Structural reorganization of the rough endoplasmic reticulum without size expansion accounts for dexamethasone-induced secretory activity in AR42J cells

The endoplasmic reticulum (ER) consists of an extensive network of interconnected membrane tubules spread throughout the cytoplasm and sheet-like cisternae, more often seen in the perinuclear region (Palade, 1956; Porter et al., 1945; Palade and Porter, 1954). The rough endoplasmic reticulum (RER), a morphologically distinct sub-domain of the ER studded with ribosomes, is a component of the endomembrane system involved in the biosynthesis of membrane and secretory proteins (Palade, 1975). Morphometric analysis has shown that an increase in secretory activity leads in many cell types to an increase in the size of the RER. For example, frog hepatocytes stimulated by 17β-estradiol to secrete vitellogenin increase their RER content 4- to 5-fold (Bergink et al., 1977; Herbener et al., 1983; Rajasekaran et al., unpublished). Similarly, rat seminal vesicle epithelial cells induced by testosterone to secrete plasma protein S and F (Falwell and Higgins, 1984), aleurone cells stimulated by gibberellic acid to secrete α-amylase (Belanger et al., 1986) and resting B-lymphocytes stimulated to secrete immunoglobulins by either mitogens or specific antigen (de Vries et al., 1983; Shohat et al., 1973) enlarge their RER, but in all cases the increase was not quantified. Two studies with stable cell lines, the murine B cell line CH12 stimulated by lipopolysaccharide (LPS) to secrete IgM (Wiest et al., 1990) and the rat pancreatic acinar carcinoma AR42J cell line induced by dexamethasone to secrete amylase (Logsdon et al., 1985; Swarovsky et al., 1988), have also reported an increase in the size of the RER.

An intriguing, but unexplained, finding is that in most very actively secreting cells the large stacked cisternal RER (SC-RER) predominates over the tubular and vesicular RER (TV-RER; Weiss, 1988). This is the case, for instance, in all the examples of induced secretion described above. Additional examples of large increases in this characteristic form of RER are found in the cells of the posterior silk gland, which is actively secreting large amounts of fibroin during larval stages of the silkworm (Morimoto et al., 1968), and in chicken primary chondrocytes induced by ascorbic acid to secrete collagen (Pacifici and Iozzo, 1988). It is unclear at present what mechanisms regulate the amplification and structural changes of the RER during enhanced secretory activity and what are the advantage(s) of the cisternal over the tubular form of the organelle. These striking changes in the morphology of the ER must be regulated by factors yet to be identified. These factors may be either constitutively expressed in cells with a high secretory activity or induced upon stimulation of an active secretory state.

The mechanisms that control these remarkable structural changes of the RER can best be studied in cells that make large amounts of cisternal RER in response to a specific stimulus. The rat pancreatic acinar carcinoma cell line, AR42J, provides a useful model for this type of study, since these cells respond with a dramatic increase in secretory activity (5- to 20-fold) upon exposure to dexamethasone. Two previous reports of morphometric analyses indicated that the increase in the secretory activity of this cell line is accompanied by a 2.4-(Logsdon et al., 1985) or 30-fold (Swarovsky et al., 1988) increase in the size of the RER. This discrepancy led us to reinvestigate the dexamethasone-induced changes in RER structure and size at the ultra-structural level and by a new biochemical assay that allowed us to determine the size of the RER by quantitating the amount of bound ribosomes (Rajasekaran et al., unpublished). Our results indicate that the RER does not increase in size in spite of a dramatic increase in the secretory activity after dexamethasone treatment. Rather, the increased secretory activity of dexamethasone-induced cells appears to depend on a change in the structure of the RER from a tubulo-vesicular to a cisternal configuration that is biosynthetically more efficient. To our knowledge, this is the first clear demonstration that reorganization of the RER, rather than an increase in surface area, is accompanied by an increase in secretory activity.

