAs and Protein-RNA Complexes in Solution. Methods Enzymol. 558, 279–331. https://doi.org/10.1016/bs.mie.2015.02.005 Dutertre, M., Chakrama, F.Z., Combe, E., Desmet, F.-O., Mortada, H., Polay Espinoza, M., Gratadou, L., Auboeuf, D., 2014. A recently evolved class of alternative 3’-terminal exons involved in cell cycle regulation by topoisomerase inhibitors. Nat. Commun. 5, 3395. https://doi.org/10.1038/ncomms4395 Emsley, P., Cowtan, K., 2004. Coot: Model-building tools for molecular graphics. Acta Crystallogr. Sect. D Biol. Crystallogr. 60, 2126–2132. https://doi.org/10.1107/S0907444904019158 Ezzeddine, N., Chang, T.-C., Zhu, W., Yamashita, A., Chen, C.-Y.A., Zhong, Z., Yamashita, Y., Zheng, D., Shyu, A.-B., 2007. Human TOB, an Antiproliferative Transcription Factor, Is a Poly(A)-Binding Protein-Dependent Positive Regulator of Cytoplasmic mRNA Deadenylation. Mol. Cell. Biol. 27, 7791–7801. https://doi.org/10.1128/MCB.01254-07 Fan, X.C., Steitz, J. a, 1998a. HNS, a nuclear-cytoplasmic shuttling sequence in HuR. Proc. Natl. Acad. Sci. U. S. A. 95, 15293–8. Fan, X.C., Steitz, J.A., 1998b. Overexpression of HuR, a nuclear-cytoplasmic shuttling protein, increases the in vivo stability of ARE-containing mRNAs. EMBO J. 17, 3448–60. https://doi.org/10.1093/emboj/17.12.3448 Farooq, F., Balabanian, S., Liu, X., Holcik, M., MacKenzie, A., 2009. p38 Mitogen-activated protein kinase stabilizes SMN mRNA through RNA binding protein HuR. Hum. Mol. Genet. 18, 4035– 4045. https://doi.org/10.1093/hmg/ddp352 Fialcowitz-White, E.J., Brewer, B.Y., Ballin, J.D., Willis, C.D., Toth, E. a, Wilson, G.M., 2007. Specific protein domains mediate cooperative assembly of HuR oligomers on AU-rich mRNA- destabilizing sequences. J. Biol. Chem. 282, 20948–59. https://doi.org/10.1074/jbc.M701751200 Figueroa, A., Cuadrado, A., Fan, J., Atasoy, U., Muscat, G.E., Muñoz-Canoves, P., Gorospe, M., Muñoz, A., 2003. Role of HuR in skeletal myogenesis through coordinate regulation of muscle differentiation genes. Mol. Cell. Biol. 23, 4991–5004. https://doi.org/10.1128/MCB.23.14.4991 Filippova, N., Yang, X., Ananthan, S., Sorochinsky, A., Hackney, J.R., Gentry, Z., Bae, S., King, P., Nabors, L.B., 2017. Hu antigen R (HuR) multimerization contributes to glioma disease progression. J. Biol. Chem. 292, 16999–17010. https://doi.org/10.1074/jbc.M117.797878 Filippova, N., Yang, X., Wang, Y., Gillespie, G.Y., Langford, C., King, P.H., Wheeler, C., Nabors, L.B., 2011. The RNA-binding protein HuR promotes glioma growth and treatment resistance. Mol. Cancer Res. 9, 648–659. https://doi.org/10.1158/1541-7786.MCR-10-0325 Gallouzi, I.E., Brennan, C.M., Steitz, J.A., 2001. Protein ligands mediate the CRM1-dependent export of HuR in response to heat shock. RNA 7, 1348–61. Gallouzi, I.E., Brennan, C.M., Stenberg, M.G., Swanson, M.S., Eversole, A., Maizels, N., Steitz, J. a, 2000. HuR binding to cytoplasmic mRNA is perturbed by heat shock. Proc. Natl. Acad. Sci. U. S. A. 97, 3073–8. Gantt, K.R., Cherry, J., Richardson, M., Karschner, V., Atasoy, U., Pekala, P.H., 2006. The regulation of glucose transporter (GLUT1) expression by the RNA binding protein HuR. J. Cell. Biochem. 99, 565–574. https://doi.org/10.1002/jcb.20950

