A New Method for Cell Permeabilization Reveals a Cytosolic Protein Requirement for Ca2+-activated Secretion in GH3 Pituitary Cells*

Ca2+ is a major regulator of exocytosis in secretory cells, however, the biochemical mechanisms underlying regulation remain to be identified. To render the secretory apparatus accessible for biochemical studies, we have developed a cell permeabilization method (cell cracking) which utilizes mechanical shear. GHJ pituitary cells subjected to cracking were permeable to macromolecules but retained a normal cytoplasmic ul- trastructure including secretory granules. Incubation of the permeable cells at 30-37 OC with 0.1-1.0 pM Ca2+ and millimolar MgATP resulted in the release of the secretory proteins, prolactin (PRL) and a proteo- glycan, but not lysosomal enzymes. store

imental evidence available to evaluate their role.
The major obstacles to biochemical studies of regulated secretion are the inaccessibility of the secretory apparatus in intact cells and the failure of cell disruption techniques to adequately preserve the structural integrity required for function. Several techniques which have been employed previously for accessing the intracellular environment of secretory cells have limitations for biochemical studies. Patch clamp electrode recording of capacitance changes provides access only in individual secretory cells (7,8). High voltage fields stably permeabilize cells only to small molecules (9). Permeabilization with detergents (10)(11)(12) or hydrophobic peptides (13) provides macromolecular access but is limited by potential effects on membrane-associated processes. Homogenization techniques are generally too disruptive except for highly specialized cell types (14,15).
The motivation for the present work was to develop a permeabilization method for GHs pituitary cells which would allow biochemical studies of regulated exocytosis. The regulation of prolactin (PRL)' secretion by Ca2+-activated or protein kinase C-mediated pathways in GH, cells has been After washing by centrifugation (800 X g) in a cold isotonic buffer containing 0.1 mM EGTA, cells were resuspended in cold Kglu buffer (0.02 M Hepes, pH 7.2, 0.12 M potassium glutamate, 0.02 M NaC1, 0.005 M glucose, 0.002 M EGTA, 0.1% BSA), a buffer previously found to be optimal for Ca2+-dependent PRL release from electrically permeabilized GH3 cells (18)(19)(20). Permeabilization was achieved by single passage of a suspension (10'-106 cells/ml) of chilled cells through a stainless steel ball homogenizer. The ball homogenizer has been previously described (22) and consisted of a bored (0.3750 inch) chamber into which was fitted a 0.3749-inch tungsten carbide ball, establishing an overall clearance of 0.0001 inch. Cell permeabilization, monitored with 0.4% (w/v) trypan blue and phase contrast optics, was routinely 95-99%. Alternatively, permeable cells were viewed under fluorescence optics in the presence of 20 pg/ml fluoresceinwheat germ agglutinin (Vector Laboratories) as shown in Fig. 1A.
PRL Secretion Assays-Permeabilized cells were used in either of two formats for PRL release experiments. Either cracked cells with residual cytoplasm were added directly to reactions or the cells were washed 3 times by centrifugation (800 X g ) using 20 volumes of Kglu buffer for each wash. In the latter protocol, the washed permeable cells (termed cell ghosts) were resuspended in Kglu buffer and added to reactions. PRL release incubations were conducted at 30 'C for 15 min or 37 'C for 5 min (see Fig. 2) in a total volume of 0.2 ml which consisted of 0.02 M Hepes, pH 7.2, 0.12 M potassium glutamate, 0.02 M NaCl, 0.005 M glucose, 0.1% BSA, 0.002 M EGTA, CaC12 (to yield indicated values of free Ca2+, see Ref. 19), lo6 cracked cells or cell ghosts, 0.002 M MgC12, 0.002 M ATP, and 1-50 pg of cytosol protein or other fractions as indicated. Assays at 30 "C for 15 min were more sensitive to stimulation by low levels of cytosol than those at 37 'C for 5 min. Following incubation, tubes were chilled on ice and contents were transferred to chilled 1-ml polycarbonate tubes for centrifugation at 100,000 X g for 90 min in a Beckman 25 rotor. Supernatants were removed and stored frozen at -20 "C. Cell pellets were solubilized in detergent solution (0.04 M potassium phosphate, pH 7.2,0.15 M NaCl, 0.002 M PMSF, 10 pg/ml leupeptin, 0.5% Nonidet P-40, and 5 mg/ml deoxycholate) by vortexing and sonication; samples were clarified by centrifugation prior to assay. The PRL content of solubilized pellets and supernatants was determined by radioimmunoassay using reagents supplied by the National Hormone and Pituitary Program (University of Maryland School of Medicine) supported by the National Institute of Diabetes and Digestive and Kidney Diseases. PRL was iodinated using lactoperoxidase (Worthington) with NalZ5I (Du Pont-New England Nuclear). IgGsorb (The Enzyme Center) was employed as the second reagent in the immunoassay. The standard curve for the assay exhibited displacement over the range of 0.1-4 ng of PRL (RP-3 standard).
