Cyclic AMP modulates the rate of 'constitutive ... - Semantic Scholar

1931

Journal of Cell Science 108, 1931-1943 (1995) Printed in Great Britain © The Company of Biologists Limited 1995

Cyclic AMP modulates the rate of ‘constitutive’ exocytosis of apical membrane proteins in Madin-Darby canine kidney cells Mirtha Brignoni1, Omar P. Pignataro2, Marcelo L. Rodriguez1, Adriana Alvarez1, Dora E. Vega-Salas4, Enrique Rodriguez-Boulan3 and Pedro J. I. Salas4,* 1Instituto de Investigaciones Bioquímicas L.F. Leloir, Fundación Campomar, Buenos Aires, Argentina 2Instituto de Biología y Medicina Experimental, Buenos Aires, Argentina 3Cornell University Medical College, Department of Cell Biology and Anatomy, New York, USA 4University of Miami School of Medicine, Department of Cell Biology and Anatomy, PO Box 016960, Miami,

FL 33101, USA

*Author for correspondence

SUMMARY Madin-Darby canine kidney and other epithelial cell lines (e.g. Caco-2, MCF-10A and MCF-7) develop intracellular vacuoles composed of apical membrane displaying microvilli (VACs) when impaired from forming normal cell-to-cell contacts. In a previous publication, we showed that VACs are rapidly exocytosed upon treatment with 8Br-3′,5′-cyclic adenosine monophosphate (8-Br-cAMP), a membrane-permeable analog of cAMP, and that this exocytosis correlates with variations in the cellular cAMP concentration in response to the cell-cell contacts. In the present work, we tested the hypothesis that cAMP may be a positive modulator of the ‘constitutive’ exocytic pathway. To mimic conditions in cells with incomplete intercellular contacts, the intracellular levels of cAMP were decreased by means of two independent approaches: (i) pores were induced in the plasma membrane with the polypeptidic antibiotic subtilin, thus allowing small molecules (including cAMP) to permeate and move out of the cytoplasm; and (ii)

INTRODUCTION It has been proposed that the apical domain of the plasma membrane in epithelial cells could be considered as a specialized ‘organelle’ because it comprises many of the proteins responsible for specific epithelial functions such as luminal digestive enzymes and specific transport systems. The basolateral domain, conversely, shares many of the components common to non-polarized cells involved in maintenance, regulatory and nutritional functions (Rodriguez-Boulan and Powell, 1992; Simons and Wandinger-Ness, 1990). Therefore, conceivably, epithelial tissues may be able to modulate the amount of apical membrane protein and, in general, apical membrane surface according to their differentiation or the degree of confluency. It is generally accepted that cell-to-cell contacts provide essential information for epithelial morphogenesis (for a review see Schmidt et al., 1993), and that the microenvironment determines gene expression and differentiation in epi-thelial cells

adenylate cyclase and protein kinase A were blocked with specific inhibitors. In all cases, the intracellular levels of cAMP were measured and, in porated cells, equilibrated to simulate the corresponding physiological intracellular concentrations. The decrease in cAMP within the physiological range resulted in a decreased rate of transport of an apical marker of the constitutive pathway (influenza virus hemagglutinin) from the trans-Golgi network to the apical plasma membrane. Likewise, the delivery of a number of cellular apical proteins to the plasma membrane was retarded at low cAMP concentrations. The inhibitors of adenylate cyclase failed to block basolateral delivery of vesicular stomatitis virus G protein. This differential modulatory effect may represent a differentiation-dependent control of the insertion of apical membrane in epithelial cells. Key words: epithelial polarity, apical exocytosis, cAMP

(Howlett and Bissell, 1993). Epithelial Madin-Darby canine kidney cells can be prevented from forming intercellular contacts by either incubation in low calcium medium (Gonzalez-Mariscal et al., 1985), or seeding very sparse cultures in medium with normal calcium concentrations. The resulting cells lack normal cell-cell contacts and develop large (1-5 µm caliper diameter) intracellular vacuoles, which we termed Vacuolar Apical Compartments (VACs) (Vega-Salas et al., 1987b). VACs contain microvilli and apical membrane markers, but exclude basolateral markers (Vega-Salas et al., 1987a). They are surrounded by a cortical cytoskeleton of actin, similar to that underlying the apical membrane and, unlike simple phagocytic vacuoles, VACs can be rapidly exocytosed upon restoration of cell-cell contacts. Then, their membrane follows an elaborate pathway of insertion to be finally located in the apical domain (Vega-Salas et al., 1988). VACs are not restricted to MDCK cells. They have been observed in other cell lines (Gilbert and Rodriguez-Boulan, 1991; Vega-Salas et al., 1993), and they are morphologically

