Amplification of the Golgi complex in MDCK cells secreting human growth hormone

In 1898 Camillo Golgi first described a novel cellular component that has come to be known as the Golgi apparatus, or complex. In the typical somatic cell of a multicellular animal the anatomy of this intracellular network has been defined by electron microscopy to consist of stacks, or pro-files, of flattened saccules (cisternae), also known as dictyosomes, which appear to be interconnected into a functional Golgi apparatus (see Mollenhauer and Morré, 1991 for review). The stacks lie in the pericentriolar, juxtanuclear region of the cell. Both biochemical and structural studies have demonstrated that each profile is a highly ordered and polarized structure consisting of cis, medial, and trans faces. Although the number of profiles per-cell in eukaryotes may vary from none in certain fungi (Bracker, 1967) to over 25,000 in some algae (Sievers, 1965), the average number in a typical animal cell is probably fewer than 10 interconnected stacks (Lucocq and Warren, 1987; Novikoff et al., 1971; Rambourg et al., 1981), which seem, within a given cell, to function synchronously (Mollenhauer and Morré, 1991).

Also residing in the perinuclear region of animal cells are tubular/vesicular networks associated with the trans pole of each profile. These have been implicated in membrane sorting and recycling, and have been described as the Golgi Endoplasmic Reticulum Lysosome System, or GERL (Novikoff, 1976), Compartment of Uncoupling of Ligand and Receptor, or CURL (Geuze et al., 1983), Partially Coated Reticulum, or PCR (Pesacreta and Lucas, 1984), Trans Golgi Network, or TGN (Griffiths and Simons, 1986), and MPR/lgp-enriched structure (Griffiths et al., 1988).

In recent years much has been learned about Golgi function at the molecular level by examining protein transport between Golgi compartments that have been reconstituted in a cell-free system. This complex reconstitution was first accomplished by Fries and Rothman in 1980. Prior to this time the Golgi apparatus had been implicated as playing roles in the secretory process, plasma membrane protein transport and lysosome biogenesis, primarily using a combination of electron microscopy and cell fractionation methods (Jamieson and Palade, 1967, 1977; Palade, 1975). Protein molecules which travel the normal pathway from rough endoplasmic reticulum (RER) to their final destinations must first be routed through the Golgi complex strictly in the cis-to-trans direction to be sorted at the tubular/vesicular network at the trans pole. On the way, most protein molecules are covalently modified in sequential fashion by activities which seem to reside in specific Golgi cisternae. Although the reason for most of these modifications is not known, one notable exception is the assembly of the oligosaccharide chain terminating in mannose 6-phosphate (M-6-P) which designates its possessor as a lysosomal enzyme to be sorted as such via a M-6-P receptor (Dalms et al., 1989). In addition to the well documented cases of Golgi export functions, there is mounting evidence that some protein molecules internalized by endocytosis may pass through the tubular/vesicular network (specifically the TGN) at some stage before arriving at their ultimate dispositions (Bomsel and Mostov, 1991).

The Golgi complex and the associated tubular/vesicular networks must be viewed as organelles with central and very complex roles in the cell. However, although the details of stack morphology and function are now better defined, questions remain as to how, when and under what conditions stacks are formed and maintained in animal cells, how secretory proteins move through the stack(s), and what role the tubular/vesicular networks play. With respect to the first question, there have been few reports of how Golgi cisternae and/or stacks increase or decrease in response to altered growth states. Two exceptions have been the electron microscopic study of differentiating Dictyostelium discoideum prespore cells, in which a fully formed Golgi apparatus was shown to develop from a group of vesicles and vacuoles around the nucleus (Takemoto et al., 1985) and the observations in HeLa cells concerning Golgi replication as a function of mitosis (Lucocq et al., 1989). The latter research has shown that fragmentation of the Golgi at metaphase yields multivesicular clusters and free vesicles which reassemble in a two-step process involving initial formation of cisternal clusters near endoplasmic reticulum “buds” during telophase and then congregation and fusion of these clusters into the juxtanuclear stack early in G1.