Preparation of Cytosol and Purification of Cytosolic Factor-Cytosol was prepared by homogenization of cells or tissue in ice-cold buffer (0.02 M Hepes, pH 7.5,0.002 M EGTA, 0.001 M EDTA, 0.001 M DTT, 0.0001 M phenylmethylsulfonyl fluoride, 0.5 pg/ml leupeptin). Homogenates were centrifuged at 30,000 X g for 30 min followed by centrifugation at 100,000 X g for 90 min.
Partial purification of the rat brain cytosolic factor was conducted by homogenizing 20 rat brains in 50 ml of buffer. Cytosol was adjusted to 0.006 M CaC12 plus 20 pg/ml leupeptin and loaded onto a 20-ml phenyl-Sepharose column (Pharmacia LKB Biotechnology Inc.) equilibrated with 0.02 M Hepes, pH 7.5,O.OOl M CaC12, 0.001 M DTT, and 0.0004 M phenylmethylsulfonyl fluoride. Loading was conducted by recirculating the cytosol overnight at a flow rate of 1-2 ml/min. The column was washed with 100 ml of equilibration buffer and then with 100 ml of 0.02 M Hepes, pH 7.5, 0.001 M CaC12, 0.001 M DTT, 1.0 M NaCl. Elution was conducted with 0.02 M Hepes, pH 7.5, 0.002 M EGTA, 0.001 M DTT. Fractions of the EGTA eluate containing protein were pooled (37 ml) and reduced in volume to 1 ml using Centriprep-30 concentrators (Amicon Corp.).
Crude rat brain cytosol fractions were analyzed by chromatography on a TSK G3000-SW column (0.75 X 30 cm, Phenomenex) using Kglu buffer lacking BSA. One mg of cytosol protein was injected onto the column at a flow rate of 1 ml/min. Individual fractions (0.5 ml) eluting between the void and salt volumes of the column were concentrated using Centricon-30 devices and tested for activity. Standard proteins used to calibrate the column (see Fig. 9) were from Sigma.
Sulfated proteoglycan release was monitored as previously described (19) except that cells were labeled with 10 pCi/ml "SO?-(Du Pont-New England Nuclear). This assay was conducted with extensively washed GH3 cell ghosts due to the necessity of removing unincorporated 35S radioactivity. Following incubation of labeled ghosts and centrifugation, supernatants were analyzed by determining phosphotungstic acid (0.5%)/trichloroacetic acid (6%)-insoluble material by filtration onto Whatman GFA filters (19 For TEM analysis, cell pellets were washed with buffer, post-fixed in 2% Os04, dehydrated with ethanol, embedded in Durcupan ACM (Fluka Chemical Corp.), sectioned, and stained with uranyl acetate and lead citrate using standard techniques (24). Samples were examined using a Hitachi H-600 microscope. For HVEM analysis, washed cell pellets were post-fixed in 0.1% 0~01, stained with 1% uranyl acetate, dehydrated in ethanol, and critical point dried as described (25). Samples were examined with an AEI EM-7 high voltage electron microscope, operated at 1 MeV. For SEM analysis, fixed cell pellets were washed, dehydrated in ethanol, critical point dried, and sputter-coated with a thin film of platinum as described (26). Samples were examined at low voltage (1-2 kV) with a Hitachi S-900 microscope.
