Identification of markers facilitating the study of support cells in adult olfactory sensilla in Drosophila

Abstract

In the Drosophila antenna, olfactory sensilla are the fundamental unit of odor detection and processing. Sensilla house one to four olfactory neurons enwrapped by three distinct support cells: thecogen, trichogen, and tormogen cells. While olfactory neurons and their odor receptors have been extensively studied in Drosophila, and increasingly in other insect species, much less is known about the support cells and their functional roles within the mature sensillar unit. A critical barrier to studying their function has been the lack of validated markers to distinguish and manipulate these cells in adult flies. Here, we systematically identify and characterize tools to differentially label specific olfactory support cells in the antenna, including sv^T2A-GAL4 as a broad support cell marker. First, we confirm that ASE5-GAL4 labels olfactory tormogen cells, which are also strongly labeled by an anti-Su(H) antibody. Next, we demonstrate that a prospero antibody broadly and exclusively marks thecogen cells, unlike the previously used marker nompA-GAL4. The pros-positive cells are also labeled by a recently developed pros^T2A-GAL4 line, but surprisingly, they are distinct from those labeled by a widely used pros-GAL4 line, which instead labels olfactory neurons. Third, we generate and validate the first reporter line for olfactory trichogen cells, atk^GAL4. Finally, we provide evidence that each tool may function similarly in the maxillary palp. Together, our findings create a framework for distinguishing olfactory support cell classes in the adult fly antenna, a crucial step towards understanding their individual contributions to sensillar function.

Graphical Abstract

1. Introduction

Insects have cuticular hair-like organs known as sensilla, which function as isolated sensory units. Olfactory, gustatory, mechanosensory, thermosensory, and hygrosensory sensilla are distributed across the insect body in stereotypical locations and play indispensable roles in insect behavior. Each sensillum contains sensory neurons as well as three non-neuronal support cells: thecogen, trichogen, and tormogen cells (Hartenstein, 2005; Shanbhag et al., 2000). A large body of research in Drosophila has established that the sensillar cells arise from a Sensory Organ Precursor (SOP) cell, which undergoes successive asymmetric cell divisions regulated by Notch signaling (Hartenstein, 2005; Rodrigues and Hummel, 2008). During development the tormogen cell forms the socket at the base of the sensillum, whereas the trichogen cell secretes the hair shaft. The thecogen cell forms a tight sheath around the sensory neurons. Support cells are identified during development by their stereotypical locations and markers, which can vary by sensory system (Hartenstein, 2005; Rodrigues and Hummel, 2008).

Olfactory sensilla are found on two appendages, the third antennal segment of the antenna and the maxillary palp. In Drosophila, olfactory sensilla are categorized by morphology, which corresponds to distinct odor sensitivities: trichoid sensilla respond to pheromones, basiconic sensilla respond to food odors, and coeloconic sensilla respond to acids and amines (Su et al., 2009; Vosshall and Stocker, 2007). Much research over the past few decades has revealed the molecular processes underlying the development and function of olfactory neurons in olfactory sensilla (Benton et al., 2025). We have a nuanced understanding of how odors that enter the sensillum lymph through pores in the shaft cuticle activate odor receptors on olfactory neuron dendrites, initiating neuron depolarization and action potential firing followed by the transfer of information to stereotyped antennal lobe glomeruli in the brain. In Drosophila, most olfactory neurons are developmentally programmed to express one odor receptor whose response to large odor panels has been characterized (Hallem and Carlson, 2006; Silbering et al., 2011; Talross and Carlson, 2025). Different functional classes of sensilla contain 1-4 olfactory neurons with stereotyped receptor pairings (Benton et al., 2025; Couto et al., 2005; Fishilevich and Vosshall, 2005; Silbering et al., 2011). Studies on the function, evolution, and behavioral relevance of different classes of olfactory neurons have been immeasurably aided by the ability to specifically label and manipulate each neuron class using GAL4 reporter lines for the odor receptor that uniquely labels each neuron population.

In contrast, we have a rudimentary understanding of how the three support cells co-housed in each olfactory sensillum contribute to odor signaling. In Drosophila olfactory sensilla, the thecogen cell neighbors olfactory neurons and enwraps their somata and inner dendrites. At the level of the inner dendrites, the trichogen cell enwraps the thecogen cell, while the tormogen cell surrounds the trichogen cell at the base of the sensillar shaft (Nava Gonzales et al., 2021; Shanbhag et al., 2000) (Fig. 1A). The tormogen and trichogen cell plasma membranes form numerous microvilli and microlamellae, suggesting they may be important for transport or secretion (Nava Gonzales et al., 2021; Shanbhag et al., 2000). These cells are thought to support olfactory neuron function by secreting Odorant Binding Proteins (OBPs), producing putative Odorant Degrading Enzymes (ODEs), and regulating ion concentrations in the sensillar lymph (Jain et al., 2024; Larter et al., 2016; Leal, 2013; Prelic et al., 2021; Shanbhag et al., 2000; Vogt, 2003). However, functional studies on the distinct roles of thecogen, trichogen, and tormogen cells in olfaction have been limited by the lack of validated genetic markers and tools to manipulate these cells in mature flies. Markers for these support cell types have been identified in other types of sensory sensilla and early in olfactory system development, but their specificity and expression in the adult antenna remain poorly defined. As a result, the three types of support cells in olfactory sensilla have often been considered as a monolithic group (“support cells”) in adult flies when considering these functions.

Fig. 1.