132

References

Gao, F.B., Keene, J.D., 1996. Hel-N1/Hel-N2 proteins are bound to poly(A)+ mRNA in granular RNP structures and are implicated in neuronal differentiation. J. Cell Sci. 109 ( Pt 3, 579–89. García-Mauriño, S.M., Rivero-Rodríguez, F., Velázquez-Cruz, A., Hernández-Vellisca, M., Díaz- Quintana, A., De la Rosa, M.A., Díaz-Moreno, I., 2017. RNA Binding Protein Regulation and Cross-Talk in the Control of AU-rich mRNA Fate. Front. Mol. Biosci. 4, 1–9. https://doi.org/10.3389/fmolb.2017.00071 Garman, E.F., Schneider, T.R., 1997. Macromolecular Cryocrystallography. J. Appl. Crystallogr. 30, 211–237. https://doi.org/10.1107/S0021889897002677 Gerstberger, S., Hafner, M., Tuschl, T., 2014. A census of human RNA-binding proteins. Nat. Publ. Gr. 15, 829–845. https://doi.org/10.1038/nrg3813 Grammatikakis, I., Abdelmohsen, K., Gorospe, M., 2017. Posttranslational control of HuR function. Wiley Interdiscip. Rev. RNA 8, 4091. https://doi.org/10.1002/wrna.1372 Grosset, C., Chen, C.Y., Xu, N., Sonenberg, N., Jacquemin-Sablon, H., Shyu, A.B., 2000. A mechanism for translationally coupled mRNA turnover: interaction between the poly(A) tail and a c-fos RNA coding determinant via a protein complex. Cell 103, 29–40. https://doi.org/10.1016/S0092-8674(00)00102-1 Gryk, M.R., Abseher, R., Simon, B., Nilges, M., Oschkinat, H., 1998. Heteronuclear relaxation study of the PH domain of beta-spectrin: restriction of loop motions upon binding inositol trisphosphate. J. Mol. Biol. 280, 879–96. https://doi.org/10.1006/jmbi.1998.1731 Handa, N., Nureki, O., Kurimoto, K., Kim, I., Sakamoto, H., Shimura, Y., Muto, Y., Yokoyama, S., 1999. Structural basis for recognition of the tra mRNA precursor by the Sex-lethal protein. Nature 398, 579–85. https://doi.org/10.1038/19242 Heck, A.M., Wilusz, J., 2018. The Interplay between the RNA Decay and Translation Machinery in Eukaryotes. Cold Spring Harb. Perspect. Biol. a032839. https://doi.org/10.1101/cshperspect.a032839 Horiuchi, M., Takeuchi, K., Noda, N., Muroya, N., Suzuki, T., Nakamura, T., Kawamura-Tsuzuku, J., Takahasi, K., Yamamoto, T., Inagaki, F., 2009. Structural basis for the antiproliferative activity of the Tob-hCaf1 complex. J. Biol. Chem. 284, 13244–13255. https://doi.org/10.1074/jbc.M809250200 Hsia, T.-C., Tu, C.-Y., Chen, Y.-J., Wei, Y.-L., Yu, M.-C., Hsu, S.-C., Tsai, S.-L., Chen, W.-S., Yeh, M.-H., Yen, C.-J., Yu, Y.-L., Huang, T.-C., Huang, C.-Y., Hung, M.-C., Huang, W.-C., 2013. Lapatinib-Mediated Cyclooxygenase-2 Expression via Epidermal Growth Factor Receptor/HuR Interaction Enhances the Aggressiveness of Triple-Negative Breast Cancer Cells. Mol. Pharmacol. 83, 857–869. https://doi.org/10.1124/mol.112.082743 Hyeon, H.K., Yang, X., Kuwano, Y., Gorospe, M., 2008. Modification at HuR(S242) alters HuR localization and proliferative influence. Cell Cycle 7, 3371–3377. https://doi.org/10.4161/cc.7.21.6895 Ikematsu, N., Yoshida, Y., Kawamura-Tsuzuku, J., Ohsugi, M., Onda, M., Hirai, M., Fujimoto, J., Yamamoto, T., 1999. Tob2, a novel anti-proliferative Tob/BTG1 family member, associates with a component of the CCR4 transcriptional regulatory complex capable of binding cyclin- dependent kinases. Oncogene 18, 7432–7441. https://doi.org/10.1038/sj.onc.1203193 Ikeya, T., Ban, D., Lee, D., Ito, Y., Kato, K., Griesinger, C., 2018. Solution NMR views of dynamical ordering of biomacromolecules. Biochim. Biophys. Acta - Gen. Subj. 1862, 287–306. https://doi.org/10.1016/j.bbagen.2017.08.020 Inoue, M., Muto, Y., Sakamoto, H., Yokoyama, S., 2000. NMR studies on functional structures of the AU-rich element-binding domains of Hu antigen C. Nucleic Acids Res. 28, 1743–50. Izquierdo, J.M., 2010. Cell-specific regulation of Fas exon 6 splicing mediated by Hu antigen R.

133

References

Biochem. Biophys. Res. Commun. 402, 324–8. https://doi.org/10.1016/j.bbrc.2010.10.025 Jain, A., Vale, R.D., 2017. RNA phase transitions in repeat expansion disorders. Nature 546, 243– 247. https://doi.org/10.1038/nature22386 Joung, I.S., Cheatham, T.E., 2008. Determination of alkali and halide monovalent ion parameters for use in explicitly solvated biomolecular simulations. J. Phys. Chem. B 112, 9020–9041. https://doi.org/10.1021/jp8001614 Kabsch, W., 1993. Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. J. Appl. Crystallogr. 26, 795–800. https://doi.org/10.1107/S0021889893005588 Karaca, E., Bonvin, A.M.J.J., 2013. Advances in integrative modeling of biomolecular complexes. Methods 59, 372–381. https://doi.org/10.1016/j.ymeth.2012.12.004 Kasashima, K., Sakashita, E., Saito, K., Sakamoto, H., 2002. Complex formation of the neuron- specific ELAV-like Hu RNA-binding proteins. Nucleic Acids Res. 30, 4519–26. Katsanou, V., Milatos, S., Yiakouvaki, A., Sgantzis, N., Kotsoni, A., Alexiou, M., Harokopos, V., Aidinis, V., Hemberger, M., Kontoyiannis, D.L., 2009. The RNA-binding protein Elavl1/HuR is essential for placental branching morphogenesis and embryonic development. Mol. Cell. Biol. 29, 2762–2776. https://doi.org/10.1128/MCB.01393-08 Kawakubo, H., Carey, J.L., Brachtel, E., Gupta, V., Green, J.E., Walden, P.D., Maheswaran, S., 2004. Expression of the NF-κB-responsive gene BTG2 is aberrantly regulated in breast cancer. Oncogene 23, 8310–8319. https://doi.org/10.1038/sj.onc.1208008 Kay, L.E., Torchia, D.A., Bax, A., 1989. Backbone Dynamics of Proteins As Studied by 15N Inverse Detected Heteronuclear NMR Spectroscopy: Application to Staphylococcal Nuclease. Biochemistry 28, 8972–8979. https://doi.org/10.1021/bi00449a003 Khabar, K.S. a, 2005. The AU-Rich Transcriptome: More Than Interferons and Cytokines, and Its Role in Disease. J. Interferon Cytokine Res. 25, 1–10. https://doi.org/10.1089/jir.2005.25.1 Kim, H.H., Abdelmohsen, K., Lal, A., Pullmann, R., Yang, X., Srikantan, R.S., Martindale, J.L., Blethrow, J., M, K., Gorospe, M., Francisco, S., 2008. Nuclear Retention of HuR Through Phosphorylation by Cdk1. Genes Dev. 1–30. https://doi.org/10.1101/gad.1645808.that Kim, H.H., Kuwano, Y., Srikantan, S., Lee, E.K., Martindale, J.L., Gorospe, M., 2009. HuR recruits let-7/RISC to repress c-Myc expression. Genes Dev. 23, 1743–8. https://doi.org/10.1101/gad.1812509 Kim, H.S., Wilce, M.C.J., Yoga, Y.M.K., Pendini, N.R., Gunzburg, M.J., Cowieson, N.P., Wilson, G.M., Williams, B.R.G., Gorospe, M., Wilce, J. a, 2011. Different modes of interaction by TIAR and HuR with target RNA and DNA. Nucleic Acids Res. 39, 1117–30. https://doi.org/10.1093/nar/gkq837 King, P.H., Levine, T.D., Fremeau, R.T., Keene, J.D., 1994. Mammalian homologs of Drosophila ELAV localized to a neuronal subset can bind in vitro to the 3’ UTR of mRNA encoding the Id transcriptional repressor. J. Neurosci. 14, 1943–52. Kotta-Loizou, I., Vasilopoulos, S.N., Coutts, R.H.A., Theocharis, S., 2016. Current Evidence and Future Perspectives on HuR and Breast Cancer Development, Prognosis, and Treatment. Neoplasia 18, 674–688. https://doi.org/10.1016/j.neo.2016.09.002 Kozlov, G., Safaee, N., Rosenauer, A., Gehring, K., 2010. Structural basis of binding of P-body- associated proteins GW182 and ataxin-2 by the Mlle domain of poly(A)-binding protein. J. Biol. Chem. 285, 13599–13606. https://doi.org/10.1074/jbc.M109.089540 Kozlov, G., Trempe, J.F., Khaleghpour, K., Kahvejian, a, Ekiel, I., Gehring, K., 2001. Structure and function of the C-terminal PABC domain of human poly(A)-binding protein. Proc. Natl. Acad. Sci. U. S. A. 98, 4409–4413. https://doi.org/10.1073/pnas.071024998

