/ournal of lnternal Medicine 1990: 228. Suppl. 1: 17-26

Membrane and protein recycling associated with gastric HCl secretion J. G. FORTE, D. K. HANZEL, C. OKAMOTO, D. CHOW & T. URUSHIDANI From the Department of Molecular and Cell Biology. Division of Biophysics and Cell Physiology. Universit.y of Calgornia. Berkeley. CA. USA

Abstract. Forte JC. Hanzel DK. Okamoto C. Chow D. Urushidani T. (Department of Molecular and Cell Biology, Division of Biophysics and Cell Physiology, University of California, Berkeley, CA. USA) Membrane and protein recycling associated with gastric HCI secretion. /ournal of Internal Medicine 1990: 228, Suppl. 1: 17-26. Stimulation of the gastric parietal cell requires massive membrane transformations as H+pumps from the domain of cytoplasmic tubulovesicles are recruited into the apical plasma membrane domain. The recycling of membrane pools, through fusion and fission processes that accompany stimulation and inhibition of HCI secretion, also involves highly selective events of protein incorporation and segregation. This manuscript describes several proteins that have been identified with the apical plasma membrane from maximally stimulated parietal cells, and broadly characterizes them either as permanent resident proteins of the apical membrane, or transient proteins that move into and out of the apical membrane as the cell progresses through the secretory cycle. A typical example of transient association with the apical membrane concerns the pump proteins, including the 9 4 kDa catalytic a-subunit of the H+K+-ATPaseand its n e w l d discovered /?-subunit glycoprotein. which move between tubulovesicles. Proteins that remain associated with the apical plasma membrane during rest and secretion include actin. and an 80-kDa phosphoprotein, which has been variously called 80 K, ezrin, p8J and cytovillin. and whose phosphorylation is increased by the histamine/cAMP pathway of parietal cell stimulation. An example of a cytosolic protein that becomes associated with the apical plasma membrane after stimulation is a 120-kDa protein, which appears to have protein kinase activity. Note that the identification, localization and characterization of the K+ and CI- transport proteins. which participate in net HCI secretion, are of immediate importance. Keywords: ATPase. ezrin. glycoprotein. parietal, phosphoprotein. proton transport.

Introduction The membrane recycling hypothesis of gastric HCI secretion proposes that the primary proton pump (H'K'-ATPase) progresses through a cytoplasmic membrane domain, the tubulovesicles. at rest, and the apical plasma membrane when the parietal cell is stimulated [ l ,21. A schematic representation of the morphological events characterizing stimulationdependent parietal cell membrane transformations is shown in Fig. 1. These massive membrane rearrangements, presumably mediated through fusion and fission processes, suggest an important role for

the cytoskeletal components that serve to organize and direct the morphological states. It is clear that the fundamental processes for the selection and segregation of intrinsic and associated membrane proteins underlie physiological membrane recycling in the parietal cell. These selection processes not only involve the pump enzyme, but also the segregation of other catalytic or structurally important elements that move into and out of the apical plasma membrane, or remain permanently attached to it. In this report, we will catalogue several proteins (as listed in Table 1)that were found to be associated with the apical plasma membrane of the stimulated 17 IMBS

18

J. G. FORTE et al.

Fig. 1. Schematic representation of parietal cells at rest and after stimulation. The membrane transformations that occur between cytoplasmic tubulovesicles and apical plasma membrane bordering the secretory canaliculi form the basis of the membrane recycling hypothesis of HCI secretion. (Adapted from reference 2).

Table 1. Proteins of the apical membrane of parietal cells

Protein (kDa)

Name

Function

94*

H K'-ATPase

60-80 44*

gp 60-XO Actin X O K, errin. PX 1, cytovillin 120 K 210 K

x-subunit of gastric H + pump Psubunit of H' pump Microvillar structure Microtilament/plasma membrane attachment Protein kinase ? C1- channel?

XO*

120

210

+

Residence in apical membrane Transient Transient Resident Resident Transient Resident ?

* Primary sequence determined.

parietal cell and review the evidence that describes their identification as a resident or a transient protein in the membrane translocation process. In this context we define a resident protein as one which remains in, or associated with, the apical membrane at all stages of the secretory cycle, and a transient protein as one that moves into and out of the apical membrane in phase with stimulation of the cell and return to the resting state.

