Quantitative principles of cis-translational control by general mRNA sequence features in eukaryotes

Predicting the translation efficiency of messenger RNA in mammalian cells

The degree to which translational control is specified by mRNA sequence is poorly understood in mammalian cells. Here, we constructed and leveraged a compendium of 3,819 ribosomal profiling datasets, distilling them into a transcriptome-wide atlas of translation efficiency (TE) measurements encompassing >140 human and mouse cell types. We subsequently developed RiboNN, a multitask deep convolutional neural network, and classic machine learning models to predict TEs in hundreds of cell types from sequence-encoded mRNA features, achieving state-of-the-art performance (r=0.79 in human and r=0.78 in mouse for mean TE across cell types). While the majority of earlier models solely considered 5' UTR sequence, RiboNN integrates contributions from the full-length mRNA sequence, learning that the 5' UTR, CDS, and 3' UTR respectively possess ~67%, 31%, and 2% per-nucleotide information density in the specification of mammalian TEs. Interpretation of RiboNN revealed that the spatial positioning of low-level di- and tri-nucleotide features (i.e., including codons) largely explain model performance, capturing mechanistic principles such as how ribosomal processivity and tRNA abundance control translational output. RiboNN is predictive of the translational behavior of base-modified therapeutic RNA, and can explain evolutionary selection pressures in human 5' UTRs. Finally, it detects a common language governing mRNA regulatory control and highlights the interconnectedness of mRNA translation, stability, and localization in mammalian organisms.

Discovery of regulatory motifs in 5' untranslated regions using interpretable multi-task learning models

The sequence in the 5' untranslated regions (UTRs) is known to affect mRNA translation rates. However, the underlying regulatory grammar remains elusive. Here, we propose MTtrans, a multi-task translation rate predictor capable of learning common sequence patterns from datasets across various experimental techniques. The core premise is that common motifs are more likely to be genuinely involved in translation control. MTtrans outperforms existing methods in both accuracy and the ability to capture transferable motifs across species, highlighting its strength in identifying evolutionarily conserved sequence motifs. Our independent fluorescence-activated cell sorting coupled with deep sequencing (FACS-seq) experiment validates the impact of most motifs identified by MTtrans. Additionally, we introduce "GRU-rewiring," a technique to interpret the hidden states of the recurrent units. Gated recurrent unit (GRU)-rewiring allows us to identify regulatory element-enriched positions and examine the local effects of 5' UTR mutations. MTtrans is a powerful tool for deciphering the translation regulatory motifs.

Improving deep models of protein-coding potential with a Fourier-transform architecture and machine translation task

Ribosomes are information-processing macromolecular machines that integrate complex sequence patterns in messenger RNA (mRNA) transcripts to synthesize proteins. Studies of the sequence features that distinguish mRNAs from long noncoding RNAs (lncRNAs) may yield insight into the information that directs and regulates translation. Computational methods for calculating protein-coding potential are important for distinguishing mRNAs from lncRNAs during genome annotation, but most machine learning methods for this task rely on previously known rules to define features. Sequence-to-sequence (seq2seq) models, particularly ones using transformer networks, have proven capable of learning complex grammatical relationships between words to perform natural language translation. Seeking to leverage these advancements in the biological domain, we present a seq2seq formulation for predicting protein-coding potential with deep neural networks and demonstrate that simultaneously learning translation from RNA to protein improves classification performance relative to a classification-only training objective. Inspired by classical signal processing methods for gene discovery and Fourier-based image-processing neural networks, we introduce LocalFilterNet (LFNet). LFNet is a network architecture with an inductive bias for modeling the three-nucleotide periodicity apparent in coding sequences. We incorporate LFNet within an encoder-decoder framework to test whether the translation task improves the classification of transcripts and the interpretation of their sequence features. We use the resulting model to compute nucleotide-resolution importance scores, revealing sequence patterns that could assist the cellular machinery in distinguishing mRNAs and lncRNAs. Finally, we develop a novel approach for estimating mutation effects from Integrated Gradients, a backpropagation-based feature attribution, and characterize the difficulty of efficient approximations in this setting.