134

References

Krepl, M., Havrila, M., Stadlbauer, P., Banas, P., Otyepka, M., Pasulka, J., Stefl, R., Sponer, J., 2015. Can we execute stable microsecond-scale atomistic simulations of protein-rna complexes? J. Chem. Theory Comput. 11, 1220–1243. https://doi.org/10.1021/ct5008108 Krissinel, E., Henrick, K., 2007. Inference of Macromolecular Assemblies from Crystalline State. J. Mol. Biol. 372, 774–797. https://doi.org/10.1016/j.jmb.2007.05.022 Kühn, U., Wahle, E., 2004. Structure and function of poly(A) binding proteins. Biochim. Biophys. Acta 1678, 67–84. https://doi.org/10.1016/j.bbaexp.2004.03.008 Kührová, P., Best, R.B., Bottaro, S., Bussi, G., Šponer, J., Otyepka, M., Banáš, P., 2016. Computer Folding of RNA Tetraloops: Identification of Key Force Field Deficiencies. J. Chem. Theory Comput. 12, 4534–4548. https://doi.org/10.1021/acs.jctc.6b00300 Kundu, P., Fabian, M.R., Sonenberg, N., Bhattacharyya, S.N., Filipowicz, W., 2012. HuR protein attenuates miRNA-mediated repression by promoting miRISC dissociation from the target RNA. Nucleic Acids Res. 40, 5088–100. https://doi.org/10.1093/nar/gks148 Lal, A., Kawai, T., Yang, X., Mazan-Mamczarz, K., Gorospe, M., 2005. Antiapoptotic function of RNA-binding protein HuR effected through prothymosin α. EMBO J. 24, 1852–1862. https://doi.org/10.1038/sj.emboj.7600661 Leandersson, K., Riesbeck, K., Andersson, T., 2006. Wnt-5a mRNA translation is suppressed by the Elav-like protein HuR in human breast epithelial cells. Nucleic Acids Res. 34, 3988–3999. https://doi.org/10.1093/nar/gkl571 Lebedeva, S., Jens, M., Theil, K., Schwanhäusser, B., Selbach, M., Landthaler, M., Rajewsky, N., 2011. Transcriptome-wide analysis of regulatory interactions of the RNA-binding protein HuR. Mol. Cell 43, 340–52. https://doi.org/10.1016/j.molcel.2011.06.008 Lee, W., Tonelli, M., Markley, J.L., 2015. NMRFAM-SPARKY: Enhanced software for biomolecular NMR spectroscopy. Bioinformatics 31, 1325–1327. https://doi.org/10.1093/bioinformatics/btu830 Legnini, I., Morlando, M., Mangiavacchi, A., Fatica, A., Bozzoni, I., 2014. A feedforward regulatory loop between HuR and the long noncoding RNA linc-MD1 controls early phases of myogenesis. Mol. Cell 53, 506–14. https://doi.org/10.1016/j.molcel.2013.12.012 Léveillé, N., Elkon, R., Davalos, V., Manoharan, V., Hollingworth, D., Vrielink, J.O., Le Sage, C., Melo, C.A., Horlings, H.M., Wesseling, J., Ule, J., Esteller, M., Ramos, A., Agami, R., 2011. Selective inhibition of microRNA accessibility by RBM38 is required for p53 activity. Nat. Commun. 2. https://doi.org/10.1038/ncomms1519 Levy, N.S., Chung, S., Furneaux, H., Levy, A.P., 1998. Hypoxic stabilization of vascular endothelial growth factor mRNA by the RNA-binding protein HuR. J. Biol. Chem. 273, 6417–6423. https://doi.org/10.1074/jbc.273.11.6417 Li, H., Park, S., Kilburn, B., Jelinek, M.A., Henschen-Edman, A., Aswad, D.W., Stallcup, M.R., Laird-Offringa, I. a., 2002. Lipopolysaccharide-induced methylation of HuR, an mRNA- stabilizing protein, by CARM1. J. Biol. Chem. 277, 44623–44630. https://doi.org/10.1074/jbc.M206187200 Li, X., Lu, L., Bush, D.J., Zhang, X., Zheng, L., Suswam, E.A., King, P.H., 2009. Mutant copper- zinc superoxide dismutase associated with amyotrophic lateral sclerosis binds to adenine/uridine-rich stability elements in the vascular endothelial growth factor 3’- untranslated region. J. Neurochem. 108, 1032–44. https://doi.org/10.1111/j.1471- 4159.2008.05856.x Lim, N.S., Kozlov, G., Chang, T.C., Groover, O., Siddiqui, N., Volpon, L., De Crescenzo, G., Shyu, A. Bin, Gehring, K., 2006. Comparative peptide binding studies of the PABC domains from the ubiquitin-protein isopeptide ligase HYD and Poly(A)-binding Protein: Implications for HYD function. J. Biol. Chem. 281, 14376–14382. https://doi.org/10.1074/jbc.M600307200