Monitoring shifts in membrane during the secretory cycle The simplest and most direct procedure to evaluate shifts in the membrane from tubulovesicles to the apical plasma membrane uses differential centrifugation to separate organelles on the basis of size [3, 41. When parietal cells are homogenized without using excessive shearing force, plasma membranes are broken into relatively large fragments, whereas

the cytoplasmic tubulovesicles are, by their very nature, small vesicles that sediment as microsomes. Thus differential centrifugation provides a quick, convenient separation, and permits assessment of the distribution of the total amount of activity or protein between the fractions, with the major drawback being that the fractions are very crude. Additional procedures, such as density gradient centrifugation, can be used to purify further a given organelle or membrane type. These are obviously more timeconsuming and the yield can be rather poor, but they are essential for specifically localizing a given function or protein. The results of an experiment using the crude fractionation scheme are shown in Fig. 2. The objective of these experiments was to examine secretion-dependent membrane redistribution of H'K+-ATPase. using the in-situ rabbit stomach to measure the H' secretory rate directly, as well as the source of mucosal tissue for subsequent membrane

M E M B R A N E A N D PROTEIN RECYCLING

histaminestlmuloted

1

t

P3 53

PI P2 P3 s3

v

I

PI P2 P3 s3

PI P2 P3 s3

P I / P 3 ratlo

Fig. 3. Correlation between the HCI secretory rate and the membrane redistribution index of H'K'-ATPase ( P l/ P 3 ratio) for rabbit gastric mucosa. Experimental procedures were similar to those described in Fig. 1. Anaesthetized rabbits were prepared with gastric fistulas for measurement of HCI secretion. H' secretory rates were stimulated with histamine and subsequently inhibited by a histamine H,-receptor antagonist. At designated times the HCI secretory rates were recorded and the stomach immediately taken for homogenization and membrane fractionation as described in Fig. 2. To quantify the extent of redistribution of H'K'-ATPase among the membrane fractions. we introduced the P1/P3 ratio. i.e. the ratio of total K-pNPPase in the low speed PI pellet to the total K-pNPPase in the microsomal P3 pellet.

19

Fig. 2. Distribution of total H'K'ATPase activity among crude cell fractions isolated from rabbit gastric mucosal homogenates taken at rest (non-secreting), during maximal histaminestimulated acid secretion. and at various times after acid secretion was inhibited by a histamine. H,receptor antagonist. Actual rates of H+ secretion were measured via a gastric fistula for each stomach prior to tissue fractionation. Ouabain-insensitive K+-stimulated pNPPase was used as the marker of H+.K'-ATPase activity. Centrifugal forces for harvesting cell fractions were: P1.40OO g x 10 min: P2. 14500 g x 10 min: P3. 50000 g x 90 min; S3, remaining supernatant. At rest most of the K+-pNPPase was in the P3 fraction (crude microsomes). whereas for maximally stimulated stomach most of the activity was distributed to the low speed P1 fraction. The activity redistributed back to P3 as H' secretion was progressively inhibited.

separation and analysis. The distribution of H+K+ATPase between the crude cell fractions was assayed using the convenient marker of the pump enzyme, K+-stimulated p-nitrophenylphosphatase [4, 51. The results of Fig. 2 clearly show that there was a secretion-related change in the distribution of pump enzyme between the crude microsomal fraction (P3) and the low speed fraction containing apical membranes (P1). For resting, or non-secreting stomachs, the majority of H+K+-ATPasewas in the P3 microsoma1 fraction, thus the ratio of total H+K+ATPase activity in the P1 versus P3 fractions, i.e. the P1/P3 ratio, was very low. At high rates of acid secretion there was a complete reversal of distribution so that the P1/P3 ratio was high, i.e. H+K+-ATPase moved from the microsomal fraction to the apical membrane-containing fraction. The P1 /P3 ratio for the pump enzyme is plotted as a function of the rate of H+ secretion measured at various stages of HCl secretion and inhibition (Fig. 3). Results such as these establish the validity of the P l / P 3 ratio as a biochemical index of HCl secretion and they are wholly consistent with the membrane recycling model for secretion [3, 4, 6. 71. Analysis of the crude cell fractions by SDS-PAGE 2-2

J. G. FORTE et al.

20

microsomes

kOa

200

116 96

66

45

-

--

-

w

-

0

-

- H, K-ATPase - 80 kOa

-

%I,

M.W. StdS.

bands are readily apparent. Major bands occur at 80 and 44kDa. and numerous minor bands can be seen. We will discuss some of the functional properties of the two H+K'-ATPase-enriched membrane vesicle populations, their relationship to HCI secretion, and the activities of their proteic components.

s.a. vesicles

-

CI a

b

C

. I-actin

d

Fig. 4. Protein composition in purified gastric microsomes and stimulated apical plasma membrane (s.a.)vesicles revealed by SDS-PAGE. Lanes a and c have 5 0 jig total protein: lanes b and d have 2 5 jig protein. The single most prominent band in microsomes is the catalytic a-subunit of the H'K'-ATPase at about 94 kDa. There is also a very broad band of glycoprotein in the region of 60-80 kDa molecular weight that does not stain well with Coomassie blue, which we propose as the ,+subunit of the H'K'-ATPase. The s.a. vesicles are also rich in H'K'-ATPase. but in addition there are prominent bands at 40 kDa (actin) and 80 kDa (see text). as well as several minor bands.