Unraveling the influences of sequence and position on yeast uORF activity using massively parallel reporter systems and machine learning

Upstream open-reading frames (uORFs) are potent cis-acting regulators of mRNA translation and nonsense-mediated decay (NMD). While both AUG- and non-AUG initiated uORFs are ubiquitous in ribosome profiling studies, few uORFs have been experimentally tested. Consequently, the relative influences of sequence, structural, and positional features on uORF activity have not been determined. We quantified thousands of yeast uORFs using massively parallel reporter assays in wildtype and ∆upf1 yeast. While nearly all AUG uORFs were robust repressors, most non-AUG uORFs had relatively weak impacts on expression. Machine learning regression modeling revealed that both uORF sequences and locations within transcript leaders predict their effect on gene expression. Indeed, alternative transcription start sites highly influenced uORF activity. These results define the scope of natural uORF activity, identify features associated with translational repression and NMD, and suggest that the locations of uORFs in transcript leaders are nearly as predictive as uORF sequences.

Improving deep models of protein-coding potential with a Fourier-transform architecture and machine translation task

Ribosomes are information-processing macromolecular machines that integrate complex sequence patterns in messenger RNA (mRNA) transcripts to synthesize proteins. Studies of the sequence features that distinguish mRNAs from long noncoding RNAs (lncRNAs) may yield insight into the information that directs and regulates translation. Computational methods for calculating protein-coding potential are important for distinguishing mRNAs from lncRNAs during genome annotation, but most machine learning methods for this task rely on previously known rules to define features. Sequence-to-sequence (seq2seq) models, particularly ones using transformer networks, have proven capable of learning complex grammatical relationships between words to perform natural language translation. Seeking to leverage these advancements in the biological domain, we present a seq2seq formulation for predicting protein-coding potential with deep neural networks and demonstrate that simultaneously learning translation from RNA to protein improves classification performance relative to a classification-only training objective. Inspired by classical signal processing methods for gene discovery and Fourier-based image-processing neural networks, we introduce LocalFilterNet (LFNet). LFNet is a network architecture with an inductive bias for modeling the three-nucleotide periodicity apparent in coding sequences. We incorporate LFNet within an encoder-decoder framework to test whether the translation task improves the classification of transcripts and the interpretation of their sequence features. We use the resulting model to compute nucleotide-resolution importance scores, revealing sequence patterns that could assist the cellular machinery in distinguishing mRNAs and lncRNAs. Finally, we develop a novel approach for estimating mutation effects from Integrated Gradients, a backpropagation-based feature attribution, and characterize the difficulty of efficient approximations in this setting.

Fine mapping of meiotic crossovers in Brassica oleracea reveals patterns and variations depending on direction and combination of crosses

Meiotic recombination is crucial for assuring proper segregation of parental chromosomes and generation of novel allelic combinations. As this process is tightly regulated, identifying factors influencing rate, and distribution of meiotic crossovers (COs) is of major importance, notably for plant breeding programs. However, high-resolution recombination maps are sparse in most crops including the Brassica genus and knowledge about intraspecific variation and sex differences is lacking. Here, we report fine-scale resolution recombination landscapes for 10 female and 10 male crosses in Brassica oleracea, by analyzing progenies of five large four-way-cross populations from two reciprocally crossed F1s per population. Parents are highly diverse inbred lines representing major crops, including broccoli, cauliflower, cabbage, kohlrabi, and kale. We produced approximately 4.56T Illumina data from 1248 progenies and identified 15 353 CO across the 10 reciprocal crosses, 51.13% of which being mapped to <10 kb. We revealed fairly similar Mb-scale recombination landscapes among all cross combinations and between the sexes, and provided evidence that these landscapes are largely independent of sequence divergence. We evidenced strong influence of gene density and large structural variations on CO formation in B. oleracea. Moreover, we found extensive variations in CO number depending on the direction and combination of the initial parents crossed with, for the first time, a striking interdependency between these factors. These data improve our current knowledge on meiotic recombination and are important for Brassica breeders.

Jasmonic acid enhances osmotic stress responses by MYC2-mediated inhibition of protein phosphatase 2C1 and response regulators 26 transcription factor in tomato

The jasmonic acid (JA) signaling pathway is involved in the plant response to drought stress. JA and other hormones synergistically regulate the drought response in plants. However, the molecular mechanism underlying this synergism remains poorly defined. In the present study, transcriptome analyses of guard cells and quantitative PCR experiments revealed that MYC2 negatively regulated the negative regulator of ABA signaling, SlPP2C1, and the type-B response regulator in the cytokinin pathway, SlRR26, and this negative regulation was direct. SlRR26 overexpression reduced drought tolerance in transgenic tomatoes, whereas slrr26^cr lines were more tolerant to drought. SlRR26 negatively modulated reactive oxygen species levels in stomata and stomatal closure through RobhB. Moreover, SlRR26 overexpression counteracted JA-mediated stomatal closure, suggesting that SlRR26 played a negative role in the JA-mediated drought response. These findings suggest that MYC2 plays a key role in JA-regulated stomatal closure under drought stress by inhibiting SlPP2C1 and SlRR26.

