Expression of Purinergic Receptor Channels and Their Role in Calcium Signaling and Hormone Release in Pituitary Gonadotrophs INTEGRATION OF P 2 CHANNELS IN PLASMA MEMBRANE- AND ENDOPLASMIC RETICULUM-DERIVED CALCIUM OSCILLATIONS*

The role of ATP as a positive feedback element in Ca 2 (cid:49) signaling and secretion was examined in female rat pi- tuitary gonadotrophs. ATP and ADP, but not AMP or adenosine, induced a dose- and extracellular Ca 2 (cid:49) -de-pendent rise in [Ca 2 (cid:49) ] i in identified gonadotrophs in a Mg 2 (cid:49) - and suramin-sensitive manner. ATP, adenosine-5 (cid:42) - O -(3-thiotriphosphate), adenosine-5 (cid:42) - O -(1-thiotriphos-phate), 2-methylthio-ATP, and 3 (cid:42) - O -(4-benzoyl)benzoyl-ATP were roughly equipotent in rising [Ca 2 (cid:49) ] i in gonadotrophs, while ADP was effective only at submil- limolar concentration range, and none of these com-pounds permeabilized the cells. On the other hand, (cid:97) , (cid:98) -methylene-ATP, (cid:98) , (cid:103) -methylene-ATP, and UTP were unable to induce any rise in [Ca 2 (cid:49) ] i . This pharmacological profile is consistent with expression of P2X 2 and/or P2X 5 purinergic receptor channels. Patch-clamp exper- iments showed that ATP induced an inward depolarizing current in gonadotrophs

leased into the extracellular medium, ATP may bind to its plasma membrane purinergic receptors (P 2 ), 1 leading to increases in [Ca 2ϩ ] i and potentiation of the exocytosis (1)(2)(3). ATP receptors belong to two major groups: G protein-coupled receptors (P 2U and P 2Y ), whose activation causes Ca 2ϩ release from its intracellular stores, and plasma membrane receptor channels (P 2X ), whose activation promotes Ca 2ϩ influx (4,5). Thus, the action of ATP during exocytosis may represent a positive feedback mechanism or self-potentiation of the calcium signaling and secretion (autocrine regulation). In addition, ATP may diffuse and activate P 2 receptors in neighboring cells (paracrine regulation). On the other hand, ATP cannot act as a typical hormone, since it is rapidly degraded by several ecto-ATPases (4).
Purinergic receptor channels have been identified in both excitable and non-excitable cells (6 -9). Calcium has been shown to carry current through these channels, and the permeability of Ca 2ϩ versus Na ϩ varies widely among cells (10). Furthermore, since ATP-controlled channels lead to depolarization, [Ca 2ϩ ] i response to ATP could result from activation of voltage-sensitive calcium channels (VSCC), in addition to the direct entry of Ca 2ϩ through purinergic channels (10). In several cell types that secrete by exocytosis, including pituitary cells (11), spontaneous firing of action potentials (APs) at resting potential is associated with [Ca 2ϩ ] i oscillations, but the possible participation of P 2 receptor channels in such Ca 2ϩ spiking (termed the plasma membrane oscillator) has not been addressed. In general, these channels could modulate secretion by changing the frequency, amplitude, or duration of APs and their resulting increases in [Ca 2ϩ ] i . In accord with this, an ATP analog was found to increase the firing rate in spontaneously active noradrenergic neurons at the resting potential (12).
ATP-gated Ca 2ϩ channels may also represent the Ca 2ϩ influx pathway that sustains InsP 3 -induced and endoplasmic reticulum (ER)-derived Ca 2ϩ spiking and Ca 2ϩ -controlled exocytosis. Such oscillations, that have frequency of spiking controlled by agonist concentration, were observed in both excitable and nonexcitable cells (11,13,14). In excitable cells, VSCC supply the cells with Ca 2ϩ during sustained stimulation (15), while in several non-excitable cells, Ca 2ϩ entry occurs through store-operated calcium channels (16). However, the role of P 2 receptor channels as exclusive or additional Ca 2ϩ influx path-ways in ER-derived oscillations in both cell types has not been addressed.
Calcium-mobilizing P 2U receptors have been identified in pituitary cells (17)(18)(19)(20)(21). We have recently observed that rat pituitary gonadotrophs express P 2 receptor channels. Since gonadotrophs are excitable (22), and agonist stimulation leads to the activation of ER-derived oscillations (15), these cells were employed to address several of the questions raised above. We have examined ATP-induced activation of the plasma membrane oscillator in silent cells, and the modulation of its activity in spontaneously active cells. Furthermore, the effects of exogenous ATP on gonadotropin-releasing hormone (GnRH)-induced and InsP 3 -controlled Ca 2ϩ oscillations were analyzed. We have also evaluated the impact of modulation of the plasma membrane and ER oscillators by ATP on gonadotropin secretion. In addition, the cosecretion of ATP during agonist stimulation was monitored. Our results indicate that ATP has a role as a positive feedback element in agonist-induced Ca 2ϩ signaling and gonadotropin secretion.

MATERIALS AND METHODS
Chemicals-GnRH was obtained from Peninsula Laboratories, fura-2 AM and mag-fura-2 AM from Molecular Probes; streptolysin O from Difco; tetrodotoxin, suramin, and adenosine-5Ј-O-(1-thiotriphosphate) from Calbiochem; and pyruvate kinase and myokinase from Boehringer Mannheim. All other chemicals were purchased from Sigma.