This report reviews work arising from the observation that Madin-Darby canine kidney (MDCK) cells that had been transfected with the human growth hormone (hGH) gene appeared to have an alteration in the relative number of Golgi profiles as compared to untransfected cells. Furthermore, by controlling the growth conditions, it is possible to manipulate elaboration of the organelle. This system opens up possibilities for studying Golgi apparatus biogenesis and perhaps will aid in the elucidation of the role of the tubular/vesicular network.

Dulbecco’s modified Eagle’s medium (DMEM), minimal essential Eagle’s medium (MEM) lacking L-leucine, L-lysine and L-methionine, the tissue culture antibiotics, the defined medium components, fluoresceinated goat anti-rabbit IgG, rhodamine-labelled anti-mouse IgG, phenylmethylsulfonyl fluoride (PMSF) and cycloheximide were purchased from Sigma Chemical Co. (St. Louis, MO). Fetal bovine serum (FBS) was from Hyclone Labs, Inc. (Logan, UT); G418 was obtained from GIBCO BRL/Life Technologies (Gaithersburg, MD); and Mowiol and 1,4-diazobi-cyclo[2.2.2]-octane (DABCO) were from Pierce (Rockford, IL). PNgase F (Endo F) was purchased from GlycoSystems, Inc. (Rosedale, NY). Dipicolinate (2,6-pyridinedicarboxylic acid (neu-tralized with sodium hydroxide) came from Aldrich Chemical Co. (Milwaukee, WI). Polybed 812 (EPON) was purchased from Poly-sciences (Warrington, PA), and trans[³⁵S] was from ICN Bio-medical (Irvine, CA). Rabbit anti-hGH antibody was obtained from NIADDA (Bethesda, MD) and monoclonal anti-p52 kDa antibody was from Upstate Biotechnology, Inc. (Lake Placid, NY). Vectastain ABC-alkaline phosphatase system was purchased from Vector Laboratories (Burlingame, CA). All other chemicals were reagent grade.

MDCK cells were obtained from American Type Culture Collec-tion (Lake Placid, NY) and were routinely grown in complete medium (CM) consisting of DMEM containing 10% FBS and penicillin/streptomycin/amphotericin B solution at 37°C in a humidified atmosphere of 5% CO2/95% air. Transfection of MDCK cells and selection of clones permanently expressing the hGH gene were obtained, as previously described (Rudick et al., 1991). The transfected cells, designated clone 3A, were maintained in the same medium containing 450 μg/ml (486 μg/μg) G418. In some cases cells were grown in defined medium (DM) consisting of DMEM supplemented with 5 μg/ml transferrin, 3.25 pg/ml triiodothyronine, 25 ng/ml prostaglandin E1, 10 nM selenate, with or without 5 μg/ml insulin and 20 ng/ml hydrocortisone (Taub et al., 1979). Cells were sometimes exposed to medium (either CM or DM) to which dipicolinate, a chelator of zinc ion (Yamaguchi and Matsui, 1989), had been added to a final concentration of either 0.1 mM or 1.0 mM, or to which 10 mM (final concentration) cycloheximide had been added. For some experiments cells were grown on coverslips or in culture dishes, and for others they were grown in 45 mm Millicell-HA culture inserts (Millipore Corp., Bedford, MA). Attainment of confluency on the inserts was determined in two ways. Firstly, the electrical resistance across the cell sheet was determined to be at least 450–500 Ω cm² using a Millicell electrical resistance system. Secondly, because the resistance was not always a reliable indication of the tightness of the cell sheet, at least one filter culture was stained with hematoxylin (Pitt and Gabriels, 1986) and examined microscopically.

Generation times of untransfected and clone 3A cells were determined as follows. Cells were grown in 35 mm dishes in either CM or DM, trypsinized at various times after seeding at a low density, and placed in filtered 0.85% sodium chloride for counting using a Model ZM Coulter Counter (Coulter Electronics, Hialeah, FL). The medium was changed every three days until confluency was reached. Cellular volumes were estimated using a multichannel analyzer linked to the Coulter Counter.

