Perturbation of cellular calcium blocks exit of secretory proteins from the rough endoplasmic reticulum.

In the cultured human hepatoma HepG2, Ca2+ ionophores block secretion of different secretary proteins to different extents, alpha 1-antitrypsin secretion being more sensitive to A23187 and ionomycin than is alpha 1-antichymotrypsin, and albumin secretion the least of the three proteins studied. As judged by subcellular fractionation experiments and by treatment of pulse chase labeled protein with endoglycosidase H, A23187 and ionomycin cause newly made secretory proteins to remain within the rough endoplasmic reticulum (ER). Experiments in which A23187 is added at different times during a pulse or chase show that secretion of newly made alpha 1-antitrypsin becomes resistant to the ionophore, on average, 15 min after synthesis; this is about 20 min before it reaches the trans-Golgi, and while it is still within the rough ER. We speculate that a high concentration of Ca2+ within the ER may be essential for certain secretory proteins to fold properly, that folding is inhibited when ER Ca2+ levels are lowered by ionophore treatment, and that unfolded proteins, particularly alpha 1-antitrypsin, cannot exit the rough ER. Treatment of murine 3T3 fibroblasts or human hepatoma HepG2 cells with the Ca2+ ionophores A23187 or ionomycin also induces a severalfold accumulation of the ER lumenal protein Bip (Grp78). These findings disagree with a recent report that Ca2+ ionophores cause secretion of Bip and other resident ER proteins, but is consistent with other reports that A23187 causes accumulation of mRNAs for Bip and other ER lumenal proteins.

The rough endoplasmic reticulum is the site of synthesis of plasma, Golgi, and lysosome membrane proteins, as well as of secretory and lysosomal proteins and resident lumenal ER' proteins. These proteins leave the ER from transitional elements that bud into transition vesicles. These, in turn, generate an intermediate compartment that mediates transport into the Golgi, from which ER-synthesized proteins are sorted to their final destination.
Exit of proteins from the ER is a step at which maturation of secretory and membrane proteins is subject to multiple types of regulation (reviewed by Pfeffer and Rothman, 1987;Lodish, 1988;Pelham, 1989;Hurtley and Helenius, 1989). Folding and assembly of secretory and membrane proteins occurs in the ER and is essential for efficient transport; unfolded or unassembled polypeptides are retained because they form aggregates unable to enter transport vesicles or because they bind to resident ER proteins such as Bip (Grp78). Production of unglycosylated or misfolded secretory or plasma membrane proteins frequently results in induction of Bip synthesis (Kozutsumi et al., 1988). Unassembled subunits of oligomeric membrane proteins frequently are degraded in a pre-Golgi compartment (Lippincott-Schwartz et al., 1988;Chen et al., 1988;Amara et al., 1989).
On the other hand, certain normal ER proteins can be induced to leave the ER by deletion of specific amino acid residues (Munro and Pelham, 1987;Poruchynsky and Atkinson, 1988;Paabo et al., 1987;Nilsson et al., 1989). These and other results indicate that there can be "bulk flow" of proteins from the ER to the Golgi (Wieland et al., 1987) and that certain ER proteins have one or more ER "retention signals." ER lumenal proteins such as Bip have a carboxyl-terminal sequence, KDEL required for their retention in the ER. Evidence suggests that KDEL-containing proteins are not permanently anchored to an immobile ER membrane receptor or lumenal matrix; rather, these proteins might normally "escape" into ER-to-Go& transport vesicles and then be selectively returned to the ER (reviewed by Pelham, 1989).
Furthermore, at least in human and mouse hepatoma cells and in the exocrine pancreas, newly made secretory proteins are exocytosed at very different rates; the rate-limiting and distinctive step in intracellular maturation for each protein is its transport from the rough ER (Strous and Lodish, 1980;Lodish et al., 1983;Scheele and Tartakoff, 1985;Ledford and Davis, 1983;Fries et al., 1984). Many integral plasma membrane and viral surface glycoproteins mature to the cell surface at different rates and, again, the distinctive and limiting step is exit from the rough ER (Fitting and Kabat, 1982;Williams et al., 1985). Whether these "natural" differences are due to protein folding, selective retention in the ER, or selective binding to transport receptors, is not known.
