דוקטור לפילוסופיה Doctor of Philosophy

עבודת גמר )תזה( לתואר Thesis for the degree דוקטור לפילוסופיה Doctor of Philosophy

מוגשת למועצה המדעית של Submitted to the Scientific Council of the מכון ויצמן למדע Weizmann Institute of Science רחובות, ישראל Rehovot, Israel

מאת By אורה חיימוב Ora Haimov

חקר מנגנון התרגום הייחודי המתווך על-ידי TISU

Elucidating the unique translation initiation mechanism directed by TISU element

מנחה: :Advisor פרופ' רבקה דיקשטיין Prof.Rivka Dikstein

כסלו תשע"ז December, 2016 Acknowledgments

First and foremost, I would like to express my deep gratitude and appreciation to my advisor, Prof. Rivka Dikstein, for her excellent guidance, passion for science, critical and creative thinking, dedication, encouragement, support and mostly for being humane. Rivka, the things I learned from you will remain with me throughout my entire career.

I would also like to thank my former and current colleagues for their scientific assistance, good advice and mostly for their friendship. Special thanks to Dr. Anat Bahat for being our ultimate staff scientist and a good friend, to Shaked Ashkenazi for her good spirit and contribution to the lab atmosphere, to Ana Tamarkin Ben-harush for many wonderful moments that I will always cherish, to Dr. Hadar Sinvani for collaboration and to Dr. Rafi Emmanuel for the initiation of the “TISU Project”. I owe many thanks to Anna Uzonyi and Adi Jacob, two young and talented scientists who contributed to this study.

Last but not least, I wish to thank my family, especially to my parents, Mordechai and Geula, from whom I learned the value of knowledge and education, to my husband, Reuven, for his love, and to my four precious children, Ariel, Eitan, Naama and Yehonatan, who made this whole period challenging, meaningful and much more satisfying.

Table of Contents

Abbreviations ...... 4

Abstract ...... 5

Introduction ...... 6

The translation initiation process ...... 6

Regulation of translation initiation by eIF1 and eIF1A ...... 10

TISU- The Translation Initiator of Short 5’UTR ...... 11

Objectives ...... 13

Materials and Methods ...... 14

Results ...... 25

Analysis of TISU element sequence requirements ...... 25

The roles of eIFs in the regulation of translation initiation fidelity and ribosomal scanning ...... 27

The effect of eIF1A and eIF1 overexpression on translation initiation fidelity and ribosomal scanning ...... 28

The effect of eIF1 and eIF1A overexpression on H2B mRNA translation ...... 33

The effect of eIF1A and eIF1 knockdown on translation initiation fidelity and ribosomal scanning ...... 35 eIF1 and eIF1A protein levels are cell cycle regulated ...... 40 eIF1 and eIF1A protein levels in different mouse cell types and Human tissue culture cell types ...... 41

The effect of eIF3c knockdown on translation initiation fidelity and ribosomal scanning ...... 42

The role of eIF4A in translation initiation fidelity and ribosomal scanning ...... 44

The effect of eIF5 knockdown on translation initiation fidelity and ribosomal scanning ...... 48 eIF4GI has differential effects on translation initiation fidelity and ribosomal scanning ………………………………………………………………………………...50 Analysis of eIF1-eIF4GI interaction...... 54

The role of eIF4E in translation initiation fidelity and ribosomal scanning ...... 58

Interaction of eIF4E and eIF1 with eIF4GI is mutually exclusive…………………..61

Discussion ...... 64

Bibliography ...... 74

Declaration ...... 81

List of publications ...... 81

Abbreviations

TISU- Translation Initiator of Short 5’ UTR eIF- Eukaryotic translation initiation factor Met-tRNAi- Initiator methionyl tRNA m7G cap- 7 methylguanosine cap UTR- Untranslated region PIC- Preinitiation complex TC- Ternary complex PABP-Poly A binding protein ORF- Open reading frame US-AUG- Upstream AUG DS-AUG-Downstream AUG RL-Renilla Luciferase FL- Firefly Luciferase 4E-BP- eIF4E- Binding Protein

Abstract

Translation Initiator of Short 5’UTR (TISU) is a unique regulatory element of both transcription and translation initiation. The core of the element has an invariable ATG sequence, and it directs efficient translation initiation from extremely short 5’UTR in a cap-dependent manner, but without scanning. The major goal of this study is to investigate the sequence and the factor requirements of the non-canonical mechanism of TISU. Comprehensive mutagenesis established TISU as a robust translation initiator of short 5’UTR mRNAs and revealed that all TISU-AUG flanking nucleotides contribute to its translational strength; however, position -3, +4, +5 and +6 are critical for its high- fidelity. We tested the role of several eIFs on translation of mRNAs with various AUG contexts and 5’UTR lengths using overexpression and knockdown experiments. Our findings show that the effects of eIF1A, eIF1, eIF4GI and eIF4E are not general but dependent on mRNA characteristics. For example, elevated levels of eIF1A promoted translation from cap-proximal AUG within canonical AUG context, while eIF1 overexpression exhibited an antagonistic effect. Regarding TISU translation, overexpression of either eIF1A or eIF1 had no significant effect. Interestingly, knockdown of eIF1, eIF1A, eIF4GI or eIF4E caused a dramatic decrease in TISU mediated translation, whereas translation from cap-proximal canonical AUG was almost unaffected, suggesting that TISU activity is highly dependent on those eIFs. The knockdown experiments also uncovered the regulatory roles that eIF4GI and eIF4E play in scanning. We show that eIF1 specifically interacts with the middle conserved domain of eIF4GI. Furthermore, our results suggest the existence of a novel eIF4E binding site which is also located in the middle conserved domain of eIF4GI. Interestingly, the interaction of eIF4GI with eIF4E and eIF1 is mutually exclusive. An eIF1 mutant impaired in eIF4GI binding fails to promote scanning but facilitates TISU activity, suggesting that eIF1-eIF4GI binding is required for scanning but not for TISU. Strikingly, eIF4E-eIF4GI antagonizes the scanning promoting activity of eIF1-eIF4GI. Therefore, we propose that the entrance into the scanning phase necessitates an exchange of eIF4GI from eIF4E to eIF1. These findings expand our understanding of scanning-dependent and independent translation initiation process and uncover important mechanistic aspects of the unique translation initiation directed by TISU.

Introduction

The translation initiation process Translation of mRNA to protein in eukaryotes requires a complex apparatus consisting of mRNA, ribosomes, tRNAs and protein factors. mRNA translation is a cyclic process which can be divided into initiation, elongation, termination and recycling. Within this framework the initiation stage is considered as a major regulatory target. This stage involves different initiation factors (eIFs) and different mRNA features, all of which play important roles in translational control. The predominant form of eukaryotic translation initiation is the canonical cap-dependent scanning mechanism (Aitken & Lorsch, 2012; Hinnebusch, 2014; Marintchev & Wagner, 2004; Sonenberg & Hinnebusch, 2009) which depends on the m7G cap structure, present at the 5ʹ-end of the mRNA, and on ribosomal scanning. Translation initiation begins with formation of the 43S preinitiation complex (PIC) that is assembled from the ternary complex (eIF2–GTP–Met–tRNAi), several initiation factors (eIFs) that include eIF1, 1A, 3 and 5 and the 40S small ribosomal subunit (Fig. 1). This 43S PIC, then attaches the cap-proximal region of activated mRNAs. mRNA activation is driven by eIF4F which is composed of eIF4E- the cap-binding protein, eIF4GI -a scaffold protein, and eIF4A- a DEAD-box ATPase and RNA dependent helicase. The first step of activation occurs when eIF4E binds the m7G cap structure. Subsequently, eIF4A unwinds the cap-proximal secondary structures (Fig. 1). The role of eIF4GI is to form a ‘closed-loop’ mRNP complex consisting of eIF4F, mRNA and PABP [poly(A) binding proteins] since it interacts with the mRNA body directly and with the m7G cap and 3′- poly(A) tail indirectly through interactions with eIF4E and PABPs (Kaye, Emmett, Merrick, & Jankowsky, 2009; Park et al., 2010; Wells, Hillner, Vale, & Sachs, 1998; Yanagiya et al., 2009). eIF4GI also interacts with eIF4A and strongly enhances its helicase activity (Marintchev, 2013; Marintchev et al., 2009; Oberer, Marintchev, & Wagner, 2005; Özeş, Feoktistova, Avanzino, & Fraser, 2011; Parsyan et al., 2011; Rogers, Richter, Lima, & Merrick, 2001; Rozovsky, Butterworth, & Moore, 2008). In mammals, eIF4GI serves as a bridge between the cap-complex and the PIC through direct interaction with eIF3 (He et al., 2003; Lefebvre et al., 2006; Villa, Do, Hershey, &

Fraser, 2013). Attachment of the ribosome to the mRNA is tightly linked to the cell physiology and is a major target for regulation. A central regulatory mechanism of this stage involves 4E-Binding Proteins (4E-BPs) which bind eIF4E with high affinity and interfere with its binding to eIF4GI, thereby inhibiting cap-dependent translation. 4E-BP is controlled by the mammalian target of rapamycin (mTOR), a protein kinase that phosphorylates it and diminishes its ability to bind eIF4E. mTOR is active under conditions that permit cell growth. When mTOR activity is inhibited, as occurs under conditions of limited nutrients or certain stress conditions, 4E-BP activity is enhanced and cap-dependent translation is suppressed. Despite the requirement of eIF4E for all cap-dependent translation the influence of diminished eIF4E availability varies considerably between different mRNAs. The mechanistic basis underlying differential regulation by eIF4E is not fully understood.

Figure 1: Schematic representation of the eukaryotic cap-dependent translation initiation mechanism. The translation initiation is divided into several stages as indicated. The 43S PIC is assembled from the 40S subunit, a ternary complex consisting of eIF2–GTP–Met- tRNAi, eIF1, eIF1A, eIF3 and eIF5. The mRNA activation stage involves cap-binding and unwinding of cap- proximal region (an ATP-dependent process) by eIF4F subunits. RNA circularization is mediated by PABP–eIF4GI interaction. Attachment of the PIC to the mRNA is mediated by the cap complex and is followed by an ATP-dependent scanning of the 5′ UTR in a 5′ to 3′ direction until an AUG is selected through codon-anti-codon base pairing with the Met-tRNAi. AUG recognition switches the scanning complex to a ‘closed’ conformation and is accompanied by eIF5-assisted hydrolysis of eIF2-bound GTP, Pi release and eIF1 displacement. The 60S subunit joining to the 48S complex is associated with release of eIF2.GDP, eIF3, eIF4F and eIF5 and is mediated by eIF5B and eIF1A. GTP hydrolysis by eIF5B triggers its own and eIF1A release rendering the 80S ribosome ready to elongate.

After mRNA binding, the 43S PIC travels along the mRNA 5′ UTR in a 5′ to 3′ direction, examining the RNA sequence for an AUG start codon through base pairing with the anticodon in the Met-tRNAi (Fig. 1). The 43S movement along the 5′ UTR is an ATP dependent process known as scanning. Secondary structures present in the mRNA 5′ UTR are unwound by eIF4A and other helicases such as yeast Ded1 and mammalian DExH-box helicase DHX29 (Berthelot, Muldoon, Rajkowitsch, Hughes, & McCarthy, 2004; Chuang, Weaver, Liu, & Chang, 1997; Hilliker, Gao, Jankowsky, & Parker, 2011; Pisareva, Pisarev, Komar, Hellen, & Pestova, 2008). The scanning compatible conformation of the 43S is an ‘open’ conformation in which the Met-tRNAi does not fully occupy the P-site (Pout state). During scanning, eIF2 partially hydrolyzes its bound GTP with the assistance of eIF5 (GTPase activating protein), resulting in a stable eIF2.GDP.Pi state. Complementation with the anticodon of the Met-tRNAi results in scanning arrest and commitment to the start codon (Cigan, Feng, & Donahue, 1988). Usually, translation initiates at the first 5′-proximal AUG codon. However, many mammalian mRNAs do not follow this ‘first-AUG’ rule. In those mRNAs, translation initiates at a downstream AUG, a phenomenon known as leaky scanning. The extent of leaky scanning depends on the nucleotide context surrounding the AUG and on the 5′ UTR length (Elfakess et al., 2011; Marilyn Kozak, 2002; Wang & Rothnagel, 2004). For many years the optimal nucleotide context for translation initiation in mammals was the Kozak context RCCAUGG, in which the most significant nucleotides are the purine R in position −3 and G in position +4 relative to the A of the AUG codon that is designated as +1. These two positions distinguish between a ‘strong’ and a ‘weak’ translation initiation site. A weak AUG context allows more leaky scanning, whereas a strong one prevents it (Elfakess et al., 2011; Marilyn Kozak, 2005). The role of the purine in position −3 and the G in position +4 is probably to stabilize the 48S following recognition of the start codon (Pisarev et al., 2006). It has been demonstrated that a 5′ UTR length of at least 20 nucleotides is needed for efficient recognition of an AUG with a favorable context (M Kozak, 1991a). The preferences for optimal AUG context and a minimal 5′UTR length are conserved features of the translation apparatus (Elfakess et al., 2011). By leaky translation the cell can produce two or more completely different proteins or protein variants from one single mRNA. The presence of an AUG upstream to the main ORF

AUG in most cases reduces its translation (Calvo, Pagliarini, & Mootha, 2009; Iacono, Mignone, & Pesole, 2005; Marilyn Kozak, 2002), a finding which strongly supports the scanning mechanism. About half of all human and rodent mRNAs have at least one uAUG in their 5′ UTR (Calvo et al., 2009; Tamarkin-Ben-Harush, Schechtman, & Dikstein, 2014; Wethmar, Barbosa-Silva, Andrade-Navarro, & Leutz, 2014). Following scanning arrest eIF1, eIF2.GDP, its bound Pi and eIF5 are displaced from the PIC (Fig. 1). These rearrangements switch the ‘open’ conformation to a ‘closed’ committed 48S initiation complex in which the small ribosomal subunit is locked onto the mRNA and the

Met-tRNAi is fully accommodated in the P-site (Pin state). eIF2B is a guanosine nucleotide exchange factor, which recycles eIF2.GDP to eIF2.GTP, in order to initiate another cycle of translation. The major regulators of the 48S conformational changes involved in scanning and start codon selection are eIF1 and eIF1A. The eIF5B GTPase and eIF1A promote joining of the 60S large subunit to create the 80S elongation competent ribosome complex. GTP hydrolysis by eIF5B stimulates its own dissociation and that of eIF1A. The 80S can now accept an appropriate aminoacyl-tRNA into the A site and begin the elongation phase. Although most mRNAs are translated via the conventional cap-dependent scanning mechanism, there are eukaryotic mRNAs that utilize alternative mechanisms which are non-canonical. In recent years the number of such mRNAs is growing. The distinction between canonical and non-canonical translation is primarily dictated by sequence elements and structural features present in the mRNA.

Regulation of translation initiation by eIF1 and eIF1A In eukaryotes the cooperative action of several factors is needed to recognize the initiation codon. By using yeast genetics, biochemical and structural techniques the roles of the two major regulators of scanning and start codon recognition, eIF1 and eIF1A, have been elucidated. Early studies identified mutations in yeast eIF1 that reduce the stringency of start codon recognition. More recent genetic studies have identified two types of eIF1A mutations: mutations in the C-terminal tail (CTT) which decrease leaky scanning, and mutations in the N-terminal tail (NTT) which enhance leaky scanning. It was shown that the N and C- terminal tails of eIF1A have opposing effects on the fidelity of start codon selection (Fekete et al., 2005, 2007; Mitchell & Lorsch, 2008). The CTT of

10 eIF1A increases the stringency of start codon selection by promoting the scanning compatible ‘open’ conformation, while the NTT decreases the stringency by promoting the scanning incompatible ‘closed’ conformation (Fekete et al., 2005, 2007). The importance of eIF1 and eIF1A in scanning and start codon selection was also demonstrated using an in vitro reconstituted mammalian translation system with purified translation components. It was shown that in the absence of both eIF1 and eIF1A the ribosomal scanning was almost abrogated. In the addition of eIF1, the 43S complexes retained some capacity of scanning, but they could no longer discriminate between cognate and non-cognate initiation codons. The further addition of eIF1A was required for the proper start codon recognition (T V Pestova, Borukhov, & Hellen, 1998; Tatyana V Pestova & Kolupaeva, 2002). Moreover, these in vitro studies revealed that eIF1 also discriminates against AUG triplets that have weak context or located within 8 nucleotides from the mRNA 5’-end. It was suggested that eIF1 and eIF1A act cooperatively during scanning and promote scanning-competent ‘open’ conformation of the 43S complex. Upon establishment of a stable codon-anticodon base- pairing eIF1 displaced from its near P-site position, eIF1A’s interaction with the 48S complex is tightened and the ribosomal complex undergoes conformational changes in order to achieve ‘closed’ committed form of the 48S (Hussain et al., 2014; Lomakin, Kolupaeva, Marintchev, Wagner, & Pestova, 2003; Lomakin & Steitz, 2013; Passmore et al., 2007; Rabl, Leibundgut, Ataide, Haag, & Ban, 2011; Weisser, Voigts-hoffmann, Rabl, Leibundgut, & Ban, 2013; Yu et al., 2009).

TISU- The Translation Initiator of Short 5’UTR Previous studies in our lab identified and characterized a unique element (SAASATGGCGGC, in which S is C or G) called Translation Initiation of Short 5’UTR – TISU. This element is located downstream and close to the transcription start site and remarkably, controls the initiation rates of both transcription and translation. TISU is present in 4.5% of mRNA genes, most of them with an unusually short 5’UTR (12 nucleotides median length) (Elfakess & Dikstein, 2008). TISU genes are specifically enriched in mRNAs encoding for proteins involved in basic cellular functions such as respiration, protein metabolism and RNA synthesis. It has been found that TISU is essential for transcription, since mutation in TISU substantially decreased the relative

11 amount of all TSSs in two endogenous promoters that were tested. In addition, TISU is a functional Ying Yang 1 (YY1) transcription regulatory element (Elfakess & Dikstein, 2008). Interestingly, TISU is also a strong translation initiation element and it was shown that the AUG core of the TISU element and its flanking nucleotides, in addition to the -3 purine and the +4 G, create a strong translation initiation context that has the distinguished ability to direct high fidelity, accurate translation initiation without leakage from a short 5’UTR (Elfakess & Dikstein, 2008). These features of TISU element are overwhelming considering the fact that initiation from short 5’UTR with other strong AUG context including the Kozak element is highly inefficient and prone to leaky translation (Elfakess & Dikstein, 2008; M Kozak, 1991a, 1991b). Leaky scanning can be explained by the requirement of the 40S ribosomal subunit to form contacts with ~ 30 nucleotides around the initiation start site for efficient initiation (Pisarev, Kolupaeva, Yusupov, Hellen, & Pestova, 2008) . A recent detailed molecular analysis of TISU revealed that this element directs a unique mode of cap-dependent translation initiation which operates without scanning (Elfakess et al., 2011). This study led to a notion that TISU mediates some non-canonical translation initiation mechanism. However the sequence and specific translation initiation factors requirements for TISU’s activity are unknown.

