A key goal of this project was to understand how the levels of transfer RNAs (tRNAs) are regulated to support accurate and efficient protein production in different human cell types. This fundamental question had remained unanswered due to the lack of methods capable of precise, high-throughput measurement of tRNA abundance. tRNAs contain chemical modifications that hinder traditional sequencing approaches by blocking reverse transcription or causing errors, and many tRNAs have over 98% sequence similarity, complicating read alignment and quantification. To overcome these barriers, we developed a novel method called modification-induced misincorporation tRNA sequencing (mim-tRNAseq). This technique allows full-length sequencing of tRNAs while preserving their modification signatures. We also created a user-friendly, open-source computational pipeline for automated tRNA read mapping, quantification, and visualization. We demonstrated in yeast, flies, and human cells that mim-tRNAseq accurately measures tRNA abundance, charging levels, and identifies modification patterns—all within a single sequencing experiment. This breakthrough has enabled many laboratories to begin investigate previously challenging aspects of tRNA biology.
Applying this technology to human stem cell differentiation, we combined mim-tRNAseq with measurements of RNA Polymerase III (Pol III) binding at tRNA genes with ChIP-Seq and translation rates with ribosome profiling. We found that during hiPSC differentiation into neural or cardiac cells, tRNA pools are extensively remodeled at the individual transcript level. Intriguingly, however, the overall tRNA populations at the anticodon level remained largely stable across different cell types. This stability was maintained by high expression of the most abundant tRNA transcripts within each anticodon family, ensuring consistent decoding speeds regardless of cell identity. Furthermore, we classified the 619 predicted human tRNA genes into three groups based on their expression patterns: only a third were “housekeeping” tRNAs expressed across all cell types, while the rest showed more variable expression. We discovered that Pol III occupancy at human tRNA genes is governed by specific sequence features, which we identified through computational analysis and verified experimentally. During differentiation, reduced activity of the mTORC1 signaling pathway activates the Pol III repressor MAF1, which restricts tRNA expression primarily to housekeeping tRNAs. These findings provide new insights into how tRNA levels are tightly controlled in human cells and set the stage for investigating how tRNA dysregulation may contribute to diseases.
Another key question we addressed was aimed at understanding how the protein production system adapts to the diverse and rapidly changing demands during cell differentiation. To explore this, we established workflows that allow us to control and study gene function over time using inducible CRISPR interference (CRISPRi) screens in hiPSCs and their differentiated progeny. Using this approach, we discovered that certain quality control pathways linked to mRNA translation are only crucial in specific cellular contexts. In particular, human stem cells heavily rely on pathways that manage and resolve ribosomes that have stalled or collided during protein synthesis. We identified a previously unrecognized type of ribosome collisions, which occur on mRNAs with high ribosome density and are caused by the slow transition of ribosomes from initiation to elongation. Our work has highlighted the importance of translation initiation as a physiological source of ribosome collisions and has provided valuable insights into the specialized pathways that maintain the fidelity of protein production in different cell types.