135

References

López de Silanes, I., Zhan, M., Lal, A., Yang, X., Gorospe, M., 2004. Identification of a target RNA motif for RNA-binding protein HuR. Proc. Natl. Acad. Sci. U. S. A. 101, 2987–92. https://doi.org/10.1073/pnas.0306453101 Lunde, B.M., Moore, C., Varani, G., 2007. RNA-binding proteins: modular design for efficient function. Nat. Rev. Mol. Cell Biol. 8, 479–90. https://doi.org/10.1038/nrm2178 Ma, W., Cheng, S., Campbell, C., Wright, A., Furneaux, H., 1996. Cloning and Characterization of HuR , a Ubiquitously Expressed Elav-like Protein 271, 8144–8151. https://doi.org/10.1074/jbc.271.14.8144 Ma, W.J., Chung, S., Furneaux, H., 1997. The Elav-like proteins bind to AU-rich elements and to the poly(A) tail of mRNA. Nucleic Acids Res. 25, 3564–9. Mackereth, C.D., Madl, T., Bonnal, S., Simon, B., Zanier, K., Gasch, A., Rybin, V., Valcárcel, J., Sattler, M., 2011. Multi-domain conformational selection underlies pre-mRNA splicing regulation by U2AF. Nature 475, 408–11. https://doi.org/10.1038/nature10171 Mackereth, C.D., Sattler, M., 2012. Dynamics in multi-domain protein recognition of RNA. Curr. Opin. Struct. Biol. 22, 287–296. https://doi.org/10.1016/j.sbi.2012.03.013 Maier, J.A., Martinez, C., Kasavajhala, K., Wickstrom, L., Hauser, K.E., Simmerling, C., 2015. ff14SB: Improving the Accuracy of Protein Side Chain and Backbone Parameters from ff99SB. J. Chem. Theory Comput. 11, 3696–3713. https://doi.org/10.1021/acs.jctc.5b00255 Mangus, D.A., Evans, M.C., Jacobson, A., 2003. Poly(A)-binding proteins: multifunctional scaffolds for the posttranscriptional control of gene expression. Genome Biol. 4, 223. https://doi.org/10.1186/gb-2003-4-7-223 Maris, C., Dominguez, C., Allain, F.H.-T., 2005. The RNA recognition motif, a plastic RNA-binding platform to regulate post-transcriptional gene expression. FEBS J. 272, 2118–31. https://doi.org/10.1111/j.1742-4658.2005.04653.x Masliah, G., Maris, C., König, S.L., Yulikov, M., Aeschimann, F., Malinowska, A.L., Mabille, J., Weiler, J., Holla, A., Hunziker, J., Meisner‐Kober, N., Schuler, B., Jeschke, G., Allain, F.H., 2018. Structural basis of siRNA recognition by TRBP double‐stranded RNA binding domains. EMBO J. e97089. https://doi.org/10.15252/embj.201797089 Matsuda, S., Rouault, J.-P., Magaud, J.-P., Berthet, C., 2001. In search of a function for the TIS21/PC3/BTG1/TOB family. FEBS Lett. 497, 67–72. https://doi.org/https://doi.org/10.1016/S0014-5793(01)02436-X Mazroui, R., Di Marco, S., Clair, E., von Roretz, C., Tenenbaum, S. a, Keene, J.D., Saleh, M., Gallouzi, I.-E., 2008. Caspase-mediated cleavage of HuR in the cytoplasm contributes to pp32/PHAP-I regulation of apoptosis. J. Cell Biol. 180, 113–27. https://doi.org/10.1083/jcb.200709030 McCoy, A.J., Grosse-Kunstleve, R.W., Adams, P.D., Winn, M.D., Storoni, L.C., Read, R.J., 2007. Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674. https://doi.org/10.1107/S0021889807021206 Meisner, N.-C., Hackermüller, J., Uhl, V., Aszódi, A., Jaritz, M., Auer, M., 2004. mRNA openers and closers: modulating AU-rich element-controlled mRNA stability by a molecular switch in mRNA secondary structure. Chembiochem 5, 1432–47. https://doi.org/10.1002/cbic.200400219 Meisner, N.-C., Hintersteiner, M., Seifert, J.-M., Bauer, R., Benoit, R.M., Widmer, A., Schindler, T., Uhl, V., Lang, M., Gstach, H., Auer, M., 2009. Terminal adenosyl transferase activity of posttranscriptional regulator HuR revealed by confocal on-bead screening. J. Mol. Biol. 386, 435–50. https://doi.org/10.1016/j.jmb.2008.12.020 Meng, Z., King, P.H., Nabors, B.L., Jackson, N.L., Chen, C.Y., Emanuel, P.D., Blume, S.W., 2005.

136

References

The ELAV RNA-stability factor HuR binds the 5’-untranslated region of the human IGF-IR transcript and differentially represses cap-dependent and IRES-mediated translation. Nucleic Acids Res. 33, 2962–2979. https://doi.org/10.1093/nar/gki603 Ming, X., Stoecklin, G., Lu, M., Looser, R., Lu, M.I.N., Moroni, C., 2001. Parallel and Independent Regulation of Interleukin-3 mRNA Turnover by Phosphatidylinositol 3-Kinase and p38 Mitogen-Activated Protein Kinase Parallel and Independent Regulation of Interleukin-3 mRNA Turnover by Phosphatidylinositol 3-Kinase and p38 Mitoge. Mol. Cell. Biol. 21, 5778– 5789. https://doi.org/10.1128/MCB.21.17.5778 Misquitta, C.M., Iyer, V.R., Werstiuk, E.S., Grover, A.K., 2001. The role of 3′-untranslated region (3′-UTR) mediated mRNA stability in cardiovascular pathophysiology. Mol. Cell. Biochem. 224, 53–67. https://doi.org/10.1023/A:1011982932645 Mukherjee, K., Ghoshal, B., Ghosh, S., Chakrabarty, Y., Shwetha, S., Das, S., Bhattacharyya, S.N., 2016. Reversible HuRmicroRNA binding controls extracellular export of miR122 and augments stress response. EMBO Rep. 17, 11841203. https://doi.org/10.15252/embr Mukherjee, N., Corcoran, D.L., Nusbaum, J.D., Reid, D.W., Georgiev, S., Hafner, M., Ascano, M., Tuschl, T., Ohler, U., Keene, J.D., 2011. Integrative regulatory mapping indicates that the RNA-binding protein HuR couples pre-mRNA processing and mRNA stability. Mol. Cell 43, 327–39. https://doi.org/10.1016/j.molcel.2011.06.007 Murata, K., Wolf, M., 2018. Cryo-electron microscopy for structural analysis of dynamic biological macromolecules. Biochim. Biophys. Acta - Gen. Subj. 1862, 324–334. https://doi.org/10.1016/j.bbagen.2017.07.020 Muto, Y., Yokoyama, S., 2012. Structural insight into RNA recognition motifs: Versatile molecular Lego building blocks for biological systems. Wiley Interdiscip. Rev. RNA 3, 229–246. https://doi.org/10.1002/wrna.1107 Oberstrass, F.C., Auweter, S.D., Erat, M., Hargous, Y., Henning, A., Wenter, P., Reymond, L., Amir-Ahmady, B., Pitsch, S., Black, D.L., Allain, F.H.-T., 2005. Structure of PTB bound to RNA: specific binding and implications for splicing regulation. Science 309, 2054–7. https://doi.org/10.1126/science.1114066 Okochi, K., Suzuki, T., Inoue, J.I., Matsuda, S., Yamamoto, T., 2005. Interaction of anti-proliferative protein Tob with poly(A)-binding protein and inducible poly(A)-binding protein: Implication of Tob in translational control. Genes to Cells 10, 151–163. https://doi.org/10.1111/j.1365- 2443.2005.00826.x Otwinowski, Z., Minor, W., 1997. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326. https://doi.org/10.1016/S0076-6879(97)76066-X Parker, R., 2012. RNA degradation in Saccharomyces cerevisae. Genetics 191, 671–702. https://doi.org/10.1534/genetics.111.137265 Pérez, A., Marchán, I., Svozil, D., Sponer, J., Cheatham, T.E., Laughton, C.A., Orozco, M., 2007. Refinement of the AMBER force field for nucleic acids: Improving the description of α/γ conformers. Biophys. J. 92, 3817–3829. https://doi.org/10.1529/biophysj.106.097782 Prévôt, D., Voeltzel, T., Birot, A.M., Morel, A.P., Rostan, M.C., Magaud, J.P., Corbo, L., 2000. The leukemia-associated protein Btg1 and the p53-regulated protein Btg2 interact with the homeoprotein Hoxb9 and enhance its transcriptional activation. J. Biol. Chem. 275, 147–153. https://doi.org/10.1074/jbc.275.1.147 Protter, D.S.W., Parker, R., 2016. Principles and Properties of Stress Granules. Trends Cell Biol. 26, 668–79. https://doi.org/10.1016/j.tcb.2016.05.004 Rideau, A.P., Gooding, C., Simpson, P.J., Monie, T.P., Lorenz, M., Hüttelmaier, S., Singer, R.H., Matthews, S., Curry, S., Smith, C.W.J., 2006. A peptide motif in Raver1 mediates splicing repression by interaction with the PTB RRM2 domain. Nat. Struct. Mol. Biol. 13, 839–848.