reveals that they are heterogeneous with respect to proteins. However, further density-gradient separations conveniently provide membrane vesicle fractions that are much more highly purified and homogeneous [7]. The protein pattern of highly purified tubulovesicles, derived from the crude P3 fraction by two successive sedimentations on a sucrose density gradient, is shown in Fig. 4. The dominant tubulovesicular protein is clearly the 94 kDa catalytic unit of the H'K+-ATPase. In heavily loaded gels one can also see a faintly staining, broad band of glycoprotein ranging from 60 to 80 kDa, which we propose is an associated B-unit of the H+K'-ATPase (see below). Apical membrane vesicles purified from the P1 fraction, from a maximally stimulated rabbit stomach, are also shown in Fig. 4. These highly enriched stimulated apical plasma membrane vesicles (called s.a. vesicles for simplicity) are also rich in H'K'-ATPase, but additional protein

Transport properties of H'K'-ATPase-rich membrane vesicles Both tubulovesicles from the resting parietal cells and apical membrane vesicles from stimulated cells (s.a. vesicles) are rich in the proton pump enzyme which utilizes ATP to pump H+ into the vesicles in exchange for K'. Moreover, in comparing these resting and stimulated vesicle types, there seems to be no functional difference in the activity of the enzyme itself [4, 81. However, the vesicles do differ markedly in their permeability properties to K+ and C1-: s.a. vesicles have intrinsic pathways for the rapid flux of K+ and C1-, whereas tubulovesicles are relatively impermeable to these ions [8-101. It is these permeability properties that provide the principal functional difference in vesicle transport activity, and represent a basis for the regulation of HCl transport in the intact cell. A model of the changes in membrane transport activity during the secretory cycle is depicted in Fig. 5. In the resting parietal cell, the abundant H+K'ATPase is constrained to the cytoplasmic domain of tubulovesicles but because of very limited K' and C1permeability, and the closed nature of the system, the turnover of ATP and secretion of HCI is very low or non-existent. Stimulation of the parietal cell leads to a marked expansion of the canalicular surface as H'K'-ATPase-rich tubulovesicles are incorporated into the apical plasma membrane. High permeability to K' and CI- provides the means for flux of these ions into the canalicular space: K' is actively exchanged for H' by the H+K+-ATPase,and HCI progressively moves by bulk flow into the glandular and gastric lumina as water follows the osmotic gradient provided by the diffusional and pump forces. When the stimulus is withdrawn, the process reverses itself as the H'K+-ATPase is sequestered back into tubulovesicles (e.g. see Fig. 2). Thus the catalytic pump unit, the H+K+-ATPase,is a transient protein that moves in and out of the apical plasma membrane through the secretory cycle. Turning on HCl secretion (and the converse of turning off) involves a sequence of events collectively

MEMBRANE AND PROTEIN RECYCLING

ADP

t POH

21

ADP t POH

I

I

cell activation

I

*

low K +

permeo bilit y

Fig. 5. Schematic representation of parietal cell activation as the cell is transformed from rest to active HCI secretion. In the resting cell.

tubulovesicles contain H'K'-ATPase. but because of low membrane permeability to K+ (and a-) there is very little H' accumulation and virtually no ATP turnover. Cell activation brings about a fusion of tubulovesicles with the apical plasma membrane, transferring the H+K'-ATPase to that surface. In addition. the participation of conductive pathways (possibly activated?)for K+ and c1- movement provides the means for KCI flux into the secretory canaliculi. The H+/K+ exchange pump recycles K+ back into the cytoplasm with the net effect of HCI transfer and ATP turnover. Water flux into the canaliculus is osmotically driven by a net solute flux.

described as stimulus-secretion coupling. The cyclic AMP (CAMP) pathway, which operates through protein A kinase, is seminal to parietal cell activation, although alternative pathways (e.g. Ca2+,protein C kinase) may also be involved directly or synergistically. Secretory effector mechanisms at the apical cell surface must certainly involve membrane movements and stabilization, membrane fusion, and regulation of permeability pathways and channel proteins. In the following sections we will describe several proteins that have been identified in parietal cell apical membranes, as they may serve in some of these effector functions.

Actin Actin has a central role in parietal cell function, forming the structural base of microfilaments within the apical microvilli [ 1, 111. Application of cytochalasin B to isolated gastric mucosa grossly disrupts actin microfilaments and inhibits HCl secretion [ l 11. Measurements have shown that parietal cells are rich in filamentous. or F-actin [12]. This is particularly apparent in cytolocalization studies, such as those displayed in Fig. 6(a), where gastric glands have been tagged with fluorescent-labelled phalloidin which binds tightly and specifically to F-actin. For the parietal cells the pattern of staining reveals a definitive interconnecting network throughout the parietal cell, corresponding to the tortuous secretory

canaliculi, and there is also a halo of basolateral staining. In addition, F-actii staining occurs along the entire glandular lumen, including the apical borders of chief and mucous neck cells lining the lumen. F-actin microfilaments remain associated with apical plasma membrane vesicles after they have been isolated from stimulated or resting parietal cells [12]. As F-actin is a major proteic component of parietal cell apical membrane vesicles and could not be released from these vesicles by treatment with salt or 1.5 M guanidine-HC1 [7, 121, we would characterize F-actin as a resident protein of the apical membrane, remaining at that locus as other membrane components are recruited to, or removed from, the apical surface. The absence of actin in isolated tubulovesicles [7, 121, and the ultrastructural localization of F-actin within parietal cells [ l , 111, support this notion. While we describe actin as a resident apical membrane protein we do not imply that it is an intrinsic membrane protein, but that it is associated with the membrane through some linking or tethering protein.