AR42J cells (ATCC, CRL 1492) were obtained from the American type culture collection (Rockville, MD). AR42J cells were also provided by Dr Craig Logsdon (University of Michigan, USA) and Dr Horst Kern (University of Marburg, Germany). Cells were maintained at 37°C as subconfluent monolayer cultures in DMEM containing 10% fetal calf serum supplemented with glutamine (2 mM), penicillin (100 i.u./ml), streptomycin (100 μg/ml), Fungizone (2 μg/ml) and polymyxin B (50 μg/ml). Cells obtained from ATCC required insulin (4 μg/ml; Irvine Scientific, Santa Anna, CA), EGF (2 ng/ml; Boehringer Mannheim, Indianapolis, IN) and HEPES (20 mM, pH 7.4) for normal growth. Cells from Dr Logsdon’s and Dr Kern’s laboratories were maintained at 5% CO₂ while the cells from the ATCC were maintained at 10% CO₂. The doubling times of these cultures was about 24 h. Cells were grown in 75 cm² flasks (Corning, NY) and fed twice a week. Cells were detached with trypsin (0.25%) in Hanks’ buffer containing EDTA (2 mM, pH 8; trypsin-EDTA) and plated at a density of 5×10⁶ to 6×10⁶, 2×10⁶ to 3×10⁶ or 0.5×10⁶ to 10⁶ cells per 100 mm, 60 mm or 35 mm dish (Corning, NY), respectively. The medium was changed 2-3 h before the addition of dexamethasone (Sigma Chemical Co., St. Louis, MO) to a final concentration 10 nM. Dexamethasone was disolved in 50% ethanol (5 μM stock solution) and added 24-36 h after splitting the cells. An equivalent volume (2 μl/ml) of 50% ethanol was added to control cultures. The medium was changed once at 36 h and dexamethasone or 50% ethanol was added immediately to experimental or control cultures, respectively. Cells were harvested at 72 h after the initial addition of dexamethasone. RU 38486 (kindly provided by Dr D. Philibert, Roussel Uclaf, France) was dissolved in 50% ethanol and added (1 μM final concentration) 1 h prior to the addition of dexamethasone.

Intracellular and secreted amylase was measured according to the method of Logsdon et al. (1985), with the following changes. Cells grown in 60 mm dishes for 69 h in the presence of dexamethasone were washed twice with MEM containing no phenol red. Cells were then incubated for 3 h in the same medium (1.2 ml) containing BSA (5 mg/ml) and soybean trypsin inhibitor (0.1 mg/ml; Sigma Chemical Co., St. Louis, MO). The media were removed, PMSF (Sigma Chemical Co., St. Louis, MO) was added to a final concentration of 1 mM and the cells were stored at −20°C. The cells were washed twice with PBS and scraped in 1 ml of 50 mM sodium phosphate (pH 6.9) and 50 mM NaCl containing 1 mM PMSF. Cells kept on ice were then sonicated three times for 15 s each with cooling intervals of 10 s, using a W185 cell disrupter (Heat System Ultrasonics Inc., Plainview, NY) equipped with a microtip. A sample of 500 μl was used for DNA analysis. The remaining 500 μl was used to measure amylase activity as described in the Worthington enzyme manual (Worthington, 1988).

To determine cell numbers the cells were washed twice in PBS and trypsinized for 3-4 min with trypsin-EDTA in Hanks’ buffer (see above). After the cells were detached, normal growth medium 5 times the volume of the trypsin-EDTA-containing buffer was added. Cells were recovered by centrifugation, resuspended in the normal growth medium and counted using a hemocytometer. The protein content of the cell lysates was quantified according to the method of Lowry et al. (1951). RNA was analyzed according to the procedure described by Munro and Fleck (1966). The precipitate of the alkaline digest with cold PCA (perchloroacetic acid) was suspended in 0.6 M PCA, incubated in boiling water for 10 min, chilled to about 4°C in ice water, and centrifuged at 3000 r.p.m. for 15 min in the cold. The supernatant was transferred into a clean tube, the pellet was resuspended in 0.6 M PCA and the suspension was centrifuged. The supernatant was combined with the previous one and used for DNA quantification by either UV absorbance or the diphenylamine colorimetric method of Burton (1968).