Objectives

The general goal of this research is to elucidate the non-canonical mechanism underlying translation initiation directed by TISU. To achieve this goal I focused on the following specific aims:

1. To analyze sequence requirements of TISU element 2. To determine the translation initiation factors requirements for TISU-mediated translation. 3. To revisit the role of eIFs in the regulation of translation of various mRNA classes which differ in their AUG context and 5’UTR length.

Materials and Methods

Plasmid construction

To generate a GFP reporter gene suitable for translation assays in vitro and in vivo we used the pEGFP-N1 vector (Clontech) that was modified by Rafi Emanuel (Formerly Rofa Elfakess) as follows: an oligo containing 35 repeats of adenosine was generated by PCR and inserted via XbaI and AflII sites downstream of the GFP coding sequence in order to generate poly-A tail in the in vitro synthesized mRNA. This oligo also contained SpeI site immediately upstream to AflII site. The multiple cloning site of the pEGFP-N1 was replaced with CAA repeats via HindIII and AgeI site to remove undesirable secondary structures. In addition, ScaI site was inserted into the vector, using the Quikchange-Site Directed Mutagenesis kit (Stratagene), at position 538 to 543 upstream to the CMV TATA-box element. For in vitro transcription the CMV core promoter was replaced by a T7 promoter via ScaI and NheI sites. Various AUG contexts were constructed by inserting the appropriate oligo into Eco47III and BglII sites. The constructs were linearized by SpeI prior to in vitro transcription. Using the T-PCR cloning method (Erijman, Dantes, Bernheim, Shifman, & Peleg, 2011), each position of the TISU element, except the central AUG, was substituted to all the other three possible nucleotides yielding 27 mutants. For construction of eIF1 expression plasmids the eIF1 cDNA was isolated by RT-PCR from HEK293T RNA and then cloned either in pET28a vector (Novagen) via Ecl136 and XhoI sites for bacterial expression, or in the pCRUZ-HA-A vector (Santa Cruz Biotechnology, Inc.) via ScaI and Bgl II sites and in pCDNA3 via EcoRV and XhoI sites for expression in mammalian cells. For cloning of eIF1A expression plasmids the eIF1AX cDNA was isolated by RT-PCR from HEK293T RNA and then cloned either in pCRUZ-HA-A vector (Santa Cruz Biotechnology, Inc.) via ScaI and Bgl II sites or in pCDNA3 via BamHI and XhoI sites. eIF4A1 WT and dominant negative mutant were previously described (Svitkin et al., 2001). For expression in mammalian cells eIF4A1 WT and mutant were sub-cloned into pCRUZ-HA-A via EcoRV and Bgl II sites.

For the cloning of H2B histone, its native promoter together with the coding sequence was first isolated by PCR from XP12RO genomic DNA. Then the PCR product was cloned into pEGFP-N1 vector (Clontech) via AseI and HindIII sites, so that the entire CMV promoter was replaced with the native histone gene promoter and the H2B ORF is in-frame with that of the EGFP protein. For the context optimization of H2B to H2B- TISU-like a PCR with oligonucleotide containing the optimized sequence was conducted. Four different fragments of Homo sapiens eIF4GI (4GI 84-197, 4GI 198-674, 4GI 675- 1129, 4GI 1130-1599) were amplified by PCR and cloned into pET28a expression vector in frame to its C-terminal 6XHis tag sequence. GST-eIF1 and GST-eIF1A were constructed by cloning PCR fragments encoding eIF1 and eIF1A into pGEX6p1 expression plasmid in frame to its N-terminal GST sequence using the T-PCR cloning method (Erijman et al., 2011). eIF1 K57A/K58A and K109A/H111A mutants were cloned into pGEX6p1 and pCRUZ HA-A expression plasmids using the T-PCR method (Erijman et al., 2011). eIF4E was also cloned into pCRUZ HA-A expression plasmid using the T-PCR method (Erijman et al., 2011). The split-Renilla luciferase fusion plasmids were constructed by two-step PCR using the RSV-C/N Renilla luciferase (pRL-C/N) plasmids and either pCRUZ eIF1, pCRUZ eIF4E, pET28a eIF4GI (84-1129) or pET28a eIF4GI(675-1129) as backbones. As proposed by Jiang et al. (Jiang et al. 2010), the N terminus of the Renilla luciferase contains positions 1 to 229, and the C terminus contains positions 230 to 311. In each construct the split Renilla sequence and the target protein sequence are distinguished by a linker, GGTGGCGGAGGGAGC, corresponding to amino acids GGGGS.

Preparation of mRNA for in vivo translation assay For synthesis of capped mRNA the constructs containing the T7 promoter were linearized with SpeI and used with the RiboMAXTM Large Scale RNA Production Systems T7 (Promega) supplemented with a methylated Cap Analog (New England Biolabs). The reaction was stopped by addition of RQ1 RNase-Free DNaseI (Promega) and the mRNA extracted with phenol: chloroform and precipitated with ethanol. The capped mRNAs were denatured at 65oC for 10 min and then placed on ice for 2 min. The concentration of the synthesized mRNAs was determined and their integrity was confirmed by agarose gel electrophoresis.

Cells, in vivo translation assays and antibodies HEK293T and HeLa cell lines were maintained in DMEM supplemented with 10 % fetal calf serum, 100 units/ml Penicillin and 100μg/ml Streptomycin. For the in vivo translation assays 0.5-1 μg of the in vitro transcribed mRNA were transfected into cells that had been previously seeded on 12-well plates, using ICAFectin441 transfection reagent (In-Cell-Art). Cells were harvested and cell lysates were subjected to 12% SDS- polyacrylamide gel electrophoresis. The levels of GFP protein were determined using the mouse monoclonal-anti-GFP antibody (ab290, Abcam). Antibodies against eIF4GI (ab47649), eIF4E (ab33766), eIF1A (ab38976) and HA-tag (ab9110) are from Abcam, anti-eIF4AI (NBP2-24632) is from Novus, anti-eIF3c antibody (sc74507) is from SantaCruz, eIF1 antibody is a kind gift from Ariel Stanhill (Technion, Haifa) and anti-His is from Qiagen. The anti-eIF5 antibody is from Antibody verify (AAS43534C).

Knockdown of eIFs For knocking-down eIFs, HEK293T cells were seeded on a 6-well plate and transfected with either eIF1 (M-015804-01) , eIF1B (M-019996-00), eIF1A (M-011262-02), eIF3c (M-009036-01), eIF4AI (M-020178-01), eIF5 (M-021336-00), eIF4GI (M-019474-01) or eIF4E (M-002000-00) Dharmacon siGENOME SMART pool siRNA (Thermo Scientific) using DharmaFECT1 transfection reagent. The Dharmacon ON-TARGETplus Non- targeting siRNA #3 was used as a negative control. 48h after the initial transfection, cells were transfected with the indicated GFP reporter plasmid using the standard CaPO4 method. Cells were harvested 24h after the second transfection. For the eIF1 knockdown-rescue assay HEK293T cells were transfected with both eIF1 and eIF1B siRNAs or the non-targeting siRNA #3. 48h after the initial transfection, cells were co-transfected with the siRNAs together with eIF1 WT or K109A/H111A mutant and the indicated GFP reporter plasmids. Cells were harvested 24h after the second transfection.

Cell transfections Overexpression of eIF4A (WT and DN) was done in a 6-well plate transfected with 50 ng of each GFP reporter plasmid and the indicated amounts of the translation initiation factor using the CaPO4 method. Cells were harvested 24h after transfection. Overexpression of eIF1, eIF1A and N-terminal mutated eIF1A was done in a 12-well plate transfected with 50 ng of each GFP reporter plasmid and the indicated amounts of the translation initiation factor using the CaPO4 method. The amount of expression plasmid was kept constant with the empty expression plasmid. Cells were harvested 24h after transfection. Since the Renilla from the pRL-CMV reporter was found be refractory to the inhibition by the DN- eIF4A and to the overexpression of W.T/ N-mutated eIF1A we used it to normalize for transfection efficiency. However pRL-CMV was highly sensitive to eIF1 expression and therefore was not used for normalization in eIF1 studies. To determine the effect of eIF1 on translation of H2B Histone gene, we co-transfected HEK293T cells seeded on a 12-well plate, with 1µg of H2B-GFP reporter plasmid and the indicated amounts of eIF1expression plasmid. The amount of expression plasmid was kept constant with the empty expression plasmid. Cells were harvested 48h after transfection. For the analysis of 4EGI-1 effects on translation, HEK293T cells seeded on 6 well plates were transfected with the indicated GFP reporter plasmid using the standard CaPO4 method. The 4EGI-1 inhibitor (Apexbio, B3696) was added to the media to the final indicated concentrations, six hours after transfection. Cells were harvested 24h after transfection and subjected to 12% SDS-PAGE followed by western blot.

Co-immunoprecipitation and GST pull down assay HEK293T cells were transfected with plasmids directing expression of either HA-tagged eIF1, eIF1A, eIF4E or eIF4AI. 24h later cells were lysed using IP buffer (20mM Tris pH8,125 mM NaCl, 10% Glycerol,0.5% NP-4, 0.2mM EDTA) to which fresh protease inhibitor cocktail (Sigma, 1:100) and PMSF 200μM (1:100) were added. Protein extract was subjected to immunoprecipitation assay using either HA antibody (Sigma) or control antibody in IP buffer, at 4oC for 16h. Each reaction was then washed three times with IP buffer. After the washes 30μl of protein sample buffer were added to each sample. 5% of

17 the input and 30% of each IP sample were then subjected to 8% and 15% SDS-PAGE followed by western blot. GST-eIF1 (WT and mutants), GST-eIF1A and various eIF4GI protein fragments were expressed in E. Coli BL-21 DE3 strain. GST fusion proteins were then purified using gluthatione sepharose beads (GE healthcare). GST pull-down reactions were performed in HEMG buffer (20mM HEPES, 12.5mM MgCl2, 0.5mM EDTA, 0.1% NP-40, 10% Glycerol and 100 mM KCl), at 4oC for 2h. Each reaction was then washed three times, after which 30μl of protein sample buffer was added. 1% input amount and 30% of each pull-down sample were then subjected to 12% SDS-PAGE followed by western blot. In addition, 30% of each pull-down assay sample was also subjected to another SDS-PAGE followed by coomassie blue staining.

Expression and purification of His- eIF1

His-eIF1 was purified from E.coli BL21(DE3) bacteria transformed with eIF1-pET28a construct. Bacteria were grown in 2YT + Kanamycin medium at 370C up to O.D=0.6. Then, IPTG (0.1 mM) was added and bacteria were harvested 4h later. The samples were sonicated and the soluble fractions were purified on Nickel His Trap HP column (Qiagen) as recommended by the manufacturer. Samples from the input, flow-through, wash and 5 elution fractions were analyzed by 15% SDS-PAGE.

Global and gene specific translation analysis HEK293T cells seeded on 6 well plate were transfected, as described above, with either siRNA against the indicated eIF or with a non-targeting siRNA #3 which served as a control . 24h after transfection the cells were trypsinaized and seeded on 10 cm plate. 72h after the siRNA transfection, the cells were incubated with 100 g/ml Cycloheximide for 5 minutes and then washed twice with cold buffer containing 20mM Tris pH 8, 140mM KCl, 5mM MgCl2 and 100 g/ml Cycloheximide. The cells were collected and lyzed with 500 l of same buffer that also contains 0.5% Triton, 0.5% DOC, 1.5mM DTT, 150 units RNAse inhibitor and 5l of protease inhibitor (Sigma). The lyzed samples were centrifuged at 12,000g at 4C for 5 minutes. The cleared lysates were loaded onto 10- 50% sucrose gradient and centrifuged at 41,000 RPM in a SW41 rotor for 90 minutes at 4C. Gradients were fractionated and the optical density at 254 nm was continuously

18 recorded using ISCO absorbance detector UA-6. The free(F), the light(L), and the heavy(H) polysome fractions were pooled. The RNA from each fraction was isolated using Trizol and Direc-Zol RNA mini-prep kits (Zymo Research). cDNA was prepared using High-capacity cDNA reverse transcription kit (Applied Biosystems). Real-Time PCR was done with Power-SYBR green master mix (Applied Biosystems) in 7300 Real- Time PCR System. Primers sequences used in the Real-Time PCR are shown below. To each siRNA treatment the Ct was calculated as follows: Ct(F)= Ct(F)- Ct(F) Ct(L)= Ct(L)- Ct(F) Ct(H)= Ct(H)-Ct(F) Then the Relative Quantification [RQ= (Ct)2] values: RQ(F), RQ(L) and RQ(H) were determined. The RQ values were used to quantify the relative distribution of each TISU mRNA in the polysomal free, light and heavy fractions as follows: % RNA(of gene of interest) in F fraction= RQ(F)*100/[ RQ(F)+RQ(L)+RQ(H)] % RNA(of gene of interest) in L fraction= RQ(L)*100/[ RQ(F)+RQ(L)+RQ(H)] % RNA(of gene of interest) in H fraction= RQ(H)*100/[ RQ(F)+RQ(L)+RQ(H)]

Primers used in Real-Time PCR PPP2R2A: Forward 5’-GGCTATGTGACATGAGGGCAT Reverse 5’-GGGATCTTCAGGTTCTTCAAACA AMPKa1: Forward 5’-ACCATTCTTGGTTGCTGAAACACC Reverse 5’-CTAAATGCCATTTTGCTTTCCTTA AMPKa2: Forward 5’-CGTTCCTGTTCTGCTGCTGGCT Reverse 5’-TGTGACTGCCCAGGCGAGGT ATP5I: Forward 5’-CTACGGAGCCACGCGCTA Reverse 5’-CAATCCGTTTCAGTTCATCCTG CAPZA2: Forward 5’-CAGAACATGGCGACTTGGGA Reverse 5’-AGGGTCTTGGATCAGTTGCCT ARFGAP: Forward 5’-CAAGCATCCGGCCTATTTCA Reverse 5’-TCCTGCGAATTTCTGACGC HMGCL: Forward 5’-ACGTCTCCTGTGCTCTTGGC

Reverse 5’-GATCTCGTAGCAGCCCATTGA CYC1: Forward 5’-CTTGGACCACACCAGCATCC Reverse 5’-CCATGCTGTGGCAGGAGG

Cell separation by elutriation Cell separation by elutriation to G1, S, G2/M cell cycle phases was performed by Dr. N. Diamant from Z. Livneh lab.

Measurement of ATP levels HEK293T cells in 96-well plate (7,000 cells/well) were transfected with control or eIF4GI siRNA as described above. Then, 72h later cells were subjected to ATP levels measurement using CellTiter-Glo Luminescent Assay (promega).

Split-Renilla (RL) complementation assay HEK293T cells seeded on 24 well plate (70,000 cells/well) were co-transfected with 350 ng of each of the indicated split-renilla constructs together with 50 ng of miR 22-Firefly Luciferase (FL) reporter, which served as a control for transfection efficiency. Twenty- four hours after transfection, the cells were harvested using 50 µl/well of the commercial Reporter lysis buffer (Promega). 5 µl of each cell lysate was added to assay plates and RL and FL activities were measured. RL activity was measured using 5 µg/ml CTZ reagent (Gold Biotechnology, Olivette, MO) in 80mMK2HPO4, 20mMKH2PO4 which was added to each well. For the analysis of 4EGI-1 effect on eIF4GI (600-1129)-N-RL and C-RL-eIF4E interaction, 4EGI-1 inhibitor was added to 10 µl of whole cell lysate to the final indicated concentrations, after 10 minutes of incubation at room temperature Renilla Luciferase activity was measured. eIF1 and eIF4E binding competition HEK293T cells seeded on 24 well plate (70,000 cells/well) were transfected with C-RL- eIF4E plasmid together with either eIF4GI (600-1129)-N-RL or eIF4GI (600-1129)-N-RL plasmid. Twenty-four hours after transfection, the cells were harvested using

50 µl/well of the commercial Reporter lysis buffer (Promega). 3µg of His-eIF1 were added to 10 µl of the whole cell extracts. After 30 minutes of incubation at room temperature Renilla Luciferase activity was measured.