A rise in cytosolic Ca*+ frequently is the trigger for exocytosis: fusion of regulated secretory vesicles with the plasma membrane. Less is known of the role of Ca2+ in earlier steps of the secretory pathway. The endoplasmic reticulum lumen is one of the major stores of cellular Ca'+. Long-term exposure of cells to Ca*+ ionophores result in overexpression of several lumenal ER proteins, including Bip and Grp94 (Wu et al., 1981;Welch et al., 1983;Lee, 1987;Watowich and Morimoto, 1988;Mater and Koch, 1988). A recent report shows that treatment of murine 3T3 cells for 4-8 h with calcium ionophores results in efficient secretion of several ER proteins, including Grp94, Bip, and protein disulfide isomerase (Booth and Koch, 1989). Results presented here disagree with these latter findings; we observe, both in 3T3 and in human hepatoma HepG2 cells, an increase in accumulation of Bip within 4-8 h. We also show that Ca*+ ionophores block secretion of several hepatoma proteins and cause accumulation of newly made polypeptides within the endoplasmic reticulum. We speculate that lumenal Gas+ may be important in folding certain secretory proteins and that folding is disrupted when ER Ca*+ levels are lowered by ionophores.

EXPERIMENTAL PROCEDURES
Materials-Endoglycosidase H (endo H) was obtained from Genzyme Corp. (Boston, MA). Rabbit antisera directed toward human albumin, transferrin, al-antitrypsin, and ai-antichymotrypsin were purchased from Accurate Chemicals (Westburv. NY). Fixed Stanhviococcus aureus cells were purchased from the New England Enzyme Center, Inc. (Boston, MA), and ["'Slmethionine was from the Radiochemical Center, Amersham Corp. A23187 and ionomycin were purchased from Calbiochem. Growth and Labeling of HepG2 Cells-Culture flasks of 25 cm2 area were seeded with 1 x lo6 HepG2 cells and incubated in minimal essential medium with 10% fetal calf serum (Schwartz et al., 1981). Cultures were generally fed on the 2nd and 4th days after seeding and used on the 5th, at which time the cells had approximately quadrupled. Similar to our previous study (Lodish and Kong, 1984), the cells were washed once in methionine-free medium, placed in 2.0 ml of methionine-free medium (containing 10% dialyzed fetal calf serum), and incubated at 32 "C. After 10 min, 20 ~1 of dimethyl sulfoxide containing the desired amount of ionophore was added. After an additional 10 min, 40 NCi of [?S]methionine was added, and the culture was incubated for a further 10 min (pulse). The plates were washed once with chase medium, and 2 ml of chase medium (growth medium plus 10% fetal calf serum and 1 mM extra methionine) was added to each. After incubation at 32 "C for the appropriate time (chase), the cells were placed at O-4 "C, washed 3 times in complete phosphate-buffered saline, and dissolved in 2.0 ml Trisbuffered saline (Owen et al., 1980) containing 1% sodium deoxycholate and 1% Nonidet P-40 (lysis buffer). Both the cell lysate and medium were clarified by centrifugation at top speed in a microcentrifuge for 10 min at 4 "C.
Zmmunoprecipitation-To 200 ~1 of lysate or medium was added 5 ~1 of normal rabbit serum and 300 /Al oflysis buffer. After incubation at 4 "C for 1 h, the mixture was me-cleared bv addition of 50 ul of a 10% suspension of washed S. a&ecu cells (Owen et al., 1980): After centrifugation to remove the bacteria, 3 ~1 of specific antiserum was added. After a 2-h to overnight incubation at 4 "C, 50 ~1 of a 10% suspension of S. aureus was again added. After a l-h incubation at 4 "C, the immunoprecipitate was recovered by centrifugation in a microcentrifuge, washed twice in buffer 2 (0.142 M NaCl, 0.24 M KCl, 0.008 M Na2P0,, 0.5% sodium deoxycholate, 1% Triton X-100), resuspended in 50 ~1 of gel sample buffer containing SDS, and boiled. Analysis was by electrophoresis through 10% polyacrylamide gels.