Pituitary Cell Culture-Experiments were performed on anterior pituitary cells from normal and ovariectomized female Sprague-Dawley rats obtained from Charles River Inc. (Wilmington, MA). Groups of 50 anterior pituitary glands were rinsed several times in dispersion medium: Dulbecco's phosphate-buffered saline, without Ca 2ϩ and Mg 2ϩ , containing 0.1% bovine serum albumin (BSA), 100 units/ml penicillin, and 100 g/ml streptomycin. They were then minced into 0.7-mm cubes with a tissue chopper (Brinkmann Instruments Inc, Westbury, NY), washed in dispersion medium, and incubated for 1 h in a shaking waterbath at 37°C in 50 ml of dispersion medium containing 75 mg of trypsin. This medium was discarded and replaced by 50 ml of fresh dispersion medium containing 75 mg of trypsin inhibitor and 10 g of DNase for an additional 10-min incubation. The pituitary tissue was then disrupted by repeated aspiration into a plastic transfer pipette and the dispersed cells were filtered through nylon mesh (80 m). Viable cells were counted in the presence of trypan blue and cultured in medium 199 containing Earle's salts, sodium bicarbonate, 10% horse serum, and antibiotics.
Measurements of Calcium Ion Concentration-For cytosolic and luminal Ca 2ϩ concentration measurements, cells (10 6 /dish) were plated on coverslips coated with poly-L-lysine and cultured in medium 199 containing Earle's salts, sodium bicarbonate, 10% horse serum, and antibiotics. The next day, the cells were incubated at 37°C for 60 min with 2 M fura-2 AM for cytosolic Ca 2ϩ measurements, and for 90 min with 7 M mag-fura-2 AM for luminal Ca 2ϩ measurements. The extracellular buffer used in Ca 2ϩ measurements was phenol red-free M199 with Hank's salts or modified Krebs-Ringer without Mg 2ϩ . When the effects of sodium ions were studied, NaCl was substituted with Nmethyl-D-glucamine chloride. Coverslips with cells were washed with phenol red-free buffer and mounted on the stage of an Axiovert 135 microscope (Carl Zeiss, Oberkochen, Germany) attached to the Attofluor Digital Fluorescence Microscopy System (Atto Instruments, Rockville, MD). Cells were examined under a 40ϫ oil immersion objective during exposure to alternating 340 and 380 nm light beams, and the intensity of light emission at 520 nm was measured. In this way, light intensities and their ratio, F 340 /F 380 , which reflects changes in Ca 2ϩ concentration, were followed in several single cells simultaneously. GnRH was added as a final stimulus in each experiment where [Ca 2ϩ ] i kinetics was followed to identify the cell as a gonadotroph and to compare relative [Ca 2ϩ ] i responses induced by ATP (or its analogues) to the maximal [Ca 2ϩ ] i amplitudes induced by GnRH.
"Intracellular medium" used in experiments with permeabilized cells contained 124 mM potassium glutamate, 6 mM magnesium acetate, 0.2 mM EGTA, 20 mM PIPES, 0.1% BSA, and its pH was adjusted by KOH. In cells loaded with mag-fura-2, decreases in fluorescence were observed 1-5 min after addition of streptolysin O; F 340 dropped by about 60%, whereas F 380 decreased by about 80%, increasing the F 340 /F 380 ratio. Such a decrease in fluorescence is due to permeabilization and consequent leakage of mag-fura-2 from the cytosol, and the remaining dye is believed to be compartmentalized within the endoplasmic reticulum (23). Whereas the initial F 340 /F 380 ratio reflects the concentrations of Mg 2ϩ and Ca 2ϩ in the cytosol, ratios in permeabilized cells reflect primarily the concentration of Ca 2ϩ in ER (23,24).
Simultaneous Measurements of [Ca 2ϩ ] i and ATP-induced Current-Simultaneous measurements of [Ca 2ϩ ] i and current in single isolated gonadotrophs were made as described previously (25). The cells were plated on glass coverslips (10 6 cells/dish) and kept in culture for 2-3 days at 37°C in an atmosphere of 95% air and 5% CO 2 . Experiments were carried out at room temperature (20 -25°C). Cells were loaded with indo-1 AM at room temperature (20 -25°C) by incubation for 30 -60 min in a loading solution containing 2 M indo-1 AM and 0.02% pluronic acid (Molecular Probes, Inc., Eugene, OR), and the following composition of salts (in mM): 140 NaCl, 5 KCl, 2.6 CaCl 2 , 1 MgCl 2 , 5 NaHepes, 5 glucose, pH ϭ 7.4. Coverslips were mounted in an open chamber designed to fit in the stage of an inverted epifluorescence microscope (Carl Zeiss, ICM 405, Oberkochen, Germany) equipped with a photon counter system (Photon Technology International Inc., Brunswick, NJ) to measure simultaneously the intensity at 405 and 480 nm from the acid form of indo-1 trapped inside the cells and excited at 365 nm. The data were acquired using a PC provided with an acquisition board. The software (OSCAR) provided with the system allowed corrections to be made for the background intensity of both the light emitted at 405 and 485 nm. Results were expressed as the corrected ratio between the fluorescent signals recorded at 410 and 485 nm.