1932 M. Brignoni and others identical to the ‘intracellular lumens’ described in a number of carcinoma cells in vivo (Remy, 1986). This similarity was recently highlighted by experiments indicating that breast carcinoma cells in tissue culture display VACs even when they develop apparently normal intercellular contacts, while their non-tumorigenic counterparts behave like MDCK cells; that is, they present VACs only when the formation of cell-cell contacts is impaired (Vega-Salas et al., 1993). Therefore, we formed the hypothesis that VACs may be reporting on a control mechanism that, under the signalling of cell-cell contacts, modulates the polarized insertion of apical membrane components on the cell surface. Under this model, carrier vesicles transporting apical proteins from the trans-Golgi network (TGN) to the plasma membrane (Orci et al., 1987; RodriguezBoulan and Nelson, 1989; Simons and Fuller, 1985; Simons and Wandinger-Ness, 1990) may be prevented from reaching their target, and then fuse to each other to form VACs in isolated cells. To test this hypothesis, we searched for a second messenger involved in the transduction of a signal originating at intercellular contacts. Total and protein-bound cAMP were found to increase twofold when intercellular contacts were restored, which is the same extracellular signal that triggers VAC exocytosis. Moreover, 8-Br-cAMP, a membranepermeable analog of cAMP mimicked the effect of cell-cell contacts, thus inducing the exocytosis of VACs (Brignoni et al., 1993). Traditionally, exocytosis has been divided into ‘constitutive’ and ‘regulated’ pathways (Burgess and Kelly, 1987). The effects of extracellular signals - often hormones - on the latter are frequently mediated by cAMP (for examples, see BaldysWaligorka et al., 1987; Macrae et al., 1990; Miyata et al., 1990; Takuma, 1990). The constitutive exocytic pathway, on the other hand, is thought not to be regulated. However, the recent discovery of G-proteins associated with this pathway suggests that it may be regulated in some unknown manner (Bourne, 1988; Ferro-Novick and Novick, 1993; Pimplikar and Simons, 1993). That cAMP is involved in the regulation of vesicular traffic has been shown in epithelial intestinal cells (Lencer et al., 1992) and in melanophores (Sammak et al., 1992). It is also known that cAMP mediates cellular responses to cell-cell contacts (such as contact inhibition and cell growth), although the nature of the receptors is still elusive (Abell and Monahan, 1973; Froehlich and Rachmeler, 1974; Matsukawa and Bertram, 1988; Oey et al., 1974; Otten et al., 1971; Tung and Fritz, 1987). Recently, it has been shown that cAMP regulates fluid secretion in kidney cysts (Neufeld et al., 1992). This body of evidence strongly encouraged us to pursue the idea that VACs may result from a down-regulation of the constitutive exocytic pathway. Our general approach was to experimentally manipulate the cAMP cellular concentrations, keeping them within the physiological range, in cells with normal cell-cell contacts (confluent and in the presence of normal extracellular calcium concentrations). Because cells that display VACs showed a reduction in cAMP levels, we decreased cAMP concentration in these cells (which of course lack VACs) to the same extent. Two independent approaches were used: (i) the cells were porated with a polypeptidic antibiotic subtilin, which forms small pores in the plasma membrane and allows the equilibration of cytoplasmic molecules of Mr ~1,000 or less with the extracellular medium; and (ii) inhibitors of adenylate cyclase and protein kinase A (PKA; Taylor et al., 1992) were