Controls established that these conditions were sufficient to recover over 85% of the labeled protein in question. In particular, less than 10% additional labeled protein was recovered if the first supernatant from the immunoprecipitation was reacted with additional antiserum. In some cases, the immunoprecipitate was split into two portions; one was digested with endo H and the other mock-digested, as detailed previously (Zilberstein et al., 1980). Homogenization and Gradient Analysis-These procedures were similar to those described in detail in a previous paper (Lodish et al., 1987). Culture dishes of 100 mm diameter were labeled and washed as described in this reference; all subsequent procedures were conducted at O-4 "C. The pooled cell pellets from two dishes were resuspended in 2.2 ml of homogenization buffer (0.01 M Hepes, pH 7.4, 0.25 M sucrose) and allowed to swell for 10 min. They were homogenized with 15 strokes of a tight-fitting Dounce homogenizer, resulting in over 85% cell breakage. Nuclei were removed by centrifugation at 1,000 rpm for 10 min. Of the postnuclear supernatant, 2.0 ml was layered on a gradient for the SW 41 Beckman rotor, consisting of 1 ml of each of the following sucrose solutions (w/v) all in D20 containing 10 mM Hepes,pH 7.4: 10,14,18,22,26,30,'35,40,45,and 50%. After centrifugation at 4 "C for 3 h at 36,000 rpm, the gradient was fractionated into I7 0.75-ml fractions using an ISCO gradient collector. Of each fraction 150 ~1 was used for immunoprecipitation with each antiserum, as described above.

Preparation
of Antipeptide Antibodies to Grp78-Polyclonal rabbit antibodies were made against the amino-and carboxyl-terminal peptides of the rat ~72 protein (Munro and Pelham, 1986); the rat ~72 shares identity with Grp78. The amino-terminal peptide KKEDVGTVVGIDLGC and the carboxyl-terminal peptide CPPPTGEEDTSEKDEL were synthesized by Dr. Peter Kim, of the Whitehead Institute, Cambridge, MA, using solid phase synthesis (Barany and Merrifield, 1979). A cysteine residue was added to the carboxyl terminus of the amino-terminal peptide and to the amino terminus of the carboxyl-terminal peptide to facilitate coupling. The peptides were coupled to the carrier protein, keyhole limpet hemocyanin, using the chemical cross-linker succinimidyl 4-(N-maleimidomethyl)cyclohexane-I-carboxylate (Pierce Chemical Co.) following the procedure of Green et al. (1982). Rabbits were immunized with the peptide conjugates in Freund's complete adjuvant and boosted every 2-3 weeks for up to 6 months. The antisera was checked for the ability to immunoprecipitate labeled cellular proteins (data not shown) as well as to identify proteins by Western blotting; only Western blotting generated positive Bip signals.

Quantification of Bip by Western
Blotting-Fresh growth medium (1.5 ml) was added to HepG2 cells growing on 60-mm dishes, together with the desired amount of A23187. After subsequent incubations at 37 "C, the medium was recovered, the cells were dissolved in 1.5 ml of lysis buffer (see above), and the lysate and medium centrifuged for 10 min at 4 "C in a microcentrifuge. Lysate and medium were then concentrated by a modification of the procedure of Wessel and Fhigge (1984). Briefly, to 150 ~1 of sample was added 600 ~1 of methanol followed by vigorous mixing. Then, 150 ~1 of chloroform was added with mixing, followed by 450 J of water. Following mixing and lowspeed centrifugation, the upper (aqueous) layer was discarded. To the interphase and lower phase was added 1.0 ml of methanol. After mixing, the precipitated proteins were recovered by centrifugation, washed once with methanol, dried under vacuum, and dissolved in 35 ~1 of SDS gel sample buffer. Analysis was by SDS-PAGE through 10% polyacrylamide gels and electrophoretic transfer to nitrocellulose filters.

RESULTS
The Ca2+ Ionophores A23187 and Ionomycin Inhibit Secretion of HepG2 Proteins to Different Extents-In most of our experiments we utilized the human hepatoma cell HepG2, which synthesizes and secretes a number of well characterized serum proteins: albumin, a nonglycosylated protein, and the glycoproteins transferrin, al-antitrypsin, and al-antichymotrypsin (Knowles et al., 1980). Previously, we showed that newly made albumin and cY,-antitrypsin are secreted rapidly (tllz = 45 min at 32 "C), while 50% of ai-antichymotrypsin is released in 90 min and 50% of transferrin only at 140 min. All of these proteins require the same time (20 min) for passage through the Golgi apparatus and exocytosis (Lodish et al., 1983). Virtually all of the cell-associated transferrin and al-antichymotrypsin contain one or more high mannose asparagine-linked oligosaccharides.
