INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The chemoreceptor (glomus) cells of the rat carotid body are electrically coupled because of the presence of gap junctions between them (Kondo and Iwasa 1996; McDonald 1981). During hypoxia or extracellular acidification, most cells (~70%) partially uncouple, whereas the rest show tighter coupling (Abudara and Eyzaguirre 1996b, 1998; Abudara et al. 2001; Monti-Bloch et al. 1993). It has been suggested that coupling and uncoupling are integral processes in the secretion of transmitters by the glomus cells because they regulate intercellular exchanges (Eyzaguirre and Abudara 1999).

In this study, we sought to explore the behavior of intercellular channels during normoxia and hypoxia induced by the superfusion of cultured and clustered glomus cells with sodium dithionite (Na₂S₂O₄) or saline equilibrated with 100% N₂. Furthermore, we tried to get an insight into possible mechanisms underlying the effects of decreased O₂ on intercellular coupling. Two possibilities are widely accepted as likely factors in triggering the effects of hypoxia on cells. One is intracellular acidification, and the other is an increase in intracellular Ca²⁺ concentration ([Ca²⁺]_i). In cultured and clustered glomus cells, hypoxia induced by either Na₂S₂O₄ or 100% N₂ reduces intracellular pH (pH_i) in ~60% of the cells, whereas it increases pH_i in the others (He et al. 1991b; Pang and Eyzaguirre 1993a). Concerning [Ca²⁺]_i, this ion also increases during hypoxia (Buckler and Vaughan-Jones 1994; Pietruschka 1985; Sato et al. 1991; Zhang and Eyzaguirre 1999) and this effect depends on the severity of the hypoxia (Dasso et al. 2000). Also, the coupling between glomus cells is sensitive to the levels of extracellular Ca²⁺ ([Ca²⁺]_o), decreasing during superfusion with high [Ca²⁺]_o and tightening when [Ca²⁺]_o is removed (Abudara and Eyzaguirre 1996a). Because [Ca²⁺]_i follows the extracellular concentrations of this ion (Jiang and Eyzaguirre, unpublished observations), we needed to establish whether there was a correlation between hypoxia and changes in [Ca²⁺]_i.

In the experiments described below, we established that 100% N₂-induced hypoxia was more effective than Na₂S₂O₄-induced hypoxia in uncoupling or coupling glomus cells. However, both agents had similar effects on [Ca²⁺]_i. Consequently, hypoxic uncoupling may be influenced by factors other than those produced by pH_i and/or [Ca²⁺]_i changes, as presented in DISCUSSION.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The methods used have been described in detail in a recent publication (Abudara et al. 2001). For the convenience of the readers, the salient points are presented here. Two procedures were employed: 1) voltage clamping of two adjoining glomus cells and 2) [Ca²⁺]_i measurements.

Voltage clamping

Cultures of glomus cell clusters were prepared as described previously (Abudara and Eyzaguirre 1998; Abudara et al. 2001). Briefly, carotid bodies were removed from Wistar rats (40-60 g) anesthetized with 50 mg/kg (ip) pentobarbital sodium. The organs were thoroughly rinsed with sterile Hank's balanced salt solution and then immersed in serum-free growth medium for mechanical dissociation. After the cells settled at the bottom of collagen-coated Petri dishes, the cultures were incubated in a humid atmosphere (5% CO₂-95% air at 37°C) for 1-7 days. The culture medium was a mixture of Dulbecco's modified Eagle's medium and Ham's F-12 (Sigma, St. Louis, MO), supplemented with 0.014 M NaHCO₃, 80 U/l of insulin, and 1% penicillin-streptomycin-fungizone (Gibco). The pH was adjusted to 7.4 with HEPES-NaOH. dB-cAMP (1-3 mM) was added to the cultures to facilitate the formation of intercellular junctions (Chanson et al. 1996; Kessler et al. 1985; Matesic et al. 1996) as has been done in previous work (Abudara and Eyzaguirre 1998; Abudara et al. 1999, 2000).

The cultures were transferred to a 1-ml chamber mounted on the stage of an inverted phase-contrast microscope. The preparation was superfused at room temperature with Ham's F-12 equilibrated with 100% O₂ (pH 7.4), flowing at 1 ml/min. PO₂ in the bath was ~300 Torr. Two microelectrodes (filled with 3 M KCl, 10-20 M), were mounted on the same micromanipulator and positioned on the tissue at ×450 magnification. The intracellular microelectrodes were independently connected to the input stages of feedback amplifiers (WPI S-7050A, Patch Clamp Systems) for current or voltage clamping. The output of each amplifier was low-pass filtered at 100 Hz to reduce noise. Hypoxia was produced by superfusing the preparations with 1 mM Na₂S₂O₄ or by equilibrating the superfusate with 100% N₂. NaOH was used to adjust the extracellular pH to 7.43 when superfusing with Na₂S₂O₄. pH and PO₂ were monitored in the bath with small specific electrodes.

