Proc. NatI. Acad. Sci. USA Vol. 76, No. 12, pp. 6689-6693, December 1979 Physiological Sciences Cellular site of secretion (H+ transport/membrane transformation/H+,K+-adenosinetriphosphatase//canaliculus) D. R. DIBONA*, S. ITOt, T. BERGLINDHW, AND G. SACHS* *Department of Physiology and Biophysics, Nephrology Research and Training Center, and *Laboratory of Membrane Biology, University of Alabama in Birmingham, University Station, Birmingham, Alabama 35294; and tDepartment of Anatomy, Harvard Medical School, Boston, Massachusetts 02115 Communicated by Alex B. Novikoff, August 30,1979

ABSTRACT Isolated gastric glands of the rabbit were ex- emission spectrum shifts in an analogous way as the local dye amined both with differential interference-ontrast microscopy concentration increases, either in free solution or when the dye and with electron microscopy to describe the morphologic is bound to an anionic polymer such as RNA or correlates of acid secretion. Stimulation of the glands with heparin. histamine resulted in the development of intracellular spaces In this report, we applied these approaches in discrete within the parietal cells. A similar transformation was produced combinations in an attempt to specifically identify the site of by addition of 1 mM aminopyrine, whether the weak base was gastric acid secretion. added in the presence of normal-K+ (5.4 mM) or high-K+ (108 mM) solutions. The intracellular space was compatible with the METHODS expanded canaliculus described in stimulated parietal cells. Confirmation that the space produced by histamine is the site Rabbit gastric glands were prepared by described methods (4). of acid secretion was gained by combining fluorescence and These involved high-pressure perfusion of the through interference-contrast methods in the presence of the dye acri- the gastric artery with phosphate-buffered saline at 370C, dine orange, which displays a pH-dependent metachromasia mincing the separated epithelial layer with scissors, and di- in its emission spectrum. Human I resulted in an ob- gestion with collagenase. The isolated glands were incubated servable discharge of peptic granules. at 370C in a stirred vessel in a medium containing: Na+, 143.4 Although it is generally assumed that the parietal cell is re- mM; Cl-, 139.8 mM; K+, 5.4 mM; PO42-, 6.0 mM; Ca2+, 1.0 sponsible for acid secretion by the stomach, exact localization mM; Mg2+, 1.2 mM; So42-, 1.2 mM; glucose, 2 mg/ml, and of the process not rabbit albumin, 2 mg/ml. In some experiments, the K+ con- has been previously demonstrated in intact centration was increased to 108 mM at the expense of Na+; in tissue (1). It is clear that there is a dramatic trapsformation of some experiments, 15 mM NaSCN was also present. Uptake of parietal cell morphology accompanying stimulation of acid amino[14C]pyrine was monitored as described (6). Acridine secretion (2), in which a complex intracellular array of mem- orange was added at 100 ,M to the incubation medium. branes (the tubulovesicles) is replaced by an extensive micro- Tetraphenylphosphonium bromide was added at 100 ,uM. villous infolding of the luminal plasma membrane to form an Accumulation of 36C1- was measured by addition of 1 MCi (3.7 expanded canaliculus. A direct investigation of the significance X 104 becquerels) of -`Cl- to 1 ml of gland suspension, with or of this structural change in terms of its relationship to acid se- without the addition of secretagogue. cretion seemed possible, given recent improvements in cell and For differential interference-contrast (DIC) and fluorescence gland isolation techniques (3, 4), methods for assessment of acid microscopy, a Zeiss inverted microscope (1M-35) was used. secretion (5, 6), and the practical application of microscopic Glands were sampled from the incubation mixture at 370C at examination of living epithelial tissues (7, 8). prescribed times after treatment; for each sample, photographs Acid secretion levels can be readily measured by the con- were taken of glands as they were randomly encountered within centration uptake by the glands of the weak base aminopyrine 3 min after sampling. DIC microscopy was performed