henhaobio
Sign in

bioRxivMolecular BiologyPreprint

An integrated pipeline to count individual transcripts with single-cell resolution

Eva SheardowniD, Jane Yan To Ling, Servaas N. van der BurghtiD, Heli VaikkineniD, Beth GowingiD, Julie AhringeriD, Fursham HamidiD, QueeLim Ch'ngiD

PDFDOICode
In-depth full-text digest available · Jump to digest ↓

Abstract

Quantifying transcript abundance at single-cell resolution is important for understanding gene regulation in intact multicellular organisms. In Caenorhabditis elegans, RNA fluorescence in situ hybridization has been widely used to visualize transcripts, but conventional smFISH approaches can be limited by low signal-to-noise, poor performance with short transcripts, and workflows that do not readily support absolute transcript counting in identified cells. Here, we present an integrated experimental and computational pipeline for quantitative transcript analysis in whole-mount C. elegans embryos, larvae, and adults. The pipeline combines Hybridization Chain Reaction (HCR), confocal microscopy, RS-FISH spot detection, manual cell annotation in FIJI, and custom MATLAB-based spot assignment to quantify individual transcripts within defined cells. We show that this approach enables sensitive, specific, and multiplexed detection of transcripts, including short insulin-like peptide mRNAs, with single-cell resolution. Known spatial expression patterns were resolved in embryos, larvae, and adults, and probe specificity was validated using deletion mutants that remove HCR probe-binding sites. Applying this pipeline to ins-4, ins-6, and daf-28 in ASI and ASJ sensory neurons revealed cell-specific and context-dependent regulatory relationships across multiple mutant backgrounds. This workflow provides an accessible method for absolute transcript counting in anatomically intact C. elegans and should support mechanistic studies of gene regulation, cellular heterogeneity, and transcriptional network function.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (23 Sept 2026) · CC BY 4.0

HCR3.0 RNA-FISH linked to RS-FISH spot detection labels transcripts in C. elegans embryos, larvae and adults, and yields per-neuron counts of short insulin-like peptide mRNAs in L2 larvae.

Standard smFISH in C. elegans has limited signal-to-noise against autofluorescence and needs many probes per transcript, and some studies rely on noisy whole-ROI fluorescence rather than molecule counts. This work combines stage-specific fixation, low-cost HCR3.0 probe sourcing and RS-FISH spot counting into one pipeline, aiming to make quantitative transcript counting easier to adopt.

Key results

  1. ShownCounting RS-FISH spots was much less sensitive to experimental batch than background-corrected mean fluorescence on the same ins-4 L2 neuron images. Counting also separated genotypes more clearly and found more significant differences.

    Same L2 ins-4 images analysed both ways: experimental batch accounted for 28.8% of model-scale variance under the Gamma model of corrected mean fluorescence vs 2.2% under the Negative Binomial model of spot counts (Supplemental Table 1). Counting also flagged more significant genotype differences. · Quantitative analysis of gene regulation; Statistical Analysis; Tables

  2. ShownPer-cell counts showed that insulin-pathway mutations affect ins-4 differently in the ASI and ASJ sensory neurons of L2 larvae. Loss of daf-2 had opposite effects in the two neuron classes.

    Negative Binomial GLM(M) vs wild type: ins-6(tm2416) lowered ins-4 in ASJ but not ASI; daf-28(tm2308) raised it in ASI but not ASJ; daf-2(e1370) raised it in ASI and lowered it in ASJ. P-values were Benjamini-Hochberg corrected; Figure 5 marks significance as *** (P < 0.001) and refers to Supplemental Table 2 for details. · Quantitative analysis of gene regulation; Statistical Analysis; Figure legends

  3. ShownMultiplexed HCR labelling reproduced expected cell-specific patterns in embryos and L2 larvae and labelled germline nos-1 in adults, with little bleed-through between fluorescence channels.

    26-cell embryos: ref-1 in D, ABaraa and ABarap; end-1 in Ea and Ep; pes-10 in all somatic cells. L2 larvae: ins-4 mainly in ASI (low in ASJ), ins-6 high in ASI and ASJ, daf-28 in ASI, ASJ and additional head neurons. Adults (single probe): nos-1 in the germline syncytium. Evidence is qualitative imaging. · Multiplexed transcript visualization at…; Specific labelling of distinct transcripts; Figure legends

  4. ShownStage-specific fixation gave high staining rates in L2 larvae, and an added methanol step improved embryo nuclear morphology and adult detection; amplified spots stayed near diffraction-limited size, supporting single-molecule counting.