used in non-porated cells. In all cases the resulting levels of cAMP were monitored and the reduction resulted in a marked inhibiton of the apical constitutive exocytic pathway. MATERIALS AND METHODS Reagents Inhibitor of protein kinase A: adenosine-3′,5′-cyclic monophosphorothioate, Rp-isomer (Rp-cAMPS), sodium salt (Biolog-Life Science Institute D-2800 Bremen, Germany). Inhibitors of adenylate cyclase: 2′,3′-isopropylidene adenosine (IA; Aldrich Chemical Co., Milwaukee, WI); 9-β-D-arabinofuranosyladenine (AFA; Sigma Chemical Company, St Louis, MO). Anti-cAMP antibody (Chemicon International, Inc., Temecula, CA). Cell culture and infections Madin-Darby canine kidney (MDCK) cells (ATCC CCL 34), from passages 58-83 were cultured as described elsewhere (Vega-Salas et al., 1987a). Cells were infected with influenza virus (WSN strain) or with vesicular stomatitis virus (ts045 mutant, Indiana strain) were carried out as described before (Rodríguez-Boulan, 1983b), except that a multiplicity of infection (MOI) of 1 was used for immunofluorescence experiments to employ non-infected cells as an internal negative control. In all other experiments, MOIs were 10. The hybridoma producing a monoclonal antibody against influenza hemagglutinin H15-IRI was kindly donated by Dr Walter Gerhardt, Wistar Institute, PA. For experiments, MDCK cells were seeded on translucent filters separating two chambers (Transwell-Clear, Costar, Cambridge, MA), glass coverslips or plastic multiwell strip plates. Immunofluorescence and RIA (radioimmunoassay) The techniques for immunofluorescence and RIA, described in detail in a previous publication (Salas et al., 1986), were followed, with the exception that MDCK cells were grown on 96-well strip plates (cat. no. 9102, Costar, Cambridge, MA) for RIA experiments, and each point was taken by quadruplicate. Subtilin preparation and use The protocol for subtilin preparation and purification was slightly modified from the method of Sahl and Brandis (1981) for polypeptidic antibiotics from Streptococcus. Briefly, Bacillus subtillis (ATCC 6633) was grown in mineral medium in logarithmic phase at 37°C for 72 hours, until the culture reached an optical absorbance (at 600 nm) of 3.0. Then, 500 ml of the culture supernatant were passed through a Servachrome XAD4 (Rohm and Haas, Philadelphia, PA) (2.5 cm × 23 cm, exclusion volume 50 ml). The column was washed with culture medium and then with 150 ml of medium supplemented with 50% (v/v) methanol. Subtilin was eluted with 50 ml 70% (v/v) methanol in 1 mM sodium acetate buffer, pH 2.0. Methanol was evaporated and the precipitated protein was removed by centrifugation at 12,000 g for 20 minutes. The pH of the translucent supernatant was then adjusted to 6.0 with 200 mM K2HPO4. The resulting solution (~100 ml) was passed through a column (1 cm × 25 cm) of CM-Sephadex C-25 (Pharmacia, Piscataway, NJ) previously equilibrated with 200 mM potassium phosphate buffer, pH 5.8. The column was washed with 10 exclusion volumes of the same buffer and then eluted with a linear gradient of the same buffer with concentrations ranging from 0.2 to 1.4 M. Eighty 1.5 ml fractions were collected and assayed for subtilin activity (see next paragraph). The active fractions (usually fractions 24-30) were pooled and slowly supplemented with solid (NH4)2SO4 up to 90% saturation, under gentle agitation on ice. This was spun at 12,000 g for 15 minutes at 4°C. The pellet was re-dissolved in 3 ml 100 mM KCl, 100 mM citrate/phosphate buffer, pH 4.5, and passed through a column (1 cm × 25 cm) of Sephadex G-50 pre-equilibrated with the same buffer. Fractions of 1 ml were taken and assayed for