by (pK, 5.0) (6). Under maximal histamine stimulation the transillumination with a tungsten/halogen source while fluo- gland-to-medium ratio may be as high as 200. Knowing both rescence was studied with an epi-illumination system and an the size of this compartment and the gland-to-medium distri- HBO-50 light source. Excitation was with Hg lines 404.7 and bution ratio of aminopyrine would allow a direct calculation 435 nm; emission was observed and photographed through a of the pH of the compartment contents. In this respect, it is also high-pass reflecting filter (cut off at 510 nm; Zeiss component of interest to examine the consequences of increased medium pair no. 48-77-07). For electron microscopy, the glands were K+ concentration. The role of K+ in acid secretion by the fixed in a solution containing 2% (wt/wt) formaldehyde, 2.5% stomach (9) has been specified by the transport characteristics (wt/wt) glutaraldehyde, and 0.02% 2,3,6-trinitrophenol in pH of the gastric H+,K+-ATPase (10) and it has been demonstrated 7.2 sodium phosphate buffer for 2 hr, postfixed in 1% OS04 for that increased medium K+ results in an aminopyrine accu- 1 hr. dehydrated, and embedded in Epon for thin sectioning. mulation nearly equivalent to that found in the secretagogue- stimulated state (6). RESULTS Characterization of an identified compartment can be fa- cilitated by the use of acridine orange, a fluorescent dye that Stimulation of Acid Secretion by Histamine. The time has been used as an indicator of energization in many mem- course of histamine stimulation of acid secretion has been brane systems and notably for everted submitochondrial par- documented with measurements of the rate of amino['4C]- ticles (11). In gastric vesicles, acridine orange is distributed pyrine accumulation (6). After addition of the secretagogue at across and bound on membranes as a function of imposed pH 100 AM, there is a rapid increase in weak base accumulation, gradients (12). At low pH the green fluorescence at 530 nm is which reaches a maximum at 25-30 min. The morphologic quenched and a red fluorescence at 624 nm appears. The transformation of histamine-stimulated glands that we visual- ized by DIC microscopy follows a similar course. Resting glands, The publication costs of this article were defrayed in part by page as in Fig. la, allow identification of both parietal and peptic charge payment. This article must therefore be hereby marked "ad- (chief) cells. The parietal cells are conical, showing triangular vertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact. Abbreviation: DIC, differential interference contrast. 6689 Downloaded by guest on September 30, 2021 6690 Physiological Sciences: DiBona et al. Proc. Natl. Acad. Sci. USA 76 (19*

FIG. 1. Gastric gland. (DIC optics; X650.) (a) Resting gland suspended in normal Ringer's solution. Characteristic features: P, parietal cell; Z, peptic cell; M, mitochondria; G, pepsinogen granules. (b) Gland 30 min after histamine Addition. Note the appearance of vacuoles (V). profiles, and contain characteristic 1-am particles-mito- so that by 30 min some cells were observed to contain one or chondria-densely packed toward the basal end of the cell. The more intracellular fluid compartments that involved as much peptic cells are recognized as those containing granules of about as half of the parietal cell volume. Typically, maximally stim- 2 ,um diameter, concentrated toward the luminal surface. No ulated glands appeared as in Fig. lb. Electron micrographs of morphologic changes in the peptic cells were evident after glands fixed under similar conditions showed a comparable histamine addition, while a progressive development of intra- transformation with a loss of the tubulovesicles of the resting cellular vacuoles was noted in the parietal cells. Individual cell and formation of canalicular profiles lined with microvilli vacuoles became prominentin many parietal cells by 10 min much as has been previously described (2). These results suggest to and coalesce that the enlarging vacuolar space may be'the site of aminopy- after histamine addition; these appeared enlarge rine accumulation.