AR42J cells grown in 60 mm dishes in the absence or presence of dexamethasone for 72 h were washed twice in methionine-free DMEM and starved for 30 min. The medium was replaced by the same medium (1.2 ml/dish) containing 125 μCi/ml of [³⁵S]methionine (specific activity 1209 Ci/mmol). After 5 min of labelling the medium was removed and the dishes were cooled to 0°C. Lysis of the cells and immunoprecipitation of amylase was done as reported earlier (Tsao et al., 1992). A rabbit anti-human amylase antibody (Sigma Chemical Co., St. Louis, MO) diluted 200-fold was used for immunoprecipitation.

Cells treated with or without dexamethasone for different time periods were washed once with PBS and total RNA was isolated using the acid/phenol/guanidinium thiocyanate procedure (Chomczynski and Sacchi, 1987). RNA was dissolved in autoclaved water and the concentration was determined by measuring the absorbance at 260 nm. After electrophoretic separation on a 1.5% formaldehyde gel the RNA was transferred to a GeneScreen hybridization transfer membrane (NEN, Boston, MA), according to the procedure specified by the manufacturer. The filters were baked at 80°C for 2 h and hybridized with full-length ribophorin I (RI), or ribophorin II (RII) cDNAs labelled with [³²P]dCTP (specific activity 3000 Ci/mmole) using the BRL nick-translation kit (BRL, Gaithersburg, MD). The filters were washed four times for 10 min each at room temperature with 2× SSC containing 0.1% SDS, once for 15 min at 65°C with 0.1× SSC containing 0.1% SDS and then exposed at −70°C using Kodak X-Omat AR film.

Cell lysates were prepared from control or dexamethasone-treated cultures according to the method of Tsao et al. (1992). Protein samples were separated by SDS-PAGE (10%) according to the method of Laemmli (1970) and transferred to nitrocellulose paper at 200 mA for 12-15 h. The blots were blocked for 1 h in buffer A (10% Carnation milk in PBS) and incubated with rabbit antirat RI and RII polyclonal antibodies in buffer B (buffer A containing 0.3% Tween 20) for 2 h. The blots were washed in buffer C (PBS containing 0.3% Tween 20) 8 times for 5 min each and incubated in buffer B containing 50,000 c.p.m./ml of ¹²⁵I-Protein A (specific activity 2.59-3.70 MBq/μg) for 60-90 min. After washing (8 times, 5 min each) in buffer C the blots were exposed to Kodak X-Omat AR film.

All steps were carried out at 4°C unless otherwise specified. Three plates were selected from control and treated with dexamethasone for cell counting to obtain the average cell number per plate. The same number of cells were used per sample (about 300 million cells). Prior to harvesting, the cells were treated with cycloheximide (final concentration 1 mM) to stop protein synthesis and washed twice in PBS containing the same concentration of cyclo-heximide. Washed control and dexamethasone-treated cells were homogenized in a homogenizing solution (10 mM Tris-HCl, pH 7.5, 10 mM KCl, 1 mM MgCl₂) using a Dounce homogenizer. The homogenate was quickly mixed with 2.5 M sucrose to bring the homogenate sucrose concentration to 0.25 M. Post-nuclear supernatant (PNS) was obtained by centrifuging the homogenate on a 1 ml of 1 M sucrose-LSB (50 mM Tris-HCl, pH 7.5, 50 mM KCl, 5 mM MgCl₂) cushion in a 15 ml Corex tube at 2500 r.p.m. for 5 min at 4°C in a Sorvall HB-4 rotor. The nuclear pellet was stored on ice for further analysis. The PNS was adjusted to a final concentration of 2.1-2.2 M and 1 mM of sucrose and MgCl₂, respectively. The mixture (2.1 M.S-PNS) was used to make sucrose step gradients as follows: 2.5 M sucrose-LSB (1.5 ml), 2.1 M.S-PNS (8.0 ml), 1.99 M sucrose-LSB (1.5 ml) and 0.7 M sucrose-LSB (1.5 ml). The gradients were centrifuged in a SW-41 rotor (Beckman, USA) at 4°C for 20 h or longer at 36,000 r.p.m.