137

References

https://doi.org/10.1038/nsmb1137 Rissland, O.S., 2017. The organization and regulation of mRNA–protein complexes. Wiley Interdiscip. Rev. RNA 8, 1–17. https://doi.org/10.1002/wrna.1369 Rossi, P., Swapna, G.V.T., Huang, Y.J., Aramini, J.M., Anklin, C., Conover, K., Hamilton, K., Xiao, R., Acton, T.B., Ertekin, A., Everett, J.K., Montelione, G.T., 2010. A microscale protein NMR sample screening pipeline. J. Biomol. NMR 46, 11–22. https://doi.org/10.1007/s10858-009- 9386-z Rouault, J.P., Prévôt, D., Berthet, C., Birott, A.M., Billaud, M., Magaud, J.P., Corbo, L., 1998. Interaction of BTG1 and p53-regulated BTG2 gene products with mCaf1, the murine homolog of a component of the yeast CCR4 transcriptional regulatory complex. J. Biol. Chem. 273, 22563–22569. https://doi.org/10.1074/jbc.273.35.22563 Rouault, J.P., Rimokh, R., Tessa, C., Paranhos, G., Ffrench, M., Duret, L., Garoccio, M., Germain, D., Samarut, J., Magaud, J.P., 1992. BTG1, a member of a new family of antiproliferative genes. EMBO J. 11, 1663–70. Safaee, N., Kozlov, G., Noronha, A.M., Xie, J., Wilds, C.J., Gehring, K., 2012. Interdomain allostery promotes assembly of the poly(A) mRNA complex with PABP and eIF4G. Mol. Cell 48, 375– 86. https://doi.org/10.1016/j.molcel.2012.09.001 Sagnol, S., Yang, Y., Bessin, Y., Allemand, F., Hapkova, I., Notarnicola, C., Guichou, J.-F., Faure, S., Labesse, G., de Santa Barbara, P., 2014. Homodimerization of RBPMS2 through a new RRM-interaction motif is necessary to control smooth muscle plasticity. Nucleic Acids Res. 42, 10173–84. https://doi.org/10.1093/nar/gku692 Sawazaki, R., Imai, S., Yokogawa, M., Hosoda, N., Hoshino, S.-I., Mio, M., Mio, K., Shimada, I., Osawa, M., 2018. Characterization of the multimeric structure of poly(A)-binding protein on a poly(A) tail. Sci. Rep. 8, 1–13. https://doi.org/10.1038/s41598-018-19659-6 Scheiba, R.M., de Opakua, A.I., Díaz-Quintana, A., Cruz-Gallardo, I., Martínez-Cruz, L. a, Martínez-Chantar, M.L., Blanco, F.J., Díaz-Moreno, I., 2014. The C-terminal RNA binding motif of HuR is a multi-functional domain leading to HuR oligomerization and binding to U- rich RNA targets. RNA Biol. 11, 1250–61. https://doi.org/10.1080/15476286.2014.996069 Schlundt, A., Tants, J.N., Sattler, M., 2017. Integrated structural biology to unravel molecular mechanisms of protein-RNA recognition. Methods 118–119, 119–136. https://doi.org/10.1016/j.ymeth.2017.03.015 Shen, Y., Delaglio, F., Cornilescu, G., Bax, A., 2009. TALOS+: A hybrid method for predicting protein backbone torsion angles from NMR chemical shifts. J. Biomol. NMR 44, 213–223. https://doi.org/10.1007/s10858-009-9333-z Smart, O.S., Womack, T.O., Flensburg, C., Keller, P., Paciorek, W., Sharff, A., Vonrhein, C., Bricogne, G., 2012. Exploiting structure similarity in refinement: automated NCS and target- structure restraints in BUSTER. Acta Crystallogr. Sect. D Biol. Crystallogr. 68, 368–380. https://doi.org/10.1107/S0907444911056058 Smith, G.R., Sternberg, M.J.E., 2002. Prediction of protein – protein interactions by docking methods. Elsevier Sci. 12, 28–35. Smyth, M.S., Martin, J.H., 2000. x ray crystallography. Mol. Pathol. 53, 8–14. https://doi.org/10.1136/mp.53.1.8 Sokoloski, K.J., Dickson, A.M., Chaskey, E.L., Garneau, N.L., Wilusz, C.J., Wilusz, J., 2010. Sindbis virus usurps the cellular HuR protein to stabilize its transcripts and promote productive infections in mammalian and mosquito cells. Cell Host Microbe 8, 196–207. https://doi.org/10.1016/j.chom.2010.07.003 Soufari, H., Mackereth, C.D., 2016. Conserved binding of GCAC motifs by MEC-8, couch potato

