Molecular mechanism of the ribosome and functions of translational regulation

National Institute Of Diabetes And Digestive And Kidney Diseases

Investigators

Linked publications, trials & patents

Abstract

Translation begins with the loading (initiation) of ribosomal subunits onto a RNA transcript, continues with the elongation of the peptide by the ribosome, and terminates with release of the completed protein and removal (“recycling”) of the ribosomal subunits from the transcript. We have a longstanding interest in the mechanism of translation termination and ribosome recycling. We also are interested in understanding how ribosomes that stall during the elongation phase of translation are detected by quality control machinery in the cell and what signaling outcomes result from this detection. Our innovative approach utilizes advanced forms of ribosome profiling (high-throughput sequencing of ribosome footprints), including multiple conformations of stalled 80S ribosomes, collided ribosomes (disomes), and scanning 40S subunits. Our recent work reveals new forms of translational regulation and shows how signaling downstream of stalled elongation and premature termination have important functional roles in human cells and budding yeast. The termination step of translation is critical for releasing the completed protein from the ribosome so that it can carry out functions in the cell. Our recent work has focused on the related problem of premature translation termination. In premature termination, the translating ribosome encounters a stop codon that lies upstream of the canonical stop codon. Ribosomes that undergo termination at these premature stop codons can be sensed by a pathway called nonsense-mediated decay (NMD) that leads to degradation of the transcript. NMD is important for eliminating aberrant transcripts that do not encode full-length proteins. It also plays a key role in human health since about 11% of inherited genetic disease is caused by a premature stop codon. Loss of NMD in budding yeast was found to stabilize >500 transcripts but the exact premature termination event underlying most cases was not readily apparent. As ongoing drug development efforts are aimed at inhibiting NMD, this question is critical for understanding how such therapies would affect normal gene expression. Therefore, our aim was to identify cryptic (apart from the main ORF) translation events that trigger NMD and broadly describe the biological pathways that are regulated by NMD. We hypothesized that cryptic translation events could lead to premature termination so we developed a 40S ribosome profiling technique that allows us to identify stop codons that are actively used for premature translation termination. 40S ribosome profiling does not capture footprints from elongating ribosomes and therefore offers a sensitive method to identify start and stop codons of cryptic ORFs not visible with 80S profiling. To identify translation termination events that could cause NMD, we used cells lacking NMD (upf1 KO background), where transcripts harboring cryptic translation would accumulate. One class of cryptic translation events that we found was translation of out-of-frame internal ORFs (iORFs) within main ORFs, such as on the gene TCA17. These events likely occur due to “leaky scanning” when the 40S ribosomal subunit skips the canonical start codon. Another class included 5’-extended (non-canonical) transcript isoforms where ribosomes only translate upstream ORFs (uORFs) in the 5’-extended region. These transcripts appeared to be regulatory in nature and have been previously called long undecoded transcript isoforms (LUTIs) that, when transcribed, repress a downstream (canonical) promoter. Our work establishes many new LUTIs that were previously not visible due to NMD. In addition, we identified cases where a long 3’UTR downstream of the canonical stop codon likely results in targeting of the transcript to NMD. Overall, our combined approach using computational analysis and ribosome profiling allowed us to account for the vast majority of NMD cases in the cell. Of particular interest, we found at least 10 NMD-sensitive LUTIs that are involved in nitrogen metabolism, including one associated with the gene DAL5. The DAL5 LUTI mechanism appears to be conserved across many yeast species, including the distantly related S. pombe. We therefore tested its role in nitrogen regulation by terminating LUTI transcription and measuring the de-repression of the canonical transcript (which encodes a nitrogen permease) after change to low nitrogen media. We found the kinetics of this process (and the inverse repression process) was modulated by the LUTI. These data show important roles for NMD-sensitive LUTIs and therefore expand our understanding of what biological functions are linked to NMD. Our work has also examined how translation changes during viral infection and promotes the innate immune response. In particular, we focused on RNase L, an endonuclease that is activated when viral RNA is detected in human cells. RNase L widely cleaves single-stranded RNAs at U bases, leading to the loss of most transcripts in the cell. Ultimately, the loss of RNA leads to cell death and benefits the host by eliminating the infected cell. However, it was unclear how widespread RNA degradation activates apoptotic pathways. Earlier work from my lab offered a model where ribosomes translating fragmented transcripts stalled and then activated a cell-death pathway known as the ribotoxic stress response (RSR). We showed this pathway to be activated, based on phosophorylation of a sensor of ribosome stalling (and resultant collisions between ribosomes), ZAK-alpha. While our work showed RNase