Introduction

Gap junctions are clusters of cell-cell channels that enable neighboring cells to exchange small signalling molecules (≤1 kDa) such as Ca2+, cAMP, IP3, and to propagate electric activities (Bruzzone, White & Paul, 1996; White & Paul, 1999; Harris, 2001). The gap junctional channels are composed of connexins, which are homologous proteins. So far, 19 and 20 connexin (Cx) genes have been found in the mouse and the human genome, respectively, that are specifically expressed and regulated in different cell types (Willecke et al., 2002). The connexins hexamerize and form connexons (hemichannels) that are inserted in the membrane of a single cell. The cell-cell channel is formed when hemichannels of two adjacent cells associate by their extracellular domains. Since in all tissues analyzed so far, at least two types of connexins are expressed (Bruzzone et al., 1996; White & Paul, 1999), connexons and gap junctional channels with complex stoichiometries can be expected. A connexon can be homomeric or heteromeric, depending on the presence of one or several connexin types, respectively. The gap junction channel formed by equally homomeric connexons is called homotypic. Two homomeric connexons composed of different connexin types build a heterotypic gap junctional channel and a cell-cell channel composed of two heteromeric connexons will be called heteromeric (Goodenough, Simon & Paul, 1999; White & Paul, 1999). Oocytes of Xenopus laevis have been very useful to characterize the regulation (e.g., gating by voltage) and the transport capacity of homotypic gap junctional channels. But the results obtained from heterologous expression systems did not allow a reliable prediction of the properties of gap junctional channels composed of different connexin isoforms as expressed in original tissues (White & Bruzzone, 1996; Elenes et al., 2001).

The most abundant cells of the follicular system, the granulosa cells, express different connexins during their hormone-regulated proliferation and differentiation (Wright et al., 2001). The gap junctional channels in granulosa cells of rat are mainly composed of the phosphoprotein Cx43, though Cx45 is also expressed. During the follicular development, the expression pattern and posttranslational modifications, like phosphorylation of connexins, change (Schreiber et al., 1993; Okuma et al., 1995; Sommersberg et al., 2000; Wright et al., 2001). As demonstrated by Ackert et al. (2001), the gap junctional communication between the granulosa cells plays an important role in folliculogenesis. They showed that follicular growth could not be induced in Cx43- deficient mice. Furthermore, the induction of atresia, which corresponds to the elimination of granulosa cells through apoptotic/necrotic processes (Tilly, 1996), is concomitant with a reduced expression of Cx43 in rat granulosa cells (Wiesen & Midgley, 1994). It was therefore hypothesized that a loss of gap junctional communication contributes to the apoptotic processes during atresia. Despite the crucial role of gap junctional communication for follicular physiology and atresia, the regulation of gap junctional conductance between granulosa cells has not yet been studied. In order to understand the effects of gonadotropins and estrogens on the gap junctional communication of granulosa cells, hormonal effects on the conducting activity of the gap junctional channels have to be analyzed.

For the first time we quantitatively analyzed the voltage gating and intracellular regulation of gap junctional coupling of granulosa cells, using the cell line GFSHR-17. GFSHR-17 cells express the receptor for the follicular stimulating hormone (FSH). This cell line is considered to be suitable for analysis of the FSH-dependent maturation and as an in vitro cell model of granulosa cells (Keren et al., 1993; Sommersberg et al., 2000). We applied the double whole-cell patch-clamp technique in combination with a simultaneous measurement of [Ca2+]i. Establishing the double whole-cell configuration caused a spontaneously occurring gap junctional uncoupling in parallel with a sustained rise of [Ca2+]i and cell shrinkage. These cellular reactions could be suppressed by blockage of K+ efflux or by use of a nominally Ca2+-free bath solution. Both modifications could be substituted by addition of 8-Br-cGMP to the pipette solution. The data indicate that cGMP depletion is related to Ca2+ entry and to K+ release, which causes cell shrinkage and induces gap junctional uncoupling in granulosa cells.

