A pilot study for whole proteome tagging in C. elegans

This pilot study demonstrates the high efficiency and utility of a scalable approach for simultaneously tagging 30 essential *C. elegans* genes with three distinct fluorophores, successfully revealing novel expression patterns and subcellular localizations to pave the way for whole-genome proteome tagging.

The Gist

Imagine you have a massive, incredibly complex city (the C. elegans worm) with 20,000 different types of workers (proteins) doing jobs to keep the city running. For years, scientists have been able to read the "shift schedules" (RNA transcripts) to guess who is working and where. But here's the problem: just because a worker is scheduled to be on shift doesn't mean they actually show up, or that they are doing the job they were hired for. Sometimes the schedule is wrong, or the worker is hiding in a different building than expected.

To solve this, scientists Matthew Eroglu and Oliver Hobert decided to do something bold: put a glowing flashlight on every single worker in the city so they could watch them in real-time.

The Big Challenge: Too Many Workers, Too Little Time

In the past, scientists had to go to the city, find one worker, attach a flashlight, and wait for them to grow up. Doing this for all 20,000 workers one by one would take about 100 years. That's too slow!

The researchers asked: Can we attach flashlights to three workers at the same time?

The "Three-in-One" Strategy

Think of this like a high-speed assembly line. Instead of sending one car down the track, they built a machine that attaches three different colored lights to three different cars simultaneously.

1. The Tools: They used a molecular pair of scissors (CRISPR/Cas9) to cut the DNA and glue a glowing tag onto the protein.
2. The Colors: They chose three super-bright colors: Blue, Gold, and Red.
   • Blue was for the "loud" workers (highly abundant proteins) that are easy to spot from far away.
   • Gold was for the "medium" workers.
   • Red was for the "quiet" workers (low abundance) that are hard to see.
3. The Test: They picked 30 random workers (genes) from the city's schedule. They grouped them into 10 teams of three. In each injection, they tried to tag all three members of a team at once.

The Results: A Surprise Party

The experiment was a huge success. They managed to successfully tag 24 out of the 30 workers. But the real magic wasn't just that they could do it; it was what they saw when they looked.

The "shift schedules" (RNA data) were often lying to them. Here are some of the surprises they found:

• The "Universal" Worker who was actually a "Gym Rat": One protein, EEF-1A.1, was thought to be everywhere, like a janitor cleaning every room. But when they turned on the flashlight, they saw it was actually only hanging out in the "gym" (the reproductive organs/germline), working out hard there.
• The "Gym Rat" who was actually a "Kitchen Chef": Another protein, ACDH-10, was predicted to be in the kitchen (intestine) helping with digestion. Instead, it was missing from the gym entirely and was only working in the kitchen and the city walls (skin).
• The "City Manager" with a Secret Office: A protein called HXK-1 was expected to be in the main office (muscles and nerves). Instead, it was found exclusively in the "sheath" surrounding the reproductive organs, like a manager who only visits the VIP lounge.
• The "Double Agent": Two proteins that were supposed to live in the mitochondria (the cell's power plants) were found there, but they weren't hanging out together. They were in different parts of the power plant. It's like finding two managers in the same building, but one is in the basement and the other is on the roof, doing completely different jobs.

Why This Matters

This study is like a pilot test for a massive project. It proved that:

1. Speed is possible: You can tag three genes at once without breaking the system.
2. The maps are wrong: The "schedules" (RNA) don't always match the reality of where proteins actually live.
3. Discovery is waiting: By looking at where proteins actually are, rather than where we think they are, we can find new biological secrets.

The Future

The authors are saying, "We proved we can tag 30 genes quickly. Now, let's scale this up to tag all 20,000." If they succeed, they will create the first-ever 3D, live, glowing map of a whole animal's body, showing exactly where every single protein is, what it's doing, and how it moves as the animal grows.

It's like going from reading a static phone book to watching a live, glowing movie of every person in a city, revealing secrets that were hidden in the dark.

Technical Summary

1. Problem Statement

While transcriptomic atlases provide valuable insights into animal biology, there is a well-documented discordance between mRNA levels and actual protein abundance, localization, and dynamics due to post-transcriptional regulation. Current efforts to map the C. elegans proteome are fragmented, with individual labs tagging specific genes of interest. At the current rate, covering the entire ~20,000-gene genome would take approximately 100 years, leading to redundant efforts on popular genes and leaving the majority of the proteome uncharacterized. The field lacks a coordinated, high-throughput strategy to scale protein tagging to the whole genome efficiently.

2. Methodology

The authors developed a pooled, high-throughput CRISPR/Cas9 tagging pipeline designed to insert fluorescent protein (FP) tags into multiple genomic loci simultaneously.