L causes apoptosis by activating ZAKα, it remained unclear where exactly ribosomes were stalling. In our recent work, we aimed to directly find ribosomes that stalled due to RNase L activation and could therefore cause ZAK-alpha activation. Consistent with our prior models, we showed transcript fragments were detectable in human lung cells via nanopore sequencing and that they sedimented with ribosomes in a sucrose density gradient. We also found that these translated fragments were actively degraded by the exonuclease XRN1. To look for cases where ribosomes stalled on the 3’ ends of transcript fragments, we applied an advanced form of ribosome profiling that I previously showed in yeast can reveal these ribosomes because they hang over the 3’ end of the RNA and therefore protect a shorter footprint (16 vs 28 nt). The data clearly revealed these ribosomes accumulate at preferred UU motifs. In addition, these stalled ribosomes are expected to be sensed and removed by PELO, a protein that is known to rescue stalled ribosomes. This activity could mitigate activation of the RSR since ribosomes would be removed before they could trigger collisions and be sensed by ZAK-alpha. In cells where PELO was knocked down or knocked out, we found amplification of both ribosome stalling at cleavage sites and enhanced activation of the RSR, in support of this hypothesis. This work therefore establishes the mechanism of how ribosome stalling leads to the RSR in cells where RNase L is active and suggests that the ribosome rescue factor PELO tunes the response. Changes in translation are also important during viral infection since the virus and host must compete for ribosomes. We are therefore examining how infection of human lung cells with respiratory syncytial virus (RSV) changes translation. RSV is a single-stranded, negative-sense RNA virus that expresses 10 individual transcripts that are 5’-capped and polyadenylated, much like host transcripts. Unlike other viruses, RSV is not thought to shut down translation of host transcripts nor employ non-canonical means (such as internal ribosome entry sites) to recruit ribosomes to its own transcripts. Therefore, RSV and the host directly compete for ribosomes and associated translational machinery. Our aim was to characterize this competition and determine whether and how it affected which host transcripts were translated. Our work in collaboration with the lab of Mark Bayfield used sucrose gradient sedimentation to directly measure how many ribosomes were associated with RNAs and used RNA-Seq across the gradient to determine whether specific transcripts were strongly or weakly associated with ribosomes. Surprisingly, we found that the overall amount of polysomes in the cell increased during infection, indicating that ribosomes were being transferred from a pool of non-translating or poorly-translating ribosomes to heavily translated polysomes. These heavier polysomes could be forming due to enhanced recruitment of ribosomes to host transcripts (pre-existing or newly transcribed) or slower-translating ribosomes. It is also conceivable that the extra RNA transcripts from the virus could be contributing to the higher number of polysomes. Intriguingly, we found that the heaviest polysomes during infection were enriched in host RNAs that were longer and rich in AU nucleotides, as opposed to shorter transcripts rich in GC-nucleotides seen in the absence of infection. As viral transcripts are also rich in AU-rich nucleotides, this finding suggests the virus is changing the preference of the translational machinery in some way, such as by altering the expression of RNA-binding proteins that recognize AU-rich sequences. Our most recent data show that stress pathways that induce an overall reduction in translation affect the virus and the host equally, highlighting the importance of avoiding these pathways since they could reduce the ability of the virus to replicate. Consistent with this, despite the added translational load created by RSV infection, our preliminary data show minimal activation of ribosome collision-sensing stress pathways. We also found that viral transcripts recruit ribosomes with greatly varying efficiencies. This variation may reflect an optimization to avoid ribosome collisions that would turn on stress signaling pathways and reduce overall translation. We have also embarked on new projects to investigate the role of translation in insulin signaling. Prior work has shown roles for proteins that can sense ribosomes that stall (and collide with each other), including ZAK-alpha, GCN2, and PELO in diabetes in mice and humans. In particular, GCN2 and PELO have been proposed to be protective against disease. In contrast, knockout of ZAK-alpha in mice fed a high-fat diet were protected from multiple symptoms of metabolic disease, including loss of blood glucose regulation. However, the underlying mechanism of why ribosomes stall in these mice and activate ZAK-alpha is not clear. In addition, the ribosome inhibitor anisomycin, which causes ribosome collisions, causes phosphorylation of IRS1, the downstream target of the insulin receptor. We have therefore carried out ribosome profiling of HepG2 liver cells that were treated with insulin to establish a framework for understanding the role of translation in insulin signaling. Our data revealed that changes in how efficiently the transcriptome is translated in response to insulin depends on the kinase mTOR, consistent with models of insulin function. Intriguingly, a number of the transcripts translated with altered efficiency are not established targets of the mTOR pathway. These data therefore suggest additional levels of translational regulation by insulin.