Other Materials-H-7 and W-7 were obtained from Seikagaku America, Inc., trifluoperazine from Smith, Kline and French, calmidazolium from Behring Diagnostics, and ATP analogs from Boehringer Mannheim GmbH. H. Schulman (Stanford University) generously provided purified Ca2+, calmodulin-dependent protein kinase I1 (54); purified calpactin I (46) was kindly provided by D. S. Drust and C. E. Creutz (University of Virginia). Calmodulin was purchased from Sigma, and B. W. Porter (University of Wisconsin) provided purified rat brain protein kinase C (approximately 40% pure).

RESULTS
Permeabilization by Cell Cracking-The permeabilization method described under "Experimental Procedures" allowed the preparation of highly permeable GH3 cells which were structurally well preserved. When GH3 cells were passed a single time through the ball homogenizer, 95-98% of the cells were rendered trypan blue stainable ("cracked" cells). TO further assess the extent of permeabilization, cracked cells were incubated with fluorescein-wheat germ agglutinin (33 kDa). As shown in Fig. L4, this fluorescent lectin bound to the surface of intact cells (left panel) whereas both plasma membrane and nuclear membrane of the cracked cells were labeled (right panel). Hence, cell cracking results in the permeabilization of cells to a 33 kDa probe.
Scanning electron microscopy showed that a large tear in the surface membrane was present in cracked cells (Fig. 1B). Biochemical studies of cracked cells also indicated that there was substantial structural preservation. Markers for the Golgi (UDP-galactosyltransferse), lysosomes (acid phosphatase, N-acetylglucosaminidase), and PRL vesicles/granules (immunoreactive PRL) were each present at a level at least 80% of the total detected in the same number of intact cells (not shown).

Ca2+-activated PRL Release Is Preserved in Cracked Cells-
Incubation of the cracked cells at 30-37 "C resulted in an increase of PRL measured in the high speed supernatant of the reaction mix. The rate and extent of PRL release were enhanced by inclusion of Ca2+ and MgATP in the reaction mixtures (Fig. 2). The maximal extent of PRL release observed under optimal incubation conditions (see below) represented 25 * 6% (mean zk S.D. of six determinations) of the total intracellular PRL pool.
With MgATP present, Ca2+ stimulated the release of PRL 2-10-fold. Free ionic Ca2+ was stimulatory over the range of 0.1-1 PM (Fig. 3A) with half-maximal activation observed at 0.6 PM. A Hill plot of data pooled from several experiments indicated cooperativity of Ca2+ activation with a Hill coefficient equal to 2 (not shown). At Ca2+ concentrations which exceed those of an activated cell (>lo p~) , PRL release was suboptimal (Fig. 3B). In the absence of MgATP, Ca2+ had little influence on PRL release by cracked cells (Table 11). These results demonstrate that cracked cells retain a PRL release mechanism which is similar in its Ca2+ sensitivity to that of intact or electropermeabilized GH3 cells (18-20).
Requirement of a Cytosolic Factor for Ca2+-activated PRL Release-Cracked GH3 were stable to manipulation and could be washed extensively by centrifugation. Washed cracked cells are membranous "ghosts" which are devoid of soluble cytoplasmic factors but which retain secretory granules and a normal cytoplasmic ultrastructure ( A limited tissue survey showed that cytosol fractions prepared from GH, cells, rat liver, rat brain ( Fig. 4B), bovine pituitary, and PC12 cells (not shown) reconstituted PRL release in GH3 cell ghosts. Cytosol from GH3 cells contained contaminating rat PRL which increased nonincubation control values in the assay. Because of its greater activity, abun- dance, and the absence of PRL, rat brain cytosol was used for subsequent characterization studies.