Detailed studies on oligosaccharides on intracellular transferrin showed that they are the forms (GlciManaGlcNAcz, ManaGlcNAcz, MansGlcNAcz, and ManTGlcNAcs) characteristic of proteins localized to the rough ER (Lodish et al., 1983(Lodish et al., , 1987. At a concentration of 5 pM, the calcium ionophore A23187 inhibits secretion of all newly made HepG2 secretory proteins, but different proteins are affected differently (Fig. 1). Secretion of albumis is delayed about 25 min, and by 90 min of chase albumin secretion is reduced only about 20-25% (Figs. 1 and 2). Secretion of oi-antitrypsin, in contrast, is blocked 67-72% during that period, and that of cY,-antichymotrypsin is reduced to an intermediate value of 50-60% ( Figs. 1 and 2). The extent of inhibition increases with the concentration of A23187; secretion of albumin is affected less than that of Lu,-antitrypsin or aiantichymotrypsin at all concentration of A23187 (Fig. 2). The Ca2+ ionophore ionomycin at 10 FM caused an inhibition of protein secretion similar to that obtained with 5 FM of A23187 (Fig. 3) the presence of 10 PM ionomycin, secretion of albumin is inhibited 27%, while that of al-antichymotrypsin and alantitrypsin is inhibited 53 and 56%, respectively.
In these experiments serum was included in both the pulse and chase medium; however, results similar to these in Figs. l-3 were obtained in serum-free media (data not shown). Importantly, in serum-free medium the effects of A23187 and ionomycin were only slightly sensitive to a reduction in the concentration of extracellular Ca*+ over the range 2.6-0.1 mM (Table I). Importantly, the lower the extracellular Ca'+, the greater the inhibition of cY1-antitrypsin secretion caused by A23187 or ionomycin. This result suggests that it is the drop in concentration of Ca*+ in an intracellular store such as the ER that is important in inhibiting protein secretion, rather

ER-Since the secreted forms of both al-antitrypsin and al-antichymotrypsin
have complex Nlinked oligosaccharides, properties of the intracellular forms of these proteins that accumulate in the presence of A23187 or ionomycin can be inferred from their gel mobilities in SDS-PAGE and their sensitivities to endo H (Figs. 4 and 5). The forms of al-antitrypsin and cui-antichymotrypsin secreted by control cells are resistant to endo H (Fig. 4, A and B, lanes 1  and 2; Fig. 5, A and B, lanes 1 and 2); they migrate more slowly than do the endo H-sensitive ER-localized precursor forms (Fig. 4, A and B, lane 7; Fig. 5, A and B, lane 5). Importantly, the bulk of cell-associated cu,-antitrypsin and cylantichymotrypsin that accumulates in the presence of A23187 migrates identically to the normal ER precursor (compare Fig. 4, A and B, lanes 9 and 11 to lane 7) and are similarly sensitive to endo H (lanes 10 and 12). Since the enzyme that confers resistance to endo H, N-acetylglucosaminyltransferase I, is localized to the cis-or medial-Go&, these results establish that maturation of secretory glycoproteins is blocked by A23187 prior to this compartment. Similar results (Fig. 5, were pulse-labeled with [ "Slmethionine and chased for 90 min at 32 "C m the presence of zero A23187 (lanes 2, 2, 7, and 8), 2 pM A23187 (lanes 3, 4, 9, and IO), and 5 PM A23187 (lanes 5, 6, II, and 12). Aliquots of the medium (lanes Z-6) and cell lysate (lanes 7-12) were immunoprecipitated with antisera to a,-antitrypsin (A) or <r,-antichymotrypsin (B). One-half of the immunoprecipitates were incubated with endo H (lanes 2, 4, 6, 8, 10, and 12) and the others without.
Shown are the radioautographs of the dried gels; nonadjacent lanes of the same exposure of the same gel were reassembled for photography. The experiment was performed as in Fig.  4, except that 10 pM ionomycin was added 10 min before the pulse and was present during the chase. Lanes I and 2, immunoprecipitates of medium from control cells chased for 210 min; lanes 3 and 4, medium from cells chased 210 min in presence of ionomycin; lanes 5 and 6, lysates from control cells chased 30 min; lanes 7 and 8, lysates from cells chased 120 min in the presence of ionomycin. Lanes 2, 4, 6, and 8, immunoprecipitates digested with endo H; lanes I, 3, 5, and 7, mock-digested.