The open recording chamber allowed access to cells for electrophysiological studies. Whole-cell currents under voltage-clamp conditions were measured using the nystatin-perforated membrane patch method (58). Membrane currents were recorded using an EPC-7 amplifier (List-Electronics, Darmstadt-Eberstadt, Germany). Patch pipettes were pulled using a BB-CH-PC (Mecanex, Geneva, Switzerland) from microhematocrit capillary tubes. Pipettes filled with a high [K ϩ ] solution (in mM: 20 KCl, 120 potassium aspartate, 10 KHepes, 3 MgCl 2 , pH 7.2, and 200 g/ml nystatin) had tip resistance in the range from 2 to 4 M⍀. Acquisition of the light signals emitted at 405 and 485 nm, and electrophysiological recording were synchronized using a common external trigger logic pulse.
Measurements of LH and ATP-Column perifusions were performed on 3-day-old cultured cells under previously reported conditions (26). Briefly, 2 ϫ 10 7 cells were incubated with preswollen Cytodex-1 beads in 60 mm culture dishes, and perifused with Hanks' M199 containing 20 mM HEPES and 0.1% BSA for 60 min at a flow rate of 0.6 ml/min. After agonist and/or ATP stimulation, fractions were collected every minute and stored at Ϫ20°C. LH was determined by radioimmunoassay, using the reagents and standards provided by the National Pituitary Agency (Baltimore, MD).
For measurement of ATP secretion and degradation, columns were perifused with Krebs-Ringer medium without Mg 2ϩ at a flow rate of 0.15 ml/min. ATP secretion and degradation was also examined in static cultures (0.5 ϫ 10 6 cells/well). The concentration of ATP in the effluent was determined by an ATP bioluminescent assay kit (Sigma) in an AutoLumat LB 953 (Berthold, Wildbad, Germany) by injection of 100 M assay solution into an aliquot of 100 M sample. Calibration curves were constructed from measurements in standard solutions, which were diluted in the same medium as the corresponding solutions of unknown ATP concentration. Detection limit of the assay was 0.02 nM.
Total concentration of adenosine nucleotides was also measured in several experiments, by employing the following enzymatic reactions to convert ADP and AMP to ATP (27,28), where PEP is phospho(enol)pyruvate. Aliquots of 50 l of sample were added to 50 l of the rephosphorylation mixture (10 units/ml myokinase, 6 units/ml pyruvate kinase, 1.5 mM PEP, 25 mM Hepes, 15 mM KOH, 10 mM MgSO 4 ) and after 1 h at room temperature, ATP concentration was measured as described above. Due to relatively high bioluminescence of the enzyme solution, detection limit of the measurements of total adenosine nucleotides concentration was increased to 10 nM. In the control experiments, the recovery of AMP was higher than 95%.

Extracellular Ca 2ϩ Dependence of ATP Actions in Gonadotrophs-Addition of ATP induced a rapid increase in [Ca 2ϩ
] i in single rat pituitary gonadotrophs, which were identified by addition of GnRH at the end of recording (Figs. [1][2][3][4][5]. Different patterns of [Ca 2ϩ ] i responses to ATP in single rat pituitary gonadotrophs bathed in Ca 2ϩ -and Mg 2ϩ -containing medium were observed (Fig. 1, left panel). A Ca 2ϩ spike followed by a plateau was elicited by addition of 100-1000 M ATP, and a monophasic long-lasting [Ca 2ϩ ] i increase or low amplitude fluctuations in [Ca 2ϩ ] i were observed at the 1-100 M concentration range. The plateau induced by ATP was always lower than GnRH-induced spike. Furthermore, GnRH-induced [Ca 2ϩ ] i responses after prolonged exposure to ATP were comparable to those induced by GnRH itself. This indicates that no obvious depletion of the intracellular Ca 2ϩ pool was induced by ATP, as usually observed upon activation of G protein-coupled receptors.
The dependence of Ca 2ϩ response on ATP concentration was not only indicated by the pattern of Ca 2ϩ response, but also by its amplitude; exposure of a single gonadotroph to increasing ATP concentrations caused a step-like increase in [Ca 2ϩ ] i ( Fig.  1, upper right panel). A concentration-response curve ( Fig. 1, lower right panel) was derived by expressing ATP-induced [Ca 2ϩ ] i increases in Mg 2ϩ -deficient medium in relation to the GnRH-induced response of the same cell. The lowest ATP concentration that gave a detectable response was 1 M, and a plateau was observed at 100 M ATP; fitting a logistic function to the data gave an EC 50 of 8 M.
In Ca 2ϩ -deficient medium, ATP did not increase [Ca 2ϩ ] i ( Fig.  2A), but the subsequent introduction of 1.2 mM Ca 2ϩ into the medium increased [Ca 2ϩ ] i in a manner comparable to the ATPinduced response (Fig. 2B, upper trace). Addition of Ca 2ϩ alone had a negligible effect on [Ca 2ϩ ] i (Fig. 2B, lower trace). The ATP-induced [Ca 2ϩ ] i response was therefore generated by an influx of Ca 2ϩ across the plasma membrane and not by Ca 2ϩ released from its intracellular stores, indicating that gonadotrophs from ovariectomized female rats express purinergic P 2 receptor channels.
