Strong stimulation triggers full fusion exocytosis and very slow endocytosis of the small dense core granules in carotid glomus cells

Abstract

Chemosensory glomus cells of the carotid bodies release transmitters, including ATP and dopamine mainly via the exocytosis of small dense core granules (SDCGs, vesicular diameter of ∼100 nm). Using carbon-fiber amperometry, we showed previously that with a modest uniform elevation in cytosolic Ca²⁺ concentration ([Ca²⁺]_i of ∼0.5 µM), SDCGs of rat glomus cells predominantly underwent a “kiss-and-run” mode of exocytosis. Here, we examined whether a larger [Ca²⁺]_i rise influenced the mode of exocytosis. Activation of voltage-gated Ca²⁺ channels by a train of voltage-clamped depolarizations which elevated [Ca²⁺]_i to ∼1.6 μM increased the cell membrane capacitance by ∼2.5%. At 30 s after such a stimulus, only 5% of the added membrane was retrieved. Flash photolysis of caged-Ca²⁺ (which elevated [Ca²⁺]_i to ∼16 μM) increased cell membrane capacitance by ∼13%, and only ∼30% of the added membrane was retrieved at 30 s after the UV flash. When exocytosis and endocytosis were monitored using the two-photon excitation and extracellular polar tracer (TEP) imaging of FM1–43 fluorescence in conjunction with photolysis of caged Ca²⁺, almost uniform exocytosis was detected over the cell’s entire surface and it was followed by slow endocytosis. Immunocytochemistry showed that the cytoplasmic densities of dynamin I, II and clathrin (key proteins that mediate endocytosis) in glomus cells were less than half of those in adrenal chromaffin cells, suggesting that a lower expression of endocytotic machinery may underlie the slow endocytosis in glomus cells. An analysis of the relative change in the signals from two fluorescent dyes that simultaneously monitored the addition of vesicular volume and plasma membrane surface area, suggested that with an intense stimulus, SDCGs of glomus cells underwent full fusion without any significant “compound” exocytosis. Therefore, during a severe hypoxic challenge, glomus granules undergo full fusion for a more complete release of transmitters.

Keywords:

• Dense core granules
• type I cells
• chemosensory organ
• dopamine secretion

Introduction

Studies on exocytosis in a wide-range of neurosecretory systems have revealed full fusion, kiss-and-run and compound exocytosis as the three major modes of exocytosis (Wu, Hamid, Shin, & Chiang, 2014). In the full fusion mode, the fused granules merged with the plasma membrane, and the vesicular membrane was eventually retrieved via endocytosis. In the kiss-and-run mode, granules fuse transiently with the plasma membrane without complete loss of shape, protein or lipid (Harata, Aravanis, & Tsien, 2006). The compound mode involves the exocytosis of large granules formed via granule-granule fusion (Wu et al., 2014), or the fusion of a granule with another granule that has already fused with the plasma membrane (aka “sequential compound” exocytosis) (Kishimoto et al., 2006; Lam et al., 2013; Takahashi et al., 2004). The various forms of exocytosis may be utilized in the same cells for distinct modes of secretion (Kasai, Takahashi, & Tokumaru, 2012).

A class of less well-studied secretory granules is the small dense core granules (SDCGs). SDCGs have a mean diameter (∼100 nm) that is intermediate between the small synaptic vesicles (SV; ∼ 50 nm) and the large dense core granules (LDCGs; ≥200 nm). In vertebrates, these type of granules are present in the chemosensory cells (glomus cells) of the carotid bodies (McDonald & Mitchell, 1975; Pallot, Al Neamy, & Blakeman, 1986) and the interneurons (small intensely fluorescent cells) of various autonomic ganglia (Doupe, Patterson, & Landis, 1985). Note that however, granules with a dense core and a mean diameter of ∼100 nm were classified as “large dense core vesicles” in Retzius cells of the leech (Bruns & Jahn, 1995; Bruns, Riedel, Klingauf, & Jahn, 2000; Wen, Saltzgaber, & Thoreson, 2017; Xia, Lessmann, & Martin, 2009; Zhang, Li, & Tsien, 2009).

In a previous study, we employed carbon-fibre amperometry to compare the kinetics of quantal catecholamine release from the SDCGs of rat glomus cells and the LDCGs of rat adrenal chromaffin cells (Wang et al., 2012). Our results showed that with a modest stimulus (whole-cell dialysis of a Ca²⁺-buffered solution that elevated [Ca²⁺]_i to ∼0.5 µM), SDCGs and LDCGs had key differences in their release kinetics. For glomus SDCGs, ∼12% of the amperometric signals was the stand-alone foot signals and ∼56% of the amperometric spikes had very rapid decay kinetics (decay τ < 1 ms), suggesting that the fusion pore of a large fraction of glomus SDCGs underwent either flickering or abrupt closure, which reflected the kiss-and-run mode of exocytosis. In contrast, the majority (>98%) of the chromaffin LDCGs underwent significant fusion pore dilation when [Ca²⁺]_i was uniformly elevated to ∼0.5 μM (Wang et al., 2012). The same study also showed that stimulation with 50 mM extracellular K⁺ which elevated [Ca²⁺]_i to ∼1 µM, increased the mean quantal size of glomus SDCGs by ∼17% and reduced the proportion of amperometric signals with very rapid decay kinetics by ∼20% (Wang et al., 2012). The last observation raised the possibility that, a larger [Ca²⁺]_i rise may prevent, or slow, the closure of the fusion pores of glomus SDCGs, resulting in more catecholamine release.

In the current study, we examined whether a large [Ca²⁺]_i rise could shift the mode of exocytosis of glomus SDCGs to full fusion or compound exocytosis. We found that activation of voltage-gated Ca²⁺ channels (VGCCs) via a train of depolarizations or flash photolysis of caged Ca²⁺ increased the cell membrane capacitance (C_m; an electrical measurement of the cell membrane surface area), reflecting fusion of SDCGs. Our C_m measurements and experiments with the two-photon excitation and extracellular polar tracer (TEP) imaging of FM1–43 fluorescence showed that glomus cells exhibited a very slow form of endocytosis. The slow endocytosis was correlated with low cytoplasmic densities of clathrin and dynamin. Our TEP results suggest that following a strong stimulus, SDCGs mainly undergo full fusion without any significant compound exocytosis. Such a dominant mode of exocytosis among glomus SDCGs at high [Ca²⁺]_i can contribute to a more complete release of transmitters.

