Whole-organism spatial transcriptomics at single-cell resolution in C. elegans

This paper presents a scalable single-molecule fluorescence in situ hybridization workflow that achieves whole-organism spatial transcriptomics at single-cell resolution in *C. elegans*, enabling the multiplexed profiling of up to 40 genes to identify neuronal classes and reveal sex- and neuron-specific expression patterns.

The Gist

Imagine you have a tiny, transparent city called C. elegans (a microscopic worm). This city is so small that you can see the entire thing at once, and every single building (cell) has a specific job and a known address. Scientists have long wanted to know exactly what "signs" (genes) are posted on every single building in this city to understand how the city functions.

However, reading these signs is tricky. The city is wrapped in a tough, waxy fence (the cuticle) that keeps things out, and the signs are so small and numerous that trying to read them all at once is like trying to read a library's worth of books in a single second.

This paper introduces a new, clever way to read these signs inside the whole worm, cell by cell, without taking the city apart. Here is how they did it, explained simply:

1. Breaking the Fence (Making the Worm Permeable)

The worm's outer shell is like a fortress wall that blocks the scientists' tools. To get inside, the team used a chemical "key" (TCEP) to soften the wall, followed by a gentle enzymatic "dissolver" (collagenase) to create tiny holes. They then wrapped the worm in a soft, clear gel (like Jell-O) to hold everything in place while they worked. This is like putting the city in a clear, protective bubble so they can poke holes in the fence without the city falling apart.

2. The "Flashlight" Game (Sequential Imaging)

Imagine you want to read 40 different signs in a room, but you only have two flashlights (fluorescent colors). You can't read them all at once.

• The Trick: The scientists use a "tagging" system. They attach a unique, invisible "handle" to each of the 40 signs they want to find.
• The Process:
  1. They shine a flashlight that only lights up the handles for Sign #1 and Sign #2. They take a picture.
  2. They wash away the flashlights (the "readout" probes) so the room goes dark again.
  3. They shine the flashlights on the handles for Sign #3 and Sign #4. They take another picture.
  4. They repeat this 20 times.
• The Result: By stacking these 20 pictures on top of each other, they can see all 40 signs at once, knowing exactly where each one is located. It's like taking 20 photos of a crowded room, each time highlighting a different pair of people, and then combining them to see everyone clearly.

3. The "Address Book" (Finding the Right Cells)

Once they have the pictures of the glowing signs, they need to know which cell is holding them.

• The Problem: In many tissues, cells are like a messy pile of clay; it's hard to tell where one ends and another begins.
• The Solution: The scientists used a special dye (DAPI) that makes the "nucleus" (the cell's brain/command center) glow blue. They used a smart computer program (AI) to draw a 3D outline around every single blue nucleus.
• The Assignment: If a glowing sign is found inside a blue outline, the computer says, "This sign belongs to this cell." If it's floating nearby, the computer uses a bit of math (probability) to guess which cell it belongs to based on how close it is.

4. The "ID Card" System (Identifying Neurons)

The worm has about 302 neurons (brain cells), and many look identical. How do you know if a cell is a "Chef" or a "Baker"?

• The Strategy: The team created a curated list of "ID genes." These are genes that only specific types of neurons turn on.
• The Process: By checking which ID genes are glowing in a specific cell, and looking at where that cell is sitting in the worm (head vs. tail), they can say with high confidence: "Ah, this glowing cell is a 'Sensory Ray Neuron'."
• The Achievement: They successfully identified up to 86 different types of neurons, including 27 types that only exist in male worms.

5. Why This Matters (The Big Picture)

Before this, scientists had to either:

• Crush the whole worm to read the genes (like blending a smoothie to taste the fruit—you know what's in it, but you lose the shape).
• Look at one cell at a time under a microscope (very slow and tedious).

This new method is like having a high-resolution, 3D map of the entire city where you can see exactly which shop is selling what, without ever breaking the city apart.

What did they discover?

• They confirmed that male and female worms have different "signs" on their cells, specifically in the neurons used for mating.
• They found that some genes are only turned on in very specific, tiny groups of cells, which helps scientists understand how the worm's nervous system is wired.

In short: The scientists built a "molecular camera" that can take a 3D photo of a whole living worm, label every single cell, and read the genetic instructions inside them, all while keeping the worm's structure perfectly intact. This opens the door to understanding how complex behaviors (like mating or sensing food) are controlled by specific cells in a way that was previously impossible.

Technical Summary

1. Problem Statement

While spatial transcriptomics has revolutionized the understanding of tissue organization, applying these techniques to whole, intact organisms remains a significant challenge. Existing methods often suffer from:

• Limited Spatial Resolution: Inability to resolve individual cells within a dense tissue context.
• Low Multiplexing Capacity: Difficulty profiling multiple genes simultaneously without extensive image alignment or complex barcoding workflows.
• Sample Integrity: The need to section tissues destroys the native spatial context of the whole organism.

In Caenorhabditis elegans (C. elegans), despite its compact size, transparency, and well-mapped connectome, current approaches struggle to profile multiple gene expression patterns across the entire intact worm while preserving single-cell spatial context.