To monitor macrojunctional conductance (G_j) between glomus cells, both cells were voltage clamped at a value intermediate between their respective resting potentials, and command pulses were applied to one cell, inducing a voltage drop (V_j) across the intercellular junction. The current produced in the coupled cell was equal in amplitude to the junctional current (I_j); thus G_j = I_j/V_j (Abudara and Eyzaguirre 1998). To study channel conductances, a constant (DC) V_j was used, thus clamping each cell at a different level. The activity of intercellular channels (gating) appeared as simultaneous current step changes (i₁ and i₂) of opposite polarity (mirror images). Thus intercellular channel activity could be distinguished from nonjunctional noise, which had the same polarity in both recordings. To calculate intercellular channel conductance (g_j), we used one-half of the difference in activity of both coupled cells; thus g_j = (i₁-i₂)/2 × V_j¹. This method takes into account outward and inward currents and eliminates or sharply reduces nonjunctional noise, which can alter results if one uses only currents entering one of the coupled cells.

Two methods were employed to estimate the conductance of single channels. 1) The g_j variance (pS²) was divided by the mean g_j (Nicholls et al. 1992), and 2) current transitions from one level to the next (without obvious intermediate steps) were measured by hand from the Scope records. Channel conductance (g_j) was calculated by dividing the amplitudes of current steps (pA) by V_j (mV).

The mean open time of single-junction channels was calculated with noise (fluctuation) analysis. The variances of the conductances (² in pS²) of each experiment were transferred to the Scope program, plotting variance versus time. This program permitted calculation of the mean variance of all sweeps in the time domain. A cosine tapered fast Fourier transform gave the power density of these values (pS² · s) as a function of frequency (Hz). A log/log plot was constructed with a Cricket Graph program to obtain an exponential fit of the data. The corner frequency (F_c), taken as one-half of the low-frequency asymptote, permitted the conversion of Hz to ms as  = 1/[2F_c] (Anderson and Stevens 1973; Gold and Martin 1983; Hille 1992; Nicholls et al. 1992).

Voltage and current recordings were simultaneously stored on a Vetter videotape system and on-line in a Macintosh computer through a Mac Lab/4 interphase. Data were acquired by the Scope v. 3.2.8 program. Once stored in the computer, the information was retrieved in digital form and transferred to Statview spread sheets for quantification and analyses.

Measurement of [Ca²⁺]_i

Carotid bodies were removed and immersed in an ice-cold solution containing (in mM) 98 NaCl, 47 Na-glutamate, 4.6 KCl, 3 CaCl₂, 1.1 MgCl₂, 7 glucose, and 5 HEPES, pH 7.4. The organs were then placed in Ca²⁺- and Mg²⁺-free Hanks' medium (Sigma) at room temperature for 30 min before they were moved to a Ca²⁺- and Mg²⁺-free Hanks' solution containing 0.3% collagenase Type II (Sigma) for 40 min at 37°C. After being washed in Ham's F-12, the carotid bodies were gently dissected in Ham's F12 nutrient medium, which also contained 80 µl of insulin (Sigma), 10% bovine serum, and 1% antibiotic-antimycotic (Gibco) at pH 7.43. The cells were plated onto poly-L-lysine-coated glass coverslips and kept in an incubator at 37°C for 1-2 h for adhesion to the glass surface.

The cells were loaded with 1 µM Fura-2 AM and 0.02% pluronic F-127 (Molecular Probes, Eugene, OR) for 10-20 min at 37°C. The preparation was mounted in a superfusion chamber (100 µl) placed on the stage of an inverted digital fluorescence microscope (Attofluor TM; Zeiss) equipped with a CCD camera and computer. It was continuously superfused at 0.5-1 ml/min with solutions containing (in mM) 135 NaCl, 5 KCl, 2 CaCl₂, 2 MgCl₂, and 10 glucose, pH adjusted to 7.43 with 10 mM HEPES-NaOH at 27-30°C. As in the previous series, PO₂ was lowered by either Na₂S₂O₄ or 100% N₂. Bathing solutions flowed through a 500-µl loop, exposing the tissues for 20-30 s to the environments.