Secreted hGH was determined by radioimmunoassay (Nichols Institute, San Juan Capistrano, CA).

Cells were grown for 24 h on coverslips, fixed for 10 min with freshly prepared 4% formaldehyde in phosphate buffered saline (PBS), pH 7.4, and permeabilized with 0.2% Triton X-100 in PBS for 10 min. After washing with PBS, intracellular hGH was then detected by incubating the cells for 1 h at room temperature with either a 1:200 or 1:500 dilution in PBS/3% bovine serum albumin (BSA) of rabbit anti-hGH serum or a 1:100 dilution of monoclonal anti-p52 kDa (a resident Golgi protein) in PBS/3% BSA, followed by fluorescein-labelled goat anti-rabbit IgG (1:100 in PBS/3% BSA) or rhodamine-labelled goat anti-mouse IgG (1:200 in PBS/3% BSA), respectively. The coverslips were then mounted in Mowiol containing 2.5% DABCO for viewing with a Zeiss Photomicroscope II (Zeiss Instruments, Thornwood, NY). Normal rabbit serum or mouse ascites fluid served as control, when substituted for anti-hGH serum or monoclonal anti-p52 kDa, respectively. Cells were photographed using Kodak Tmax 3200 film.

MDCK cells grown on Millicell-HA filters were washed in Eagle’s balanced saline solution, then fixed in 3% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.3, for 1 h, followed by three 10 min rinses in 0.1 M cacodylate buffer alone. Cells were post-fixed for 1 h in 1% osmium tetroxide in cacodylate buffer, rinsed with three 5 min washes of distilled water, and then stained for 1 h en bloc using a 0.5% aqueous uranyl acetate solution. Following distilled water washes, the material was dehydrated in a series of graded ethanol solutions through 100% ethanol. To infiltrate with Epon, three successive mixtures of 100% ethanol/Epon 812 (2:1, 1:1 and 1:2) were used. The membranes were then placed in fresh plastic before being polymerized at 65°C for 24 h. All procedures were performed at room temperature except for the polymerization.

Polymerized samples were sectioned on a Sorvall MT2-B Ultra-microtome (Research and Manufacturing, Tucson, AR) with a diamond knife. Thick sections (0.5-1.0 μm) were stained with 1% toluidine blue in a 1% sodium borate solution. Thin sections (60–100 nm) were cut, collected on uncoated 150M nickel grids, and then stained with a 50% ethanolic saturated uranyl acetate solution for 5 min, followed by Sato’s lead stain (Sato, 1968) for 5 min. Sections were then observed with a Hitachi 600 electron microscope (Hitachi Instuments, San Jose, CA) and pictures were taken using Kodak SO-163 negative film.

Most stereological procedures require collection of images of randomly oriented sections of isotropically oriented cells. However, filter-grown cell monolayers are highly oriented and both surface features, such as microvilli, and internal structures, such as Golgi stacks, are anisotropically distributed. Since the monolayer is most easily sectioned perpendicular to the filter, randomly ordered sectioning cannot be assumed. A solution to the problem of anisotropy is to employ sampling procedures for vertical sections as outlined by Gundersen et al. (1988) and to choose a stereological test system for analyzing the sections that will allow unbiased determination of particle number, in this case, Golgi profiles. In the present study, the most important morphometric results are comparative estimates where a structure is compared under different conditions, and, thus, any outside influences will affect the structure in a similar manner in all conditions. But to minimize error, the fractionator was chosen as an unbiased estimator of the total number of items, i.e. Golgi profiles, in a specimen (Gundersen, 1986), since with the use of the fractionator one need not measure the distance between sections, the section thickness, the volume of the specimen, or assume anything about shrinkage/swelling, sectioning compression or lost caps (Cruz-Orvive and Weibel, 1990). It has successfully been used to count particles in hamster lungs (Geiser et al., 1990) and to estimate the total number of lymphatic valves in infant lungs (Ogbuihi and Cruz-Orvive, 1990).

