Evolutionary emergence and preservation of microproteins encoded by upstream ORFs

By integrating ribosome profiling and Nanopore direct RNA sequencing across multiple *Saccharomyces* species, this study identifies thousands of translated upstream ORFs in yeast that encode conserved, functional microproteins under purifying selection, while demonstrating that downstream ORFs are largely non-conserved and less likely to be functional.

The Gist

The Big Picture: Finding Hidden Gems in the "Junk" Drawer

Imagine your cell's DNA as a massive library of instruction manuals. For a long time, scientists thought these manuals only had one important story per page: the "Main Character" (the main protein that does the heavy lifting in the cell). Everything else on the page—the stuff before the story starts and after it ends—was considered "junk" or just background noise.

This paper is like a team of detectives who decided to look closer at that "background noise." They discovered that hidden in the margins of these manuals are tiny, secret stories called microproteins. Specifically, they found these secrets hiding in the Upstream ORFs (uORFs)—the little notes written before the main story begins.

The Mystery: Are These Notes Real or Just Scribbles?

For years, scientists saw ribosomes (the cell's "readers") stopping to read these little notes. But they weren't sure if the notes were important instructions or just random scribbles that happened to get read by accident.

• The Problem: To know if a note is important, you need to see if it's been copied and kept by many different people over time. If a note is just a random scribble, it won't survive. If it's a vital instruction, it will be preserved.
• The Limitation: The researchers only had detailed "reading logs" (Ribo-Seq data) for one species of yeast (S. cerevisiae). Without logs from its relatives, they couldn't tell if the notes were being ignored or cherished by evolution.

The Investigation: A Family Reunion of Yeast

To solve this, the team went on a massive field trip. They didn't just study one yeast; they studied seven different species of Saccharomyces yeast, which are like cousins separated by about 16 million years of evolution.

1. Mapping the Territory: First, they used a high-tech microscope (Nanopore sequencing) to map out the entire "page" of the instruction manuals, including the margins (the 5' and 3' ends) that were previously missing from the maps.
2. Listening to the Readers: They then listened to the ribosomes reading these manuals in all seven species. They found thousands of instances where ribosomes were reading the "margin notes" (uORFs) and turning them into tiny proteins.
3. The "Conservation" Test: They asked: "Did the cousins also read these notes?"
   • The Result: Most of the notes were indeed just random scribbles (low translation, not conserved).
   • The Surprise: However, they found 195 specific notes that were being read by all the cousins, from the closest relatives to the distant ones. These weren't accidents; they were important instructions that evolution had decided to keep.

What Do These Microproteins Do?

The team found that these "surviving" microproteins have some special traits:

• They are tough: They are often made of sticky, water-repelling ingredients (hydrophobic), suggesting they might stick to cell membranes like little anchors.
• They are efficient: In the cases where the note was conserved, the ribosome read it just as loudly and clearly as the main story.
• They are selected for: When they compared the DNA of these notes across species, they found that the "spelling" was being carefully preserved. Nature was actively fixing mistakes in these notes, proving they have a job to do.

The Analogy: Think of a main protein as a full-length novel. These microproteins are like the chapter summaries or footnotes that appear before the story starts. For a long time, we thought footnotes were just random thoughts. But this study found that some footnotes are actually essential plot points that have been copied into every edition of the book for millions of years because they help the story make sense.

A Twist in the Plot: The "Side-Story" Isoforms

The researchers also found something very strange. In some cases, the cell doesn't just read the whole manual from start to finish. Sometimes, it cuts the manual short!

• The Discovery: They found "alternative endings" (alternative transcript isoforms). In these cases, the cell stops reading right after the "margin note" and never gets to the main story.
• The Implication: This means the cell can produce the microprotein (the footnotes) without producing the main protein (the novel). It's like a factory that can make just the cover art for a book without printing the whole book inside. This gives the cell a new way to control how much of each product it makes.

The Takeaway

This paper changes how we see the genetic code. It suggests that:

1. Evolution is a tester: Cells constantly try out tiny, random proteins. Most fail and disappear.
2. Survival of the fittest: A few of these tiny proteins turn out to be so useful that evolution keeps them, refining them over millions of years.
3. New Tools: We now have a list of 195+ "hidden gems" (microproteins) that are likely doing important work in yeast, and potentially in humans too.

In short, the cell isn't just writing one story per gene; it's writing a whole anthology of tiny, powerful stories, and we are just starting to read them.

Technical Summary

1. Problem Statement

While ribosome profiling (Ribo-Seq) has revealed widespread translation of non-canonical open reading frames (ncORFs), such as upstream ORFs (uORFs) and downstream ORFs (dORFs), in eukaryotic mRNAs, the biological significance of this translation remains largely unknown. A primary limitation in the field is the lack of comparative Ribo-Seq data from closely related species, making it difficult to distinguish between:

• Biologically functional microproteins: Those preserved by purifying selection.
• Translational noise: Pervasive, low-level translation events that are merely by-products of ribosome scanning and lack functional constraints.

Most existing studies focus on single species (typically Saccharomyces cerevisiae), preventing the identification of phylogenetically conserved translation events that would indicate functionalization.

2. Methodology

The authors employed a multi-omics approach combining long-read transcriptomics, ribosome profiling, and evolutionary genomics across seven Saccharomyces species.