Method

Cell preparation

Male Sprague-Dawley rats (200–250 g) were euthanized in accordance to the standards of the Canadian Council on Animal Care. The carotid bodies were removed and dissociated enzymatically as described in our previous studies (Xu, Tse, & Tse, 2007; Yan, Lee, Tse, & Tse, 2012). Single carotid cells or cell clusters were plated onto glass coverslips and kept in a medium containing F-12/Dulbecco’s modified Eagle’s medium (1:1), supplemented with 5% fetal calf serum, 50 U/ml of penicillin G and 50 µg/ml of streptomycin. All culture materials were from Gibco (Grand Island, NY, USA). Recordings were performed on cells maintained in culture for 24–40 h. The carotid body contains glomus (type I) and the glial-like sustentacular (type II) cells. We identified glomus cells based on their ovoid shape and/or the presence of voltage-gated Ca²⁺ channels as described previously (Xu, Xu, Tse, & Tse, 2005).

Chemicals and solutions

Indo-1 or indo1-FF was obtained from Teflabs (Austin, TX, USA). Nitrophenyl-EGTA (NP-EGTA) was a gift from Dr. Graham Ellis-Davis (Mount Sinai Hospital, NY, USA). All other chemicals were obtained from Sigma–Aldrich (Oakville, ON, Canada). The standard bath solution contained (in mM): 150 NaCl, 4.5 KCl, 2.5 CaCl₂, 1 MgCl₂, 8 glucose and 10 Na-HEPES (pH 7.4). The whole-cell pipette solution contained (in mM): 70 Cs-aspartate, 20 TEA-Cl, 40 Cs-HEPES, 1 MgCl₂, 2 Na₂ATP and 0.1 Na₄GTP (pH 7.4).

Measurement of C_m and [Ca²⁺]_i

Single glomus cells were whole-cell voltage clamped at −80 mV using an EPC-9 amplifier and C_m was measured using the built-in lock-in software in the EPC-9 Pulse software (List-electronic, Darmstadt, Germany). A 1-kHz, 10-mV peak-to-peak sinusoid and the sine wave plus direct current method were used to calculate C_m as described previously (Yan et al., 2012). When triggering exocytosis via activation of VGCCs, a train of voltage steps (each to +10 mV for 200 ms) was delivered to the cell and the pipette solution contained 0.1 mM indo-1. When triggering exocytosis via flash photolysis of caged Ca²⁺, the pipette solution contained 10 mM NP-EGTA (80–95% saturated with Ca²⁺) and 0.1 mM indo-1 FF. To photolyse Ca²⁺-NP-EGTA, a UV flash from a modified XF-10 xenon flash lamp (Hi-Tech Ltd, Salisbury, UK) was delivered to the cell via a fused silica focusing lens which replaced the microscope’s condenser. Details of the instrumentation and calibration of the Ca²⁺ signal were as described in our previous studies (Lee & Tse, 2001; Yan et al., 2012). A junction potential correction of −10 mV was applied. All experiments were conducted at room temperature (20–23 °C).

TEP imaging and TEP imaging-based quantification (TEPIQ)

The equipment and procedures were identical to those described previously for imaging FM1–43 fluorescence and photolysis of NP-EGTA (Hatakeyama, Takahashi, Kishimoto, Nemoto, & Kasai, 2007; Liu et al., 2005). Cells were incubated with 30 µM of the acetoxymethyl ester (AM) form of NP-EGTA for 30 min. In separate experiments, cells were incubated with 10 µM of the AM form of fura-2 FF and we found that the duration of UV illumination needed to elevate [Ca²⁺]_i to >10 µM (Takahashi et al., 2004) was ∼0.5 s. Therefore, in all TEP imaging experiments, cells were illuminated for 0.5 s with UV for the photolysis of NP-EGTA.

The individual cells selected for TEP imaging must have no detectable shape change after photolysis of caged Ca²⁺. For cells with increase in FM1–43 fluorescence (ΔF1) > 50% after photolysis, blebs were frequently detected on the plasma membrane. The blebs changed the profile of the cell in the Z-plane of focus and reduced the overall sharpness of the fluorescent plasma membrane. Such gross changes in cell shape are not suitable for TEPIQ. Therefore, we restricted our analysis to cells with ΔF1 < 50%. When estimating the amount of exocytosis, we corrected for any “constitutive endocytosis” while the cell was exposed to FM1–43 with the following procedure. In separate experiments (n = 3), we measured the average change in cellular fluorescence in individual glomus cells before and after 3 min of focal application of FM1–43, as well as after washing off the extracellular FM1–43. On average, after the removal of extracellular FM1–43, glomus cells retained 3.9 ± 1.7% of the steady state fluorescence during the duration (3 min) of FM1–43 application, indicating that constitutive endocytosis increased FM1–43 fluorescence with a rate of 1.3% per minute. We assumed that glomus cells had a constant rate of constitutive endocytosis, and calculated the anticipated increase in cellular fluorescence arising from constitutive endocytosis by scaling the rate of 1.3% per minute with the appropriate duration of FM1–43 exposure in individual cells. This value was subtracted from the increase in FM1–43 fluorescence after photolysis to obtain the net increase in FM1–43 fluorescence.

The procedures for TEPIQ that involved the simultaneous measurement of two extracellular dyes (FM1–43 and sulforhodamine-B (SRB)), and for the calibration of the scaling factors (F_M and F_E; which converted the increase in FM1–43 and SRB fluorescence into ΔS and ΔV respectively), were as described previously (Kasai et al., 2005; Liu et al., 2005). The values of F_E (9,960,588/µm²) were obtained from the fluorescent intensity of the bath solution. The values of F_M (3110/µm²) was obtained from the fluorescent intensity of 20 µM of FM1–43 (in 40 mM CHAPS) and the conversion coefficient from the fluorescence of FM1–43 solution to that of the plasma membrane (m_C) was 0.28.