138

References

and the RBPMS protein family. RNA 105–109. https://doi.org/10.1261/rna.059733.116 Šponer, J., Bussi, G., Krepl, M., Banáš, P., Bottaro, S., Cunha, R.A., Gil-Ley, A., Pinamonti, G., Poblete, S., Jurečka, P., Walter, N.G., Otyepka, M., 2018. RNA Structural Dynamics As Captured by Molecular Simulations: A Comprehensive Overview. Chem. Rev. acs.chemrev.7b00427. https://doi.org/10.1021/acs.chemrev.7b00427 Šponer, J., Krepl, M., Banáš, P., Kührová, P., Zgarbová, M., Jurečka, P., Havrila, M., Otyepka, M., 2017. How to understand atomistic molecular dynamics simulations of RNA and protein–RNA complexes? Wiley Interdiscip. Rev. RNA 8, 1–17. https://doi.org/10.1002/wrna.1405 Srikantan, S., Gorospe, M., 2012. HuR function in disease. Front. Biosci. 17, 189–205. https://doi.org/10.2741/3921 Srikantan, S., Tominaga, K., Gorospe, M., 2012. Functional interplay between RNA-binding protein HuR and microRNAs. Curr. Protein Pept. Sci. 13, 372–9. Struckmann, K., Schraml, P., Simon, R., Elmenhorst, K., Mirlacher, M., Kononen, J., Moch, H., 2004. Impaired expression of the cell cycle regulator BTG2 is common in clear cell renal cell carcinoma. Cancer Res 64, 1632–1638. Stupfler, B., Birck, C., Séraphin, B., Mauxion, F., 2016. BTG2 bridges PABPC1 RNA-binding domains and CAF1 deadenylase to control cell proliferation. Nat. Commun. 7, 10811. https://doi.org/10.1038/ncomms10811 Taylor, G., 2003. The phase problem. Acta Crystallogr. - Sect. D Biol. Crystallogr. 59, 1881–1890. https://doi.org/10.1107/S0907444903017815 Teplova, M., Farazi, T.A., Tuschl, T., Patel, D.J., 2016. Structural basis underlying CAC RNA recognition by the RRM domain of dimeric RNA-binding protein RBPMS. Q. Rev. Biophys. 49, e1. https://doi.org/10.1017/S0033583515000207 Toba, G., White, K., 2008. The third RNA recognition motif of Drosophila ELAV protein has a role in multimerization. Nucleic Acids Res. 36, 1390–9. https://doi.org/10.1093/nar/gkm1168 Trewhella, J., 2016. Small-angle scattering and 3D structure interpretation. Curr. Opin. Struct. Biol. 40, 1–7. https://doi.org/10.1016/j.sbi.2016.05.003 Turnbull, W.B., Daranas, A.H., 2003. On the Value of c: Can Low Affinity Systems Be Studied by Isothermal Titration Calorimetry? J. Am. Chem. Soc. 125, 14859–14866. https://doi.org/10.1021/ja036166s Uversky, V.N., 2017. Intrinsically disordered proteins in overcrowded milieu: Membrane-less organelles, phase separation, and intrinsic disorder. Curr. Opin. Struct. Biol. 44, 18–30. https://doi.org/10.1016/j.sbi.2016.10.015 Vagin, A.A., Steiner, R.A., Lebedev, A.A., Potterton, L., McNicholas, S., Long, F., Murshudov, G.N., 2004. REFMAC5 dictionary: Organization of prior chemical knowledge and guidelines for its use. Acta Crystallogr. Sect. D Biol. Crystallogr. 60, 2184–2195. https://doi.org/10.1107/S0907444904023510 Van Der Giessen, K., Di-Marco, S., Clair, E., Gallouzi, I.E., 2003. RNAi-mediated HuR Depletion Leads to the Inhibition of Muscle Cell Differentiation. J. Biol. Chem. 278, 47119–47128. https://doi.org/10.1074/jbc.M308889200 Van Zundert, G.C.P., Rodrigues, J.P.G.L.M., Trellet, M., Schmitz, C., Kastritis, P.L., Karaca, E., Melquiond, A.S.J., Van Dijk, M., De Vries, S.J., Bonvin, A.M.J.J., 2016. The HADDOCK2.2 Web Server: User-Friendly Integrative Modeling of Biomolecular Complexes. J. Mol. Biol. 428, 720–725. https://doi.org/10.1016/j.jmb.2015.09.014 Venter, J.C.,et al., 2001. The sequence of the human genome. Science (80-. ). 291, 1304–1351. https://doi.org/10.1126/science.1058040

139

References

Vitali, F., Henning, A., Oberstrass, F.C., Hargous, Y., Auweter, S.D., Erat, M., Allain, F.H.T., 2006. Structure of the two most C-terminal RNA recognition motifs of PTB using segmental isotope labeling. Embo J. 25, 150–162. https://doi.org/Doi 10.1038/Sj.Emboj.7600911 Wahle, E., Winkler, G.S., 2013. RNA decay machines: Deadenylation by the Ccr4-Not and Pan2- Pan3 complexes. Biochim. Biophys. Acta - Gene Regul. Mech. 1829, 561–570. https://doi.org/10.1016/j.bbagrm.2013.01.003 Wakamatsu, Y., Weston, J.A., 1997. Sequential expression and role of Hu RNA-binding proteins during neurogenesis 3460, 3449–3460. Wang, H., Zeng, F., Liu, Q., Liu, H., Liu, Z., Niu, L., Teng, M., Li, X., 2013. The structure of the ARE-binding domains of Hu antigen R (HuR) undergoes conformational changes during RNA binding. Acta Crystallogr. D. Biol. Crystallogr. 69, 373–80. https://doi.org/10.1107/S0907444912047828 Wang, I., Hennig, J., Jagtap, P.K.A., Sonntag, M., Valcárcel, J., Sattler, M., 2014. Structure, dynamics and RNA binding of the multi-domain splicing factor TIA-1. Nucleic Acids Res. 42, 5949–5966. https://doi.org/10.1093/nar/gku193 Wang, J., Guo, Y., Chu, H., Guan, Y., Bi, J., Wang, B., 2013. Multiple Functions of the RNA-Binding Protein HuR in Cancer Progression, Treatment Responses and Prognosis. Int. J. Mol. Sci. 14, 10015–41. https://doi.org/10.3390/ijms140510015 Wang, W., Caldwell, M.C., Lin, S., Furneaux, H., Gorospe, M., 2000. HuR regulates cyclin A and cyclin B1 mRNA stability during cell proliferation. EMBO J. 19, 2340–2350. https://doi.org/10.1093/emboj/19.10.2340 Wang, W., Furneaux, H., Cheng, H., Caldwell, M.C., Hutter, D., Liu, Y., Holbrook, N., Gorospe, M., 2000. HuR Regulates p21 mRNA Stabilization by UV Light. Mol. Cell. Biol. 20, 760–769. https://doi.org/10.1128/MCB.20.3.760-769.2000 Wang, X., Tanaka Hall, T.M., 2001. Structural basis for recognition of AU-rich element RNA by the HuD protein. Nat. Struct. Biol. 8, 141–5. https://doi.org/10.1038/84131 Waudby, C.A., Ramos, A., Cabrita, L.D., Christodoulou, J., 2016. Two-Dimensional NMR Lineshape Analysis. Sci. Rep. 6, 1–8. https://doi.org/10.1038/srep24826 Webster, M.W., Chen, Y.H., Stowell, J.A.W., Alhusaini, N., Sweet, T., Graveley, B.R., Coller, J., Passmore, L.A., 2018. mRNA Deadenylation Is Coupled to Translation Rates by the Differential Activities of Ccr4-Not Nucleases. Mol. Cell 70, 1089–1100.e8. https://doi.org/10.1016/j.molcel.2018.05.033 Wells, S.E., Hillner, P.E., Vale, R.D., Sachs, A.B., 1998. Circularization of mRNA by Eukaryotic Translation Initiation Factors. Mol. Cell 2, 135–140. https://doi.org/10.1016/S1097- 2765(00)80122-7 Williamson, M.P., 2013. Using chemical shift perturbation to characterise ligand binding. Prog. Nucl. Magn. Reson. Spectrosc. 73, 1–16. https://doi.org/10.1016/j.pnmrs.2013.02.001 Wlodawer, A., Minor, W., Dauter, Z., Jaskolski, M., Physics, B., 2008. Protein crystallography for non-crystallographers, or how to get the (but not more) from published macromolecular structures. Febs J 275, 1–21. https://doi.org/10.1111/j.1742-4658.2007.06178.x. Wu, L., Fan, J., Belasco, J.G., 2006. MicroRNAs direct rapid deadenylation of mRNA. Proc. Natl. Acad. Sci. 103, 4034–4039. https://doi.org/10.1073/pnas.0510928103 Yaman, I., Fernandez, J., Sarkar, B., Schneider, R.J., Snider, M.D., Nagy, L.E., Hatzoglou, M., 2002. Nutritional control of mRNA stability is mediated by a conserved AU-rich element that binds the cytoplasmic shuttling protein HuR. J. Biol. Chem. 277, 41539–46. https://doi.org/10.1074/jbc.M204850200 Yang, X., Morita, M., Wang, H., Suzuki, T., Yang, W., Luo, Y., Zhao, C., Yu, Y., Bartlam, M.,

