Different sensillum classes and their representation across SBEM volumes

(A) Schematic of antennal regions sampled in each SBEM volume analyzed in this study (pink-shaded areas). Inset: the largest volume, showing 3D reconstructions of the antenna and sensillum cuticles for basiconic (yellow), coeloconic (orange), intermediate (blue), and trichoid (teal) sensilla. Scale bar, 15 µm. (B) Representative 3D models of the major morphological classes of olfactory sensilla and their associated ORNs. Scale bar, 2 µm. (C) Proportional representation of sensillum classes across SBEM volumes. In volumes 2–4, only a subset of sensilla within each volume was sampled and analyzed. The total number of analyzed sensilla is indicated above each bar.

Morphological comparison of EVs and NVEPs

(A–B) Representative SBEM images of EVs (A) and NVEPs (B), arranged in ascending order of diameter. The scale bar shown in the first panel applies to all images. Scale bar, 250 nm. (C) Distributions of EV and NVEP diameters. Curves represent kernel density estimates. Median diameters are 177 nm for EVs and 111 nm for NVEPs. Group differences were assessed using the Mann- Whitney U test, and effect sizes were quantified using the rank-biserial correlation (rrb) (n = 1,397 EVs; n = 1,914 NVEPs). (D) Distributions of EV and NVEP sphericity, with representative 3D models illustrating increasing sphericity from left to right. Median sphericity values are 0.88 for EVs and 0.87 for NVEPs. Statistical analysis as in (C).

Comparison of EVs and NVEPs across sensillum classes

(A) Representative 3D models of major olfactory sensillum classes, showing sensillum cuticle, lumen, ORN outer dendrites, EVs (purple), and NVEPs (blue). Insets: magnified views. Scale bar, 1 µm. (B) EV and NVEP prevalence. The EV ratio is defined as the number of EVs divided by the total number of EVs and NVEPs. Each datapoint represents one sensillum; only sensilla with ≥5 particles were considered. Numbers of analyzed sensilla are indicated. Median ± 95% CI. Dashed line indicates an EV ratio of 0.5. Significance was assessed by Wilcoxon signed-rank test against 0.5 (*p < 0.05). (C) EV and NVEP distribution. Left: Schematic illustrating three categories of particle localization: 1) no EVs/NVEPs (gray); 2) EVs/NVEPs restricted to the sensillum lumen below the cuticle base (light shade); and 3) EVs/NVEPs present in the lumen both above and below the cuticle base (dark shade). Right: Stacked bar graphs across sensillum classes. Different letters indicate statistically significant differences between groups (Chi-square test for independence with FDR-corrected Chi-square post hoc tests, p < 0.05). (D) EV and NVEP number. Only sensilla containing at least one EV or NVEP were considered and outliers were removed using Tukey’s extreme fences. Distributions are represented as kernel density estimates with individual data points shown along the x-axis. Kruskal–Wallis test followed by FDR- corrected Dunn’s post hoc tests, p < 0.05. Scale bar, 10% density. (E) Volume-normalized EV and NVEP abundance. Only sensilla containing at least one EV or NVEP above the cuticle base were included. Each datapoint represents a single sensillum; median ± 95% CI. Statistical analysis as in (D).

Variation in EV properties across sensillum classes

(A) Representative SBEM images illustrating three categories of EV intraluminal density: dense, lucent, and cargo-filled. Two independent examples are shown for each category. Scale bar, 100 nm (applies to all panels). (B) Size distributions of EVs by intraluminal density category, shown as kernel density estimates with individual data points plotted along the x-axis. Sample sizes for each category are indicated. Different letters denote statistically significant differences between groups (Kruskal–Wallis test with FDR- corrected Dunn’s post hoc comparisons, p < 0.05). (C) Stacked bar plots showing the proportion of EV density categories across sensillum classes. Numbers of sensilla analyzed per class are indicated. Sensillum classes: large basiconic (ab1, ab2/3), small basiconic (SB), intermediate (I), trichoid (T), and coeloconic (C). Chi-square test for independence with FDR-corrected Chi-square post hoc tests, p < 0.05. (D) Representative SBEM images and corresponding 3D models of EV structural variants: double EVs (left) and ball-and-socket EVs (right). Scale bars, 250 nm. (E) Stacked bar plots showing the distribution of intraluminal density categories within double EVs and ball-and-socket EVs, highlighting enrichment of the cargo-filled category. (F) Venn diagram showing overlap between cargo-filled EVs, double EVs, and ball-and-socket EVs. Numbers indicate counts of EVs in each category and their intersections.