As shown in Fig. 3B, the Ca2+-dependence of PRL release by GH3 cell ghosts shifted to lower Ca2+ concentrations as increasing amounts of rat brain cytosolic protein were used  Following indicated treatments, rat brain cytosol was tested for activity in a PRL release assay utilizing cell ghosts, 2 mM MgATP and either 1 p M or 1 nM ca2+. PRL release observed at 1 nM Ca2+ was subtracted from that at 1 p~ Ca2+ and compared with that of untreated cytosol (100%).
* Cytosol was incubated with papain-agarose (Sigma, activated for 1 h at 37 "C with 0.05 M cysteine) for 15 min at 35 "2 at a ratio of 1 unit of papain/50 mg of protein. Reactions were terminated by chilling and addition of antipain (2 pg/ml) and leupeptin (0.5 pg/ml) followed by centrifugation. An incubation of cytosol without papain addition was conducted in parallel (control).
in the incubations. In addition, Ca2+-independent PRL release (that observed at 1 nM Ca2+) increased with increasing concentrations of cytosolic protein, as was also evident in the experiment of Fig. 4B. It is unclear whether Ca*+-dependent and Ca2+-independent modes of PRL release are mediated by the same or different factors in the cytosol. Subsequent studies focused on the characterization of the cytosolic factor responsible for the Ca2+-dependent mode of PRL release.
The activity of cytosol in supporting Ca2+-dependent PRL release was found to be nondialyzable (not shown), thermolabile (Table I), and susceptible to inactivation by treatment with proteases (Table I) but not with RNase or DNase (not shown), indicating that a protein(s) in the cytosol was responsible for activity. The general presence of protein was not adequate since a variety of purified proteins failed to exhibit similar activity (e.g. see Table V). These data indicate that PRL release from GHs cell ghosts requires a specific cytosolic protein(s) in addition to micromolar Ca2+.
Requirement for Nucleotides in Ca2+-activated PRL Release-Ca'+-activated PRL release by both cracked cells and cell ghosts (supplemented with cytosol) was found to be dependent upon inclusion of MgATP in the reactions (Fig.  5A), with maximal release observed by 1-2 mM MgATP. In contrast, MgGTP, MgUTP, and MgCTP were much less effective (Fig. 5A). A requirement for the y-phosphate of ATP was evident from the findings that MgADP, MgAMP, Mg-AMPPCP (not shown), MgATP[S], and MgAMPPNP (Fig.  5B) failed to substitute for MgATP. Nonhydrolyzable ATP analogs were found to inhibit ATP-dependent PRL release (Fig. 5B) and the inhibition by ATP[S] was not readily reversed (Fig. 5C), similar to results reported for catecholamine release from saponin-permeabilized chromaffin cells (27).
A representative experiment in Table I1 illustrates the interdependence of Ca'+, MgATP, and cytosol factor. Optimal PRL release by GH3 cell ghosts was Ca'+, MgATP, and cytosol dependent.
Evidence for PRL Release by Exocytosis-Several criteria have been utilized to assess whether PRL release by cracked cells occurred by exocytotic discharge. We have found that Ca2+-activated PRL release required structural integrity of the cracked cells, was accompanied by the Ca2+-, MgATP-, and cytosol-dependent release of other secretory proteins, and occurred under conditions which did not result in the release of lysosomal proteins. As shown in Fig. 6, passage of GH3 cells a second or third time through the ball homogenizer resulted in a substantial reduction of Ca2+-activated PRL release in spite of the fact that PRL was retained in sedimentable structures (see Fig. 6, legend). Light microscopic examination indicated that, by a second passage through the ball homogenizer, there was a marked disruption of cellular integrity (not shown). Multiple passes of the cells through the ball homogenizer resulted in the preparation of PRL vesicles/ granules and membranes, however, these did not participate in a Ca2+-activated process which releases PRL.
Intact and electropermeable GH3 cells secrete a sulfated proteoglycan by a Ca2+-regulated mechanism (19). As shown in Table 111, Ca2+-activated, MgATP-and cytosol-dependent release of the proteoglycan by cracked cells was observed.
