Canton-S (CS) was used as the wild type (wt) control. The following flies were previously described: Ir21a-Gal4 (Ni et al., 2016), Ir68a-Gal4 (Knecht et al., 2017), Ir21a^Δ1 (Ni et al., 2016), Ir25a² (Benton et al., 2009), Ir93a^Mi (Knecht et al., 2016), UAS-Ir21a (Ni et al., 2016), UAS-Ir25a (Abuin et al., 2011), UAS-Ir93a (Knecht et al., 2016), UAS-GFP [p(10XUAS-IVS-Syn21-GFP-p10)attP2] (Pfeiffer et al., 2012), UAS-TrpA1 (Hamada et al., 2008), and UAS-GCaMP6m [p(20XUAS-IVS-GCaMP6m)attP40 and p(20XUAS-IVS-GCaMP6m)VK00005] (RRID:BDSC_42748 and BDSC_42750) (Chen et al., 2013).

Calcium imaging and analysis of larval neurons to temperature changes were performed as described (Tyrrell et al., 2021; Omelchenko et al., 2022a; Omelchenko et al., 2022b). Fly larvae at 72 h AEL were collected using 10 mL 20% w/v sucrose solution. The duration in 20% sucrose solutions was counted from the moment that the 20% sucrose solution was added to the fly vial. Approximately 15–20 min later, larvae were transferred into 100 μL drops of 20% sucrose; each drop contained 20–30 larvae. If other hypertonic solutions (10% sucrose, 40% sucrose, and 0.5 M NaCl) were applied, larvae were picked using a small paint brush and then placed in the hypertonic solutions. The duration in hypertonic solutions was counted from the moment that larvae were placed in the hypertonic solutions. When calcium imaging was conducted, a 3rd-instar larva was selected using a stereo microscope and transferred from hypertonic solutions to 75 μL of 1× PBS on a slide for imaging, and the time the larva spent in the hypertonic solutions was recorded. We combined several criteria to distinguish between A-type and B-type DOCCs. Within a DOG, the B-type DOCC is the most anterior DOCC and tends to have shorter dendrites than A-type DOCCs (Hernandez-Nunez et al., 2021). In addition, A-type DOCCs also typically exhibit quicker responses than B-type DOCCs, and the calcium changes in two A-type cells generally occur at concurrent or similar time points (Klein et al., 2015).

Calcium imaging was conducted on a Zeiss LSM 880 using the Airyscan Fast mode, equipped with a z-axis piezo stage (432339-9000-000, Wienecke & Sinske) and a stage insert (432339-9030-000, Wienecke & Sinske). Z-stacks were captured at 7–8 fps with a 512 × 512 resolution and 1.5× zoom using a 25× water objective. A custom-built thermoelectric cooler was made to deliver the temperature stimuli by attaching a thermoelectric module (30 mm × 30 mm, TE-127-1.0–0.8, TE Technology) to a modified heat sink (12.9 cm × 5.5 cm, ATS2193-ND, Digi-Key). The cooler was placed on the slide covering the larvae, and a 2A current was supplied using a power supply (CSI1802X, Circuit Specialists). The temperature was monitored using a data acquisition device (USB-TEMP, Measurement Computing) and DAQami software (Measurement Computing). Images were analyzed using ImageJ plugins TrackMate and TACI (Tinevez et al., 2017; Omelchenko et al., 2022a).

Immunostaining was performed as described (Kang et al., 2011). The following antibodies were used: guinea pig anti-IR21a (1:100) (Budelli et al., 2019), rabbit anti-IR93a (1:100) (Knecht et al., 2016), guinea pig anti-IR25a (1:100) (Benton et al., 2009), chicken anti-GFP (1:500; Abcam), goat anti-guinea pig Cyanine Cy™3 (1:100; Jackson ImmunoResearch), goat anti-rabbit Alexa Fluor^® 647 (1:100; Jackson ImmunoResearch), and goat anti-chicken FITC (1:500; Invitrogen). IR21a, IR93a, and IR25a antibodies were kind gifts from Dr. Richard Benton. To quantify immunostaining, the center plane of each soma was determined in ImageJ. The mean intensity was then read (Analyze > Measure) and the background intensity was subtracted.