Immunocytochemistry

The adrenal medulla and the carotid bodies were harvested from three rats. The procedures for dissociating carotid cells and adrenal chromaffin cells were as described previously (Wang et al., 2010; 2012). Cells were plated onto glass coverslips and kept in culture for 24 h. The cultured cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 and then stained with at least two of the following primary antibodies. Three primary antibodies were purchase from Abcam: rabbit anti-dynamin I (ab52611), rabbit anti-dynamin II (ab65556) and mouse anti clathrin (ab2731). Two primary antibodies were purchase from Sigma: mouse anti-myosin light chain (M4401) and mouse anti-tubulin (T5168). The primary antibody rabbit anti-actin (R49) and all Alexa Fluor tagged second antibodies (Molecular Probes A21201 and A21441) were gifts from the laboratory of Dr. Luc Berthiaume (Department of Cell Biology, University of Alberta). The immunostained cells (on coverslips) were mounted onto glass slides with PoLong Gold anti-fade with the nuclear stain DAPI (P36931, Invitrogen). Images were acquired in the Cellular Imaging Centre (Faculty of Medicine and Dentistry, University of Alberta) with a Perkin Elmer ERS6 spinning disc confocal on a Leica DMI inverted base and using a CSU-22 disc (Woodbridge, ON). Fluorophores were excited with 405, 488, 561, 647 nm lasers line matched dichroics and paired excitation filters, and imaged using a 20×/0.5 NA objective onto a Hamamatsu c9100 EMCCD camera. All images were 0.8 μm thick optical sections at approximately half the Z-axis depth of individual cells. The Volocity Sofware (Perkin, Elmer) was used to acquire and analyze images.

Data analysis

Origin program 8 (OriginLab Corp., Northampton, MA) was employed for plotting and statistical procedures. The two-sample Student’s t-test was used to compare mean values between two populations of cells. Any difference with p < .05 was considered statistically significant and was marked with an asterisk in the figures. All values shown were mean ± SEM.

Results

Activation of VGCCs evoked full fusion exocytosis in glomus cells

In the full fusion mode of exocytosis, the fused granules significantly merge with the plasma membrane, resulting in an increase in the cell surface membrane area, which can be recorded electrically as an increase in C_m. Therefore, we employed C_m measurement to examine whether some glomus SDCGs could undergo this mode of fusion. Since hypoxia triggered rise in [Ca²⁺]_i in rat glomus cells via depolarization and activation of VGCCs (Buckler & Vaughan-Jones, 1994; Xu et al., 2005), we stimulated exocytosis in single glomus cells by delivering a train of depolarization to activate VGCCs. Figure 1(A) shows an example of a simultaneous measurement of C_m and [Ca²⁺]_i in a glomus cell. The cell was voltage clamped at −80 mV and a train of 20 depolarization (200 ms step to +10 mV at 2 Hz) elevated [Ca²⁺]_i to ∼1.5 µM. The rise in [Ca²⁺]_i was accompanied by an increase in C_m, indicating that many glomus granules underwent full fusion exocytosis. In this example, the amplitude of exocytosis (ΔC_mexo) was 108 fF (2.6% of the initial C_m). Assuming that SDCGs in glomus cells have a mean diameter of ∼100 nm and their membrane has a specific capacitance similar to that of LDCGs of chromaffin cells (0.9 fF/μm²) (Albillos et al., 1997), then the full fusion of a single SDCG should increase C_m by 0.28 fF. Thus, the C_m increase shown in Figure 1(A) corresponded to the full fusion exocytosis of ∼380 SDCGs. After the termination of depolarization, [Ca²⁺]_i returned to the basal level in ∼10 s while C_m declined very slowly, reflecting a slow form of endocytosis. At 30 s after the termination of depolarization, the amount of endocytosis (ΔC_mendo) was 4 fF. Thus, endocytosis retrieved only 3.7% of the vesicular membrane that was added to the plasma membrane during exocytosis (i.e. ΔC_mendo normalized to ΔC_mexo). On average, the peak [Ca²⁺]_i evoked by a train of depolarization was 1.6 ± 0.5 µM (n = 6; Figure 1(Bi)) and exocytosis (ΔC_mexo normalized to initial C_m) was 2.5 ± 0.6% (Figure 1(Bii)). At 30 s after the termination of the depolarization, endocytosis retrieved only 5.1 ± 1.7% of the added vesicular membrane (Figure 1(Biii)).

Figure 1. A train of depolarization elevated [Ca²⁺]_i and triggered an increase in the cell membrane capacitance (C_m). (A), Simultaneous measurement of [Ca²⁺]_i (indo-1 fluorometry) and C_m. The cell was whole-cell voltage-clamped at −80 mV. A train of 20 depolarization steps (each to +10 mV for 200 ms) was delivered to activate VGCCs (time indicated by the bar). [Ca²⁺]_i was elevated to ∼1.5 µM and the total increase in C_m (ΔC_mexo) is 0.108 pF. At 30 s after the termination of the train of depolarization, the amount of endocytosis (ΔC_mendo) was 0.004 pF. (B) The values of (i) peak [Ca²⁺]_i, (ii) exocytosis (ΔC_mexo normalized to initial C_m) and (iii) endocytosis at 30 s after the termination of the train of depolarization (ΔC_mendo normalized to ΔC_mexo) averaged from six experiments.

