$$\rightleftharpoonup{xx}$$
$$\longleftharp{xx}$$,
$$\longrightharp{xx}$$,
Understanding how the brain is assembled requires identifying both the diversity of cell types present and the gene repertoires expressed within each cell type1. Recent advances in single-cell RNA sequencing have enabled researchers to address these questions at high resolution2. However, many single-cell approaches do not preserve spatial information, requiring additional experiments to map transcriptional cell-type clusters to their locations within the developing brain.
Binary gene expression systems such as GAL4-UAS can provide positional information but are not quantitative3. In addition, delays between gene expression and reporter detection can obscure physiologically relevant expression patterns during early developmental stages. Immunostaining is another widely used approach for determining tissue-specific expression patterns4. However, secreted proteins are often diffuse and difficult to localize accurately.
Nucleotide-labeling approaches, such as in situ hybridization (ISH), have also been widely used to obtain spatial information about gene expression5. In ISH, labeled DNA oligonucleotides bind to reverse-complementary RNA sequences of interest. Early ISH methods used radiolabeled DNA probes6, whereas later protocols incorporated safer labels such as digoxigenin, fluorescein, and rhodamine7,8,9. Scientists rapidly adapted ISH for use in Drosophila melanogaster, where it was applied to characterize polytene chromosome organization and map the expression patterns of gap genes such as hunchback in whole-mount embryos10,11.
Despite its widespread utility, several technical limitations hindered the adoption of ISH in thicker whole-mount preparations such as the larval optic lobe. Many in situ protocols relied on diffusible catalytic reporters, such as horseradish peroxidase, which often resulted in low signal-to-noise ratios12. In addition, many approaches used long probes targeting single sequences to improve specificity. Although effective in Drosophila embryos and polytene chromosomes, long probes penetrate poorly into the larval brain and often produce uneven labeling10,11,13.
The development of single-molecule fluorescent ISH (smFISH) and hybridization chain reaction fluorescent in situ hybridization (HCR-FISH) addressed many of these limitations14,15. By using sets of short probes tiled across a single gene region, smFISH reduces average probe length from approximately 1,000 nt to approximately 40 nt16. This substantially improves probe penetration while maintaining strong signal intensity. HCR-FISH further improves signal-to-noise ratios by using fluorescently conjugated amplifier probes that increase positional accuracy and amplify signal output. Importantly, this amplification step appears to scale linearly with transcript concentration, enabling quantitative measurement of gene expression levels17.
HCR-FISH begins with initiator probes complementary to the target sequence. These unlabeled DNA oligonucleotides contain a single-stranded initiator sequence that binds to the input domain of a fluorescently labeled H1 amplifier probe. H1 amplifiers adopt a hairpin structure that maintains catalytic inactivity until binding to the initiator sequence. Upon binding, the H1 amplifier opens and exposes a sequence that binds to a partner probe, H2. This interaction opens the H2 hairpin, allowing it to bind another H1 probe (Figure 1)15. Repeated cycles of H1/H2 binding generate a stable polymer that amplifies the fluorescence signal while reducing background noise. Because each initiator sequence is specific to a corresponding H1/H2 pair, multiple probes with distinct initiator sequences and fluorophores can be used simultaneously to label several genes within the same sample.

Figure 1: Graphical representation of the HCR-FISH workflow. DNA probes containing initiator sequences hybridize to a target RNA transcript. Fluorescently labeled DNA amplifier hairpins bind to the initiator sequence and undergo sequential polymerization to form stable double-stranded amplification chains. Direct fluorophore conjugation improves signal amplification and signal-to-noise ratio. Please click here to view a larger version of this figure.
Recent improvements to HCR-FISH have further increased its utility. Split initiator probe design improves amplifier specificity, and secondary antibodies conjugated to HCR initiators enable simultaneous immunohistochemistry and HCR-FISH18. In addition, expanded fluorophore options now permit multiplexed imaging of up to 10 genes simultaneously19,20.
Although initial proof-of-concept experiments for HCR-FISH were performed in zebrafish (Danio rerio), the method has gained widespread use across many systems, particularly in non-model organisms where antibodies are expensive or unavailable15. Probe synthesis also offers a faster alternative to antibody generation, which can require months of development. Consequently, HCR-FISH provides a rapid approach for validating candidate genes identified through single-cell RNA sequencing datasets21. Even in established genetic model organisms, HCR-FISH provides a valuable method for spatially mapping newly born neurons when GAL4 reporters or immunohistochemistry fail to adequately label specific cell populations.
This article describes an HCR-FISH protocol optimized for the third instar larval Drosophila optic lobe. The protocol adapts methods previously published in other organisms for use in the Drosophila visual system22. Fixation conditions commonly used for Drosophila immunohistochemistry were optimized for FISH applications. Probe and hairpin concentrations were increased to improve signal intensity and tissue penetration. PCR tubes were used instead of spot plates to reduce reagent consumption and simplify handling of small tissue samples. In addition, wash and hybridization times were shortened to accommodate the smaller size of the Drosophila brain relative to other insect systems commonly used for HCR-FISH.
Representative applications demonstrate how this approach can provide positional information for specific neuronal populations during development. These experiments indicate that certain transcripts, such as abrupt, can be detected earlier with HCR-FISH than with antibody staining or reporter lines. Potential challenges, troubleshooting strategies, and applications of HCR-FISH for studying visual system development in Drosophila are also discussed.