All work performed on animals was conducted in accordance and with approval of Wright State University’s Institutional Animal Care and Use Committee. Carotid bodies and then Type I cells were isolated from ninety two rats across nine litters. Carotid bodies were resected from anesthetized (5% isoflurane in oxygen; 1 L/min) neonatal Sprague-Dawley rats (11–23 days old) before the sedated animals were euthanized via decapitation, see Burlon et al. (2009). Carotid bodies were held in ice-cold Dulbecco’s Phosphate Buffered Saline (DPBS; without calcium or magnesium, Gibco) before Type I cells were dissociated using a digestive enzyme solution composed of: 4mg of collagenase type I (220-240 u/mg, Worthington), and 2mg trypsin from bovine pancreas (8000-12,000 u/mg, Sigma) dissolved in 9.65 ml of DPBS (0.65 ml with calcium and magnesium, 9 ml without calcium and magnesium; Gibco). Isolated CB Type I cells were maintained in Ham’s F12 media (supplemented with 10% fetal bovine serum) and plated onto poly-d-lysine-coated (0.1 mg/ml aqueous, Sigma-Aldrich): 22 × 22 mm microscope glass coverslips (Fisherbrand) for immunocytochemistry experiments, 10mm glass FluoroDish coverslips (WPI) for thermal imaging experiments, or 15 mm round no.1 glass coverslips (Warner Instruments) for electrophysiology experiments. Cells were held in an incubator at 37°C, 5% CO₂ in humidified air until used for experimentation (2–6 h). These methods reliably isolate viable CB Type I cells that retain their oxygen-sensitivity (Burlon et al., 2009).

Glass coverslips with CB Type I cells adhered were removed from the incubator and immediately submerged in freezing methanol (−20 °C) for 15 min, then washed three consecutive times (5 minutes each) with a blocking-buffer consisting of 1% bovine serum albumin (Sigma) and 0.3% Triton X-100 (Alfa Aesar A16046) in DPBS with calcium and magnesium (Gibco). Freezing methanol fixation was selected to fix the cells due to its ability to preserve antibody binding sites. However, no fixing method is perfect (Whelan and Bell, 2015) and future studies using alternate fixing methods would confirm or contrast our data. Mitochondria and TASK-1 or Mitochondria and TASK-3 potassium channels were immunofluorescently labeled in separate but identical experiments by first incubating cells with primary antibodies (diluted in blocking-buffer): anti-TOMM20 (1:100 dilution, Abcam AB56783); anti-KCNK3 (1:500 dilution, Alomone APC-024); anti-KCNK9 primary antibody (1:500 dilution, Alomone APC-044), respectively, for 14 h at 4°C in a humidified chamber. Coverslips were then washed with blocking buffer four separate times, for 5 minutes each before secondary antibodies (diluted in blocking-buffer) were applied for TOMM20 (1:400, Alexa Fluor 594, Invitrogen A21201; excitation/emission: 594 nm/617 nm), and TASK-1 or TASK-3 (1:500, Alexa Fluor 488, Invitrogen A21441; excitation/emission: 493 nm/519 nm) and left at room temperature in darkness for 2 hours. Secondary antibodies were washed from coverslips using phosphate buffered saline with calcium and magnesium (Gibco), five separate times for 5 minutes each. In separate control experiments, primary antibodies were omitted, and only secondary antibodies exposed to cells to ensure no cross-reactivity or off targeted binding, of which none was observed (i.e., no signal present when imaged; data not shown). Micro slides (25 × 75 × 1 mm thick, VWR 48312) were labeled and a drop (∼50 µL) of DAPI-containing Vectashield Antifade Mounting Media (Vector Laboratories) was added before coverslips were mounted and then sealed with clear nail polish. Slides were then protected from light and stored at 4°C until imaged (1–7 days post-staining).