Flash photolysis of caged Ca²⁺ evoked more full fusion exocytosis that was followed by slow endocytosis

In the experiment described above, activation of VGCCs was employed to trigger exocytosis. It is generally accepted that activation of VGCCs generates a spatial Ca²⁺ gradient such that the local [Ca²⁺] is higher for granules that are in close proximity to the VGCCs. Thus, it is conceivable that the number of SDCGs that undergo full fusion exocytosis may be limited by their proximity to the VGCCs. Therefore, we examined whether a large uniform rise in [Ca²⁺]_i evoked by flash photolysis of caged Ca²⁺ could drive more SDCGs to undergo exocytosis. Figure 2(A) shows an example of the increase in C_m in a glomus cell when [Ca²⁺]_i was elevated via flash photolysis of caged Ca²⁺. A flash of UV light was delivered at the time indicated by the arrow. [Ca²⁺]_i elevated rapidly to ∼25 μM and this was accompanied by a robust increase in C_m. Within 10 s after the UV flash, C_m reached a maximum (ΔC_mexo =296 fF) and it was followed a slow decline in C_m, reflecting slow endocytosis. At 30 s after the UV flash, the amount of endocytosis (ΔC_mendo) was only 71 fF. Thus, endocytosis retrieved ∼24% of the vesicular membrane that was added to the plasma membrane during exocytosis. On average, the peak [Ca²⁺]_i evoked by flash photolysis of caged Ca²⁺ was 16.5 ± 2.4 µM (n = 9; Figure 2(Bi)) and exocytosis was 13.2 ± 2.0% of the initial C_m (Figure 2(Bii)), approximately 5-fold larger than the exocytosis triggered via activation of VGCCs (Figure 1(Bii)). In other secretory cells such as pituitary corticotropes, a large [Ca²⁺]_i elevation (>10 μM) triggered a ‘fast’ form of endocytosis that manifested as multiple step-like decreases in C_m (Lee & Tse, 2001). In contrast, we observed only the slow form of endocytosis in glomus cells. At 30 s after the UV flash, endocytosis in glomus cells was 29.5 ± 5.5% (Figure 2(Biii)) which was approximately 6-fold larger than that detected during VGCC activation (Figure 1(Biii)). At 100 s after the UV flash, endocytosis was only 40.3 ± 9.0% (n = 6). In these cells, [Ca²⁺]_i decayed with a time constant of 8.4 ± 1.3 s to a plateau of 2.3 ± 0.3 µM. Thus, the slowing of endocytosis between 30 to 100 s after the UV flash may be related the lower level of [Ca²⁺]_i.

Figure 2. Flash photolysis of caged Ca²⁺ triggered a robust increase in C_m. (A), Simultaneous measurement of [Ca²⁺]_i (indo-1 FF fluorometry) and C_m. The cell was whole-cell voltage-clamped at −80 mV. An UV flash elevated [Ca²⁺]_i to ∼25 µM and increased C_m by 0.296 pF. At 30 s after the UV flash, the amount of endocytosis (ΔC_mendo) was 0.071 pF. (B) The values of (i) peak [Ca²⁺]_i, (ii) exocytosis (ΔC_mexo normalized to initial C_m) and (iii) endocytosis (ΔC_mendo normalized to ΔC_mexo) at 30 s after the UV flash averaged from nine cells.

TEP imaging showed that robust exocytosis was followed by very slow endocytosis

Results from our C_m measurements in glomus cells showed that following an increase in C_m, the decline of C_m was very slow. While this finding can be interpreted as a very slow form of endocytosis, an alternative explanation is that endocytosis is underestimated due to the temporal overlap of exocytosis. To distinguish between these two scenarios, we employed TEP imaging to monitor the accumulation of FM1–43 fluorescence (Kasai et al., 2006). This experiment is based on the rationale that endocytosis would result in accumulation of FM1–43 stained granules inside the cell even after the removal of extracellular FM1–43. Figure 3(A) shows the typical fluorescent signals from two-photon excitation at the equatorial Z-plane of a glomus cell. In the absence of extracellular FM1–43, there was no detectable fluorescence (Figure 3(Ai)). Following the focal application of extracellular FM1–43, fluorescence signal was clearly detected at the optical section of the plasma membrane (basal FM1–43 fluorescence; Figure 3(Aii)). The long and short axes of this cell were 7.9 and 6.3 μm, indicating that glomus cell had roughly half the linear dimensions of a PC-12 cell (Liu et al., 2005). Note that in these experiments, an UV illumination of 0.5 s was employed to photolyse NP-EGTA (see Method) and t = 0 denoted the termination of the UV illumination. At 0.9 s after photolysis, a net increase in FM1–43 fluorescence (obtained by subtracting the basal FM1–43 signal in Figure 3(Aii) from Figure 3(Aiii)) could be detected clearly (Figure 3(Bii)), and additional increase in fluorescence could be detected at 2.4 s after photolysis (Figure 3(Biii)). Considering that TEP imaging cannot resolve the exocytosis of individual granule with diameter <200 nm (Kasai et al., 2005), and the diameter of SDCGs is only ∼100 nm, the capture of any image of exocytosis from individual SDCGs was not anticipated. As shown in Figure 3(Bii, iii), the increase in FM1–43 fluorescence after photolysis of caged Ca²⁺ occurred at the entire cell surface. The application of FM1–43 was terminated at ∼20 s after photolysis, and within a few seconds, the fluorescence at the rim of the cell decreased. At ∼23 s following the removal of FM1–43 (Figure 3(Av); i.e. 42.7 after photolysis), the fluorescence signal from the entire field decreased to a level close to that before FM1–43 application (Figure 3(Ai)) and there was no detectable net increase in fluorescence at any particular location inside the cell (Figure 3(Biv)). These observations suggested that very little FM1–43 dye was endocytosed in ∼20 s. To quantify the change in FM1–43 signals at different regions of the cell, we separated the cell into three regions. The first region is an annulus that included mainly the plasma membrane (outlined in Figure 3(Aii)). The second region is the cytoplasmic area of the cell (i.e. the smaller area surrounded by the annulus). The third region is the cross-sectional area of the entire cell (outlined by the outer rim of the annulus shown in Figure 3(Aii)). As shown in Figure 3(C), changes of the FM 1–43 fluorescence occurred predominantly at the plasma membrane area. There were only small parallel changes of FM1–43 signal in the cytoplasmic area, indicating that the changes in average fluorescence of the entire cell in the optical section arose mainly from the plasma membrane region. Note that the linear dimension of glomus cell was typically about half that of other secretory cells (e.g. PC-12, chromaffin and pancreatic β-cells) previously examined by TEP imaging (Hatakeyama et al., 2007; Kishimoto et al., 2006; Liu et al., 2005). Therefore, the small change in FM1–43 signals detected in the cytoplasmic region was probably due to the spread of scattered fluorescence from the plasma membrane. To estimate the amount of exocytosis triggered by the photolysis of caged Ca²⁺, we calculated the percentage increase in FM1–43 fluorescence by normalizing the amplitude of the post-photolysis increase in FM1–43 fluorescence in the plasma membrane (ΔF2; Figure 3(C)) to that of the steady-state signal associated with the application of FM1–43 before photolysis (F2; Figure 3(C)). Similar calculation was also applied to the entire cell (ΔF1 normalized to F1). As shown in Figure 3(D), the percentage increase in FM1–43 fluorescence for the annulus of plasma membrane area (solid line) and for the entire cell (dotted line) was essentially identical. At ∼3 s after photolysis, the increase in FM1–43 signal was ∼20%, and this increase approached a plateau of ∼30% just before the termination of FM1–43 application.