Further evidence that ATP does not induce Ca 2ϩ mobilization in gonadotrophs was obtained in experiments with permeabilized cells where changes in luminal Ca 2ϩ concentration were measured. It has been shown that GnRH-induced Ca 2ϩ release from the intracellular pool occurs in permeabilized rat gonadotrophs under certain conditions and is primarily dependent on the concentration of free Ca 2ϩ in the medium (24). Similar responses were observed in permeabilized gonadotrophs upon stimulation by endothelin-1, as well as by InsP 3 . In contrast to these agonists, ATP did not induce detectable Ca 2ϩ release in such a system (data not shown).
Pituitary gonadotrophs express several plasma membrane channels, including two types of VSCC and voltage-sensitive Na ϩ channels (22). To exclude the possible participation of these channels in the ATP-induced [Ca 2ϩ ] i response, patchclamp techniques with simultaneous measurements of current and [Ca 2ϩ ] i responses were employed. Fig. 2C shows a wholecell recording from a nystatin-perforated gonadotroph held at Ϫ50 mV. At this potential, 10 nM GnRH evoked a typical outward I K(Ca) current (upper panel) that was driven by changes in [Ca 2ϩ ] i (bottom panel). In this experiment, 100 nM ATP was added in the continuous presence of the agonist, and after the oscillatory response was extinguished due to depletion of intracellular Ca 2ϩ stores. Prior to application of 100 M ATP, the cell was hyperpolarized to exclude any participation of VSCC in current response. Under these conditions, ATP induced an inward (depolarizing) current of about 50 pA in cells bathed in normal Krebs-Ringer buffer. An increase in [Ca 2ϩ ] i was observed in parallel to the current response, confirming that Ca 2ϩ entry occurs through the plasma membrane channels that are distinct from VSCC. However, the amplitude of this response was reduced when compared to responses in intact cells. Furthermore, in contrast to cells expressing multiphasic inward currents (29), ATP induced a monophasic response in gonadotrophs, suggesting the expression of a single class of purinergic channels.
In additional studies, we have employed two blockers of VSCC, nifedipine and verapamil, and a blocker of voltagesensitive Na ϩ channels, tetrodotoxin. The ATP-induced [Ca 2ϩ ] i response was not abolished in gonadotrophs treated with nifedipine (Fig. 3A) and verapamil (results not shown). However, the addition of nifedipine to ATP-stimulated cells decreased the plateau of [Ca 2ϩ ] i response, but the newly achieved plateau was always above basal [Ca 2ϩ ] i . These observations, and results presented in Fig. 2C, indicate a dual action of ATP in gonadotrophs: direct stimulation of Ca 2ϩ influx through its own channels and consequent influx through dihydropyridinesensitive VSCC.
In contrast to the VSCC blockers, tetrodotoxin was ineffective when applied before or during ATP stimulation (Fig. 3B), indicating that ATP does not act through Na ϩ channels. The effects of ATP on the [Ca 2ϩ ] i response were further examined in normal Krebs-Ringer buffer (controls) and in buffer in which Na ϩ was replaced by the non-permeant ion, N-methyl-D-glucamine. As shown in Fig. 3C, the addition of Na ϩ per se was ineffective, while its addition to ATP-stimulated gonadotrophs further increased [Ca 2ϩ ] i . These results indicate that P 2 receptor channels conduct both Na ϩ and Ca 2ϩ ions, which are responsible for depolarization of the cells and subsequent activation of VSCC.
Pharmacological Characterization of P 2 Receptor Channels in Gonadotrophs-Recently, several subtypes of P 2X and P 2Y receptors were cloned, and it is possible to a certain extent to distinguish them pharmacologically, employing several P 2 agonists and antagonists. In pituitary cell types other than gonadotrophs, UTP induced [Ca 2ϩ ] i response (results not shown), but no response was observed in freshly dispersed and cultured gonadotrophs from ovariectomized rats (Fig. 4A), confirming the absence of G-protein-coupled P 2U and P 2Y receptors in these cells. Two additional ATP agonists, ␣,␤-methylene-ATP (Fig. 4B) and ␤,␥-methylene-ATP (results not shown) did not induce a [Ca 2ϩ ] i response, arguing against the expression of P 2Y phospholipase C-coupled receptors in gonadotrophs. The action of ATP and its analogs was also analyzed in gonadotrophs from cycling female rats, and the same extracellular Ca 2ϩ dependence of [Ca 2ϩ ] i response was found. In both cell types, the Ca 2ϩ dependence of ATP action was observed over temperature range of a 22-35°C, and in cells immediately after dispersion, as well as cells cultured for 1-4 days. Thus, it is unlikely that G protein-coupled purinergic (P 2Y or P 2U ) receptors are coexpressed in pituitary gonadotrophs from Sprague-Dawley rats. The lack of effects of ␣,␤-methylene-ATP and ␤,␥-methylene-ATP also argues against the expression of P2X 1 (30) and P2X 3 (31, 32) receptor channels in pituitary gonadotrophs. Furthermore, inhibiting effects of suramin, a P 2 receptor antagonist, on ATP-induced [Ca 2ϩ ] i response (Fig. 4C) exclude the possibility that P2X 4 and P2X 6 receptor subtypes (33)(34)(35) are expressed in these cells. The highly potent agonist for the P2X 7 , i.e. P 2Z purinergic receptor channels (36), 3Ј-O-(4-benzoyl)benzoyl-ATP (BzATP) induced rise in [Ca 2ϩ ] i in gonadotrophs (Fig. 4D), but it was equally potent as ATP. Except for inducing an inward current associated with a rise in [Ca 2ϩ ] i , ATP and BzATP also induce lysis of cells expressing native and cloned channels through the formation of the membrane pores permeable to large molecules (28,36). In gonadotrophs, ATP and BzATP did not show the lysic effects at 22-35°C, arguing against the expression of P2X 7 channels. As shown in Fig. 4E, addition of Mg 2ϩ to ATP-stimulated cells caused a rapid drop in [Ca 2ϩ ] i to its basal value, consistent with conclusion that ATP 4Ϫ is responsible for the [Ca 2ϩ ] i increase. Moreover, the ability of suramin and Mg 2ϩ to decrease the ATP-induced rise in [Ca 2ϩ ] i confirms that gonadotrophs were not permeabilized by ATP. Furthermore, the rate of fluorescence decrease was not enhanced by application of ATP (Fig. 4E), showing that leakage of fura-2 did not occur.