• Strategy: Instead of tagging one gene per injection, the team pooled three distinct genetic targets into a single microinjection mix.
• Fluorophore Selection: To facilitate visual screening without co-CRISPR markers, they selected three spectrally distinct, high-brightness fluorophores:
  • mTagBFP2 (Blue): Assigned to the highest expression genes (≥500 FPKM).
  • mStayGold (Green/Yellow): Assigned to modestly expressed genes (100–500 FPKM).
  • mScarlet3 (Red): Assigned to the lowest expressed genes (18–100 FPKM).
  • Rationale: Red fluorophores offer the best signal-to-noise ratio in C. elegans due to minimal background autofluorescence and bacterial (OP50) fluorescence in the red spectrum.
• Screening Pipeline:
  1. Injection: 30 genes were divided into 10 pools of 3. Each pool was injected into 5 worms.
  2. Primary Screen: F1 progeny were screened via low-magnification fluorescence stereomicroscopy. The brightest tag (usually mTagBFP2 or mStayGold) served as a "marker" to identify positive worms.
  3. Secondary Screen: Worms positive for the bright marker were further screened via higher-magnification imaging or PCR to detect the dimmer tags (mStayGold or mScarlet3) within the same pool.
• Target Selection: 30 essential genes were selected based on bulk transcriptomics (FPKM) and single-cell RNA-seq (scRNA-seq) data to cover a wide range of expression levels and predicted subcellular localizations (nuclear, cytosolic, transmembrane).
• Tagging Design: Tags were inserted at the N- or C-terminus (preferring C-terminus) with an 18-amino acid linker to minimize functional disruption.

3. Key Contributions

• Proof of Concept for Pooled Tagging: Demonstrated that three independent genomic loci can be successfully tagged in a single injection round using novel guides and repair templates without pre-validated reagents.
• Efficient Screening Workflow: Established a "bright-to-dim" screening hierarchy that eliminates the need for co-CRISPR markers (like dpy-10), significantly reducing the complexity and time of strain generation.
• Scalability Analysis: Provided a cost and time analysis suggesting that a coordinated effort could tag the entire C. elegans proteome in approximately 5–6 years, compared to the projected 100 years under current decentralized practices.
• Discovery of Protein-Transcript Discordance: Validated that protein expression patterns often deviate significantly from transcriptomic predictions, highlighting the necessity of direct protein atlases.

4. Key Results

• Tagging Efficiency:
  • Success Rate: 24 out of 30 (80%) targets were successfully isolated.
  • Pool Success: In 4 out of 10 pools, all three tags were successfully isolated in a single worm. In others, tags were isolated in pairs or singles.
  • Visibility: All successful tags were visible via fluorescence stereomicroscopy, confirming that modern fluorophores can detect proteins even at low expression levels (down to ~18 FPKM).
  • Failure Analysis: Failures were attributed to either technical issues (e.g., poor repair template purity, which was resolved upon re-injection) or biological constraints (loss of protein function leading to lethality).
• Unanticipated Expression Patterns:
  • Tissue Specificity: Many genes predicted to be ubiquitous based on scRNA-seq showed strong tissue-specific enrichment.
    • EEF-1A.1 (translation factor) showed strong germline enrichment.
    • HXK-1 (glucokinase) was highly enriched in the gonadal sheath, contrary to predictions of muscle/neuron enrichment.
    • ACDH-10 was depleted in the germline despite broad somatic expression.
  • Germline Bias: There was a general trend of underestimating protein levels in the germline when relying solely on bulk or single-cell transcript data.
• Unanticipated Subcellular Localization:
  • Mitochondrial Heterogeneity: Two proteins predicted to be mitochondrial (GDH-1 and KARS-1) localized to distinct, non-overlapping subsets of mitochondria in the germline, suggesting metabolic specialization of organelles.
  • Dynamic Localization: VHA-2 (V-ATPase) showed specific perinuclear localization in the germline, and HAT-1 (histone acetyltransferase) shifted from cytoplasmic in early embryos to nuclear in adults.
  • Unexpected Compartments: TDO-2 was found to be cytoplasmic, and ZK632.9 showed diffuse localization rather than the predicted nuclear confinement.

5. Significance

This study serves as a critical roadmap for the systematic annotation of the C. elegans proteome.

• Methodological Impact: It proves that high-throughput, pooled CRISPR tagging is feasible, efficient, and scalable. The "bright marker" screening strategy drastically reduces the labor required to generate tagged strains.
• Biological Insight: The pilot revealed that transcriptomic data alone is insufficient for predicting protein localization and abundance. The observed discrepancies (e.g., germline enrichment, subcellular compartmentalization) underscore the value of direct protein imaging for understanding cellular physiology.
• Future Outlook: The authors propose that a centralized, distributed effort using this protocol could complete a whole-proteome atlas within a few years. Such an atlas would provide an unprecedented resource for studying development, metabolism, and disease mechanisms in a multicellular organism, bridging the gap between genomics and functional proteomics.