Slides with immunofluorescent-labeled TASK potassium channels (TASK-1 or TASK-3) and mitochondria (TOMM20) were placed on a confocal microscope stage (Olympus FV1000 Confocal LSM). Cells were visualized using a ×60 objective (Nikon), and TASK-1 or TASK-3 staining via Alexa Fluor 488 was visualized with a 488 nm laser line (15% laser power, HV: 700), and mitochondrial staining (TOMM20) via Alexa Fluor 594 was visualized with a 561 nm laser line (15% laser power, HV: 700). Z-stack (50nm steps) images were acquired sequentially for each channel, Kalman filter was activated (set to average two by line; reduces noise), aspect ratio of 1024 × 1024 pixels, 7.5x zoom, pixel size X/Y: 47 nm², and 4.0 µs/pixel dwell time. Raw images were saved in a hyperstack (Olympus.oib format) with microscope parameters and imaging data included in an Olympus metafile. Image deconvolution was accomplished using Huygens Essential Software (SVI, Hilversum, Netherlands v4.2), separately for each channel per z-stack image, using the Classic Maximum Likelihood Estimation algorithm (maximum iterations: 30). Deconvolved images were next imported into Imaris computer software (Oxford Instruments Imaris v7.7.0) and three-dimensional, volumetric rendering was accomplished via software generation of maximum intensity projections. Voxels with fluorescent-signal meeting software criterion for illumination were rendered as voluminous red or green objects (representative of mitochondria or TASK-channels, respectively) and visualized in a 3D-modeled arena topographically representative of the imaged CB Type I cells. Three-dimensional rendering and visualization of cells allowed measurements to be acquired of distances between mitochondria and TASK-1 or TASK-3 channels (Firth et al., 2009). Imaris software’s “measurement tool” was used to generate measurements, only in the “X-Y” plane, of distances between the surface of a TASK-1 or TASK-3 channel signal (i.e., Alexa Fluor 488) and the surface of the nearest mitochondria signal (i.e., Alexa Fluor 594). Data were recorded, averaged by group (i.e., mitochondria distance to TASK-1 vs TASK-3), and an unpaired student’s t-test used to statistically test if measured distances between mitochondria and respective TASK-channels were significantly different from one another. Data were analyzed and figures generated in software program Prism (Graphpad Prism v9.1.2). Data are presented as the averaged distance measured between CB Type I cell mitochondria and TASK-1 or TASK-3 ion channels plus or minus the standard error of the mean. Statistical significance was denoted if a p-value of <0.05 (n.s.: not significant, p < 0.05: *, p < 0.01: **, p < 0.001: ***, p < 0.0001: ****, for comparisons in all figures).

FluoroDish coverslips (FD3510 WPI) with CB Type I cells adhered were removed from the incubator and cells loaded with 250nM ERthermAC (EMD Millipore SCT057). ERthermAC is a commercially available, membrane-permeable and endoplasmic reticulum-targeted, BODIPY-based dye; fluorescence is inversely proportional to change in temperature whereby a decrease in temperature is visualized as a reversible increase in ERthermAC relative signal fluorescence (Kriszt et al., 2017). After loading with ERthermAC, cells were washed twice with 37°C HEPES-buffered extracellular solution (see electrophysiology methods for components), for 10 minutes each. For control experiments, loaded cells were next fixed with 4% paraformaldehyde and washed twice thereafter with 37°C extracellular-solution. Coverslips were then viewed using an Olympus FV1000 microscope. Gravity-fed extracellular solution was perfused (5 ml/min) over the coverslip for the entire length of the experiment with temperature maintained as it was passed through a thermistor-feedback controlled inline heater (Warner Instruments TC-344), set to maintain the solution in the dish at 37°C or 22°C, depending upon the experiment. Images of cells loaded with ERthermAC were acquired over time (every 5 s) with a water-immersion ×25 objective lens (Nikon XLPlan N). Cells were allowed 10 min to equilibrate to the bath (volume 1.0 ml) and baseline measurements were made before rapidly switching to a “stimulus” perfusate (i.e., identical solution with 2mM cyanide added or bubbled with 100% nitrogen and 0.2mM sodium dithionite for the anoxic solution). Regions of interest (ROI) were drawn around whole CB Type I cells (Figure 3) and pixel intensities were spatially averaged post-hoc within ImageJ (NIH). Results were interpreted as group average relative fluorescent signal intensity units (RFU) during time of interest and from baseline fluorescent intensity measurements (averaged in 30-s bins). Paired Student’s t-tests tested for significant differences between the relative fluorescent signal averaged values from peak responses at the time of interest and during baseline recordings, within each cell recording. Significance was determined with a p-value < 0.05.