Identification of multiple novel genetic mechanisms that regulate chilling tolerance in Arabidopsis

INTRODUCTION

Cold stress adversely affects the growth and development of plants and limits the geographical distribution of many plant species. Accumulation of spontaneous mutations shapes the adaptation of plant species to diverse climatic conditions.

METHODS

The genome-wide association study of the phenotypic variation gathered by a newly designed phenomic platform with the over six millions single nucleotide polymorphic (SNP) loci distributed across the genomes of 417 Arabidopsis natural variants collected from various geographical regions revealed 33 candidate cold responsive genes.

RESULTS

Investigation of at least two independent insertion mutants for 29 genes identified 16 chilling tolerance genes governing diverse genetic mechanisms. Five of these genes encode novel leucine-rich repeat domain-containing proteins including three nucleotide-binding site-leucine-rich repeat (NBS-LRR) proteins. Among the 16 identified chilling tolerance genes, ADS2 and ACD6 are the only two chilling tolerance genes identified earlier.

DISCUSSION

The 12.5% overlap between the genes identified in this genome-wide association study (GWAS) of natural variants with those discovered previously through forward and reverse genetic approaches suggests that chilling tolerance is a complex physiological process governed by a large number of genetic mechanisms.

The interplay between translational efficiency, poly(A) tails, microRNAs, and neuronal activation

Neurons provide a rich setting for studying post-transcriptional control. Here, we investigate the landscape of translational control in neurons and search for mRNA features that explain differences in translational efficiency (TE), considering the interplay between TE, mRNA poly(A)-tail lengths, microRNAs, and neuronal activation. In neurons and brain tissues, TE correlates with tail length, and a few dozen mRNAs appear to undergo cytoplasmic polyadenylation upon light or chemical stimulation. However, the correlation between TE and tail length is modest, explaining <5% of TE variance, and even this modest relationship diminishes when accounting for other mRNA features. Thus, tail length appears to affect TE only minimally. Accordingly, miRNAs, which accelerate deadenylation of their mRNA targets, primarily influence target mRNA levels, with no detectable effect on either steady-state tail lengths or TE. Larger correlates with TE include codon composition and predicted mRNA folding energy. When combined in a model, the identified correlates explain 38%-45% of TE variance. These results provide a framework for considering the relative impact of factors that contribute to translational control in neurons. They indicate that when examined in bulk, translational control in neurons largely resembles that of other types of post-embryonic cells. Thus, detection of more specialized control might require analyses that can distinguish translation occurring in neuronal processes from that occurring in cell bodies.

Direct analysis of ribosome targeting illuminates thousand-fold regulation of translation initiation

Translational control shapes the proteome in normal and pathophysiological conditions. Current high-throughput approaches reveal large differences in mRNA-specific translation activity but cannot identify the causative mRNA features. We developed direct analysis of ribosome targeting (DART) and used it to dissect regulatory elements within 5' untranslated regions that confer 1,000-fold differences in ribosome recruitment in biochemically accessible cell lysates. Using DART, we determined a functional role for most alternative 5' UTR isoforms expressed in yeast, revealed a general mode of increased translation via direct binding to a core translation factor, and identified numerous translational control elements including C-rich silencers that are sufficient to repress translation both in vitro and in vivo. DART enables systematic assessment of the translational regulatory potential of 5' UTR variants, whether native or disease-associated, and will facilitate engineering of mRNAs for optimized protein production in various systems.