Cells were viewed through a ×40 oil-immersion objective and excited through this lens. Fluorescence ratios were obtained by illumination through intermittent 334- and 380-nm excitation filters and a 520-nm emission filter. Ratios were converted to [Ca²⁺]_i by a two-point calibration formula
where R is the measured fluorescence ratio K_d is the dissociation constant of the dye-Ca²⁺ complex, R_(Lo) is the fluorescence ratio given by low-Ca²⁺ solutions, R_(Hi) is the ratio obtained with high-Ca²⁺ standards, Den_(Lo) is the denominator intensity for low Ca²⁺, and Den_(Hi) the denominator intensity for high Ca²⁺. Calibrations were done with standard Ca²⁺ solutions as reported by Brown and Owen (1979). Results were stored in the computer and analyzed after transfer to Statview spread sheets and to the Maclab Scope program.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We studied 28 electrically coupled cell pairs with different degrees of coupling between them. The membrane potential (E_m) of the cells varied from 79 to 14 mV [mean 28 ± 0.8 mV (SE)], with a mean input resistance of 181.6 ± 18.74 M. These values were lower than those previously obtained from single impalements in cultures (34.7 ± 0.53 mV; n = 101) (He et al. 1991a,b) and those calculated by Fieber and McCleskey (1993) with noninvasive methods (approximately 32 mV). However, our impalements (even producing low E_m) were characterized by sharp and stable negative shifts of the baseline. In any case, and concerning the results presented below, there was no correlation between the E_m of the cells and intercellular coupling within ±40 mV from the resting potential (Abudara and Eyzaguirre 1998) as is generally the case (Spray and Bennett 1985).

Macroconductance and channel activity during normoxia

The junctional macroconductance (G_j) varied from 0.04 to 24.36 ns, with a mean (±SE) of 3.0 ± 1.01 ns (see also Abudara et al. 2001). There was multiple-channel activity in our recordings because, as shown below, there are from very few to >1,000 junction channels between glomus cells. Channel activity (flickering) was seen only when a mean transjunctional voltage (V_j) of 100.1 ± 10.9 mV (range 40-190 mV) was applied (Abudara et al. 2001).

Figure 1 shows six double traces of a long recording of two coupled glomus cells (traces 1 and 2). Cell 1 was voltage clamped at 30 mV (its resting potential), whereas cell 2 was held at 180 mV, creating a transjunctional voltage of 150 mV. Intercellular activity was voltage dependent, but the threshold was high. Channel flickering appeared only during a DC V_j of from 40 to 190 mV, with no signs of channel desensitization or rectification for seconds or minutes. Intercellular channel currents appeared as deflections of similar amplitude and opposite polarity (mirror images) in both recordings. Cell 2 showed inward currents, whereas outward currents were recorded in cell 1. Thus when the traces separate, the channels opened. The largest and slowest deflections had superimposed multiple flickering, probably representing partial channel closing. The durations of the openings varied from <1 s to several seconds. These multiple channel recordings are similar to those obtained in coupled heart myocytes (Veenstra and DeHaan 1988).

View larger version (27K): [in this window] [in a new window]   Fig. 1. Six samples of intercellular channel activity recorded from same impalements of two coupled cells (1 and 2). Cell 1 was voltage clamped at its resting potential (Em1 and VH1 = 30 mV), whereas cell 2 (Em2 = 44 mV) was clamped at 180 mV from Em2 = 44 mV, creating a voltage drop (Vj) of 150 mV. Currents flowed from cell 1 to cell 2, and the channels opened when the traces are seen separated.

In 17 control experiments, multiple intercellular channel conductances (g_j) were calculated as i_j/V_j. The mean g_j was 104.44 ± 10.165 pS, and there was a significant (P < 0.04; r = 0.513) and direct correlation with intercellular macroconductance (G_j) as shown in Fig. 2A.

View larger version (30K): [in this window] [in a new window]   Fig. 2. A: significant and positive correlation (r = 0.513) between mean microconductances (gj; pS) of intercellular channels and the intercellular macroconductances (Gj; ns) in 17 pairs of coupled glomus cells under normoxic conditions. Exact coincidences are shown by different symbols. B: percentile distribution of changes in intercellular Gj during superfusion with 100% N2 and 1 mM Na2S2O4 (Na-DTN). Changes are expressed as test (T)/control(C) ratios (Gj/Gj).

Effects of hypoxia on intercellular coupling

G_J CHANGES. Short superfusions with 1 mM Na₂S₂O₄ decreased PO₂ in the bathing saline to ~10 Torr in 200 s (Abudara and Eyzaguirre 1998) and partially uncoupled (decreased G_j) ~65% of glomus cell pairs, whereas G_j increased in the rest (solid surface in Fig. 2B). A less severe hypoxia (~60 Torr) occurred during superfusion with saline equilibrated with 100% N₂, and it took a bit longer (250 s) to reach this value (Abudara and Eyzaguirre 1998). Coupling between glomus cells followed a similar pattern because the same proportion of cells uncoupled (65%) or coupling tightened (25%). However, both effects were more marked under 100% N₂ (dotted surface in Fig. 2B (Abudara and Eyzaguirre 1996b).

Intercellular channel activity during Na₂S₂O₄

Figure 3 illustrates two different experiments in which this reducing agent induced opposite effects on coupling between glomus cells. The mean g_j values (obtained from 6.4-s recordings) were plotted against the recording time. The depressant effect of Na₂S₂O₄ on intercellular coupling and the increased coupling produced by this agent are shown in Fig. 3, A and B, respectively.

View larger version (27K): [in this window] [in a new window]   Fig. 3. Different effects of hypoxia induced by 1 mM Na-DTN on 2 coupled pairs in the time domain. Intercellular channel conductance decreased (A) but increased in (B). Each point represents mean gj obtained during a 6.4-s interval.

Figure 4 shows an example of the depressant effect of 1 mM Na₂S₂O₄ on total intercellular channel activity. Cell 1 was voltage clamped at +6 mV (E_m1 = 24 mV), whereas cell 2 (E_m2 = 27 mV) was held at 99 mV, producing a V_j of 105 mV. Thus current flowed from cell 1 to cell 2. A 38.4-s recording, split into six double traces, during the control period is shown in Fig. 4A. When the channels are shown open, the traces are separated. Figure 4B depicts channel activity during superfusion with 1 mM Na₂S₂O₄. Inspection of the left and right traces clearly shows that channel activities, seen in the control situation were blunted by Na₂S₂O₄. The control and Na₂S₂O₄ i_j were measured from a baseline (lowest values) after digitizing all traces, and g_j were calculated as i_j/V_j. As seen in the illustration, the mean g_j in the controls (102.6 ± 4.6 pS) significantly (P < 0.035 by Kolmogorov-Smirnov test) decreased to 65.9 ± 4.67 pS.

View larger version (29K): [in this window] [in a new window]   Fig. 4. Example of the depressant effect of 1 mM Na-DTN on intercellular channel conductance. A: 6 double traces of conductances during control period. B: traces recorded during the effect of the reducing agent. Mean conductance went from 102.6 ± 4.6 pS in controls to 65.9 ± 4.67 pS during hypoxia. A and B, bottom, values for resting and holding potentials and Vj. C: same experiment shows an amplitude distribution histogram (%occurrence) of control conductances (thin line) and conductances obtained during hypoxia (thick line). Note that hypoxia shifted the distribution curve to the left. D: difference of histograms presented in C (%occurrence), showing a marked increase in smaller conductances and disappearance of larger ones.

The channel conductances obtained in this experiment were grouped in amplitude histograms at 5-pS intervals, giving counts and percentage of occurrence for each interval. These analyses are shown in Fig. 4C. The thin line represents the distribution of conductances (in %) under normoxic conditions, and the thick trace is the conductance distribution during the severe hypoxia induced by 1 mM Na₂S₂O₄. It should be noted that the peak of the distribution in the controls (~100 pS) shifted to the left (~60 pS) because the larger conductances decreased, and there was predominance of the smaller conductances. This is better seen in Fig. 4D, which is a differential histogram of these measurements showing the shift in the proportions of conductances before and after Na₂S₂O₄. Further evidence that the cells uncoupled was established by the fact that G_j decreased from 2.3 to 1.4 nS.

The opposite effect, that is, increased intercellular channel activity induced by Na₂S₂O₄, is illustrated in Fig. 5. Figure 5A shows two superimposed 6.5-s traces in which currents have been converted to conductances in the control (bottom trace) and during superfusion with Na₂S₂O₄ (top trace). The intercellular channel conductance practically doubled, increasing from ~75 to 130 pS. The statistical significance between the two traces was high (P < 0.0002 by Kolmogorov-Smirnov test). Figure 3B presents histograms of conductance distribution (%) during normoxia (thin line) and during Na₂S₂O₄ superfusion (thick continuous trace). Differences in occurrence are shown by the broken thick line. The smaller conductances (round a peak of 50 pS) decreased and eventually disappeared during Na₂S₂O₄ hypoxia. Also, this reducing agent induced the appearance of larger conductances that were absent during normoxia. In this experiment, G_j increased from 0.41 to 0.57 nS, denoting increased coupling.

View larger version (35K): [in this window] [in a new window]   Fig. 5. A: example of enhancement of intercellular channel conductance by Na-DTN shown by superimposed control and test traces. Control trace shows mean values of 21 sweeps in time domain; Na-DTN (1 mM) trace shows means of 28 sweeps over time. µ, Overall mean gj that went from 73.2 ± 0.15 pS in controls to 128.9 ± 0.42 during hypoxia. B: amplitude distribution of conductances (%occurrence) in controls (thin line) and during Na-DTN (thick line). - - -, Decrease of smaller conductances (<100 pS) and marked increase in larger conductances (%occurrence).

Figure 6A is a composite histogram, grouping results from all experiments on the effects of Na₂S₂O₄ on intercellular conductance. There were no significant differences (Wilcoxon signed-rank test) between the controls and during Na₂S₂O₄ hypoxia for the simple reason that conductance decreased in some experiments but increased in others. This variability is well illustrated in Fig. 6B, which is the differential histogram of results presented in Fig. 6A.

View larger version (28K): [in this window] [in a new window]   Fig. 6. A: amplitude distribution (%occurrence) of intercellular conductances in all experiments with Na-DTN. Thin line, distribution of control gj (n = 144); thick line, same distribution during Na-DTN (n = 186). Control and test means are approximately the same because, as in B, Na-DTN increased and decreased gj almost equally.

Effects of 100% N₂ on intercellular junctions

When measuring intercellular macroconductances, it became clear that 100% N₂ was more effective than Na₂S₂O₄ in depressing or enhancing intercellular coupling (Fig. 2B). The conductance of intercellular channels was studied in six experiments where, in four cases, this parameter was depressed by 100% N₂. The areas in Fig. 7, A and B, show the results obtained from all six experiments. The mean conductances from each sweep (1,280 points) were plotted against the sweep number in the controls (A) and during hypoxic hypoxia (B). The amplitude distribution histograms of the conductances (%occurrence) are presented in Fig. 7C. The thin line is the g_j distribution under normoxic conditions, and the thick trace presents changes in this parameter during superfusion with 100% N₂. The distribution curve significantly (P < 0.002 by U-test and P < 0.007 by Kolmogorov-Smirnov test) shifted to the left, because the larger conductances decreased drastically, and there were more numerous smaller conductances. Figure 7D is the differential histogram depicting this effect: the increase in smaller conductances and decrease of larger ones.

View larger version (23K): [in this window] [in a new window]   Fig. 7. A: mean channel conductance (gj) obtained from each sweep in 82 control recordings. B: mean channel conductance (gj) obtained from each sweep in 55 recordings under 100% N2. C: amplitude histogram of conductances covering all experiments with 100% N2. Thin line, distribution of conductances during normoxia; thick line, conductance distribution during 100% N2. There is a major shift to the left because, as in D, conductances <75 pS increased, whereas those above this level markedly decreased. This is reflected in the significant decrease of the mean gj.

Single-junction channel (g_j) in normoxia and hypoxia

As indicated in METHODS, single intercellular channel g_j was calculated as g_j variance/mean g_j from digitized recordings (Nicholls et al. 1992) and by manual measurements in the controls and during hypoxia induced by Na₂S₂O₄ or 100% N₂. Different conductance values were obtained from digitized records and from manual measurements, as is the case when membrane channels are measured by fluctuation analysis and from patch recordings (Fenwick et al. 1983) (see also DISCUSSION). Nevertheless, in spite of these differences, the significance of the effects induced by Na₂S₂O₄ and 100% N₂ were similar.

For digitized measurements of the effects of hypoxia produced by Na₂S₂O_4, we used 145 control sweeps (1,280 points/sweep) and 190 sweeps during Na₂S₂O₄ superfusion. The mean g_j control value (17.6 ± 0.94 pS) did not change significantly during superfusion with Na₂S₂O₄ (g_j = 16.6 ± 0.78 pS). There were increases and decreases in conductance. For 100% N₂ studies, we computed 82 control sweeps and 55 sweeps during superfusion with 100% N₂. In this case, the control g_j was similar (17.7 ± 1.68 pS), but N₂-induced hypoxia significantly decreased it to 10.34 ± 0.87 pS. Manual measurements made in physiological solutions before (n = 422) and during (n = 199) superfusion with 1 mM Na₂S₂O₄ gave a control g_j of 38.9 ± 0.96 pS that was not changed significantly by the reducing agent (g_j = 41.6 ± 1.29 pS). However, as in the case of digitized measurements, 100% N₂ significantly decreased single channel g_j, which went from a control value of 32.5 ± 1.29 pS (n = 283) to 20.6 ± 1.39 pS (n = 109). These results are presented in graphic form in Fig. 8.

View larger version (39K): [in this window] [in a new window]   Fig. 8. Effects of hypoxia induced by Na-DTN and 100% N2 on single-channel conductance, calculated as gj variance/mean gj (Digitized) and manually measured (Manual). Only hypoxic hypoxia significantly decreased single-channel conductance (*P < 0.014 in both measurements).

Number of intercellular channels under normoxic and hypoxic conditions

In 17 experiments, the number of channels (n) was calculated as G_j/single channel g_j, following Hille's (1992) suggestion for membrane currents. We estimated that in control solutions the mean was 452.2 ± 38 channels (range 1-2,471). This number did not significantly change (P < 0.12 by U-test) during hypoxia induced by Na₂S₂O₄. The mean was 575.6 ± 55.8 (range 6-3,292). During 100% N₂, there was a significant (P < 0.001, same test) decrease in the number of intercellular channels (83.6 ± 8.9 (range 4-262). These effects are illustrated in Fig. 9, as shown by the open bars.

View larger version (22K): [in this window] [in a new window]   Fig. 9. Number (n) of intercellular channels (calculated as Gj/mean gj in 17 experiments) in controls, during superfusion with 1 mM Na-DTN, and during applications of 100% N2. Application of Na-DTN increased or decreased n, resulting in a nonsignificant change. 100% N2 significantly decreased n. Mean n for each experimental situation is shown within each column.

Mean open time of intercellular channels

Noise analysis (see METHODS) was used to calculate the mean open time of junction channels. In a previous publication (Abudara et al. 2001), we reported that saline acidification to pH 6.3 did not change the open time of channels between glomus cells. Similar results were obtained in this study with Na₂S₂O₄ but not with 100% N₂. During normoxia, the mean open time measured in 17 junctions varied from 3.98 to 30 ms (mean 11.34 ± 1.34 ms). In 12 of these junctions, the preparations were superfused with 1 mM Na₂S₂O₄, and the mean open time did not change because this parameter decreased in five instances, increased in another five, and there was no change in two. In other experiments, three preparations were superfused with saline equilibrated with 100% N₂. In all cases the mean open time decreased.

Figure 10 illustrates these experiments. Figure 10A shows a log/log plot of the g_j variance versus Hz obtained from 12 coupled cell pairs. The exponential curves obtained during normoxia and during Na₂S₂O₄ hypoxia yielded mean open times of 11.9 ± 1.88 ms in the controls and 11.8 ± 1.83 ms during hypoxia. Figure 10B describes the mean open time, in milliseconds, obtained in each of the experiments in the controls and during superfusion with Na₂S₂O₄, showing the variability of the results. Figure 10C is another log/log plot of the g_j variance versus Hz obtained from three coupled cell pairs during normoxia and during superfusion with 100% N₂. The mean open time during normoxia was 10.9 ± 0.54 ms; that was reduced to 9.5 ± 0.55 during hypoxic hypoxia. Figure 10D shows differences in the variances presented in Fig. 10C. In practically all cases, these values were smaller during hypoxia, resulting in a significant difference between control and hypoxic variance values (P < 0.007 by Wilcoxon signed-rank test).

View larger version (39K): [in this window] [in a new window]   Fig. 10. Mean open time of single-junction channels. A: power density spectra of mean gj variances (12 coupled pairs) plotted against frequency (Hz). , Controls; , 1 mM Na-DTN. Curves are the exponential fit of the data: thin line, controls; thick line, hypoxia by Na-DTN. In each case, corner frequency (Fc) was calculated as 1/2 the Y-value of each curve. Mean open time (OT) in ms was calculated from  = 1/[2Fc]. B: mean open time (in ms) of channels in each experiment during control experiments (open bars) and Na-DTN hypoxia (solid bars). There are increases and decreases in (OT). C: same as in A, but with hypoxia induced by 100% N2. D: plot showing that spectral changes in channel conductance (gj variance [ps2 · s]) induced by hypoxic hypoxia occurred throughout the spectrum but especially at frequencies <10 Hz.

Effects of hypoxia on [Ca²⁺]_i

([Ca²⁺]_i) was measured in 48 cells from 18 experiments, resulting in a mean [Ca²⁺]_i of 46.2 ± 4.84 nM. During superfusion with 1 mM Na₂S₂O₄ and 100% N₂, the values were 77.32 ± 11.27 (P < 0.001 by Wilcoxon test) and 66.4 ± 11.64 nM (P < 0.005, same test), respectively. With both stimuli, ~60% of the cells showed an increase in [Ca²⁺]_i, whereas the rest showed either little change or a decrease in [Ca²⁺]_i. In most, but not all, cases nifedipine (10 µM) blocked or depressed the [Ca²⁺]_i changes induced by Na₂S₂O₄ or 100% N₂, suggesting that at least some of their effects were mediated through voltage-gated L-type channels (R. G. Jiang and C. Eyzaguirre, unpublished observations). Thus we found similar effects of Na₂S₂O₄ and 100% N₂ on [Ca²⁺]_i in spite of the fact that Na₂S₂O₄ decreased PO₂ to much lower values. Dasso et al. (2000) found graded responses of [Ca²⁺]_i to graded changes in PO₂. However, we did not conduct dose-response experiments with the same hypoxic stimulus. Therefore, future experiments with graded stimuli, using Na₂S₂O₄ or 100% N₂, may reveal similar or different properties (see DISCUSSION).

Figure 11, A and B, illustrates examples of two cells stimulated with Na₂S₂O₄ and 100% N₂. Figure 11C shows the mean [Ca²⁺]_i in each of the controls, giving an overall mean (µ) of 46.2 ± 4.84 nM. Figure 11D presents a percentile distribution of ratios-test (hypoxia) values over control measurements. The effects of hypoxia induced by Na₂S₂O₄ and 100% N₂ were not statistically different. However, the curves show larger effects of Na₂S₂O₄ above the 60th percentile level, suggesting that this agent may be more effective than 100% N₂ in increasing [Ca²⁺]_i.

View larger version (34K): [in this window] [in a new window]   Fig. 11. A and B: samples of optical recordings of intracellular calcium ([Ca2+]i) in normoxia and during brief applications of Na-DTN (solid arrows) and 100% N2 (open arrows). C: mean [Ca2+]i (2+]i) of each experiment with an overall mean (µ) of 46.2 ± 4.84 nM. D: percentile distribution of ratio of changes in [Ca2+]i (T/C [Ca2+]i) during application of 1 mM Na-DTN and 100% N2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It is important to discuss or justify the choice of techniques in this study. Intracellular microelectrodes were used instead of the more commonly employed patch-type pipettes. In intercellular junctions, the intracellular medium is important in regulation, and it could be argued that microelectrodes disturb this medium much less than do patch pipettes. Although we know the ionic composition of the cytosol in glomus cells (He et al. 1991a; Oyama et al. 1986a,b; Pang and Eyzaguirre 1993b; Zhang and Eyzaguirre 1999; Zhang et al. 1995), we only have limited qualitative knowledge concerning the second messengers, proteins, or peptides that are bound to influence the intercellular junctions (see Pérez-García and González 1997; Zapata 1997). Also, we used fluctuation (noise) analysis to establish the properties of single-junction channels, realizing it is an indirect method. However, we chose to use this analysis because it does not rely on artificial procedures to reduce the number of channels (e.g., long-chain alcohols or other junction blockers). However, because there is a discrepancy in cell membrane studies when channels are indirectly measured by fluctuation analyses and, directly, from membrane patches (Fenwick et al. 1982), we also measured channel currents and conductances by hand. As in a previous study (Abudara et al. 2001), manual measurement of single intercellular channels gave larger values than those detected by fluctuation analysis. In our case, manual measurements were likely biased toward larger values because it was difficult at times to be certain whether small deflections had partners of equal amplitude and opposite polarities. When there was uncertainty, those deflections were not considered in the analyses. Nevertheless, computerized and manual measurements gave similar results, showing that 100% N₂ was more effective than Na₂S₂O₄ in depressing intercellular conductances.

Hypoxic hypoxia (induced by 100% N₂) was considerably more effective than hypoxia elicited by Na₂S₂O₄ in affecting the junction channels between glomus cells. This happened in spite (or because) of the fact that 100% N₂ decreased saline PO₂ much less than Na₂S₂O₄. The g_j changed more drastically during N₂-induced hypoxia than during Na₂S₂O₄ hypoxia (see also Abudara and Eyzaguirre 1998). Also, 100% N₂ significantly depressed intercellular microconductances and reduced the number of active intercellular channels and their mean open time, whereas Na₂S₂O₄ had variable effects. We do not know the reasons for the different effects of these two hypoxic agents. However, it is important to recall that, in the lung, Na₂S₂O₄ produces superoxide anions and hydrogen peroxide (Archer et al. 1995), which may also happen in the carotid body. Applications of hydrogen peroxide to pairs of glomus cells increases intercellular coupling in >80% of the pairs (L. Monti-Bloch, V. Abudara, and C. Eyzaguirre, unpublished observations). Furthermore, Na₂S₂O₄ evokes Ca²⁺ influx into glomus cells (see above), regardless of PO₂ levels (Carpenter et al. 2000), which could contribute to cell uncoupling (Abudara and Eyzaguirre 1996a). Consequently, the uncoupling effects induced by Na₂S₂O₄ hypoxia may have been blunted by release of hydrogen peroxide and enhanced by Ca²⁺ influx. The effects of 100% N₂ were clearer, possibly because this agent had a more straightforward uncoupling action (e.g., Ca²⁺ influx), with little or no release of coupling agents. However, concerning possible mechanisms of hypoxic coupling changes, the following information is pertinent.

It is generally agreed that intracellular acidity and increases in [Ca²⁺]_i contribute to cell uncoupling in many tissues (Francis et al. 1999; Lazrak and Peracchia 1993; Obaid et al. 1983; White et al. 1990). Na₂S₂O₄ and 100% N₂ change the pH_i of cultured and clustered glomus cells, inducing intracellular acidification in ~60% of the cases and alkalinization in the others (He et al. 1991b; Pang and Eyzaguirre 1993a). Therefore, it would be tempting to assume that hypoxia acts on intercellular coupling via pH_i changes that uncouple most glomus cells during acidification (Abudara and Eyzaguirre 1998; Monti-Bloch et al. 1993). However, both hypoxia-inducing agents also increase [Ca²⁺]_i (which uncouples glomus cells), an action blocked by cobalt (Abudara and Eyzaguirre 1996a, 1998). Therefore, both high [Ca²⁺]_i and low pH_i may act in synchrony. It is still puzzling why 100% N₂ is more effective than Na₂S₂O₄ in uncoupling glomus cells when it induces weaker pH_i changes and similar increases in [Ca²⁺]_i. Therefore, an explanation of mechanisms based only on changes in [Ca²⁺]_i and/or pH_i is not satisfactory because hypoxic uncoupling (or increased coupling) may also be influenced by other factors.