cAMP modulates constitutive exocytosis 1933 subtilin activity. The active fractions were pooled and de-salted in a Sephadex G-25 column (50 ml) pre-equilibrated with 20 mM citrate/phosphate buffer, pH 4.5. The fractions from this column were tested again for subtilin activity. The final step of purification was performed by reverse-phase HPLC, in a C-18 column with a 0 to 60% acetonitrile gradient in 0.2% trifluoroacetic acid. The fractions were dried in a Speed Vac (Savant Instruments Inc., Farmingdale NY), resuspended in PBS and assayed for subtilin activity. Subtilin activity assay Staphylococcus aureus 29 CCM-A-748-ATCC25923 was grown in nutrient broth (3 g Bacto beef extract; 5 g Bacto peptone; 1 l water). In a 96-well dish, we seeded in triplicate: 120 µl of fresh nutrient broth, 30 µl of a S. aureus culture in logarithmic phase, and either 30 µl PBS (positive control) or 30 µl of a fraction to be assayed for subtilin activity. The dish was incubated for 24 hours at 37°C, then the turbidity was read by eye or the absorbance was determined at 600 nm. Application of subtilin to MDCK cells The rationale for the method of use of subtilin was based on two premises: (i) pore formation requires a trans-membrane potential difference of −80 to −100 mV (Schuller et al., 1989); and (ii) cells require buffer that mimics intracellular conditions once the pores open to maintain exocytic functions (Gravota et al., 1990). Thus, to induce pore formation, confluent MDCK cell monolayers grown in DME (Dulbecco’s modified minimal essential medium, normal calcium concentration), 5% horse serum, were washed twice in saline phosphate buffer, and incubated in intracellular buffer (which mimicks the ionic composition of cytoplasm; ICB: 6 mM CaCl2, 0.32 µM Fe(NO3)3, 5.4 mM KCl, 0.8 mM MgSO4, 44 mM NaHCO3, 0.9 mM NaH2PO4, 5.5 mM D-glucose, 15 mg/l Phenol Red, 1 mM sodium pyruvate, 10 mM EGTA, 1 mM DME/amino acids, DME/vitamins, 1 mM ATP, 0.2 mM GTP, 0.05 mM UTP, 0.05 mM UDP-N-acetylglucosamine, pH 7.4) supplemented with 140 mM choline chloride, to induce membrane hyperpolarization, and 1/10 (v/v) purified subtilin extract in PBS. The cells were kept in this medium for 30 minutes and then shifted to ICB supplemented with 140 mM KCl instead of choline chloride (ICB-K). The efficiency of pore formation was assayed by incubation in FITC-phalloidin (Mr 1,177), Trypan Blue (Mr 960), and Lucifer Yellow CH (Mr 454). The first two stained subtilin-treated cells faintly (not shown), while the latter gave a clear fluorescence (see Fig. 4). Thus, it was concluded that subtilin was inducing pores with a cut off of ~1,000 daltons, which closely correlates with the pore size determined by Schuller and coworkers (1989). Trypan Blue staining in subtilin-treated cells faded within 30-45 minutes, presumably because of diffusion of the stain back to the extracellular compartment. However, because it was convenient to read the results under the inverted microscope, it was routinely used as a positive control for all experiments involving subtilin poration. To further test that cellular proteins were not leaking out of the cytoplasm of subtilin porated cells, MDCK monolayers were labelled with [35S]methionine for 60 minutes at 37°C, chased in DME for 30 minutes, and incubated with subtilin for 30 minutes. Then, the monolayers were incubated in ICB-K for various times at 37°C. The cell supernatants were precipitated with 10% trichloroacetic acid, and the cpm in the precipitates were measured. Typically the cpm released to the medium were ~0.6% of the cpm in the cellular pellet. The cpm released to the medium from control monolayers incubated in ICB-K without subtilin treatment exceeded those of porated cells by a factor of 1.8, suggesting that subtilin may inhibit normal MDCK cell secretion (Gottlieb et al., 1986), and that no significant release of cytoplasmic proteins was taking place. Determinations of cAMP Total intracellular cAMP and protein-bound cAMP were measured by