The amphotericin-B (240 μg/ml; Sigma) mediated perforated-patch clamp technique was performed using glass electrodes (resistance: 6.5–9.5 MΩ) back-filled with intracellular solution composed of (in mM): 55 K₂SO₄, one EGTA, 30 KCl, 20 HEPES, five MgCl₂, 10 glucose, adjusted to pH 7.2 with KOH. Coverslips with isolated CB Type I cells adhered were loaded onto a perfusion chamber, mounted over an inverted microscope (Nikon TE2000-U), and gravity-fed, HEPES-buffered solution perfused (5 ml/min) for the entire experiment. Recording chamber volume was 0.2ml. Solution temperature was maintained as it passed through a thermistor-feedback controlled inline heater (Warner Instruments TC-344). HEPES-buffered solutions contained the following (in mM): 140 NaCl, 4.5 KCl, 2.5 CaCl₂, one MgCl₂, 11 glucose, 10 HEPES, and were adjusted to pH 7.4 with NaOH at respective temperature. Resting membrane potentials were recorded in current-clamp mode (“I = 0”; gap-free at 2 KHz) in 37°C HEPES-buffered solution before rapidly switching to a 24°C HEPES-buffered solution. All data were analyzed off-line using Clampfit 10 (Molecular Devices). Data were filtered at 1KHz and analyzed in 30-s binned averages at baseline and peak response. Paired Student’s t-tests determined if average resting membrane potentials recorded at 24°C were different from those recorded at 37°C for the same cell. Data are reported as group averages plus or minus the standard error of the mean. Statistical significance was denoted if a p-value of <0.05 was observed.

Distances between fluorescently labeled mitochondria (TOMM20) and TASK-1 (Figure 1 A & B) or TASK-3 channels (Figure 1 C & D) were measured. It was of interest to determine if CB Type I cell oxygen-sensing microdomains formed whereby mitochondria closely associate with TASK-potassium ion channels forming chemotransductive-signaling arenas (Gao et al., 2017; Rakoczy & Wyatt, 2018). TASK-1 and -3-channels were targeted, as they are the major “oxygen-sensitive” current-carriers in rodent CB Type I cells (Kim et al., 2009; Turner & Buckler, 2013). Details of CB Type I cells microdomains, defined as a region whereby a TASK-Channel (i.e., TASK-1 or -3 signal) was within 1 μm of mitochondria (i.e., TOMM20 signal), are provided in Figure 2. Distribution of data in Figure 2 show ∼72% of TASK-1 and ∼79% of TASK-3 measured distances from mitochondria were ≤0.5µm. As a TASK-1/3 heteromultimer-specific antibody is not commercially available, distances between mitochondria and TASK-1 and TASK-3 channels were made separately but in identical experiments. Average distances found between mitochondria and TASK-1: 0.33 ± 0.04µm (n = 47; from 16 cells over six coverslips) and TASK-3 potassium channels: 0.32 ± 0.03 µm (n = 54, from 15 cells over five coverslips), shown in Figure 3, were not significantly different from one another, and suggest both CB Type I cells’ TASK-1 and TASK-3 monomeric, potentially as well as TASK-1/3 heteromultimer potassium ion channels reside in microdomains.

Rapid intracellular thermal fluxes have been demonstrated to occur, in living cells, at up to 3-fold the distances, compared those measured here (i.e., 1 μm: Sotoma et al., 2021). As CB Type I cells contain a dense population of mitochondria, and these organelles are known to produce heat as a byproduct of respiratory chain activity (i.e., mitochondrial thermogenesis; Chrétien et al., 2018), it was of interest that rodent CB Type I cells TASK-Channels resided within distances from a heat-source wherein rapid changes in intracellular temperature have been shown to occur (Sotoma et al., 2021). Therefore “heat flux for short-distance thermal signaling” in CB Type I cells may be possible whereby temperature-sensitive TASK-Channel activity is influenced by oxygen-sensitive mitochondrial thermogenesis (Maingret et al., 2000; Chrétien et al., 2018; Sotoma et al., 2021).