Statistical details of experiments are described in figure legends. The normality of distributions was assessed by the Shapiro–Wilk test (p ≤ 0.05 rejected normal distribution). Statistical comparisons of normally distributed data were performed by the one sample t-test, the two-tailed unpaired t-test, the Welch’s t-test, or the ordinary one-way ANOVA test followed by the Tukey test. For data that did not conform to a normal distribution, statistical comparisons were performed by the one sample Wilcoxon test, the Mann–Whitney test, or the Kruskal–Wallis test followed by the Dunn’s test. Chi-square test for trend or Fisher’s exact test was used in analyzing the detectability of DOCCs. Data analysis was performed using GraphPad Prism 9.

We first tested the temperature responses of DOCCs in 20% sucrose using calcium imaging. The duration of sucrose exposure affected the temperature responses of B-type DOCCs. When larvae stayed in the sucrose solution for no more than one hour, most animals exhibited detectable GCaMP baseline signals in B-type DOCCs with robust calcium responses to cool temperatures, similar to A-type responses (Figures 1B, C, H, J, K). However, when the duration of exposure to sucrose was extended to three hours, more than half of the animals no longer exhibited detectable B-type DOCCs (Figures 1F, G, K). The undetectability of B-type DOCCs was not due to the absence of B-type DOCCs but because their GCaMP signals fell below the detection threshold for two reasons. First, B-type DOCCs remained labeled by IR21a, IR25a, and IR93a antibodies under hypertonic conditions (Supplementary Figure 1A). Second, B-type DOCCs that were undetectable at baseline exhibited warm-activated responses when expressing a warm-activated temperature receptor TRPA1 (Supplementary Figure 2). For the B-type DOCCs that were detectable by calcium imaging, their temperature responses significantly decreased and were smaller than those of A-type DOCCs (Figures 1D, E, I, J). Sucrose exposure did not decrease the cooling responses or the baseline detectability of A-type DOCCs (Figures 1B–J).

We then tested the dose effect using different concentrations of sucrose solutions. When larvae were exposed to 10% sucrose for three hours, we did not observe a reduction in the proportion of detectable B-type cells or their temperature responses (Figures 1L, M). This result is as expected because 10% sucrose is a relatively weaker hypertonic solution and, thus, does not exert a significant effect within the tested time frames. In contrast, exposure to 40% sucrose influenced the responses and detectability of B-type DOCCs within the first hour of incubation (Figures 1N, O). One hour of incubation in 40% sucrose led to similar effects on B-type cells to those exposed to 20% sucrose for three hours: neither the temperature responses in detectable B-type cells (Mann–Whitney test, p = 0.4286) nor the detectability (Fisher’s exact test, p > 0.9999) showed a significant difference between these two conditions. We did not observe a further decrease in the proportion of B-type DOCCs with a detectable GCaMP baseline following prolonged exposure to 40% sucrose (Figure 1O). In addition, the responses of detectable B-type DOCCs did not further diminish (Figure 1N). This indicate that one-hour incubation in 40% sucrose has already exerted the maximum impact on B-type DOCCs, and thus no further decrease in detectability or responses can be observed.

In addition to sucrose, we tested another hypertonic solution 0.5 M NaCl. Similar to 20% sucrose, 0.5 M NaCl had no effect on A-type DOCCs after three hours of incubation (Figures 1P–R). However, the responses of detectable B-type DOCCs significantly declined (Figures 1P–R). Furthermore, the detectability of B-type DOCCs also decreased (Figure 1S). These findings suggest hypertonic environments selectively modulate B-type DOCCs.