ASE5-GAL4 labels antennal tormogen support cells across sensillar classes. (A) Schematic of an olfactory sensillum showing two olfactory receptor neurons (orange and red) with branched dendrites protruding into the sensillar shaft. The neurons are surrounded by thecogen (pink), trichogen (blue), and tormogen (green) support cells. Epithelial cells (grey) occupy the space between sensilla. (B) Single-nucleus RNA sequencing data from the Fly Cell Atlas (Li et al., 2022) visualized with SCope (Davie et al., 2018) showing t-SNE plot of nuclei from the adult Drosophila antenna (10x stringent dataset), containing auditory and olfactory segments. Expression of elav (green) marks neuronal clusters, and expression of sv (magenta) marks support cells clusters. (C) Similar to (B) but expression of Su(H) (magenta). (D) Antennal section from a fly in which ASE5-GAL4 drove expression of mCherry.NLS (red) immunostained with anti-Su(H) (green) and anti-elav (magenta). ASE5-positive cells express Su(H), but not elav. (E) Antennal section from a fly in which sv^T2A-GAL4 drove mCherry.NLS (red) stained with anti-Su(H) (green) and anti-elav (magenta). Arrows identify support cells that express Su(H) and mCherry.NLS, but not elav. Arrowheads identify neurons that express Su(H) and elav, but not mCherry.NLS. (F) Antennal section from a fly in which ASE5-GAL4 drove mCherry.NLS (magenta). (G)–(K) Antennal sections from flies in which mCD8::RFP (magenta) was driven by ASE5-GAL4. Representative examples are shown of staining near (G) large basiconic sensilla, (H) small basiconic sensilla, (I) trichoid sensilla, (J) coeloconic sensilla, and (K) sensilla in sacculus chambers II and III. Ir8a-positive olfactory neurons in coeloconic sensilla and sacculus chamber III were identified by anti-Ir8a expression (green) in (J) and (K). The DIC channel is overlaid to reveal cuticular outlines for sensillar identification. (L) Fly Cell Atlas snRNA-seq expression of basiconic support cell marker Obp28a (blue), trichoid support cell marker lush (green), and coeloconic and sacculus support cell marker a10 (red). Scale bars, (D) and (E) 10 μm, (F) 40 μm, (G)-(K), 5 μm.

Recent single nucleus RNA sequencing (snRNA-seq) data from the Fly Cell Atlas show that neurons (marked by elav) form distinct clusters from support cells (marked by shaven (sv)) (Li et al., 2022; Scalzotto et al., 2022) (Fig. 1B). However, there are more than ten support cell clusters, and it is unclear which clusters correspond to each of the three support cell classes. Therefore, we aimed to systematically identify and validate specific genetic markers for antennal olfactory thecogen, trichogen, and tormogen cells. Our work provides a firm foundation for future studies on the differential olfactory functions of these enigmatic sensillar cells.

2. Methods

2.1. Drosophila stocks

Drosophila melanogaster fly stocks were reared on standard cornmeal-molasses-agar food in a 25°C incubator with a 12:12 hour light/dark cycle. ASE5-GAL4 (Barolo et al., 2002) and nompA-GAL4 (Todi et al., 2005) were obtained from Craig Montell. Flies with the UAS-mCD8::GFP transgene on either chromosome II or III (Lee and Luo, 1999) were obtained from John Carlson. The following lines were obtained from the Bloomington Drosophila Stock Center: #78901 TI{CRIMIC.TG4.2}sv^{CR00370-TG4.2} (Lee et al., 2018), #80572 pros-GAL4 (Shiga et al., 1996), #78970 TI{CRIMIC.TG4.2}pros^{CR00631-TG4.2} (Lee et al., 2018), #66679 Obp19d-GAL4 (Shanbhag et al., 2001b), #38425 P{UAS-mCherry.NLS}2, #38424 P{UAS-mCherry.NLS}3, #8546 P{UAS-RedStinger}4, #8545 P{UAS-RedStinger}3, #32218 P{10XUAS-IVS-mCD8::RFP}attP2 and #32219 P{10XUAS-IVS-mCD8::RFP}attP40. GAL4 reporter lines for atk and Osi23 were generated as described below. Genotypes of fly lines used in this study are listed in Table S1.

2.2. Generation of atk^GAL4 and Osi23^GAL4

Knock-in atk^GAL4 and Osi23^GAL4 transgenic lines were designed and generated by WellGenetics Inc. with CRISPR/Cas9 genome engineering using homology directed repair procedures that were modified from (Kondo and Ueda 2013). For targeting atk, upstream GAAGATGGGCAGTAACATCA and downstream AAGAAACGAGACTTCAACGG gRNA sequences were cloned into U6 promoter plasmids. For targeting Osi23, upstream GCGAACAATGCGTGTTCCGT and downstream AGTGGCGCCCACTTTGGATA gRNA sequences were cloned. For each gene, cassette GAL4-3xP3-RFP, which contains a ribosome binding site, GAL4, SV40 3’UTR, and a floxed 3xP3-RFP, along with two homology arms were cloned into pUC57-Kan as a donor template for repair. The gRNA plasmids, donor template plasmid, and hs-Cas9 were microinjected into embryos of control strain w¹¹¹⁸. Successful integration was determined by ocular RFP expression, genomic PCR, and sequencing. To prevent potential interference with histological experiments, the 3xP3-DsRed cassette was excised using Cre/loxP recombination and validated with sequencing for both lines. In the atk^GAL4 and Osi23^GAL4 lines, >95% of the coding sequences were replaced by cassette GAL4-3xP3-RFP; therefore, heterozygous atk^GAL4 and Osi23^GAL4 flies were used for histology experiments.

2.3. Antennal immunohistochemistry

Equal numbers of male and female flies of a single genotype were aged 6-8 days old and placed in an alignment collar. Their heads were encased in OCT (Tissue-Tek) in a silicone embedding mold, frozen over dry ice, and separated from their bodies. Blocks of tissue were stored in 1.5 mL microcentrifuge tubes at −80°C until sectioning. A Leica CM3050S cryostat was used to collect 20 μm thick tissue sections at −20°C. The following steps were then performed at room temperature unless otherwise specified. Tissue sections were fixed for 10 minutes in a solution of 1X PBS containing 4% paraformaldehyde. Residual paraformaldehyde was washed away with three 5-minute washes in 1X PBS. The tissue was permeabilized with a solution of 1X PBS containing 0.2% Triton X-100 (PBST), followed by blocking in PBST with 1% BSA. Primary antibodies were diluted into PBST with 1% BSA, and 200 μL of primary antibody solution was pipetted under bridged coverslips. Primary antibodies were the following: mouse anti-Su(H) (1:50, Santa Cruz Biotechnology, sc-398453, RRID:AB_3086636), mouse anti-prospero (1:50, Developmental Studies Hybridoma Bank, MR1A supernatant, RRID:AB_528440), rat anti-elav antibody (1:10, Developmental Studies Hybridoma Bank, 7E8A10 supernatant, RRID:AB_528218), and guinea pig anti-Ir8a antibody (1:100-250, (Abuin et al., 2011)). Slides were incubated overnight in a 4°C refrigerator.