80 and 120 kDa phosphoproteins Several laboratories have pursued the nature of effectorsites forstimulus-secretion coupling by searching for proteins that are phosphorylated in concert with secretagogue activation [ 13-1 61. A number of

22

J. G. FORTE et al.

--

histone

* protomine * 30' I'

30'

histone protamine

I'

30' I'

30' I'

I

52 Pautoradiogrophy

Fig. 6. Simultaneous localization of F-actin (a) and 80 K (b) in the same gastric gland. (a) Fixed gastric glands were probed with FITC-phalloidin to localize F-actin. Actin staining was most intense on the secretory canaliculi of parietal cells and along the entire luminal aspect of the gland. Basolateral aspect of parietal cells also showed some actin-staining. (b) 80 K was localized in the same glands as in (a) using a mouse monoclonal anti-80 K. followed by antimouse-IgC labelled with Texas red. Note that 80 K is restricted to the canalicular structures that also contain F-actin. Each of the two images was produced with the filters most appropriate for the fluorophore.

phosphoproteins have been identified with kinetics of phosphorylation. which suggests correlation with the cAMP/protein A kinase pathway of HCI activation :however, for most of these a cytolocalization or functional role has not been specified. Our laboratory was intrigued to find that there was an increase in the incorporation of 32Pinto two proteins. with molecular weights of 80 and 120 kDa. respectively, in the apical membrane fraction isolated from gastric glands treated with secretagogues that act via the CAMP pathway (e.g. histamine, IBMX, dibutyryl CAMP, forskolin) [ 131. In the case of the 120-kDa phosphoprotein (120 K), we have identified the same protein in the cytosolic fraction from parietal cell-rich homo-

protein stain

Fig. 7. Protein kinase activity of a fraction enriched in 120 K by immunoprecipitation. The supernatant (cytosolic) fraction from a gastric homogenate was reacted with a monoclonal antibody against the 120 K protein. The 120 K-antibody complex was adsorbed to Staph A, washed, and added to the protein kinase assay mixture. (The mixture included 1 mM AT3*P. 5 m~ Mg. 1 mM EGTA, 0.1 mM MT. and either 1 mg ml-' histone or 1 mg d-' protamine as acceptors.) Individual reactions were stopped at 1 min ( 1 ') with 2 % SDS or allowed to react for 30 min ( 3 0 ) at 37 "C before stopping. Samples were then separated by SDS-PAGE. The gel was stained for proteins and autoradiographed. 120 K contained the highest 3zP specific activity. In addition to 120 K. other proteins identified on the gel are immunoglobulin heavy chain (IgC h.c.). light chain (IgC I.c.) and acceptor proteins (acc. prot.).

genates; in fact, the majority of 120 K is in the cytosol [171. Interestingly, the secretagogue-stimulation data show that there is an increase in both '*P and the quantity of 120 K protein in the stimulated apical membrane vesicles, which suggests a redistribution of 120 K from cytosol to apical plasma membrane at the time of stimulation [13]. The association between 120 K and the s.a. vesicles is relatively tight in that 120 K could not be easily removed by sonication or high salt treatment [17]. Recently, we have developed monoclonal antibodies against 120 K for use in cytolocalization and immunopurilication. Immunostaining of gastric glands revealed that 120 K was confined to parietal cells and appeared to be distributed throughout the cytoplasm [ 181.When 120 K was purified by immunoa n i t y chromatography, the adsorbed fraction contained kinase activity which not only phosphorylated substrates, such as histone and protamine, but also phosphorylated 120 K itself, as shown in Fig. 7. Furthermore, phosphorylation of 120 K was markedly