In addition to BzATP, ATP␣S (Fig. 5A), ATP␥S (Fig. 5B), and 2 methylthio-ATP (2-MeS-ATP) (Fig. 5C), were found to be roughly equipotent as agonists. Such pharmacological profile is comparable but not identical with those of P2X 2 (37) and P2X 5 (35) cloned channels. P2X 2 channel show a lower sensitivity to ATP (EC 50 of 60 M), that was increased in the presence of Zn 2ϩ (37). In contrast, addition of Zn 2ϩ (10 M) did not affect ATP (1-100 M)-induced [Ca 2ϩ ] i responses in gonadotrophs (data not shown). There is also an obvious difference in the amplitude of current triggered by ATP in cells expressing P2X 2 channel and gonadotrophs. On the other hand, the amplitude of P 2 current in gonadotrophs was comparable to that observed in cells expressing P2X 5 channel. In addition, half-maximal concentrations for ATP, 2-MeS-ATP, and ATP␥S for this channel are comparable to that of gonadotrophs. P2X 5 channel was reported to be responsive to ADP, with an EC 50 of 270 M (35). We have also observed the effects of ADP at submillimolar concentration range (Fig. 5D), but in our experiments this observation should be considered with reservation, because ADP contained 0.4% ATP. Thus, such quantitative differences in pharmacological profiles are not sufficient to distinguish which subtype(s) of P 2 channels are expressed in gonadotrophs, P2X 5 , and/or P2X 2 .
Effects of ATP on Oscillatory Ca 2ϩ Signaling-Pituitary gonadotrophs show spontaneous firing of APs and AP-driven low amplitude oscillatory Ca 2ϩ signals (22). To evaluate the effects of ATP on this oscillator, the depolarizing action of P 2 receptor channels was evaluated in quiescent and spontaneously active cells. As shown in Fig. 6A depolarizes the cells and activates the plasma membrane oscillator. Nevertheless, ATP is not essential for the endogenous operation of the plasma membrane oscillator.
In cells depolarized by 50 mM KCl, the operation of the plasma membrane oscillator was inhibited and a non-oscillatory high amplitude increase in [Ca 2ϩ ] i was detected (38). Similarly, in both ATP-activated and spontaneously active cells, addition of high concentrations of ATP caused the transformation of the oscillatory [Ca 2ϩ ] i patterns to a high amplitude non-oscillatory increase in [Ca 2ϩ ] i (Fig. 6C). The subsequent addition of suramin or Mg 2ϩ frequently recovered the oscillatory response (Fig. 6D). These results further indicate that the P 2 influx current is depolarizing and that the operation of the plasma membrane oscillator can be initiated, facilitated, or inhibited, depending on the ambient ATP 4Ϫ concentration.
In addition to the plasma membrane oscillator, an ER oscillator is also operative in gonadotrophs. Ca 2ϩ -mobilizing agonists including GnRH, ET-1, and PACAP induce rapid (5-25 spikes/min), [Ca 2ϩ ] i oscillations that may last 3-10 min in cells bathed in Ca 2ϩ -deficient medium, and up to 40 min in Ca 2ϩcontaining medium (15). It is therefore reasonable to expect that extracellular ATP will affect signaling by enhancing the existing Ca 2ϩ influx in GnRH-stimulated cells. As shown in Fig. 7A, addition of ATP during the sustained extracellular Ca 2ϩ -dependent phase of signaling induced two types of responses. In some cells, it increased [Ca 2ϩ ] i above spike amplitude and abolished spiking. In others, addition of ATP reinitiated or remodulated agonist-induced Ca 2ϩ spiking. In general, the first type of response was observed at higher (above 100 M) and the second type of response at lower (below 100 M) ATP concentrations in cells bathed in Mg 2ϩ -containing medium. Low concentrations of ATP not only raised the base line of the oscillations, but also increased the frequency of spiking (Fig.  7B). Thus, if ATP-induced Ca 2ϩ influx is lower, the ER oscillator can integrate it without disrupting the oscillatory mode of signaling. Otherwise, the oscillatory response is abolished, presumably due to interference by Ca 2ϩ influx with the bidirectional action of [Ca 2ϩ ] i on InsP 3 channel activity (39 -41).