The following day, remaining antibody solution was tapped onto a Kimwipe, and slides were washed in PBST three times for 10 minutes. Secondary antibodies were diluted 1:500 in PBST with 1% BSA, and 200 μL was pipetted under bridged coverslips, followed by a two-hour incubation in the dark. Secondary antibodies were the following: donkey anti-mouse Alexa Fluor 488 (Invitrogen, A21202, RRID:AB_141607), goat anti-mouse Alexa Fluor Plus 488 (Invitrogen, A32723, RRID:AB_2633275), goat anti-guinea pig Alexa Fluor 488 (Invitrogen, A11073, RRID:AB_2534117), donkey anti-rat Alexa Fluor 488 (Invitrogen, A21208, RRID:AB_2535794), goat anti-rat Alexa Fluor 647 (Jackson ImmunoResearch, 112-605-167 RRID:AB_2338404) and goat anti-rat Alexa Fluor 647 (Invitrogen, A21247, RRID:AB_141778). Excess antibody solution was removed by tapping, and slides were washed in PBST three times for 5 minutes. Slides were dried for 30 minutes in the dark prior to mounting with VectaShield Mounting Medium with DAPI (Vector Labs, #1200-10) or ProLong Diamond Mountant (Invitrogen, #P3965). Stacks of images with a 0.5 μm z-step size were collected for stained antennal sections using a Nikon AXR confocal microscope in the UConn Advanced Light Microscopy Facility. Images were analyzed using ImageJ/FIJI software. Quantitative analysis was done on ten consecutive sections from each antennal stack. All quantification is reported as the mean ± SEM. The FIJI cell counter tool was used to manually label cells during quantification. Each experiment was repeated at least three independent days.

2.4. Imaging endogenous expression of fluorophores

A shortened protocol was used to image atk^GAL4>UAS-mCD8::GFP, Osi23^GAL4>UAS-mCD8::GFP, pros^T2A-GAL4>10XUAS-mCD8::RFP, and Obp19d-GAL4>10XUAS-mCD8::RFP antennae because an antibody was not applied in these experiments. Blocks of tissue were prepared and cryosectioned in the same way as for immunohistochemistry. Then, antennal sections were fixed for 10 minutes in 4% paraformaldehyde in 1X PBS and washed three times for 5 minutes with 1X PBS. Slides were then dried for 30 minutes in the dark prior to mounting. The sections were imaged with a confocal microscope using the same protocol as for immunohistochemistry imaging. Each experiment was repeated at least three independent days.

3. Results

3.1. ASE5-GAL4 is a broad marker of olfactory tormogen cells

Earlier research has established that an enhancer region of the Suppressor of Hairless (Su(H)) transcription factor, known as ASE5, is expressed in tormogen cells of mechanosensory bristles in developing and mature Drosophila (Barolo et al., 2002; Barolo et al., 2000). Numerous recent studies have used ASE5-GAL4 as a marker for tormogen cells in antennal olfactory sensilla in adult flies (Ha et al., 2023; Jain et al., 2024; Larter et al., 2016; Prelic et al., 2024; Prelic et al., 2021; Scalzotto et al., 2022); however, the ability of this transgene to specifically label this cell population has not been verified. Fly Cell Atlas snRNA-seq data (Li et al., 2022) indicate that Su(H) is expressed in two olfactory support cell clusters and multiple olfactory neuron populations within the antenna (Fig. 1C), raising the possibility that the antennal ASE5-GAL4 expression may not conform to its mechanosensory bristle profile. We therefore sought evidence that ASE5-GAL4 indeed selectively labels tormogen cells in adult antennal olfactory sensilla. Using a nuclear localized mCherry (UAS-mCherry.NLS), we found that ASE5-GAL4 is expressed in cells that are strongly labeled by an anti-Su(H) antibody and do not express the neuronal marker anti-elav (Fig. 1D and Fig. S1A). We also noted that some elav-positive neurons were weakly labeled by anti-Su(H), consistent with the expression pattern of this gene on the Fly Cell Atlas. Thus, ASE5-GAL4 expression is limited to the non-neuronal Su(H)-positive cells.

We next sought evidence that cells labeled by ASE5-GAL4 are support cells, which can be marked by expression of shaven (sv) (Li et al., 2022). There are seven reported splice variants of the sv gene, and it is unknown which isoforms are expressed in the antenna. Therefore, we obtained a newly developed knock-in CRIMIC line, TI{CRIMIC.TG4.2}sv^{CR00370-TG4.2} (hereafter sv^T2A-GAL4), which was designed to capture expression of all sv isoforms (Lee et al., 2018). As expected, in flies in which UAS-mCherry.NLS is driven by sv^T2A-GAL4, mCherry-expressing cells are widely distributed across the antenna (Fig. S1B) and are not labeled by elav (Fig. S1C). Co-staining antennae from sv^T2A-GAL4>mCherry.NLS flies with both anti-Su(H) and anti-elav revealed that nearly all cells that are both anti-Su(H)-positive and elav-negative (i.e. the ASE5-positive cells described above) express mCherry (95.9% ± 1.5%, n = 8 antennae) (Fig. 1E). These anti-Su(H)-positive and elav-negative cells are 33.7% ± 1.5% (n = 8 antennae) of all mCherry-positive cells, consistent with anti-Su(H) (and therefore ASE5-GAL4) labeling one of three types of sv-expressing support cells in each sensillum. Taken together, our findings support the idea that ASE5-GAL4 labels tormogen support cells in the adult olfactory sensilla, similar to its well-characterized role in other sensillar types.