Sections were cut perpendicular to the filter and stained as described (see Transmission electron microscopy). Representative micrographs of the sections were taken at a primary magnification of 6000 following the guidelines for random sampling as outlined by Gundersen (1986). The negatives were enlarged by a factor of 2.7 and a double square lattice grid (B100) with interline spacing equivalent to 1.49 mm was applied (Weibel, 1979). The grid was aligned with the surface of the filter so that it had the same orientation over each cell, and then Golgi profiles were counted in random, but distinct, cells as described for the fractionator procedure. In this manner, a single cell was used only one time and each Golgi stack within that cell section was counted only once. The counting was performed independently by two or three workers.

The average numbers of Golgi profiles in the two lines or in cells under different growth conditions were statistically evaluated using the Mann-Whitney U test. This two-sample ranked test for nonparametric data is designed for use with small experimental populations. The statistical reference is Zar (1984).

Cells grown on Millicell-HA inserts were washed twice with methionine-free MEM and were then placed into a new multiwell dish in 1.2 ml of methionine-free MEM containing 2% dialyzed FBS. It had been determined that 2% dialyzed FBS supports the same level of hGH secretion in clone 3A cells as does 10% FBS. After incubating for 1 h, the basolateral medium was removed and replaced with 1.2 ml of MEM/2% FBS containing 100 mCi/ml of trans[³⁵S]. With time, 100 μl samples of apical and basolateral media were removed to microfuge tubes, PMSF was added to a final concentration of 2 mM, and proteins were precipitated overnight with an equal volume of cold 10% TCA containing 0.1% cold D,L-methionine. Precipitates were centrifuged and washed twice with cold 5% TCA, and then the pellets were dissolved in dissociation buffer for resolution on 12% SDS-polyacrylamide gels, as described by Laemmli (1970). The gels were fixed in 40% methanol/10% acetic acid for 1 h, placed into 10% glycerol for 30 min, and impregnated with 1 M sodium salicylate for 30 min. After drying the gels were overlayed with Kodak XAR film. At each of the assay times, aliquots of apical and basolateral media were also collected to determine hGH levels and total radioactivity. Fluorographs were scanned and the area under the peaks integrated with a GS 300 transmittance/reflectance scanning densitometer using the GS-365 data system (Hoefer Scientific Instruments, San Francisco, CA).

For Endo F treatment, basolateral and apical media from cells grown with trans[³⁵S] for 6 h in the absence of FBS were collected, desalted and dried in a SpeedVac (Savant Instruments, Inc., Hicksville, NY). After dissolution in 100 ml of 20 mM phosphate buffer, pH 7.5, containing 0.1% SDS, the samples were heated at 100°C for 10 min, cooled, and then Triton X-100 was added to 0.5%. Each sample was separated into two 50 μl aliquots and 25 units of Endo F was added to one. All preparations were incubated overnight at 37°C according to the manufacturer’s protocol. After being dried, the samples were heated in dissociation buffer, and analyzed on a 12% SDS-polyacrylamide gel, as described above.