Effects of Extracellular ATP on Basal and GnRH-induced LH Release-As already reported (19), perifused pituitary cells attached to beads release LH in response to ATP stimulation. In our experiments, the threshold concentration of ATP to activate exocytosis in cells bathed in Mg 2ϩ -containing medium was higher than 5 M. The peak of ATP-induced LH concentration was comparable to the value of the GnRH-induced plateau when the ATP concentration was 10 M or higher (Fig. 8A). Interestingly, although ATP induced long lasting and slowly falling [Ca 2ϩ ] i increases in isolated cells, the kinetics of ATPinduced LH secretion in perifused cells had a different profile; a peak with a rapid drop to near-basal values (see also "Degradation of ATP by Pituitary Cells").
In GnRH-stimulated cells, addition of 10 M ATP further enhanced the secretory response. In perifused cells isolated from random cycling female rats, simultaneous addition of GnRH and ATP led to an amplification of secretion during the sustained extracellular Ca 2ϩ -dependent phase (Fig. 8B). In pituitary cells from ovariectomized rats, addition of ATP during the sustained phase of GnRH stimulation also enhanced gonadotropin secretion (Fig. 8C). The effect of ATP on basal and GnRH-induced gonadotropin secretion was concentrationdependent, with an EC 50 similar to that observed in [Ca 2ϩ ] i measurements, and was abolished in cells bathed in Ca 2ϩdeficient medium (data not shown).
Secretion and Degradation of ATP by Pituitary Cells-The finding that ATP, by enhancing Ca 2ϩ influx, can prolong and amplify GnRH-induced [Ca 2ϩ ] i oscillations, as well as induce LH secretion, makes ATP a candidate for a positive feedback element in agonist-induced Ca 2ϩ signaling and hormone secretion. However, under conditions normally used to measure LH release, no measurable (detection limit 0.02 nM) ATP was secreted from GnRH-stimulated cells. Further experiments revealed that detection of ATP released from GnRH-stimulated cells was prevented by its rapid degradation in the presence of pituitary cells. When a column with 20 million cells was perifused at a rate of 0.15 ml/min, a 10-min pulse of 10 M ATP was reduced to around 2 M in the effluent, while a 1.2 M pulse was degraded to less then 0.2 M ATP. When the same pulses were applied to columns without cells, no significant degradation or adsorption of ATP was observed (Fig. 9A). Such a rapid degradation of ATP in the presence of the cells indicated the operation of ecto-ATPases. To permit the measurement of ATP se-cretion, several ecto-ATPase inhibitors were employed: N-ethylmaleimide, trifluoroperazine, etharinic acid, and the removal from external medium of Ca 2ϩ and Mg 2ϩ , the ions required for activation of ecto-ATPases (27). Only a minor reduction of ecto-ATPase activity was observed in cells treated with these inhibitors or in cells perifused with Ca 2ϩ /Mg 2ϩdeficient medium. However, the presence of Ca 2ϩ /Mg 2ϩ chelators, EGTA or EDTA, in the medium significantly inhibited ATP degradation (Fig. 9B). Furthermore, EGTA increased the basal secretion of ATP from non-detectable to about 1 nM.
An increase in basal ATP was also detected in the effluent of EDTA-perifused cells; 10 min after switching to the perifusion medium containing 1 mM EDTA, ATP concentration in the effluent was still rising. Nevertheless, a GnRH pulse further increased ATP levels (Fig. 9C). Cessation of GnRH stimulation was followed by a fall in ATP concentration to a plateau value, and a subsequent GnRH pulse induced an additional ATP response. After being washed, the cells were perifused with a 50 nM ATP pulse, and only abut 10% of applied ATP remained in the effluent. The next pulse contained 200 nM ATP and the corresponding plateau in the effluent was 25 nM (data not shown). Therefore, the estimated concentration of GnRH-induced ATP secretion in medium containing 1 mM EDTA under the described perifusion conditions is about 40 nM, and most of the secreted amount is quickly degraded, leaving about 4 nM in the effluent. GnRH-induced ATP secretion was observed in two out of three perifusion experiments (where all perifusion media contained EDTA), but was never detected in experiments where EDTA was not used.
To estimate the concentration of total adenosine nucleotides in the column perfusate and in culture medium of the cells plated on dishes, AMP and ADP in the medium were enzymatically converted to ATP prior to luminometric determination of ATP concentration. In the absence of the cells, 1 h of exposure to myokinase and pyruvate kinase was sufficient to rephosphorylate 96 -99% of 1 M AMP to ATP. However, in the presence of the cells in static cultures only about 17% of added 100 nM or 1 M ATP (in the medium containing 1.2 mM Ca 2ϩ ) was recovered (Table I), while in perifused cells no recovery was observed. These observations are consistent with conclusion that, in addition to ecto-ATPase, pituitary cells also express ecto-ADPase and ecto-5Ј-nucleotidase, thus, degrading ATP to adenosine.
Like ecto-ATPase, ecto-ADPase and ecto-5Ј-nucleotidase are also inhibited by depletion of Ca 2ϩ and Mg 2ϩ (27). In accord with this, in cells perifused with or bathed in EDTA-containing medium, more than 90% recovery of added ATP was detected. Furthermore, in Ca 2ϩ -containing medium, addition of 50 M ␣,␤-methylene-ADP, an inhibitor of 5Ј-ectonucleotidase (27,28) led to almost complete recovery of exogeneously added ATP in static cultures and to about 40% recovery in perifusate. As shown in Table I, basal secretion of ATP also increased in cells exposed to ␣,␤-methylene-ADP. These observations are consist-  ent with conclusion that ATP is secreted by pituitary cells and rapidly degraded by ectonucleotidases to adenosine.