radioimmunoassay as described elsewhere (Brignoni et al., 1993; Pignataro and Ascoli, 1990). Metabolic labelling The cells were starved in methionine/cysteine-free MEM (Gibco) for 20 minutes. For surface biotinylation, the cells were then labelled with 0.1 mCi/ml [35S]methionine (NEN Research products, Du Pont, Wilmington, DE) in 0.05 mM cold methionine-containing medium. For surface immunoprecipitations, we used the same label added with 0.1 mCi/ml [35S]cysteine (NEN Research products) in the presence of 0.05 mM cold cysteine-containing medium. Surface biotinylation MDCK cells (2.5×106) were grown at confluency in DME on polycarbonate filters (Transwell) separating two chambers under standard culture conditions for 24 hours. Under these conditions the cells acquire full apical (but not basolateral) polarization (Vega-Salas et al., 1987b). The monolayers were washed three times in saline buffer. Then, the cells were metabolically labelled for 10 minutes at 37°C and for 110 additional minutes at 19°C. Finally, the monolayers were chased in DME (ICB-K for porated cells) for 30 minutes at 37°C. If drug treatments were used, the inhibitors of adenylate cyclase or PKA were added 30 minutes before the shift to 37°C (i.e. during the last 30 minutes of the pulse), and the cells were kept under the same treatment during the chase. The monolayers to be permeabilized with subtilin received the antibiotic only from the basolateral side at the end of the pulse. In this case, poration was performed in the cold for 30 minutes (to allow pore formation and cAMP diffusion) before shifting the cells to 37°C. The last step consisted of biotinylization of the apical components as described elsewhere (Rodriguez-Boulan et al., 1989; Sargiacomo et al., 1989). Briefly, the filters were cooled to 0°C on ice and washed for 30 minutes in PBS supplemented with 0.1 mM CaCl2 and 1 mM MgCl2, and with 1% BSA only on the basolateral side. Then, sulfo-NHS-biotin (Pierce, Rockford, IL) was added to the apical chamber (from a 10 mg/ml stock solution in DMSO) to a final concentration of 0.5 mg/ml and incubated for 30 additional minutes. Finally, the monolayers were washed three times in ice-cold PBS, and extracted in 1% Triton X-100 in 100 mM KCl, 2 mM MgCl2, 2 mM EGTA, 60 mM PIPES, pH 6.9, and 1 mM PMSF, with the basolateral chamber left empty. The supernatant was collected, spun in the cold in an Eppendorf microcentrifuge for 5 minutes, and the supernatant was subjected to affinity purification with streptavidin-agarose beads (Pierce) overnight. The beads were then washed twice in ice-cold PBS, 1% Triton X-100, twice in the same buffer supplemented with 0.5 M NaCl and two more times in PBS alone. Then, the beads were eluted in SDS, 15 mM β-mercaptoethanol and 8 M urea sample buffer, and the eluates were run in a 7% to 14% acrylamide gradient gel. The gels were fixed, soaked with a scintillation enhancer, dried and exposed for autoradiography. Whenever necessary, absorbance was determined in autoradiograms with a laser scanning densitometer (Personal Densitometer, Molecular Dynamics, Sunnyvale, CA). When biotinylation was performed on cells porated with subtilin, the possibility arose that sulfo-NHS-biotin might enter the cytoplasm via the subtilin pores. Therefore, in those cases, subtilin was added from the basolateral side and biotin from the apical side. Even though it is known that some biotin may permeate through the tight junctions, it was expected that the basolateral excess of BSA would quench the leak. To further control the possibility that biotin might enter the cells via the tight junctions and the basolateral subtilin pores the following experiments were performed. Confluent MDCK monolayers grown on polycarbonate filters were porated (or not) with subtilin and apically biotinylated. Then the cells were extracted in Triton X-100, and the extracts were blotted and analyzed with 125I-streptavidin. The pattern of biotinylated proteins did not differ at all in porated cells from that in cells that had not received subtilin. If any biotin had entered the

1934 M. Brignoni and others cells, a number of cytoplasmic proteins should have been biotinylated, which was not the case. Surface immunoprecipitation The cells were grown on Transwell-Clear filters at full confluency for 3 days and infected with ts045 VSV (vesicular stomatitis virus, Indiana strain) mutant from the apical and basolateral sides. Then the cells were kept at 37°C for 2.5 hours, labelled with [35S]methionine and [35S]cysteine, for 30 minutes at 19°C, and chased at 19°C for 1 additional hour. After 30 minutes of the chase period, inhibitors of adenylate cyclase were added. Finally, to allow exocytosis, the monolayers were transferred to 37°C for 30 minutes in DME. The cells were chilled to 0°C, incubated for 15 minutes in ice-cold PBS supplemented with 1% globulin-free BSA (Sigma) added from both the apical and basolateral sides, and incubated with a monoclonal antibody against the ectodomain of VSV G protein from the basal side for 2.5 hours (PBS-BSA was kept on the apical side continuously) with vigorous agitation. Then, the monolayers were washed for an equal period in ice-cold PBS, and extracted for 10 minutes in PBS supplemented with 0.5% sodium deoxycholate, 0.1% Triton X-100, 1 mM EGTA, 1% globulin-free BSA, 1 mM AEBSF (anti-protease), 10 µM E-64 (anti-protease), 2 µg/ml aprotinin, and 1 µM pepstatin (all anti-proteases were from Calbiochem, La Jolla, CA). This extract was spun at 3,000 g for 10 minutes to remove debris and further incubated overnight at 4°C with agarose beads covalently coupled to affinitypurified goat anti-mouse Ig antibodies. Finally, the beads were washed 4 times in the same extraction buffer, and once in the same buffer supplemented with 0.6 mM NaCl. The beads were eluted with SDS sample buffer containing 8 M urea and 0.1% SDS, and the eluates were analyzed by SDS-PAGE and autoradiography.