140

References

Yamamoto, T., Rao, Z., 2008. Crystal structures of human BTG2 and mouse TIS21 involved in suppression of CAF1 deadenylase activity. Nucleic Acids Res. 36, 6872–6881. https://doi.org/10.1093/nar/gkn825 Yi, H., Park, J., Ha, M., Lim, J., Chang, H., Kim, V.N., 2018. PABP Cooperates with the CCR4- NOT Complex to Promote mRNA Deadenylation and Block Precocious Decay. Mol. Cell 70, 1081–1088.e5. https://doi.org/10.1016/j.molcel.2018.05.009 Yoon, J.-H., Abdelmohsen, K., Srikantan, S., Guo, R., Yang, X., Martindale, J.L., Gorospe, M., 2013. Tyrosine phosphorylation of HuR by JAK3 triggers dissociation and degradation of HuR target mRNAs. Nucleic Acids Res. 38, 1–13. https://doi.org/10.1093/nar/gkt903 Yoshida, Y., Nakamura, T., Komoda, M., Satoh, H., Suzuki, T., Tsuzuku, J.K., Miyasaka, T., Yoshida, E.H., Umemori, H., Kunisaki, R.K., Tani, K., Ishii, S., Mori, S., Suganuma, M., Noda, T., Yamamoto, T., 2003. Mice lacking a transcriptional corepressor Tob are predisposed to cancer. Genes Dev. 17, 1201–1206. https://doi.org/10.1101/gad.1088003 Young, L.E., Moore, A.E., Sokol, L., Meisner-Kober, N., Dixon, D.A., 2012. The mRNA stability factor HuR inhibits microRNA-16 targeting of COX-2. Mol. Cancer Res. 10, 167–80. https://doi.org/10.1158/1541-7786.MCR-11-0337 Zgarbová, M., Otyepka, M., Šponer, J., Mládek, A., Banáš, P., Cheatham, T.E., Jurečka, P., 2011. Refinement of the Cornell et al. Nucleic acids force field based on reference quantum chemical calculations of glycosidic torsion profiles. J. Chem. Theory Comput. 7, 2886–2902. https://doi.org/10.1021/ct200162x Zhang, H., Curreli, F., Zhang, X., Bhattacharya, S., Waheed, A. a, Cooper, A., Cowburn, D., Freed, E.O., Debnath, A.K., 2011. Antiviral activity of a-helical stapled peptides designed from the HIV-1 capsid dimerization domain. Retrovirology 8, 28. https://doi.org/10.1186/1742-4690-8- 28 Zhang, W., Zeng, F., Liu, Y., Zhao, Y., Lv, H., Niu, L., Teng, M., Li, X., 2013. Crystal structures and RNA-binding properties of the RNA recognition motifs of heterogeneous nuclear ribonucleoprotein L: Insights into its roles in alternative splicing regulation. J. Biol. Chem. 288, 22636–22649. https://doi.org/10.1074/jbc.M113.463901 Zhu, H., Zhou, H.L., Hasman, R.A., Lou, H., 2007. Hu proteins regulate polyadenylation by blocking sites containing U-rich sequences. J. Biol. Chem. 282, 2203–2210. https://doi.org/10.1074/jbc.M609349200 Zou, T., Mazan-Mamczarz, K., Rao, J.N., Liu, L., Marasa, B.S., Zhang, A.H., Xiao, L., Pullmann, R., Gorospe, M., Wang, J.Y., 2006. Polyamine depletion increases cytoplasmic levels of RNA-binding protein HuR leading to stabilization of nucleophosmin and p53 mRNAs. J. Biol. Chem. 281, 19387–19394. https://doi.org/10.1074/jbc.M602344200 Zucal, C., D’Agostino, V., Loffredo, R., Mantelli, B., Thongon, N., Lal, P., Latorre, E., Provenzani, A., 2015. Targeting the Multifaceted HuR Protein, Benefits and Caveats. Curr. Drug Targets 16, 1–17. https://doi.org/10.2174/1389450116666150223163632

141

7. Curriculum Vitae

Name Nina Ripin Date of birth 06.08.1986 Citizenship German

Address: Institute of Molecular Biology and Biophysics ETH, CH-8093 Zürich, Switzerland Phone: +41 44 633 07 21 Email: [email protected]