Modeling the ribosomal small subunit dynamic in Saccharomyces cerevisiae based on TCP-seq data

Translation Complex Profile Sequencing (TCP-seq), a protocol that was developed and implemented on Saccharomyces cerevisiae, provides the footprints of the small subunit (SSU) of the ribosome (with additional factors) across the entire transcriptome of the analyzed organism. In this study, based on the TCP-seq data, we developed for the first-time a predictive model of the SSU density and analyzed the effect of transcript features on the dynamics of the SSU scan in the 5′UTR. Among others, our model is based on novel tools for detecting complex statistical relations tailored to TCP-seq. We quantitatively estimated the effect of several important features, including the context of the upstream AUG, the upstream ORF length and the mRNA folding strength. Specifically, we suggest that around 50% of the variance related to the read counts (RC) distribution near a start codon can be attributed to the AUG context score. We provide the first large scale direct quantitative evidence that shows that indeed AUG context affects the small sub-unit movement. In addition, we suggest that strong folding may cause the detachment of the SSU from the mRNA. We also identified a number of novel sequence motifs that can affect the SSU scan; some of these motifs affect transcription factors and RNA binding proteins. The results presented in this study provide a better understanding of the biophysical aspects related to the SSU scan along the 5′UTR and of translation initiation in S. cerevisiae, a fundamental step toward a comprehensive modeling of initiation.

Quantitative model suggests both intrinsic and contextual features contribute to the transcript coding ability determination in cells

Gene transcription and protein translation are two key steps of the 'central dogma.' It is still a major challenge to quantitatively deconvolute factors contributing to the coding ability of transcripts in mammals. Here, we propose ribosome calculator (RiboCalc) for quantitatively modeling the coding ability of RNAs in human genome. In addition to effectively predicting the experimentally confirmed coding abundance via sequence and transcription features with high accuracy, RiboCalc provides interpretable parameters with biological information. Large-scale analysis further revealed a number of transcripts with a variety of coding ability for distinct types of cells (i.e. context-dependent coding transcripts), suggesting that, contrary to conventional wisdom, a transcript's coding ability should be modeled as a continuous spectrum with a context-dependent nature.

Widespread translational control regulates retinal development in mouse

Retinal development is tightly regulated to ensure the generation of appropriate cell types and the assembly of functional neuronal circuitry. Despite remarkable advances have been made in understanding regulation of gene expression during retinal development, how translational regulation guides retinogenesis is less understood. Here, we conduct a comprehensive translatome and transcriptome survey to the mouse retinogenesis from the embryonic to the adult stages. We discover thousands of genes that have dynamic changes at the translational level and pervasive translational regulation in a developmental stage-specific manner with specific biological functions. We further identify genes whose translational efficiencies are frequently controlled by changing usage in upstream open reading frame during retinal development. These genes are enriched for biological functions highly important to neurons, such as neuron projection organization and microtubule-based protein transport. Surprisingly, we discover hundreds of previously uncharacterized micropeptides, translated from putative long non-coding RNAs and circular RNAs. We validate their protein products in vitro and in vivo and demonstrate their potentials in regulating retinal development. Together, our study presents a rich and complex landscape of translational regulation and provides novel insights into their roles during retinogenesis.

Codon usage and protein length-dependent feedback from translation elongation regulates translation initiation and elongation speed

Essential cellular functions require efficient production of many large proteins but synthesis of large proteins encounters many obstacles in cells. Translational control is mostly known to be regulated at the initiation step. Whether translation elongation process can feedback to regulate initiation efficiency is unclear. Codon usage bias, a universal feature of all genomes, plays an important role in determining gene expression levels. Here, we discovered that there is a conserved but codon usage-dependent genome-wide negative correlation between protein abundance and CDS length. The codon usage effects on protein expression and ribosome flux on mRNAs are influenced by CDS length; optimal codon usage preferentially promotes production of large proteins. Translation of mRNAs with long CDS and non-optimal codon usage preferentially induces phosphorylation of initiation factor eIF2α, which inhibits translation initiation efficiency. Deletion of the eIF2α kinase CPC-3 (GCN2 homolog) in Neurospora preferentially up-regulates large proteins encoded by non-optimal codons. Surprisingly, CPC-3 also inhibits translation elongation rate in a codon usage and CDS length-dependent manner, resulting in slow elongation rates for long CDS mRNAs. Together, these results revealed a codon usage and CDS length-dependent feedback mechanism from translation elongation to regulate both translation initiation and elongation kinetics.