Materials and Methods

Cell Culture

Granulosa cell line GFSHR-17 (Keren et al., 1993) with passage number 5 was cultivated in petri dishes at 37°C under an atmosphere of 5% CO2–95% air. Cells were fed with Dulbecco's F12 modified Eagle's medium (DMEM, pH 7.4, Sigma) supplemented with 10% fetal calf serum and penicillin and streptomycin. The culture medium was replaced every 3 days. After 7 days a cell monolayer was formed. Then the cells were trypsinized, collected and centrifuged for 5 min at 100 × g. The pellet was resuspended in 10 ml DMEM. For the measurements, 1–2 µl of the cell suspension were placed on a cover slip in a petri dish. After 5 min, the cells had adhered and 2–3 ml of culture medium were added to the dish. The cells were cultivated as described above and used for double whole-cell patch-clamp experiments 2–5 days later. Cells of passage number 6–10 were cryoconserved and used as stock. Cells were utilized up to a total of 25 passages.

Pipette Solutions

For composition and nomenclature of the different pipette solutions see Table 1.

Table 1 Composition of control bath and different pipette solutions
Full size table

Measurement of Gap Junctional Coupling and Cytoplasmic Free Calcium [CA2+]I

Double Whole-cell Patch-clamp Analysis

A cover slip with cells was transferred to a superfusion chamber containing 0.5 ml bath solution. The chamber was mounted on a Zeiss inverted microscope (Oberkochen, Germany). The cells were washed with 10 ml (2 ml/min) of the bath solution and allowed to adapt to room temperature (~20°C) for at least 30 min. The double whole-cell patch-clamp configuration was established for an isolated cell pair according to Neyton & Trautmann (1985) (cf. Ngezahayo & Kolb, 1994; Fig. 1), using two patch-clamp amplifiers EPC 7 (List Medical, Darmstadt, Germany). The electrophysiological data were filtered at 3 kHz and digitized at 10 kHz via an interface ITC 16 (Instrutech, Minnesota, USA). The data acquisition and off-line analysis were done using Pulse Pulsefit (HEKA Electronics, Lamprecht, Germany).

Figure 1
figure 1

Model circuit of the double whole-cell configuration of a cell pair coupled by gap junctional channels. U p1 and U p2 are the potentials of electrodes 1 and 2. Serial access resistances are given by R a1 and R a2. Nonjunctional membrane resistances are denoted by R m1 and R m2, respectively. R j gives the gap junctional resistance. C m1 and C m2 represent the membrane capacities of the corresponding cells.

Full size image

For measurement of the time course of gap junctional conductance (G j), both cells were clamped to a holding potential of −40 mV near their resting potential U m (generally between −30 to −38 mV), resulting in a net zero junctional current and a zero non-junctional current. Thereafter, a voltage jump ΔU p of ±10 mV was applied every 30 s for 100 ms to one cell, respectively. A model circuit of resistances for the double whole-cell configuration is given in Fig. 1. The serial resistances R a1, R a2 varied between 5 and 12 MΩ for all pipette solutions (see Table 1). The non-gap junctional membrane resistance R m2 was in the range of 0.25 to 2.0 GΩ. To avoid significant errors in the determination of G j, only experiments with |R a1−R a2| < 1 MΩ, |R m1−R m2| < 25 MΩ and |U m1−U m2| < 1.5 mV were considered (see Veenstra 2001). If the serial resistances R a1 and R a2, the membrane resistances R m1 and R m2 and the membrane potentials U m1 and U m2 are not significantly different, G j can be calculated using the equation:

Index 1 denotes the cell at which a voltage jump from the common holding potential (−40 mV) was applied and index 2, the other member of the cell pair, which was clamped to the holding potential. To analyze the voltage dependence of G j, voltage pulses ΔU p1 between −120 and 120 mV were added to the holding potential of −40 mV, in steps of 10 mV with a duration of 1 s, and the evoked current responses I p1 and I p2 were recorded.

In some cell pairs (see Results), a voltage- and time-dependent current decrease was observed. To describe the voltage sensitivity of the corresponding gap junctional conductance, the steady-state conductance G ss was calculated from the current values recorded during the last 50 ms of the applied pulse of 1 s. For clearer presentation, G ss was normalized to the maximal value of G j (G max), which was derived as the mean initial conductance value (G inst) recorded within 20 ms after application of a voltage pulse. G ss/G max values were plotted as function of ΔU p1 and fitted by a relation that describes the voltage dependence at both polarities (Chen-Izu, Moreno & Spangler, 2001)

to provide an estimation of ΔU 1 and ΔU 2, the voltages of half-activation for each of the connexons. A 1 and A 2 are the cooperativity factor describing the voltage dependence of the open-closed equilibrium for each of the connexons. G min denotes the minimal value of G j as function of ΔU p1. Experiments were performed at room temperature. Data are given as mean ± SEM; n denotes the number of independent experiments (cell pairs).