Acid phosphatase and N-acetylglucosaminidase are not secreted by intact GH3 cells. Less than 2.7 and 4.1%, respectively, of the cellular content of these enzymes was released by cracked cells in incubations in which release of 30% of the PRL content was observed (Fig. 7). These data suggest that PRL release by permeable GH3 cells occurred by Ca2+-regulated exocytotic discharge rather than by a nonspecific lytic mechanism.
Partial Purification of the Cytosolic Factor-Reconstitution of Ca2+-activated PRL release by GH3 cell ghosts served as a quantitative bioassay (Fig. 4 B ) to monitor the purification of the cytosolic protein. Partial purification has been achieved as summarized in Table IV using phenyl-Sepharose, protamine-agarose and Mono Q chromatography. The elution characteristics from Mono Q chromatography are shown in Fig. 8 where Ca2+-dependent PRL releasing activity was found to be eluted between 0.1 and 0.3 M KCI. Overall purification achieved by sequential chromatography with three columns was approximately 40-fold (Table IV). Although activity was very stable in crude cytosol, purified fractions lost activity at 0 or -20 "C. It is likely that the degree of purification has been underestimated due to inactivation. Analysis of the partially purified material by sodium dodecyl sulfate-polyacrylamide gel electrophoresis indicated that multiple proteins were present, precluding a conclusion about the molecular characteristics of the factor at this time.
Molecular sieve chromatography of the rat brain cytosolic factor indicated that the activity displayed molecular weight heterogeneity, eluting more broadly than calibrating proteins. The major peak of activity was observed to elute at the position of a globular protein of about 350 kDa on a calibrated TSK G3000-SW column (Fig. 9). Gel filtration on S300 Sephacryl indicated an apparent molecular mass similar to catalase (210 kDa) with the activity eluting broadly on this column as well (not shown). The basis for the size heterogeneity of the cytosolic factor is not known, and clarification of this property must await full purification.
Attempts to Identify the Cytosolic Factor-A number of Ca2+-dependent enzymes and binding proteins, many of them cytosolic, have been suggested as potential Ca2+ targets involved in the regulation of secretion (see "Discussion"). As     labeling of cells and analysis of proteoglycan release was as described under "Experimental Procedures." The complete incubation was at 30 "C for 15 min and contained 35S04-labeled GH3 cell ghosts, 30 pg of rat brain cytosol, 2 mM MgATP, and 1 p M Caz+. Data shown are the means of duplicate determinations expressed as a percentage of the total secretory proteoglycan pool determined by extraction of cells with 0.1% Triton X-100. summarized in Table V, we have examined the effects of several inhibitors on Ca2+-dependent PRL release and the ability of several purified proteins to substitute for the cytosolic factor. Inhibitors which are known to interfere with hydrophobic interactions (28) required by calmodulin and protein kinase C (trifluoroperazine, calmidazolium) had no  effect of Ca2+-activated PRL release, even when tested at high concentrations. W-7 and pimozide were also without effect (not shown). H-7, an inhibitor of protein kinase C (28), also had no influence. In contrast, neomycin, a drug used to inhibit polyphosphoinositide metabolism (29), fully inhibited Ca2+activated PRL release at low concentrations (Ki = 20 p M ) .
The purified proteins which were tested (calmodulin, Ca2+calmodulin-dependent protein kinase 11, protein kinase C, calpactin I) failed to substitute for the cytosolic factor.