DISCUSSION
The expression of purinergic receptors in pituitary cells and the effects of ATP on basal LH release have already been reported (21,42). Likewise, we have also observed that ATP induces LH release in perifused pituitary cells, with peak responses lower than those induced by GnRH, and similar to GnRH-induced LH plateaus. In contrast to agonist-induced LH release, ATP-induced gonadotropin secretion rapidly declined after reaching its peak. Such kinetics of LH release from perifused pituitary cells attached on beads differ from the sustained and slowly dropping [Ca 2ϩ ] i responses to ATP in single gonadotrophs attached to coverslips. This discrepancy suggested that ATP is degraded by pituitary cells and that such degradation is faster in 2 ϫ 10 7 cells attached on beads and stimulated at 37°C (secretory studies), than in 10 6 cells cultured on coverslips and stimulated at room temperature ([Ca 2ϩ ] i measurements). Indeed, our results demonstrated substantial ecto-nucleotidase activity in perifused pituitary cells, and somewhat lower activity in cells in static cultures.
The operation of ecto-nucleotidase in pituitary cells was consistent with the hypothesis that ATP is cosecreted by pituitary cells, and that mechanism for its degradation is needed to terminate the signal upon the removal of agonist. However, the only effective way to demonstrate the cosecretion of ATP is to inhibit ecto-nucleotidase activity. Through the use of Ca 2ϩ / Mg 2ϩ chelator, EDTA, we were able to demonstrate the agonist-induced secretion of ATP by pituitary cells. The amplitude of these responses was below the threshold concentrations required for activation of purinergic receptors. However, experiments with rephosphorylation of AMP and ADP and inhibition of ecto-5Ј-nucleotidase by ␣,␤-methylene-ADP (27,28) have revealed that ATP concentration was much higher than that estimated in the presence of EDTA. It is also reasonable to speculate that the intercellular ATP concentration in intact tissue is higher than that measured in our experiments. Thus, it is possible that in pituitary tissue, where there is much less space between cells, the kinetics of ATP degradation by ectonucleotidase and its binding to the receptor have significant effects on signaling. For example, the significance of the geometry of a neuromuscular junction is shown in theoretically predicted current induced by released quanta of ATP (43).
The expression of P 2 receptors, their coupling to Ca 2ϩ signaling, gonadotropin secretion, and cosecretion of ATP with gonadotropins suggests that pituitary cells belong to group of secretory cells in which ATP serves as a positive feed-back element in agonist-induced [Ca 2ϩ ] i signaling and exocytosis. Within the cells expressing P 2 receptors, the effects observed in gonadotrophs are most comparable to those observed in chromaffin cells and PC12 cells. In these related cell types, ATP induced an inward current, increased [Ca 2ϩ ] i , and catecholamine release (44 -46). As in gonadotrophs, ATP does not release [Ca 2ϩ ] i from internal stores in a subpopulation of chromaffin cells bathed in Ca 2ϩ -deficient medium (46). Finally, ATP is cosecreted during agonist-and depolarization-induced secretion of catecholamines (47).
A general feature of purinergic receptors is that their activation is associated with an increase in [Ca 2ϩ ] i . Depending on the receptor type, increases in [Ca 2ϩ ] i can be caused by increased InsP 3 production and consequent Ca 2ϩ release from intracellular pools (G protein/phospholipase C-coupled P 2Y and P 2U receptors) or by Ca 2ϩ influx (P 2X receptor channels) (5). Three lines of evidence argue against the hypothesis that rat pituitary gonadotrophs express G protein-coupled purinergic receptors. First, ATP did not induce [Ca 2ϩ ] i responses in iden-tified gonadotrophs bathed in Ca 2ϩ -deficient medium. Second, UTP, ␣,␤-methylene-ATP, and ␤,␥-methylene-ATP failed to induce a [Ca 2ϩ ] i response. In a mixed population of pituitary cells, several cell types other than gonadotrophs responded to ATP and UTP in Ca 2ϩ -deficient medium, indicating the diversity of the expression of P 2 receptors within the subpopulations of cells. In accord with this, the expression of P 2U -purinoreceptors was found in pituitary lactotrophs (21) and mixed population of pituitary cells (17,18). Third, experiments with permeabilized gonadotrophs showed that ATP did not induce Ca 2ϩ release from the intracellular pool, whereas InsP 3 , as well as activation of G protein-coupled receptors by GnRH and ET-1, induced Ca 2ϩ mobilization.
Our pharmacological and electrophysiological characterization of ATP-induced Ca 2ϩ influx was consistent with the expression of P2X 2 and/or P2X 5 receptor channels in gonadotrophs. The comparable specificity of responses to these two cloned receptors was observed under a wide range of conditions (see "Results"). On the other hand, P 2 receptor channels in gonadotrophs are clearly distinct from the cloned P2X 1 , P2X 3 , P2X 4 , and P2X 6 receptor channels (30 -35). Furthermore, although BzATP induced a rise in [Ca 2ϩ ] i in gonadotrophs, it was equipotent with ATP, contrary to observations with native (29) and cloned (36) P2X 7 receptor channels, which are more sensitive to BzATP. Additionally, permeabilization of gonadotrophs was not observed in response ATP or any of its analogues, again arguing against the expression of P2X 7 receptor channels.