Primer list Primers used for the cloning of TISU element mutants in the context of pEGFP-N1 TISU:

C-4T: 5’ CGACTCACTATAGGGTAAGATGGCGGCAGATCTC

C-4A: 5’ CGACTCACTATAGGGAAAGATGGCGGCAGATCTC

C-4G: 5’ CGACTCACTATAGGGGAAGATGGCGGCAGATCTC

A-3T: 5’ CGACTCACTATAGGGCTAGATGGCGGCAGATCTC

A-3C: 5’CGACTCACTATAGGGCCAGATGGCGGCAGATCTC

A-3G: 5’ CGACTCACTATAGGGCGAGATGGCGGCAGATCTC

A-2T: 5’ CGACTCACTATAGGGCATGATGGCGGCAGATCTCG

A-2C: 5’CGACTCACTATAGGGCACGATGGCGGCAGATCTCG

A-2G: 5’CGACTCACTATAGGGCAGGATGGCGGCAGATCTCG

G-1T: 5’GACTCACTATAGGGCAATATGGCGGCAGATCTCGAG

G-1A: 5’ GACTCACTATAGGGCAAAATGGCGGCAGATCTCGAG

G-1C: 5’ CGACTCACTATAGGGCAACATGGCGGCAGATCTCGAG

G+4T: 5’ CACTATAGGGCAAGATGTCGGCAGATCTCGAGCTC

G+4A: 5’ CACTATAGGGCAAGATGACGGC AGATCTCGAGCTC

G+4C: 5’ CACTATAGGGCAAGATGCCGGCAGATCTCGAGC

C+5G: 5’ CTATAGGGCAAGATGGGGGCAGATCTCGAGCTCAAG

C+5A: 5’ CTATAGGGCAAGATGGAGGCAGATCTCGAGCTCAAG

C+5T: 5’ CTATAGGGCAAGATGGTGGCAGATCTCGAGCTCAAG

G+6C: 5’ CTATAGGGCAAGATGGCCGCAGATCTCGAGCTCAAG

G+6A: 5’ CTATAGGGCAAGATGGCAGCAGATCTCGAGCTCAAG

G+6T: 5’ CTATAGGGCAAGATGGCTGCAGATCTCGAGCTCAAG

G+7A: 5’ ATAGGGCAAGATGGCGACAGATCTCGAGCTCAAG

G+7T: 5’ ATAGGGCAAGATGGCGTCAGATCTCGAGCTCAAG

G+7C: 5’ ATAGGGCAAGATGGCGCCAGATCTCGAGCTCAAG

C+8T: 5’ ATAGGGCAAGATGGCGGTAGATCTCGAGCTCAAG

C+8G: 5’ ATAGGGCAAGATGGCGGGAGATCTCGAGCTCAAG

C+8A: 5’ ATAGGGCAAGATGGCGGAAGATCTCGAGCTCAAG

Primers used for the cloning of eIF1 cDNA into pET28a and pCRUZ HA-A: eIF1 Forward : 5’ ACTACCGAGGAAAAGGAATCGTATC eIF1 Reverse (XhoI): 5’ CCCCCCTCGAGTTAAAACCCATGAACCTTCAG eIF1 Reverse (BglII): 5’ CCCCCAGATCT TTAAAACCCATGAACCTTCAG

Primers used for the cloning of eIF1A cDNA into pCRUZ HA-A: eIF1A Forward : 5’ ACTGAAAGAAGTCAGAGACGCCG eIF1A Reverse (BglII): 5’ CCCCCAGATCTCCAAATTGTAGGACAATCTTC

Primers used for H2B Histone gene cloning into pEGFP-N1: H2B (NM_003519) was cloned from -127 to +407 Forward (AseI): 5’ CCCCCATTAATTTTTGATTGGGCAAACCTGGC Reverse (HindIII): 5’ CCCCCAAGCTTCTTGGAGCTGGTGTACTTG

Primers used for H2B context optimization to H2B-TISU-like: Forward (AseI): 5’ CCCCCATTAATTTTTGATTGGGCAAACCTGGC Reverse: 5’CCATCTTGAAACAATAGTGGCA

Primers used for the cloning of eIF1 into pGEX6p1: Forward: 5’-GTTCTGTTCCAGGGGCCCCTGGGATCCATGTCCGCTATC CAGAACCTC

Reverse: 5’-ATCGTCAGTCAGTCACGATGCGGCCGCTTAAAACCCATGAA CCTTCAGC

Primers used for the cloning of eIF1A into pGEX6p1: Forward: 5’-GTTCTGTTCCAGGGGCCCCTGGGATCCATGCCCAAGAATA AAGGTAAAG Reverse: 5’-ATCGTCAGTCAGTCACGATGCGGCCGCTTAGATGTCATC

AATATCTTCAT

Primers used for the cloning of eIF4GI fragments into pET28a: eIF4GI fragment-84-197aa Forward: 5’-CTGGTGGACAGCAAATGGGTCGCCCGCGGATGGGATCCCAAG TAATGATGATCC Reverse: 5’-GGCTTTGTTAGCAGCCGGATCTCAGTGGTGGTGGTGGTGGT GCATGATCTCCTCTGTGATATC eIF4GI fragment-198-674aa Forward: 5’-CTGGTGGACAGCAAATGGGTCGCCCGCGGATGTCTGGGG CCCGCACTGCC Reverse: 5’-GGCTTTGTTAGCAGCCGGATCTCAGTGGTGGTGGTGGTGGTGGC CAAGGTTGGCAAAGGATGG eIF4GI fragment-675-1129aa Forward: 5’-CTGGTGGACAGCAAATGGGTCGCCCGCGGATGGGCCGGAC AACCCTTAGCACC Reverse: 5’-GGCTTTGTTAGCAGCCGGATCTCAGTGGTGGTGGTGGTGGTG TTGTTGAAGGGCTGAGAAGCG eIF4GI fragment-1130-1599aa Forward: 5’-CTGGTGGACAGCAAATGGGTCGCCCGCGGATGGCGGTACCC ACAGAAAGCACAG Reverse: 5’-GGCTTTGTTAGCAGCCGGATCTCAGTGGTGGTGGTGGTGGTG GTGGTCAGACTCCTCCTCTGC

Primers used for the cloning of eIF1 K109A/H111A into pCRUZ HA-A: Forward: 5’-ATGTCCGCTATCCAGAACCTCC Reverse: 5’-CCTCTAGATCTTTAAAACCCAGCAACCGCCAGCTG ATCGTCCTTAGCCA

Primers used for the cloning of eIF1 K57A/K58A into pCRUZ HA-A and pGEX6p1: Forward: 5’-ATGTCCGCTATCCAGAACCTCC Reverse: 5’- CTTAAACGCCTTCACTAGAGCCGCTTTATCGTAATCATCAGCG

Primers used for the cloning of eIF1 K109A/H111A into pGEX6p1: Forward: 5’-ATGTCCGCTATCCAGAACCTCC Reverse: 5’- GATGCGGCCGCTTAAAACCCAGCAACCGCCAGCTGATCGTCC TTAGCCA

Primers used for the cloning of eIF4E into pCRUZ HA-A: Forward: 5’-GATTACGCTAGCCTCGAATTCAGTACTATGGCGACTGTCGAAC

CGGAAAC Reverse: 5’- CACTATAGAATAGGGCCCTCTAGATCTTTAAACAACAAACC TATTTTTAG

Primers used for the cloning of eIF4E into RSV-N Renilla luciferase plasmid: Forward: 5’- AGCGGTGGCGGAGGGAGCATGGCGACTGTCGAACCGG Reverse: 5’- GAAGCGGCCGCTCTAGAATTAAACAACAAACCTATTTTTAG

Primers used for the cloning of eIF4GI (600-1129aa) into RSV-C Renilla luciferase plasmid: Forward: 5’- CTCACTATAGGCTAGCCACCATGGATCAGTGGAAGCCTCCAAAC Reverse: 5’-GTGCTAAGGGTTGTCCGGCCAAGGTTGGCAAAGGATGGAG

Primers used for the cloning of eIF4GI (675-1129aa) into RSV-C Renilla luciferase plasmid: Forward: 5’-CTCACTATAGGCTAGCCACCATGGGCCGGACAACCCTTAGCAC Reverse: 5’-CACCGCTCCCTCCGCCACCTTGTTGAAGGGCTGAGAAGCG

Results Analysis of TISU element sequence requirements

The TISU consensus was initially determined by computational and block mutational analyses (Elfakess & Dikstein, 2008). To determine the sequence requirements of TISU in greater detail, each position of the element, except the central AUG, was substituted to all the other three possible nucleotides yielding 27 mutants. TISU WT and mutants were placed upstream and in frame with the initiating AUG of a GFP reporter and downstream of T7 promoter (Fig. 2A). In all the constructs the core TISU AUG is preceded by a 5 nucleotides long 5’UTR (Fig. 2A). Translation that begins from the TISU AUG generates a ~30 kDa GFP protein (US-AUG). In a case of leaky scanning, translation begins from the downstream GFP AUG produces a ~ 27 kDa protein (DS-AUG). In vitro transcribed and capped mRNAs bearing poly-A tail (Fig. 2B) were co-transfected into HEK293T cells together with a luciferase mRNA that served as an internal control for transfection efficiency. The in vivo translated GFP proteins were analyzed by western blot. The results show that the vast majority of the TISU mutants decreased translation from the US-AUG (Fig. 2C). From this analysis it appears that the best TISU consensus is CAAGAUGGCGGC. Of particular interest are mutations of the A at the -3 position in which substitutions to pyrimidine caused substantial leakage to the downstream AUG and substitution to G retained fidelity but significantly deceased strength. Mutations of the G(+4) to U/C/A, C(+5) to U and G(+6) to C also reduced translation fidelity but to a lesser extent (Fig. 2C). Remarkably, only seven out of 27 mutants, exhibit decreased translation initiation accuracy, highlighting TISU robustness and suggesting that many single nucleotide deviations from the consensus are functional.

Figure 2: The exact sequence requirement for TISU activity. A. A schematic representation of TISU WT and mutant GFP reporter genes. TISU WT or 27 different TISU mutants, each with one mismatch in the flanking nucleotides of TISU core ATG, were cloned downstream to T7 promoter and in frame with the downstream ATG of GFP. The 5’UTR length of all the constructs is 5 nt. The constructs described in A were transcribed and capped in vitro and their integrity was analyzed by agarose-gel electrophoresis as shown in B (M, denotes DNA size marker).C. The in vitro transcribed and capped mRNAs were then transfected into HEK293T cells where they underwent in vivo translation, which was analyzed by immunoblot using anti-GFP antibody. US and DS denote upstream and downstream initiation site, respectively. The table represents

26 quantification of translation from the US-AUG. The * or ** in the table denote weak or strong leaky scanning, respectively.

The roles of eIFs in the regulation of translation initiation fidelity and ribosomal scanning TISU element drives non-canonical translation initiation mechanism which is cap- dependent but scanning independent (Elfakess et al., 2011; Sinvani et al., 2015). To determine which translation initiation factors (eIFs) play specific roles in TISU-directed translation, and to investigate further the interplay between eIFs, AUG context and 5’UTR length, we perturbed eIFs levels in HEK293T human tissue culture cells by either overexpression (eIF1, eIF1A or eIF4A) or knockdown (eIF1, eIF1B, eIF1A, eIF3c, eIF4AI, eIF4GI, eIF4E or eIF5). Next, cells were transfected with GFP reporters constructs containing either a TISU or a canonical AUG context preceded by a short 5’UTR (US-AUG) followed by an in-frame downstream AUG (DS-AUG), or by a long unstructured 5’ UTR GFP reporter as shown schematically in Figure 3. Translation levels were determined by western blot. This research approach enabled us to better understand the mechanism of TISU-mediated translation and to draw more general conclusions about the roles of the different eIFs in the regulation of start codon selection and scanning.

Figure 3: A schematic representation of the GFP reporter genes driven either by TISU or by canonical AUG context, both with short 5’UTR of 17 nucleotides or by a long unstructured 5’ UTR (150 nt) GFP reporter gene in which translation begins from the native GFP start codon.

The effect of eIF1A and eIF1 overexpression on translation initiation fidelity and ribosomal scanning Previous studies indicated that the translation initiation factor eIF1A has an important role in the promotion of scanning and start codon selection (Fekete et al., 2005, 2007; Mitchell & Lorsch, 2008; T V Pestova et al., 1998). The N- and C-termini of eIF1A have opposing effects on the fidelity of start codon selection and ribosomal conformation: the N-terminal tail decreases leaky scanning and promotes the ‘closed’ ribosomal conformation, whereas the C-terminal tail increases translation fidelity and promotes the ‘open’ conformation (Fekete et al., 2005, 2007). Considering that TISU genes are characterized by a very short 5’UTR and nevertheless exhibit high fidelity translation, we examined the effect of eIF1A on translation efficiency and fidelity of mRNAs with a short 5’UTR. We cloned the cDNA of eIF1A (eIF1AX, which is the major form and ubiquitously expressed eIF1A) into mammalian expression plasmid. We utilized the GFP reporter plasmids described in Fig. 3 and the Renilla- luciferase (RL) from the pRL- HCV-FL plasmid which was found to be unaffected by eIF1A (data not shown) as an internal control. HEK293T cells were co-transfected with each GFP reporter together with RL and eIF1A expression plasmids. Cells were harvested 24h later and subjected to western blot analysis using anti-GFP antibody. The results revealed that TISU-mediated translation initiation was not significantly affected by eIF1A overexpression (Fig. 4A). However, translation initiation from the upstream canonical AUG context (Kozak) reporter was stimulated and the ratio of the canonical US-AUG to the DS-AUG was increased (see graph Fig. 4A). It has been demonstrated that translation from long unstructured 5’UTR mRNAs is more efficient than translation from short mRNAs probably due to the ability of longer 5’UTR to load more 43S subunits (M Kozak, 1991b). However, translation of long 5’UTR mRNAs requires efficient and continuous ribosomal scanning. Since eIF1A is involved in scanning (T V Pestova et al., 1998), we next examined its effect on translation of long 5’UTR GFP reporter with 5’UTR length of 150 nt, in which the translation initiation site is from the native GFP AUG generating ~27 kDa protein (Fig. 3). Interestingly, overexpression of eIF1A had no significant effect on translation of the long 5’UTR reporter (Fig. 4A).

As mentioned, the N- and C-termini of eIF1A have opposing effects on start codon selection fidelity and ribosomal conformation. Mutations in the NTT increase leaky scanning and promote the ‘open’ ribosomal conformation, whereas mutation in the CTT decrease translation fidelity and promote the ‘closed' conformation (Fekete et al., 2005, 2007). We wished to examine the effect of N-terminal mutated form of eIF1A on translation directed by TISU or canonical AUG. Since in yeast addition of a tag to the N- or C-termini of eIF1A can lead to a mutant phenotype (Fekete et al., 2005, 2007), we added HA- tag to the N-terminus of eIF1A and repeated the experiment described above. Here again, TISU-mediated translation initiation was refractory to HA-eIF1A expression (Fig. 4B). As for the canonical AUG context reporter, in contrast to the wild type protein the ‘mutated’ form stimulated initiation from the downstream AUG (Fig. 4B) and the ratio of the canonical US-AUG to the DS-AUG was decreased (see graph Fig. 4B). Our data support the previously demonstrated antagonistic roles of eIF1A N- and C-termini. Moreover, the N-terminus mutated eIF1A form significantly promotes translation from the long unstructured 5’UTR mRNA (Fig. 4B). The efficient translation from long 5’UTR depends on the ability to maintain the ‘open’ conformation of scanning complexes. Our results suggest that the N-terminus ‘mutated’ form of eIF1A can support the ‘open’ conformation better than the wild type.

Figure 4: eIF1A overexpression affects start codon selection in an AUG context and 5’UTR length dependent manner. HEK293T cells were co-transfected with short 5’UTR GFP reporter genes in TISU or canonical AUG context, or with a long 5’UTR GFP reporter together with either A. wild type eIF1A or B. N-terminus HA- tagged eIF1A expression plasmid as indicated (+). Overexpression levels are presented on the right panels. Renilla- luciferase under the CMV promoter was used to normalize transfection efficiency. The translation initiation site was determined by immune-blot using anti- GFP specific antibody. US and DS denote upstream and downstream initiation site, respectively. Representative immune-blots are shown. Densitometric analysis of several (3) independent experiments was performed. The graphs represent the average±SE translation. The overall translation in each control was set to one. The relative translation from the US-AUG is presented by light grey bars and the relative translation from the DS-AUG is presented by dark grey bars. The * denote statistically significant differences (p<0.05).

Another important factor which regulates translation fidelity and scanning is eIF1. Previous studies revealed that eIF1 stimulates leaky scanning when the AUG is preceded by a short 5’ UTR (Cheung et al., 2007; Mitchell & Lorsch, 2008; Nanda et al., 2009; T V Pestova et al., 1998). Considering that TISU genes are characterized by a strong and accurate translation regardless of the extremely short 5’UTR, we wished to examine the role of eIF1 in its translation. We cloned eIF1 cDNA into expression plasmid and performed the experiment as described above for eIF1A. Since translation of the normalizing RL reporter was significantly promoted by eIF1 overexpression (data not shown) we could not use it as an internal control. The results revealed that the mRNA with the canonical AUG context was highly sensitive to eIF1 expression: in the presence of eIF1 utilization of the upstream AUG was inhibited while initiation from the downstream AUG was increased (Fig. 5A) and the ratio of the canonical US-AUG to the DS-AUG was decreased (see graph Fig. 5A). Interestingly, eIF1 had no significant effect on TISU-mediated translation initiation, which was primarily from the upstream AUG without any leaky translation (Fig. 5A). Moreover, similar to the ‘mutated’ form of eIF1A, eIF1 also significantly promoted translation from the long unstructured 5’UTR mRNA (Fig. 5A). This result supports the previously demonstrated role of eIF1 in maintaining the ‘open’ scanning compatible ribosomal conformation. Considering the opposing effects of eIF1 and eIF1A (untagged) on the canonical AUG translation, we next wanted to elucidate which of these factors is dominant. Therefore, HEK293T cells were co- transfected with the canonical AUG reporter plasmid together with eIF1, eIF1A or equal amounts of both expression plasmids. Cells were harvested 24h later and subjected to western blot analysis with anti-GFP antibody. The analysis showed that cells transfected with both eIF1 and eIF1A exhibited eIF1 effect: utilization of the upstream AUG was inhibited while initiation from the downstream AUG was increased (Fig. 5B). Therefore, it seems that eIF1 effect is dominant over eIF1A effect. Our results reveal that eIF1 and eIF1A affect translation in an AUG context and 5’UTR length dependent manner, supporting the previously demonstrated antagonistic roles of eIF1 and eIF1A in yeast. A model of a good cop/bad cop was proposed for the actions of eIF1 and eIF1A(Mitchell & Lorsch, 2008). It was suggested that while eIF1 prevents translation initiation commitment on putative start codons, eIF1A facilitates

31 pausing at start codons long enough to proceed with initiation. Clearly among short 5’UTR mRNAs, which are generally inhibited or stimulated by eIF1 and eIF1A, respectively, TISU confers tolerance to their elevated levels.

Figure 5: eIF1 overexpression affects start codon selection in an AUG context and 5’ UTR length dependent manner. A. HEK293T cells were co-transfected with each GFP reporter together with eIF1 expression plasmid as indicated (+).Overexpression level is presented on the right panel. B. eIF1 effect is dominant over eIF1A effect. HEK293T cells were co-transfected with the canonical AUG context reporter plasmid together with eIF1, eIF1A or equal amounts of both expression plasmids. Each factor was cloned into pCDNA3 expression plasmid. Cells were harvested 24h after transfection and subjected to western blot analysis with anti-GFP antibody. The translation initiation site was determined by immune-blot using anti- GFP specific antibody. US and DS denote upstream and downstream initiation site, respectively. Representative immune- blots are shown. Densitometric analysis of several independent experiments (3-7) was performed. The graphs represent the average ±SE translation. The overall translation in each control was set to one. The relative translation from the US-AUG is presented by light grey bars and the relative translation from the DS-AUG is presented by dark grey bars. The * denote statistically significant differences (p<0.05).