The band at the interface between the 1.99 M and 0.7 M sucrose layers (membrane fraction) and the rest (free ribosome fraction) were diluted 3-fold with LSB and centrifuged at 4°C for 3 and 5 h, respectively, at 40,000 r.p.m. in a 60Ti rotor. The membrane and free ribosome pellets were resuspended in 1 ml of LSB and used for further analysis.

Cells grown on polylysine-coated glass coverslips were treated with dexamethasone for 72 h. The control cells were treated with the same volume of 50% ethanol. Cells were washed once in PBS, fixed with 4% paraformaldehyde for 30 min and permeabilized with 0.2% Triton X-100 for 5 min. Then cells were washed with PBS and blocked with 2% BSA in PBS for 15 min. The endoplasmic reticulum was visualized using an ER antibody kindly provided by Dr Daniel Louvard. Cells were incubated at 37°C with 1:50 diluted ER antibody for 30 min, washed with PBS 3 times (10 min each) and further incubated for 30 min at 37°C with an anti-rabbit biotin-conjugated goat antibody (Vectar Labs, Burlingame, CA). After washing as described above, the cells were incubated with streptavidin-conjugated Texas red (Tx-R) for 30 min at 37°C, washed 6 times (10 min each) with PBS and mounted on glass slides using FITC-guard (Testog Inc., IL) as the mounting medium.

Cells, fixed and stained as described above, were examined in a PHOIBOS 1000 laser scanning confocal microscope (Sarastro, Stockholm, Sweden). Tx-R was excited with an argon laser. The emitted signals were collected and used to create three-dimensional reconstructions of serial confocal sections using the program Vanis (Sarastro, Stockholm, Sweden).

Cells grown on 60 mm dishes were washed twice with 0.1 M cacodylate buffer (pH 7.4) and fixed with 2% glutaraldehyde in 0.1 M cacodylate buffer for 2-4 h. At 30 min after addition of the fixative the cells were scraped and spun for 4-5 min in a microfuge. The pellet was washed, fixed with 2% osmium tetroxide, processed by conventional procedures for electron microscopy and viewed with a Philips 300 electron microscope at 80 kV.

Treatment of AR42J cells with dexamethasone led to an inhibition of cell growth as measured by cell counting and DNA determination (Fig. 1A). Pretreatment of the cells with the antiglucocorticoid agent RU 38486 (Moguilewsky and Philbert, 1984; Baulieu, 1989) abolished these biochemical changes (Fig. 1A). Furthermore, the RNA and protein content, when normalized to DNA, was practically unchanged (1.2-fold higher) in dexamethasone-treated and in control cells (Fig. 1B).

Dexamethasone had a profound effect on the production and release of amylase, a major secretory product of induced AR42J cells; 24 h after this treatment an approx. 6.6-fold increase in intracellular amylase activity and an approx. 4.6-fold increase in the secreted amylase activity was observed. These values did not change significantly at later times (Fig. 2; Fig. 3B). This correlated with a 6-fold increase in the rate of incorporation of [³⁵S]methionine into amylase (Fig. 3A). On the other hand the rate of incorporation of [³⁵S]methionine into total TCA-insoluble material increased 1.8-fold after 24 h (Fig. 3A) and 2.4-fold after 72 h of dexamethasone treatment (Table 1). On the basis of our results and on published data (Swarovsky et al., 1988) we estimate that, after dexamethasone treatment, amylase and total secretory proteins (including chymotrypsinogen, trypsinogen, procarboxypeptidase and lipase; Swarovsky et al., 1988) constitute 23% and approx. 50%, respectively, of the total newly synthesized protein. This indicates that the 1.8- to 2.4-fold increase in the incorporation into total protein may be primarily accounted for by the increase in the synthesis of secretory proteins. The effect of dexamethasone on the increase in amylase secretion was completely reversed by the antiglucocorticoid RU 38486 (Fig. 2).

