Extended Data Fig. 2: Cell-type-specific expression levels of all CSP pairs across the antennal lobe.

a, mRNA expression levels of the three CSP pairs at 24–30 h APF in ORNs and PNs that target 10 glomeruli (DA1, DA4l, DL1, DL3, DC3, VA1d, VA1v, DM2, VM3, and VA2) from single-cell RNA sequencing data^3,4. The unit for x and y axis is log₂(CPM + 1). CPM: counts per million reads. Besides VA1d and VA1v, several other glomeruli relevant to experiments described in this study are indicated. b, Confocal images showing neuropil staining by N-cadherin antibody (blue) and lack of Myc staining of tagged endogenous CSPs (cyan for Toll2, Fili, Hbs, and Sns; red for Ptp10D, Kek1, and Kirre) without FLP. Annotation (+) indicate signal amplification for Myc staining. c, Confocal images showing neuropil staining by N-cadherin antibody (blue) and Myc staining of tagged endogenous CSPs (red for Kek1 and Kirre; cyan for Fili, Hbs, and Sns) using ORN-specific FLP (top tow) or PN-specific FLP (bottom row). The VA1d (white) and VA1v (yellow) glomeruli are outlined based on N-cadherin staining. Samples were chosen based on (1) their VA1d vs. VA1v preference indexes are close to the average, and (2) both VA1d and VA1v glomeruli occupy similar sizes in these single optic sections. Scale bars = 10 µm. Data from the above panels, along with data from Fig. 1c,d, contribute to the quantification of expression (Fig. 1e–g) and simplified schematic summary (Fig. 1h–j). Note that although Fili’s expression in PNs was undetectable using the conditional tag, a previous study showed that Fili exhibits higher expression in VA1d-PNs than VA1v-PNs using immunostaining of Fili antibody and cell-type-specific expression pattern using intersection of ORN- or PN-FLP and Fili-GAL4²⁷. And in DA1 glomerulus, Fili appears to be expressed in a small portion of DA1-PN dendrites neighboring the VA1d glomerulus²⁷. For Kirre–Hbs/Sns, expression of Hbs and Sns in ORNs are not detectable, so we did not draw their expression in Fig. 1j. We did not draw the Kirre expression level in VA1d- or VA1v-PNs (Fig. 1j) given its preference index is highly variable (Fig. 1g). We also note that the differential expression patterns of mRNAs and proteins are largely consistent (Fig. 1e–g). Occasional discrepancies could be caused by (1) post-transcriptional regulations (e.g., protein translation, stability) and (2) different time windows from which mRNA (24–30 h APF) and protein (42–48 h APF) data were collected. Ideally, protein staining should be done around 30 h APF when synaptic partner matching initiates. However, as glomeruli have not formed at that stage, we could not distinguish cell types in which proteins are expressed. 42–48 h is the earliest window we could use glomerular identity to infer cell-type-specific expression. It is possible that the expression of some of the CSPs for synaptic partner matching is already downregulated by then. Although protein expression data is more directly relevant to the action of these genes, mRNA expression data is more temporally relevant, and thus these data provide complementary information.

Source Data