Measurement of [Ca2+]i

Measurement of intracellular Ca2+ ([Ca2+]i) was performed according to Grynkiewicz, Poenie & Tsien (1985). To allow a measurement of [Ca2+]i prior and after whole-cell patch formation, the cells were loaded with fura 2-AM (5 µM and 1% DMSO) (Molecular Probes) for 45–60 min at room temperature. The fura 2-AM-loaded cells were then transferred to a superfusion chamber, mounted onto an inverted microscope (see above). The cells were superfused with a bath solution (2 ml/min) for at least 5 min to wash out extracellular fura 2-AM and DMSO. The loaded cells were excited at 340 nm and 380 nm using a monochromator polychrome II (T.I.L.L. Photonics, Planegg, Germany) equipped with a 75 W XBO xenon lamp. To minimize the dilution of intracellular fura-2 concentration by the whole-cell formation, 50 µM K5-fura-2 (Molecular Probes) were added to the pipette solution. This concentration of K5-fura-2 was found to be suitable to obtain an about constant fluorescence signal at 340 nm and 380 nm prior and after whole-cell configuration. The fluorescence was registered with a digital CCD camera (C4742-95, Hamamatsu Photonics K.K.; Japan). The 340 nm to 380 nm excitation ratios of the fluorescence images were then calculated and calibrated to determine [Ca2+]i, using the program Aquacosmos (Hamamatsu Photonics K.K.; Japan). The program was also used to estimate relative changes of cell diameters.

Results

Voltage Dependence of Gap Junctional Conductance

The double whole-cell patch-clamp configuration was applied to an isolated cell pair and both cells were clamped to a holding potential of −40 mV close to a membrane potential of −30 mV to −38 mV. To analyze the voltage dependence of G j, one cell was clamped to test voltages (U p1) of −160 mV to +80 mV in steps of 10 mV, applied for 1 s, while the holding potential of the adjacent cell was kept constant (see Fig. 1 A, 1B). This allowed imposition of transjunctional voltages (ΔU p1) of −120 mV to 120 mV.

Independently of the pipette solution, G j varied as function of ΔU p1 for different cell pairs. 37 cell pairs out of a total of 60 cell pairs showed a voltage independence of voltage-jump current-relaxations as well as of G j. Figure 2 C and 2D show representative series of current relaxations of cell 1 (I p1) and cell 2 (I p2) in response to test voltages applied to cell 1. The current records indicate an almost voltage-independent behavior, which is reflected in the gap junctional conductance. Figure 2 E presents the corresponding instantaneous gap junctional conductance (G inst) and the quasi steady-state values of G j (G ss) as function of ΔU p1, respectively; the two functions do not differ significantly in their amplitude.

Figure 2
figure 2

Voltage protocol applied to cell 1 (U p1) and cell 2 (U p2), evoked currents (I p1, I p2) and gapjunctional conductance (G j). (A) Voltage steps applied to cell 1. (B) Cell 2 was clamped at the constant holding potential of −40 mV. (C) and (D) Current response of cell 1 and cell 2, respectively. (E) Gap junctional conductance (G j) as function of ΔU p1. The instantaneous gap junctional conductance (G inst) was derived as mean value within the first 20 ms after application of a voltage step and the quasi-steady-state gap junctional conductance (G ss), 50 ms before the voltage step of 1 s ended. For clearer presentation only the voltage steps to −160, −120, −80, −40, 0, 40, 80 mV (A) and the corresponding current responses (C and D) are presented. Control bath and pipette solution A (Table 1) were used.