Figure 3. TEP imaging revealed that photolysis of caged Ca²⁺ triggered robust exocytosis. (A) Images of FM1–43 fluorescence (F) of a glomus cell (loaded with NP-EGTA-AM) recorded at different times before (−) or after (+) photolysis. FM1–43 dye was applied focally for ∼45 s (between −25 s and +20 s). After the removal of FM1–43, there was no detectable FM1–43 (F) in the cell (Image Av). (B) Net changes in FM1–43 fluorescence (ΔF) from the same cell. Images Bi–iii were obtained by subtracting Image Aii from Images Aii–iv. Note that following photolysis, there was a gradual increase in F (Bii–iii). The image in Biv (obtained by subtracting image Ai from Av) showed that there was very little increase in ΔF following FM1–43 removal. Note that ΔF was enhanced 5-fold relative to F in (A). (C) The increase in FM1–43 (F) triggered by photolysis of caged Ca²⁺ occurred predominantly at the plasma membrane. Plots of the changes in mean FM1–43 (F) in the cytoplasm, plasma membrane and the entire cell. Note that there was little increase in FM1–43 (F) at the cytoplasm. The horizontal bar indicates the duration of FM1–43 application. (D) The time course of increase in FM1–43 (F) at the plasma membrane (solid line; ΔF2 normalized to F2) and the entire cell (dotted line; ΔF1 normalized to F1) following photolysis.

As shown in Figure 3(C), there was a rapid decline in fluorescence after the termination of the FM1–43 application. At ∼40 s after photolysis, most of the FM1–43 signal at the plasma membrane was washed off. At this time, the average fluorescence from the entire cross section of the cell was only 4 arbitrary units above the background level (i.e. before the application of FM1–43) and this value corresponded to <10% retention of the post-photolysis increase in fluorescence averaged over the entire cell (ΔF1). At ∼100 s after photolysis (data not shown), the fluorescence averaged from the entire cell was barely detectable above background and corresponded to only 6.7% retention of the post-photolysis increase in fluorescence averaged from the entire cell (ΔF1). Since constitutive endocytosis would result in the cellular accumulation of FM1–43 even under basal (unstimulated) condition, we estimated that the amount of FM1–43 accumulation due to constitutive endocytosis in glomus cells by exposing unstimulated cells to FM1–43 (described in Methods). We found that unstimulated glomus cells retained 1.3% of the steady state FM1–43 fluorescence per minute of exposure to FM1–43. If we corrected for the anticipated signal arising from ∼45 s of constitutive endocytosis while extracellular FM1–43 was present, the amount of endocytosis for the cell shown in Figure 3 was only 3.8% of ΔF1. In 6 glomus cells examined with experimental protocol identical to Figure 3, photolysis of caged Ca²⁺ triggered a robust increase in FM 1–43 fluorescence (ΔF1 = 26.5 ± 5.3%). At ∼1 min after photolysis, the amount of endocytosis (after correction for constitutive endocytosis) was only 8.2% ± 6.8% of ΔF1.

In separate experiments, the exposure of glomus cells to the FM1–43 dye was increased to ∼3 min after photolysis. An example of the fluorescent images obtained with this experimental protocol is shown in Figure 4(A), and the net increases in FM1–43 signals (obtained by subtracting the basal FM1–43 signal in Figure 4(Aii) from Figure 4(Aii–iv)) were shown in Figure 4(B). An increase in FM1–43 signals could be clearly detected at 0.8 s and 2.4 s after photolysis (Figure 4(Bii, iii)). Note that at 390 s after photolysis (i.e. at ∼3 min the termination of FM1–43 application), there was detectable cytoplasmic retention of FM1–43 (Figure 4(Biv)). The changes in the average fluorescence at the cytoplasmic, plasma membrane and the entire cell were shown in Figure 4(C). After the FM1–43 fluorescence reached a maximum at the plasma membrane, there was a slow decline in the fluorescence, which was accompanied by a slow increase in fluorescence at the cytoplasmic area (Figure 4(C)). The slow retention of FM1–43 fluorescence in the cytoplasm suggested that the cell underwent endocytosis while FM1–43 was present in the extracellular solution. For cells exposed to the FM1–43 dye for ∼3 min after photolysis, the average increase in FM1–43 fluorescence (ΔF1) was 25.6 ± 3.7% (n = 3) similar to that detected in cells with <1 min exposure to FM1–43 (26.5 ± 5.3%; n = 6). To estimate the amount of endocytosis, the amount of FM1–43 fluorescent retained in the cells after the removal of FM1–43 (ΔFw; e.g. Figure 4(C)) was normalized to ΔF1 and then corrected for constitutive endocytosis. As summarized in Figure 4(D), the amount of endocytosis detected by TEP imaging at <1 min after the UV flash was 8.2%±6.8% (estimated from experiments similar to Figure 3(C)) and increased to 28.6%±5.5% at ∼3 min after the UV flash (estimated from experiments similar to Figure 4(C)).