Regulated Exocytosis in Cracked GH3
Cells-A variety of methods have been employed to permeabilize secretory cells. Electropermeabilization is widely employed since it produces plasma membrane pores without damaging intracellular organelles. However, the permeability achieved allows only the exchange of small molecules (9,(18)(19)(20). Methods used for inducing macromolecular lesions have employed agents which either lack a well-defined permeabilization mechanism or which have the potential for affecting intracellular mem-

Fraction Number
FIG. 8. Purification of cytosolic factor by Mono Q chromatography. One mg of cytosolic factor purified to the protamineagarose stage was applied to a Mono Q column at a flow rate of 1 ml/ min. Fractions of 0.5 ml were collected, and a KC1 gradient was used to elute protein as described under "Experimental Procedures." Absorbance at 280 nm was monitored (---) and fractions indicated by the histograms were pooled, concentrated, dialyzed, and tested for Ca2+-dependent activity (PRL release at 1 p~ minus that at 1 nM Ca2+) in the PRL release assay at 30 "C for 15 min using GH3 cell ghosts. One mg of rat brain cytosol was applied to a TSK G3000-SW column at a flow rate of 1 ml/min. One-half-ml fractions were collected, absorbance at 280 nm was monitored (---), and fractions were concentrated for assay of Ca2+-dependent PRL release activity at 30 'C for 15 min with ghosts ( U ) as described under "Experimental Procedures." The elution position of calibration proteins is indicated by the arrows: a, apoferritin (440 kDa); b, or-amylase (200 kDa); c, catalase (210 kDa); d, immunoglobulin IgG (158 kDa); e, bovine serum albumin (68 kDa); f , ovalbumin (43 kDa); g, carbonic anhydrase (30 kDa) and h. cytochrome e (12 kDa).
branes. Digitonin and saponin have been successfully used to permeabilize secretory cells by several workers (10)(11)(12)27) whereas others have found that such detergents are inhibitory to Ca2+-activated exocytosis (9). In order to avoid the use of detergents, which promote PRL release in the absence of Ca2+ and MgATP in GH3 cells,' we sought a nonchemical method for cell permeabilization. Balch and Rothman (30)   *Inhibitor studies were conducted with cytosol present. Ca2+activated PRL release (release at 1 pM Ca" minus release at 1 nM Ca") was assigned a value of 100%.
Studies with purified proteins were conducted with cytosol absent (except where indicated). PRL release at 1 pM Ca2+ in the absence of cytosol was assigned a value of 100%. Four individual experiments are shown where stimulation by cytosol alone varied from 3.27-to 3.82-fold. In each case, a range of protein concentrations was tested; results for the highest concentrations are shown as representative. Protein amounts are indicated for 0.2-ml incubations. Where indicated, 12-0-tetradecanoyl phorbol-13-acetate was included at 100 nM. buffers for the isolation of intact organelles. We have modified the homogenizer (22) and found that a single passage of cells through the device produces a large tear in the plasma membrane (Fig. 1B) without damaging intracellular organelles ( Fig. 1, C and D). Cells cracked open by this method preserve their cytoplasmic ultrastructure and retain functional Ca2+activated exocytotic mechanisms. The conclusion that Ca2+-activated, MgATP-dependent PRL release from cracked cells faithfully reflects the physiological secretory process of intact cells is based on biochemical evidence. The available evidence strongly mitigates against a nonspecific lytic mechanism for PRL release. PRL release by cracked cells was found to be activated by 0.1-1.0 NM Ca2+, precisely the [Ca"] effective with electropermeable GH, cells and measured to be the range of [Ca"] effective in intact cells (18-20). Ca2+-activated PRL release by cracked cells was also MgATP-dependent. This is in accord with the expectation provided for numerous (10,31,32) but not all (14, 15, 33) secretory cells, although the underlying role for MgATP remains to be established. Cracked cells exhibit a specificity for ATP since other nucleotides were much less effective and hydrolysis of the y-phosphate appeared to be obligatory.
Our evidence suggests that ultrastructural integrity is crit-ical for Ca2+-activated, MgATP-dependent PRL release by cracked cells. Disrupting cell structure by passing cells a second time through the ball homogenizer markedly reduced Ca2+-activated PRL release in spite of the fact that PRL granules/vesicles remained intact. This observation indicates that isolated membranous elements, in the absence of adequate structural organization, fail to engage in Ca2+-activated PRL release. Hence, it is very unlikely that Ca2+ activation in this system involves an unmasked nonspecific lytic mechanism. A requirement for structural integrity is expected for vectorial exocytosis.
Additional studies indicated that a second secretory protein, a proteoglycan, was coreleased by cracked cells by a Ca2+activated, MgATP-dependent process requiring a cytosolic factor. Although it has not been determined whether PRL and the proteoglycan are copackaged in granules, parallel release by cracked cells would be expected since parallel regulation of their secretion by intact' and electropermeable (19) GH3 cells has been observed. The observation that PRL and proteoglycans were released by cracked cells whereas lysosomal enzymes were not constitutes additional evidence against a nonspecific lytic release mechanism.