EDUCATION

Since 11/2012 ETH Zürich  PhD candidate of the Life Science Zürich Graduate School- Biomolecular Structure and Mechanism program (BSM)

10/2007-09/2012 Goethe University Frankfurt, Germany  diploma in biochemistry (MSc equivalent)

07/2006-07/2007 Metropolitan Community College - Penn Valley, Kansas City Missouri, USA  classes in biology and chemistry

LAB EXPERIENCE

Since 10/2014 ETH Zürich  PhD Thesis, lab of Prof. Dr. Frédéric Allain (ETH) Department of Biology, Institute of Molecular Biology and Biophysics, ETH Zurich Switzerland  Main project “Molecular basis for AU-rich element recognition and dimerization by the HuR C-terminal RRM domain”  Project: “Structural and Functional Characterization of the HuR- RBM38 Interaction”  Project: “Characterization of molecular interactions between BTG2 and PABPC1 RRMs to recruit CAF1 deadenylase”

11/2012-10/2014 Novartis Institutes of BioMedical Research (NIBR), Basel  External PhD Thesis  Project: “Structural and functional characterization of the adenosyl transferase and 3’-5’exonuclease activity of RNA binding proteins”

04/1012-09/2012 Goethe University Frankfurt, Germany  Diploma thesis (MSc equivalent), lab of Prof. Dr. Volker Dötsch Department of Biophysical Chemistry  Project: “Investigation of the interactions between the transcription factor WT1 and proteins from the cytosolic iron sulfur cluster assembly (CIA) system”

04/2011-11/2011 Sydney University, Australia  Internship in the lab of Prof. Dr. Jacqueline Matthews

References

Department of Structural Biology  Project: “Crystallization of the transcription factor complex CEH- 14/LIM-7 playing a role in the developing nervous system”  Project: “Structural and functional characterization of transcription factor Gata-1, a regulator of erythropoiesis”

02/2010-04/2011 Goethe University Frankfurt, Germany  Student research assistant in the lab of Prof. Dr. Volker Dötsch Department of Biophysical Chemistry  Project: “Characterization of molecular interactions between ACP and halogenase domains in the Curacin A polyketide synthase”

RESEARCH SKILLS

Lab work:  Extensive experience in Molecular Biology and Biochemistry: Cloning, Expression and Purification of proteins, FPLC, in vitro RNA Transcription, HPLC, Isothermal Titration Calorimetry (ITC), Fluorescence Anisotropy, Enzymatic assays, gel shifts, radioactive work  NMR (protein and RNA titration experiments, protein backbone assignments, dynamics/ relaxation experiments)  Crystallization, X-Ray Structure Determination (by MR)  Basic cell culture (HEK293T, Hela, Huh7 cells), qPCR, Luciferase assays, stressing assays, basic fluorescence microscopy

Computer skills:  MS Office, Adobe Illustrator, In Design, Origin  NMR: Topspin & Sparky  X-Ray Crystallography: Coot, PyMol & Phenix  bioinformatic applications (clustalw, blast, etc.)

Language skills: German/ Russian (native language), English (fluent)

ACADEMIC AWARDS AND SCHOLARSHIPS

11/2012-10/2015 Marie Curie Initial Training Networks- “RNPnet” Fellowship http://www.rnpnet.ethz.ch

04/2011-11/2011 “PROMOS (DAAD)” Scholarship – Internship at University of Sydney

2011-2012 “Deutschlandstipendium” Student Scholarship from the German Federal Ministry of Education and Research

TEACHING EXPERIENCE

2016-2017 Supervision of two ETH master students (3-4 month lab internships) (lab of Prof. Dr. Frédéric Allain, ETH Zürich, Switzerland)

143

References

2014 Co-Supervision of one apprentice – around 3 months (Novartis Institutes of BioMedical Research (NIBR), Basel, Switzerland)

2012 Co-Supervision of one bachelor student for a 3 month internship (lab of Prof. Dr. Volker Dötsch; Goethe University Frankfurt, Germany)

10/2009- 3/2010 Teaching assistant for metabolism seminar for 1st grade students (Department of Biochemistry, Goethe University Frankfurt, Germany)

UNIVERSITY SERVICE

09/2015 – 10/2018 Board member of AVETH (Academic Association of Scientific Staff at the ETH Zürich); Coordinator of the AVETH-Helpdesk; http://www.aveth.ethz.ch;

03/2015-04/2018 Until 03/2016: Vice-president of AMB (Scientific Staff Association at Department of Biology, ETH Zürich) http://www.amb.ethz.ch Until 04/2018: President of AMB

01/2015-01/2018 Organizer of the “NMR Winter Retreat of Protein-RNA Interactions” – Annual meeting with 13 academic groups (~65 participants)

10/2007-09/2012 Board member of the Student Association of Biochemistry, Goethe University Frankfurt, Germany

PUBLICATIONS

Molecular basis for AU-rich element recognition and dimerization by the HuR C-terminal RRM domain, under revision Ripin N, Boudet J, Duszczyk MM, Hinniger A , Faller M, Krepl M, Gadi A, Schneider RJ, Šponer J, Meisner-Kober NC, Allain FH

Systems NMR: reconstruction of biomolecular networks by NMR and mathematical modeling, in preparation Nikolaev Y, Iber D, Picotti P, Ripin N, Soste M, Allain FH

Aromatic side-chain conformational switch on the surface of the RNA Recognition Motif enables RNA discrimination Diarra Dit Konté N, Krepl M, Damberger FF, Ripin N, Duss O, Šponer J, Allain FH. Nat Commun. 2017; 8(1):654. doi: 10.1038/s41467-017-00631-3.

Interactions between LHX3- and ISL1-family LIM-homeodomain transcription factors are conserved in Caenorhabditis elegans Bhati M, Llamosas E, Jacques DA, Jeffries CM, Dastmalchi S, Ripin N, Nicholas HR, Matthews JM. Sci Rep. 2017 ;7(1):4579. doi: 10.1038/s41598-017-04587-8.

GATA1 directly mediates interactions with closely spaced pseudopalindromic but not distantly spaced double GATA sites on DNA Wilkinson-White L, Lester KL, Ripin N, Jacques DA, Mitchell Guss J, Matthews JM. Protein Sci. 2015, 24(10):1649-59. doi: 10.1002/pro.2760.

Characterization of molecular interactions between ACP and halogenase domains in the Curacin A polyketide synthase Busche A, Gottstein D, Hein C, Ripin N, Pader I, Tufar P, Eisman EB, Gu L, Walsh CT, Sherman DH, Löhr F, Güntert P, Dötsch V. ACS Chem Biol. 2012; 7(2):378-86. doi: 10.1021/cb200352q

144