The carotid body glomus cells contain a number of chemicals, the concentration of which changes during hypoxia, and some of them are released from the cells. For instance, glomus cells contain ACh, catecholamines [especially dopamine concentration (DA), serotonin(5-HT), enkephalins, prostaglandins, ATP, substance P, and peptides] cholecystokinins and atrial natriuretic peptide (ANP)], as shown by many authors (see González et al. 1997; Zapata 1997). In addition, hypoxia increases the intracellular levels of cAMP while decreasing those of cGMP (Delpiano and Acker 1991; Pérez-García et al. 1990; Wang et al. 1989). Most of the substances influence intercellular coupling in other tissues, and they may also participate in the carotid body. For instance, ACh and DA are released from glomus cells during hypoxia, and when exogenously applied, uncouple most glomus cells (Monti-Bloch et al. 1993). This means that the released substances can affect intercellular coupling. In other tissues, exogenous applications of ACh and DA also uncouple cells (He et al. 2000; Piccolino et al. 1984; Randriamampita et al. 1988).

Administration of dB-cAMP has long-term effects on coupling because it increases the number of gap junctions (Chanson et al. 1996; Romanello et al. 2001; van Rijen et al. 2000), a phenomenon that also occurs in junctions between glomus cells (Abudara et al. 1999, 2000). Also, acute administration of cAMP tightens coupling between glomus cells (Abudara and Eyzaguirre 1998), and this substance increases during hypoxia. Therefore it is possible that during hypoxia, an increase in cellular cAMP would tend to improve coupling between glomus cells. Likewise, a decrease in cGMP, a decoupler in other tissues (Kwak et al. 1995), would also tend to increase intercellular coupling. As a consequence, hypoxia would release two opposing forces on glomus cell coupling, one trying to decouple the cells (ACh and DA) and another having the opposite effect (increased cAMP and decreased cGMP). This may explain the variable effects of hypoxia on glomus cell coupling, depending on which one predominates at a given time.

Concerning the other agents within the glomus cell cytoplasm and their possible role in coupling, the following is pertinent: 1) ATP is contained in the dense-cored granules of glomus cells (also containing DA) and is released by exocytosis during hypoxia. Once released, part of it is converted to adenosine by ectonucleotidases. These purinergic agonists may activate the membranes of adjoining cells because of the presence of purinergic receptors for ATP (P₁) and adenosine (P₂) (see Zapata 1997). In other tissues, ATP release produces or increases Ca²⁺ waves and increases junction permeability, acting extracellularly (Cotrina et al. 2000; Guthrie et al. 1999; Homolya et al. 2000; Isakson et al. 2001; Sauer et al. 2000). A similar mechanism may be present in glomus cell junctions, leading to increased coupling during hypoxia. 2) Cyclooxygenases and prostaglandin E₂ (PGE₂) may also play a role in coupling between glomus cells and their changes during hypoxia, although this has not been studied. This stimulus increases the synthesis of endogenous PGE₂, and its exogenous application inhibits catecholamine release from glomus cells during hypoxia and inhibits inward Ca²⁺ currents. In osteocytelike MLO-Y4 cells, PGE₂ seems to be essential for intercellular communication across gap junctions (when mechanically activated) by increasing connexin 43 (Cheng et al. 2001). 3) Exogenous applications of the secretagogue cholecystokinin octapeptide (CCK-8) increases chemosensory discharges in the carotid body after a period of depression. In pancreatic acini, CCK-8 induces electrical uncoupling (Ngezahayao and Kolb 1993). However, we still are not certain if changes in chemosensory discharges are related to intercellular coupling in the carotid body. 4) Applications of ANP to the cat and rabbit carotid bodies depress or inhibit the increased sensory discharge elicited by hypoxia. This effect appeared to be elicited by an increase in cGMP because applications of the cell-permeant form of this compound had a similar effect. Interestingly, and related to intercellular coupling, de Mello (1998) found that delivery of ANP (10⁸ M) to myocytes of cardiomyopathic hamsters decreased g_j by ~48% and applications of db-cGMP (10⁴ M) reduced g_j by ~80%. Similar experiments dealing with coupling between glomus cells have not been done. 5) 5-HT has no effects on the carotid body chemosensory discharge in vitro, suggesting that the effects observed in vivo are of vascular origin. In the somatosensory cortex, 5-HT reduces dye coupling (Rorig and Sutor 1996), but in vascular smooth muscle, g_j increases (Moore and Burt 1995). There are no studies on coupling and 5-HT in glomus cells.

The information presented here clearly shows that to understand the mechanisms of hypoxic uncoupling (or increased coupling) between glomus cells, much more work has to be done. It is almost certain that hypoxia starts a cascade of events in glomus cells that would provoke uncoupling in some cases and increased coupling in others. Most of these steps are unknown. However, this complex series of events could be in place to ensure sustained activity of the receptor by releasing transmitters from some glomus cells and recharging others to replace the transmitter load (Eyzaguirre and Abudara 1999). This mechanism is needed to provide proper ventilation during prolonged hypoxia, which happens at high altitudes, because the carotid body is the main or only O₂ sensor in the body.