Figure 4. Endocytosis that occurred within 1 to 3 min after photolysis as detected in TEP imaging. (A) and (B), FM1–43 fluorescence images (F) and difference images (ΔF) were obtained as described in the legend of Figure 3. Note that, in Image Biv, (i.e. 390 s after photolysis or ∼190 s after the end of FM1–43 application) the cytoplasmic region of the cell retained detectable fluorescence. (C) Plots of the changes in mean FM1–43 (F) in the cytoplasm, plasma membrane and the entire cell (as outlined in Image Aii). The duration of FM1–43 application was indicated by the horizontal bar. The post-photolysis increase in fluorescence in the entire cell (ΔF1) and the fluorescence retained after the removal of FM1–43 (ΔF_W) were indicated. (D) The amount of endocytosis (ΔF_W normalized to ΔF1 and then corrected for “constitutive endocytosis”) at ∼45 s (n = 6) or 3 min (n = 3) after photolysis.

Glomus cells had a lower abundance of clathrin and dynamin

One possible explanation for the very slow endocytosis in glomus cell is a lower abundance of the proteins involved in endocytosis. To test this possibility, we employed immunocytochemistry to examine the expression of key proteins of endocytosis in the cytoplasm of glomus cells. In chromaffin cells, the ‘slow’ endocytosis was found to be dynamin-II and clathrin-dependent while the ‘rapid’ endocytosis was dynamin-I dependent but clathrin-independent (Elhamdani, Azizi, Solomaha, Palfrey, & Artalejo, 2006b). Therefore, we compared the cytoplasmic densities of dynamin I, II and clathrin in glomus cells with those of rat chromaffin cells. We also examined the cytoplasmic densities of several proteins involved in the cytoskeleton (actin, myosin and tubulin) because perturbation of actin and myosin had been reported to influence the kinetics of the fusion pore (Berberian, Torres, Fang, Kisler, & Lindau, 2009). Since the carotid body contains both glomus and sustentacular cells, we included the sustentacular cells in our comparison. The sustentacular cells are glial-like cells, which do not contain any dense core granules (Xu, Tse, & Tse, 2003), but have been shown to release ATP via pannexin-1 channels (Zhang, Piskuric, Vollmer, & Nurse, 2012). Thus, sustentacular cells are expected to have low densities of proteins involved in endocytosis. The fluorescent images of glomus cell, sustentacular cell and chromaffin cells when immunostained for dynamin I, chathrin and the nucleus (with DAPI) were shown in Figure 5(A). Since the cytoplasmic area for the three cell types were clearly different (e.g., the mean diameter of chromaffin cells is ∼2-fold of that of glomus cells), we measured the cytoplasmic density (i.e. excluding the area of the nucleus) of immunofluorescence signals at a 0.8 µm confocal plain at half of the depth of the Z-axis of individual cells. The average cytoplasmic density of a specific protein for the different cell types was then normalized that of glomus cells. As shown in Figure 5(B), the cytoplasmic densities of the different cytoskeletal proteins (actin, tubulin and myosin) were similar among the glomus, sustentacular and chromaffin cells. However, the cytoplasmic densities of dynamin I, II and clathrin in glomus and sustentacular cells were only about half of those in chromaffin cells. The low cytoplasmic densities of dynamin I, II and clathrin in glomus cells raised the possibility that these cells have a smaller number of endocytotic machinery.

Estimation of the vesicular diameter with TEPIQ

Previous TEPIQ studies that employed the ΔV/ΔS analysis (see Methods) at the narrow intercellular areas within a small cluster of cells were successful in estimating the mean vesicular diameter from a population of granules undergoing exocytosis, even when individual granules were too small to be resolved (Hatakeyama et al., 2007; Liu et al., 2005). We employed this analysis to verify whether the robust exocytosis triggered in glomus cells was dominated by structures with a mean vesicular diameter of ∼100 nm (as expected of SDCGs). This analysis is based on the following rationale. When a granule undergoes exocytosis with SRB (a fluid-phase fluorescent dye) in the extracellular fluid (Figure 6(A)), that granule will transiently gain fluorescence that is proportional to its vesicular volume (ΔV). If FM1–43, a fluorescent dye that selectively stains lipid membranes (and has a different emission wavelength) is also present in the extracellular fluid during the exocytosis of this granule, the addition of the vesicular membrane area to the plasma membrane (ΔS) will be reflected as an increase in FM1–43 fluorescence. Assuming that the granule is spherical, the vesicular diameter can be calculated from the value of 6ΔV/ΔS (Kasai et al., 2005). The above analysis can only be performed at cell surfaces located next to a narrow intercellular space in intact tissue or deep inside a small cluster of cells because a narrow intercellular space is crucial for reducing the otherwise large background signals of the extracellular SRB and FM1–43 dyes.

Therefore, in our TEPIQ analysis, we employed clusters of carotid cells (the linear dimension of the entire cell cluster in each axis was typically <50 µm). Previous studies on the structure of the carotid body had established that glomus cell has a roughly ovoid or polygonal cross-sectional profile (occasionally with short processes) while sustentacular cell has a spindle-shape in its long axis (Duchen et al., 1988; Gronblad, 1983; Kameda, 1996). By exploiting the optical sectioning capability of TEP imaging over the entire Z-axis of carotid cell clusters, we could identify individual cells that not only had the morphology of glomus cells, but also share at least one major intercellular space with another cell with a similar morphology (as shown by the arrows in Figure 6(A,B)). We performed ΔV/ΔS analysis on seven such intercellular spaces (from different cell pairs, summarized in Figure 5(C)) where the increase in both signals at >5 s after photolysis were at least 3-fold larger than the pre-photolysis baseline noise. Our analysis indicated that within 5–10 s after photolysis (while the fluorescent signals from both dyes were still increasing), the vesicular diameter of the granules undergoing exocytosis was 94.5 ± 10.7 nm (range of 62–137 nm). When the same quantification was performed from the same sites at 10 to 20 s after photolysis (while both signals approached a plateau), the vesicular diameter was 75.4 ± 8.9 nm (range of 57–117 nm). A possible explanation for the small decrease in the estimated vesicular diameter at longer time duration after photolysis is that the lumen of SDCGs that undergo full fusion may eventually become flattened or shrink (Shin et al. 2018) at the plasma membrane, resulting in an underestimation of ΔV/ΔS. We did not find any of the structures suggestive of compound exocytosis, such as vesicle chain from the plasma membrane into the cytosol (Lam et al., 2013; Takahashi et al., 2004) or vacuolar structures (Kishimoto et al., 2006).