• Species Selection: The study analyzed S. cerevisiae and six closely related species (S. paradoxus, S. mikatae, S. jurei, S. kudriavzevii, S. arboricola, S. uvarum), spanning an evolutionary divergence of approximately 16 million years.
• Transcriptome Reconstruction (Nanopore dRNA-Seq):
  • Generated Nanopore direct RNA sequencing data (33–49 million reads per species) to capture full-length transcripts, including 5' and 3' untranslated regions (UTRs).
  • Polished reads using Illumina RNA-Seq data to improve accuracy.
  • Reconstructed comprehensive gene annotations (5'UTR, CDS, 3'UTR) for all seven species, overcoming the limitation of existing annotations that often lack UTRs.
• Translation Profiling (Ribo-Seq):
  • S. cerevisiae: Aggregated and analyzed 100 high-quality Ribo-Seq datasets (640 million mapped reads) to identify translated ncORFs using RibORF v.2.0.
  • Other Species: Generated new Ribo-Seq data for the six additional species (85–189 million reads each) and integrated existing data for S. paradoxus.
  • Criteria: Translated ORFs were defined by significant three-nucleotide periodicity and a RibORF score > 0.6.
• Evolutionary Analysis:
  • Identified one-to-one orthologous genes and constructed pairwise mRNA alignments.
  • Mapped conserved translated ncORFs across species.
  • Calculated the ratio of non-synonymous to synonymous substitutions (dN/dS) using concatenated alignments to detect signatures of purifying selection.
• Alternative Isoform Detection:
  • Analyzed the 3' ends of Nanopore dRNA reads to identify alternative transcription termination sites (alt-TTS) using peak calling (MACS2).
  • Investigated whether these alternative isoforms excluded the main coding sequence (CDS) while retaining uORFs.

3. Key Contributions

• Comprehensive UTR Annotation: Provided the first complete 5' and 3' UTR annotations for six non-cerevisiae Saccharomyces species, enabling comparative analysis of ncORFs.
• Cross-Species Conservation Framework: Established a rigorous pipeline to identify ncORFs that are not only translated but evolutionarily conserved, serving as a proxy for functional microproteins.
• Discovery of Alternative Isoforms: Demonstrated that uORF translation is frequently associated with the production of alternative transcript isoforms that contain the uORF but lack the main CDS, suggesting a mechanism for independent microprotein production.

4. Key Results

A. Landscape of ncORF Translation in S. cerevisiae

• Prevalence: Identified 2,332 translated uORFs and 1,008 translated dORFs.
• Translation Levels: uORFs are translated at significantly higher levels than dORFs. However, most ncORFs are translated at levels lower than the main CDS.
• Sequence Bias: Translated microproteins are enriched in hydrophobic amino acids (C, F, Y, H) and positively charged residues (Arginine), while being depleted of negatively charged amino acids. This bias suggests an origin from non-coding sequences.
• Start Codons: While ATG is dominant, 27% of translated uORFs initiate at near-cognate codons (TTG, CTG, GTG).

B. Phylogenetic Conservation and Functionalization

• Conserved uORFs: Only 195 uORFs (7.6% of translated uORFs) showed evidence of translation in multiple Saccharomyces species. In contrast, dORFs were rarely conserved (2.77%).
• Translation Efficiency: Conserved uORFs are translated at levels comparable to the main CDS, whereas non-conserved uORFs are translated at much lower levels.
• Signatures of Selection:
  • Conserved uORFs displayed dN/dS ratios significantly lower than 1 (ranging from 0.47 to 0.8 depending on phylogenetic distance), indicating purifying selection acting on the encoded microproteins.
  • Specific examples of highly conserved microproteins include those encoded by KEX2 (13 aa), SCP160 (16 aa), and PUS2 (27 aa).
• Regulatory vs. Functional: The conservation of uORFs was not correlated with the regulation of the main CDS translation efficiency (TE) under oxidative stress (unlike the known GCN4 case). This suggests that many conserved uORFs encode microproteins with inherent functions rather than acting solely as regulatory elements.

C. Alternative Transcript Isoforms

• Discovery: Identified 204 genes with alternative transcription termination sites (alt-TTS) located in the 5'UTR or early CDS.
• Association: 46.5% of genes with alt-TTS contained translated uORFs (nearly double the random expectation).
• Mechanism: In genes like PHO80, ANY1, and SER3, alternative isoforms were found that included the uORFs but excluded the main CDS.
  • In SER3, the alternative isoform (containing 4 uORFs) was actually more abundant than the standard isoform containing the CDS.
  • This implies a mechanism where microproteins can be produced independently of the main protein product, potentially allowing for independent regulation of their abundances.

5. Significance

• De Novo Gene Birth Model: The study supports a model where pervasive, low-level translation of uORFs creates a "reservoir" of potential microproteins. A subset of these gains function, undergoes purifying selection, and becomes conserved, representing a pathway for de novo gene birth.
• Functional Microproteins: The identification of ~195 conserved, selected uORFs provides a high-confidence catalog of novel microproteins for functional characterization, many of which appear to be membrane-associated or involved in transport.
• Regulatory Complexity: The discovery of alternative isoforms that produce microproteins without the main protein challenges the view of uORFs solely as translational repressors. It suggests a more complex regulatory landscape where polycistronic transcripts can yield independent functional products.
• Evolutionary Insight: By spanning 16 million years of evolution, the study distinguishes between transient translational noise and biologically significant microproteins, offering a template for similar studies in other eukaryotic lineages.