Next, we wanted to test whether ASE5-GAL4 labels tormogen cells in all morphological classes of olfactory sensilla as suggested by our initial experiments with ASE5-GAL4>mCherry.NLS, in which expression was found broadly across the antenna (Fig. 1F). We therefore examined antennal sections from flies in which ASE5-GAL4 drove expression of a membrane-targeted RFP (10xUAS-mCD8::RFP). In accordance with the known location of tormogen cells, RFP-positive cells were found just under the cuticular surface at the base of olfactory sensilla (Fig. 1G-J). Strong expression was found at the base of morphologically identified basiconic and trichoid sensilla (Fig. 1G-I), with weaker expression at the base of coeloconic sensilla (identified by selective expression of the Ir8a odor receptor in coeloconic neurons) (Abuin et al., 2011; Task et al., 2022) (Fig. 1J). Similarly, weaker RFP expression was seen in sacculus chamber III, which houses sensilla containing Ir8a-positive olfactory neurons in coeloconic sensilla (Abuin et al., 2011; Ai et al., 2013), although robust expression was found in sacculus chambers I and II where hygrosensitive and thermosensitive sensilla are located (Enjin et al., 2016; Knecht et al., 2017; Knecht et al., 2016) (Fig. 1K). Weaker coeloconic expression of ASE5-GAL4 is consistent with snRNA-seq data indicating that only two antennal support cell clusters robustly express Su(H) (Fig. 1C). These two clusters express Obp28a and lush, markers of basiconic and trichoid support cells, respectively (Larter et al., 2016; Park et al., 2000; Shanbhag et al., 2001a) (Fig. 1L). In contrast, Su(H) was not detected in coeloconic support cell clusters, identified by a10 expression (McKenna et al., 1994; Pikielny et al., 1994; Scalzotto et al., 2022), likely because weak coeloconic Su(H) expression is below the detection threshold. Together, this indicates that ASE5-GAL4 broadly labels tormogen cells, but its expression varies in different morphological classes of sensilla.

A recent serial block-face electron microscopy study found large extracellular vesicles in the sensillum shaft of Drosophila basiconic and trichoid sensilla, with particularly high frequency in large basiconic sensilla (Nava Gonzales et al., 2021). Similar extracellular vesicles were also seen in mosquito olfactory sensilla (Charara et al., 2025). From one reconstructed sensillum, it was suggested that these vesicles bud from the plasma membrane of tormogen cells (Nava Gonzales et al., 2021). Intriguingly, we found strong mCD8::RFP signal filling the sensillum shafts of large basiconic sensilla in antenna from ASE5>mCD8::RFP flies (Fig. 1G). We also observed moderate RFP labeling in small basiconic sensillar shafts (Fig. 1H), and weaker labeling of trichoid sensillar shafts (Fig. 1I). In sensilla with less intense labeling, the RFP appeared discontinuous, consistent with membranous vesicles (Fig. 1I). We did not observe mCD8::RFP in the shaft of coeloconic sensilla (Fig. 1J), consistent with electron microscopy findings (Nava Gonzales et al., 2021). This membrane-targeted RFP labeling in basiconic sensilla does not arise from olfactory neuron dendrites because these neurons do not express ASE5-GAL4 (Fig. 1D). Given that tormogen cells do not extend into the sensillar shaft, our data provide additional support for the suggestion that extracellular vesicles can be derived from tormogen cell plasma membranes.

3.2. nompA selectively labels thecogen cells in coeloconic sensilla

We next sought a broad marker of olfactory thecogen cells. Several recent studies have used nompA-GAL4 (Todi et al., 2005) to label olfactory thecogen cells in adult antenna (Ha et al., 2023; Jain et al., 2024; Larter et al., 2016; Prelic et al., 2024; Prelic et al., 2021; Scalzotto et al., 2022) due to its well-established localization to thecogen cells in mechanosensory bristles and similar scolopale cells in the Johnston’s organ (Chung et al., 2001). However, antennal snRNA-seq data from the Fly Cell Atlas indicates that nompA is only expressed in two support cell clusters (Fig. 2A); these clusters likely derive from coeloconic sensilla due to their expression of coeloconic marker a10 (Fig. 2B) (Li et al., 2022; Scalzotto et al., 2022). To directly test this possibility, we stained antenna from flies in which nompA-GAL4 drove UAS-RedStinger, a nuclear-localized red fluorophore, with an anti-Ir8a antibody to identify coeloconic olfactory neurons. Unlike ASE5-GAL4, nompA-GAL4 drives expression relatively sparsely in the adult antenna (Fig. 2C). Further, each of the RedStinger-positive cells is associated with Ir8a-positive neurons (Fig. 2C and D). Together, these data are consistent with nompA-GAL4 exclusively labeling cells in coeloconic sensilla, despite its previous use as a universal olfactory thecogen cell marker.

Fig. 2.

nompA-GAL4 expression is biased toward coeloconic sensilla. (A)–(B) Single-nucleus RNA sequencing data from the Fly Cell Atlas showing t-SNE plot of nuclei from the adult Drosophila antenna showing (A) nompA expression (magenta) predominantly in olfactory sensilla support cell clusters marked by (B) a10 (red, coeloconic and sacculus), but not those labeled by Obp28a (blue, basiconic) or lush (green, trichoid). (C) Antennal section from a fly in which nompA-GAL4 drove RedStinger (magenta) and stained with anti-Ir8a (green). (D) Higher magnification view of (C). Scale bars, (C) 40 μm, (D) 8 μm.

3.3. Prospero is a marker of olfactory thecogen cells in adult flies

We therefore sought a better marker for olfactory thecogen cells in the adult antenna. Prospero (pros) is a transcription factor expressed in thecogen cells of multiple types of chordotonal organs and developmentally similar scolopale cells in the Johnston’s organ (Doe et al., 1991; Zhang et al., 2023). Early in antennal development, pros is expressed in both olfactory thecogen cells and some olfactory neurons (Chai et al., 2019; Endo et al., 2011; Hartl et al., 2011; Sen et al., 2003), whereas pros expression is limited to olfactory thecogen cells during later developmental stages of the maxillary palp, a secondary olfactory organ, (Hartl et al., 2011). We therefore wondered whether pros expression might be limited to thecogen cells in the adult antenna. Examining antennal snRNA-seq data from the Fly Cell Atlas (Li et al., 2022), we saw that pros is expressed in both clusters of a10- and nompA-positive coeloconic thecogen cells (Fig. 3A and B). Two additional support cell clusters express pros, one expressing Obp28a and the other expressing lush, suggesting these clusters contain basiconic and trichoid support cells, respectively. Unlike the developing antenna, robust pros levels are not seen in olfactory neurons, although pros is detected in some Johnston’s organ neurons. Together with previous research, this made pros a strong candidate for a selective marker of olfactory thecogen cells across morphological classes of sensilla in adult flies.