Full size image

A significant voltage-dependent inactivation of junctional currents was obtained for 9 cell pairs, which was observed in the range of −50 mV > ΔU p1 > 50 mV. Figure 3 A, B shows representative current records in response to the voltage protocol given in Fig. 2 A, B. The current relaxations of both cells decrease in a voltage- and time-dependent manner. The time course of the currents could be described by a sum of two exponential functions. The corresponding time constants τ1 and τ2 are given in Table 2. The corresponding gap junctional conductances G inst and G ss are presented in Fig. 3 C. No significant voltage dependence of G inst is observed, but G ss exhibits a bell-shaped course as function of ΔUp1, with a maximum in the range of −50 mV > ΔU p1 > 50 mV. For a more detailed analysis of the voltage dependence of G j, the corresponding normalized conductance G ss/G inst was described by Eq. 2 (Materials and Methods). From a fit of Eq. 2 to the data (Fig. 3 D), half-maximal inactivation voltages, according to the polarity of the applied voltage pulse, of ΔU 1 = −61.3 ± 0.8 mV and ΔU 2 = 66.4 ± 1.0 mV and for the corresponding voltage sensitivities A 1 = 0.13 ± 0.01 mV−1 and A 2 = 0.08 ± 0.01 mV−1 (n = 9) were derived.

Table 2 Time constants of current inactivation τ1, τ2
Full size table
Figure 3
figure 3

Evoked currents and corresponding gap junctional conductance. (A) and (B) Current responses of electrode/pipette 1 and 2 after application of the voltage protocol given in Fig. 2 A, B, respectively. (C) G inst and G ss as function of ΔU p1.(D) G ss as function of ΔU p1, which was normalized to the maximal value of G j (G max). G max was derived as the mean initial conductance value recorded within 20 ms after application of a voltage pulse ΔU j. The mean values of G ss were obtained from 9 different cell pairs. G ss/G max vs ΔU j was fitted by Eq. 2. The result is given by the lines drawn. For further explanation, see Fig. 2 and text.

Full size image

In three cell pairs, the application of the voltage protocol given in Fig. 2 A, 2B caused a time-dependent activation of the junctional current, which rapidly reached a stationary value within about 50 ms. Figure 4 shows the result of a representative experiment, which indicates a voltage-independent activation of current and conductance of the gap junctional channels.

Figure 4
figure 4

Evoked currents and corresponding gap junctional conductance. (A and B) Current responses of electrode/pipette 1 and 2 after application of the voltage protocol given in Fig. 2 A, B. (C) G inst and G ss as function of ΔU p1. For further explanation, see Fig. 2 and text.

Full size image

In 11 cell pairs, a variable superposition of a voltage-dependent inactivating and voltage-independent activating behavior was observed (data not shown). A correlation of this finding with the specific passage number could not be found.

Time Course of [CA2+]I, Cell Volume and Gj After Whole-cell Formation

Control Bath and Pipette Solution A

Using control bath and pipette solution A (see Table 1, additionally containing 50 µM K5-fura-2), the time course of [Ca2+]i, cell volume and G j could be followed on cell pairs after formation of the two consecutively performed whole-cell configurations that are necessary for establishing the double whole-cell formation (see Materials and Methods). A typical experiment is shown in Fig. 5. The formation of the first whole-cell configuration on cell 1 caused a rapid rise of [Ca2+]i within less than a minute from a resting level of 65 ± 5 nM (n = 20) to values above 350 nM. It is interesting to note and will be considered in the Discussion that despite the known permeability of gap junctional channels for Ca2+ and a large initial value of G j0 of about 40 nS, the adjacent cell 2 responded in the time scale of several minutes by a far less pronounced increase of [Ca2+]i. (see Fig. 5 A and Inset). Consecutive formation of the whole-cell configuration on cell 2 caused a rapid and steep rise of [Ca2+]i in this cell, which is partially reflected in the neighboring cell. Thereafter the time course of G j could be recorded, which is given in Fig. 5 B. Comparison of Fig. 5 A and 5B indicates that the rapid rise of [Ca2+]i causes no significant change of G j, but during the observed sustained increase of [Ca2+]i the gap junctional conductance declines to about 1% of the initial value (G j0) within 15–20 min. This finding was obtained also in the absence of recording [Ca2+]i. In parallel, cell shrinkage was observed (see Fig. 5 C), which occurs at a similar time scale. For easier quantification of the cell size, single cells of spherical shape were selected and the cellular diameter was determined after whole-cell formation (Fig. 5 D). The cell shrinkage started 2–3 min after whole-cell formation and was followed for about 10 min. The cell diameter of the cells decreased by about 75% within 10 min. Thereafter further morphological changes of the cells, like bleb formation, were observed. Occasionally (data not shown), an increase of the cell volume was observed within 1–2 min after establishing the whole-cell configuration, which was followed by the described cell shrinkage.