The effect of eIF1 and eIF1A overexpression on H2B mRNA translation To determine whether the observed effects of eIF1 and eIF1A on translation are relevant to native short 5’UTR mRNAs, we tested eIF1 and eIF1A overexpression effects on the translation of Histone H2B gene which is characterized by a short 5’UTR and a weak AUG context. We constructed an H2B-GFP plasmid in which the promoter and the entire coding sequence (from -127 to +407) are upstream and in frame to GFP reporter gene (Fig. 6A). Translation that initiates from the main ORF AUG of H2B is expected to produce a ~ 40 kDa nuclear GFP fused protein. In addition, we constructed an H2B- TISU-like plasmid in which the AUG context was optimized (Fig. 6A). First, we tested the effect of AUG context optimization on translation. HEK293T cells, transfected with either H2B-GFP or H2B-TISU-like-GFP plasmids, were harvested 48h after transfection and subjected to immunoblot analysis. Indeed, context optimization improves translation efficiency by ~3-folds (Fig. 6A). To further demonstrate the importance of AUG context to eIF1 effect we co- transfected HEK293T cells together with eIF1 expression plasmid and either with H2B- GFP or H2B-TISU-like-GFP plasmid. Our results demonstrated that eIF1 significantly inhibited translation initiation from the main ORF AUG of H2B-GFP (Fig. 6B); however, H2B-TISU-like-GFP translation level remained steady following eIF1 overexpression (Fig. 6C). From these results it seems that TISU genes are resistant to eIF1 inhibitory effect; however, native genes bearing short 5’UTR and weak AUG context are not. We also tested the effect of eIF1A on the native weak AUG of H2B gene and found no significant effect (Fig. 6D). The lack of an effect on H2B gene, which is different from the effect on the reporter gene, may be explained by other regulatory features present in the native gene.

Figure 6: eIF1 and eIF1A effects on translation of Histone H2B mRNA. A. A schematic representation of H2B-GFP reporter gene, which is driven by native weak AUG context. The transcription and translation units are schematically represented. Arrow indicates the positions of the major TSS. The expected translation initiation site is indicated in red. The native and optimized AUG contexts are also represented, positions –3/+4 are bolded. The translation start site of GFP is indicated. Context optimization of Histone H2B ATG improves translation.

HEK293T cells were transfected with H2B-GFP or H2B-TISU-like-GFP reporter plasmid. Transfection efficiency was normalized to Renilla-Luciferase (in a secreted form). B. Co- transfection of H2B-GFP together with eIF1 expression plasmid. C. Co-transfection of H2B- TISU-like-GFP together with eIF1 expression plasmid. D. Co-transfection of H2B-GFP together with eIF1A expression plasmid. Renilla- luciferase under the CMV promoter was used to normalize for transfection efficiency in eIF1A transfections. The relative translation level was determined by immune-blot analysis with anti-GFP antibody. Representative immune-blots are shown. Densitometric analysis of several (3-5) independent experiments was performed. The graphs represent relative translation level (average ±SE) of H2B-GFP or H2B-TISU-like-GFP. The translation level in each control was set to one. The * denote statistically significant differences, p<0.05.

The effects of eIF1A and eIF1 knockdowns on translation initiation fidelity and ribosomal scanning

Hadar Sinvani from our lab reconstituted TISU mediated translation in vitro using purified and recombinant factors. Her results indicated that TISU translation is diminished by omission of eIF2, initiator tRNA, eIF3, eIF4F, and by both eIF1 and eIF1A (Sinvani et al., 2015). To further establish those results in a cellular context, we tested the requirement of eIF1 and eIF1A for TISU mRNA translation by knockdown experiments using SMARTpool siRNAs. First, HEK293T cells were transfected with siRNA pool against eIF1A. 48h later cells were co-transfected with eIF1A siRNA, together with each of the GFP reporter genes described in Fig. 3. Cells were harvested 24h after the second transfection and immune-blot analysis was performed. The double knockdown approach resulted in an undetectable eIF1A level (Fig. 7, right panel). The results revealed AUG- context dependent differential effects. Remarkably, translation from TISU-AUG was dramatically decreased, without concomitant increase of leaky translation. In contrast translation from the canonical AUG was only slightly affected (Fig. 7). Moreover, the ratio between the canonical US-AUG to the DS-AUG was increased (see graph Fig. 7) suggesting that leaky scanning was inhibited. Translation from the long 5’UTR reporter was unaffected by eIF1A depletion (Fig. 7). Thus, TISU-mediated translation seems to be highly eIF1A dependent. This interesting observation should be further tested on native TISU genes.

Figure 7: eIF1A knockdown effects on translation initiation fidelity and scanning. HEK293T cells were transfected with siGENOME smart pool siRNA against eIF1A (50 nM). Non-targeting siRNA served as a control. 48h later cells were co-transfected again with the siRNA pool together with short 5’UTR GFP reporter genes in TISU or canonical AUG context, or with a long 5’UTR GFP reporter. Cells were harvested 24h after the second transfection and immune-blot analysis was performed. Representative immune-blots are shown. Knockdown efficiency is presented on the right panel. Densitomentric analysis of several independent experiments (5-6) was performed. The graphs represent the average ±SE translation. The overall translation in each control was set to one. The relative translation from the US-AUG is presented by light grey bars and the relative translation from the DS-AUG is presented by dark grey bars. The * denote statistically significant differences, p<0.05.

We next examined the necessity of eIF1 to TISU translation in vivo using knockdown experiments. HEK293T cells were transfected with siRNA pool against eIF1. 48h later cells were transfected with each of the GFP reporter genes described above. Cells were harvested 24h after the second transfection and immune-blot analysis was performed. The results revealed that although eIF1 knockdown efficiency was about 60% on average (Fig. 8A, right panel), we could detect AUG-context dependent differential effects. Translation from the upstream canonical AUG context mRNA was stimulated by ~ 20% (Fig. 8A) and the ratio of the canonical US-AUG to the DS-AUG was increased. In contrast, TISU-mediated translation decreased by ~30% (Fig. 8A). Thus, TISU- mediated translation seems to be also eIF1 dependent. Surprisingly, translation of the mRNA containing the long 5’UTR was marginally affected (Fig. 8A), which is unexpected considering that eIF1 is known to promote ribosomal scanning. We therefore considered the possibility that eIF1 activity is

36 redundant with another factor. Searching the human genome database (http://genome.ucsc.edu/) for eIF1 homology we found that eIF1 has a highly similar paralog called eIF1B (Fig. 8B) that is expressed as abundantly as eIF1. The role of eIF1B in translation was not characterized before. We therefore knocked-down eIF1B with siRNA pool (Fig. 8C). Similar to eIF1 depletion, upon eIF1B knockdown TISU-mediated translation and leaky scanning were decreased, while translation from the canonical US- AUG and from long 5’UTR were unaffected (Fig. 8C). In order to get clear differential effects, we next knocked-down both eIF1 paralogs. With the eIF1/eIF1B siRNAs, the knockdown efficiency was increased (Fig. 8D, right panel) and led to a more dramatic AUG-context dependent differential effect which is similar to the effect of eIF1A knockdown. The ratio of the canonical US-AUG to the DS-AUG was increased (see graph Fig. 8D), suggesting that leaky scanning was inhibited. In contrast to the canonical US-AUG, TISU-AUG levels were greatly diminished (Fig. 8D). Interestingly, the decrease in TISU translation levels was not followed by concomitant increase of leaky scanning. In addition, translation from the long 5’UTR which is dependent on ribosomal scanning was reduced, consistent with the role of eIF1 in scanning (Pestova and Kolupaeva 2002,1998).

Figure 8: eIF1/eIF1B knockdown effects on translation initiation fidelity and scanning. HEK293T cells were transfected with either A. eIF1 or C. eIF1B or D. mixture of both eIF1 and eIF1B Dharmacon SMART pool siRNA (50 nM each). Non-targeting siRNA served as a control. 48h later cells were transfected again with the GFP reporter genes indicated. Cells were harvested 24h after the second transfection and analyzed by western blot with GFP antibody. US and DS denote upstream and downstream initiation site, respectively. Representative immune-blots are shown. The eIF1 knockdown efficiencies are shown on the right panels. The graphs represent the average ±SE of densitometric measurements of GFP levels of (3-6) experiments. The overall translation in control cells was set to one. The relative translation from the US-AUG is presented by light grey bars and the relative translation from the DS-AUG is presented by dark grey bars. The * denote statistically significant difference. (p<0.05). B. Protein sequence alignment of eIF1 (NP_005792) and eIF1B (NP_005866) as retrieved from PRALINE server.

Taking into account that translation of TISU GFP reporter gene was highly dependent on eIF1/eIF1B, we next wanted to assess the effect of eIF1/eIF1B depletion on translation of native TISU bearing genes. HEK293T cells were transfected with either siRNA pool directed against both eIF1/eIF1B or with non- targeting siRNA as a control. 72h later cells were harvested and cell lysates were subjected to polysome profiling. eIF1/eIF1B double knockdown did not result in detectable effect on global translation (Fig. 9B) although the knockdown was efficient (Fig. 9A). This may be a consequence of compensation by other factors with redundant activities such as eIF1A. To examine the influence of eIF1/eIF1B depletion on translation of native TISU bearing genes we pooled the fractions from the free, the light, and the heavy polysomes; extracted their RNA; and performed RT-qPCR analysis of eight TISU genes. Consistent with the GFP reporter genes analysis, we found that translation of all tested TISU genes was inhibited by eIF1/eIF1B knockdown, as the amount of mRNA was shifted from the heavy polysomal fractions to the light polysomal fractions (Fig. 9C).

Figure 9: Translation efficiency of specific endogenous TISU mRNAs following eIF1/1B knockdown. HEK293T cells were transfected with eIF1/1B Dharmacon SMART pool siRNAs (50 nM of each). Non-targeting siRNA served as a control. 72h after transfection cells were harvested and subjected to A. western blot in order to determine the knockdown efficiency and B. polysomal fractionation. C. Real-Time qPCR analysis of the indicated TISU mRNAs in the (F) free polysome fractions (L) light polysomes fractions and (H) heavy polysomes fractions of the gradient.

eIF1 and eIF1A protein levels are cell cycle regulated Our results suggest that changes in eIF1 or eIF1A protein levels affect translation of mRNAs with distinct features differentially. These findings may have biological significance if there are specific physiological conditions in which eIF1 and eIF1A protein levels change. We examined the possibility that translation initiation levels change during the cell cycle. For this purpose, U2OS cells were separated by elutriation

40 according to their cell cycle phase (the cell separation and further validation by FACS was carried out by Dr. Noam Diamant from Zvi Livneh’s lab). Cell lysates were subjected to western blot using anti-eIF1 and anti- eIF1A. The results revealed that eIF1 and eIF1A protein levels are remarkably changing during cell cycle (Fig. 10). During the G2/M phase, eIF1 and eIF1A levels were elevated by 7.5 and 4- fold, respectively, in comparison to their basal level during the G1 phase.

Figure 10: eIF1 and eIF1A protein levels are cell cycle regulated. U2OS cells were separated by elutriation according to their cell cycle phase (the separation was performed by Dr. Noam Diamant). Cell lysates were subjected to immunoblot using A. anti-eIF1 and B. anti- eIF1A antibodies. Relative immune-blots are presented.

eIF1 and eIF1A protein levels in different mouse cell types and Human tissue culture cell types Next, we wanted to examine whether eIF1 and eIF1A protein levels vary between different cell types. Therefore, we examined by western blot analysis the expression of eIF1 and eIF1A in mouse brain, kidney, liver and spleen tissue lysates (the lysates were kindly provided by Rene Abramovitz from Gideon Schreiber’s lab). The results revealed substantial differences in eIF1 and eIF1A protein levels. The highest eIF1A protein level was detected in mouse liver tissue, whereas the highest eIF1 protein level was detected in mouse kidney tissue (Fig. 11A). It seems that there is a physiological variation in eIF1 and eIF1A protein levels between different cell types. In order to further establish our observations we analyzed eIF1 and eIF1A protein levels in lysates of HEK293T, HeLa and HepG2 human cell lines. The results revealed that eIF1A protein level is relatively similar among different cell types; however, eIF1 levels are significantly diverse (Fig. 11B). The lowest eIF1 protein level was detected in HEK293T cells, whereas the highest level was detected in HepG2 cells. Since eIF1 and 41 eIF1A affect translation in an opposing way, it can be anticipated that eIF1 to eIF1A ratio would determine the efficiency of translation and the extent of leaky scanning and therefore influence the protein content of the cells in a cell-type-dependent manner. In agreement to this hypothesis we noticed that translation from the canonical upstream AUG varies significantly between different cell types, constituting approximately 45% and 20% of total translation in HEK293T and HeLa cells, respectively, while being undetectable in HepG2 cells (Fig. 11B, lower panel).

Figure 11: eIF1 and eIF1A protein levels in different mouse cell types and human tissue culture cell types. A. Mouse tissue lysates from brain, kidney, liver and spleen were subjected to western blot using anti-eIF1A and anti-eIF1antibodies (the lysates were kindly provided by Rene Abramovitz). B. Upper panel-Protein lysates from HEK293T, HeLa and HepG2 cells were subjected to immune-blot using anti-eIF1A and anti -eIF1 antibodies. Lower- panel- HEK293T, HeLa and HepG2 cells were transfected with canonical AUG GFP reporter plasmid. Cells were harvested 24h after transfection and analyzed by western blot with GFP antibody. US and DS denote upstream and downstream initiation site, respectively. Coomassie blue staining and representative immune-blots are shown.

The effect of eIF3c knockdown on translation initiation fidelity and ribosomal scanning

As previously mentioned, TISU reconstitution in vitro requires eIF3 among other additional factors. Mammalian eIF3 is a large multifunctional factor composed of 13 subunits (a-m). To validate this finding and to determine the involvement of eIF3 in

42 translation initiation fidelity in vivo (Fig. 12) we chose to knockdown the core subunit eIF3c, since it was previously implicated as one of the regulators of start codon selection (Valásek, Nielsen, Zhang, Fekete, & Hinnebusch, 2004). The eIF3c knockdown resulted in an overall translation reduction driven by all GFP reporters tested, irrespective of 5’UTR length and AUG context (Fig. 12). Nevertheless TISU translation is the most sensitive to eIF3 depletion. Unlike eIF1/1B and eIF1A knockdowns, there was no effect on leaky scanning, since the proportion between the US-AUG to DS-AUG translation of the canonical AUG reporter was maintained (Fig. 12).

Figure 12: eIF3c knockdown effects on translation initiation fidelity and scanning. A. HEK293T cells were transfected with eIF3c Dharmacon SMART pool siRNA (25 nM). Non- targeting siRNA served as a control. 48h later cells were transfected again with the GFP reporter genes indicated. Cells were harvested 24h after the second transfection and analyzed by western blot with GFP antibody. US and DS denote upstream and downstream initiation site, respectively. Representative immune-blots are shown. The eIF3c knockdown efficiency is shown on the right panels. The graphs represent the average ±SE of densitometric measurements of GFP levels of (3) experiments. The overall translation in control cells was set to one. The relative translation from the US-AUG is presented by light grey bars and the relative translation from the DS-AUG is presented by dark grey bars. The * denote statistically significant difference (p<0.05).

We also analyzed the effect of eIF3c knockdown on global translation using polysome profiling as described above. Depletion of eIF3c resulted in accumulation of 80S monoribosome and a decrease in polysomes, indicating for global inhibition of

43 translation initiation (Fig. 13B). Analysis of eight native TISU mRNAs confirmed their eIF3 dependency, as the amount of these mRNAs was shifted from the heavy polysomal fractions to the light and free polysomal fractions (Fig. 13C).

Figure 13: Translation efficiency of specific endogenous TISU mRNAs following eIF3c knockdown. HEK293T cells were transfected with eIF3c Dharmacon SMART pool siRNA (25 nM). Non-targeting siRNA served as a control. 72h after transfection cells were harvested and subjected to A. western blot in order to determine the knockdown efficiency and B. polysome fractionation. C. Real-Time qPCR analysis of the indicated TISU mRNAs in the (F) free polysome fractions (L) light polysomes fractions and (H) heavy polysomes fractions of the gradient.

The role of eIF4A in translation initiation fidelity and ribosomal scanning

eIF4A is an RNA helicase that unwinds secondary structures in the mRNA 5’UTR (Lorsch & Herschlag, 1998; Marintchev et al., 2009; Oberer et al., 2005). As most TISU

44 bearing mRNAs are characterized by a short 5’UTR (Elfakess & Dikstein, 2008), one would expect that TISU-mediated translation would be less dependent on eIF4A. To test this possibility we utilized a previously characterized mutant of eIF4A that is deficient in its helicase activity but can still be incorporated into the eIF4F complex thereby acting as a dominant negative mutant (Svitkin et al., 2001). We examined the effect of eIF4A mutant on three different types of GFP reporter mRNAs. The first has a long and unstructured 5’ UTR, the second has a secondary structure embedded within long 5’UTR and the third bears TISU in a context of a short 5’UTR (Fig. 14, upper left panel). The Renilla luciferase (RL) from the pRL-CMV plasmid was not significantly affected by the dominant negative mutant of eIF4A in multiple experiments (data not shown), therefore it served as an internal normalizing control. HEK293T cells were co-transfected with the GFP and RL reporters together with wild type or increasing amounts of eIF4A dominant negative mutant expression plasmid. The expression of eIF4A wild type and mutant was validated by immunoblot with anti-HA tag antibody (Fig. 14). The results show that the dominant negative mutant of eIF4A inhibited the expression of the reporters bearing either unstructured long or structured 5’UTR, with the secondary structured 5’UTR reporter being sensitive to eIF4A inhibition at lower concentrations. In contrast, TISU mRNA was the least affected by the eIF4A mutant (Fig. 14). These findings raised the possibility that TISU is likely to confer to the genes bearing it, resistance to fluctuations in eIF4A availability. In order to test this hypothesis and to examine eIF4A requirement for TISU we performed knockdown experiments using SMARTpool siRNAs against eIF4AI (50 nM) which is the predominant form of eIF4A in the cells. The knockdown experiments were performed as described above. Interestingly, eIF4AI depletion resulted in an overall translation reduction in all GFP reporters tested, irrespective of 5’UTR length and AUG context (Fig. 15). It seems that eIF4A helicase is required for translation of short 5’UTR mRNAs, including TISU mRNAs, as well as for translation of long 5’UTR mRNAs. However, as opposed to eIF1A and eIF1/1B knockdown experiments, eIF4AI depletion did not affect the extent of leaky scanning, since the proportion between the canonical US-AUG to DS-AUG was maintained (Fig. 15). Thus, it seems that eIF4AI protein amount in the cells is required for TISU mRNAs translation.