M E M B R A N E A N D PROTEIN RECYCLING

enhanced by the presence of substrate. The kinase activity was dependent on Mg"'. but not on Ca2+ or calmodulin, and was unaffected by inhibitors of protein A kinase. Evidence to date indicates that 120 K is a parietal cell protein kinase which undergoes autophosphorylation [ 181. However, the specification of physiological substrates for the 120 K kinase, as well as its mode of regulation, remain to be established. Unlike 120 K, the 8 0 kDa phosphoprotein (80 K ) can be considered to be a membrane protein in that it co-purifies with, and is restricted to, the membrane fraction harvested as s.a. vesicles. In fact, the apical membrane fraction, whether from resting or stimulated glands, is always rich in 80 K even though the amount of "P incorporated into this protein and the amount of H+K'-ATPase associated with s.a. vesicles are both correlated with stimulation [7, 131. On the other hand, H'K'-ATPase-rich tubulovesicles are virtually devoid of 80 K. These data suggest that 8 0 K is a permanent resident of the apical plasma membrane, and that it is excluded from the endocytic recycling of H+K+-ATPaseinto tubulovesicles when parietal cells return to the resting state. Immunocytolocalization with monoclonal antibodies supports the parietal cell-apical membrane locus for 80 K in both resting and stimulated cells [19]. In Fig. 6(b), gastric glands were probed with an 8 0 K antibody conjugated with Texas Red. Staining was exclusive to parietal cells, and to the network of secretory canaliculi at the apical membrane. This may be compared with F-actin staining in the same gland (Fig. (la) where an identical localization to the secretory canaliculi is seen but, as noted previously, F-actin is also localized in other parietal and nonparietal locations. Extraction of gastric glands in 0.1% NP-40 in low ionic strength medium removed more than 90% of the total protein, including H+K+ATPase. but F-actin microfilaments and 80 K remained associated with the unextracted cellular remnants [20]. The co-localization of actin and 80 K to the apical plasma membrane and their resistance to non-ionic detergent extraction suggested to us some possible role for 8 0 K as an actin-binding protein, possibly even as a link between the apical microfilaments and the cell membrane. In searching the literature for other proteins with properties similar to 80 K, we noted that Anthony Bretscher had identified an 80-kDa microvillar protein which he called ezrin [21. 231. We recently exchanged antibodies and protein antigens with

23

Bretscher's laboratory. Our tests have established that chicken intestinal microvillar ezrin. human placental ezrin, and the 80 K protein from gastric parietal cells are all recognized by the same antibodies, either anti-80 K or anti-ezrin. It is now also apparent that 8 0 K and ezrin are homologous with a phosphoprotein called P-8 1 from A43 1 carcinoma cells [23] and a microvillar protein from choriocarcinoma cells that has been called cytovillin [24]. These variously named phosphoproteins share the same molecular weight and are associated with actin-rich microvilli. Specific homology ha5 been demonstrated for ezrin and P-8 1 by immunocrossreactivity, and interestingly they both show epidermal growth factor (EGF)-stimulated phosphorylation of tyrosine residues [2 31. Furthermore. Bretscher's recent studies of EGF-mediated response in A431 cells have detailed the correlation between the phosphorylation of ezrin and the morphological change underlying membrane ruffling [25]. In a general way this is similar to stimulation-dependent phosphorylation of gastric 8 0 K and the membrane movements associated with secretion. However, there is also one big difference, which concerns the nature of the agonist-mediated phosphorylation. In the gastric system stimulation is mediated by the CAMP-dependent pathway with phosphorylation occurring on a serine residue [7], whereas Bretscher, Hunter and their colleagues observe EGF-stimulated tyrosine phosphorylation [23, 251. There is the possibility that 8 0 K, or ezrin, may be phosphorylated (and regulated?) by both an A kinase and a tyrosine kinase. Parietal cells are known to have EGF receptors and, in fact, acid secretion is inhibited by EGF [26]. It remains for future tests to sort out among the possibilities that regulation of gastric 80 K may be under dual control (e.g. histamine-stimulation, EGFinhibition) via different kinases and phosphorylation sites. Instances of a protein serving as an acceptor for different kinases have been reported for other systems [27, 281. Whether this has a functional significance in the gastric activation/inhibition scheme remains to be established.

The 60-80 kDa glycoprotein: a proposed /%unitfor the H'K+-ATPase Although the 9 4 kDa catalytic subunit of the H+K+ATPase has been characterized extensively, and its primary sequence has been determined, basic questions regarding the structure and composition of