Post-transcriptional and translational control of the morphology and virulence in human fungal pathogens

Host-pathogen interactions at the molecular level are the key to fungal pathogenesis. Fungal pathogens utilize several mechanisms such as adhesion, invasion, phenotype switching and metabolic adaptations, to survive in the host environment and respond. Post-transcriptional and translational regulations have emerged as key regulatory mechanisms ensuring the virulence and survival of fungal pathogens. Through these regulations, fungal pathogens effectively alter their protein pool, respond to various stress, and undergo morphogenesis, leading to efficient and comprehensive changes in fungal physiology. The regulation of virulence through post-transcriptional and translational regulatory mechanisms is mediated through mRNA elements (cis factors) or effector molecules (trans factors). The untranslated regions upstream and downstream of the mRNA, as well as various RNA-binding proteins involved in translation initiation or circularization of the mRNA, play pivotal roles in the regulation of morphology and virulence by influencing protein synthesis, protein isoforms, and mRNA stability. Therefore, post-transcriptional and translational mechanisms regulating the morphology, virulence and drug-resistance processes in fungal pathogens can be the target for new therapeutics. With improved "omics" technologies, these regulatory mechanisms are increasingly coming to the forefront of basic biology and drug discovery. This review aims to discuss various modes of post-transcriptional and translation regulations, and how these mechanisms exert influence in the virulence and morphogenesis of fungal pathogens.

Learning the Regulatory Code of Gene Expression

Data-driven machine learning is the method of choice for predicting molecular phenotypes from nucleotide sequence, modeling gene expression events including protein-DNA binding, chromatin states as well as mRNA and protein levels. Deep neural networks automatically learn informative sequence representations and interpreting them enables us to improve our understanding of the regulatory code governing gene expression. Here, we review the latest developments that apply shallow or deep learning to quantify molecular phenotypes and decode the cis-regulatory grammar from prokaryotic and eukaryotic sequencing data. Our approach is to build from the ground up, first focusing on the initiating protein-DNA interactions, then specific coding and non-coding regions, and finally on advances that combine multiple parts of the gene and mRNA regulatory structures, achieving unprecedented performance. We thus provide a quantitative view of gene expression regulation from nucleotide sequence, concluding with an information-centric overview of the central dogma of molecular biology.

A translation enhancer element from black beetle virus engages yeast eIF4G1 to drive cap-independent translation initiation

Cap-independent translation initiation plays crucial roles in fine-tuning gene expression under global translation shutdown conditions. Translation of uncapped or de-capped transcripts can be stimulated by Cap-independent translation enhancer (CITE) elements, but the mechanisms of CITE-mediated translation initiation remain understudied. Here, we characterized a short 5ʹ-UTR RNA sequence from black beetle virus, BBV-seq. Mutational analysis indicates that the entire BBV-seq is required for efficient translation initiation, but this sequence does not operate as an IRES-type module. In yeast cell-free translation extracts, BBV-seq promoted efficient initiation on cap-free mRNA using a scanning mechanism. Moreover, BBV-seq can increase translation efficiency resulting from conventional cap-dependent translation initiation. Using genetic approaches, we found that BBV-seq exploits RNA-binding properties of eIF4G1 to promote initiation. Thus, BBV-seq constitutes a previously uncharacterized short, linear CITE that influences eIF4G1 to initiate 5′ end-dependent, cap-independent translation. These findings bring new insights into CITE-mediated translational control of gene expression.

Inferring efficiency of translation initiation and elongation from ribosome profiling

One of the main goals of ribosome profiling is to quantify the rate of protein synthesis at the level of translation. Here, we develop a method for inferring translation elongation kinetics from ribosome profiling data using recent advances in mathematical modelling of mRNA translation. Our method distinguishes between the elongation rate intrinsic to the ribosome’s stepping cycle and the actual elongation rate that takes into account ribosome interference. This distinction allows us to quantify the extent of ribosomal collisions along the transcript and identify individual codons where ribosomal collisions are likely. When examining ribosome profiling in yeast, we observe that translation initiation and elongation are close to their optima and traffic is minimized at the beginning of the transcript to favour ribosome recruitment. However, we find many individual sites of congestion along the mRNAs where the probability of ribosome interference can reach \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$50\%$\end{document}. Our work provides new measures of translation initiation and elongation efficiencies, emphasizing the importance of rating these two stages of translation separately.