Figure 14: eIF4A WT and DN overexpression effects on translation. A schematic representation of the GFP-reporter genes is shown in the upper-left panel. The first has a long unstructured 5’ UTR, the second has a hair-pin structure within the 5’ UTR and the third has TISU in a context of a short 5’ UTR. HEK293T cells were transfected with these GFP reporters together with increasing amounts of eIF4A-dominant negative mutant (eIF4A-DN) or wild-type eIF4A-expression plasmids as indicated. The Renilla luciferase (RL) from the pRL-CMV plasmid was not significantly affected by the dominant negative mutant of eIF4A in multiple experiments, therefore it served as internal normalizing control. GFP, eIF4A-DN and eIF4A expression were analyzed by immunoblot. Representative immuneblots are shown on the upper right section. The * in the eIF4A blot denotes a non-specific band. A graph representing the average of densitometric measurements of three independent transfection experiments is shown on the lower panel.

Figure 15: eIF4AI knockdown effects on translation initiation fidelity and scanning A. HEK293T cells were transfected with eIF4AI Dharmacon SMART pool siRNA (50 nM). Non- targeting siRNA served as a control. 48h later cells were transfected again with the GFP reporter genes indicated. Cells were harvested 24h after the second transfection and analyzed by western blot with GFP antibody. US and DS denote upstream and downstream initiation site, respectively. Representative immune-blots are shown. The eIF4AI knockdown efficiency is shown on the right panels. The graphs represent the average ±SE of densitometric measurements of GFP levels of (3- 5) experiments. The overall translation in control cells was set to one. The relative translation from the US-AUG is presented by light grey bars and the relative translation from the DS-AUG is presented by dark grey bars. The * denote statistically significant difference (p<0.05).

Next, we tested the requirement of eIF4AI for global translation and for translation of endogenous TISU genes by polysome profiling as described above. We found that eIF4AI depletion caused global translation inhibition as evident particularly from the dramatic reduction in the relative amount of polysomes (Fig. 16B). Consistent with the global inhibitory effect of eIF4AI knockdown on translation, we found that translation of all tested TISU mRNAs was inhibited, as the amount of mRNA was shifted from the heavy polysomal fractions to the light and free fractions (Fig. 16C). These findings revealed that eIF4A is required not only for translation of long and structured 5’UTR mRNAs, as previously suggested, but also for short 5’UTR mRNAs including TISU mRNAs.

Figure 16: Translation efficiency of specific endogenous TISU mRNAs following eIF4AI knockdown. HEK293T cells were transfected with eIF4AI Dharmacon SMART pool siRNA (50 nM). Non-targeting siRNA served as a control. Then, 72h after transfection cells were harvested and subjected to A. western blot in order to determine the knockdown efficiency and B. ribosomal fractionation. C. Real-Time qPCR analysis of the indicated TISU mRNAs in the (F) free polysome fractions (L) light polysomes fractions and (H) heavy polysomes fractions of the gradient.

The effect of eIF5 knockdown on translation initiation fidelity and ribosomal scanning

eIF5 is a GAP-GTPase activating protein that stimulates the hydrolysis of eIF2 bound GTP. The GTP hydrolysis is a critical step in scanning arrest and commitment of the 43S PIC to initiate translation. Previous yeast genetic studies had identified yeast strains with disordered start codon selection due to mutations in eIF5 (Huang, Yoon, Hannig, & Donahue, 1997; Saini et al., 2014). Since eIF5 is one of the regulators of

48 scanning and start codon selection we decided to test its requirement for TISU genes translation. First we performed knockdown experiments as described above. Similar to eIF3c and eIF4AI knockdown experiments, eIF5 depletion resulted in an overall translation reduction in all GFP reporters tested, irrespective of 5’UTR length and AUG context. However, it seems that TISU translation is the least sensitive to eIF5 reduced level. Here again, there was no effect on leaky scanning, since the ratio of canonical US- AUG to DS-AUG translation was maintained (Fig. 17). Regarding the global translation analysis, eIF5 knockdown did not cause detectable effect on polysome profiling (Fig. 18B). Yet, translation of five out of eight specific endogenous TISU genes tested was inhibited in the eIF5 depleted cells (Fig. 18C).

Figure 17: eIF5 knockdown effects on translation initiation fidelity and scanning. A. HEK293T cells were transfected with eIF5 Dharmacon SMART pool siRNA (50 nM). Non- targeting siRNA served as a control. 48h later cells were transfected again with the GFP reporter genes indicated. Cells were harvested 24h after the second transfection and analyzed by western blot with GFP antibody. US and DS denote upstream and downstream initiation site, respectively. Representative immune-blots are shown. The eIF5 knockdown efficiency is shown on the right panels. The graphs represent the average ±SE of densitometric measurements of GFP levels of (3- 5) experiments. The overall translation in control cells was set to one. The relative translation from the US-AUG is presented by light grey bars and the relative translation from the DS-AUG is presented by dark grey bars. The * denote statistically significant difference (p<0.05).

Figure 18: Translation efficiency of specific endogenous TISU mRNAs following eIF5 knockdown. HEK293T cells were transfected with eIF5 Dharmacon SMARTpool siRNA. Non- targeting siRNA served as a control. 72h after transfection cells were harvested and subjected to A. western blot in order to determine the knockdown efficiency and B. polysome fractionation. C. Real-Time qPCR analysis of the indicated TISU mRNAs in the (F) free polysome fractions (L) light polysomes fractions and (H) heavy polysomes fractions of the gradient. eIF4GI has differential effects on translation initiation fidelity and ribosomal scanning Previous characterization of TISU element translation strongly argues for its cap- dependent nature (Elfakess & Dikstein, 2008; Elfakess et al., 2011). Since TISU mRNAs are characterized with an extremely short 5’UTR with a median length of 12 nt, it is inevitable that upon 48S formation the ribosome and the attached eIF4F (cap-binding complex) would be situated in an overlapping positions. Such case could lead to a steric

50 clash between the cap-binding complex and the ribosome. Therefore we were interested to examine the interplay between eIF4F, TISU-AUG and the small ribosomal subunit during translation initiation. Since the scaffold protein eIF4GI is the largest subunit of eIF4F, we decided to examine its role in translation directed by TISU element. We performed eIF4GI knockdown experiments as described above. Strikingly, eIF4GI depletion caused the same phenotype as eIF1/eIF1B knockdown with respect to the different AUG contexts and 5’UTR lengths. Specifically, in eIF4GI-deficient cells, the relative translation of the canonical US-AUG to the DS-AUG was increased while TISU US-AUG was almost abolished (Fig. 19). To examine whether the similarity of the knockdown effects of eIF4GI and eIF1 is associated with indirect effect on eIF1 expression, we analyzed eIF1 levels in eIF4GI depleted cells and found that its expression was unaffected. Thus, TISU-mediated translation seems to be eIF4GI dependent. Similar to eIF1, eIF4GI depletion caused to a reduction of translation from the canonical DS-AUG and from the unstructured long 5’UTR (Fig. 19), indicating its role in the promotion of ribosomal scanning and leaky scanning. These findings raise the possibility that eIF1 and eIF4GI functionally cooperate to facilitate translation initiation from TISU and non-TISU mRNAs as well as ribosomal scanning and leaky scanning. The global translation analysis using polysome profiling showed that depletion of eIF4GI resulted in accumulation of 80S monoribosome and a decrease in polysomes, indicating for global inhibition of translation initiation (Fig. 20B). Moreover, the gene specific translation analysis showed that down-regulation of eIF4GI had a notable effect on translation of all examined TISU genes (Fig. 20C), consistent with the results obtained using the TISU GFP reporter gene.

Figure 19: eIF4GI knockdown effects on translation initiation fidelity and scanning. HEK293T cells were transfected with eIF4GI Dharmacon SMARTpool siRNA (50nM). Non- targeting siRNA was used as a control. 48h later cells were transfected again with the GFP reporter genes indicated. Cells were harvested 24h after the second transfection and analyzed by western blot with GFP antibody. US and DS denote upstream and downstream initiation site, respectively. Representative immune-blots are shown. The eIF4GI knockdown efficiency is shown on the right panel. The effect of eIF4GI knockdown on eIF1 protein levels is also shown. The graph represents the average ±SE of densitometric measurements of GFP levels of 3 experiments. The overall translation in control cells was set to one. The relative translation from the US-AUG is presented by light grey bars and the relative translation from the DS-AUG is presented by dark grey bars. The * denote statistically significant difference (p<0.05).

Previous analysis of the functional classes associated with the TISU genes in which their ORF begins from the element’s AUG showed that one of the most enriched functional classes is the mitochondrion (Elfakess & Dikstein, 2008). Within this class we found a significant enrichment of genes involved in mitochondrial activities and energy metabolism. As translation of exogenous and endogenous genes governed by TISU is highly eIF4GI dependent, it is expected that eIF4GI depletion would affect energy homeostasis. To test this possibility, we measured cellular ATP levels in cells treated with control or eIF4GI siRNA, using luminescence assay. The data, after normalization to cell number, revealed a significant reduction in endogenous ATP levels upon eIF4GI knockdown (Fig. 20D).

Figure 20: Translation efficiency of specific endogenous TISU mRNAs following eIF4GI knockdown. HEK293T cells were transfected with eIF4GI Dharmacon SMARTpool siRNA (50 nM). Non-targeting siRNA served as a control. 72h after transfection cells were harvested and subjected to A. western blot in order to determine the knockdown efficiency and B. polysome fractionation. C. Real-Time qPCR analysis of the indicated TISU mRNAs in the (F) free polysome fractions (L) light polysomes fractions and (H) heavy polysomes fractions of the gradient. D. Knockdown of eIF4GI down regulates cell ATP level. Using 96 well plate format, HEK293T cells were transfected with eIF4GI Dharmacon SMARTpool siRNA (50 nM). Non- targeting siRNA served as a control. 72h later cells were subjected to ATP levels measurement by CellTiter-Glo Luminescent Cell Viability Assay (promega). The graph represents quantified results of three experiments. The * denote statistically significant difference (p<0.05).

eIF1-eIF4GI interact in mammalian cells Depletion of eIF1A, eIF1 or eIF4GI resulted in a highly similar phenotype, displaying diminished TISU-directed translation, but not canonical AUG context preceded by a short 5’UTR. In addition all these factors appear to be required for scanning and leaky scanning (Fig. 7, 8 and 19). To investigate the underlying basis for the shared activities of eIF1A, eIF1 and eIF4GI, we examined whether eIF1 and eIF1A can interact with eIF4GI. Since eIF1 is located next to the ribosomal P-site, whereas eIF1A is located next to the ribosomal A- site (Hussain et al., 2014; Lomakin et al., 2003; Lomakin & Steitz, 2013; Passmore et al., 2007; Rabl et al., 2011; Weisser et al., 2013; Yu et al., 2009) it is not expected that these two factors interact. Initial co- immunoprecipitation (co-IP) experiments with endogenous proteins failed to reveal an association between eIF1, eIF1A and eIF4GI proteins (data not shown). Considering that such an interaction may be transient, we carried out co-IP following overexpression of HA-tagged eIF1 or HA-eIF1A. The results revealed efficient co-precipitation of endogenous eIF4GI with HA-eIF1, but not with HA-eIF1A or control antibodies (Fig. 21A). To determine whether the eIF1-eIF4GI interaction is direct we carried out GST- pull down assays. GST fussed eIF1 and eIF1A were expressed in bacteria, purified and subjected to in vitro binding reaction with recombinant His-tagged eIF4GI fragments (Fig. 21B). The results revealed an interaction between eIF1 and the middle domain of eIF4GI (amino acid 675-1129) and a weak interaction with the most C-terminal region of eIF4GI (amino acid 1130-1599). These interactions are highly specific as similar amount of GST or GST-eIF1A (Fig. 21B) failed to interact with any of the eIF4GI fragments. We also carried the reciprocal GST pull down assay (Fig. 21C), which confirmed that the central region but not the C-terminal domain is the major eIF1-binding site. These results suggest that mammalian eIF4GI specifically interacts with eIF1 but not with eIF1A.

Figure 21: eIF4GI interacts with eIF1 A. HEK293T cells were transfected with plasmids directing expression of either HA-tagged eIF1 or HA-eIF1A. 24h later cells were lysed and subjected to immunoprecipitation assays using either HA antibody or control antibody. The immune complexes were then run on SDS-PAGE followed by western blot with the indicated antibodies. B. A Schematic representation of the primary structure of Homo sapiens eIF4GI. The black boxes denote the binding sites of PABP, eIF4E, eIF3 and eIF4A.The boxes below represent the eIF4GI fragments used in the assay with their location in the sequence. Four different eIF4GI fragments were sub-cloned into pET28a vector and expressed in E.coli BL21DE3 bacteria. GST pull down assay was performed using each of the fragments and GST-eIF1 (left panel), GST- eIF1A (right panel) or GST coated glutathione sepharose beads. The pulled-down complexes were run on SDS-PAGE followed by western blot with anti-his tag antibody. The arrows indicate 1% input amount of each 4GI fragment. Coomassie blue staining representing the input amount of GST and GST fussed proteins are shown on left to each western blot C. Reciprocal assay in which GST fussed 4GI 675-1129 and 4GI 1130-1599 fragments were expressed , bound to beads and tested for binding to His-eIF1 purified protein. Lane 1 indicated 5% input amount.

eIF1-eIF4GI interaction is required for scanning but not for TISU To investigate further eIF1-eIF4GI interaction we collaborated with the lab of Katsura Asano (Kansas State University) who applied an NMR approach to identify surface residues on yeast eIF1 involved in yeast eIF4G2 binding (aa. 439-513, the minimal binding segment for eIF1)(Singh et al., 2012). Mutational studies identified two pairs of basic residues K52-R53, and K104-H106 of yeast eIF1 as being involved in eIF4G2 binding (Singh et al, manuscript in preparation). These residues are conserved in the human eIF1 protein and correspond to K57-K58 and K109-H111. On the basis of this information we mutated the homologous residues in the human eIF1 (K57A/K58A and K109A/H111A) and examined their effect on the interaction with human eIF4GI using GST pull down assay. As shown in Fig. 22A, these eIF1 mutants are also defective in their ability to interact with the human eIF4GI. We next constructed mammalian expression plasmids of WT and mutant HA- tagged eIF1 variants, transfected them into human cells and performed subsequent co-IP with an anti-HA antibody. We found that these mutated residues impaired their interaction with the endogenous eIF4GI (Fig. 22B). We also examined eIF3c, another PIC component previously shown to interact with eIF1(Singh, He, Ii, Yamamoto, & Asano, 2004), and found that eIF1 K57A/K58A mutant is also impaired in eIF3 binding, in agreement with the structural mapping studies(Selection et al., 2008), while K109A/H111A is not. Thus K109A/H111A mutant is more specific to eIF4GI and was chosen for further analysis.

Figure 22: eIF1-eIF4G1 interaction is required for scanning but not for TISU.A. GST, GST- eIF1 WT and the indicated mutants were coupled to glutathione sepharose beads and incubated eIF4G1 (aa 675-1129), with a C-terminal 6XHis tag. The pulled-down complexes were washed and then run on SDS-PAGE followed by western blot with anti-His antibody. The input represents 1% of the His-eIF4GI used for the binding reaction. The GST and the GST fusion proteins used for the binding reactions are shown in the Coomassie blue stained gel shown on the bottom. The graph represents the average ± SE binding level of 3-4 experiments. The * denotes statistically significant difference (p<0.05). B. HEK293T cells were transfected with plasmids directing expression of HA-tagged eIF1 WT or mutants as indicated. After 48h cells were lysed and subjected to immounoprecipitation using either anti-HA or control antibodies. The immune complexes were analyzed by western blot with the indicated antibodies. C. HEK293T cells were transfected with a mixture of siRNAs against eIF1 and eIF1B. 48h after the initial transfection, cells were co-transfected with these siRNAs together with a siRNA-resistant eIF1 WT or K109A/H111A mutant and GFP reporter plasmid governed by TISU or canonical AUG both with short 5’UTR. A non-targeting siRNA served as a control. Cells were harvested 24h after the second transfection and analyzed by western blot with GFP antibody. US and DS denote upstream and downstream initiation site, respectively. The ratio between the US to DS band is presented.

To determine the function of eIF1 interaction with eIF4GI we examined the ability of eIF1 variants to restore the activity of TISU or leaky scanning, following eIF1 down- regulation. This was done by depletion of endogenous eIF1 and eIF1B by siRNAs and subsequent transfection of siRNA-resistant WT eIF1 or K109A/H111A mutant, together with TISU or canonical AUG GFP reporter (Fig. 3). The upstream AUG of the GFP reporters monitors the translation of leader-less AUG in the TISU or canonical context and the downstream AUG serves as readout of leaky scanning. Consistent with previous findings (Fig. 8D), depletion of eIF1 diminished TISU activity. In contrast the canonical AUG preceded by identical short 5’UTR was unaffected while translation of the downstream, scanning-dependent AUG, was decreased consistent with eIF1’s role in scanning. Transfection of low amount of WT eIF1 restored the TISU activity while higher levels were inhibitory (Fig. 22C). Interestingly the K109A/H111A mutant fully restored TISU activity with no inhibitory activity at the higher concentrations. With the canonical AUG the reduced translation from the scanning dependent DS-AUG was restored by the WT eIF1 in a dose-dependent manner (Fig. 22C), however, the K109A/H111A mutant not only failed to rescue eIF1/1B knockdown phenotype but the effect was even exacerbated, as leaky scanning was diminished to a much greater extent (Fig. 22C). Thus it appears that eIF1-eIF4GI interaction is required only for the scanning-dependent translation but not for TISU mediated translation

The role of eIF4E in translation initiation fidelity and ribosomal scanning

Since both, eIF4E and eIF1 interact with eIF4GI (He et al., 2003; Shabalina, Ogurtsov, Rogozin, Koonin, & Lipman, 2004; Yanagiya et al., 2009) and eIF1-eIF4GI interaction seems to promote scanning, we next wished to determine the role of eIF4E- eIF4GI interaction and its effect on scanning. For this purpose we analyzed the effect of 4EGI-1, a specific allosteric inhibitor of eIF4E-eIF4GI complex formation (Moerke et al., 2007) on translation of GFP reporter genes described in Fig. 3. Cells were transfected with each of the GFP reporter genes and then treated with increasing concentrations of 4EGI-1. The exclusive translation from the US-AUG directed by TISU appears to be 4 fold more sensitive to 4EGI-1 than the canonical AUG, displaying IC-50 of 25μM and

100μM, respectively (Fig. 23A). Note that the DS-AUG that is dependent on scanning is more resistant to the drug than the scanning-free US-AUG. Regarding translation of the long unstructured 5’UTR mRNA, at a low dose of 4EGI-1 it is enhanced and it became sensitive to the drug only at the higher concentrations (Fig. 23A). Next, we depleted endogenous eIF4E levels using siRNA. We used moderate levels of eIF4E siRNA (25 nM), which result in ~50% depletion of eIF4E. With these eIF4E levels there was ~70% reduction of TISU activity while no effect was seen with the canonical US-AUG or the leaky scanning (Fig. 23B). Remarkably, as opposed to eIF4GI and eIF1 knockdown experiments which decreased scanning-dependent translation, eIF4E knockdown elevated scanning dependent translation (Fig. 23B). Interestingly, the enhancement of scanning following partial eIF4E depletion is reminiscent of the enhancement of scanning following eIF1 overexpression (Fig. 5A), suggesting that eIF1 and eIF4E affect scanning antagonistically. Thus, the differential effects of the 4EGI-1 drug and eIF4E down-regulation suggest that TISU confers strong sensitivity to the intra-cellular levels of the cap complex. Moreover, it seems that eIF4E-eIF4GI interaction imposes some constrains on scanning.