A and B, lanes 7 and 8) were obtained with the ""S-labeled CQ-antitrypsin and al-antichymotrypsin that accumulate in cells treated with 10 PM ionomycin.
Note that the small amount of a,-antitrypsin and al-antichymotrypsin secreted in the presence of A23187 is resistant to endo H digestion but migrates faster than does the normal secreted product (Fig. 4, A and B, compare lanes 3-6 to lane 1). This is due mainly to a decrease in the numbers of sialic acid residues added in the presence of A23187, since treatment with Arthrobacter ureafaciens neuraminidase increases mobility of the forms secreted in the presence and absence of A23187 to the same value (data not shown). Thus, besides blocking the initial stages of maturation of secretory proteins, A23187 also affects the trans-Golgi vesicles or trans-Golgi reticulum, where terminal sialic acid residues are added (see Kornfeld and Kornfeld, 1985). In contrast, the reduced amounts of a,-antitrypsin and Qantichymotrypsin secreted in the presence of ionomycin have the same gel mobilities as do the normal secreted products (Fig. 5, A and B, compare lanes 3 and 4 to 1 and 2). Thus, ionomycin does not affect addition of terminal sialic acids, though it, like A23187, blocks the initial stages of maturation of secretory proteins.
That A23187 causes newly made albumin and cY1-antitrypsin to accumulate in the rough ER is shown by density gradient analysis of subcellular fractions (Fig. 6). In these studies, HepG2 cells were labeled with [""Slmethionine for 10 min and chased for 60 min (at 32 "C). A postnuclear supernatant was layered atop a shallow sucrose gradient on a 50% sucrose cushion, all in D20, and centrifuged for 3 h. Radiolabeled al-antitrypsin, transferrin (Fig. 6, c and d) and albumin (Fig. 6, a and b) were immunoprecipitated from each fraction. The most dense fractions (12-18) derive from the rough ER, fractions 3-6 contain ER-to-Golgi transport vesicles, and fractions 4-9 the Golgi itself (Lodish et al., 1987). Material in the top fractions (l-3) has leaked out of vesicles and is always less than 10% of each radiolabeled secretory protein analyzed. By 60 min of chase in control cells ( ously, virtually all cell-associated transferrin co-fractionates with the rough ER (Fig. 6, c and d; see Lodish and Kong, 1984;Lodish et al., 1987). In control cells little radiolabeled albumin (Fig. 6~) remains in the rough ER, the little radiolabeled protein not yet secreted co-fractionates with the Golgi complex. When cells are incubated with A23187, by contrast, the additional albumin that remains cell-associated co-fractionates with the rough ER (Fig. 6b, fractions 12-18). In control cells little radiolabeled oll-antitrypsin remains after a 60-min chase (Fig. 6~). Strikingly, in cells treated with A23187 virtually all of the radiolabeled a,-antitrypsin remains cellassociated (Fig. 2) and virtually all of it co-fractionates with the rough ER in fractions 13-17 (Fig. 6d).
Confirmatory evidence exists that the ol,-antitrypsin ( Fig.  6d) and albumin (Fig. 6b) in the dense gradient fractions 13-17 indeed are in the rough ER, not in lysosomes or other dense organelles. Treatment of cell lysates with 15 mM EDTA and 100 pg/ml RNase for 5 min at 0 "C destroys ER-attached ribosomes and causes the ER to band at ZO-25% sucrose in DzO, equivalent to fractions 10-12; the density of lysosomes (30~50% sucrose; fractions 13-18) is not affected by this treatment (data not shown). Treatment of lysates from A23187-incubated cells with EDTA/RNase causes a shift in radiolabeled cY,-antitrypsin, albumin, and transferrin from fractions 12-17 to fractions lo-12 (data not shown). Collectively, the data in Figs. 4 and 6 thus show that A23187 causes an inhibition of maturation of secretory proteins at the level of the rough ER. Either the secretory proteins never enter ER-to-Golgi transport vesicles (fractions 4-7) or, if they do, the proteins are rapidly returned to the rough ER.
