Optimized tRNA Structure-seq reveals robust tRNA secondary structures in S. cerevisiae under mild stress conditions

This study presents an optimized tRNA Structure-seq pipeline that integrates natural tRNA modifications and refined energy parameters to achieve 94% accuracy in predicting *S. cerevisiae* tRNA secondary structures, revealing that these molecules maintain their canonical cloverleaf conformation and structural robustness even under mild stress conditions.

The Gist

Imagine your cell is a bustling, high-tech factory. Inside this factory, there are millions of tiny workers called tRNAs (transfer RNAs). Their job is to read the blueprints (DNA) and deliver the correct building blocks (amino acids) to assemble proteins, which are the machines that keep the factory running.

For these workers to do their job, they need to be folded into a very specific shape, like a paper crane. If the paper crane is crumpled or folded wrong, it can't deliver its package, and the factory stops.

This paper is about a team of scientists who wanted to take a perfect photograph of these paper cranes while they are working inside the factory to see exactly how they are folded. They also wanted to see if the cranes change shape when the factory gets a little hot, salty, or stressed.

Here is the story of their discovery, broken down simply:

1. The Problem: The "Crystal Ball" Was Wrong

For a long time, scientists tried to guess what these tRNA cranes looked like just by looking at their chemical "ingredients" (their sequence). They used computer programs that acted like a crystal ball, trying to predict the shape based on physics.

• The Result: The crystal ball was terrible. It only got the shape right about 57% of the time. It was like trying to guess the shape of a folded paper crane just by looking at a flat piece of paper; the computer kept guessing the wrong folds.

2. The New Tool: "Chemical Flash Photography"

The scientists developed a better way to take a picture. They used a chemical called DMS (Dimethyl Sulfate).

• The Analogy: Imagine the tRNA crane is made of paper. The scientists spray it with a special "ink" (DMS). The ink only sticks to the parts of the paper that are sticking out into the air (unpaired). The parts that are folded tight against each other (paired) stay clean.
• By reading where the ink stuck, they could map out exactly where the folds were. This is called tRNA Structure-seq.

3. The Big Breakthrough: Adding "Hidden Instructions"

Even with the chemical spray, the computer predictions were still only about 87% accurate. The scientists realized they were missing a crucial piece of the puzzle: Modifications.

• The Analogy: Think of the tRNA paper crane as having little "sticker notes" or "glue dots" placed on it by the factory workers after it was printed. These stickers (called natural modifications) physically prevent certain parts of the paper from folding together.
• The computer didn't know these stickers existed, so it kept trying to fold the paper in impossible ways.
• The Fix: The scientists updated their computer program to "see" these stickers. They told the computer: "Hey, there's a sticker here, so this part CANNOT fold."
• The Result: With this new rule, the computer's accuracy jumped to 94%. It finally got the picture right almost every time.

4. The Surprise: The Cranes Are Tough!

Once they had a perfect way to take pictures, they decided to test the cranes under stress. They turned up the heat, added salt to the water, and even added a drug that slows down the factory.

• The Expectation: They thought the cranes would get messy, unfold, or change shape to adapt to the stress.
• The Reality: The cranes were incredibly tough. Even when the factory got hot or salty, the tRNAs stayed in their perfect "paper crane" shape. They didn't crumple.
• The Only Change: The only thing that changed slightly was the number of "stickers" (modifications) on some cranes, but the overall shape remained rock-solid.

Why Does This Matter?

This paper is a big deal for two reasons:

1. Better Tools: They built a super-accurate "camera" (the optimized pipeline) that can now see the true shape of these tiny workers in living cells. This helps us understand how cells work at a molecular level.
2. Resilience: They discovered that these essential workers are built to last. Even when the cell is under mild stress, the tRNAs don't fall apart. This suggests that the cell has a very stable foundation, and the "stickers" (modifications) play a huge role in keeping everything stable.

In a nutshell: The scientists fixed a broken computer program by teaching it to look for "stickers" on the RNA, allowing them to see that these tiny cellular workers are surprisingly sturdy and stay the same shape even when things get a little crazy around them.

Technical Summary

1. Problem Statement

Accurate prediction of RNA secondary structure is essential for understanding RNA function, yet in silico methods relying solely on thermodynamic parameters often fail to predict complex, highly modified structures like transfer RNA (tRNA) with high fidelity.

• Specific Challenge: While tRNA is known to fold into a canonical cloverleaf secondary structure, computational predictions for Saccharomyces cerevisiae (yeast) tRNAs using standard minimum free energy (MFE) models achieve only ~57% accuracy.
• Limitations of Existing Methods: Previous attempts to improve accuracy using chemical probing data (DMS-seq) combined with standard pseudo-free energy parameters improved accuracy to ~87%, but this remained insufficient for reliable, genome-wide structural analysis. Furthermore, the impact of mild physiological stress on tRNA structural stability in eukaryotes was largely unknown.