Figure 5. Glomus cells had a lower cytoplasmic density of proteins (dynamin I, II and clathrin) that mediated endocytosis. (A) Fluorescent images of glomus cells, sustentacular cells and chromaffin cells when immunostained for dynamin I, clathrin or DAPI. The horizontal scaled width of each image shown is 28 μm. (B) Comparison of the cytoplasmic densities of actin, tubulin, myosin, dynamin I, II and clathrin among glomus cells, sustentacular cells and chromaffin cells. Immunofluorescence were averaged from 9 to 21 cells for each protein and then normalized to the values obtained from glomus cells. Each asterisk denotes significant difference from glomus cells.

Discussion

Glomus SDCGs underwent full fusion exocytosis at high [Ca²⁺]_i elevations

In this study, we examined exocytosis and endocytosis in glomus cells by employing C_m measurement and TEP imaging. Our results show that glomus SDCGs underwent full fusion when [Ca²⁺]_i was elevated to several µM. Activation of VGCCs (via a train of depolarizing voltage steps) which elevated [Ca²⁺]_i to ∼1.6 μM, increased C_m by ∼2.5% (Figure 1). For a typical glomus cell with an initial C_m of ∼3 pF, the 2.5% increase in C_m will correspond to the addition of 75 fF. Assuming that the exocytosis of a glomus SDCG contributes to an increase of 0.28 fF, a robust stimulation of VGCCs will result in the full fusion of ∼250 SDCGs. Consistent with the notion that a larger Ca²⁺ signal not only triggers more exocytosis but also shifts the mode of exocytosis from kiss-and-run to full fusion (Elhamdani, Azizi, & Artalejo, 2006a), the elevation of [Ca²⁺]_i to ∼16 µM (via flash photolysis of caged Ca²⁺), increased the C_m by ∼13% (Figure 2), reflecting the full fusion of ∼1400 SDCGs. In separate experiments, when [Ca²⁺]_i was elevated to 35.9 ± 7.6 µM, the average amount of exocytosis in glomus cells increased to 19.6 ± 1.2% (n = 3; data not shown). An estimate on the number of granules in different releasable pools in glomus cells from published electron micrograph (Kusakabe, Powell, & Ellisman, 1993; Verna, 1979) has not been possible because these studies employed chemical fixation instead of rapid freeze fixation (Plattner, Artalejo, & Neher, 1997). Rapid elevation of [Ca²⁺]_i to >15 µM via flash photolysis of caged Ca²⁺ has been reported to deplete the readily releasable pool of granules in neuroendocrine cells such as chromaffin cells (Xu, Xu, Binz, Niemann, & Neher, 1998), pituitary gonadotropes (Tse, Tse, Hille, Horstmann, & Almers, 1997) and corticotropes (Tse & Tse, 1998). The depletion of the readily releasable pool of granules in corticotropes and gonadotropes were found to increase C_m by 2.8 and 4.7% respectively (Tse et al., 1997; Tse & Tse, 1998). In view of the 13% increase in the C_m of glomus cells when [Ca²⁺]_i was elevated to ∼16 µM, it is possible that such C_m increase reflects the release from more than one pool of granules in glomus cells.

The robust exocytosis in glomus cells with photolysis of caged Ca²⁺ was confirmed with TEP imaging using the membrane dye, FM1–43. As shown in Figure 3, photolysis of caged-Ca²⁺ induced a rapid increase in FM1–43 fluorescence which was localized mostly in the plasma membrane. On average, the FM1–43 fluorescence in glomus cells increased by 26.5 ± 5.3%. To examine whether high [Ca²⁺]_i elevation triggered compound exocytosis in glomus cells, we estimated the vesicular diameter with TEPIQ analysis. Since sequential compound exocytosis of chromaffin LDCGs was found to occur preferentially at sites of plasma membrane facing the intercellular space (Kishimoto et al., 2006), we employed a semi-intact preparation of clusters of carotid cells in our TEPIQ experiments. Our TEPIQ analysis of ΔV/ΔS within 5–10 s after photolysis estimated that the mean vesicular diameter of granules that underwent exocytosis at the plasma membrane areas between pairs of glomus cells was 94.5 ± 10.7 nm (range of 62–137 nm). This value is consistent with a previous electron microscopy study that reported an average vesicular diameter of 90 ± 11 nm (range 50 to 130 nm) and 116 ± 20 nm (range 50–170 nm) in glomus cells (McDonald & Mitchell, 1975). Overall, our results suggest that glomus SDCGs do not undergo compound exocytosis (which can lead to a larger estimated vesicular diameter in the ΔV/ΔS analysis), and the dominant mode of exocytosis at high [Ca²⁺]_i is full fusion.