Figure 5
figure 5

Effect of double whole-cell formation on the time course of [Ca]i, gap junctional conductance, and shape of the cell pair in control bath and with pipette solution A (Table 1). (A) Time course of [Ca2+]i after establishing the two consecutive whole-cell configurations, which were started from cell-attached configuration of the two pipettes. After whole-cell formation at the first cell, marked by ▲, a rapid rise of [Ca2+]i occurs, while [Ca2+]i in the second cell (see lower trace in the inset between the two marks), remains almost unaffected. At the 2nd mark, the whole-cell formation of the second cell is established. The data exhibit a representative experiment of a total of n = 5. (B) Time course of G j. For clearer presentation, G j is normalized to G jo obtained 1 min after double whole-cell formation. The data represent the mean ± SEM of n = 10 cell pairs. For the total of 60 experiments, an initial value G jo = 39.25 ± 2.37 nS (n = 60) was obtained at |ΔU p1| = 10 mV. (C) Shapeof a cell pair 0 min and 10 min after double whole-cell formation. (D) Shape of a single cell 0 min and 10 min after whole-cell formation. For single cells with an almost spherical form, within 10 min, a reduction of the cell diameter to 77 ± 1.2% (n = 6) of the initial value could be estimated.

Full size image

Ca2+-free Bath and Pipette Solution A

To analyze an involvement of Ca2+ influx in the observed cellular processes initiated by whole-cell formation with pipette solution A (see Fig. 5), the experiments were repeated in nominally Ca2+-free bath containing 10 mM EGTA. Fig. 6 presents a typical experiment. Fig. 6 A shows a significant, but transient rise of [Ca2+]i for 2–3 min in both cells, which is induced by the whole-cell formation. Gap junctional conductance G j (Fig. 6 B) as well as cell volume (Fig. 6 C) appear to be unchanged during the time span of an experiment of up to 30 min. These data indicate an involvement of Ca2+ in the observed changes in cell morphology and G j (compare Fig. 5 B,C and Fig. 6 B,C).

Figure 6
figure 6

Effect of double whole-cell formation on the time course of [Ca2+]i, gap junctional conductance, and shape of the cell pair in nominally Ca2+-free control bath containing additionally 10 mM EGTA, and with pipette solution A (Table 1). (A) Time course of [Ca2+]i afterestablishing the two consecutive whole-cell configurations, which were started from cell-attached configuration of the two pipettes. After whole-cell formation at the first cell, marked by ▲, a rapid rise of [Ca2+]i occurs, while [Ca2+]i in the second cell (see lower trace between the two marks), remains almost unaffected. At the 2nd mark, the whole-cell formation of the second cell is established. The data show a representative experiment of a total of n = 5. (B) Time course of G j. For clearer presentation, G j is normalized to G jo obtained 1 min after double whole-cell formation. The data represent the mean ± SEM of n = 5 cell pairs. (C) Shape of a cell pair 0 min and 10 min after double whole-cell formation.

Full size image

Control Bath and Pipette Solution Containing Cs+

Since it is known that an increase of [Ca2+]i amplifies a Ca2+-dependent K+ efflux, the influence of the latter was investigated. A K+ efflux was blocked by addition of 40 mM Cs+ to the pipette solution (pipette solution B, Table 1). The time course of the three measured cellular parameters is presented in Fig. 7. Again a transient rise of [Ca2+]i was observed for the two adjacent cells (Fig. 7 A). It is interesting that the transient change of [Ca2+]i in one cell caused no visible change of [Ca2+]i in the adjacent coupled cell. G j (Fig. 7 B) as well as the cell volume (Fig. 7 C) stayed constant throughout the measuring time of an hour. Comparison of Figs. 6 and 7 indicates that the behavior of [Ca2+]i, cell morphology and G j after whole-cell formation under blockade of K+ efflux by Cs+ is comparable with that observed in Ca2+-free bath solution.

Figure 7
figure 7

Effect of double whole-cell formation on the time course of [Ca2+]i, gap junctional conductance, and shape of the cell pair in control bath and with pipette solution Bcontaining Cs+ (Table 1). (A) Time course of [Ca2+]i after establishing the two consecutive whole-cell configurations, which werestarted from cell-attached configuration of the two pipettes. After whole-cell formation at the first cell, marked by ▲, a rapid rise of [Ca2+]i occurs, while [Ca2+]i in the second cell (see upper trace between the two marks), remains almost unaffected. At the 2nd mark, the whole-cell formation of the second cell is established. The data give a representative experiment of a total of n = 5. (B) Time course of G j. For clearer presentation, G j is normalized to G jo obtained 1 min after double whole-cell formation. The data represent the mean ± SEM of n = 5 cell pairs. (C) Shape of a cell pair 0 min and 10 min after double whole-cell formation.