The increase in secretory activity in AR42J cells was reported to be accompanied by an increase in the size of the RER (Logsdon et al., 1985; Swarovsky et al., 1988). To determine whether dexamethasone induced an increase in the amount of RER, levels of the RER-specific proteins ribophorin I and II were measured, which were shown to increase proportionally with respect to the size of the RER (Wiest et al., 1990). Ribophorins I and II are RER-specific integral membrane proteins (Kreibich et al., 1978a) that are found in the RER in a 1:1 stochiometric ratio with respect to bound ribosomes (Marcantonio et al., 1984) and are in close proximity to these bound ribosomes (Kreibich et al., 1978b; Yu et al., 1990). Recently it has been shown that the ribophorins are part of a heterotrimeric complex that includes a 48 kDa polypeptide that has oligosaccharyl transferase activity (Kelleher et al., 1992). The content of ribophorin mRNA and protein were quantified by northern and western blot analyses, respectively, at different times after addition of the drug. Neither the mRNA (Fig. 4) nor the protein (Fig. 5) levels of ribophorin I or ribophorin II changed after 24 h or 72 h (Fig. 6), which are the times when the morphological changes of the RER were observed. Thus, the induction of secretory activity by dexamethasone in AR42J cells is not accompanied by an increase in the levels of a characteristic RER marker.

We have recently developed a biochemical procedure to measure the RER, which is based on the quantitation of bound polysomes (Rajasekaran et al., unpublished). The levels of bound polysomes can be expected to reflect the amount of RER and provide an independent estimate of the size of this organelle. We applied this procedure to AR42J cells at three time points after dexamethasone addition (48, 72 and 96 h) when the change in the structure of the RER was very prominent (see below). The same numbers of control and treated cells from three different sources (see Materials and Methods) were processed for cell fractionation, so that their sedimentation profiles can be directly compared without any further normalization. To this effect, postnuclear supernatants were subfractionated into total membranes and free ribosomes. These two fractions were collected on sucrose density gradients, treated with high salt and puromycin and the absorbance profiles of the gradients were obtained to quantify the ribosome content. As shown in Fig. 7, there was no detectable difference in the amounts of free and membrane-bound polysomes between control and treated AR42J cells. In particular, the approximate 1:1 ratio of free to bound ribosomes is not affected by the dexamethasone treatment. The nuclear fraction did not contain any detectable amount of bound ribosomes, suggesting that contamination by unbroken cells was negligible and that all the bound ribosomes were contained in the membrane fraction. Western blot analysis of membranes collected from the peak fraction did not show any difference in the ribophorin I content, when comparing control and dexamethasone-treated cells (data not shown).

We also carried out studies in order to determine whether the surface area of the RER increased without a corresponding increase in bound ribosomes. AR42J cells kept as control or treated for 72 h with dexamethasone were labelled with [¹⁴C]uridine for 24 h and the total membrane fractions were prepared from post-nuclear supernatants as described in Materials and Methods. The membrane fractions were then subjected to isopycnic density gradient centrifugation. The gradients were then fractionated (500 μl aliquots) and for each fraction the density of sucrose and the radioactivity incorporated into ribosomes was determined. As shown in Fig. 8 (A,B) the density distributions of rough microsomes derived from control or treated cells were similar, indicating that the surface area of the RER did not change after dexamathasone treatment. The finding that the radioactivity contained in the membrane fraction (Fig. 8A,B) is indeed incorporated into rRNA was confirmed by experiments showing that almost all radioactivity was distributed in proportion to the UV absorbance of large and small ribosomal subunits, when the total membranes were treated with puromycin in high salt buffer (Fig. 8C,D). These data provide independent evidence for the conclusion that dexamethasone treatment does not cause an increase in the size of the RER in AR42J cells.