Cells were grown in 100 mm dishes in CM, refed for 1 h, then washed three times with 60 mM Tris-HCl, pH 6.8, and scraped from the dish in 10 ml of the same buffer. A sample was taken for a cell count. After centrifuging at 800 g for 5 min the cell pellet was resuspended in 10 volumes of the buffer containing 2% SDS. The cells were broken and the chromatin sheared by passage first through a 20 gauge and then a 26 gauge needle. Nuclei and mitochondria were removed by pelleting at 10,000 g for 10 min and samples of supernatants were taken for estimation of protein (Lowry et al., 1951). Laemmli dissociation buffer (10 × without 2-mercaptoethanol) was added to a final concentration of 1× and, after boiling for 5 min, equal amounts of MDCK and clone 3A cell proteins were loaded onto a 12% SDS gel along with prestained molecular weight standards. Following electrophoresis, protein bands were transferred to Triton X-100-free nitrocellulose using the buffer system described by Towbin et al. (1979). All subsequent steps were carried out with gentle agitation. The blot was washed in TBS (100 mM Tris-HCl, pH 7.5, containing 0.9% NaCl) and then incubated for 1 h at room temperature in TBS/5% normal rabbit serum. After washing over a 30 min period with several changes of TTBS (TBS containing 0.1% v/v Tween 20), the blot was exposed to a 1:50 dilution of anti-p52 kDa antibody in TTBS at 4°C overnight. The blot was then washed four times with TTBS over 15 min at room temperature and transferred to a 5 μg/ml solution of biotinylated rabbit anti-mouse IgG in TTBS for 30 min at room temperature. It was then washed as above and placed into the Vectastain ABC-alkaline phosphatase reagent (avidin/biotinylated alkaline phosphatase) for 30 min. After washing, the membrane was incubated in bromochloroindolyl phos-phate/nitroblue tetrazolium (BCIP/NBT), as described by Harlow and Lane (1988), until bands appeared, whereupon the reaction was stopped with PBS containing 20 mM EDTA. Appropriate controls were also run, such as omission of primary or secondary antibody. The blot was scanned and areas under peaks integrated with the GS 300 Transmittance/Reflectance Scanning Densitometer.

A high secreting, transfected MDCK cell clone (clone 3A), that secretes hGH linearly at 5.3 ng/h per 105 cells, as measured by radioimmunoassay, was analyzed for the presence of hGH by immunofluorescence and exhibited bright juxtanuclear fluorescence (Fig. 1). The appearance of the cells suggested that most of the fluorescence was in the Golgi region. In order to confirm this, clone 3A cells were stained simultaneously with anti-hGH antibody (fluorescein) and with anti-p52 kDa (rhodamine). The latter was generated against a 52 kDa protein molecule purified from a human breast carcinoma cell line (MCF-7) Golgi apparatus. The crescent-shaped perinuclear pattern of fluorescence obtained with this monoclonal antibody is consistent with its recognition of a Golgi specific antigen. As can be seen in Fig. 2, D and E, fluorescein and rhodamine fluorescence coincided. Cells stained with fluorescein-labelled or rho-damine-labelled antibodies alone were viewed with both excitation filters to ensure that neither label was masking or enhancing the presence of the other (data not shown). A surprising result was obtained when anti-p52 kDa antibody staining of clone 3A and untransfected MDCK cells were compared. The apparent amount of rhodamine fluorescence was greater in clone 3A cells than in the untransfected cells (Fig. 2, A and D), suggesting there might be a quantitative difference in their respective Golgi complexes. These double labelling experiments were repeated several times and each time the rhodamine fluorescence was consistently less distinct in the untransfected cells as compared with the transfected ones, even though they were treated exactly the same way. Western blot analysis of proteins in post-mitochondrial supernatants from MDCK and clone 3A cells grown in CM using anti-p52 kDa antibody showed that the enhanced fluorescence demonstrated in clone 3A cells (Fig. 2) was correlated specifically with an increase in the amount of a protein molecule with an apparent mass of 52 kDa (Fig. 3). Densitometry yielded a 1.9-fold increase of the 52 kDa protein in clone 3A cells compared to MDCK cells.