    L2 larvae fixed with formaldehyde only: ~90% of animals stained in all 3 channels across 3 independent batches. An added methanol step preserved 8-cell embryo nuclei and raised adult detection (not quantified in the text). Lateral FWHM averaged ~0.3 μm, with >90% of objects below 0.5 μm using a 1.4 NA objective. · Multiplexed transcript visualization at…; Fixation and permeabilization; HCR probe design, hybridization and staining; Image Analysis

  5. SuggestedHCR detected short insulin-like peptide mRNAs that are hard to target with smFISH. The authors report seeing transcripts with far fewer probe pairs than the recommended number.

    15 or more probe pairs per transcript are recommended, but the authors report success with as few as 4. ins-4, ins-6 and daf-28 were detected in ASI/ASJ. Probe-pair counts per gene and detection efficiency are not reported. · Low-cost probe and amplifier design; Multiplexed transcript visualization at…; Discussion

Methods, data and reproducibility

Animals are frozen with fixative and fixed by stage-specific protocols, then hybridized with split-initiator HCR3.0 probes, amplified and confocal-imaged. RS-FISH detects 3D spots, FIJI ROIs outline cells, a MATLAB convex-hull script counts spots per cell, and Gamma or Negative Binomial GLM(M)s compare genotypes.

  • Wild-type N2 and mutant L2 larvae (QL1257 daf-2(e1370), QZ81 ins-6(tm2416), QZ83 daf-28(tm2308)) with ins-4 counted in ASI and ASJ neuron pairs · C. elegans
  • Wild-type L2 larvae triple-labelled for ins-4, ins-6 and daf-28 and scored for HCR staining efficiency · C. elegans · n = 3 independent experimental batches
  • Early embryos: 8-cell stage for fixation comparison; 26-cell stage multiplexed for pes-10, ref-1 and end-1 · C. elegans
  • Adult worms labelled for nos-1; fixation with formaldehyde alone vs methanol plus formaldehyde · C. elegans
Methods & tools
HCR3.0 split-initiator hybridization chain reaction RNA-FISH, RS-FISH (Radial Symmetry spot detection, Fiji plug-in with RANSAC), FIJI/ImageJ (freehand ROI tracing, batch macro), Custom MATLAB script (convex hull spot-to-cell assignment), Napari (3D visualization of segmentation and spots), R lme4 (Gamma and Negative Binomial GLM/GLMM), tidyverse, Custom HCR probe designer, Integrated DNA Technologies oligopools; Molecular Instruments hairpin amplifiers, Zeiss LSM 800 and Nikon AX R confocal microscopes, DAPI nuclear staining for cell identification, Benjamini-Hochberg multiple-testing correction
Software & environment
R with lme4 and tidyverse; RS-FISH Fiji/ImageJ plug-in run through a custom ImageJ macro; custom MATLAB script; Napari (Python). No software versions stated. Code was proofread with Claude (Anthropic, accessed 2026).
Reproducibility gaps
No repository or accession for raw images is given. The full wet-lab protocol and Supplemental Tables 1 and 2 are only in supplementary material. Software versions are not reported, and cell segmentation is manual.
Limitations (5)
  • Labelling may not be fully efficient, and two closely spaced transcripts may merge below the resolution of conventional light microscopy, which could cause undercounting. The authors suggest super-resolution or expansion microscopy as a fix. · stated by the authors
  • Because neuronal morphology is complex, cells are segmented by hand in FIJI. Operator differences in ROI boundaries are therefore the main source of variability in the pipeline; the authors suggest training to minimize this. · stated by the authors
  • The authors place detailed optimization of imaging parameters outside the paper's scope and refer readers to an external practical guide instead. · stated by the authors
  • The main text gives no per-genotype animal or neuron counts, effect sizes or exact p-values for the ins-4 comparisons; Figure 5 points to Supplemental Table 2 for details. Counting accuracy is not benchmarked against smFISH or an independent ground truth. · noted in this summary
  • The batch-robustness result (28.8% vs 2.2% batch variance) comes from a single gene at a single larval stage. The claims of detection with 4 probe pairs and of better adult fixation lack quantified detection rates in the text. · noted in this summary

Read this if you need per-cell mRNA counts for short or multiplexed transcripts in whole-mount C. elegans and want a low-cost HCR3.0 plus RS-FISH workflow with released analysis code.

Citations
0

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISAguilera '26Packer '19Barrett '25Barrett '22Bernard '26Parker '21
65% semantic matchNeuroscience
Jose David Aguirre Aguilera, Xuan Wan, Carsten H. Tischbirek+4 more·bioRxiv·Apr 11, 2026