The results from our biochemical analysis led us to study the effect of dexamethasone on the morphology of AR42J cells. It was found that both control and dexamethasone-treated cells grew as small three-dimensional aggregates, rather than as monolayers. However, the size of the aggregates was smaller after drug treatment. At the ultrastructural level the most striking difference between control and dexamethasone-treated cells was the appearance of secretory granules, and a change in the structure of the RER. Whereas in control cells secretory granules were rarely seen and the RER appeared as tubules, vesicles and isolated small cisternae (Fig. 9A), in dexamethasone-treated cells the secretory granules were very prominent and the RER appeared predominantly as large stacked cisternae. Laser scanning confocal microscopy showed that the cells were partially polarized and that amylase-containing granules were mainly localized at the apical pole (Rajasekaran et al., unpublished results). Although the cisternal stacks were already seen after 24 h of dexamethasone treatment (Fig. 9B), they became very prominent at 72 and 96 h (Fig. 9C,D) after addition of the drug. Pretreatment of the cells with the antiglucocorticoid agent RU 38486 resulted in a control phenotype (Fig. 9E), indicating that the effect of dexamethasone was specific and was mediated by the glucocorticoid receptor. No significant changes in the morphology of other organelles were detected.

In order to determine whether dexamethasone treatment of AR42J cells causes changes in the cellular distribution of the ER, laser scanning confocal microscopy of control and treated cells was performed on cells labelled with an antibody specific for the ER. In en face views of control cells, the endoplasmic reticulum had a rather uniform distribution with an apparent concentration in the perinuclear region that is caused, in part, by the increased cell thickness at this level (Fig. 10a). After dexamethasone treatment, however, the ER staining was not uniform, but appeared to be concentrated in discrete regions of the cell (Fig. 10c). In side views of control cells, obtained by a 90° rotation, the ER appeared to be preferentially localized at the base (Fig. 10b), while in dexamethasone-treated cells, ER patches were observed at various levels (Fig. 10d). These patches may represent the clumps of SC-RER observed at the ultra-structural level (Fig. 9C,D).

This report demonstrates that a cell can drastically increase its protein-synthetic and secretory activity without an increase in the size of its RER or of its bound polysome population. The increased rate of secretory protein production is accompanied by the reorganization of the RER from a tubulo-vesicular/small cisternal configuration (TV-RER) to that of large stacked cisternae (SC-RER) characteristic of ‘professional’ secretory cells. These results suggest that the organization of the RER into large cisternae allows secretory proteins made on bound polysomes to be synthesized at a higher rate.

Dexamethasone treatment of rat pancreatic acinar carcinoma AR42J cells resulted in a 6.6-fold increase in intra-cellular amylase and a 4.4-fold increase in secreted amylase activity. Since amylase constitutes about 23% of the newly synthesized proteins and dexamethasone increases the synthesis and secretion of several other secretory proteins (Swarovsky et al., 1988), the observed 2.4-fold increase in total protein synthesis may be primarily due to an increased synthesis of secretory proteins. The greater capacity of these cells to synthesize secretory proteins appears to be in part due to increased levels of the corresponding mRNAs (Swarovsky et al., 1988); in the case of amylase it is clear that the effect of dexamethasone on the mRNA content is at the transcriptional level (Logsdon et al., 1987).

Do the increased levels of secretory protein synthesis reflect an increased amount of RER and bound polysomes or an increased efficiency of the existing ribosomes without an increase in their number?