This prompted an electron microscopic analysis of the cells. Ultrastructural examples of representative untransfected (Fig. 4A) and clone 3A (Fig. 4B) cells grown on microporous supports for seven days clearly show a marked increase in the dimensions and/or number of Golgi stacks in the latter. In order to extend the analysis, both untransfected and clone 3A cells were grown on microporous supports in CM for either two or seven days, prepared for electron microscopy, viewed and photographed. Using the morphometric sampling procedure and fractionator estimator described in Materials and Methods, 11 to 17 different cells of each type were examined and the average number of Golgi stacks per cell was estimated. The results of this stereological analysis (Table 1) revealed that untransfected cells had 1.7 or 1.6 and transfected cells 3.2 or 3.3 Golgi profiles per cell in CM at two or seven days, respectively, indicating that there was a difference in the number of stacks per cell manifest as early as two days in culture (Mann-Whitney U comparisons of the transfected to the untransfected cells at two days, U13,17 = 178, P ≤0.005; at seven days, U11,17 = 146.5, P ≤0.01). These data reflect either an increase in the number of individual Golgi stacks or an elongation of the original stacks which would enhance the probability of their being included in the sections. Further-more, both volumes and doubling times of the untransfected and clone 3A cells in CM were similar (Table 1), so that a change in the size of one of the cell types could not explain the observed differences in the number of profiles. Since the general appearance of the Golgi complexes in both cell types was the same, this result did not seem to be due to an engorgement and swelling of the clone 3A Golgi. Thus, it seemed that the secretion of large quantities of hGH had led to elaboration of Golgi complex, presumably to accommodate the heightened demand for its functions. It was possible that Golgi enhancement could have been due to some kind of autocrine effect of the secreted hGH on the clone 3A cells. However, after treatment of untrans-fected MDCK cells with 5 μg hGH/ml medium for 7 days, morphometric analysis yielded 1.1 Golgi stacks per cell (n = 9), when hGH was placed into the basolateral medium, and 1.8 Golgi stacks per cell (n = 9), when hGH was in the apical medium. There was no statistical difference between these values and the ones reported for MDCK cells grown in the absence of hGH.

Since the transfected hGH gene is driven by the metal-lothionine promoter, it was hoped that modulation of secre-tory activity could be accomplished by removing zinc ion (the only heavy metal present) from CM with dipicolinate, a zinc ion chelator, or by growing the cells in a serum-free defined medium (DM) plus or minus hydrocortisone, a glucocorticoid, which can promote transcription of the hGH gene via its receptor. Dipicolinate had no effect on secretion of hGH (data not shown), but, exposure of the cells to DM (with or without hydrocortisone) did. Clone 3A cells grown in CM were washed and then placed into DM. After allowing the cells to adapt for 1 h, samples of the medium taken with time were assayed for the presence of hGH and compared to similar samples taken from a culture which had received fresh CM (Fig. 5). Cells in CM continued to secrete hGH at a high linear rate throughout the course of the experiment, but cells in DM released only a relatively small amount of hGH over the 120 min assay period. However, immunofluorescence of cells maintained in DM for the 2 h time showed that there was still hGH in the Golgi region, even though little was being secreted (data not shown). When hydrocortisone was omitted from the DM, the residual fluorescence remained. To determine if longer exposure to DM would result in disappearance of intracel-lular labelling and complete cessation of secretion, 3A clone cells were grown for seven days in either CM or DM lacking hydrocortisone and analyzed for intracellular hGH and rate of secretion. Microscopic examination of cells in CM revealed that there was still hGH associated with the perinuclear region, but that much of it was also trapped between the tightly confluent cells (Fig. 6B). While the latter effect was not seen in DM, there was still significant perinuclear fluorescence (Fig. 6A). Only inhibition of protein synthesis by treating the cells in DM with 10 mM cycloheximide for 2 h prior to fixation caused the disappearance of fluorescence (Fig. 6C). Fig. 7 illustrates that during the seven day period, there was a slow release of hGH from cells growing in DM, but it was four-fold less than that secreted by cells in CM and apparently declined rapidly after exposure to DM (Fig. 5). Thus, the presence of hGH in the Golgi region of cells even days after switching them from CM to DM appears to be due to its continued synthesis, but probably at a lower rate.