RESULTS The exocytosis of influenza hemagglutinin is delayed in cells kept in low calcium medium The appearance of VACs in cells with defective intercellular contacts takes several hours (Vega-Salas et al., 1987a, 1988). Transport from the trans-Golgi network (TGN) to the cell surface, on the other hand, requires only a few minutes (Griffiths et al., 1985). Thus, we were unable to design experiments to study the latter using the origin of VACs as a reporting phenomenon. Therefore, the first step to test the hypothesis was to measure the rate of delivery of an apical membrane marker accumulated in the TGN at 19°C to the apical plasma membrane by switching the temperature to 37°C, an approach extensively used to study Golgi-plasma membrane exocytosis (Griffiths et al., 1985). Because large amounts of viral glycoprotein are accumulated in infected cells in contrast to small amounts of cellular proteins, we chose influenza virus glycoprotein hemagglutinin (HA) as a marker apical protein (Rodríguez-Boulan, 1983a) that follows the constitutive exocytic pathway (Orci et al., 1987). To detect HA, we used H15-IRI, a monoclonal antibody that recognizes the protein only during and after passage through the Golgi apparatus (Rodríguez-Boulan et al., 1984). In MDCK cells kept in low calcium medium (S-MEM; spinner MEM, ~5 µM calcium), only a modest amount of fluorescence was observed on the cell surface after a 30 minute incubation at 37°C following a 19°C accumulation in the TGN (Fig. 1B versus A, cells kept at 19°C). In contrast, cells grown in normal calcium medium displayed bright fluorescence signals on the cell surface after the same periods at 19/37°C

(Fig. 1D versus C). These experiments were done using relatively low rates of infection (MOI=1) to show the infected cells in contrast over a background of non-infected cells. (Parallel samples were permeabilized with Triton X-100, and showed 40-50% infected cells.) Although these experiments were nonquantitative, they showed a clear difference in the exocytic rate between cells kept in normal calcium and cells grown in low calcium medium. The relative amounts of HA in intracellular compartments versus the apical surface were semi-quantitatively compared by radioimmunoassay. In Triton-permeabilized monolayers, the antibodies had accessibility to both surface and intracellular compartments (Fig. 2, Total), while in non-permeabilized cells only the epitopes on the cell surface were identified (Fig. 2, Surface). The same experimental design as for Fig. 1 was used, except that HA was determined by RIA as a function of time after temperature switch to 37°C. In monolayers kept in DME (Fig. 2A), a difference between Total and Surface samples at time 0 demonstrated the intracellular accumulation of HA. It can be assumed that this is mostly TGN accumulation for the following reasons: (i) the antibody recognizes HA in or after the Golgi step; (ii) even though some leakage to the cell surface is known to occur, these experiments were performed at relatively early times after infection, when small amounts of HA can reach the surface and, therefore, even smaller amounts can recycle back into the cell; (iii) at 19°C the exocytic pathway is blocked at the exit of the TGN; and (iv) parallel immunofluorescence images showed all infected cells displaying a peri- or supranuclear intracellular structure with location and shape compatible with those of the Golgi (e.g. Fig. 3A). Most of the epitopes in this intracellular compartment, however, became exposed to the cell surface within the first 30 minutes after switching the cells to 37°C. The amount of HA on the cell surface significantly increased (P