Slow endocytosis in glomus cells

Results from our measurements of C_m in glomus cells showed that following exocytosis, the decline of C_m was very slow. In experiments where activation of VGCCs was employed to evoke exocytosis, only ∼5% of the vesicular membrane inserted into the plasma membrane by exocytosis, was retrieved by endocytosis at 30 s after the termination of depolarization (Figure 1(Biii)). The slow endocytosis detected in glomus cells probably reflects the mode of classical endocytosis that typically follows full fusion exocytosis (Wu et al., 2014). In corticotropes (Lee & Tse, 2001) and chromaffin cells (Engisch & Nowycky, 1998), 2/3 of the vesicular membrane was retrieved within 4–6 s via the classical slow endocytosis. Thus, the rate of endocytosis in glomus cells seems to be even slower than the classical endocytosis in other secretory cells. When [Ca²⁺]_i was elevated to ∼16 µM via flash photolysis of caged Ca²⁺ in glomus cells, the larger amount of exocytosis was followed by a faster rate of endocytosis, such that ∼30% of the vesicular membrane was endocytosed at 30 s after the UV flash. The increase in the rate of endocytosis with Ca²⁺ in glomus cells is consistent with the notion that the rate of endocytosis is regulated by Ca²⁺ (Wu et al., 2014). In secretory cells such as corticotropes (Lee & Tse, 2001) and chromaffin cells (Artalejo, Henley, McNiven, & Palfrey, 1995), the classical slow endocytosis is typically preceded by large rapid decrease in C_m, reflecting rapid endocytosis (or bulk endocytosis) when [Ca²⁺]_i rise was high. In glomus cells, however, no rapid endocytosis was detected even when [Ca²⁺]_i was elevated to >20 µM (e.g. Figure 2). Consistent with a very slow rate of endocytosis in glomus cells, TEP imaging showed that there was almost no retention of cellular FM1–43 fluorescence at 20 s after the washout of the FM1–43 dye (Figure 3(C)), suggesting that there was little endocytosis within 20 s. This result contrasted that of SV in PC-12 cells in which the increase in FM1–43 was not affected by the washout of the dye, reflecting the retrieval of most of the vesicular membranes by endocytosis within 10 s (Liu et al., 2005).

We found that even at 100 s after the UV flash, glomus cells retrieved only ∼40% of the vesicular membrane. In contrast, our previous study in corticotropes has shown that C_m returned to baseline within 10 s after the UV flash (Lee & Tse, 2001). The very slow classical endocytosis in glomus cells was correlated to the low densities of dynamin I, II and clathrin. It is generally accepted that the classical endocytosis in secretory cells is clathrin-dependent (Wu et al., 2014). In bovine chromaffin cells, pharmacological inhibitions of dynamin-II or clathrin were found to reduce slow endocytosis, and rapid endocytosis could be inhibited by the disruption of dynamin-I (Artalejo, Elhamdani, & Palfrey, 2002; Elhamdani, Azizi, Solomaha, Palfrey, & Artalejo, 2006b; Tsai et al., 2009). Since the cytoplasmic densities of dynamin I, II and clathrin in glomus cells were less than half of those in chromaffin cells (Figure 5(B)), it is probably that fewer endocytic machinery contribute to the slower classical endocytosis in glomus cells. The slow rate of endocytosis in glomus cells is probably related to their normally low secretory output for the paracrine/autocrine regulation of the carotid body functions. Thus, it is not essential for the glomus SDCGs to be rapidly replenished with transmitters. In contrast, chromaffin cells release large amount of hormones into the circulation during the ‘fight or flight response’ and a rapid replenishment of chromaffin granules would be crucial.

Physiological significance

Our results show that with a large [Ca²⁺]_i rise, exocytosis of glomus SDCGs can shift from the kiss-and-run mode to full fusion, resulting in a more complete release of transmitters. During a severe hypoxic challenge, the cytosolic [Ca²⁺] of glomus cells (averaged from the entire cell) increased to ∼1 to 1.6 μM (Buckler & Vaughan-Jones, 1994). Since the hypoxia-evoked [Ca²⁺]_i rise is mediated via activation of VGCCs, it is conceivable that the Ca²⁺ entry via VGCCs generates a spatial Ca²⁺ gradient, such that the secretory granules near the vicinity of VGCCs are exposed to a local Ca²⁺ concentration which is higher than the cytosolic average. In chromaffin cells, the local Ca²⁺ near the VGCCs was estimated to be 10–100 μM (Garcia, Garcia-de-Diego, Gandia, Borges, & Garcia-Sancho, 2006). It is not clear whether secretory granules in glomus cells are tightly coupled to VGCCs. Nevertheless, it is conceivable that some glomus granules near the VGCCs may be exposed to [Ca²⁺] in the range of 10 μM and thus undergo full fusion for a more complete release of transmitters. This would result in a more robust stimulation of the carotid sinus nerve and the triggering of the respiratory and cardiovascular reflexes.

The ability of glomus SDCGs to undergo different modes of exocytosis, ranging from kiss-and-run to full fusion may also have an important role in the paracrine and autocrine signalling within the carotid body. As shown in Figure 6, exocytosis can detected at the interface between two glomus cells within a cluster of carotid cells. Glomus SDCGs release multiple transmitters including ATP, acetylcholine and dopamine (Nurse, 2010). Receptors for some of these transmitters (e.g. ATP) are not only expressed on the carotid nerve terminals, but also on glomus cells (Xu et al., 2005) and sustentacular cells (Xu et al., 2003). Therefore, the local concentrations of the transmitters released to the neighboring glomus cells, sustentacular cells and nerve terminals influence the autocrine/paracrine signaling and thus shape the response of the carotid body to hypoxia (Tse, Yan, Lee, & Tse, 2012).

Figure 6. TEPIQ analysis of vesicular diameter. A cluster of carotid cells simultaneously stained with SRB (A) and FM1–43 (B). The arrows indicated an intercellular space between two ovoid shaped glomus cells where ΔV/ΔS analysis was performed. The horizontal scale for each image shown is 50 μm. (C) The mean vesicular diameter estimated from ΔV/ΔS analysis at the intercellular space between seven different pairs of glomus cells at two ranges of post-photolysis durations: (i) 5 to <10 s (when the signals for ΔV and ΔS were still rising); and (ii) 10 to 20 s (when the signals for ΔV and ΔS reached a plateau). At each post-photolysis duration, the circles indicated the range, while the squares were the mean ± SE of the estimated vesicular diameter.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by operating grants from the Canadian Institute of Health Research (MOP-79291 to AT, MOP-12070 to FWT) and Natural Science and Engineering Research Council (RGPIN-2016-04653 to FWT); a Visiting Professorship from the University of Tokyo (to FWT), CREST (JPMJCR1652 to HK) from JST; Grant-in-Aid (No. 26221001 to HK), AMED (JP18 dm 01017120 to HK) and World Premier International Research Center Initiative (WPI) to HK from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan.