Reconstitution of acid secretion in digitonin-permeabilized rabbit gastric glands. Identification of cytosolic regulatory factors.

When isolated rabbit gastric glands were permeabilized with digitonin, they lost their ability to secrete acid, as monitored by [14C]aminopyrine accumulation, and they never recovered by supplement with cytosol prepared from gastric mucosa. However, the permeabilized glands elicited acid secretion when brain cytosol was supplemented. Fractionation of gastric cytosol by gel filtration revealed that the fraction at 30 kDa stimulated permeabilized glands by itself, whereas the 200-kDa fraction potently inhibited brain cytosol-stimulated acid secretion. Brain cytosol contained only the former stimulatory factor. With further gel filtration, the 30-kDa activator was separated into two components, 20 kDa (peak 1) and 1.8 kDa (peak 2), both of which are necessary for full activity. We purified peak 1 from bovine brain, and phosphatidylinositol transfer protein (PITP) was identified as the main component of the activity. The stimulating activity in brain and gastric mucosa correlated with the contents of PITP, and recombinant PITP mimicked the effect of peak 1, suggesting that PITP is one of the essential components in gastric acid secretion. When gastric glands were stimulated, the inhibitory activity, but not stimulatory activity, in the cytosol was increased. This suggests a regulatory mechanism such as stimulation translocates the inhibitory component from the secretory site on the membrane to cytosol. These results demonstrate a high degree of usefulness for our present model, the reconstituted digitonin-permeabilized gastric glands.

Although there have been a few candidates reported as putative targets for PKA, e.g. ezrin (1,4) and a chloride channel, ClC2G (5), it is still difficult to lay out the whole scheme of the activation process.
One advantage of another cell system, chromaffin cell, is that this cell preserves its secretory response to Ca 2ϩ even after the cell becomes permeable to large molecules (such as proteins) from treatment with digitonin. In contrast, the digitonin-permeabilized gastric gland entirely loses its ability to secrete acid (6), and this fact has restricted the kind of experimental design that uses protein probes for analyzing intracellular signal transduction. It is reasonable to suppose that cytosolic protein(s), essential for the activation of acid secretion, would leak out through the pores formed by digitonin. This follows after other permeabilized gland models with ␣-toxin (7,8) and ␤-escin (3), where much smaller pores in the plasma membrane are able to be stimulated by cAMP. From another point of view, it would be possible to identify these putative essential factors by searching for the activity that reconstitutes the responsiveness of digitonin-permeabilized glands. In fact, using permeabilized chromaffin cells that deplete cytosolic factors, candidates for regulatory proteins involved in the ATP-dependent process (9,10) or the Ca 2ϩ -dependent process (11,12) have been successfully identified.
In the present study, we report that the secretory responsiveness of digitonin-permeabilized rabbit gastric glands can be reconstituted by cytosolic factors and that this model is a powerful tool for searching for the cytosolic factors involved in the activation of acid secretion.

EXPERIMENTAL PROCEDURES
Materials-Rat cloned ␤ isoform of phosphatidylinositol transfer protein (PITP) (13) in pET-21a-d(ϩ) vector was kindly gifted by Dr. H. Arai (Graduate School of Pharmaceutical Sciences, University of Tokyo, Tokyo, Japan). The cDNA was cloned into pGEX-4T-1, and transformed XL1-blue. Expression of the ␤-PITP glutathione S-transferase (GST) fusion protein was induced by isopropyl-␤-thiogalactopyranoside (0.1 mM) for 2 h at 25°C, and the bacterial cells were collected and then resuspended in 150 mM NaCl, 3 mM K 2 HPO 4 , 0.64 mM NaH 2 PO 4 , 10 mM EDTA (pH ϭ 7.0). After freeze thawing, the sample was centrifuged at 40,000 ϫ g for 30 min at 4°C. Recombinant protein was purified from the supernatant using glutathione-Sepharose 4B resin, and then the fusion protein was cleaved by thrombin treatment.
All the chemicals were reagent grade and obtained from Sigma or Nacalai Tesque except otherwise noted.
Co., Tokyo, Japan) essentially by the method of Berglindh (14). Isolated glands, suspended in the normal medium containing 132.6 mM NaCl, 5 mM Na 2 HPO 4 , 1 mM NaH 2 PO 4 , 5.4 mM KCl, 1.2 mM MgSO 4 , 1.0 mM CaCl 2 , 25 mM HEPES-Na (pH ϭ 7.4), 11.1 mM glucose, and 1 mg/ml bovine serum albumin, were washed three times with a permeabilizing medium containing 100 mM NaCl, 20 mM KCl, 1 mM MgSO 4 , 0.5 mM EGTA, 1 mM ATP, 10 mM sodium pyruvate, 10 M cimetidine, 10 mM Tris, 20 mM Hepes, pH 7.3. Before the last wash, an aliquot was taken to measure the wet weight of the glands per milliliter of suspension, and the final gland preparation was adjusted to 20 -25 mg wet weight/ml with the permeabilizing medium. Permeabilization of the glands was performed with 50 -100 g/ml digitonin (depending upon the lot) at 37°C for 20 min. The glands were then suspended in a high K ϩ medium containing 20 mM NaCl, 100 mM KCl, 1.0 mM MgSO 4 , 0.5 mM EGTA, 2 mM ATP, 10 mM sodium pyruvate, 10 M cimetidine, and 20 mM HEPES (pH ϭ 7.4). The free Ca 2ϩ concentration in the high K ϩ medium was calculated to be as high as 90 nM using the computer program, Chelator ® , assuming that the contaminated total Ca 2ϩ in the medium was as high as 10 M.
To assure the reproducibility of the responsiveness of the permeabilized glands, the stock solution of digitonin was prepared as follows. Digitonin (Merck) was completely dissolved in boiling water at 100 mg/ml and kept at 4°C for 1 or 2 days. After the precipitates formed were removed by decantation, the supernatant was filtered through a 0.22-m membrane. The optimum concentration of this 1000ϫ stock solution (usually between 50 and 100 g/ml final, ignoring loss from precipitation) was determined for each lot.
The extent of the permeabilization was monitored by the leakage of lactate dehydrogenase, a 145-kDa cytosolic protein. Using the assay kit (Wako, Japan), the loss of the enzyme by the treatment was estimated to be 60 -70% of the total.
Acid secretion of the glands was monitored by accumulation of a weak base, [ 14 C]aminopyrine (14). One ml (0.5 ml for rough screening purpose) of suspension of the glands in the high K ϩ medium was incubated at 37°C for 30 min using a duplicated 1.5-ml Eppendorf tube.
In some experiments, isolated gastric cells were obtained from isolated glands by Pronase digestion (0.4 mg/ml; at 37°C for 15 min), and the isolated cells were separated by Percoll (Amersham Pharmacia Biotech) density gradient based on the procedure described (15). The fraction of the cells floating on 45% Percoll was designated as parietal cell-rich fraction, and that which pelleted on the bottom was termed as parietal cell-poor (mostly chief cells) fraction. Using monoclonal antibody against H,K-ATPase (16), the purity of parietal cell in the former fraction was estimated to be about 80%, whereas that in the latter was about 20%.
Preparation of Cytosol and Its Fractionation-Various tissues were homogenized with 3 volumes of ice-cold homogenizing buffer (125 mM mannitol, 40 mM sucrose, 1 mM EDTA, 5 mM PIPES, pH 7.3) including protease inhibitor mixture (100 M phenylmethylsulfonyl fluoride, 1.5 M pepstatin A, and 1 M leupeptin; added immediately before homogenization as 1000ϫ methanol solution) through a Teflon piston homogenizer (Potter-Elveheim). The homogenates were centrifuged at 12,000 ϫ g for 10 min, and the supernatants were further centrifuged at 100,000 ϫ g for 60 min. The supernatants were dialyzed against a high K ϩ medium containing 100 mM KCl, 20 mM NaCl, 10 mM Tris, 20 mM HEPES, pH 7.3, at 4°C for 3 h, and the aliquots were stored at Ϫ80°C until use. In the case of purified cells, they were transferred to a glass tube, suspended in a high K ϩ medium containing 20 mM NaCl, 100 mM KCl, 1.0 mM MgSO 4 , 0.5 mM EGTA, 2 mM ATP, 10 mM sodium pyruvate, 10 mM cimetidine, protease inhibitor mixture, and 20 mM HEPES (pH ϭ 7.4), and then sonicated using a bath sonicator (Kokusai Electric, 3 ϫ 30 s). The homogenate was centrifuged at 100,000 ϫ g for 30 min, and their activity was assayed on digitonin-permeabilized glands. In order to compare the inhibitory and stimulatory factors in the isolated glands, 4 ml of settled glands were separated into two aliquots; one was maximally stimulated with 100 M histamine plus 50 M isobutylmethylxanthine at 37°C for 30 min in 40 ml of medium, whereas the other was incubated with 100 M cimetidine to keep it in the resting state. The glands were then homogenized through a tight Teflon piston homogenizer with 1 ml of the high K ϩ medium. The 100,000 ϫ g supernatant, mixed with blue dextran 2000 (Amersham Pharmacia Biotech) and phenol red as visual markers, was loaded on a 30 cm ϫ 0.6-cm 2 Sepharose CL4B column (Amersham Pharmacia Biotech), equilibrated with the high K ϩ medium, and the eluate was collected every 1.5 ml between blue dextran 2000 and phenol red. The fractions corresponding to 30 and 200 kDa, determined by standards in advance, were assayed for stimulatory and inhibitory activity, respectively, using digitonin-permeabilized glands.
To separate partially purified material, cytosolic fraction from rabbit gastric mucosa or brain was loaded on a 75 cm ϫ 2-cm 2 Sepharose CL4B column (Amersham Pharmacia Biotech), which had been equilibrated with the high K ϩ medium. Elution was performed at 0.4 ml/min, and the fractions were collected every 15 min.
In order to fractionate smaller molecules, the cytosolic fraction without dialysis was concentrated to 50 -60 mg of protein/ml by lyophilization and loaded on a 75 cm ϫ 2-cm 2 Superdex 200-pg column (Amersham Pharmacia Biotech), which had been equilibrated with the high K ϩ medium. Elution was performed at 0.4 ml/min, and the fractions were collected every 15 min. As described under "Results," the activity of brain cytosol was separated by this procedure into two peaks, i.e. peak 1, subsequently designated as p35, and peak 2, with an apparent molecular mass of 1.8 kDa. Since activity was only evident when both fractions were present, the bioassay procedure beyond this point was performed as follows.
Cytosol was prepared from rabbit whole brain, concentrated, and loaded on the Superdex 200-pg as described above, and the fractions corresponding to ϳ1.8 kDa (20 -25 ml) were collected and designated as peak 2. For the aminopyrine accumulation assay, 50 l of peak 2 was included such that the final composition of the medium was equivalent to that of the high K ϩ medium.
Purification of p35 (Peak 2) from Bovine Brain-Fresh bovine brain was obtained from the local slaughterhouse. After removing the white matter, the brain was homogenized with 2 volumes of the homogenization buffer using a Waring blender (three bursts of 30 s at 10-s intervals). The homogenate was centrifuged with a Beckman J-1 rotor at 10,000 rpm for 1 h, and the supernatant was collected. The pellet was re-homogenized with 0.6 volumes of the buffer, and its supernatant was combined with the first supernatant and centrifuged at 100,000 ϫ g for 1 h to harvest the supernatant.
The cytosolic fraction was loaded on a DEAE -Sepharose column (Amersham Pharmacia Biotech) containing 100 ml of resin, which had been equilibrated with a buffer containing 10 mM Tris, 20 mM HEPES, pH 7.3. The column was washed with the same buffer until the absorbance at 280 nm decreased, and the elution was performed with a low pH buffer containing 40 mM KCl, 10 mM Tris, 20 mM HEPES, pH 6.7. The eluted material, about 200 ml, was loaded on a 20-ml red-agarose column (Sigma) and eluted with a buffer containing 1 M KCl, 0.2 M NaCl, 10 mM Tris, 20 mM HEPES, pH 7.3. This fraction was concentrated by using Centriprep 3 (Amicon) and loaded on a 75 cm ϫ 2-cm 2 Superdex 200-pg column, which had been equilibrated with the high K ϩ medium. Elution was performed at 0.4 ml/min, and the fractions were collected every 15 min. Fractions containing stimulatory activity, approximately 20 kDa, were collected and diluted with water to adjust the KCl concentration to 20 mM, and then loaded on a hydroxyapatite column (Sigma). Elution was performed with a linear gradient of sodium phosphate (0 -400 mM in a buffer containing 10 mM Tris, 20 mM HEPES, pH 7.3), and the active fractions (40 -70 mM) were collected and concentrated by using Centriprep 3. This material was finally purified with an HPLC gel filtration column, Diol-150 (Shimadzu, Japan) equilibrated with the high K ϩ buffer.
Preparation of Anti-p35 Polyclonal Antibody-About 10 g of p35 purified from bovine brain was suspended in TiterMAX Gold (CytRx Corp.) and subcutaneously injected into the rats. After 2 weeks, 10 g of p35 was intravenously injected as a booster. The IgG-rich fraction was purified from the serum with DEAE-Sepharose.
Peptide Sequencing of p35-Purified fraction containing p35 was separated on 15% SDS-polyacrylamide gel electrophoresis (PAGE), blotted on a nitrocellulose membrane by a semidry blotting apparatus, and visualized with Ponceau S, and the band of p35 was cut out. The p35 on the membrane was then digested by V8 protease (1:50), and the material was separated with HPLC reverse-phase chromatography using an acetonitrile gradient (0 -80%) in the presence of 0.1% trifluoroacetic acid. The peptide fragments obtained were sequenced using a peptide sequencer (Applied Biosystems 471A).

Brain Cytosol Stimulates Acid Secretion in Digitonin-perme-
abilized Glands-Acid secretion in digitonin-permeabilized rabbit gastric glands was monitored by the accumulation of [ 14 C]aminopyrine and is summarized in Fig. 1A. In contrast to the model using ␣-toxin (7,8) or ␤-escin (3), digitonin-permeabilized rabbit glands were not stimulated by 100 M cAMP, and this is consistent with previously reported results (6). To make up for the possible loss of PKA (the target of cAMP), from the cytosol, 1000 units/ml PKA catalytic subunit was added to the glands, but no secretory response was observed. The responsiveness of digitonin-permeabilized glands to 100 M cAMP never recovered with the addition of the supplement of cytosolic fraction as prepared from rabbit gastric mucosa. We then searched for other tissues as a source of cytosol and found that brain cytosol showed potent stimulatory activity in digitonin-permeabilized glands. We also prepared cytosol from skeletal muscle and pancreas, but the stimulating effect was specific for brain among the tissues tested. To understand why gastric cytosol was ineffective, we used gastric cytosol in addition to the brain cytosol and found that the stimulating activity of brain cytosol was completely abolished. This suggested that gastric mucosal cytosol contains inhibitory factors in addition to the stimulating factors, if any, which make the inhibitory activity predominant. In order to examine the possibility that the ineffectiveness of gastric cytosol observed here was due to contamination of cells other than parietal cells in the mucosal preparation, we prepared isolated gastric cells and fractionated them into parietal cell-rich (about 80%) and poor (about 20%; mainly chief cells) fractions and obtained cytosol from each. In this case, the cytosol (0.8 mg/ml) was added to 1 mg/ml brain cytosol, which showed about half the maximal stimulation (see Fig. 2), in order to pick up either stimulatory or inhibitory activity. As shown in Fig. 1B, the cytosol from highly purified parietal cell still showed inhibitory activity on brain cytosolstimulated aminopyrine accumulation; furthermore, its activity was more potent than that from the parietal cell-poor fraction. This observation also excluded the possible involvement of pepsin activity remaining, if any, in the presence of protease inhibitors. Fig. 2A shows the concentration-stimulation relationships of brain cytosol in the presence or absence of cAMP. Brain cytosol itself dose-dependently stimulated aminopyrine accumulation in permeabilized glands. Addition of 100 M cAMP further stimulated acid formation in the presence of submaximal concentration of cytosol. However, it never augmented the maximal effect of brain cytosol.
As shown in Fig. 2B, the stimulating effect of brain cytosol was not inhibited by 30 M heat-stable protein kinase inhibitor peptide, a specific PKA inhibitor, but was inhibited by 30 M omeprazole, a proton pump inhibitor, suggesting that the putative activator in the brain cytosol does not utilize the cAMPdependent pathway. Furthermore, cytosol-stimulated acid secretion was not augmented, but rather inhibited, by the inclusion of 100 M Ca 2ϩ . Considering the possible loss of calmodulin from the cell, we added 0.1-100 M calmodulin in addition to the brain cytosol, but no recovery of stimulation was In the case of the catalytic subunit of PKA (1000 units/ml), the experiment with cAMP was omitted. B, isolated gastric cells were purified into parietal cell-rich (ϳ80%) and poor (ϳ20%, mainly chief cells) fractions by Percoll gradient. Digitonin-permeabilized glands were incubated without (R) or with 1 mg/ml brain cytosol (control) to measure the aminopyrine ratio. Cytosol obtained from either parietal cell-rich or poor fraction (0.8 mg/ml) was added to brain cytosol. The aminopyrine ratio was measured in duplicate and means Ϯ S.E. of three to five independent experiments are shown (except skeletal muscle and pancreas, which were both two experiments apiece). observed (data not shown). The stimulatory effect of brain cytosol was also inhibited by 100 M GTP␥S, a non-hydrolyzable analogue of GTP, and by 50 M SM-1, a pseudo substrate of myosin light chain kinase.
Cytosol of Gastric Mucosa Contains Both Stimulatory and Inhibitory Factors-In the previous section, it was suggested that gastric mucosal cytosol contains inhibitory factors, with the essential factors for the activation of acid secretion, if any, which result in inhibitory activity with respect to brain cytosol. In order to examine this hypothesis, we fractionated gastric cytosol by gel filtration.
The cytosolic fraction was separated by Sepharose CL4B column chromatography, which works well for separation of components with relatively high molecular weights. Each fraction was assayed for stimulatory activity by itself as well as for inhibitory activity by using brain cytosol-stimulated aminopyrine accumulation in digitonin-permeabilized gastric glands. As shown in Fig. 3A, gastric cytosol obviously contained the activity to stimulate digitonin-permeabilized glands in the relatively small molecular mass range (ϳ30 kDa). An assay for inhibitory activity revealed that gastric cytosol also contained inhibitory activity at a much higher molecular mass (ϳ200 kDa) as evident in Fig. 3B. When brain cytosol was fractionated in the same way as above, the stimulatory activity eluted in the same position as above (ϳ30 kDa), whereas no inhibitory activity was observed in any position (data not shown). We then concluded that the gastric cytosol contains both stimulatory and inhibitory factors, whereas brain cytosol only contains the former.
Purification of Stimulatory Factors from Brain Cytosol-We first tried to purify the stimulatory factor(s) from gastric cytosol. However, we encountered too many difficulties and had to switch the source to brain cytosol, supposing that the stimulatory factor(s) in both tissues were in common.
The stimulating activity in the brain cytosol, which eluted at approximately 30 kDa with a Sepharose CL4B column, was separated by another gel filtration column, Superdex 200, good for the smaller molecules. As shown by open bars in Fig. 4, the stimulating activity split into two broad peaks (designated as peak 1 for the larger and peak 2 for the smaller) and the total as well as the specific activity drastically decreased. We postulated that the activity was due to multiple components. To examine this hypothesis, aminopyrine accumulation assay was repeated in the presence of peak 2 at the concentration that showed little stimulating activity by itself. As evident from the filled bars in Fig. 4, peak 1, with an apparent molecular mass of 20 kDa, had potent acid stimulating activity in the presence of peak 2. Although not shown in the figure, we confirmed in other purification experiments that the other fractions showed no such activity. As described later, peak 2 did not appear to be peptide, so we decided to purify peak 1 first, employing the bioassay in the presence of peak 2.
Purification of Peak 1 and Identification of p35-In order to increase the scale of the purification, we employed bovine brain instead of rabbit as a source. To ensure the full activity, the assay of purified material using digitonin-permeabilized glands was performed in the presence of a threshold amount of peak 2. In the pilot experiments, it was found that bovine brain A, fractions 10 -26 were tested for stimulatory activity on aminopyrine accumulation in digitonin-permeabilized glands. Note that the peak of stimulation was ϳ19 -21. B, fractions 10 -18 (300 l each) were also tested for inhibitory activity on aminopyrine accumulation stimulated by brain cytosol in digitoninpermeabilized glands. Note that the peak of inhibition was ϳ13-15. As for fractions 1-9, we confirmed in other purification experiments that no obviously active peaks were evident. was also active, and the stimulatory activity adsorbed to DEAE-Sepharose at pH ϭ 7.3 (but not at 6.7) and was retained in red-agarose.
We thus applied bovine brain cytosol to a DEAE-Sepharose column and eluted by decreasing pH, and then applied on red-agarose and eluted by high salt, as described under "Experimental Procedures." This partially purified material was then separated on a Superdex 200 gel filtration column, and it was found that the fraction corresponding to the apparent molecular mass of 20 kDa showed acid stimulating activity as in the case of rabbit brain (data not shown). This active fraction was further separated on another gel filtration column, HPLC Diol-150, and the activity was concentrated in the fraction corresponding to the apparent molecular mass of 10 kDa (fraction numbers 18 and 19 in Fig. 5A). The fractions, including these two, were analyzed on 15% SDS-PAGE, and a protein of 35 kDa correlating with activity was identified (Fig. 5B). This 35-kDa protein was designated as p35. The differences in apparent molecular masses (35 kDa in SDS-PAGE, 20 kDa in Superdex 200, and 10 kDa in Diol-150) might reflect the differences in affinity of this protein to each resin. The purification procedure is summarized in Table I, where the biological activity is expressed relative to the maximal aminopyrine ratio of the crude brain cytosol and is arbitrarily designated as 100.
Identification of p35 as PITP-The final preparation of p35 was separated on SDS-PAGE, blotted on a nitrocellulose membrane, and digested by V8 protease. The digest was separated on a reverse-phase HPLC, and 10 peptide fragments were obtained. From analysis with a peptide sequencer, two of them gave reliable sequences. By homology search, both of these two fragments showed high homology to human and rat PITPs (Fig.  6A). Although it was difficult to identify p35 either as the ␣ or ␤ isoform without knowing the amino acid sequence of its bovine counterpart, it has been reported that there is no functional difference between the ␣ and ␤ isoforms (17). As cDNA of rat ␤-PITP was available, we made this type of PITP as a GST fusion protein and then GST was cleaved off by thrombin treatment. As shown in Fig. 6B, recombinant ␤-PITP stimulated aminopyrine accumulation in the digitonin-permeabilized glands in the presence of peak 2, whereas a negative control, GST, did not. However, the maximal effect of PITP plus peak 2 reached as high as 60% of that of brain cytosol. This suggests that there exists factor(s) other than PITP (plus peak 2), which partially explain the stimulatory effect of cytosol. The stimulation by ␤-PITP was also resistant to 50 M PKI, as was that of peak 1 (data not shown). In separate experiments, it was shown that neither 10 M phosphatidylinositol nor 10 M phosphatidylinositol 4,5-bisphosphate (PIP 2 ) augmented the stimulatory effect of ␤-PITP (data not shown), suggesting that peak 1 is different from these phosphoinositides.
Using p35 purified from bovine brain as antigen, anti-p35 antiserum was raised in rats and it recognized recombinant rat PITP as well as bovine p35. Using this antiserum, the presence of anti-p35 positive bands were screened in rabbit brain fractions separated on a Superdex 200 column. As shown in Fig.  7A, it was evident that PITP existed in fractions 10 and 11, where peak 1 activity was also found (see also Fig. 4). It was confirmed that the active peak contained PITP in gastric cytosol as well (Fig. 7B). In the latter case, for cytosol fractionated by Sepharose CL4B, there was some discrepancy between the biological activity and the contents of p35. The activity in the early fractions (approximately fraction 18 in Fig. 3A) showed low stimulatory activity, whereas those in the late fractions (approximately fraction 22 in Fig. 3A) showed high activity as compared with the p35 contents. This could be explained by the presence of the inhibitory factor and peak 2. The early fractions have been completely separated from the low molecular weight activator peak 2, but not from the inhibitory factor, and subsequently the biological activity in the early fractions would have been underestimated. On the other hand, the late fractions have been completely separated from the high molecular weight inhibitor peak but not from activator peak 2. Subsequently, the biological activity in the later fractions was overestimated in terms of their p35 contents.
Characterization of Peak 2-To date, the active component in peak 2 from the rabbit brain cytosol has not been identified. To find a clue for its identification, we examined its features, and this is summarized in Table II. Peak 2 was harvested as a fraction with apparent molecular mass of 1.8 kDa on a Superdex 200 gel filtration column. Its activity was unchanged after 15 min of boiling, and was also resistant to treatment with 0.05% chymotrypsin or 0.05% carboxypeptidase at 30°C for 1 h. These indicated that the active component was non-peptide. It did not adsorb to an ODS reverse-phase column in phosphate at pH ϭ 2.6. After the extraction of lipids with methanol/chloroform (2:1), the activity remained in the aqueous phase. The activity of peak 2 was stable after freeze-drying, FIG. 5. Identification of p35 as an active component of peak 1. Bovine brain cytosol was sequentially purified by DEAE-Sepharose, red-agarose, and Superdex 200 gel filtration, and an active fraction was seen corresponding to the apparent molecular mass of 20 kDa, as in the case of rabbit brain (see Fig. 4). A, this active fraction was further separated on HPLC Diol-150, and the activity (aminopyrine accumulation in digitonin-permeabilized glands in the presence of peak 2; open column) and protein concentration (A 280 ; closed circles) was measured. The positions of the molecular size markers are shown at the top of the panel. a, blue dextran (2,000 kDa); b, bovine serum albumin (68 kDa); c, ovalbumin(45 kDa); d, cytochrome c (10 kDa). The stimulating activity was recovered in the fraction corresponding to the apparent molecular mass of 10 kDa (fraction numbers 18 and 19; asterisk). B, fractions 14 -21 in panel A were analyzed on 15% SDS-PAGE and stained with Coomassie Blue. Note that the active fraction was correlated with the bands near 35 kDa assigned by the asterisk (fractions 18 and 19).
but it disappeared within a few days as an aqueous solution even in the frozen state.
Changes in the Stimulatory and Inhibitory Components in the Gastric Glands by Activation of Acid Secretion-We examined the possible changes in activity of both the stimulatory and inhibitory components with activation of acid secretion in gastric glands. A suspension of isolated rabbit gastric glands was separated into two aliquots; one was maximally stimulated with 100 M histamine plus 50 M isobutylmethylxanthine at 37°C for 30 min, whereas the other was kept in resting state by the addition of 100 M cimetidine. After the treatments, the glands were harvested and homogenized and the cytosol was prepared. Each cytosol was separated on a gel filtration column, Sepharose CL-4B, and both stimulatory and inhibitory components were assayed using digitonin-permeabilized glands.
Unexpectedly, the activity of the stimulatory component was not changed at all by stimulation of the gastric glands (Fig. 8A). Moreover, the activity of the inhibitory component was markedly increased in the stimulated, not resting, glands (Fig. 8B).
We have tried to identify the inhibitory components, but have not yet succeeded. The main difficulty is that the inhibitory activity is unstable and always spreads into multiple peaks with little activity after the separation of any type of column chromatography. This could indicate that the active material consists of multiple components functioning synergistically.

DISCUSSION
Reconstitution of the Acid Secretory Response of Digitoninpermeabilized Glands with Brain Cytosol-It has been known that permeabilization of gastric glands with digitonin ruins TABLE I Purification of p35 from bovine brain Each fraction was assayed for aminopyrine accumulation in digitonin-permeabilized gastric glands. In order to quantify the activity, the maximal aminopyrine ratio stimulated by brain cytosol (1.5 mg/ml) was arbitrarily designated as 100 units (i.e. specific activity of 66.7 units/mg). The activity of the other fractions was expressed by this unit. Assay for steps 2-5 contained 0.1 ml of peak 2 fraction/1 ml to attain the maximal stimulation. The protein amount in step 5 was estimated by Coomassie Blue staining of the SDS-PAGE.
Step  6. Identification of p35 as PITP. A, partial peptide sequences of p35, human and rat ␣and ␤-PITP. For PITPs, the first part corresponds to the amino acids from 17 to 28, and the second part from 155 to 162. B, stimulating effect of recombinant rat ␤-PITP (50 g/ml) in the presence of peak 2 on the aminopyrine accumulation in digitoninpermeabilized glands. GST was used as a negative control, and the effects were expressed as the maximal aminopyrine ratio obtained by crude brain cytosol (1.5 mg/ml). Aminopyrine ratio was measured in duplicate, and means Ϯ S.E. of four independent experiments are shown.
FIG. 7. The presence of PITP in the active fractions from brain and stomach. A, the rabbit brain cytosol was separated on a Superdex 200 gel filtration column, and each fraction was separated on 15% SDS-PAGE, blotted onto a PVDF membrane, and probed with anti-p35 polyclonal antiserum. Lane S shows recombinant rat ␤-PITP. Note that the activity correlates with the amount of p35/␤-PITP in Fig. 4. B, the rabbit gastric mucosal cytosol was separated on a Sepharose CL4B gel filtration column and fractions 14 -24 were separated on 15% SDS-PAGE and probed with anti-p35 polyclonal antiserum. Fractions 18 -21 are enriched in p35/␤-PITP. Note that the activity roughly correlates with the amount of p35/␤-PITP in Fig. 3, and the small discrepancy is explainable by the presence of the inhibitory factor and its peak 2 (see "Results"). their responsiveness to any secretagogues, although their ability to maintain the proton gradient preformed in the luminal space was preserved (6). This phenomenon has been attributed to the leakage of the proteins essential for the activation process of acid secretion. Taking advantage of this, one might postulate that these essential proteins could be identified by "reconstitution" of secretion. In fact, several proteins involved in catecholamine secretion in the chromaffin cell have been identified employing this strategy (9 -12). However, no such success has been reported in the case of parietal cell. We have also repeated this type of reconstitution experiment using gastric cytosol and came up with nothing. We have now realized that it was due to the inhibitory component in gastric cytosol, this finding becoming possible from our discovery of the stimulating effect of brain cytosol in digitonin-permeabilized glands.
Although the responsiveness of the permeabilized glands to cAMP recovered to some extent in the presence of submaximal concentration of brain cytosol, the maximal response (which was resistant to PKA inhibitor peptide) was obtained by brain cytosol alone, suggesting that there should be a stimulatory factor downstream of PKA. Ca 2ϩ ion is essential for the secretory process in general, and there is a marked potentiating interaction between Ca 2ϩ -dependent and cAMP-dependent pathways in gastric acid secretion (18). Considering that, it appears rather odd that 100 M Ca 2ϩ does not augment, but rather inhibits, brain cytosol-stimulated acid secretion. However, a similar phenomenon was also observed in ␣-toxin-per-meabilized (6) or ␤-escin-permeabilized (3) glands, where cAMP-stimulated acid secretion was not augmented, but rather inhibited, by Ca 2ϩ . Myosin light chain kinase inhibitory peptide, SM-1, which potently inhibited cAMP-dependent acid secretion in ␤-escin-permeabilized glands (3), was also effective in the present model. This observation again supports our opinion that myosin light chain kinase-like, or SM-1-sensitive kinase, is involved in the activation step downstream of PKA. GTP␥S, a non-hydrolyzable analogue of GTP, is known to be a potent inhibitor of acid secretion in glands permeabilized by ␣-toxin (6) or ␤-escin (3), possibly due to the disturbance of the normal cycle of the small GTP-binding protein. This compound also inhibited brain cytosol-stimulated acid secretion in the digitonin-permeabilized model, although the extent of inhibition was weaker than that observed in the ␤-escin model. This might reflect the leakage or the disturbance of the normal distribution of target GTP-binding proteins, which could not be supplemented by the addition of brain cytosol. From these observations, it could be concluded that acid secretion stimulated by brain cytosol in digitonin-permeabilized glands has a similar property to that observed in our previous model, ␤-escin-permeabilized glands, except that the activation occurs at a point downstream of PKA.
Identification of Stimulatory Factors in Brain Cytosol-Using the digitonin-permeabilized gland model as an assay system, we purified the active components in the brain cytosol and finally identified PITP as one of the activation factors of acid secretion. It would be needless to mention, but we cannot conclude that the active components in the brain are completely same as that in the parietal cell. We started the purification on the assumption that there might be common substances, and in this consequence have presently identified PITP and peak 2, which could explain the stimulatory activity present in the gastric mucosal cytosol, at least in part.
PITPs function in transporting phosphatidylinositol and phosphatidylcholine between intracellular membranes and subsequently regulate the activity of phospholipase C via the synthesis of PIP 2 (19). Using the permeabilized cell model, this protein was also identified as the factor involved in the ATPdependent priming step in chromaffin cell (9) as well as in the budding of the vesicles in PC12 cells (20). It has been clarified that phosphatidylinositides play important roles in vesicular transport (21). It would be reasonable to postulate that PITPs play an important role in gastric acid secretion, since vesicular traffic is considered to be the crucial step in the activation of parietal cell (1,22). In the chromaffin cell, PITP is essential for the priming step by ATP, but it does not directly regulate its Ca 2ϩ -dependent secretion step. In the present digitonin-permeabilized glands, the effect of PITP was independent of cAMP. This might indicate that PITP is not the regulatory switch of acid secretion, but is indeed an essential component of the vesicular transport system common to all the secretory cell types. It is necessary in future study to elucidate the possible involvement of PITP in the intracellular membrane traffic. Interestingly, our recent work (23) revealed that the K ϩ per-TABLE II Characteristics of peak 2 activity Peak 2 was purified from the rabbit brain cytosol as described under "Experimental Procedures." The sample was assayed by following aminopyrine uptake of digitonin-permeabilized gastric glands after various treatments.  8. Effects of stimulation of gastric glands on their contents of stimulatory and inhibitory activities in the cytosol. Resting (incubated with 100 M cimetidine for 30 min) or stimulated (100 M histamine plus 50 M isobutylmethylxanthine) rabbit isolated glands were homogenized to prepare cytosol, which was separated on a Sepharose CL4B pre-equilibrated with the assay buffer. A, the fraction near 30 kDa was tested for stimulating activity on aminopyrine accumulation in digitonin-permeabilized glands and expressed as the percentage of maximum obtained by crude brain cytosol (1.5 mg/ml). B, the fraction near 200 kDa was tested for inhibitory activity on aminopyrine accumulation stimulated by crude brain cytosol (1.5 mg/ml) in digitonin-permeabilized glands and expressed as percentage of inhibition. Mean Ϯ S.E. of four pairs of experiments are shown. The difference between resting and stimulated group in B was statistically significant (Student's t test, p Ͻ 0.05). meability, essential for the activity of H ϩ ,K ϩ -ATPase on the apical membrane of parietal cell, was dependent on PIP 2 . This observation indicates the critical role of phosphoinositides in acid secretion and supports the potential involvement of PITP in parietal cell function.
Peak 2, the Low Molecular Weight Active Component-By using separation with a Superdex 200 column, it was revealed that the active component in brain cytosol contained a factor with apparent molecular mass of 1.8 kDa (peak 2). As peak 2 was resistant to treatments with proteases and heat, it seems not to be peptide. Following our observation that peak 2 potently augmented the PITP-stimulated acid secretion, we then examined the possibility that peak 2 was either PI or PIP 2 , but neither was the case. In general, PITP does not require exogenous PI in the cell level since PI is synthesized within endoplasmic reticulum, and PITP is able to transport it to the target membrane (17).
We have tried to purify peak 2, but to no avail. Looking at its physical properties, it appears to be similar to phosphoinositides, although any of derivatives presently available fail to mimic the effect of peak 2. Further work is obviously necessary to identify this interesting factor.
Inhibitory Factor in the Gastric Glands-In the present study, we revealed that gastric mucosal cytosol contains an acid-stimulatory factor similar to that in brain cytosol in the low molecular weight range, and it also contains an acid-inhibitory factor in the high molecular weight range. This indicates that gastric acid secretion may be regulated by a balance between these positive and negative factors. Walent et al. (2) found an inhibitory factor of catecholamine secretion in the flow-through fraction of a Matrix Green column when they purified CAPS (calcium-dependent activator protein for secretion), which activates catecholamine secretion in chromaffin cell. Similar factors might exist in various secretory cell types.
Based on the fact that the stimulatory factor is independent of cAMP, it might be possible that the inhibitory factor, and not the stimulatory factor, is involved in the switching of activation. To examine this hypothesis, we compared the activity of these factors in resting and stimulated gastric glands, and we found that the inhibitory activity in the cytosol was markedly increased by stimulation, whereas the stimulatory activity was not. This result was somewhat puzzling because it appeared that stimulation increased the inhibitory activity. However, it would be reasonable to interpret this in such a way that the inhibitory factor, which exists and inhibits acid secretion on the membrane of parietal cell, translocates to the cytosol in association with stimulation and subsequently the inhibition is canceled. In fact, it has been reported that Rab 3 translocates with stimulation from the membrane fraction to cytosol in the neuronal cell (24), and Rab 3 on the membrane inhibits catecholamine secretion in the resting stage (25). Another example is syncollin, which has been identified as a Ca 2ϩ -sensitive inhibitory factor in the fusion process of rat pancreatic zymogen granule (26). Recombinant syncollin inhibits the fusion of zymogen granule by binding to syntaxin. When it dissociates from syntaxin with addition of Ca 2ϩ , the inhibition of the fusion is released. In case of rat parotid gland, where cAMPdependent secretion occurs, there exists a protein factor, which dissociates from VAMP2 (vesicle associated membrane protein 2) depending on activation by PKA (27). In analogy, it would be possible in the parietal cell that inhibitory factor(s), regulated by PKA, can be translocated from the membrane to cytosol during the activation process.
In conclusion, we established a new system useful for assaying essential factors of regulation of gastric acid secretion with reconstituted digitonin-permeabilized rabbit gastric glands. Using this system, we identified PITP as an activation factor of acid secretion. Furthermore, we found an inhibitory component in gastric cytosol and postulated a hypothesis that the acid secretion was regulated via this inhibitory component. The present system is considered to be a powerful tool to identify unknown factors and to analyze their mode of action in the regulation of gastric acid secretion.