In contrast to well characterized structure and pharmacology, the physiological significance of these receptor channels is not well understood. In this regard, gonadotrophs represent an excellent cell model, especially since both plasma membraneand ER-controlled Ca 2ϩ oscillators are operative in these cells (11). We (22) and others (48 -50) have identified the essential elements of the plasma membrane oscillator in gonadotrophs: T and L voltage-sensitive calcium channels, delayed-rectifier, and apamin-sensitive potassium channels, and plasma membrane (Ca 2ϩ )ATPase. Simultaneous measurements of electrical activity and [Ca 2ϩ ] i in gonadotrophs demonstrated that a single AP is sufficient to drive a [Ca 2ϩ ] i spike with an average amplitude of about 200 nM above the basal level (22). Since spontaneous firing in lactotrophs and somatotrophs is sufficient to trigger exocytosis in these cells (51), it is reasonable to speculate that ATP is cosecreted with GH and prolactin and is involved in modulation of such electrical activity in an autocrine and/or paracrine manner.
The present data indicate that P 2 receptor channels in gonadotrophs are not essential for spontaneous electrical activity, but that their stimulation can activate the plasma membrane oscillator and modulate its activity. As shown in patch-clamp studies, ATP gating of P 2 receptor channels leads to an increase in [Ca 2ϩ ] i that depolarizes gonadotrophs to levels that are sufficient to activate their plasma membrane oscillator in silent cells. In further accord with this, depolarizing current initiates firing of APs in quiescent gonadotrophs and modulates the frequency pattern of firing in spontaneously active cells (22). These observations in gonadotrophs parallel those in noradrenergic neurons, where enzymatically more stable analogues of ATP were found to increase the AP firing rate (12). Thus, ATP increases [Ca 2ϩ ] i by two mechanisms, directly through P 2 receptor channels and indirectly by activation of VSCC.
Gonadotrophs also express an ER oscillator, the operation of which is controlled by InsP 3 and [Ca 2ϩ ] i (52) in a manner comparable to that observed in oocytes, hepatocytes, and pancreatic acinar cells (53)(54)(55). During the prolonged agonist stim-ulation, non-excitable cells are supplied by Ca 2ϩ through storeoperated calcium channels (16), while in excitable gonadotrophs VSCC play such a role. Results from voltageclamped gonadotrophs have indicated that Ca 2ϩ entry through VSCC plays a minor role during the initial phase of intracellular Ca 2ϩ mobilization. However, it sustains the operation of the ER oscillator as the InsP 3 -sensitive Ca 2ϩ pool becomes depleted (15). The expression of the P 2 receptor channels in the same cells may serve as an additional system to provide Ca 2ϩ entry. As shown here, the ER oscillator is able to integrate Ca 2ϩ entry through both VSCC and P 2 channels. Similarly, the P 2 receptor channels in non-excitable cells (7-9) may enhance Ca 2ϩ influx controlled by store-operated calcium channels.
InsP 3 -controlled Ca 2ϩ release from ER is highly synchronized with Ca 2ϩ influx. Such synchronization is essential for prevention of possible remodulation of InsP 3 -induced [Ca 2ϩ ] i spiking by Ca 2ϩ entry (39). In general, these interactive regulatory events are complex. Our experiments dealing with the phase resetting of InsP 3 -dependent Ca 2ϩ oscillations by membrane depolarization during the sustained phase revealed modulation of the frequency and amplitude of Ca 2ϩ spiking in gonadotrophs (25,56). Similarly, Ca 2ϩ entry through storeoperated calcium channels was shown to modulate the frequency of spiking in oocytes (57). Calcium entry through P 2 receptor channels can also modulate the frequency of InsP 3controlled Ca 2ϩ spiking even at high and presumably nonphysiological concentrations of ATP. Thus, the ER oscillator in gonadotrophs has a high, but not unlimited, capacity to accommodate Ca 2ϩ influx through both VSCC and P 2 receptor channels without disrupting the oscillatory nature of Ca 2ϩ signaling.
In conclusion, the present data indicate that, in addition to two subtypes of VSCC, pituitary gonadotrophs also express P 2 calcium channels. ATP gating of these channels leads to an increase in Ca 2ϩ influx and depolarization of the cells. Such ATP-induced depolarization is sufficient to initiate AP-driven [Ca 2ϩ ] i oscillations in silent cells and to increase the firing rate in cells that are spontaneously active at the resting membrane potential, as well as to stimulate LH release in the absence of GnRH. Furthermore, ATP is able to amplify GnRH-induced Ca 2ϩ signaling in single cells and LH secretion in perifused cells. The physiological significance of such an action of ATP is supported by the finding that ATP is cosecreted with gonadotropins during GnRH stimulation. To control the action of this positive feedback system, the ecto-ATPases in pituitary cells provide an efficient mechanism for the termination of signaling. Thus, pituitary gonadotrophs express a complete purinergic system for the self-potentiation of signaling and secretion. The loading of secretory granules with ATP and the expression of distinct subtypes of P 2 receptors in other pituitary cell types indicate that ATP may also have an important role in cell-tocell signaling within the pituitary gland.