Fig. 3.

anti-Prospero is a marker of olfactory thecogen cells. (A)–(B) Single-nucleus RNA sequencing data from the Fly Cell Atlas showing t-SNE plot of nuclei from the adult Drosophila antenna showing that (A) pros expression (green) if found in four olfactory sensilla support cell clusters marked by (B) a10 (red, coeloconic and sacculus), Obp28a (blue, basiconic) and lush (green, trichoid). Additionally, pros is detected in putative mechanosensory neurons (orange #) and scolopale cells (orange *) from the Johnston’s organ in the second antennal segment. (C) Antennal section from a fly in which RedStinger (magenta) expression was driven by nompA-GAL4 and stained with anti-pros (green). (D) Antennal cryosection immunostained with anti-pros (green) and anti-elav (magenta) revealing distinct cells. (E) Pros immunofluorescence (green) on antennal cryosection in which support cells were labeled with sv^T2A-GAL4>mCherry.NLS (magenta). (F) Cells labeled by anti-pros (green) are distinct from tormogen cells labeled by mCherry.NLS (magenta) driven by ASE5-GAL4. Scale bars, (C) 30 μm, (D)-(F) 10 μm.

To examine this possibility, we first immunostained antenna in which nompA-GAL4 drove expression of UAS-RedStinger with an anti-pros antibody (Spana and Doe, 1995). We found that anti-pros broadly labels nuclei across the antenna, a subset of which express nompA (Fig. 3C). Staining antennae with both anti-pros and anti-elav revealed that few of the anti-pros-positive nuclei in the third antennal segment express elav (3.0% ± 0.7%, n = 10 antennae) (Fig. 3D), in agreement with the snRNASeq data. Occasionally, cells with weak anti-pros labeling exclusively outside the nucleus were seen; such cells were invariably elav-positive neurons. This staining pattern may indicate non-specific antibody labeling; we therefore focused further analysis on the anti-pros-positive nuclei. Immunostaining antennae from sv^T2A-GAL4>UAS-mCherry.NLS flies with anti-pros revealed that nearly all pros-positive nuclei also express the support cell marker sv (88.3% ± 1.6%, n = 9 antennae) (Fig. 3E). Consistent with anti-pros labeling one in three support cells broadly across sensillar classes, 38.2% ± 1.1% (n = 9 antennae) of the mCherry-positive cells were anti-pros positive. Finally, we confirmed that the pros-expressing nuclei are distinct from those expressing the tormogen cell marker ASE5 (Fig. 3F). Taken together, our data are consistent with pros expression being restricted to thecogen cells in adult antennae, similar to its expression in later stages of the developing maxillary palp. In contrast, few olfactory neurons maintain pros expression after early pupal development.

We next sought a pros reporter line to express RFP in the plasma membrane of the putative thecogen cells to examine their morphology. We initially tested a commonly used pros-GAL4 line (Shiga et al., 1996) by using this driver to express mCherry.NLS. Surprisingly, the mCherry-positive cells were distinct from those labeled by anti-pros, but instead were labeled by anti-elav (Fig. 4A); this indicated that pros-GAL4 primarily labels neurons in this tissue. We considered that pros-GAL4 may not accurately report pros expression in this tissue and developmental time point because there are six reported pros transcripts, and the exact sequence used to generate the pros-GAL4 line has not been described (Shiga et al., 1996). We therefore assessed a newly developed reporter, TI{CRIMIC.TG4.2}pros^{CR00631-TG4.2} (hereafter, pros^T2A-GAL4), a transgenic line from the CRIMIC collection (Lee et al., 2018). In the CRIMIC reporter system, a cassette encoding a splice acceptor and T2A-GAL4 is precisely inserted between coding exons of an endogenous gene, allowing for accurate reporting of the gene’s expression pattern while simultaneously truncating the protein. The pros^T2A-GAL4 transgene is inserted into an intron shared by all known splice variants and thus should function as a universal reporter of pros expression. When crossed to UAS-mCherry.NLS, the pros^T2A-GAL4 driver labels 89.0% ± 3.1% (n = 10 antennae) of the anti-pros-positive nuclei, and conversely, most of the mCherry-positive cells are labeled by anti-pros (82.3% ± 1.9%, n = 10 antennae) (Fig. 4B). A minority of the cells labeled by pros^T2A-GAL4 express elav (15.4% ± 2.1%, n = 10 antennae) (Fig. 4B). Thus, pros^T2A-GAL4 can generally be used to gain genetic access to the same cells as those labeled by anti-pros, and it is a more faithful reporter of pros expression in the adult antenna than the pros-GAL4 reporter.

Fig. 4.

Differential expression of two prospero driver lines. (A) Antennal section from a fly in which pros-GAL4 drove mCherry.NLS (magenta) immunostained with anti-pros (green) and anti-elav (cyan). Most mCherry-expressing cells express elav, but not pros. (B) Antennal section from a fly in which pros^T2A-GAL4 drove mCherry.NLS (magenta) immunostained with anti-pros (green) and anti-elav (cyan). Most mCherry express pros, with a minority expressing elav. (C) Schematic of a thecogen cell (pink) ensheathing the soma and inner dendrites of two olfactory neurons (grey). (D)-(F) Antennal cells labeled by mCD8::RFP (magenta) driven by pros^T2A-GAL4 have a morphology consistent with thecogen cells and are found at the base of (D) trichoid, (E) large basiconic, and (F) coeloconic sensilla (white arrowheads). The DIC channel is overlaid to reveal cuticular outlines for sensillar identification. Scale bars, (A) and (B) 10 μm, (D)-(F) 5 μm.