2. Methodology

The authors extended their previously developed tRNA Structure–seq pipeline (originally validated in E. coli) to S. cerevisiae and implemented a multi-step optimization strategy:

• Experimental Workflow:

  • Chemical Probing: In vivo and in vitro DMS (dimethyl sulfate) treatment was performed on S. cerevisiae cells to probe nucleotide accessibility (modifying A, C, and G).
  • Sequencing: tRNAs were isolated, deacylated, and subjected to library preparation using Induro reverse transcriptase (a highly processive enzyme from a bacterial group II intron). This enzyme is capable of reading through heavily modified tRNAs and inducing misincorporations at both DMS-modified and native modification sites.
  • Data Processing: ShapeMapper2 was used to calculate mutation rates, distinguishing between DMS-induced modifications and natural post-transcriptional modifications (detected in "minus DMS" controls).

• Computational Optimization:

  • Native Modification Constraints: The pipeline explicitly incorporated data on three conserved eukaryotic tRNA modifications (m1G9, m22G26, and m1A58). Since these modifications block Watson-Crick base pairing, the authors converted these nucleotides to lowercase in the input sequence for RNAstructure Fold, forcing the algorithm to treat them as unpaired.
  • Parameter Optimization: The standard pseudo-free energy equation ($\Delta G_{pseudo} = m \ln[reactivity + 1] + b$) used to integrate DMS data was re-parameterized. The authors systematically varied the slope ($m$) and intercept ($b$) to find values that maximized prediction accuracy for tRNAs, resulting in a unified parameter set ($m=2.4, b=-0.8$) distinct from the parameters used for general RNAs.
  • Ensemble Analysis: The Rsample algorithm was employed to predict conformational ensembles and assess structural heterogeneity.

• Stress Conditions: The optimized pipeline was applied to cells under mild stress: heat (42°C), osmotic (3% NaCl), and antibiotic (cycloheximide) stress.

3. Key Contributions

1. Extension to Eukaryotes: Successfully adapted tRNA Structure–seq for S. cerevisiae, demonstrating its applicability beyond bacteria.
2. Novel Optimization Pipeline: Developed a refined analysis workflow that integrates native modification constraints and optimized pseudo-free energy parameters specifically for tRNA.
3. Discovery of Structural Robustness: Provided the first comprehensive, high-resolution view of the in vivo tRNA structurome in yeast under various stress conditions.

4. Key Results

• DMS Reactivity Patterns: DMS reactivity profiles in yeast mirrored those in E. coli, showing high reactivity at the CCA terminus, anticodon loop, and A58 (T-loop), consistent with the canonical cloverleaf structure.
• Dramatic Accuracy Improvement:
  • In silico only: 56.9% accuracy.
  • In silico + DMS restraints: 87.4% accuracy.
  • In silico + DMS + Native Modification Constraints: 89.4% accuracy.
  • Optimized Pipeline (DMS + Mod Constraints + Optimized Parameters): Achieved 94% average prediction accuracy.
• Role of Native Modifications: Incorporating constraints for m1G9, m22G26, and m1A58 alone (without DMS data) improved accuracy by >12 percentage points, suggesting these modifications act as an "epigenetic structural code" that suppresses non-canonical folding.
• Conformational Homogeneity: Rsample analysis revealed that while in vitro conditions sometimes allowed for multiple non-canonical conformations, in vivo conditions strongly favored the canonical cloverleaf structure (e.g., tRNA$^{Arg}$ and tRNA$^{Ser}$ showed >80% probability of the canonical fold in vivo vs. mixed populations in vitro).
• Stress Response:
  • Global Stability: Mild stress conditions (heat, osmotic, antibiotic) caused minimal changes to the global secondary structure of the tRNA population.
  • Modification Dynamics: While structure remained stable, the modification level of m1G9 increased significantly under stress (detected via mutation rates and LC-MS/MS), whereas m22G26 and m1A58 levels remained constant.
  • Local Perturbations: A small subset of tRNAs (e.g., tRNA$^{Pro}$, tRNA$^{Val}$) showed minor local structural changes (e.g., in the acceptor stem or D-loop) under specific stresses, but the overall cloverleaf topology was preserved.

5. Significance

• Methodological Advancement: The study establishes a highly accurate (94%) pipeline for predicting tRNA secondary structures in vivo. This resolves the long-standing issue of poor in silico prediction for highly modified RNAs by treating native modifications as structural constraints.
• Biological Insight: The findings challenge the notion that tRNAs are highly dynamic or plastic under mild stress. Instead, they demonstrate that tRNA secondary structures are intrinsically robust and maintained in the canonical cloverleaf form even under physiological perturbations.
• Epigenetic Code: The results support the hypothesis that conserved tRNA modifications function as a structural code that guides folding and prevents misfolding, acting as a quality control mechanism.
• Future Applications: This optimized pipeline provides a powerful tool for investigating tRNA structure regulation in development, disease, and diverse environmental conditions across all domains of life.