Ca" Ionophores Block Secretion of al-Antitrypsin Only if Added During or Immediately After Synthesis-In the previous experiments, Ca*+ ionophores were present both during the period of protein synthesis (pulse) and the subsequent chase. Since, at the concentrations we use, these ionophores should alter cellular Ca2+ concentrations within seconds, one can ask how soon after their synthesis secretory proteins acquire resistance to inhibition of secretion by Ca*+ ionophores. The study in Fig. 7 -antitrypsin (m). In the other part (dotted lines) 5 gM A23187 was added at the time indicated on the abscissa and kept in until 60 min of chase. The fraction of pulse-labeled albumin (0) and oil-antitrypsin (0) secreted at this time was quantified, as was the fraction of al-antitrypsin, secreted and intracellular, in the mature form (m). of oli-antitrypsin by A23187 requires that it be present during or very soon after synthesis, while the protein is still in the rough ER.
Without ionophore addition (solid lines) one-half of the newly made cu,-antitrypsin acquires resistance to endo H, i.e. has moved into or through the Golgi, in 34 min, and one-half is secreted in 46 min (at 32 "C). Secretion of one-half of the newly made albumin requires 50 min. These numbers are very similar to those in our earlier studies (Lodish et al., 1983).
Secretion of albumin is, as noted above, relatively insensitive to A23187. Fig. 7 (dotted line) shows that the extent of albumin secretion (at 60 min) is the same whether A23187 is added during the pulse or at various times during the chase.
In contrast, addition of 5 pM A23187 during the pulse inhibits over 90% the secretion of ai-antitrypsin, and the extent of inhibition is reduced the later the ionophore is added. Quantitatively, the ability of half of the cu,-antitrypsin molecules to "mature" to the Golgi-processed form becomes resistant to A23187 addition 9 min after synthesis (measured from the end of the pulse). Secretion of one-half of the newly made (piantitrypsin became resistant to A23187 addition 13 min after synthesis. By 10 min of chase, none of the newly made cyiantitrypsin has yet reached the Golgi, as judged by acquisition of complex, sialic acid-containing oligosaccharides ( Fig. 7; see also Lodish et al., 1987). We conclude that newly made alantitrypsin acquires resistance to inhibition of secretion by Ca" ionophores while it is in a pre-Golgi organelle, presumably while it is still in the rough ER. As noted under "Discussion," this is most simply explained by a requirement of Ca*+ for folding of ai-antitrypsin while it is in the rough ER. A.23187 Causes Accumulation, Not Secretion, of Bip (Grp78)-The original impetus for these studies was the report of Rudolph et al. (1989) that yeast strains deleted for a plasma membrane Ca*+ ATPase secrete mammalian secretory proteins and mutant yeast secretory proteins that are normally retained in the rough ER. We hypothesized that mammalian cells treated with A23187 which, like the yeast mutation, causes an elevation in cytosolic Ca*+, might cause secretion of Bip or other normal ER proteins. In contrast, our studies indicated that there is little to no secretion of Bip, but rather an increase in the amount of cellular Bip protein.
To detect Bip (Grp78), we generated antisera against the chemically synthesized 15-amino acid amino-terminal peptide and the 16-amino acid carboxyl-terminal peptide, using the sequence of the rat protein derived from that of the cDNA (Munro and Pelham, 1986). On Western blots of HepG2 (Fig.  8) and 3T3 (Fig. 9) cell proteins both antisera recognized a major polypeptide of apparent M, 78,000 that clearly is Bip. The antibody against the carboxyl terminus also recognized a protein of apparent M, 95,000 that most likely is Grp94 (endoplasminf, since the two proteins share the same carboxyl-terminal four amino acids. As expected, the immunoreactive M, 68,000 and 95,000 species co-fractionated on an equilibrium density gradient with the rough ER (data not shown). Fig. 8 shows the amount of cell-associated Bip as a function of incubation of HepGP cells with 5 pM A23187. As judged by scanning of the autoradiographs, the amount after 8 h (lane 7) was 1.4 times that of control cultures similarly incubated (average of three experiments).