Full size image

Control Bath and Pipette Solution A Containing 8-Br-cGMP

The data presented in Figs. 5, 6 and 7 indicate that whole-cell formation generates a Ca2+-dependent K+ efflux that is related to gap junctional uncoupling. Since it can be assumed that by whole-cell formation regulatory molecules are washed out by dialysis against the pipette solution, a corresponding substitute was evaluated. 3′,5′-Cyclic guanosine monophosphate (cGMP) was considered a suitable molecule, since it is known that cGMP controls various cellular functions like regulation of [Ca2+]i and K+ efflux. The activation of both ions' fluxes has been related to the early phase of apoptosis (Hirsch et al., 1999; Moreno et al., 2001). 50 µM 8-Br-cGMP, the non-hydrolyzable form of cGMP, was added to the pipette solution (pipette solution C, Table 1). Figure 8 shows that by use of pipette solution C the time course of [Ca2+]i, cell volume as well as of G j was unaffected by whole-cell formation. Also the early transient change of [Ca2+]i, as observed in Figs. 6 A and 7A, was suppressed by the presence of intracellular 8-Br-cMP (Fig. 8 A). Replacement of cGMP by cAMP up to 0.1 mM did not suppress any of the described cellular responses (data not shown).

Figure 8
figure 8

Effect of double whole-cell formation on the time course of [Ca2+]i, gap junctional conductance, and shape of the cell pair at control bath and pipette solution C containing 5 µM 8-Br-cGMP (Table 1). (A) Time course of [Ca2+]i after establishing the two consecutive whole-cell configurations, which were started from cell-attached configuration of the two pipettes. After whole-cell formation at the first cell, marked by ▲, no significant rise of [Ca2+]i occurs. At the 2nd mark, the whole-cell formation of the second cell is established. Again the whole-cell formation does not induce an elevation of [Ca2+]i. The data are a representative experiment of a total of n = 5. (B) Time course of G j. For clearer presentation, G j is normalized to G jo obtained 1 min after double whole-cell formation. The data represent the mean ± SEM of n = 5 cell pairs. (C) Shape of a cell pair 0 min and 10 min after double whole-cell formation.

Full size image

Discussion

Voltage Dependence of Gap Junctional Coupling

In the GFSHR-17 granulosa cell line variable transjunctional-voltage gating properties were observed (compare Figs. 2, 3 and 4). One population showed a bell-shaped behavior of the gap junctional conductance G j at voltages |ΔU p1| ≥ 60 mV (Fig. 3 C). This property is known for specific connexins forming gap junctional channels. In rat granulosa cells, coexpression of at least two connexins, Cx43 and Cx45, and their concomitant presence within some gap junctional plaques have been demonstrated (Okuma et al., 1996). The simplest configuration is the formation of homotypic channels, which result from the association of homomeric hemichannels or connexons. This configuration can be achieved after expression of connexins in Xenopus oocytes or by transfection of connexins in various cell systems. For Cx43 and Cx45 the corresponding homotypic gap junctional channels exhibit a significant voltage dependence (Banach & Weingart, 1996; Steiner & Ebihara, 1996; Stergiopoulos et al., 1999; Elenes et al., 2001), which in the following will be compared with our findings.