Intracellular temperature of isolated CB Type I cells was visualized using the heat-sensitive dye, ERthermAC. The first set of experiments utilized paraformaldehyde-fixed, and previously dye-loaded, CB Type I cells to test the temperature-dependence of the dye in non-living cells. Cells were imaged while the bath temperature was reduced from 37°C to 22°C (example trace shown in Figure 4A). Relative signal intensity was averaged for each condition (37°C vs 22°C; 30 s bins) and fluorescent signal intensity found to significantly increase in fixed cells as they were passively cooled from 37°C to 22°C (p < 0.001; n = 10; 755.1 ± 67.3 vs 1007.8 ± 93.7 relative fluorescence intensity units; Figure 4B). ERthermAC signal intensity (250nM) was shown to be inversely related to temperature: as the bath temperature was decreased, signal intensity was found to significantly increase. Next, it was of interest to test whether whole-cell temperature changes occur as a result of inhibition of mitochondrial activity (i.e., electron chain transport). Baseline fluorescent measurements were made before cells were exposed to 2mM cyanide while bath temperature was held constant at 37°C for the entire length of the experiment (example trace shown in Figure 5A). Cyanide was found to significantly increase ERthermAC fluorescence, suggesting a decrease in intercellular temperatures (p < 0.01; n = 9; 679.3 ± 74.3 vs 1131.9 ± 96.6 RFU; Figure 5B). Control experiments exposed ERthermAC-loaded, paraformaldehyde-fixed CB Type I cells to 2 mM cyanide (trace shown in Figure 5C) and showed no significant effect of the compound on signal fluorescence (n.s.; 1391.9 ± 110.2 vs 1370.1 ± 115.5 RFU.; n = 9; Figure 5D). As CB Type I cells depolarize in response to low oxygen levels, we next decided to test whether anoxia was capable of decreasing whole-cell temperatures. Baseline measurements were made again in ERthermAC-loaded, isolated CB Type I cells with bath temperature held constant at 37 °C for the entire experiment, before rapidly switching to a hypoxic HEPES-buffered solution (N₂ bubbled and 0.2mM sodium dithionite added; trace shown in Figure 6A). Anoxia was shown to significantly increase ERthermAC signal intensity, again suggesting a decrease in intercellular temperatures (p < 0.0001, n = 8; 843.3 ± 61.1 vs 1044.8 ± 57.3 RFU; Figure 6B). Control experiments exposed ERthermAC-loaded, paraformaldehyde-fixed CB Type I cells to the anoxic solution (trace shown in Figure 6C) and showed no significant effect of anoxia on signal fluorescence (n.s.; 789.1 ± 72.8 vs 776.3 ± 74.4 RFU; n = 7; Figure 6D). Together, results suggested that inhibition of CB Type I cell mitochondrial activity was capable of significantly reducing whole-cell temperatures. Furthermore, as mitochondria in Type I cells are found near temperature sensitive TASK potassium channels (see Figures 1, 2 and Maingret et al., 2000), oxygen-dependent temperature reductions in CB Type I cell microdomains may play a role in chemotransduction in these cells, and potentially others (Firth et al., 2009).

Resting membrane potentials of isolated CB Type I cells were recorded via perforated-patch whole-cell current-clamp (I = 0 mode) recordings. Electrophysiological recordings demonstrated sensitivity of resting membrane potentials to temperature (example trace shown in Figure 7A): lowering the temperature of the extracellular solution from 37° to 24°C induced consistent and reversible cellular depolarizations. Averaged resting membrane potentials measured of isolated CB Type I cells at 37°C were significantly different from depolarized potentials measured when bath temperature was lowered to 24°C (-48.4 ± 4.1 mV vs V_m 24°C: -31.0 ± 5.7 mV; n = 5; p < 0.01; Figure 7B) this represents an average depolarization of 17.4 ± 2.2 mV.