2. Methodology

The authors developed a robust, scalable workflow combining sequential single-molecule fluorescence in situ hybridization (smFISH) with advanced computational analysis.

A. Experimental Workflow (Wet Lab)

• Sample Preparation & Permeabilization: To overcome the impermeability of the C. elegans cuticle, the protocol employs a multi-step chemical treatment:
  1. TCEP Treatment: Chemically reduces disulfide bonds in cuticle collagen fibers.
  2. SDS & Collagenase IV: Further permeabilizes the cuticle to allow reagent entry.
  3. Post-fixation: Uses BS(PEG)5 to stabilize tissue integrity after harsh permeabilization.
  4. Proteinase K: Digests cellular proteins to reduce autofluorescence and background noise.
• Probe Design & Hybridization:
  • Primary Probes: Target specific mRNAs and contain unique overhang sequences.
  • Readout Probes: Fluorescently labeled probes bind to the overhangs.
  • Sequential Imaging: The system uses two fluorescent channels (e.g., Alexa 647 and Cy3B). By performing 20 hybridization rounds, the team can image up to 40 genes (20 genes per channel) in a single experiment.
  • Stripping: Readout probes are stripped via formamide washes between rounds to allow the next set of probes to hybridize.
• Embedding: Samples are embedded in a hydrogel to maintain structural integrity during the sequential stripping and imaging cycles.

B. Computational Analysis Pipeline

• Image Processing: Background correction, contrast enhancement, and Gaussian filtering.
• Transcript Detection: Uses the Big-FISH library to detect and decompose fluorescent spots into single-molecule signals.
• Nuclear Segmentation: Instead of relying on membrane staining (which is difficult in whole-mount worms), the authors trained a custom StarDist model on DAPI-stained nuclei to generate 3D nuclear masks.
• Transcript-to-Cell Assignment:
  • Hard Assignment: Transcripts inside the nuclear mask are assigned directly.
  • Probabilistic Soft Assignment: Transcripts in the perinuclear region (within 2x nuclear size) are assigned using a Gaussian Mixture Model (GMM) that accounts for nuclear geometry and size, minimizing bias.
• Neuron Identification: A curated panel of marker genes is integrated with anatomical position data to manually identify and label specific neuronal classes (e.g., AVA, PLM).

3. Key Contributions

1. Whole-Organism smFISH Pipeline: A validated protocol enabling multiplexed imaging (up to 40 genes) across the entire C. elegans body with single-cell resolution.
2. Curated Marker Gene Panel: A set of validated marker genes designed to reliably identify 86 distinct neuronal classes (including 27 male-specific classes) across the head and tail regions.
3. Probabilistic Assignment Algorithm: A novel method for assigning transcripts to cells based on nuclear masks and local geometry, overcoming the lack of membrane segmentation in whole-mount samples.
4. Sex-Specific Profiling: The first comprehensive spatial map of sexually dimorphic gene expression in C. elegans at single-cell resolution.

4. Key Results

• Validation of Specificity: Colocalization assays using the housekeeping gene ama-1 (targeted by two independent probe sets) demonstrated a 77.4% ± 4.7% colocalization efficiency, confirming probe specificity.
• Correlation with Sequencing Data:
  • Gene expression profiles generated by sequential smFISH showed strong concordance with both bulk RNA-seq and single-cell RNA-seq (CeNGEN) datasets.
  • Heatmaps of neuron class expression in hermaphrodites and males closely matched CeNGEN reference data.
  • Quantitative comparisons showed high correlation ($R^2$ values along the diagonal) between smFISH counts and RNA-seq TPM values.
• Neuronal Classification: The method successfully identified up to 86 neuronal classes in intact worms.
• Sexually Dimorphic Expression: The study confirmed and spatially resolved the expression of male-specific genes (e.g., trf-1, paml-2, F26C11.3) in sensory ray neurons and CEM neurons, as well as shared expression patterns in sex-common neurons.
• Restricted Expression Patterns: The method identified highly specific expression patterns (e.g., mec-7 in PLM neurons), demonstrating its utility for generating genetic driver lines for specific neuron types.

5. Significance

• Scalability and Reproducibility: This framework provides a reproducible, scalable approach for spatial transcriptomics in intact organisms, bypassing the need for tissue sectioning.
• Bridging Molecular and Circuit Biology: By linking gene expression directly to the known C. elegans connectome and cell lineage, this method allows researchers to map molecular changes to specific circuit functions and behaviors within the same animal.
• Foundation for Future Studies: The curated marker panels and optimized protocols serve as a resource for the community to:
  • Generate specific genetic driver lines.
  • Study inter-tissue signaling (e.g., gut-brain axis).
  • Investigate cellular asymmetry (e.g., ASEL/ASER neurons) in their native spatial context.
• Overcoming Limitations: While the method currently relies on nuclear masks (potentially missing transcripts in long axonal projections), it establishes a critical baseline for whole-organism spatial analysis that can be further refined with improved membrane staining techniques in the future.

In summary, this paper presents a transformative toolkit that enables high-resolution, spatially resolved gene expression profiling in the intact C. elegans nervous system, bridging the gap between single-cell genomics and whole-organism physiology.