After 16 h (lane 8) the amount was about 1.6-fold that of control cultures. We do not know why the anti-amino-terminal serum recognizes two closely spaced Bip bands; in A23187-treated cells it is the intensity of the slowest-migrating species that is induced the most. Both of these species, as with the species recognized by the anti-carboxyl-terminal antibody, co-fractionate with the rough ER on equilibrium density gradients (not shown). Fig. 9 shows a similar induction of Bip levels obtained with 3T3 murine fibroblasts. After 8 h incubation in the presence of 5 pM A23187, cells accumulated 1.4 times as much Bip as did control cultures (average of three experiments).
In several experiments we looked for Bip in the medium using Western blotting; in these studies we incubated 3T3 or HepG2 cells for up to 8 h in serum-free medium in order to reduce the background from serum proteins. Never did we see more than 15% of Bip in the medium. When we did see Bip in the medium, it was correlated with the presence of lo-20% (or more) of the cells that could not exclude trypan blue and were probably killed by prolonged incubation with A23187 in serum-free medium. We conclude that A23187 does not induce Bip secretion, as has been claimed in another report (Booth and Koch, 1989). Rather, A23187 causes accumulation of Bip in the rough ER, consistent with accumulation in the rough ER of newly made secretory proteins.

DISCUSSION
Our most important result is that the calcium ionophores A23187 and ionomycin block movement of hepatoma secretory proteins from the rough ER. Inhibition is selective, in that secretion of albumin is only marginally reduced, about 20% after a 90-min chase, while secretion of cY1-antitrypsin is inhibited 67-72% over the same period (Figs. l-3). That secretion is slowed or inhibited at a pre-Golgi level is shown by two types of studies. First, equilibrium density gradient analyses show that the bulk of the newly made albumin, CQantitrypsin (Fig. 6), and a,-antichymotrypsin (data not shown) in ionophore-treated cells accumulates in vesicles that have a density equivalent to that of the rough ER, and whose density decreases when RNA and ribosomes are destroyed by prior incubation of the homogenate with ribonuclease. Second, the glycoproteins cY,-antitrypsin and ocl-antichymotrypsin accumulate in a form that is sensitive to endo H, and that co-migrates with the normal ER precursor that contains highmannose oligosaccharides (Figs. 4 and 5). Our results do not eliminate the possibility that, in the presence of ionomycin or A23187, secretory proteins exit the ER into a pre-Golgi "intermediate" or "salvage" compartment (Pelham, 1989), but such proteins must then be returned at high efficiency to the rough ER.
The rates of later stages of protein maturation, movement of secretory proteins through the Golgi complex and exocytosis, is much less affected by A23187 or ionomycin. For both newly made a,-antitrypsin and a,-antichymotrypsin, little cell-associated protein accumulates that has the mobility of the sialic acid-containing secreted material (Fig. 4, A and B, lanes 9 and 11; Fig. 5, A and B, lanes 7). A kinetic analysis of the accumulation of this material in a pulse-chase experiment (analyses similar to those in Lodish and Kong, 1984) indicates that, in the presence of A23187, these endo H-resistant species require 20-25 min to be secreted from the cell, a value similar to that in control cells obtained in the present work and also in previous studies (data not shown; see Lodish et al., 1983). Similarly, during a chase in the presence of A23187 little pulse-labeled al-antitrypsin or cY,-antichymotrypsin accumulates with the gel mobility of the rough ER precursor but that is resistant to endo H. Such forms are characteristic of secretory glycoproteins that have passed through the medial but not the trans-Golgi (Lodish et al., 1987;Kornfeld and Kornfeld, 1985). This is additional evidence that once secretory proteins leave the rough ER, whether in the presence or absence of Ca2+ ionophores, they are rapidly matured through the Golgi and exocytosed.
Clearly A23187 does affect terminal stages of maturation of N-linked oligosaccharides, since al-antitrypsin and cylantichymotrypsin secreted in the presence of the ionophore migrate faster than the form made in its absence (Fig. 4, A and B, lanes 1, 3, and 5). The difference is due largely to incomplete addition of sialic acid, since treatment with neuraminidase causes all secreted forms to acquire the same faster gel mobility. Possibly A23187 blocks synthesis of CMP sialic acid or its transport into the trans-Golgi (reviewed by Hirschberg and Snider, 1987); possibly the ionophore alters the structure of these terminal Golgi cisternae such that sialyl transferase cannot get access to secretory glycoproteins.