To explore the mechanisms that underlie the discrepancy in cooling responses between A-type and B-type DOCCs under hypertonic conditions, we examined the expression levels of IRs (IR21a, IR25a, and IR93a) in the soma. The specificity of IR antibodies was validated using the corresponding Ir mutants (Supplementary Figure 3). A-type DOCCs (yellow arrows) paired with DOWCs, labeled by Ir68a > GFP (Figures 2A, B). B-type DOCCs (white arrowheads) paired with an IR25a-positive cell that did not express IR21a, IR93a, or Ir68a-Gal4 (Figures 2A, B). When we quantified the somal expression of each IR, we observed that IR93a expression was significantly higher in A-type than B-type DOCCs, while IR21a and IR25a exhibited similar levels in both (Figure 2C and Supplementary Table 1). Hypertonic conditions had no impact on the expression patterns of IRs (Supplementary Figure 1A and Supplementary Table 1). These results suggest that the levels of IR subunits may play a role in determining the different cooling responses of A-type and B-type DOCCs under hypertonic conditions.

To explore whether the discrepancies in the detectability and temperature responses between A-type and B-type DOCCs were due, at least in part, to the different levels of IRs, we expressed UAS-Ir21a, UAS-Ir25a, and UAS-Ir93a in both A-type and B-type DOCCs using Ir21a-Gal4. Overexpressing Ir21a did not impact calcium responses of detectable B-type DOCCs (Figures 3A–C) or the proportion of animals with detectable B-type DOCCs compared to wt (Figure 3D). Therefore, overexpression of Ir21a did not alter B-type responses with or without prolonged hypertonic exposure.

Overexpression of Ir25a had a noticeable impact on the responses and detectability of both A-type and B-type DOCCs. When Ir25a was overexpressed, more than half of the animals failed to display detectable A-type DOCCs, but the proportion of animals with detectable A-type DOCCs did not decline with prolonged exposure to sucrose (Supplementary Figure 4A). Moreover, the responses of A-type DOCCs remained unchanged over time in sucrose (Figures 3E–G). We could not observe B-type DOCCs in animals that did not exhibit detectable A-type DOCCs. Unlike A-type DOCCs, the proportion of animals with detectable B-type DOCCs decreased with extended exposure to sucrose (Supplementary Figure 4B). When considering only animals with detectable A-type DOCCs, the proportion of animals exhibiting detectable B-type DOCCs resembled that of wt animals (Figure 3H). Like in wt, the responses of detectable B-type DOCCs diminished over time in sucrose. After three hours, no detectable B-type DOCCs were observed (Figures 3E–G). Therefore, the overexpression of Ir25a did not change how hypertonic conditions impacted their responses, even though it led to a reduction in the proportion of animals with detectable A-type and B-type DOCCs and their responses without extended exposure to hypertonic solutions.

The overexpression of Ir93a had a selective impact on B-type DOCCs. The detectable B-type DOCCs maintained their temperature responses after extended sucrose exposure (Figures 3I–K) despite the proportion of animals with detectable B-type DOCCs being reduced to the wt level (Figure 3L). Unlike in wt larvae and those overexpressing Ir21a and Ir25a, the expressions of IR21a and IR25a were comparable in both A-type and B-type DOCCs, and IR93a had a higher level in A-type than in B-type DOCCs (Supplementary Figures 1A–C and Supplementary Table 1). Overexpressing Ir93a eliminated the discrepancy in IR93a expression levels between A-type and B-type DOCCs (Figure 4, Supplementary Figure 1D, and Supplementary Table 1), and restored the detectable B-type calcium responses to cool temperatures after three hours of exposure to sucrose. However, the proportion of B-type DOCCs containing detectable GCaMP baseline was not restored by overexpressing Ir93a. Therefore, overexpression of Ir93a restores the responses of detectable B-type DOCCs under hypertonic conditions but not their detectability.