Figure 23: Cap-complex availability affects translation differentially. A. HEK293T cells were transfected with TISU, canonical or long 5’UTR GFP reporter gene. Six hours later, increasing amounts of 4EGI-1 were added to the media as indicated. Cells were harvested 24h after transfection and subjected to western blot using anti-GFP antibody. Representative immune-blots are shown. The graphs denote the average ± SE of densitometric measurements of GFP levels of 3-5 experiments. The overall translation in control cells was set to one. Light and dark grey bars represent the relative translation from the US-AUG and DS-AUG, respectively B. eIF4E knockdown effects on translation initiation fidelity and scanning. HEK293T cells were transfected with moderate levels of eIF4E siRNA (25nM). A non-targeting siRNA was used as a control. 48h later cells were transfected again with TISU, canonical or long 5’UTR GFP reporter genes as indicated. Cells were harvested 24h after the second transfection and cell lysates were analyzed by western blot with GFP, eIF4E and Tubulin antibodies as indicated. US and DS denote upstream and downstream initiation site, respectively. Representative immune-blots are shown. The graph represents the mean ± SE of 3-4 independent experiments. The * denotes statistically significant difference (p<0.05). .

Interaction of eIF4E and eIF1 with eIF4GI is mutually exclusive Both, eIF4E and eIF1 interact with eIF4GI (He et al., 2003; Shabalina et al., 2004; Yanagiya et al., 2009). While eIF1 and eIF4GI co-operate in the promotion of scanning and leaky scanning it seems that eIF4E suppresses scanning (Fig. 8, 19 and 23A). To examine the interplay between eIF4E, eIF1 and eIF4GI we set to investigate further the interaction between these factors. We transfected into cells HA-tagged eIF1 and analyzed the co-immunoprecipitated proteins. The results revealed that while eIF4GI and eIF3c were co-precipitated with eIF1, the cap complex subunits eIF4E and eIF4A were not (Fig. 24A). To validate this observation cells were transfected with HA-tagged eIF4E or HA-tagged eIF4A and associated proteins were analyzed. eIF4GI and eIF4A as well as eIF3c were co-precipitated with eIF4E, however eIF1 was undetected in the immune complex (Fig. 24B). Similarly, eIF4GI, eIF3c and eIF4E were co-precipitated with eIF4A but eIF1 was not (Fig. 24C). These findings suggest that eIF4GI exists in two distinct mutually exclusive complexes, one with eIF4F and another with eIF1, which regulate scanning and leaky scanning in an antagonistically manner.

Figure 24: Interaction of eIF1 and eIF4E/A with eIF4GI is mutually exclusive. A. HEK293T cells were transfected with plasmids directing expression HA-tagged eIF1 and 24h later cells were lysed and subjected to immunoprecipitation assays using either anti-HA or control antibodies. The immune complexes were then run on SDS-PAGE followed by western blot with the indicated antibodies. B. The same experiments as in A with HA-eIF4E C. The same experiments as in A with HA-eIF4A1. The * denotes unspecific band.

Those striking results prompted us to test whether eIF4E and eIF1 compete with each other on the binding to eIF4GI. Therefore we established split-Renilla (RL) protein- protein interaction system (Paulmurugan 2003) in which fragment of eIF4GI (600-1129 aa) which contains both eIF4E and eIF1 binding sites (612-618 aa and 675-1129 aa, respectively) was fused to the RL N-termini and eIF4E was fused to the RL C-termini (Fig. 25A). In this assay, interaction of the target proteins brings the RL N and C termini into close proximity, restoring RL enzymatic activity. HEK293T cells were co-transfected with both constructs together with a firefly luciferase (FL) reporter as a normalizing control. The cells were also transfected with eIF4GI (600-1129)-N-RL and eIF4E-C-RL constructs and their empty counterparts to determine the background levels of RL activity. As shown in Fig. 25B, the strongest activity was achieved by the eIF4GI (600- 1129)-N-RL and eIF4E-C-RL. To validate that the RL activity is dependent on eIF4GI- eIF4E interaction we performed additional tests. First we examined the effect of 4EGI-1, a specific allosteric inhibitor of eIF4E-eIF4GI complex formation (Moerke et al., 2007), on RL enzymatic activity. Indeed in a dose-dependent assay we confirmed that 4EGI-1 inhibits RL activity with an IC-50 of ~50 µM (Fig. 25C). Next, we cloned the middle part of eIF4GI (675-1129 aa) which lacks the eIF4E binding site, fussed to the N-RL and used it to co-transfect cells together with eIF4E-C-RL plasmid. Indeed we observed a ~5-fold reduction in RL activity. Interestingly we found some residual RL activity which is significantly higher than the background control (Fig. 25B). This observation suggests the existence of two eIF4E binding sites on eIF4GI, one of which is the major binding site (4E1) and the other is a novel site which is located in the middle part of eIF4GI (4E2). Considering the mutual exclusive interaction of eIF4GI with eIF4E and eIF1 we tested whether eIF1 influences eIF4GI-eIF4E interaction. We set an in vitro competition assay in which purified eIF1 was added to cell lysates containing either eIF4GI (600- 1129)-N-RL or eIF4GI (675-1129)-N-RL paired with eIF4E-C-RL. The luminescence assay shows that the addition of eIF1 consistently decreases eIF4GI (600-1129 aa)-eIF4E interaction by ~ 10%. Interestingly, the effect of eIF1 addition to eIF4GI (675-1129 aa)- eIF4E pair resulted in much greater inhibition of the RL activity (~ 30%) suggesting that the 4E2 binding domain of eIF4G1 is more sensitive to eIF1 (Fig. 25D). Thus, our results suggest that eIF1 and eIF4E compete on the binding to the middle conserved domain of eIF4GI.

Figure 25: eIF1 competes with eIF4E on the binding to eIF4GI. A. Schematic representation of the constructs used in the experiment and schematic illustration of the split-Renilla luciferase principle. B. HEK293T cell were co-transfected with the indicated plasmids together with miR- 22–FL, which served as an internal control. RL and FL activities were measured after 24 h. The graph represents the average ± SE of Renilla Luminescence level of 3-4 experiments. The activity of eIF4GI (600-1129)-N-RL and C-RL-eIF4E pair was set to 1. C. Dose-response of eIF4GI (600-1129)-N-RL and C-RL-eIF4E pair to 4EGI-1 inhibitor. D. Exogenous His-eIF1was added to the indicated cell lysates. Renilla luminescence was measured after 30 minutes of incubation at room temperature. The graph represents the average ± SE of Renilla Luminescence level of 4-5 experiments. The activities of the split-renilla pairs without His-eIF1 were set to 1. The * denotes statistically significant difference (p<0.05).

Discussion

Previously our lab had identified and characterized TISU, an element optimized to direct efficient and accurate translation initiation from mRNAs with a 5’ UTR as short as 5 nucleotides (Elfakess & Dikstein, 2008; Elfakess et al., 2011). These properties of TISU are unprecedented in the light of previous studies showing that an AUG within a favored context requires a 5' UTR of at least 20 nt for accurate translation initiation (M Kozak, 1991a). Importantly, it was shown in several different assays that TISU activity is cap-dependent (Elfakess et al., 2011). These findings suggest a non-canonical translation initiation mode that is cap-dependent however scanning independent (Elfakess et al., 2011). Translation from short 5’UTR is a poorly investigated phenomenon, in particular in higher eukaryotes. In this study I examined the sequence and translation initiation factors requirements for the non-canonical translation initiation mechanism driven by the TISU element. In addition, I revisited the role of AUG context, 5’UTR length and translation initiation factors in regulation of scanning and start codon selection. Most of the previous studies addressing the role of translation initiation factors in translation initiation fidelity and scanning used either yeast genetics or in vitro translation systems; whereas in this study we analyzed those aspects of translation initiation, within human tissue culture cells. For many years the optimal nucleotide context for translation initiation in mammals was the Kozak context (RCCAUGG) (Kozak 1991). TISU consensus (SAASAUGGCGGC) contains Kozak element’s key positions, purine at position -3 and G at position +4, however, its -2 and -1 positions are distinct from those of Kozak and positions -4 and +5 to +8 are unique to TISU and absent from Kozak (Elfakess & Dikstein, 2008). Here we show that many AUG flanking nucleotides contribute to the translational strength of TISU element. The most critical nucleotides for the high fidelity of TISU are purine at position -3 and nucleotides +4, +5 and +6, which cooperate to support initiation accuracy. It had been shown that mutations in positions -3 or +4 of Kozak element reduce translation; however, relatively lesser effects were observed for substitutions at position +4. Interestingly, in yeast the nucleotide context is less important for initiation codon recognition, but purine at position -3 is critical for start codon recognition (Cavener & Ray, 1991; Marilyn Kozak, 1986). Consistently, a recent genome wide comparative analysis of orthologous sets of mammalian and yeast mRNAs revealed

64 high preference for purine at position -3 (Shabalina et al., 2004). Thus, it seems that the common preference for purine at position -3 is a conserved feature of eukaryotic mRNAs. It was suggested that the purine in position -3 and the G in position +4 stabilize the 48S following recognition of the start codon by direct interactions they make with 48S components (Pisarev et al., 2006). Therefore, the translational strength and high fidelity of TISU can be explained by unique interactions between AUG flanking nucleotides of TISU to different ribosomal proteins (Sinvani Hadar, unpublished data), which compensate for the lack of 5’UTR contacts and contribute to TISU recognition and utilization. It seems that many TISU mutants exhibit decreased translational strength due to weaker association with the ribosomal proteins. Most substitutions of TISU nucleotides result with no or mild leaky scanning. This finding is overwhelming since translation from Kozak short 5’ UTR mRNA (5’UTR=5 nucleotides) results with more than 50% leaky scanning to DS-AUG. Our detailed mutagenesis analysis of TISU element emphasizes its robustness and further establishes it as an optimal initiator of short 5’ UTR mRNAs which is distinguished from the Kozak element in its sequence requirements. Besides the sequence requirement there are additional features which discriminate TISU from Kozak element including TISU strict location in the 5’UTR, its dual function in transcription and translation and its mode of translation which is scanning-free (Dikstein, 2012). It would be interesting to determine how these TISU mutants influence its activity in transcription. Using in vitro reconstitution of the 48S complex from purified 40S, purified eIFs and Met-tRNAi our lab showed that the basal translation machinery has intrinsic capability to initiate from TISU indicating that no additional cellular factor is required for the basal activity of TISU (Sinvani et al., 2015). These findings prompted us to focus on the role that basal translation initiation factors play in controlling translation driven by distinct AUG contexts and 5’UTR lengths, using overexpression and knockdown experiments. Both eIF1A and eIF1 are key players in the regulation of scanning and start codon selection. Recent structural studies reveal that these factors bind to a different surface of the small ribosomal subunit (Hussain et al., 2014; Lomakin & Steitz, 2013; Passmore et al., 2007; Rabl, Leibundgut, Ataide, Haag, & Ban, 2011; Weisser, Voigts- hoffmann, Rabl, Leibundgut, & Ban, 2013; Yu et al., 2009), suggesting that the mechanism by which they affect scanning and start codon selection is different. eIF1A

65 displays two opposing functions through the competing influences of its two termini (Fekete et al., 2005, 2007; Mitchell & Lorsch, 2008). The C-terminal tail increases the stringency of initiation, induces the ‘open’ ribosomal conformation and promotes scanning, whereas the N-terminal tail decreases the accuracy of initiation, induces the ‘closed’ ribosomal conformation and promotes scanning arrest (Fekete et al., 2005, 2007; Mitchell & Lorsch, 2008). Consistent with the opposing roles that eIF1A tails play in translation, we found that, overexpression of the WT eIF1A stimulated translation initiation from canonical AUG context preceded by short 5’UTR, while overexpression of the N-terminus HA-tagged/mutated eIF1A promoted leaking to the downstream AUG. Interestingly, both forms of eIF1A had no significant effect on translation initiation from TISU containing short 5’UTR mRNA. Thus, their effect is AUG context dependent. As mentioned above eIF1 is also a major regulator of scanning and start codon selection. Previous studies have shown that eIF1 facilitates AUG codons discrimination by the initiation complex on the basis of their nucleotide context (favoring the canonical Kozak sequence in the mRNA) and their location relative to the cap (Tatyana V. Pestova & Kolupaeva, 2002). It promotes the dissociation of the 43S from non-AUG codons, from AUG codons in a poor context and from AUG codon located very close to the 5’ end of the mRNA (Lomakin et al., 2003; Tatyana V. Pestova & Kolupaeva, 2002). In agreement with these previous observations we found that elevated levels of eIF1 inhibited AUG selection in a canonical context when it was located close to the cap (17 nt). Remarkably, eIF1 failed to inhibit initiation from TISU under the same conditions. Thus its effect is also AUG context dependent. We suggest that TISU mRNA are refractory to the inhibitory effects of either N-terminus mutated eIF1A or eIF1 due to the strong binding TISU makes with the small ribosomal subunit (Sinvani, unpublished data). This binding induces rapid transition into the ‘closed’ ribosomal conformation which is no longer responsive to the effects of eIF1A or eIF1. Moreover, our results support the previously demonstrated antagonistic roles of eIF1 and eIF1A since overexpression of eIF1 decreased translation from the canonical AUG preceded by short 5’UTR, while eIF1A promoted it. A model of a good cop/bad cop was proposed for the actions of eIF1 and eIF1A (Mitchell & Lorsch, 2008). It was suggested that while eIF1 prevents translation initiation commitment on putative start codons, eIF1A facilitates pausing at start codons long enough to proceed with initiation. Thus the degree of leaky scanning is likely to be

66 governed by the relative levels of these factors. Indeed we found that eIF1 and eIF1A protein levels vary between different cell types and that the extent of leaky scanning in human tissue culture cells varies accordingly. Therefore it seems that eIF1 to eIF1A ratio can influence the protein milieu of the cells in a cell-type-dependent manner. Our findings suggest that the effects of eIF1A and eIF1 overexpression are also 5’UTR length dependent. We show that both N-terminus mutated eIF1A and eIF1 promote translation from long 5’UTR. This observation is in agreement with their ability to promote scanning. It seems that efficient and continuous scanning results in higher translation since it enables the mRNA to load more ribosomal PICs. Given that WT eIF1A had no significant effect on translation from long 5’UTR mRNA, but the N-terminus mutated form significantly promoted it; we again demonstrate the opposing effects of eIF1A tails and the importance of the C-terminal tail in the stimulation of scanning. It would be interesting to further examine eIF1A and eIF1 effects on translation of structured long 5’UTR mRNA. Interestingly, although eIF1A and eIF1 overexpression had no effect on TISU mediated translation, down-regulation of either eIF1A or eIF1 and its paralog eIF1B, caused severe impairment of TISU mRNA translation, suggesting that TISU driven translation is eIF1A and eIF1 dependent and that the protein levels of both factors in HEK293T cells are not rate-limiting and actually sufficient to support efficient translation from TISU. Here again, we detected AUG context dependent differential effects since translation from the canonical AUG preceded by short 5’UTR barely changed upon eIF1A or eIF1/1B knockdown. The effect of eIF1 depletion on TISU translation was seemingly unexpected considering ours and others data demonstrating that eIF1 inhibits translation from an AUG preceded by a short 5’UTR (Elfakess et al., 2011; Tatyana V. Pestova & Kolupaeva, 2002). However, those findings are in agreement with in vitro analysis of initiation factors requirements which was conducted in our lab and indicated that removal of eIF1 decreases initiation from TISU AUG, while translation from the canonical AUG is enhanced (Sinvani et al., 2015). Another support for the results comes from endogenous genes translation analysis that shows that translation of specific TISU genes is eIF1/eIF1B dependent. Similar analysis should be further done in order to demonstrate the requirement of eIF1A for translation of endogenous TISU genes. The protein levels of eIF1 and eIF1A considerably fluctuate during the cell cycle. In

67 vertebrates, entry into mitosis (i.e., G2/M transition) is controlled by some checkpoints during cell cycle progression. When cells cannot satisfy the mitotic checkpoints they are delayed in mitosis or stop further cell cycle progression. Recent live cell studies revealed that normal and cancer cell including U2OS exhibit mitotic arrest (Rieder, 2011). It is possible that the elevated levels of eIF1 during G2/M transition are part of some checkpoint control pathway. Moreover, during the S phase there is rapid and strict high expression of histone genes (Jaeger, Barends, Giegé, Eriani, & Martin, 2005). It is possible that elevated eIF1 levels contribute to the down-regulation of histone genes translation during the G2/M phase of the cells cycle. Furthermore, it would be interesting to examine whether TISU activity is also cell cycle regulated. Our previous findings, strongly argue for the cap-dependent nature of initiation directed by TISU. Taking into account the short length of the 5’ UTR of TISU mRNAs, it is expected that both the cap and the 5’ UTR nucleotides will be within the ribosomal exit channel upon 80S formation. This situation is incompatible with continued association of eIF4F with the mRNA through the cap. We tested the requirement of eIF4GI and eIF4E, two subunits of eIF4F complex, for TISU activity. Our data shows that TISU translation is highly dependent on eIF4GI and eIF4E, hence, it is unclear how cap-dependent translation of TISU short 5’ UTR mRNAs is possible. Analysis of translation initiation fate in 48S and 80S TISU ribosomal complexes that was conducted in our lab supports a model in which the cap-binding complex is evicted from the 5’ end of the mRNA upon 48S initiation complex formation (Sinvani et al., 2015) enabling translation from 5’- proximal AUG as illustrated in Fig. 26 (left box). The release of the cap complex occurs possibly due to steric interference between the cap-eIF4F complex and the small ribosomal subunit (Sinvani et al., 2015). It is therefore expected that in TISU mRNA each new translation initiation cycle necessitates de novo recruitment of eIF4F to the mRNA cap (Fig. 26, left). On the other hand, when leaky scanning occurs as with the canonical AUG, the cap complex is not evicted and actually it is retained on the mRNA ready for the next translation cycle. In this case recruitment of eIF4F de novo is less frequent (Fig. 26, right). This model predicts that the effect of variations in the intra- cellular concentrations of the cap complex on TISU would be stronger. The experiments reported here on the differential sensitivity of TISU and non-TISU mRNA to eIF4GI and eIF4E levels, support this model further. As TISU efficiently prevents leaky scanning,

68 each new translation initiation cycle requires recruitment of eIF4F from the beginning, rendering TISU highly dependent on high concentration of active cap-complex. On the other hand, a short 5’UTR that is followed by an AUG with a context that permits leaky scanning, retains the cap complex on the mRNA at a higher frequency and is therefore less sensitive to changes in eIF4E levels. Apparently, the least sensitive mRNA is the one that has sufficiently long 5’UTR as in this case the dissociation of the cap complex from the mRNA is much less recurrent. These findings are consistent with a recent study that examined the global effect of mTOR inhibitors on translation and reported that in addition to the TOP mRNA, TISU containing genes are among the most sensitive to pharmacological inhibition of mTOR complex 1, a positive regulator of eIF4E(Gandin et al., 2016). Our findings therefore provide a mechanistic basis to the differential sensitivity displayed by different classes of mRNA to intra-cellular levels of active eIF4E even though it is required for the translation of all cap-dependent translation.

