Secretagogue-induced Exocytosis Recruits G Protein-gated K+ Channels to Plasma Membrane in Endocrine Cells*

Stimulation-regulated fusion of vesicles to the plasma membrane is an essential step for hormone secretion but may also serve for the recruitment of functional proteins to the plasma membrane. While studying the distribution of G protein-gated K+ (KG) channels in the anterior pituitary lobe, we found KG channel subunits Kir3.1 and Kir3.4 localized on the membranes of intracellular dense core vesicles that contained thyrotropin. Stimulation of these thyrotroph cells with thyrotropin-releasing hormone provoked fusion of vesicles to the plasma membrane, increased expression of Kir3.1 and Kir3.4 subunits in the plasma membrane, and markedly enhanced KG currents stimulated by dopamine and somatostatin. These data indicate a novel mechanism for the rapid insertion of functional ion channels into the plasma membrane, which could form a new type of negative feedback control loop for hormone secretion in the endocrine system.

The G protein-gated K ϩ (K G ) 1 channel, a member of the inwardly rectifying K ϩ (Kir) channel family, is directly activated by pertussis toxin-sensitive G proteins. This system was first discovered in the muscarinic deceleration of the heartbeat (1). More recently, it is considered to play an essential role in the hormone-mediated inhibitory regulation of neural excitability (1,2). Electrophysiological studies have revealed that a variety of inhibitory receptors, including M 2 -muscarinic, A 1purinergic, D 2 -dopamine, ␣ 2 -adrenergic, serotonin, ␥-aminobutyric acid type B, opioid, and somatostatin receptors in the brain, are coupled to K G channels (3). In endocrine organs such as the anterior pituitary lobe and pancreatic islet, it was also reported that some neurotransmitters including dopamine and somatostatin hyperpolarize the membrane by activating K G channels, which results in the inhibition of hormone secretion (4 -7).
The main subunit of K G channels has been cloned from the heart and designated GIRK1/Kir3.1 (8,9). In the brain and heart, Kir3.1 forms functional K G channels by assembling with other Kir3.0 subunits such as GIRK2/Kir3.2, GIRK3/Kir3.3 (10,11), and GIRK4/CIR/Kir3.4 (12). Because Kir3.1 mRNA was detected in the anterior pituitary lobe (13) and pancreatic islet (14), this subunit may also contribute to the formation of K G channels in endocrine cells. The anterior pituitary lobe contains several kinds of endocrine cells: lactotrophs, somatotrophs, corticotrophs, and thyrotrophs. Although electrophysiological experiments have shown that somatostatin or dopamine activate K G currents in lactotrophs (4) and also in several cell lines derived from the anterior pituitary lobe, such as GH3 (5) and AtT20 (6), the cellular and subcellular localizations of K G channels in in vivo pituitary endocrine cells have not been examined.
In this study, using a polyclonal antibody specific to Kir3.1, we found that this subunit was expressed only in thyrotroph cells and, surprisingly, was localized predominantly on intracellular secretory vesicles. Kir3.4 was co-localized on the vesicles with Kir3.1. Thyrotropin-releasing hormone (TRH) stimulation of thyrotrophs caused fusion of the vesicles to the cell membrane, increase of cell capacitance, and enhancement of dopamine-or somatostatin-induced K G current. These data indicate a novel mechanism for the rapid insertion of functional ion channels into the plasma membrane, which could form a new type of negative feedback control loop for thyrotrophs by tuning up their inhibitory regulatory signaling system in response to the stimulatory signal.

EXPERIMENTAL PROCEDURES
Preparation of Polyclonal Antibodies-The polyclonal antibody for Kir3.1 (aG1C-1) was raised in rabbit against a synthetic peptide corresponding to amino acid residues 488 -501 (LPAKLRKMNSDRFT) of Kir3.1/GIRK1 (15,16). We have also developed the antibody to Kir3.4 (aG4N-10) in rabbit using the antigenic peptide, DSRNAMNQD-MEIGV, which corresponds to amino acids 4 -17 of Kir3.4 (17). These antibodies have been successfully used for immunoprecipitation and Western blotting analyses of Kir3.1 protein (15,16) and Kir3.4 protein 2 in the brain and heart. Furthermore, in the rat heart, the immunoreactivities were detected with the antibodies only in the atrium but not in the ventricle (data not shown). This finding coincides with the expression of the cardiac K G channel, which is composed of the subunits Kir3.1 and Kir3.4 in the heart atrium (13).