J. G. FORTE et al.

24

kDo

200 116 97

-ATPof

)-a

66

43

31

--

M W I

2

A NP-40

3

4

5

B DTAB

6

.DO

peptide

7

C TFMS

Fig. 8. Characterization of glycoproteins from gastric microsome membranes. Proteins and glycoproteins were separated by SDSPAGE, stained with Coomassie blue: molecular weight standards (MW) are as shown. In the first two experiments microsomes were solubilized in detergent and passed over a wheat germ lectin (WGA) affinity column (adsorbs N-acetylglucosaminerich glycoproteins). Material was separated into a non-binding fraction and a binding fraction. the latter was subsequently eluted with 1.O M N-acetylglucosamine in 0.5 M KCI. Experiment A. Microsomes solubilized in 2% NP-40: lane 1. total material: lane 2. non-binding material: lane 3. WGAbinding material. Note that the H+K+-ATPasewas primarily extracted in the WGA-binding fraction along with the poorly focused glycoprotein. gp 60-80. as well as others of higher molecular weight. Experiment B. Microsomes were s o l u b i l i in 2 % DTAB: lane 4. total material: lane 5. non-binding material: lane 6, WGA-binding material. Note that in this detergent the H'K'-ATF'ase was primarily partitioned to the non-binding fraction while there was apparent enrichment of gp 60-80 and other glycoproteins in the WGA-binding fraction. Experiment C. The WGA-binding material from DTAB-solubilized microsomes (i.e. similar to lane 6) was treated with trifluoromethanesulfonic acid for 2 h under nitrogen in order to effect complete deglycosylation of the glycoprotein. The resulting peptides were run in lane 7 where a 34-kDa core peptide. derived from gp 60-80. is the dominant peptide. Lanes 1-6 have 50 pg protein/lane: lane 7 has 25 pg protein/lane.

the functional holoenzyme remain. Recently we have assembled evidence that the 60-80 kDa tubulovesicular protein (show earlier in an SDS-gel) is a non-covalently associated, heavily glycosylated subunit of the H+K+-ATPase,and is in many ways similar to the P-subunit of the Na+K+-ATPase. The non-covalent interaction between the H+K+ATPase and the 60-80 kDa glycoprotein (gp 60-80) was initially characterized by lectin affinity chromatography and immunoprecipitation of detergentsolubilized tubulovesicular proteins [29]. As shown by the SDS-gels in Fig. 8 (lanes 1. 2 and 3). when tubulovesicular proteins were solubilized in the non-

ionic detergent Nonidet P-40 (NP-40) and chromatographed on a wheat germ lectin affinity column, the glycoproteins were obviously adsorbed, but surprisingly the H+K+-ATPasewas also recovered in the lectin-binding fraction. However, as shown by lanes 4, 5 and 6 of Fig. 8. when solubilization was performed with the cationic detergent dodecyltrimethylammonium bromide (DTAB). the H+K+ATPase was largely recovered in the non-binding fraction, while the glycoproteins were still adsorbed by the lectin. These results are consistent with a noncovalent interaction, stable in NP-40 but labile in DTAB, between the ATPase and a tubulovesicular glycoprotein. Further support for this idea comes from the converse experiment where tubulovesicular proteins were solubilized and then adsorbed with a monoclonal antibody to the H+K+-ATPase[29]. When the solubilizing agent was NP-40, both the ATPase and g p 60-80 appeared in the immunoprecipitate, supporting a non-covalent association exclusively between the ATPase and gp 60-80. On the other hand. when immunoprecipitation was performed on DTABsolubilized proteins, only the ATPase was immunoprecipitated, thus confirming that the association between the H+K+-ATPaseand gp 60-80 is labile in DTAB. Additional tests have revealed that the interactive association between the ATPase and gp 60-80 is also stable in the detergents Triton X-100, octaethyleneglycododecyl ether, and cholate [301. Immunohistochemical localization of g p 60-80 within rabbit gastric mucosal sections and isolated gastric glands, with a monoclonal antibody to g p 60-80. has revealed that the glycoprotein is localized in parietal cells [3 11. Furthermore, the distribution of gp 60-80 within parietal cells closely resembles that of the H+K+-ATPase,including the change in distribution from a more diffuse pattern in resting cells to the apparent condensation to the intracellular canalicular network in stimulated cells (data not shown). Unequivocal evidence to support an identical locus for gp60-80 and H+K+-ATPase awaits immunolocalization at the ultrastructural level. Biochemical analyses of gp 60-80 reveal several similarities between this gastric glycoprotein and the P-subunit of the Na+K+-ATPase[30]. As shown in lane 7 of Fig. 8, chemical deglycosylation of g p 60-80 results in the generation of a 34-kDa core protein. A 34-kDa core protein is also the product of deglycosylation of g p 60-80 by endoglycosidase F.

MEMBRANE AND PROTEIN RECYCLING

Furthermore, analysis of the time course of deglycosylation by Endo-F suggests that gp 60-80 possesses multiple asparagine-linked (or N-linked ') oligosaccharides. Thus, gp60-80 is similar to the 8subunit of the Na+K+-ATPasewith respect to the size of the core peptide and the presence of multiple Nlinked oligosaccharides. Both also appear to be transmembrane proteins. In the light of these findings, we propose that gp60-80 is the 8-subunit of the H'K+-ATPase. Preliminary data suggest that the stoichiometry between the 94-kDa a-subunit and the 8-subunit is 1 : 1 ; however, it is uncertain whether the holoenzyme occurs as a8 or ad2. Available information suggests that the 8-subunit behaves like the catalytic a-unit with respect to turnover during the secretory cycle, i.e. moving from tubulovesicles to apical plasma membrane as pumps are recruited during cell activation. However, at the present time the precise functional role of a /?-subunit for the H+K'-ATPase remains speculative. The heavy glycosylation might imply a very specific gastric function, such as a role in cytoprotection, but it is more likely that its function has more in common with other glycoproteins closely associated with cationtransporting ATPases. Thus, as has been suggested for the 8-subunit of Na+K+-ATPase,gp 60-80 might be essential for cation transport sites, stability of ATPase activity, insertion and maintained orientation of the H+K+-ATPasein the membrane, or targeting the H+K+-ATPaseto the correct cellular membrane. This last possibility poses an intriguing problem in the biology of the oxyntic cell, because the H+K+-ATPaseis sorted to the tubulovesicular membrane compartment (and apical membrane), while the Na+K+-ATPaseis almost always found in the basolateral membrane. Application of molecular biological tools may provide new clues to the function of the 8-subunits associated with cation-transporting ATPases.