Given the decrease in hGH secretion from cells maintained in DM, would there be an effect of this change in secretory activity on the Golgi complex itself? To answer this, both untransfected and clone 3A cells were kept in DM on microporous supports for various periods of time and examined by electron microscopy. Fig. 8 illustrates examples of representative untransfected (Fig. 8A) and clone 3A cells (Fig. 8B) grown for seven days in DM. It became apparent that the number of Golgi stacks in the transfected cells had decreased. To confirm this observation, the same type of analysis to determine Golgi stack numbers that was performed on cells in CM was undertaken on cells growing on microporous supports in DM for either two or seven days. The results (Table 1) revealed that for the untransfected cells there was no statistical difference between the number of Golgi stacks in cells grown in DM or CM. However, in as little as two days of exposure to DM, the number of Golgi profiles began to decline in clone 3A cells, and by seven days there was no statistical difference between the number found in transfected and untransfected cells.

From the foregoing it seemed likely that the pressure of increased demand for Golgi function had induced the amplification of the organelle in the transfected cells. On this basis it might have been assumed that the rate of secretion of endogenous proteins would be increased proportionally. This was investigated by labelling clone 3A and untransfected cells grown on microporous supports with trans[35S] and analyzing fluorographs of basolaterally and apically secreted proteins after SDS-PAGE (Fig. 9). Subsequently, the fluorographs were scanned with a densitometer, as described in Materials and Methods (Fig. 10). Apical secretion of endogenous proteins was markedly attenuated in clone 3A cells, except for a 40 kDa protein which appeared to be induced in tandem with hGH. Densitometric scans supported these observations. For example, scans of MDCK and clone 3A apical lanes confirmed that MDCK cells have a higher rate of secretion of endogenous proteins than clone 3A cells (compare Fig. 10, A and B), while subtraction of the MDCK scan from that of clone 3A illustrated both the presence of hGH and an increase in the secreton of an endogenous 40 kDa protein (Fig. 10, C). Since the subtraction program converts negative values to zero, peaks enhanced in the untransfected cells did not appear. Thus, a reciprocal densitometric subtraction was done and revealed that there were in fact higher amounts of several protein molecules secreted by MDCK cells (data not shown). In contrast, the rate of basolateral secretion was the same in clone 3A and MDCK cells, except for what perhaps may be a slight increase in a 18 kDa band in the transfected cells. Clone 3A cells secrete hGH through both apical and basolateral plasma membrane domains, as previously described (Rudick et al., 1991). Estimation of the areas under the peaks in the clone 3A scans revealed that hGH accounted for about 10% of total protein secreted either apically or basolaterally.

In order to determine if there was a difference in the glycosylation of secreted endogenous proteins between MDCK and 3A cells, both types were labelled for 6 h in serumfree medium and the secreted proteins were then treated with Endo F to remove N-linked oligosaccharide chains. It was found that in both cell types most of the secreted endogenous proteins are N-glycosylated judging from their increased mobilities (Fig. 11). Thus, there is no apparent change in glycosylation as a result of hGH expression. It should be emphasized that the relative intensities of MDCK and clone 3A bands in this experiment should not be compared to those depicted in Fig. 9, since the labelling conditions were not the same. It is well documented that hGH is not a glycoprotein (Denoto et al., 1981; Moore and Kelly, 1986), and these data confirm it.

Immunofluorescence comparison of clone 3A and untransfected cells using anti-p52 kDa antibody revealed a much more intensely stained perinuclear cap in clone 3A cells than in untransfected cells. This was shown to correlate with an actual increase in 3A cells of a protein molecule having an apparent mass of 52 kDa. Thus, we were prompted to examine the cells ultrastructurally, in order to ascertain the possible basis for this result. At first glance the Golgi stacks in clone 3A cells seemed to be larger and/or more numerous than in untransfected cells, and this was borne out after determination of the average number of Golgi profiles per cell. It is significant that the cisternae in clone 3A cells did not appear to be swollen, as might be expected if the observed effect were due simply to their engorgement by hGH in transit. Such an effect has been seen, for example, in cells infected with vesicular stomatitis virus in which the transport through the Golgi of a prominent envelope glycoprotein becomes blocked, resulting in enlargement of the affected cisternae (Griffiths et al., 1985) or in secretory cells treated with secretagogue, which increases GERL (Hand and Oliver, 1984). Instead, the Golgi saccules of clone 3A cells were structurally indistinguishable from those of untransfected cells. This suggested that there were increased Golgi stacks in the clone 3A cells or an amplification of Golgi, presumably brought about by the increased demand for basic Golgi function in the hGH secreting cells.