The transjunctional voltage-gating properties derived from fitting Eq. 2 to the data of Fig. 3 C can be compared with the corresponding data obtained for Cx43/Cx43 and Cx45/Cx45 homotypic and Cx43/Cx45 heterotypic gap junctional channels formed in paired Xenopus oocytes (Elenes et al., 2001). For homotypic gap junctional channels composed of Cx43 a striking similarity is found for the transjunctional-voltage gating concerning the half-maximal inactivation voltages and the corresponding voltage sensitivities for the transitions between the conductive states. In addition, a similar agreement is found for the corresponding time constants of current inactivation after a transjunctional voltage step (compare also Table 2 and Elenes et al., 2001). Such an agreement could be identified for the transjunctional-voltage gating properties of neither Cx45/ Cx45 homotypic nor of Cx45/Cx43 heterotypic channels (Elenes et al., 2001). Therefore, it appears likely that the transjunctional-gating properties presented in Fig. 3 C, which are representative for 15% of a total of 60 experiments, are related to formation of homotypic Cx43/Cx43 gap junctional channels between adjacent granulosa cells. In all the other cases, the steady-state behavior of G j appears to be voltage-independent. But besides the majority of cell pairs (60%), which showed an ohmic behavior of gap junctional coupling (Fig. 2), two further populations of cell pairs were identified, according to their voltage-dependent response after a voltage step. In these two populations the instantaneous and corresponding steady-state values of G j differ significantly. One population of cell pairs (5%) showed a gap junctional current or conductance that became activated in a time-dependent but voltage-independent manner (Fig. 4). The other population of cell pairs (15%) exhibited a superimposition of a time-dependent activation and inactivation (data not shown).

Variable voltage dependence is known for primary cell systems and has been observed for e.g., coupled Hensen cells of the organ of Corti in the inner ear (Zhao & Santos-Sacchi, 2000; Todt et al., 2001). This variability has been interpreted as a position-related differentiation. In the follicle, granulosa cells are organized in concentric layers. Gap junctional coupling could also vary within the spatial pattern of these cell layers. However, in a cell culture without the necessity of a three-dimensional organization, such a pattern of variable gap junctional coupling is difficult to understand. At the molecular level, the variability in transjunctional voltage-dependent channel gating has been associated with the coexpression of different connexin types, which form homotypic, heterotypic and heteromeric gap junctional channels (Sosinsky; 1995; White & Bruzzone, 1996; Harris, 2001). Furthermore, posttranslational modifications like phosphorylation have also been related to variations in voltage-dependent gating of gap junctional channels (Godwin et al., 1993; Ngezahayo et al., 1998). These modifications of connexins are dynamic processes that can be modulated by the metabolic and the developmental state of the cells (Kanemitsu, Jiang & Eckhart, 1998). It can be speculated that Cx45 assembles with Cx43 to form heteromeric channels whose voltage dependence varies depending on the stoichiometry and posttranslational modification of the connexins. Since posttranslational modifications of the connexins may reflect developmental state of the cells (Sommersberg et al., 2000), further studies, including with synchronized GFSHR-17 cells, are necessary to elucidate gap junctional coupling of granulosa cells as a function of cell cycle.

Effect of Whole-cell Formation on [CA2+]I and Gap Junctional Coupling

With control bath and absence of cGMP in the pipette solution (Solution A, Table 1), the formation of whole-cell configuration caused several significant cellular changes: a sustained increase in [Ca2+]i (Fig. 5 A) that reached the maximal level within less than 3 minutes; cell shrinkage within 2–4 minutes after whole-cell formation, which was completed after about 5 minutes. Within this phase, the cell diameter decreased by about 25% (Fig. 5 B). Thereafter gap junctional coupling decreased significantly to less than 1% of the starting value within 15 minutes (Fig. 5 C). It is interesting that this spontaneous gap junctional uncoupling by use of a pure electrolyte as pipette solution was not affected by addition of cAMP. The addition of this cyclic nucleotide had been found to be essential for stable gap junctional coupling during double whole-cell configuration in other cells, like pancreatic acinar cells (Somogyi, Batzer & Kolb, 1989).

The observed cellular changes could be suppressed if the whole-cell configuration was established using a pipette solution containing Cs+ (Fig. 7; Table 1, Solution B) or if Ca2+ in the extracellular solution was chelated by EGTA (Fig. 6). However, even use of a pipette solution containing Cs+ or Ca2+-free extracellular solution did not suppress the generation of a Ca2+ spike (Figs. 7 B and 6A) by formation of whole-cell configuration. Surprisingly, the time course and amplitude of the Ca2+ increase differed for the two cells of a coupled pair during the consecutive formation of the corresponding whole-cell configuration. At the first whole-cell formation a steep and significant rise in [Ca2+]i was registered, but only a slow and minor response of [Ca2+]i could be observed in the neighboring cell, while the second pipette was still being kept in the cell-attached configuration (Fig. 5 A). Since the two cells are coupled by gap junctional channels, it should be expected that Ca2+ diffuses from one cell into the other, so that the [Ca2+]i response should be of similar amplitude in both cells. Fig. 5 B could indicate that Ca2+ either does not diffuse freely into the neighbouring cell or becomes rapidly stored in intracellular compartments. Lack of visible diffusion is not surprising since it is known that gap junctions are differently permeable to ions and small molecules (Veenestra et al., 1994; White & Paul, 1999). But it should be noted that whole-cell formation of the 2nd cell again caused a significant rise of [Ca2+]i in this cell, which in this case partially was accompanied by an increase in the first cell. Such an asymmetric response of the two cells to whole-cell formation with pipette solution A was recorded in all investigated cell pairs (n = 10).