In contrast, ionomycin does not affect addition of sialic acid to secreted glycoproteins, as judged by the unaltered gel mobility of cY,-antitrypsin and a,-antichymotrypsin (Fig. 5, A and B). Thus, the alterations of terminal glycosylation induced by A23187 are probably not caused by the altered Ca2+ levels but are due to some other effect of the ionophore itself.
It is not surprising that movement from the rough ER of cY,-antitrypsin is much more sensitive to A23187 and ionomycin than is exit of the other secretory proteins studied. Exit of a,-antitrypsin from the ER is blocked by tunicamycin, an inhibitor of formation of N-linked oligosaccharides, or by drugs that block removal of the terminal glucose residues from the oligosaccharide after its transfer to protein (Lodish and Kong, 1984). Several single amino acid changes in CQ-antitrypsin also block its exit from the ER, and the secreted protein is folded into a conformation deficient in binding to trypsin (reviewed by Curie1 et al., 1990). If one assumes that a native or "native-like" conformation of cr,-antitrypsin is essential for it to enter ER-to-Golgi transport vesicles, then each of the above treatments could alter protein conformation in a manner incompatible with ER exit. Similarly, a reduction in Ca2+ concentration within the ER, engendered by Ca*+ ionophores, might slow or block "sensitive" proteins such as a,-antitrypsin from assuming a native conformation. Although there are other possibilities, Ca2+ ionophores probably block exit of proteins from the ER by reducing the Ca*+ content within the ER. One reason for thinking SO is that synthesis of the lumenal ER protein Bip (also known as Grp78) is induced by a variety of conditions which have in common the production of unfolded or unassembled proteins with the ER: glucose deprivation or treatment of cells with glycosylation inhibitors; anoxia; overproduction of abnormally folded recombinant proteins; or even overexpression of more-or-less normally folded glycoproteins within the ER, such as is induced by infection with many lipid-enveloped viruses (Lee 1987;Kozutsumi et al., 1988;Watowich and Morimoto, 1988;Shiu et al., 1977;Sciandra et al., 1984). Ca'+ ionophores also induce expression of Bip (Figs. 8 and 9;Welch et al., 1983;Wu et al., 1981), and it is reasonable to assume that they also do so by virtue of causing ER accumulation of unfolded or misfolded proteins. We have tried, unsuccessfully, to detect by co-immunoprecipitation an interaction between Bip and the ocl-antitrypsin and al-antichymotrypsin that accumulate in the ER in the presence of A23187; such negative results, of course, do not disprove that such an interaction occurs in vivo.
Both A23187 (Fig. 7) and ionomycin (data not shown) block a very early stage in the secretion of cY1-antitrypsin. Secretion of one-half of the newly made al-antitrypsin becomes resistant to ionophore addition about 10 min after synthesis. At this time, all of the newly made cyi-antitrypsin is sensitive to endo H ( Fig. 7 and Lodish et al., 1983Lodish et al., , 1987 and, as judged by sucrose density gradient analysis, virtually all is in the rough ER (Lodish et al., 1987). The simplest interpretation of these data is that Ca*+ is required for proper folding, a step that occurs within lo-13 min after synthesis, and that this folding is essential for subsequent movement of the protein to and through the Golgi, a process that requires an additional 25 min (Fig. 7).
Our results showing Bip accumulation and the absence of Bip secretion in the presence of A23187 or ionomycin disagree with those recently reported by Booth and Koch (1989). We have no explanation for the difference, other than to point out that these workers found that only in 3T3 cells did Ca2+ ionophores induce secretion of Bip. In our strain of 3T3 cells A23187 had virtually the same effect on Bip accumulation as in HepG2 cells, a 1.4-1.6-fold increase in Bip levels after 8-16 h. We saw GO% of Bip secreted into the medium, and we suppose that the secretion of Bip and other KDEL-containing ER proteins observed by Booth and Koch must be confined to only a few cell lines. Certainly enhanced accumulation of Bip within the ER, induced by Ca2+ ionophores, is consistent with ER accumulation of (presumably unfolded or misfolded) secretory proteins caused by these ionophores, particularly if Bip indeed is bound to one or more of these polypeptides.
It would be of interest to measure the free Ca*+ concentration within the ER after addition of A23187 to the cells, but currently this is not possible. More importantly, a study of the role of Ca2+ ions in cell-free folding reactions of denatured secretory proteins, such as ol,-antitrypsin, could be revealing.