Immunohistochemical Study-Immunohistochemistry was performed according to a method described elsewhere (15). Rats were anesthetized with sodium pentobarbital and perfused transcardially with 4% (w/v) paraformaldehyde, 0.5% (w/v) glutaraldehyde, and 0.2% (w/v) picric acid, 0.1 M phosphate buffer (pH 7.4). The anterior pituitary lobe was removed, * This work was supported by grants from the Ministry of Education, Science, Sports and Culture of Japan (to Y. K.), the "Research for the Future" Program of The Japan Society for the Promotion of Science, and the Human Frontier Science Program. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
postfixed in a fixative containing 4% paraformaldehyde and 0.2% picric acid for 48 h at 4°C, and transferred to 0.1 M phosphate buffer containing 15% (w/v) sucrose and 0.1% (w/v) sodium azide at 4°C. Sections were cut with a cryostat at 20 m of thickness and stored at 4°C in the same solution until used. After rinsing, the sections were placed in free-floating states into aG1C-1 that had been diluted (1:10,000) in phosphate-buffered saline containing 0.3% (v/v) Triton X-100 and 1% (w/v) bovine serum albumin and incubated for 3 days at 4°C. After incubation with biotinylated goat anti-rabbit IgG, staining was accomplished using the avidinbiotin complex method (Vectastain Elite kit, Vector Laboratories, Burlingame, CA), with nickel-diaminobenzidine as the chromogen. Control procedures consisted of preabsorbing aG1C-1 (1:10,000 dilution) by incubating with a saturating concentration of the antigenic peptide. For double immunofluorescent staining, sections were first incubated with aG1C-1 and a monoclonal antibody for each pituitary hormone, i.e. anti-prolactin monoclonal antibody (QED Bioscience Inc., San Diego, CA), anti-growth hormone monoclonal antibody (Quartett, Berlin, Germany), anti-luteinizing hormone monoclonal antibody (Quartett), anti-adrenocorticotropic hormone (ACTH) monoclonal antibody (YLEM, Rome, Italy), or anti-thyrotropin (TSH) monoclonal antibody (Quartett). The sections were washed thoroughly and incubated with fluorescein isothiocyanate-labeled anti-rabbit IgG (E. Y. Laboratories, San Mateo, CA) and Texas Red-labeled anti-mouse IgG (Protos Immu-noResearch, San Francisco, CA). These sections were examined with a confocal microscope (MRC-1024, Bio-Rad Laboratories, Hertfordshire, U. K.).
Electron Microscopic Study-For electron microscopy, sections were processed according to the method described elsewhere (18). TRH stimulation involved the injection of 10 g of TRH into the femoral vein 10 min before the perfusion, which was enough to stimulate the thyrotroph in vivo (19). The vibratome sections were dehydrated in a graded series of ethanol and incubated in a mixture of ethanol and LR Gold resin (London Resin, Berkshire, U. K.) followed by flat embedment in fresh LR Gold containing 0.1% (v/v) benzyl on silicon-coated glass slides. The embedded sections were polymerized for 6 h in an ultraviolet cryochamber (Pelco, Ted Pella Inc., Redding, CA) at Ϫ20°C. Small blocks containing anterior pituitary lobe were cut out and glued to black resin. Ultrathin sections were cut using an ultramicrotome (Ultracut UCT, Leika, Vienna, Austria) and mounted on collodion and carbon-coated nickel grids (thin bar grid, Nishin EM Co. Ltd., Tokyo, Japan). The nickel grids with ultrathin sections were incubated with 3% (v/v) normal goat serum in 0.1 M phosphate-buffered saline containing 0.2% bovine serum albumin, 0.2% (w/v) saponin, and 0.05% (w/v) NH 4 Cl for 30 min at room temperature. The grids were then incubated with aG1C-1 (1:100) or aG4N-10 (1:50) in the same buffer for 2 h and then with a 15-nm immunogold conjugated goat anti-rabbit IgG (British BioCell International Inc., Cardiff, U. K., diluted 1:40) for 1.5 h at room temperature. In the case of double immunolabeling, the grids were incubated with aG1C-1 (1:100) or aG4N-10 (1:50), and anti-TSH monoclonal antibody (1:10) followed by incubation with immunogold-conjugated anti-rabbit IgG (15-nm gold particles) and immunogold-conjugated anti-mouse IgG (10 nm gold particles). The grids were washed with 0.1 M phosphate-buffered saline and then with distilled water before staining with uranyl acetate and Reynold's lead citrate. These sections were examined with an electron microscope (H7100TE, Hitachi, Hitachinaka, Japan).
Whole Cell Patch-Clamp Recordings-Anterior pituitary glands were dissected from adult female Wistar rats. Cells were dispersed enzymatically using a non-trypsin dissociation protocol modified from that used by Einhorn and Oxford (4). In brief, the glands were removed and minced in sterile Hanks' balanced salt solution that was free of calcium and magnesium (Hanks' CMF). After washing, the fragments were incubated in a shaking water bath for 1 h in 37°C in Hanks' CMF containing trypsin inhibitor (Sigma, 0.1 mg/100 ml), collagenase (Worthington, 0.3%), and DNase I (Sigma, 1.0 mg/100 ml). Fragments were then mechanically dispersed by trituration with a siliconized Pasteur pipette, washed with Dulbecco's modified Eagle's medium, filtered through nylon mesh (20 m), and harvested via centrifugation. Cells were incubated in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Whole cell recordings were made utilizing the gigaohm seal patch-clamp technique (20). Recordings were made using a patch-clamp amplifier (EPC-7, List Electronics, Darmstadt, Germany) and recorded on videocassette tapes with a PCM converter system (RP-880, NF Electronic Circuit Design, Yokohama, Japan), reproduced and low pass-filtered at 1 kHz (Ϫ3 db) by a filter with Bessel characteristics (48 db/octave slope attenuation) for an analysis and illustration. Experiments were performed at 22-24°C in a bath solution composed (in mM) of the following: NaCl (91), KCl (50), MgCl 2 (1), glucose (5.5), and Hepes buffer (5), pH 7.2. The composition of the pipette filling solution was as follows (in mM): KCl (140), MgCl 2 (8), EGTA (0.1), CaCl 2 (0.0415), GTP (0.1), and Hepes (5), pH 7.2.
Single-cell RT-PCR Analysis-Cytoplasmic RNAs of a dissociated single pituitary cell, which was aspirated and transferred into a microcentrifuge tube with a glass tip used for patch-clamp experiments, were extracted, and cDNAs were synthesized as described elsewhere (21). To eliminate the contamination of genomic DNAs, we treated RNA samples with DNase I (Takara) followed by cDNA synthesis with reverse transcriptase (Life Technologies, Inc.). To ascertain that cytoplasmic RNAs were correctly aspirated, samples were conducted to RT-PCR of ␤-actin by the ␤-actin PCR kit (CLONTECH). We could detect the ␤-actin transcripts in all of the Type I cells, 16 .4), or 56°C (TSH␤) for 1 min, and at 72°C for 1 min and then at 72°C for 8 min. Because the products from the single pituitary cell were insufficient for visualization, second cycles of reactions using the second PCR primers and the same amplification conditions were performed with a part of the first PCR products. The nucleotide sequence of the PCR products was confirmed with the dye primer method and DNA sequencer (A-381, Perkin-Elmer) after TA cloning (Invitrogen).
Cell Capacitance Measurement-Simultaneous measurements of changes in membrane capacitance (C m ), conductance (G m ), and membrane current (I m ) were carried out using a patch-clamp amplifier (EPC-7, List Electronics) and a two-phase lock-in amplifier (NF5610B, NF Electronic Circuit Design) as described previously (24). The capacitance transients originating from the capacitance of the plasma membranes were minimized by adjusting the capacitance and the time constant of the capacitance cancellation circuit in the patch-clamp amplifier while applying voltage pulses with an amplitude of 10 mV. This adjustment gave whole cell capacitance values of 9.0 -13.4 picofarads and series resistance values of 5-9 megaohms. After the cancellation, a 600-Hz, 3.4-mV peak to peak sine-wave voltage was superimposed on the holding potential of Ϫ100 mV. The resulting current output was fed into the lock-in amplifier. The phase offset of the lock-in amplifier was adjusted so that when the capacitance of the cancellation circuit was modified to calibrate C m , there was no change in the output for G m (closed circles at the C m trace in Fig. 3A).

RESULTS AND DISCUSSION
K G channels are heteromultimeric proteins composed of the subunits Kir3.1 ϩ Kir3.4 in the heart (17) and Kir3.1 ϩ Kir3.2 in the brain (25,26). The Kir3.1 subunit may therefore be common to the K G channels of various tissues (8,9). We examined the distribution of Kir3.1 immunoreactivity in sections of rat pituitary using a specific polyclonal antibody (aG1C-1) (15, 16) (Fig.  1). The Kir3.1-positive cells were scattered in the anterior pituitary lobe and made up ϳ3% of the anterior pituitary cells (Fig.  1A). Under the light microscope Kir3.1 immunoreactivity was detected diffusely in the cytosol of each positive cell (Fig. 1, A and  B). The Kir3.1-positive cells had a polygonal or stellate shape between 20 and 30 m in size. To identify which types of pituitary cells express Kir3.1, sections were double-stained for Kir3.1 and various pituitary hormones. Fig. 1, B and C, shows that Kir3.1 immunoreactivity (green) was detected only in the cells that were stained with anti-TSH antibody (red). In 176 TSHpositive cells (i.e. thyrotrophs), 149 cells (ϳ85%) were also Kir3.1positive. No Kir3.1 immunoreactivity was detected in cells stained with anti-prolactin, anti-growth hormone, anti-luteinizing hormone, or anti-ACTH antibodies. Prominent orange-yellow signals generated by the double staining of green (Kir3.1) and red (TSH) suggest that they may be localized in close proximity in thyrotrophs (Fig. 1D).
The subcellular localization of Kir3.1 was examined using immunoelectron microscopy (Fig. 1, E, F, and H). The immunoreactive signals (gold particles) were detected predominantly on secretory vesicles and rarely on the plasma membrane of Kir3.1-positive pituitary cells (Fig. 1E). The cells were ϳ30 m in diameter and polygonal in shape. The reactive intracellular secretory vesicles were ϳ100 nm in diameter. These morphological features of Kir3.1-positive cells are those of the thyrotroph (27). Furthermore, with double immunolabeling under the electron microscope, we found that Kir3.1 (detected by 15-nm gold particles) and TSH (10-nm gold particles) were localized in the same secretory vesicles (Fig. 1E, inset). In contrast, the adjacent cell, which contained large 200 -300-nmdiameter electron dense vesicles, did not show any immunoreactivity to aG1C-1. Fig. 1, F and H, depicts examples of Kir3.1 immunoreactivity in the thyrotrophs obtained from TRH-administered rats. TRH stimulation caused an increase in the detection of gold particles on the plasma membrane (Fig. 1F, arrows). The number of gold particles detected on the plasma membrane in the control was 0.25 Ϯ 0.25% (mean Ϯ S.E., n ϭ 15 cells) of the total number of particles that existed within 1.5 m of the plasma membrane. This number increased to 7.56 Ϯ 1.80% (n ϭ 15 cells) after TRH stimulation (Fig. 1G). Furthermore, as indicated by the arrowheads in Fig. 1H, after TRH stimulation various stages of fusion of the gold particle-positive vesicles with the plasma membrane were detected. These results suggest that the K G channels on the secretory vesicles of thyrotrophs could be translocated to the plasma membrane during TRH-induced exocytosis.
It is known that dopamine-D 2 and somatostatin receptors can couple to K G channels via pertussis toxin-sensitive G proteins in pituitary cells (2). Therefore, TRH-induced recruitment of the K G channels to the plasma membrane of thyrotrophs may in turn facilitate the effect of inhibitory transmitters such as dopamine and somatostatin. To examine this possibility, electrophysiological techniques were applied to dissociated pituitary cells (Fig. 2). From 82 cells examined, we observed three distinct types of cells in terms of their responses to bromocriptine, somatostatin, and TRH (Types I-III). Fig. 2A depicts a representative cell current record of the response of Type I cells (n ϭ 6). The dopamine receptor agonist bromocriptine induced a small inwardly rectifying K ϩ (Kir) current, which was markedly enhanced upon the addition of TRH to the bath. The enhanced bromocriptine-induced Kir current was inhibited by sulpiride, an antagonist for the dopamine receptor. After ϳ10 min of wash-out of bromocriptine and sulpiride, somatostatin induced a small Kir current similar to that induced by bromocriptine under control conditions. The somatostatin-induced Kir current was also enhanced by the addition of TRH. In the six Type I cells, TRH increased the bromocriptine-induced K G current by 6.0 Ϯ 0.5-fold (mean Ϯ S.E.) and the somatostatin-induced K G current by 7.0 Ϯ 0.8-fold, measured at the command pulse to Ϫ100 mV. TRH alone, in the absence of bromocriptine or somatostatin, did not induce any appreciable Kir current (data not shown). Both bromocriptineand somatostatin-induced currents reversed at approximately Ϫ30 mV close to the equilibrium potential for K ϩ with 50 mM extracellular K ϩ and exhibited clear inward rectification (Fig.  2A, lower right panel). The currents activated slowly during hyperpolarizing command voltage pulses, which is a feature of K G channels containing Kir3.1 subunits (Fig. 2A, lower left panels) (17,25). In RT-PCR analysis of cytoplasmic RNA aspirated from the cell presented in Fig. 2A, we detected transcripts of Kir3.1 and TSH␤ as shown in lane 1 of Fig. 2D. The same results were obtained in the remaining five Type I cells (including the result shown in lane 4 of Fig. 2D). Thus, Type I cells are thyrotrophs that may possess K G channels containing the Kir3.1 subunit.
Because it is known that homomeric K G channels composed of Kir3.1 are not functional (17) but that heteromers of Kir3.1 and another Kir3.0 subunit form functional K G channels, we have tried to identify the Kir3.0 subunit that may assemble with Kir3.1 in the vesicular K G channels. Kir3.3 was not detected by RT-PCR in the mRNA from whole anterior pituitary lobe (data not shown). Therefore, the expression of Kir3.2 and Kir3.4 subunits was examined by single-cell RT-PCR. All six Type I cells expressed a Kir3.4 transcript, and one of them also expressed Kir3.2. Furthermore, we detected Kir3.4 immunoreactivity (15-nm gold particles) on the vesicles containing TSH (10-nm particles) with electron microscopic immunocytochemistry (Fig. 2E). Kir3.2 immunoreactivity was not detected on the vesicles (data not shown). These results suggest that Kir3.4 may be the subunit that makes up the vesicular K G channel with Kir3.1.
In Type II cells (n ϭ 21), bromocriptine or somatostatin induced a very small Kir current, which was not affected by TRH (Fig. 2B). Lanes 2 and 5 in Fig. 2D 6-4). B, in a Type II cell, perfusion of bromocriptine induced a small inward current, which was not affected by TRH. Lower panels of B and C, superimposed whole cell currents evoked by bromocriptine before (trace 2-1) and after TRH stimulation (trace 3-1) with the same voltage steps as in A. C, in most of the dissociated pituitary cells (Type III cells), dopamine and TRH did not affect the membrane current. D, representative results of single-cell RT-PCR. The cytosols were sucked through the patch pipettes after current recording, and then RT-PCR was done. All of the Type I cells (n ϭ 6) expressed the Kir3.1 and TSH␤ transcripts (lanes 1 and 4). Lane 1 exhibits the transcripts obtained from the cell whose current record is shown in A. Type II (lanes 2 and 5) and Type III (lanes 3 and 6) cells did not express the Kir3.1 transcripts, although a few cells expressed the TSH␤ mRNAs (lane 2). Lanes 5 and 3 illustrate transcripts obtained from the cells whose current recordings are shown in B and C, respectively. Lane P, PCR product of whole pituitary cDNA; lane C, PCR product of each cDNA. E, Kir3.4 immunoreactivity (detected by 15-nm gold particles) was also located on the TSH (10-nm particles)-containing vesicles. Scale bar ϭ 0.25 m.
response cells (n ϭ 55), bromocriptine or somatostatin did not induce any Kir current, and TRH showed no appreciable effect (Fig. 2C). We could not detect any transcripts of Kir3.1, Kir3.2, or TSH␤ in 10 Type III cells examined (lane 3 in Fig. 2D is from the cell shown in Fig. 2C). Two of the 10 Type III cells expressed Kir3.4 mRNA. In conclusion, TRH enhances bromocriptine or somatostatin activation of K G current specifically in thyrotrophs, which express Kir3.1 mRNA.
To examine whether the TRH-induced augmentation of K G channel activity in Type I thyrotrophs is derived from the exocytosis of secretory vesicles, as was suggested by the electron microscopic examination (Fig. 1, E-H), we simultaneously measured C m , G m , and I m in Type I cells (Fig. 3A). The cells were bathed in 5 mM K ϩ bathing solution and held at Ϫ100 mV. A 600-Hz sine wave with a peak to peak amplitude of 3.4 mV was applied to the cell under voltage clamp. The application of bromocriptine caused some increase of G m and of the small inward K ϩ current without any significant effect on C m . When TRH was added in the continued presence of bromocriptine, the inward K ϩ current was markedly enhanced in parallel with the simultaneous increase of both C m and G m . The increase of C m may reflect the fusion of secretory vesicles to the plasma membrane induced by TRH (28). In the continued presence of TRH, C m , G m , and I m fluctuated in parallel and slowly returned toward their basal levels. The decrease of C m may represent endocytosis of membrane containing K G channels. This TRHinduced response was observed in 5 of 45 cells examined. In these 5 cells, expression of mRNAs of Kir3.1, Kir3.4, and TSH␤ was confirmed by retrospective single-cell RT-PCR (data not shown). The remaining 40 cells did not express any Kir3.1transcript. The simultaneous measurement of C m , G m , and I m indicates that TRH-induced augmentation of K G channel activity in Type I cells is associated with the exocytosis of secretory vesicles.
TRH-induced TSH secretion may be composed of voltage-dependent and voltage-independent components, as is the case for prolactin secretion by TRH in the lactotroph (29). The increase in C m measured in Fig. 3A must represent a voltageindependent component because these experiments were conducted under voltage clamp. The TRH-induced voltage-independent fusion of secretory vesicles causes an increase in the number of K G channels on the plasma membrane. The increased K G channels should then facilitate the hyperpolarization induced by dopamine and somatostatin, which may suppress the depolarization-induced secretion. TRH-induced recruitment of vesicular K G channels to the plasma membrane may act as a feedback control mechanism to avoid excessive stimulation of the cell (Fig. 3B).
At the crustacean neuromuscular junction, repetitive stimulation of motor nerves resulted in an increase of the number of active zones containing a voltage-dependent Ca 2ϩ channel at the axonal terminus, which was implicated in the long term facilitation of neurotransmitter release (30). The increase of active zones is thought to be the result of the translocation to the plasma membrane of a voltage-dependent Ca 2ϩ channel localized on secretory vesicles (31). Similarly, it was shown that the number of postsynaptic ␥-aminobutyric acid type A receptors was rapidly increased by insulin (32) or in an experimental model of temporal lobe epilepsy (33). The stimulation-induced increase of ␥-aminobutyric acid type A receptors may underlie the long term modification of synaptic transmission in hippocampal inhibitory synapses (32,33). These previous studies require a mechanism that enables stimulation to translocate ion channels from an intracellular pool to the plasma membrane. This study provides, for the first time, clear evidence to indicate that the localization of ion channels on secretory vesicles actually plays a role in the physiological regulation of membrane excitability. Further studies are needed to elucidate the molecular mechanisms localizing ion channels on secretory vesicles. This line of study may provide novel insights into the stimulation control of cellular excitability, including long term modification of synaptic transmission.