The induced pronounced Ca2+ spike by whole-cell formation is probably caused by washout of regulatory molecules like cGMP (see below), which induces a Ca2+ release from intracellular Ca2+ stores. Therefore, it is assumed that washout of cGMP, but not of cAMP, by the control pipette solution induces a Ca2+ release from intracellular stores, which activates a sustained Ca2+ influx and Ca2+-dependent K+ efflux. The activation of these ion fluxes is followed by volume shrinkage and gap junctional uncoupling. Presence of cGMP could directly regulate a K+ efflux, since a cGMP-dependent inhibition of K+ channels has been observed in immortalized human kidney epithelial cells (Hirsch et al., 1999). The question arises whether a cGMP-dependent activation of protein kinase, PKG, known for phosphorylation of Cx43, is the key mechanism in gating gap junctional coupling of GFSHR-17 cells. But the observed stable coupling after washout of cGMP when K+ efflux or Ca2+ influx was inhibited seems to indicate the involvement of further gating mechanisms not related to cGMP-dependent connexin phosphorylation or Ca2+-dependent closure of gap junctional channels. Gap junctional uncoupling by altering the osmolarities of extracellular solutions or pipette pressure have been observed in some cell systems like pancreatic acinar cells (Ngezahayo & Kolb, 1990) and supporting cells in Corti's organ (Zhao & Santos-Sacchi, 1998), but in the presented experiments, morphological changes were not generated by changing the membrane tension.

Interestingly, increase of [Ca2+]i and a Ca2+-dependent K+ efflux have been demonstrated to serve as a trigger for cell shrinkage and caspase activation, which are major characteristics of apoptosis/necrosis (Dallaporta et al., 1998; Bortner and Cidlowski, 1999; Gomez-Angelats, Bortner & Cidlowski, 2000; Maeno et al., 2000; Krick et al., 2001). Thus, the whole-cell formation at control pipette solution could induce apoptotic/necrotic processes, e.g., activation of caspase and other proteases that break down the cytoskeletal system, and thereby mediates a decrease of gap junctional communication. This assumption is supported by the finding that 30–60 min after gap junctional uncoupling the whole-cell configuration induced formation of blebs (data not shown) similar to apoptotic bodies.

Addition of cGMP to the control pipette solution resulted in unchanged morphological cell shape, the absence of a significant change of [Ca2+]i and stable gap junctional coupling (Fig. 8). cGMP is one of the messenger molecules known to diffuse through gap junctional channels (Bevans et al., 1998), whose expression is ubiquitous and which regulates different cellular functions like K+ efflux (Hirsch et al., 1999) and apoptosis (McGee et al., 1997; Estévez et al., 1998). There are several lines of evidence that loss of cytoplasmic K+ is responsible for cell shrinkage and that it also is a major characteristic of apopotosis (Bortner, Hughes & Cidlowski, 1997; Bortner & Cidlowski, 1999), especially in granulosa cells (Perez et al., 2000). An anti-apoptotic effect of cGMP has been reported in cultured astrocytes (Takuma et al., 2001). In granulosa cells, cGMP is synthesized during the follicular development (Gutkowska et al., 1999; Noubani, Farookhi & Gutkowska, 2000). To account for the presented results, we postulate that whole-cell formation in the absence of cGMP in the pipette solution reduces intracellular cGMP concentration. Depletion of cGMP causes liberation of Ca2+ from intracellular stores, a Ca2+ influx and promotion of a Ca2+-dependent K+ efflux. We propose that superposition of these ion fluxes leads to volume shrinkage, disruption of the cytoskeleton and thereby, by an unknown mechanism, to gap junctional uncoupling.