b JPP";; ^; FIG. 2. Gastric gland. (DIC optics; X650.) (a) Gland incubated in 5.4 mM K+, recorded after the addition of 1 mM aminopyrine. Formed vacuoles (V) were restricted to a small fraction of the parietal cells. (b) Gland incubated in 108 mM K+ Ringer's solution, recorded 5 min after addition of 1 mM aminopyrine. Vacuolation extends to all observed parietal cells. Downloaded by guest on September 30, 2021 Physiological Sciences: DiBona et al. Proc. Natl. Acad. Sci. USA 76 (1979) 6691

2.;. *te,s =

FIG. 3. (a) Electron micrograph of a glandular parietal cell incubated in 5.4 mM K+ Ringer's solution and 1 mM aminopyrine. Stages of development of osmotically generated canaliculi are seen. Large canaliculi have few microvilli, conceivably due to stretching of the membrane. (x2300.) (b and c) Glandular parietal cells incubated in 108 mM K+ Ringer's solution and 1 mM aminopyrine, showing large expanded intracellular canaliculi with few microvilli (b) as well as canaliculi with multiple microvilli (c). (b, X2300; c, X3600.) Effects of Addition of Aminopyrine. In unstimulated Use of Acridine Orange. The pH-dependent distribution glands, aminopyrine accumulation increases with increasing and fluorescence spectrum of the dye acridine orange offered K+ concentration of the incubation medium (6). The amino- an exceptional opportunity to further characterize the parietal pyrine accumulation ratio (when the base is added at 1 MM) is cell vacuolization noted above. In unstimulated resting glands about 20 in normal medium, and it rises to 75 when K+ con- (not shown here) addition of acridine orange resulted in a rather centration is increased from 5.4 mM to 108 mM at the expense uniform green fluorescence of all of the glandular components. of Na+. Should this accumulation be limited to a compartment The lack of selective nuclear staining by acridine orange was corresponding to about 5% of glandular volume, higher con- surprising; cytoplasmic components of these cells may have an centrations of the impermeant RNH3+ form would result in a unusually high binding affinity. Fig. 5a illustrates a gastric solution buffering which, with continued pump activity, would gland 30 min after histamine stimulation; the DIC image has locally cause an increase in osmolality and consequent swelling been recorded in the simultaneous presence of epi-illuminating of the aminopyrine space. It was noted, as in Fig. 2a, that, after fluorescence optics. One can see that two parietal cells in par- the addition of 1 mM aminopyrine to glands incubated in 5.4 ticular show signs of extensive vacuolization and that vacuolar mM K+, a minor fraction of the parietal cells became vacuo- contents are red relative to the surrounding cytoplasm. In Fig. lated; most of the observed parietal cells, however, were not involved in these unstimulated preparations. In high-K+ solu- tions (108 mM), glands were largely unaffected in the absence of aminopyrine except that a diffuse swelling of cells was oc- casionally evident. Within 5 min of aminopyrine addition, as in Fig. 2b, however, vacuolar formation was widespread and glandular structure was comparable to that after histamine stimulation. Addition of SCN-, an inhibitor of acid secretion and of aminopyrine accumulation, prevented the morphologic changes noted above. Fig. 3 a and b shows the electron micro- scopic appearance of the active glands in normal and high-K+ media in the presence of 1 mM aminopyrine. Again, it appeared that the basal level of aminopyrine accumulation was due to its concentrated sequestration in a few cells rather than diffuse accumulation among the population at large. -"Cl Uptake. There was an increase in glandular Cl- uptake after histamine stimulation. The equilibrium level of Cl- was 46 mM, and this level increased to 66 mM after stimulation. Assuming that the measured increase was due to accumulation of isotonic HCO in a newly formed space, a shift from 46 to.66 mM would correspond to the development of a space that constituted about 20% of glandular volume-a value in general agreement with the morphologic findings above. Note that, even at rest, given an internal cell potential of 60-72 mV (in- terior negative) (13), Cl- is in electrochemical disequilibrium and active Cl- accumulation is indicated. Effect of Gastrin. Addition of gastrin alone does not effect glandular acid secretion (4), and its morphologic expression was quite different from that of the stimulatory conditions studied above. The hormone's activation of pepsin secretion was evi- FIG. 4. Gastric gland by DIC microscopy after 30-min treatment denced by a degranulation of the peptic cells that was nearly with 0.1 ,uM human gastrin I. Peptic granules are lost and an open complete by 30 min. Fig. 4 illustrates such a gland. lumen (L) is visible. (X800.) Downloaded by guest on September 30, 2021 6692 Physiological Sciences: DiBona et al. Proc. Natl. Acad. Sci. USA 76 (1979)