The amount of PRL released by cracked cells has been found to correlate with the size of the releasable PRL pool in intact cells. Maximally, 30% of the PRL content was released by permeable cells during incubation under optimal conditions. Secretagogue treatment of intact cells prior to permeabilization resulted in a corresponding reduction in Ca'+-activated PRL release by permeable cells.' Growth of GH, cells under conditions known to promote increased secretory granule formation (21) also resulted in correspondingly enhanced Ca2+-activated PRL release by permeable cells. 2 Requirement for a Cytosolic Protein in Regulated Secretion-GH3 cells permeabilized by cell cracking retain normal intracellular PRL pools, however, Ca2+-activated PRL release by cytoplasm-free ghosts was found to be completely dependent upon the addition of a protein(s) present in the cytosol fraction of cells. The precise intracellular distribution of this protein(s) in cells is not known, so we cannot exclude the possibility of a membrane association in intact cells (see below) which is disrupted upon permeabilization in chelatorcontaining buffer. The participation of a cytosolic protein in Ca2+-regulated secretion appears to extend to other secretory cells, although the available evidence is limited. Paramecium tetraurelia requires a cytosolic factor which allows regulated trichocyst discharge. This was established by the curing of a mutant nd9 phenotype by microinjection of wild type cytoplasm (34). The cytoplasmic factor was reported to be heat labile and macromolecular and to differ from a protein phosphatase also believed to be involved in the induction of exocytosis (15). Patch clamp recording studies of membrane capacitance changes in individual mast cells (7) or GH3 pituitary cells (35) have also identified a requirement for soluble factors which are "washed out" by intracellular dialysis. Secretagogue-induced capacitance changes in GH3 cells could be prolonged by inclusion of aqueous cellular extracts in the pipette barrel, although the nature of the required soluble material was not determined (35). Peppers and Holz (11) reported .that Ca2+-activated catecholamine release by digitonin-permeabilized PC12 cells deteriorated with preincubation and speculated that this loss of function corresponded to the loss of a required cytosolic factor. Sarafian et al. (12) extended these observations to chromaffin cells and demonstrated that nondialyzable material eluted from digitonin-permeabilized cells could be used to reconstitute Ca2+-activated catecholamine release. These studies of Ca2+-regulated chromaffin granule exocytosis appear to document the role of an uncharacterized cytosolic factor which had not been detected in previous patch clamp studies where intracellular dialysis was performed (8). However, Sarafian et al. (12) noted that the loss of Ca2+-activated secretion from digitonin-permeabilized cells occurred slowly, and they suggested that a required soluble factor may associate with intracellular binding sites which delay its elution upon cell permeabilization. Howell and Gomperts (13) also observed a time-dependent loss of Ca'+-activated histamine release from streptolysin 0-treated mast cells which was slightly delayed in comparison to the loss of cytosolic lactate dehydrogenase. Direct evidence for loss of a cytosolic factor as opposed to rundown of function in streptolysin 0-treated cells was not provided in this study. In contrast to the results of these studies with chromaffin (12) and mast (13) cells, GH, cells exhibit a requirement for a cytosolic protein which is evident immediately upon permeabilization at 0 "C. It may be that the extremely large membrane lesions created by cell cracking allow the rapid exodus of a protein of 200-350 kDa, whereas the smaller lesions of digitoninor streptolysis 0treated cells may retard the loss of this protein.
In another well-characterized system for Ca2+-regulated exocytosis, sea urchin eggs, an obvious requirement for a cytosolic protein is less evident (14). However, Sasaki (36) reported the presence of a KC1-extractable, heat labile 100-kDa protein in egg cortices which was required for cortical granule exocytosis at low [Ca"]. Fractionation studies indicated that most of this material was actually cytosolic. It is unclear whether each of these cited studies has identified a common protein factor similar to that characterized by our study. If so, this protein might well be distributed as a cytosolic as well as a peripheral membrane protein. This suggestion is further supported by our recent finding3 that the cytosolic factor required for PRL release exhibits reversible Ca2+-dependent binding to liposomes. In some systems, the distribution of the factor may allow sufficient membrane association to support exocytosis in the absence of cytosol or cause slow wash-out kinetics in permeable cell or patch clamp studies.