When used to drive mCD8::RFP, we found that pros^T2A-GAL4 is expressed in cells that are located further from the cuticular surface compared to the ASE5-expressing tormogen cells, consistent with the relative locations of these cells within the sensillar unit (Figs. 4C-F and 1A). Additionally, these RFP-positive cells taper towards the base of olfactory sensilla (Fig. 4D-F), and thus morphologically resemble thecogen cells, further confirming their identity. RFP-positive cells were found at the base of morphologically identified trichoid (Fig. 4D), basiconic (Fig. 4E) and coeloconic (Fig. 4F) sensilla. Taken together, our data demonstrate that pros is broadly expressed in mature olfactory thecogen cells.

3.4. Artichoke is a selective marker for olfactory trichogen cells

A marker for olfactory trichogen cells in adult flies has not been previously identified, preventing functional studies of these cells (Fig. 5A). Literature searches revealed two candidate genes that could serve as markers for these cells. Osiris23 (Osi23) is expressed by olfactory trichogen cells during pupal development and is necessary for the formation of pores within the sensillum shaft (Ando et al., 2019). Artichoke (atk) is expressed in the trichogen cells of type I embryonic sensory organs, including chemosensory sensilla organs (Andres et al., 2014). However, neither gene is detected with either single nuclear RNA-sequencing (Fig. 5B) (Li et al., 2022) or bulk transcriptome sequencing of the adult antenna (Menuz et al., 2014). However, both genes are expressed in a recent snRNA-seq analysis of antennal sensilla in pupal development (Mermet et al., 2025). In this dataset, Osi23 and atk are expressed in support cells labeled by sv, with Osi23-expressing cells forming a subset of the broader atk-positive cells (Fig. S2A-C). At this developmental time, atk-expressing clusters are largely distinct from those expressing Su(H) or pros as well as the glial maker repo (Fig. S2D-F). We therefore decided to investigate Osi23 and atk as potential trichogen cell markers and generated knock-in GAL4 lines for each gene (Fig. 5C-D). When crossed to a mCD8::GFP reporter, Osi23^GAL4 drove weak expression in the adult antenna (Fig. 5C), whereas atk^GAL4 drove strong expression in most antennal regions (Fig. 5D). We therefore focused on determining the identity of the antennal cells that express atk^GAL4.

Fig. 5.

Identification of atk^GAL4 as a reporter for antennal trichogen cells. (A) Schematic of a trichogen cell (blue) ensheathing two olfactory neurons (grey). (B) Candidate trichogen markers Osi23 (magenta) and atk (green) are not detected in a tSNE plot of single-nucleus transcriptomes of the adult antenna from the Fly Cell Atlas. (C) Above, schematic showing the generation of a knock-in GAL4 reporter lines for Osi23. Cassette integration was identified by 3xP3-RFP, which was removed through Cre recombinase excision using flanking loxP sites (orange). See Methods for details. Below, antennal section from a fly in which mCD8::GFP was driven by Osi23^GAL4 showing weak antennal labeling. (D) Similar to (C), but for atk. Strong antennal labeling of the antenna is seen, except in the region containing large basiconic sensilla (*). (E)-(G) Antennal sections from flies in which atk^GAL4 drove expression of mCherry.NLS (magenta) and immunostained with (E) anti-elav, (F) anti-Su(H), or (G) anti-pros. (H)-(M) Antennal sections from flies in which mCD8::RFP (magenta) was driven by atk^GAL4. Representative examples are shown of staining near (H) large basiconic sensilla, (I) small basiconic sensilla, (J) trichoid sensilla, (K) coeloconic sensilla, and (L) sensilla in sacculus chambers II and III. Ir8a-positive olfactory neurons in coeloconic sensilla and sacculus chamber III were identified by anti-Ir8a expression (green) in (K)-(M). The DIC channel is overlaid to reveal cuticular outlines for sensilla (arrowheads) in (H)-(K). Counterstaining with DAPI (blue) in (M) showing a coeloconic sensillum containing two cells (arrowheads) labeled by atk^GAL4>mCD8::RFP. (N) Schematic of a coeloconic sensillum showing four support cell populations ensheathing two olfactory neurons. Scale bars, (C)-(D) 30 μm, (E)-(L) 8 μm, (M) 3 μm.

We first tested whether the cells labeled by atk^GAL4 are distinct from other types of antennal cells. First, we immunostained antennal sections from flies in which atk^GAL4 drove expression of UAS-mCherry.NLS. We found that atk^GAL4 labeled cells distinct from neurons labeled with anti-elav (Fig. 5E), tormogen cells marked by anti-Su(H) (Fig. 5F), and thecogen cells labeled with anti-pros (Fig. 5G), consistent with the atk-expressing clusters in the pupal antenna (Mermet et al., 2025) (Fig. S2). We next examined the localization and morphology of atk-expressing cells by using atk^GAL4 to drive expression of mCD8::RFP. The RFP-positive cells were found directly under olfactory sensilla shafts (Fig. 5H-J), consistent with expression in support cells rather than in epithelial cells. Taken together with the previously reported expression of atk during development, our data indicate that atk is a marker of trichogen cells in the adult antenna.

Next, we determined whether atk is expressed in all morphological classes of olfactory sensilla. We found RFP expression at the base of large and small basiconic sensilla (Fig. 5H and I), trichoid sensilla (Fig. 5J), coeloconic sensilla (identified by Ir8a-positive coeloconic neurons) (Fig. 5K), and sensilla in the sacculus (Fig. 5L). We observed that atk-driven mCD8::RFP expression was weakest at the base of basiconic sensilla, consistent with the relatively low expression of mCD8::GFP driven by atk^GAL4 in the proximal medial region of the antenna, which predominantly contains large basiconic sensilla (Fig. 5D). Interestingly, we observed that some sensillar shafts were labeled by mCD8::RFP when driven by atk^GAL4 (Fig. 5H-J). This is unlikely to be due to expression of RFP in either neurons or tormogen cells, as atk^GAL4 does not label either cell type (Fig. 5E and F). This data may suggest that in addition to tormogen cells, trichogen cells can also release membranous vesicles into the sensillum shaft.