FIG. 5. (a) Histamine-stimulated gastric gland incubated with 100 /AM acridine orange, recorded with combined DIC and fluorescence optics. Cell A reveals a well-developed canaliculus, cell B contains three prominent vacuoles; these structures show a strong red fluorescence against the green of the surrounding cytoplasm. (X800.) (b) Fluorescence micrograph of the field in a. Without DIC optics, the shallowness of field is lost, the fluorescent image is intensified, and an impression of the three-dimensional distribution ofthe dye is gained, showing the ramifying nature of the canaliculus. (X800.) (c) A gland as above but subsequently exposed to 500 mM sucrose, resulting in loss ofvacuoles and disappearance of the red fluorescence from the cells and the appearance of a yellowish-orange color in the glandular lumen. Cytoplasmic contents are now green as in resting glands. (Optics as in a; X800.) (d) A gland in a but after treatment with 100 1AM tetraphenylphosphonium. The red fluorescent vacuoles have disappeared, but peptic granules show an orange fluorescence indicating acridine orange accumulation. (Optics as in a; X500.) Downloaded by guest on September 30, 2021 Physiological Sciences: DiBona et al. Proc. Natl. Acad. Sci. USA 76 (1979) 6693 5b, for which the transilluminating interference-contrast in high K+ medium, therefore, appears to be sufficient for condenser has been removed, the strongly contrasted fluo- modification of the intracellular membrane system. This might rescent pattern is observed alone; the marked difference in be due to an induction of fusion through an increase in surface fluorescence emission between cytoplasmic and vacuolar free energy of osmotically stressed vesicles or, alternatively, to contents is far more striking. osmotic expansion of an otherwise collapsed tubular system. The The fluorescence characteristics of acridine orange depend latter hypothesis, which would imply one site of attachment for upon pH, concentration, and the degree of binding to negative several collapsed tubules, avoids the need for selective pro- sites; a pH gradient is expected to result in dye accumulation cessing of the membrane as required by a fusion-segregation in the acidic region, as has been shown for gastric vesicles (12). model (2, 15). The limiting factor for osmotic expansion could Fig. 5 a and b suggests that the vacuolar contents are markedly be the flux of KCI into the tubular space followed by a pump- acidic, and, therefore, that the vacuolar membrane-the driven K+-for-H+ exchange across the surface of the canaliculus margin of the secretory canaliculus by electron microscopy-is (16). In this scheme osmotic expansion is coupled to the con- the site of acid secretion. tinuing influx of KCI driven by the K+ gradient, which is In the isolated preparation, in which the lumen is virtually maintained by the H+-K+ pump. The buffering action of the closed and in which there are no physical constraints on the weak base aminopyrine would decrease the steady-state H+ contraluminal surface, vacuolar (canalicular) distension appears gradient and the pump, in an effort to maintain a fixed H+ to proceed without obligatory extrusion of formed acid. We gradient, would then drive the hyperosomolarity. noted, however, that if the preparation was abruptly challenged The observed action of gastrin is also of interest. This hor- with hypertonic sucrose, as in Fig. 5c, the rapid osmotic mone was able to effect a degranulation of the peptic cells but shrinkage collapsed the vacuoles and produced a yellowish- produced no morphologic changes indicative of acid secretion. orange fluorescence along the luminal margins. Similarly, it has been noted (unpublished-data) that gastrin does The lipid-permeant cation tetraphenylphosphonium dis- not stimulate acid secretion in these isolated glands unless a tributes across both the plasma and mitochondrial membranes phosphodiesterase inhibitor (e.g., isobutylmethylxanthine) is of cells (14) and glands (13). With 100 uM tetraphenylphos- also present. It appears that, although the gastrin receptor is phonium the mitochondria are uncoupled via dissipation of the intact and the peptic cell is responsive, essential factors for mitochondrial potential gradient and acid secretion is inhibited. gastrin