Figure 26: A model for multiple cycles of translation initiation from TISU and non-TISU short 5’UTR mRNAs. In the case of TISU, recruitment of the 43S PIC by the cap-binding complex is followed by an immediate and efficient recognition of the 5’ proximal TISU-AUG and eviction of eIF4F from the cap to avoid steric clash between the cap and the ribosome (Sinvani et al., 2015). Therefore each of the next translation cycles requires engagement of eIF4E with the cap from the beginning. In translation initiation from short 5’UTR with canonical AUG context there are two possible states. In the first, translation occurs from the cap-proximal AUG and involves eIF4F detachment. In this case, the next translation cycle requires eIF4E recruitment to the cap de-novo

69 as in TISU. In the second, translation initiates at a downstream AUG as a consequence of leaky scanning. In this scenario eIF4F is retained on the cap ready for the next translation initiation cycle. Note that an exchange of eIF4GI from eIF4E to eIF1 is necessary in order to promote leaky scanning.

Unexpectedly, we also uncovered a pivotal role of eIF4GI in translation initiation fidelity regulation in mammalian cells, eIF4GI promotes leaky translation initiation when the AUG is too close to the cap, while facilitating TISU- mediated translation as well as ribosomal scanning. All these features of eIF4GI are common to those attributed to eIF1 and eIF1A. Importantly, we show that eIF1 directly interacts with the middle conserved domain of eIF4GI as in yeasts, (He et al., 2003). We found that eIF1-eIF4GI interaction is dispensable for translation of TISU, but seems to promote leaky scanning. We propose that eIF1-eIF4GI interaction is important for promotion of leaky scanning probably by induction of the ‘open’ scanning-compatible conformation. Our findings also revealed that eIF4GI is present in two mutually exclusive complexes, one as part of the cap binding eIF4F complex and the second with eIF1. Remarkably, these complexes appear to have opposing effects on scanning. While eIF1-eIF4GI complex promotes scanning, eIF4E-eIF4GI suppresses it. Considiring this and the split-Renilla complementation assay results which suggest that simillarly to 4E-Binding Proteins (4E-BPs)(Peter D et al., 2015) eIF4GI also possesss two eIF4E binding sites, one of which is the major binding site (4E1) (Yanagiya et al., 2009, Mader et al., 1995) and the other one (4E2) is located in the middle domain of eIF4GI (amino acids 675-1129) we assume that 4E2 site and eIF1 binding site are orientated in an over-lapping manner. On the basis of these findings we suggest a model in which the cap complex recruits eIF1-containing 43S complex. Subsequently an exchange of eIF4GI from eIF4F to eIF1 occurs, resulting in dissociation of the small ribosomal subunit from the cap and entrance into the scanning phase (Fig. 27). Considering that the binding affinity of eIF4E towards eIF4GI is higher than that of eIF1, the transition may also involve other 43S components, such as eIF3 with RNA- binding subunits(Asano et al., 1997; Lee, Kranzusch, & Cate, 2015) and eIF5 which binds the same segment of eIF4G in yeast (He et al., 2003; Singh et al., 2012; Yamamoto et al., 2005). Further work is warranted to fine map 4E2 position and its orientation relative to eIF1 binding site and to delineate the mechanism of eIF4E-eIF1 exchange.

Figure 27: A model describing the interplay between eIF4E, eIF4GI and eIF1 in scanning- dependent translation. The 43S PIC is recruited to the mRNA through the cap-binding complex. The transition of eIF4GI from eIF4F to eIF1 facilitates the detachment of the 43S ribosomal from the cap and entrance to the scanning phase.

Unlike the AUG context and 5’UTR length dependent effects of: eIF1A, eIF1, eIF4GI and eIF4E, it seems that eIF3c and eIF5 are required for the translation of all tested mRNAs, irrespective of 5’UTR length and AUG context. Therefore, apparently eIF1, eIF1A, eIF4GI and eIF4E play some unique roles in translation of TISU short 5’ UTR mRNA, however, these roles are still unclear. There is a possibility that TISU mediated translation relies on the ‘open’ conformation those factors induce. The ‘open’ conformation may be necessary in order to enable high affinity binding of TISU sequence to 48S components and subsequent recognition of TISU AUG. It would be of much interest to further elucidate the ribosomal conformation while being engaged with TISU short 5’UTR mRNA by Cryo-EM analysis. The short 5’UTR length in TISU mRNAs evoked the possibility that translation through TISU would be less dependent on eIF4A, an ATP dependent RNA helicase that unwinds secondary structures on the mRNA 5’UTR and facilitates scanning (Tatyana V. Pestova & Kolupaeva, 2002; Svitkin et al., 2001). In accordance with this expectation, TISU mRNA translation was the least affected by overexpression of eIF4A Dominant- negative mutant, while translation from unstructured or structured long 5’UTR mRNAs was sensitive to eIF4A inhibition. However efficient and dramatic down-regulation of eIF4AI by siRNAs resulted in global inhibition of translation, meaning, translation of

71 either short or long 5’UTR mRNAs including TISU was decreased upon eIF4AI depletion. Therefore, it seems that translation through TISU is eIF4A dependent. In agreement with these results, in vitro translation of total yeast RNA showed that most yeast mRNAs are eIF4A dependent. Moreover, in vitro translation of 8 nucleotides short 5’ UTR mRNA was also found to be eIF4A dependent (Blum et al., 1992). It was assumed that eIF4A is needed to enable ribosome attachment. A scanning-free mechanism that is cap-dependent was already described for the short 5′ UTR mRNA of the protozoan G. lamblia which ranges between 0-14 nt without any consensus sequence. This unusual short 5′ UTR appears to be necessary for efficient translation since increasing its length resulted in a dramatic decrease in translation, contrary to other eukaryotes in which translation efficiency increases with larger 5′ UTR length (Elfakess et al., 2011; M Kozak, 1991b; Marilyn Kozak, 2002; Tatyana V. Pestova & Kolupaeva, 2002). In this protozoan, the translation mechanism involves the cap, however scanning is absent or irrelevant since accurate initiation, without leaky translation, occurs even with a 1 nt long UTR(Li & Wang, 2004). Presently the exact mechanistic details of this initiation mode are not known. Another interesting example for cap-dependent, scanning independent translation initiation regards Histone H4 mRNA. Martin et al. used H4 mRNA and demonstrated in vitro that the cap complex is important for its translation but not for ribosomal recruitment (Martin et al., 2011). This study showed that H4 mRNA contains two structural elements within its ORF, which recruit eIF4E and promote ribosome binding and positioning on the AUG initiation codon. Still, the in vivo relevance of such non-canonical mechanism should be further examined. Few recent studies examined the physiological significance of TISU mediated translation. One such study revealed that mitochondrial mRNAs are enriched with TISU element which confers translational resistance to global inhibition of translation in response to energy stress but not to other stresses(Sinvani et al., 2015). Another study which studied the establishment of rhythmic gene regulation in mouse liver by circadian and feeding rhythms found that TISU and TOP (5′-Terminal Oligo Pyrimidine tract) mRNAs present transcription-independent rhythmic translation mainly regulated by feeding (Atger et al., 2015). It would be interesting to uncover additional physiological settings in which TISU mRNAs are differentially regulated and to unravel the mechanistic events which are associated.

In summary, in this study we showed that the effects of eIFs may be general or differential. The differential effects are mRNA characteristics dependent. It seems that the overall effect on translation is achieved through the cooperative actions of translation initiation factors and features present in the mRNA. Future studies of the roles of translation initiation factors will shed more light on TISU and canonical translation initiation mechanism. From the experiments presented here and from additional data accumulated in our lab, a unique mode of translation initiation through TISU has emerged. TISU directs high efficiency and fidelity translation initiation from short 5’UTR mRNAs and displays differential sequence and factor requirements. These features clearly discriminate between TISU and other AUG contexts including Kozak, which cannot support accurate translation initiation from leader-less mRNAs.

Bibliography

Aitken, C. E., & Lorsch, J. R. (2012). A mechanistic overview of translation initiation in eukaryotes. Nature Structural & Molecular Biology, 19(6), 568–76. doi:10.1038/nsmb.2303 Asano, K., Vornlocher, H., Richter-cook, N. J., Merrick, W. C., Hinnebusch, A. G., & Hershey, J. W. B. (1997). Structure of cDNAs Encoding Human Eukaryotic Initiation Factor 3 Subunits, 272(43), 27042–27052. Atger, F., Gobet, C., Marquis, J., Martin, E., Wang, J., Weger, B., & Lefebvre, G. (2015). Circadian and feeding rhythms differentially affect rhythmic mRNA transcription and translation in mouse liver, 6579–6588. doi:10.1073/pnas.1515308112 Berthelot, K., Muldoon, M., Rajkowitsch, L., Hughes, J., & McCarthy, J. E. G. (2004). Dynamics and processivity of 40S ribosome scanning on mRNA in yeast. Molecular Microbiology, 51, 987–1001. doi:10.1046/j.1365-2958.2003.03898.x Blum, S., Schmid, S. R., Pause, a, Buser, P., Linder, P., Sonenberg, N., & Trachsel, H. (1992). ATP hydrolysis by initiation factor 4A is required for translation initiation in Saccharomyces cerevisiae. Proceedings of the National Academy of Sciences of the United States of America, 89(August), 7664–7668. doi:10.1073/pnas.89.16.7664 Calvo, S. E., Pagliarini, D. J., & Mootha, V. K. (2009). Upstream open reading frames cause widespread reduction of protein expression and are polymorphic among humans. Proceedings of the National Academy of Sciences of the United States of America, 106(18), 7507–7512. doi:10.1073/pnas.0810916106 Cavener, D. R., & Ray, S. C. (1991). Eukaryotic start and stop translation sites. Nucleic Acids Research, 19(12), 3185–3192. doi:10.1093/nar/19.12.3185 Cheung, Y.-N., Maag, D., Mitchell, S. F., Fekete, C. a, Algire, M. a, Takacs, J. E., … Hinnebusch, A. G. (2007). Dissociation of eIF1 from the 40S ribosomal subunit is a key step in start codon selection in vivo. Genes & Development, 21(10), 1217–1230. doi:10.1101/gad.1528307 Chuang, R. Y., Weaver, P. L., Liu, Z., & Chang, T. H. (1997). Requirement of the DEAD- Box protein ded1p for messenger RNA translation. Science (New York, N.Y.), 275(March), 1468–1471. doi:10.1126/science.275.5305.1468 Cigan, a M., Feng, L., & Donahue, T. F. (1988). tRNAi(met) functions in directing the scanning ribosome to the start site of translation. Science (New York, N.Y.), 242(4875), 93–7. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/3051379 Dikstein, R. (2012). Transcription and translation in a package deal: The TISU paradigm. Gene, 491(1), 1–4. doi:10.1016/j.gene.2011.09.013 Elfakess, R., & Dikstein, R. (2008). A translation initiation element specific to mRNAs with very short 5′UTR that also regulates transcription. PLoS ONE, 3(8), e3094. doi:10.1371/journal.pone.0003094

Elfakess, R., Sinvani, H., Haimov, O., Svitkin, Y., Sonenberg, N., & Dikstein, R. (2011). Unique translation initiation of mRNAs-containing TISU element. Nucleic Acids Research, 39(17), 7598–7609. doi:10.1093/nar/gkr484 Erijman, A., Dantes, A., Bernheim, R., Shifman, J. M., & Peleg, Y. (2011). Transfer-PCR ( TPCR ): A highway for DNA cloning and protein engineering. Journal of Structural Biology, 175(2), 171–177. doi:10.1016/j.jsb.2011.04.005 Fekete, C. a, Applefield, D. J., Blakely, S. a, Shirokikh, N., Pestova, T., Lorsch, J. R., & Hinnebusch, A. G. (2005). The eIF1A C-terminal domain promotes initiation complex assembly, scanning and AUG selection in vivo. The EMBO Journal, 24(20), 3588–3601. doi:10.1038/sj.emboj.7600821 Fekete, C. a, Mitchell, S. F., Cherkasova, V. a, Applefield, D., Algire, M. a, Maag, D., … Hinnebusch, A. G. (2007). N- and C-terminal residues of eIF1A have opposing effects on the fidelity of start codon selection. The EMBO Journal, 26(6), 1602– 1614. doi:10.1038/sj.emboj.7601613 Gandin, V., Masvidal, L., Hulea, L., Gravel, S., Cargnello, M., Mclaughlan, S., … Topisirovic, I. (2016). nanoCAGE reveals 5 ′ UTR features that define specific modes of translation of functionally related MTOR-sensitive mRNAs, 636–648. doi:10.1101/gr.197566.115.Freely He, H., von der Haar, T., Singh, C. R., Ii, M., Li, B., Hinnebusch, A. G., … Asano, K. (2003). The yeast eukaryotic initiation factor 4G (eIF4G) HEAT domain interacts with eIF1 and eIF5 and is involved in stringent AUG selection. Molecular and Cellular Biology, 23(15), 5431–5445. doi:10.1128/MCB.23.15.5431 Hilliker, A., Gao, Z., Jankowsky, E., & Parker, R. (2011). The DEAD-Box Protein Ded1 Modulates Translation by the Formation and Resolution of an eIF4F-mRNA Complex. Molecular Cell, 43(6), 962–972. doi:10.1016/j.molcel.2011.08.008 Hinnebusch, A. G. (2014). The scanning mechanism of eukaryotic translation initiation. Annual Review of Biochemistry, 83, 779–812. doi:10.1146/annurev-biochem- 060713-035802 Huang, H. K., Yoon, H., Hannig, E. M., & Donahue, T. F. (1997). GTP hydrolysis controls stringent selection of the AUG start codon during translation initiation in Saccharomyces cerevisiae. Genes and Development, 11(18), 2396–2413. doi:10.1101/gad.11.18.2396 Hussain, T., Llácer, J. L., Fernández, I. S., Munoz, A., Martin-Marcos, P., Savva, C. G., … Ramakrishnan, V. (2014). Structural Changes Enable Start Codon Recognition by the Eukaryotic Translation Initiation Complex. Cell, 597–607. doi:10.1016/j.cell.2014.10.001 Iacono, M., Mignone, F., & Pesole, G. (2005). uAUG and uORFs in human and rodent 5???untranslated mRNAs. Gene, 349, 97–105. doi:10.1016/j.gene.2004.11.041 Jaeger, S., Barends, S., Giegé, R., Eriani, G., & Martin, F. (2005). Expression of metazoan replication-dependent histone genes, 87, 827–834.

doi:10.1016/j.biochi.2005.03.012 Kaye, N. M., Emmett, K. J., Merrick, W. C., & Jankowsky, E. (2009). Intrinsic RNA binding by the eukaryotic initiation factor 4F depends on a minimal RNA length but not on the m7G cap. Journal of Biological Chemistry, 284(26), 17742–17750. doi:10.1074/jbc.M109.009001 Kozak, M. (1986). Kozak M. Point mutations define a sequence flanking the AUG initiator codon that modules translation by eukaryotic ribosomes. Cell 1986; 44:283- 92., 44, 283–292. Kozak, M. (1991a). A short leader sequence impairs the fidelity of initiation by eukaryotic ribosomes, Kozak M. 1991 Gene Expr.1(2):p111-5. Gene Expr., 1(2), 111–5. Kozak, M. (1991b). Effects of long 5’ leader sequences on initiation by eukaryotic ribosomes in vitro. Gene Expression, 1(2), 117–25. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/1820209 Kozak, M. (2002). Pushing the limits of the scanning mechanism for initiation of translation. Gene, 299(1-2), 1–34. Kozak, M. (2005). Regulation of translation via mRNA structure in prokaryotes and eukaryotes. Gene, 361(1-2), 13–37. doi:10.1016/j.gene.2005.06.037 Lee, A. S. Y., Kranzusch, P. J., & Cate, J. H. D. (2015). eIF3 targets cell-proliferation messenger RNAs for translational activation or repression. doi:10.1038/nature14267 Lefebvre, A. K., Korneeva, N. L., Trutschl, M., Cvek, U., Duzan, R. D., Bradley, C. a., … Rhoads, R. E. (2006). Translation initiation factor eIF4G-1 binds to eIF3 through the eIF3e subunit. Journal of Biological Chemistry, 281(32), 22917–22932. doi:10.1074/jbc.M605418200 Li, L., & Wang, C. C. (2004). Capped mRNA with a single nucleotide leader is optimally translated in a primitive eukaryote, Giardia lamblia. The Journal of Biological Chemistry, 279(15), 14656–14664. doi:10.1074/jbc.M309879200 Lomakin, I. B., Kolupaeva, V. G., Marintchev, A., Wagner, G., & Pestova, T. V. (2003). Position of eukaryotic initiation factor eIF1 on the 40S ribosomal subunit determined by directed hydroxyl radical probing Ivan. Genes & Development, 17, 2786–2797. doi:10.1038/sj.emboj.7601751 Lomakin, I. B., & Steitz, T. a. (2013). The initiation of mammalian protein synthesis and mRNA scanning mechanism. Nature, 500(7462), 307–11. doi:10.1038/nature12355 Lorsch, J. R., & Herschlag, D. (1998). The DEAD box protein eIF4A. 1. A minimal kinetic and thermodynamic framework reveals coupled binding of RNA and nucleotide. Biochemistry, 37(8), 2180–2193. doi:10.1021/bi972430g Marintchev, A. (2013). Roles of helicases in translation initiation: A mechanistic view. Biochimica et Biophysica Acta (BBA) - Gene Regulatory Mechanisms, 1829(8), 799–809. doi:10.1016/j.bbagrm.2013.01.005