Multivesicular bodies within auxiliary cells

(A) 3D reconstructions of the sensillum cuticle (gray), ORNs (bronze), thecogen cell (light pink), and MVBs (dark pink). Inset: enlarged view showing the cutting plane for the representative SBEM image. MVBs are marked by thick white arrows. Scale bars: 2 µm (3D models) and 1 µm (SBEM image). (B C) Same as (A), except different auxiliary cells are shown: (B) trichogen cell (turquoise) and (C) tormogen cell (green). (D) Percentage of auxiliary cells containing MVBs, observed more frequently in trichogen and tormogen cells. Sample sizes for each auxiliary cell type are indicated (from 304 sensilla). The higher number of trichogen cells reflects the presence of two trichogen cells per coeloconic sensillum, compared to one per sensillum in other sensillum types. Different letters denote statistically significant differences between groups (Chi-square test for independence with FDR-corrected Chi-square post hoc tests, p < 0.05). (E) Number of MVBs per auxiliary cell among cells containing at least one MVB. Sample sizes indicate the number of auxiliary cells with at least one MVB. Data are shown as median ± 95% CI. Kruskal– Wallis test, p > 0.05. (F) Percentage of sensilla containing MVBs (n = 304). MVBs were more frequently observed in coeloconic sensilla. Sample sizes indicate the number of surveyed sensilla. Statistical analysis as in (D). (G) Pie charts showing the distribution of auxiliary cell types containing MVBs in coeloconic sensilla (top, n = 29) and in all other sensillum types combined (bottom, n = 24). (H) Number of MVBs per auxiliary cell (cells with ≥1 MVB) across sensillum types. Median ± 95% CI. Sample sizes indicate the number of auxiliary cell with at least one MVB. Statistical analysis as in (E).

Putative EV budding from auxiliary cells

(A) 3D reconstructions of the sensillum cuticle (gray), ORNs (bronze), thecogen cell (light pink), and a putative budding EV (purple) that remains physically connected to the auxiliary cell membrane. Inset: enlarged view. Representative SBEM image is shown below. Scale bars: 2 µm (3D models) and 400 nm (SBEM image). (B C) Same as (A), except different auxiliary cells are shown: (B) trichogen cell (turquoise) and (C) tormogen cell (green). In (B), the section shown captures a plane in which the EV membrane is no longer continuous with the trichogen cell membrane; membrane continuity is visible in adjacent serial sections (Figure 6—figure supplement 1). (D) Percentage of auxiliary cells with putative budding EVs, observed more frequently in tormogen cells. Sample sizes for each auxiliary cell type are indicated (from 304 sensilla). Different letters denote statistically significant differences between groups (Chi-square test for independence with FDR-corrected Chi-square post hoc tests, p < 0.05). (E) Number of budding EVs per auxiliary cell among cells with at least one putative budding EV. Sample sizes indicate the number of auxiliary cells with at least one putative budding EV. Data are shown as median ± 95% CI. Kruskal–Wallis test, p > 0.05. (F) Percentage of sensilla containing putative budding EVs (n = 304), which were more frequently observed in intermediate sensilla. Sample sizes indicate the number of surveyed sensilla. Statistical analysis as in (D). (G) Distribution of auxiliary cell types associated with putative budding EVs in intermediate sensilla (top, n = 7) and all other sensillum types combined (bottom, n = 21). (H) Number of putative budding EVs per auxiliary cell (cells with ≥1 budding EV) across sensillum types. Median ± 95% CI. Sample sizes indicate the number of auxiliary cell with at least one putative budding EV. Statistical analysis as in (E).