K' and C1- transport proteins Pathways for the transport of K+ and CI- have clearly been identified in apical membrane vesicles from stimulated parietal cells. Studies from our own laboratory identified these pathways as conductive channels for the respective ions [9, lo], but additional work has suggested there may also be a nonconductive KCl symport in operation [32-341. Moreover. a number of questions regarding the location and regulation of these transport proteins remain

25

unanswered. Thus, K+ and/or C1- transport proteins may be part of the tubulovesicular system or they may be permanent residents of the apical plasma membrane. Moreover, it is not known whether these transport proteins require modification (e.g. phosphorylation) for activation of the conductive pathways. It is also possible that the K+ and C1- transport proteins reside in a totally separate membrane population in the resting cell, and that these too are recruited into the apical membrane when the parietal cell is activated by intracellular signals. To address these questions fully, it will be essential to identify the transport proteins, localize them within the cell, and reconstitute them into liposomes or lipid bilayers for analysis of transport characteristics. In this regard a n interesting, but very preliminary, observation has recently surfaced. Arthur Finn's group has produced and studied a monoclonal antibody that recognizes a C1- conductance channel identified in Necturus gallbladder and several other cell types [35]. When western transblots of the gastric membrane proteins were probed with this antibody, Finn and Reenstra found a positive reaction with a 2 10-kDa protein, the same molecular weight as the putative C1- channel in the other tissues (personal communication). While this observation certainly needs further verification and study, it is mentioned here as a n approach to identify these important K+ and C1- transport pathways in the parietal cell.

References 1 Forte TM. Machen TE. Forte JG. Ultrastructural changes in oxyntic cells associated with secretory function: a membrane recycling hypothesis. Gastroenterology 1977: 73: 941-55. 2 Forte JG. Black JA. Forte TM. Machen TE. Wolosin JM. Ultrastructural changes related to functional activity in gastric oxyntic cells. Am J Physiol 1981: 241 (Gastroinkst Liver Physiol. 4): G349-58. 3 Wolosin JM. Forte JG. Changes in the membrane environment of the (H+K+)-AWasefollowing stimulation of the gastric oxyntic cell. J Biol Chem 1981 : 256: 3149-52. 4 Hirst BH. Forte JG. Redistribution and characterization of (H'K')-ATPase membranes from resting and stimulatedgastric parietal cells. Biochern J 1985: 231 : 641-68. 5 Forte JG. Ganser AL. Beesley RC. Forte TM. Unique enzymes of purified microsomes from pig fundic mucosa. Gastroenterology 1975: 69: 175-89. 6 Urushidani T. Forte JG. Stimulation-associatedredistribution of H+K+-ATPaseactivity in isolated gastric glands. Am J Physiol 1987: 252 (Gastrointest Liver Physiol. 15): G458-65. 7 Urushidani T. Hanzel DK. Forte JG. Characterization of an 80 kDa phosphoprotein involved in parietal cell stimulation. Am J Physiol 1989: 256 (Gastrointest Liver Physiol. 19): G107O-8 1 .