So close, no matter how far: multiple paths connecting transcription to mRNA translation in eukaryotes

Transcription of DNA into mRNA and translation of mRNA into proteins are two major processes underlying gene expression. Due to the distinct molecular mechanisms, timings, and locales of action, these processes are mainly considered to be independent. During the last two decades, however, multiple factors and elements were shown to coordinate transcription and translation, suggesting an intricate level of synchronization. This review discusses the molecular mechanisms that impact both processes in eukaryotic cells of different origins. The emerging global picture suggests evolutionarily conserved regulation and coordination between transcription and mRNA translation, indicating the importance of this phenomenon for the fine-tuning of gene expression and the adjustment to constantly changing conditions.

Control of translation by eukaryotic mRNA transcript leaders-Insights from high-throughput assays and computational modeling

Eukaryotic gene expression is tightly regulated during translation of mRNA to protein. Mis-regulation of translation can lead to aberrant proteins which accumulate in cancers and cause neurodegenerative diseases. Foundational studies on model genes established fundamental roles for mRNA 5' transcript leader (TL) sequences in controlling ribosome recruitment, scanning, and initiation. TL cis-regulatory elements and their corresponding trans-acting factors control cap-dependent initiation under unstressed conditions. Under stress, cap-dependent initiation is suppressed, and specific mRNA structures and sequences promote translation of stress-responsive transcripts to remodel the proteome. In this review, we summarize current knowledge of TL functions in translation initiation. We focus on insights from high-throughput analyses of ribosome occupancy, mRNA structure, RNA Binding Protein occupancy, and massively parallel reporter assays. These data-driven approaches, coupled with computational analyses and modeling, have paved the way for a comprehensive understanding of TL functions. Finally, we will discuss areas of future research on the roles of mRNA sequences and structures in translation. This article is categorized under: Translation > Translation Mechanisms RNA Evolution and Genomics > Computational Analyses of RNA RNA Structure and Dynamics > Influence of RNA Structure in Biological Systems.

Quantitative global studies reveal differential translational control by start codon context across the fungal kingdom

Eukaryotic protein synthesis generally initiates at a start codon defined by an AUG and its surrounding Kozak sequence context, but the quantitative importance of this context in different species is unclear. We tested this concept in two pathogenic Cryptococcus yeast species by genome-wide mapping of translation and of mRNA 5' and 3' ends. We observed thousands of AUG-initiated upstream open reading frames (uORFs) that are a major contributor to translation repression. uORF use depends on the Kozak sequence context of its start codon, and uORFs with strong contexts promote nonsense-mediated mRNA decay. Transcript leaders in Cryptococcus and other fungi are substantially longer and more AUG-dense than in Saccharomyces. Numerous Cryptococcus mRNAs encode predicted dual-localized proteins, including many aminoacyl-tRNA synthetases, in which a leaky AUG start codon is followed by a strong Kozak context in-frame AUG, separated by mitochondrial-targeting sequence. Analysis of other fungal species shows that such dual-localization is also predicted to be common in the ascomycete mould, Neurospora crassa. Kozak-controlled regulation is correlated with insertions in translational initiation factors in fidelity-determining regions that contact the initiator tRNA. Thus, start codon context is a signal that quantitatively programs both the expression and the structures of proteins in diverse fungi.

G-quadruplex located in the 5'UTR of the BAG-1 mRNA affects both its cap-dependent and cap-independent translation through global secondary structure maintenance

The anti-apoptotic BAG-1 protein isoforms are known to be overexpressed in colorectal tumors and are considered to be potential therapeutic targets. The isoforms are derived from alternative translation initiations occuring at four in-frame start codons of a single mRNA transcript. Its 5′UTR also contains an internal ribosome entry site (IRES) regulating the cap-independent translation of the transcript. An RNA G-quadruplex (rG4) is located at the 5′end of the BAG-1 5′UTR, upstream of the known cis-regulatory elements. Herein, we observed that the expression of BAG-1 isoforms is post-transcriptionally regulated in colorectal cancer cells and tumors, and that stabilisation of the rG4 by small molecules ligands reduces the expression of endogenous BAG-1 isoforms. We demonstrated a critical role for the rG4 in the control of both cap-dependent and independent translation of the BAG-1 mRNA in colorectal cancer cells. Additionally, we found an upstream ORF that also represses BAG-1 mRNA translation. The structural probing of the complete 5′UTR showed that the rG4 acts as a steric block which controls the initiation of translation at each start codon of the transcript and also maintains the global 5′UTR secondary structure required for IRES-dependent translation.