Previous electron microscopy studies have revealed that coeloconic sensilla have four support cells, unlike the three found in basiconic and trichoid sensilla (Nava Gonzales et al., 2021; Shanbhag et al., 2000). Based on their morphology, it was suggested that the fourth cell was either an extra tormogen cell (Shanbhag et al., 2000) or an extra trichogen cell (Nava Gonzales et al., 2021). Close examination of an isolated coeloconic sensillum revealed two atk-positive trichogen cells enwrapping the Ir8a-positive coeloconic neuron (Fig. 5M), supporting the latter interpretation (Fig. 5N).

3.5. Obp19d-GAL4 as an epithelial cell marker

Although our primary focus was to identify tools to label support cells in mature olfactory sensilla, we also examine whether an Obp19d-GAL4 line (Shanbhag et al., 2001b) specifically labels one of the other frequently encountered antennal cell types, epithelial cells. Such cells are not directly associated with sensilla, but instead separate sensilla from each other (Fig. 6A). Previous electron microscopy studies used an anti-Obp19d antibody to reveal that Obp19d is predominantly found in epithelial cells in the antenna (Park et al., 2000; Shanbhag et al., 2001a). This expression pattern is supported by snRNA-seq from the Fly Cell Atlas, which confirms that Obp19d-expressing cells are distinct from sv-expressing support cells and elav-positive neurons (Fig. 6B). An existing Obp19d-GAL4 line was reported to be expressed in all chemosensory tissues, including antennae, maxillary palps, labella, and legs (Shanbhag et al., 2001b), but to our knowledge, its labeling of antennal epithelial cells has not been closely examined. We first confirmed that Obp19d-GAL4 drives UAS-mCherry.NLS broadly throughout the antenna (Fig. 6C). When Obp19d-GAL4 was used to express mCD8::RFP, the labeled cells had a more columnar morphology than support cells, and they were found in between sensilla rather than directly underneath the sensillar shaft (Fig. 6D). Next, we examined antennal sections from flies in which Obp19d-GAL4 drove UAS-mCherry.NLS expression to examine more directly whether these cells are distinct from other antennal cell types. Our histology revealed that the mCherry-positive cells are not elav-positive neurons (Fig. 6E), Su(H)-positive tormogen cells (Fig. 6F), or pros-positive thecogen cells (Fig. 6G). Thus, Obp19d-GAL4 can be used as a marker of antennal epithelial cells.

Fig. 6.

Obp19d-GAL4 labels antennal epithelial cells. (A) Schematic of an olfactory sensillum showing epithelial cells (grey) surrounding the sensillar support cells and neurons (colored). (B) Single-nucleus RNA sequencing data from the Fly Cell Atlas showing t-SNE plot of nuclei from the adult Drosophila antenna with distinct cell clusters labeled by Obp19d (epithelial cells), nSyb (neurons), and sv (support cells). (C) Antennal section from a fly in which mCherry.NLS (magenta) was driven by Obp19d-GAL4. (D) Antennal section from a fly in which mCD8::RFP (magenta) was driven by Obp19d-GAL4 revealing cells with a columnar morphology adjacent interspersed between the base of sensilla. (E)-(G) Antennal sections from flies in which Obp19d-GAL4 drove expression of mCherry.NLS (magenta) and immunostained with (E) anti-elav, (F) anti-Su(H), or (G) anti-pros. Scale bars, (C) 30 μm, (D)-(G) 10 μm.

3.6. Expression in the maxillary palp

Drosophila flies have a secondary olfactory organ, the maxillary palp. We wondered if the tools that we identified in the antenna might also be useful in this tissue. First, we found that the support cell reporter sv^T2A-GAL4 drives mCherry.NLS in cells distinct from neurons in sections of the maxillary palp (Fig. 7A). We also found that support cell markers anti-pros (Fig. 7B), pros^T2A-GAL4 (Fig. 7C), atk^GAL4 (Fig. 7D), and ASE5-GAL4 (Fig. 7E) label a subset of non-neuronal cells in the maxillary palp. Finally, we also found that Obp19d-GAL4 labels cells distinct from anti-elav positive neurons in this tissue (Fig. 7F). Together, these data provide initial support for the ability of these tools to label maxillary palp cell populations for olfactory studies.

Fig. 7.

Olfactory support cell markers are expressed in the maxillary palp. (A) Section of maxillary palps from flies in which mCherry.NLS (magenta) was driven by sv^T2A-GAL4 and immunostained with anti-elav (green). (B) Maxillary palp immunostained with anti-elav (green) and anti-pros (red). (C)-(F) Similar to (A), but with (C) pros^T2A-GAL4, (D) atk^GAL4, (E) ASE5-GAL4, or (F) Obp19d-GAL4. Scale bars, 10 μm.

4. Discussion

This study provides a validated genetic framework for identifying and manipulating the three classes of support cells (tormogen, thecogen, and trichogen cells) within the mature Drosophila antennal olfactory system (Fig. 8A). We have systematically characterized reporter lines and antibodies that can be used with high specificity to label individual classes of olfactory support cells in adult tissues. Our data also suggest that such tools may also function in the maxillary palp, a secondary olfactory appendage. The identification of these tools will allow researchers to overcome previous barriers to studying the contributions of sensillar support cells to olfaction.

Fig. 8.

Support cell markers for antennal olfactory sensilla in adult flies. (A) Schematic summarizing the best identified markers for broad labeling each type of olfactory support cell in the adult antenna. (B) Inferred identities of olfactory support cells clusters in the Fly Cell Atlas antennal dataset based on cell-type markers including a10 (coeloconic and sacculus), Obp28a (basiconic), lush (trichoid), and sv (support cells) depicted in previous figures. Thecogen cells were identified by pros expression and tormogen cells by Su(H) expression. The third remaining basiconic and trichoid clusters were inferred to be trichogen cells. Coeloconic tormogen and trichogen cells could not be distinguished.