Marintchev, A., Edmonds, K. a., Marintcheva, B., Hendrickson, E., Oberer, M., Suzuki, C., … Wagner, G. (2009). Topology and Regulation of the Human eIF4A/4G/4H Helicase Complex in Translation Initiation. Cell, 136(3), 447–460. doi:10.1016/j.cell.2009.01.014 Marintchev, A., & Wagner, G. (2004). Translation initiation: structures, mechanisms and evolution. Quarterly Reviews of Biophysics, 37(3-4), 197–284. doi:10.1017/S0033583505004026 Martin, F., Barends, S., Jaeger, S., Schaeffer, L., Prongidi-Fix, L., & Eriani, G. (2011). Cap-assisted internal initiation of translation of histone h4. Molecular Cell, 41(2), 197–209. doi:10.1016/j.molcel.2010.12.019 Mitchell, S. F., & Lorsch, J. R. (2008). Should I stay or should I go? Eukaryotic translation initiation factors 1 and 1A control start codon recognition. The Journal of Biological Chemistry, 283(41), 27345–27349. doi:10.1074/jbc.R800031200 Moerke, N. J., Aktas, H., Chen, H., Cantel, S., Reibarkh, M. Y., Fahmy, A., … Wagner, G. (2007). Small-Molecule Inhibition of the Interaction between the Translation Initiation Factors eIF4E and eIF4G. Cell, 128(2), 257–267. doi:10.1016/j.cell.2006.11.046 Nanda, J. S., Cheung, Y. N., Takacs, J. E., Martin-Marcos, P., Saini, A. K., Hinnebusch, A. G., & Lorsch, J. R. (2009). eIF1 Controls Multiple Steps in Start Codon Recognition during Eukaryotic Translation Initiation. Journal of Molecular Biology, 394(2), 268–285. doi:10.1016/j.jmb.2009.09.017 Oberer, M., Marintchev, A., & Wagner, G. (2005). Structural basis for the enhancement of eIF4A helicase activity by eIF4G. Genes and Development, 19(18), 2212–2223. doi:10.1101/gad.1335305 Özeş, A. R., Feoktistova, K., Avanzino, B. C., & Fraser, C. S. (2011). Duplex unwinding and ATPase activities of the DEAD-box helicase eIF4A are coupled by eIF4G and eIF4B. Journal of Molecular Biology, 412(4), 674–687. doi:10.1016/j.jmb.2011.08.004 Park, E., Walker, S. E., Lee, J. M., Rothenburg, S., Lorsch, J. R., & Hinnebusch, A. G. (2010). Multiple elements in the eIF4G1 N-terminus promote assembly of eIF4G1 PABP mRNPs in vivo. The EMBO Journal, 30(2), 302–316. doi:10.1038/emboj.2010.312 Parsyan, A., Svitkin, Y., Shahbazian, D., Gkogkas, C., Lasko, P., Merrick, W. C., & Sonenberg, N. (2011). mRNA helicases: the tacticians of translational control. Nature Reviews. Molecular Cell Biology, 12(4), 235–245. doi:10.1038/nrm3083 Passmore, L. a., Schmeing, T. M., Maag, D., Applefield, D. J., Acker, M. G., Algire, M. A., … Ramakrishnan, V. (2007). The Eukaryotic Translation Initiation Factors eIF1 and eIF1A Induce an Open Conformation of the 40S Ribosome. Molecular Cell, 26(1), 41–50. doi:10.1016/j.molcel.2007.03.018 Pestova, T. V, Borukhov, S. I., & Hellen, C. U. (1998). Eukaryotic ribosomes require

initiation factors 1 and 1A to locate initiation codons. Nature, 394(6696), 854–859. Pestova, T. V, & Kolupaeva, V. G. (2002). The roles of individual eukaryotic translation initiation factors in ribosomal scanning and initiation codon selection. Genes & Development, 16(22), 2906–2922. doi:10.1101/gad.1020902 Pestova, T. V., & Kolupaeva, V. G. (2002). The roles of individual eukaryotic translation initiation factors in ribosomal scanning and initiation codon selection. Genes and Development, 16, 2906–2922. doi:10.1101/gad.1020902 Pisarev, A. V, Kolupaeva, V. G., Yusupov, M. M., Hellen, C. U. T., & Pestova, T. V. (2008). Ribosomal position and contacts of mRNA in eukaryotic translation initiation complexes. The EMBO Journal, 27(11), 1609–1621. doi:10.1038/emboj.2008.90 Pisarev, A. V, Pisarev, A. V, Kolupaeva, V. G., Kolupaeva, V. G., Pisareva, V. P., Pisareva, V. P., … Merrick, W. C. (2006). − 3 + 4. Genes & Development, 624–636. doi:10.1101/gad.1397906.60S Pisareva, V. P., Pisarev, A. V., Komar, A. a., Hellen, C. U. T., & Pestova, T. V. (2008). Translation Initiation on Mammalian mRNAs with Structured 5???UTRs Requires DExH-Box Protein DHX29. Cell, 135(7), 1237–1250. doi:10.1016/j.cell.2008.10.037 Rabl, J., Leibundgut, M., Ataide, S. F., Haag, A., & Ban, N. (2011). Crystal Structure of the Eukaryotic 40 S Ribosomal Subunit in Complex with Initiation Factor 1, 331(February), 730–737. Rieder, C. L. (2011). Mitosis in vertebrates : the G2 / M and M / A transitions and their associated checkpoints, 291–306. doi:10.1007/s10577-010-9178-z Rogers, G. W., Richter, N. J., Lima, W. F., & Merrick, W. C. (2001). Modulation of the Helicase Activity of eIF4A by eIF4B, eIF4H, and eIF4F. Journal of Biological Chemistry, 276(33), 30914–30922. doi:10.1074/jbc.M100157200 Rozovsky, N., Butterworth, A. C., & Moore, M. J. (2008). Interactions between eIF4AI and its accessory factors eIF4B and eIF4H. RNA (New York, N.Y.), 14(10), 2136– 2148. doi:10.1261/rna.1049608 Saini, a. K., Nanda, J. S., Martin-Marcos, P., Dong, J., Zhang, F., Bhardwaj, M., … Hinnebusch, a. G. (2014). Eukaryotic translation initiation factor eIF5 promotes the accuracy of start codon recognition by regulating Pi release and conformational transitions of the preinitiation complex. Nucleic Acids Research, 42(15), 1–18. doi:10.1093/nar/gku653 Selection, A. U. G., Reibarkh, M., Yamamoto, Y., Singh, C. R., Rio, F., Fahmy, A., … Asano, K. (2008). Eukaryotic Initiation Factor ( eIF ) 1 Carries Two Distinct eIF5- binding Faces Important for Multifactor Assembly and, 283(2), 1094–1103. doi:10.1074/jbc.M708155200 Shabalina, S. a., Ogurtsov, A. a., Rogozin, I. B., Koonin, E. V., & Lipman, D. J. (2004). Comparative analysis of orthologous eukaryotic mRNAs: Potential hidden

functional signals. Nucleic Acids Research, 32(5), 1774–1782. doi:10.1093/nar/gkh313 Singh, C. R., He, H., Ii, M., Yamamoto, Y., & Asano, K. (2004). Efficient incorporation of eukaryotic initiation factor 1 into the multifactor complex is critical for formation of functional ribosomal preinitiation complexes in vivo. Journal of Biological Chemistry, 279(30), 31910–31920. doi:10.1074/jbc.M313940200 Singh, C. R., Watanabe, R., Chowdhury, W., Hiraishi, H., Murai, M. J., Yamamoto, Y., … Asano, K. (2012). Sequential Eukaryotic Translation Initiation Factor 5 ( eIF5 ) Binding to the Charged Disordered Segments of eIF4G and eIF2 ␤ Stabilizes the 48S Preinitiation Complex and Promotes Its Shift to the Initiation, 32(19), 3978– 3989. doi:10.1128/MCB.00376-12 Sinvani, H., Haimov, O., Svitkin, Y., Sonenberg, N., Tamarkin-Ben-Harush, A., Viollet, B., & Dikstein, R. (2015). Translational Tolerance of Mitochondrial Genes to Metabolic Energy Stress Involves TISU and eIF1-eIF4GI Cooperation in Start Codon Selection. Cell Metabolism, 21(3), 479–492. doi:10.1016/j.cmet.2015.02.010 Sonenberg, N., & Hinnebusch, A. G. (2009). Regulation of Translation Initiation in Eukaryotes: Mechanisms and Biological Targets. Cell, 136(4), 731–745. doi:10.1016/j.cell.2009.01.042 Svitkin, Y. V, Pause, a, Haghighat, a, Pyronnet, S., Witherell, G., Belsham, G. J., & Sonenberg, N. (2001). The requirement for eukaryotic initiation factor 4A (elF4A) in translation is in direct proportion to the degree of mRNA 5’ secondary structure. RNA (New York, N.Y.), 7, 382–394. doi:10.1017/S135583820100108X Tamarkin-Ben-Harush, A., Schechtman, E., & Dikstein, R. (2014). Co-occurrence of transcription and translation gene regulatory features underlies coordinated mRNA and protein synthesis. BMC Genomics, 15(1), 688. doi:10.1186/1471-2164-15-688 Valásek, L., Nielsen, K. H., Zhang, F., Fekete, C. a, & Hinnebusch, A. G. (2004). Interactions of eukaryotic translation initiation factor 3 (eIF3) subunit NIP1/c with eIF1 and eIF5 promote preinitiation complex assembly and regulate start codon selection. Molecular and Cellular Biology, 24(21), 9437–9455. doi:10.1128/MCB.24.21.9437-9455.2004 Villa, N., Do, A., Hershey, J. W. B., & Fraser, C. S. (2013). Human eukaryotic initiation factor 4G (eIF4G) protein binds to eIF3c, -d, and -e to promote mRNA recruitment to the ribosome. Journal of Biological Chemistry, 288(46), 32932–32940. doi:10.1074/jbc.M113.517011 Wang, X.-Q., & Rothnagel, J. a. (2004). 5’-untranslated regions with multiple upstream AUG codons can support low-level translation via leaky scanning and reinitiation. Nucleic Acids Research, 32(4), 1382–1391. doi:10.1093/nar/gkh305 Weisser, M., Voigts-hoffmann, F., Rabl, J., Leibundgut, M., & Ban, N. (2013). BRIE F COMMUNICATIONS The crystal structure of the in complex with eIF1 and eIF1A. Nature Publishing Group, 20(8), 1015–1017. doi:10.1038/nsmb.2622

Wells, S. E., Hillner, P. E., Vale, R. D., & Sachs, A. B. (1998). Circularization of mRNA by Eukaryotic Translation Initiation Factors. Molecular Cell, 2(1), 135–140. doi:10.1016/S1097-2765(00)80122-7 Wethmar, K., Barbosa-Silva, A., Andrade-Navarro, M. a., & Leutz, A. (2014). UORFdb - A comprehensive literature database on eukaryotic uORF biology. Nucleic Acids Research, 42(October 2013), 60–67. doi:10.1093/nar/gkt952 Yamamoto, Y., Singh, C. R., Marintchev, A., Hall, N. S., Hannig, E. M., Wagner, G., & Asano, K. (2005). The eukaryotic initiation factor (eIF) 5 HEAT domain mediates multifactor assembly and scanning with distinct interfaces to eIF1, eIF2, eIF3, and eIF4G. Proceedings of the National Academy of Sciences of the United States of America, 102, 16164–16169. doi:10.1073/pnas.0507960102 Yanagiya, A., Svitkin, Y. V, Shibata, S., Mikami, S., Imataka, H., & Sonenberg, N. (2009). Requirement of RNA binding of mammalian eukaryotic translation initiation factor 4GI (eIF4GI) for efficient interaction of eIF4E with the mRNA cap. Molecular and Cellular Biology, 29(6), 1661–1669. doi:10.1128/MCB.01187-08 Yu, Y., Marintchev, A., Kolupaeva, V. G., Unbehaun, A., Veryasova, T., Lai, S. C., … Pestova, T. V. (2009). Position of eukaryotic translation initiation factor eIF1A on the 40S ribosomal subunit mapped by directed hydroxyl radical probing. Nucleic Acids Research, 37(15), 5167–5182. doi:10.1093/nar/gkp519

Declaration

I declare that this thesis summarizes my independent research.

Collaborations:

The elutriation separation of the U2OS cells according to their cell cycle phase (Figure 10) was performed by Dr. Noam Diamant .

The eIF4A overexpression (Figure 14) experiments were performed in collaboration with Dr. Hadar Sinvani.

The cloning of the split-renilla constructs and further experiments (Figure 25) were conducted in collaboration with Dr. Anat Bahat, Anna Uzonyi and Adi Jacob.

List of publications

Haimov O*, Sinvani H*, Dikstein R.Cap-dependent, scanning-free translation initiation mechanisms. Review Biochim Biophys Acta. 2015 Nov;1849(11):1313-8.

Sinvani H*, Haimov O*, Svitkin Y, Sonenberg N, Tamarkin-Ben-Harush A, Viollet B, Dikstein R. Translational tolerance of mitochondrial genes to metabolic energy stress involves TISU and eIF1-eIF4GI cooperation in start codon selection. Cell Metabolism. 2015 Mar 3;21(3):479-92

Elfakess R*, Sinvani H*, Haimov O, Svitkin Y, Sonenberg N, Dikstein R. Unique translation initiation of mRNAs containing TISU. Nucleic Acids Res. 2011 Sep 1;39(17):7598-609

* Equal contribution

תקציר TISU הינו אלמנט )רצף נוקלאוטידים( ייחודי המבקר את שלבי התחלת השעתוק והתרגום. בתצורתו המשועתקת האלמנט מכיל במרכזו AUG ומביא לידי אתחול תרגום באופן יעיל ביותר מרנ"א שליח בעל UTR’5 קצר במיוחד במנגנון תלוי cap אך ללא scanning. המטרה העיקרית של המחקר הנ"ל הייתה לבחון את העמדות ברצף האלמנט שתורמות ליעילותו באתחול תרגום ואת פקטורי האיניציאציה המעורבים במכניזם הייחודי אותו הוא מתווך. אנליזה נרחבת שכללה מוטגנציה של כל אחד מהנוקלאוטידים החופפים את ה- TISU -AUG הראתה שמדובר באלמנט בקרה עמיד והדגישה את תרומתן של כל העמדות לחוזקו. עם זאת, נראה שעמדות -3, +4, +5, +6 הן קריטיות לדיוק בתרגום. כמו כן, בחנו את ההשפעה של מספר פקטורי איניציאציה על תרגום של קבוצות רנ"א שליח שונות הנבדלות מבחינת הרצפים החופפים את ה- AUG )קונטקסט ה-AUG( ואורך ה-UTR’5 באמצעות הגברה או הורדה של ביטוי הפקטורים בתאים. אנו מראים שההשפעות של eIF4GI ,eIF1 ,eIF1A ו- eIF4E על תרגום אינן כלליות אלא תלויות במידה רבה במאפיינים של הרנ"א שליח. לדוגמא: עלייה ברמות eIF1A הביאה לעליה בתרגום מ- AUG בקונטקסט קנוני הנמצא קרוב ל cap , לעומת זאת הגברת הביטוי של eIF1 הביאה לאפקט אנטגוניסטי. הגברת הביטוי של eIF1A או eIF1 לא השפיעה באופן משמעותי על תרגום של רנ"א שליח המכיל TISU. מעניין לציין שהורדת הביטוי של eIF1 והפרלוג שלו eIF4GI ,eIF1A ,eIF1B או eIF4E גרמה לירידה דרמטית בתרגום המתווך ע"י TISU, לעומת זאת תרגום מ- AUG בקונטקס קנוני הנמצא קרוב ל cap כמעט שאינו הושפע. הדבר מרמז על כך שפעילותו של TISU תלויה רבות בפקטורים הללו. בנוסף לכך, בעזרת סדרת הניסויים שערכנו מצאנו לראשונה ש-eIF4GI ו-eIF4E מבקרים את תהליך ה-scanning . אנו מראים שeIF1 נקשר באופן ספציפי לאזור מרכזי ושמור ב- eIF4GI. בנוסף לכך הממצאים שלנו מצביעים על הימצאותו של אתר קישור חדש של eIF4E בחלק המרכזי והשמור של eIF4GI. הקישור של eIF1 אינו יכול להתקיים בו זמנית לקישור של eIF4E . מוטנט של eIF1 הפגוע ביכולת להיקשר ל-eIF4GI אינו מסוגל לקדם scanning אך תומך בתרגום המתווך ע"י TISU, על כן נראה שהקישור בין eIF1-eIF4GI נחוץ ל- scanning אך לא ל-TISU. הממצאים שלנו מצביעים על כך שהפעילות של eIF4E-eIF4GI הינה אנטגוניסטית לפעילות של eIF1-eIF4GI בקידום scanning. על כן אנו מציעים מודל לפיו הכניסה לשלב ה-scanning דורשת ניתוק של האינטרקציה בין eIF4GI לבין eIF4F וקישור של eIF4GI לeIF1- . לפיכך העבודה שלנו מרחיבה את הבנתנו במנגנונים השונים לאתחול תרגום במנגנון תלוי scanning או במנגנון של TISU שאינו תלוי scanning.