action on the parietal cell are missing in the isolated This is shown in Fig. 5d, in which it also appears that this gland preparation. treatment results in an accumulation of acridine orange by the Finally, our results with acridine orange provide evidence peptic granules, perhaps indicative of the presence of a pH that the canalicular contents are at a low pH and that the gradient or of a modification of binding to negative sites. Red bounding membrane is the site of acid secretion. Previous work fluorescence of the parietal cells is also lost here. SCN- also with this dye's distribution characteristics (12) supports these inhibited parietal cell accumulation of acridine orange. conclusions. In this respect, the combined use of DIC and flu- orescence methods has greatly facilitated unambiguous inter- DISCUSSION pretation of results and illustrated the potential utility of this The data presented here provide direct evidence in living tissue approach for similar problems in the correlation of structure that the parietal cell is the acid-secreting component of the and function. gastric mucosa and establish the validity of the secretagogue- induced morphological transformation of the parietal cell, This research was supported in part by National Institutes of Health which has previously been investigated only in fixed tissue. It Grants AM 25788, AM 15878, and AM 21588 and by National Science appears that formation of a discrete canaliculus is preceded by Foundation Grant PCM 78-09208. the development of individual vacuoles. Although we have 1. Rehm, W. S. (1972) Arch Int. Med. 129,270-278. noted several instances of vacuoles coalescing to form a single 2. Sedar, A. W. (1965) Fed. Proc. Fed. Am. Soc. Exp. Biol. 24, intracellular network, canalicular development seems less well 1360-1367. directed than in the intact stomach. An explanation for this may 3. Blum, A. L., Shah, G., Wiebelhaus, V. D., Brennan, F. T. & Sachs, lie in the fact that, in the isolated gland, physical constraints on G. (1971) Gastroenterology 61,189-200. 4. Berglindh, T. & Obrink, K. J. (1976) Acta Physiol. Scand. 96, the glandular walls have been removed and the lumen is no 150-159. longer patent. In the intact stomach, in which connective tissue 5. Soil, A. H. (1977) Gastroenterology 72, 1166. and muscularis mucosae preclude expansion of the conical 6. Berglindh, T. (1978) Acta Physiol. Scand. Suppl., Proc. Symp. parietal cells, vacuolar coalescence and fusion with the luminal Gastric Ion Transport, 55-68. cell membrane to form the canalicular network might be fa- 7. Allen, R. D., David, G. B. & Namarski, G. (1969) Z. Wiss. Mi- cilitated. Prevention of this transformation with uncouplers of krosk. 69, 193. oxidative the view that mor- 8. DiBona, D. R. (1978) J. Membr. Biol. 40, 45-70. phosphorylation strengthens the 9. Harris, J. B., Frank, H. & Edelman, I. S. (1958) Am. J. Physiol. phologic change is a manifestation of acid secretion. 195,499-504. Although aminopyrine has been used often as a probe for acid 10. Sachs, G., Chang, H., Rabon, E., Schackmann, R., Lewin, M. & secretion in these isolated systems, its site of accumulation has Saccomani, G. (1976) J. Biol. Chem. 251, 7690-7698. not been previously identified. The addition of high amino- 11. Dell'Antone, P., Colonna, R. & Azzone, G. F. (1972) Eur. J. pyrine concentrations to glands suspended in normal K+ me- Biochem. 24, 553-565. 12. Rabon, E., Chang, H. & Sachs, G. (1978) Biochemistry 17, dium suggests that the "basal" aminopyrine accumulation is 3345-3353. due to a small number of cells with high individual rates of 13. Berglindh, T. & Sachs, G. (1979) Fed Proc. Fed. Am. Soc. Exp. pump activity. One limitation for generalized pump activity Biol. 38, 962. seems to be the K+ concentration, as might be predicted from 14. Lichtshtein, D., Kaback, H. R. & Blum, A. J. (1979) Proc. Natl. the properties of the H+,K+-ATPase (10). Thus, elevation of Acad. Sci. USA 76, 650-654. medium K+ stimulated aminopyrine accumulation (6), indi- 15. Sachs, G., Rabon, E., Hung, H., Schackmann, R., Sarau, H. M. an increased The addition of & Saccomani, G. (1977) in Hormonal Receptors in Digestive cating pH gradient. aminopyrene Tract Physiology, eds. Bonfils, S., Fromageot, P. & Rosselin, G. (in the presence of high K+) induced a morphological trans- (Elsevier, Amsterdam), pp. 347-360. formation of the parietal cell, from a resting to an apparent 16. Helander, H. F., Saccomani, G. & Sachs, G. (1979) Gastroen- secretory state. The generation of an osmotic gradient, at least terology 76, 1153. Downloaded by guest on September 30, 2021