Degenerating ORNs across sensillum classes

(A) Representative 3D models of coeloconic sensillum showing the sensillum cuticle, ORN outer dendrites, EVs (purple), and NVEPs (blue; generally much smaller than EVs). Left: control sensillum with intact ORNs. Right: sensillum containing degenerating ORNs. The outer dendrites of both ORNs are truncated distally, and EVs and NVEPs are both more abundant relative to the control sensillum. Scale bar, 2 µm. (B E) Same as (A), showing additional sensillum classes. (B) Small basiconic sensillum, with both ORN outer dendrites truncated distally. (C) Large basiconic sensillum, with one ORN outer dendrite fragmented in the middle region and exhibiting distal blebbing, defined as the presence of rounded membrane protrusions along the dendritic shaft. (D) Intermediate sensillum, with one ORN outer dendrite truncated distally. (E) Trichoid sensillum, with both ORN outer dendrites truncated proximally.

Comparison of extracellular particles between intact and degenerating sensilla

(A) EV and NVEP abundance per sensillum. Distributions are shown as kernel density estimates with individual data points plotted along the x-axis. Comparisons were made between intact sensilla (gray; n = 257 for NVEPs, n = 230 for EVs) and degenerating sensilla (colored; n = 32 for both) using Mann– Whitney U tests. Scale bar, 2.5% density. (B) Distributions of EV and NVEP diameters. Curves represent kernel density estimates. Differences between intact (gray) and degenerating sensilla (colored) were assessed using Mann–Whitney U tests, with effect sizes quantified by rank-biserial correlation (rrb) (n = 3,396 EVs; n = 1,116 NVEPs). Intact sensilla data are the same as in Figure 2C. (C) EV diameter categories. Stacked bar plots show proportions of maximum EV size categories in intact and degenerating sensilla. EVs >1000 nm fall within an apoptotic body-like size range. Statistical comparison was performed using Chi-square tests for independence. (D) Size distributions of EVs grouped by intraluminal density (degenerating sensilla). Gray line: distribution for intact sensilla (as in Figure 4B). When comparing intact vs degenerating sensilla within each intraluminal density, statistical analysis is as in (B). When comparing between intraluminal density categories within degenerating sensilla, different letters denote statistically significant differences between groups (Kruskal–Wallis test with FDR-corrected Dunn’s post hoc comparisons, p < 0.05). Inset: Stacked bar plots showing the distribution of intraluminal density categories between intact and degenerating sensilla, highlighting enrichment of the dense EV class in degenerating sensilla. Statistical analysis as in (C). (E) Percentage of sensilla containing auxiliary cells with MVBs. Only sensilla with intact auxiliary cells were included. Statistical comparison was performed using Fischer exact test. (F) Left: Percentage of auxiliary cells containing MVBs in degenerating sensilla. Chi-square test for independence, p > 0.05. Right: Number of MVBs per auxiliary cell, shown as median ± 95% CI. Kruskal–Wallis test, p > 0.05. (G H) Same as (E F), except analyses were performed for putative budding EVs. Statistical comparisons could not be performed for the right panel of (H) because all auxiliary cells contained a single budding EV.

Sensilla containing degenerating auxiliary cells but intact ORNs

(A B) Representative 3D models (A) and corresponding SBEM images (B) of two basiconic sensilla from the same SBEM volume, showing the sensillum cuticle and intact ORNs. Left: control sensillum with intact auxiliary cells. Right: sensillum containing degenerating auxiliary cells. Cells are pseudocolored by identity: tormogen cell (green), trichogen cell (turquoise), thecogen cell (pink), and ORNs (bronze). Dashed outlines in images 3 and 4 indicate disintegrated membranes of the tormogen and trichogen cells, respectively. Scale bars, 2 µm for 3D models and 1 µm for SBEM images. (C) EV and NVEP abundance per sensillum, as in Figure 8A. Arrows indicate datapoints from the three sensilla containing degenerating auxiliary cells but intact ORNs.