When clone 3A cells were grown in DM for seven days, the rate of secretion of hGH declined 4-fold, but did not completely halt. There remained a much reduced perinuclear hGH fluorescence which could not be abolished by omission of glucocorticoid from the DM or by chelation of zinc ion with dipicolinate. Only inhibition of protein synthesis with cycloheximide resulted in disappearance of Golgi hGH fluorescence. It seems that even in the absence of two potent positive effectors of metallothionein promoter function, there was some hGH gene transcription, although probably at a reduced level, leading to continued hGH synthesis. The hGH fluorescence seen in the Golgi under these conditions undoubtedly represented a drastically reduced quantity of Golgi hGH compared to that in cells grown in CM, as reflected in the hGH secretory rates. Ultrastructural examination of cells grown in DM revealed that the average number of Golgi in clone 3A cells had declined to that of untransfected cells, which supports the earlier supposition concerning the relationship between Golgi amplification and the rate of hGH secretion. However, it is also possible that the decrease in the number of Golgi profiles in DM grown 3A cells was due to there being less hGH in the medium. Such an autocrine effect was ruled out by counting the number of Golgi stacks in untransfected cells exposed for 7 days to exogenous hGH. Thus, the results suggest that the Golgi complex of MDCK cells, and probably other cell types as well, has a finite capacity for handling the flow of protein molecules en route to their post-Golgi destinations and that any increase in demand for Golgi function beyond this requires more of the organelle itself. In this regard, Griffiths et al. (1989) have observed an accumulation of Semliki Forest Virus (SFV) spike protein resulting in the enlargement of the TGN in SFV infected MDCK cells. They propose that this might be due to the saturation of the capacity to transport proteins from the TGN to the cell surface.

In the present study the rate of secretion of the majority of apically released endogenous proteins was lower in clone 3A cells than in untransfected cells. This could be a result of the competition of hGH and endogenous secretory proteins for post- or intra-Golgi vesicles, and, since there is such a high concentration of hGH and a limited vesicular volume, bulk flow would tend to cause hGH to be included in the vesicles, to the detriment of endogenous proteins. However, as determined densitometrically, hGH seems to account for only about 10% of total secreted proteins. Of course this is a rough estimate, since fluorograph band intensities depend not only on the amounts of the proteins secreted, but also on the relative number of methionine and cysteine residues in the various protein species. Nevertheless, the increase in total protein secretion would hardly seem to warrant the two-fold elaboration of Golgi by the clone 3A cells. However, it is not known what proportion of the hGH synthesized in the clone 3A cells is actually secreted. It is possible that much of the hGH is degraded and never exits from the cell. There are several sites of possible protein degradation, including the cytosol, the endoplasmic reticulum and the lysosomes. As ours and previous studies have shown, hGH is not targetted to any particular cell surface in MDCK cells, presumably because the hGH does not possess a signal recognized by the secretory machinery. Bulk flow through membranous organelles via vesicular carriers would require that hGH enter any forming vesicle regardless of its destination, and this would include those carrying lysosomal proteins from the TGN. Thus, an unknown amount of hGH might end up in lysosomes and there may be a much higher hGH flux through the Golgi than is evident in the amount that is secreted, making the necessity for the observed increase in the number of Golgi understandable. This explanation does not seem to hold true for the basolateral pathway where there is little, if any, difference in the rates of endogenous protein secretion between the two cell types. We cannot explain why there is a differential effect on the rates of apical and basolateral secretion of endogenous proteins. However, we are currently investigating the intracellular location of hGH in clone 3A cells, especially with regard to post-Golgi compartments.

Support for this study was provided by National Institutes of Health grant R15GM42109-01. The authors wish to acknowledge the technical help of Shikha Chakraborty.