Our results necessitate a re-evaluation of some markers previously used to study support cells. Although our data confirm that ASE5-GAL4 can be used to label tormogen cells, our histology unambiguously shows that nompA-GAL4 is exclusively expressed in coeloconic sensilla despite its previous use as a general marker of olfactory thecogen cells (Ha et al., 2023; Jain et al., 2024; Larter et al., 2016; Prelic et al., 2024; Prelic et al., 2021; Scalzotto et al., 2022). Instead, our work shows that an anti-pros antibody labels olfactory thecogen cells in all morphological classes of olfactory sensilla and that these cells can be genetically accessed with pros^T2A-GAL4. Unexpectedly, an older, commonly used pros reporter line, pros-GAL4, inappropriately labels anti-pros negative olfactory neurons, highlighting the superior fidelity of the newer knock-in pros^T2A-GAL4 CRIMIC line for targeting this cell population.

We have also generated and validated the first tool, atk^GAL4, to label and manipulate olfactory trichogen cells in adult flies. Using this tool, we resolved the identity of the “extra” support cell in coeloconic sensilla, which possess four support cells (rather than the typical three) according to previous electron microscopy studies (Nava Gonzales et al., 2021; Shanbhag et al., 2000). Our observation of two atk-positive cells enwrapping Ir8a neurons confirms that the "extra" support cell in these sensilla is indeed a second trichogen cell.

Although the Fly Cell Atlas is a useful resource for identifying many types of cells in the antenna, shaven-positive clusters were grouped as “support cells” (Li et al., 2022). Using markers for trichoid, basiconic and coeloconic olfactory sensilla, we could assign clusters to specific morphological classes of sensilla (Fig. 1L). Combining this knowledge with expression of the thecogen cell marker pros expression (Fig. 3A) allowed identification of two clusters of coeloconic or sacculus thecogen cells and one cluster each of thecogen cells in basiconic and trichoid sensilla (Fig. 8B). We could also use Su(H) expression (Fig. 1B) to identify one cluster each of basiconic and trichoid tormogen cells (Fig. 8B). We could then infer that the unassigned third basiconic and trichoid clusters are very likely trichogen cells (Fig. 8B), despite the inability to detect atk (Fig. 5B). However, due to the low expression of Su(H) in coeloconic and sacculus sensilla, tormogen and trichogen cells in these sensilla could not be distinguished (Fig. 8B).

Despite the utility of these tools, some limitations remain. Although each tool is expressed in each sensillar morphological class, we observed varying expression levels across morphological classes of sensilla. For example, ASE5-GAL4 and atk^GAL4 show weaker expression in coeloconic and basiconic sensilla, respectively. Additionally, we have only examined expression in mature flies, and these markers may exhibit dynamic expression patterns during development that potentially label additional or fewer cell populations. We may see evidence of this with pros, where snRNA-seq data indicates that it is additionally expressed by some neuron populations in the developing antenna (Fig. S2) (Li et al., 2022), consistent with previous reports (Chai et al., 2019; Endo et al., 2011; Hartl et al., 2011; Sen et al., 2003). This likely explains why anti-pros predominantly labels thecogen cells in the mature antenna, but pros^T2A-GAL4 additionally drives sporadic fluorophore expression in some neurons; the fluorophore may perdure longer than the pros protein. To use pros^T2A-GAL4 to gain genetic access to olfactory thecogen cells for functional experiments, it should be combined with a tool such as the GAL4 repressor elav-GAL80 to avoid off-target neuronal labeling (Rideout et al., 2010; Yang et al., 2009). Fluorophore perdurance may also explain our ability to detect atk with atk^GAL4, which is detectable in snRNA-seq of the developing antenna, but not in adults (Figs. 5B and S2C) (Li et al., 2022; Mermet et al., 2025). A practical limitation of the weak or absent atk expression in the adult antenna is that atk^GAL4 may be insufficient to drive adult-stage UAS-RNAi constructs to answer functional questions regarding trichogen cell physiology. Finally, we also emphasize that while these markers and tools are effective in the antenna and likely in the maxillary palp, they are unlikely to be universal markers for support cells in other sensory systems. For example, data from the Fly Cell Atlas indicates that pros is highly expressed in some Johnston’s organ neurons (Fig. 3A and B).

Support cells are thought to carry out important functions within olfactory sensilla, including secretion of Odorant Binding Proteins, degradation of odorant molecules, and regulation of sensillar lymph ions. Our identification of tools to individually label and manipulate antennal olfactory thecogen, trichogen, and tormogen cells opens new avenues to a more comprehensive understanding of how the entire sensillar unit —neurons and support cells alike— detects and processes olfactory information.

Supplementary Material

1

Fig. S1. sv broadly labels antennal support cells. (A) Antennal section from a fly in which ASE5-GAL4 drove expression of mCherry.NLS (magenta) and immunostained with anti-elav (green). Note this is the same image as in Fig. 1D but shown here with just two channels for clarity. (B) Antennal section from a fly in which sv^T2A-GAL4 drove expression of mCherry.NLS (magenta). (C) Similar to (A), but a sv^T2A-GAL4 instead of ASE5-GAL4. Scale bars, (A) and (C) 10 μm, (B) 30 μm.

2

Fig. S2. Expression of candidate trichogen markers during development. (A)-(F) Single-nucleus RNA sequencing data visualized with SCope (Davie et al., 2018) showing a UMAP plots of sensillar nuclei from the developing Drosophila antenna (broad developing antenna dataset), which contains auditory and olfactory neurons, but few epithelial cells (Mermet et al., 2025). (A) The support cell marker sv (magenta) and neuronal marker elav (green) are expressed in largely distinct clusters. (B) Osi23 (magenta) labels a subset of support cells (A) that express (C) atk (magenta). (D)-(F) Cells expressing atk (magenta) are found in distinct clusters from those that express (D) Su(H) (green), (E) pros (green), or the glial marker repo (green).

• Identification of markers for support cells in adult Drosophila olfactory sensilla

• Prospero, but not nompA, is expressed broadly in thecogen cells

• A new artichoke driver line is the first tool to label adult trichogen cells

• Tormogen cells confirmed to be selectively labelled by ASE5-GAL4