26

J. G. FORTE et al.

8 Wolosin JM. Forte JG. Functional differences between K+-

ATPase-rich membrane isolated from resting or stimulated rabbit fundic mucosa. FEBS k t t s 1981: 125: 208-12. 9 Wolosin JM. Forte JG. Stimulation of oxyntic cell triggers K' and CI- conductances in apical (H' +K')-ATPase membrane. A m / Physiol 1984: 246 (Cell Physiol. 15): (253745. 1 0 Wolosin JM. Forte JG. K' and CI- conductances in the apical membrane from secreting oxyntic cells are concurrently inhibited by divalent cations. /Membr Biol 1985: 83: 261-72. 1 1 Black JA. Forte TM. Forte JG. The effects of microfilament disrupting agents on HCI secretion and ultrastructure of piglet oxyntic cells. Gastrmnterology 1982 : 83: 595-604. 12 Wolosin JM. Okamoto C. Forte TM. Forte JG. Actin and associated proteins in gastric epithelial cells. Biochern Biophgs Acta 1983; 761: 171-82. 1 3 [Jrushidani T. Hanml DK. Forte JG. Protein phosphorylation associated with stimulation of rabbit gastric glands. Biochem Hiophys Acta 1987: 930: 209-19. 14 Chew CS, Brown MR. Histamine increases phosphorylation of 2 7 and 4 0 kDa parietal cell proteins. A m / Physiol 1987 : 2 5 3 (Gastrointest Liver Physiol. 16): G823-9. 1 5 Cuppoletti J. Malinowska DH. Cytoskeletal determinants of control of gastric acid secretion. Prog Clin Hiol Res 1988: 258: 23-37. I6 Modlin IM. Oddsdottir M. Adrian TE. Zdon MJ. Zuker KA. Goldenring JR. A specific histamine-stimulated phosphoprotein in isolated parietal cells. / Surg Res 1987: 42: 348-5 3. 17 Urushidani T. H a n d DK. Forte JG. A 120 kDa phosphoprotein associated with the apical membrane of stimulated parietal cells. Biophgs / 1988: 53: 524a. 18 Chow DC. Okamoto C. Forte JC. Immunological characterization of a 60-80 kDa glycoprotein associated with the gastric microsomal H.K-ATPase. FASEB / 1989: 3: A873. 19 Hanrel DK. Urushidani T. Usinger WR. Smolka A. Forte JG. Immunological localization of an 80 kDa phosphoprotein to the apical membrane of gastric parietal cells. A m / Physiol 1989: 256 (Gastrointest Liver Physiol. 19): (31082-9. 20 Hanwl DK. Forte JC.Localiration of a parietal cell apical membrane protein to the actin cytoskeleton. / Cell Biol 1989: 107: 638a. 2 1 Bretscher A. Purification of an 80.000-dalton protein that is a component of the isolated microvillus cytoskeleton and its localization in non-muscle cells. / Cell Biol 1983: 97: 425-32. 22 Btetscher A. Purification of the intestinal microvillus

cytoskeletal proteins villin. fimbrin and ezrin. Methods Enzymol 1986: 134: 24-37. 23 Could KL. Cooper JA. Bretscher A. Hunter T. The proteintyrosine kinase substrate, p81. is homologous to a chicken microvillar core protein. / Cell Biol 1986 : 102 : 660-9. 24 Pakkanen R. Hedman K. Turunen 0. Wahlstrom T. Vaheri A. Microvillus-specific M, 7 5.000 plasma membrane protein of human choriocarcinoma cells. / Historhem Cytochem 1987: 3 5 : 809-16. 2 5 Bretscher A. Rapid phosphorylation and reorganiration of ezrin and spectrin accompanying morphological changes induced in A431 cells by epidermal growth factor. / Cell Biol 1989: 108: 921-30. 26 Hatt JF, Hanson PJ. Inhibition of gastric acid secretion by epidermal growth factor: effects on CAMP and on prostaglandin production in rat isolated parietal cells. Hiochem / 1988: 255: 789-94. 27 Gaehlen RL. Allen SM. Krebs EG. Etrect of phosphorylation on the regulatory subunit of the type 1 CAMP-dependent protein kinase. / Biol Chem 1981: 256: 4536-40. 28 Garrison JC. Wagner JD. Glucagon and the Ca"-linked hormones angiotensin 11. norepinephrine, and vasopressin stimulate the phosphorylation of distinct substrates in intact hepatocytes. / Biol Chem 1982: 257: 1 3 1 3 5 4 3 . 29 Karpilow J, Okamoto C. Smolka A. Forte JG. Glycoprotein associated with gastric H.K-ATPase. Fed Proc 1987: 46: 363. 3 0 Okamoto C. Reenstra WW. Li W. Forte JG. Partial characterization of a 60-80 kDa glycoprotein associated with hog gastric microsomal H+K'-ATPase. / Cell Biol 1989: 107: 125a. 31 Chow DC, Okamoto C. Forte JG. Immunological characterization of a 60-80 kDa glycoprotein associated with the gastric microsomal H'K'-ATPase. FASEB / 1989: 3: A873. 32 Wolosin JM, Forte JG. Kinetic properties of the KCI transport of the secreting apical membrane of the oxyntic cell. / Membr Biol 1983: 71 : 195-204. 3 3 Perez A. Blissard D. Sachs G. Hersey S. Evidence for a chloride conductance in the secretory membrane of parietal cells. A m / Physiol 1989: 256 (Gastrointest Liver Physiol. 19): G2 99-30 5. 34 Demarest JR. Loo DDF. Sachs G. Activation of apical chloride channels in the gastric oxyntic cell. Science 1989: 245: 4024. 35 Finn AL. Falk RJ, Tsai LM. Chloride conductance in toad and mudpuppy gallbladders is inhibited by a common monoclonal antibody. FASEB / 1989: 3: A1149.

Correspondence : Professor J. G. Forte. Department of Molecular and Cell Biology. Division of Biophysics and Cell Physiology. University of California. Berkeley, CA 94720. USA.