Nature of the Ca" Effector System in Exocytosis-It has long been known that the secretion of many neurotransmitters, hormones, and enzymes requires extracellular Ca2+. These observations led to the suggestion that cytoplasmic [Ca2+] elevations mediate the stimulatory effect of many secretagogues (37). This suggestion has received direct experimental support by measurements of intracellular [Ca"] ( 5 ) and by the demonstration that Ca2+ can activate secretion in permeable cells (9). Nonetheless, the events which are Ca2+ regulated and which trigger exocytosis in secretory cells remain to be identified. The primary motivation for the studies reported here was the development of a permeable cell model in which the macromolecular requirements for Ca2+-activated secretion could be examined.
Our results with permeable GH3 cells have a direct bearing on several of the proposed mechanisms for Ca2+-regulated exocytosis. We have been unable to demonstrate inhibition of Ca2+-activated PRL release using high concentrations of calmodulin inhibitors. Direct addition of purified calmodulin or calmodulin-dependent protein kinase I1 was unable to substitute for the cytosolic factor in our system. These results argue against a major role for calmodulin in Ca2+-activated PRL release and indicate that the cytosolic factor is not likely to be a calmodulin-regulated enzyme.
Several results also argue against a role for protein kinase C as the Ca2+ effector system involved in Ca2+-activated PRL release. H-7, a protein kinase C inhibitor (28), had no effect on PRL release supported by the cytosolic factor, and purified protein kinase C was unable to substitute for the cytosolic factor. In spite of the conclusion that protein kinase C may not itself constitute the Ca2+-dependent effector, several studies document that protein kinase C activation can stimulate PRL secretion in GH3 cells (16,20). Possibly, as suggested by Baker and Knight (50) for chromaffin cells, protein kinase C activation can sensitize the Ca2+-dependent effector system to low Ca2+ concentrations. Our preliminary results indicate that PRL release from cracked cells can also be stimulated by protein kinase C activators at low Ca2+ concentrations.
Neomycin was found to be a very potent inhibitor of Ca2+activated PRL release. Neomycin inhibition has previously been reported for a large number of secretory systems (41,51,52). It is known that neomycin binds phosphatidylinositol 4,5-bisphosphate and inhibits the metabolism of this lipid by several enzymes, including phospholipase C (29). Hence, neomycin inhibition might indicate the involvement of phospholipase C in Ca2+-activated PRL release (42). The cytosolic factor described in this study exhibits chromatographic properties similar to at least one of the brain cytosolic isoenzymes (53). However, the cationic nature of neomycin and the possibility of inhibitory actions not involving phospholipase C preclude definitive interpretation at this time.
One of the steps of cytosol factor purification involves Ca2+dependent hydrophobic interaction chromatography on phenyl-Sepharose. As detailed el~ewhere,~ interaction of the factor with this hydrophobic matrix at low ionic strength was Ca2+-dependent, a property which implies that the cytosolic factor is itself a Ca2+-dependent enzyme or binding protein.
In addition, the cytosolic factor was shown to interact with liposomes in a Ca2+-dependent manner." Hence, the cytosolic factor may be similar to one of the recently characterized Ca2+-dependent membrane-binding proteins (annexins, [44][45][46][47]. However, direct tests have shown that calpactin I is incapable of substituting for the cytosolic factor. In addition, the native molecular mass of the cytosolic activity (200-350 kDa) is substantially greater than those reported for many of these Ca2+-binding proteins (3, 5, 6, 44-47).
At present, neither the subunit molecular weight nor the identity of the cytosolic factor with a previously identified protein are known. Future efforts to fully purify this Ca2+dependent cytosolic factor should indicate the nature of Ca2+dependent events associated with regulated secretion. Rk and P. Cooke for HVEM studies and by W. V. Welshons and J.