To address this problem it was necessary to utilize a reliable method to quantify the RER after induction of the secretory activity by dexamethasone. In previous work on AR42J cells the size of the RER was determined using morphometric procedures (Logsdon et al., 1985; Swarovsky et al., 1988). These methods, while reliable, are tedious and laborious and are subject to considerable sampling errors. We utilized, instead, two biochemical procedures to determine the size of the RER. The first involved the measurement of RNA and protein levels of ribophorins by northern and western blot analyses, respectively. Ribophorins I and II are present in the RER in a 1:1 ratio with respect to bound ribosomes (Marcantonio et al., 1984), to which they can be cross-linked (Kreibich et al., 1978b). Recent evidence indicates that they are components of the oligosaccharyl transferase (Kelleher et al., 1992), which is part of the protein-translocating apparatus in the RER. Furthermore, a good correlation has been shown between ribophorin levels and the amount of RER in CH12 cells (Wiest et al., 1990) and in B lymphocytes at different stages of development (Zhou et al., unpublished results). Our results showed no change in the levels of ribophorins I and II after induction with dexamethasone, and therefore suggest that the number of ribosomes bound to the RER did not change in spite of an induction of secretory activity by dexamethasone.

To determine directly the effect of dexamethasone on the amount of bound polysomes, we utilized sucrose density gradient analysis of a total cell membrane fraction, from which ribosomes are released with puromycin in a buffer of high ionic strength. This is a very sensitive method that can detect changes of only 0.25-fold in the size of the RER, as measured by the amount of bound polysomes recovered form microsomes (Rajasekaran et al., unpublished). Using this procedure the 4.2-fold increase in the population of membrane-bound ribosomes in Xenopus laevis hepatocytes, caused by steroid hormone treatment, is in good agreement with a 4-fold increase obtained by morphometric analysis performed on hepatocytes from a different species, Rana pipiens (Herbener et al., 1983). A similar correlation between an increase in the size of the RER and the amounts of bound polysomes was obtained by stereological analysis of thin-section electron micrographs of hepatocytes derived from rats treated with phenobarbital (Staubli et al., 1969). Our results, both on the quantification of ribophorin levels and the direct determination of the amount of membrane-bound ribosomes support the conclusion that the large steroid-induced increase in secretory activity occurs without an enlargement in the population of bound ribosomes.

We then considered the possibility that dexamethasone treatment causes an increase in the size of the RER surface area without a change in the amount of membrane-bound ribosomes. However, this would result in a decrease in the buoyant density of the rough microsomal fraction. Using sucrose density gradient analysis Wibo et al. (1971) have shown a correlation between the density of rat liver microsomes and the number of bound ribosomes associated with the microsomal vesicles (see also Amar-Costesec et al., 1984). Since the total membrane fractions from control and treated AR42J cells had a very similar isopycnic density distribution in sucrose density gradients (Fig. 8), this possibility was ruled out. In addition, a significant increase in the amount of the bound ribosomes would have been easily detected by a biochemical determination of total RNA. Our analysis of the RNA content shows no significant difference between control and dexamethasone-treated cells (Fig. 1b). Finally, since the ratio of free to membrane-bound ribosomes in dexamethasone-treated cells is 1:1 (Fig. 7), which is the same as in control cells, the possibility that free ribosomes are converted into bound ribosomes can be excluded. On the basis of all of these criteria we conclude that dexamethasone treatment of AR42J cells causes a large increase in the synthesis and secretion without an increase in the size of the RER.

Our results thus contradict those reported by Logsdon et al. (1985) and Swarovsky et al. (1988), who, respectively, reported 2.4-fold and 30-fold increases in RER, using morphometric analysis. The reason for the different results obtained by these two groups, or between these groups and us is not clear. Biased sampling or too small a number of samples can lead to a considerable error; these factors were not extensively discussed by these authors. The fact that two independent biochemical procedures failed to detect an increase in the amount of bound polysomes and that no shift in the density distribution of the bound polysomes observed are very strong arguments against an increase in the size of the RER.

The increase in secretory activity of dexamethasone-treated AR42J cells must, therefore, reflect a biosynthetically more efficient RER. On the other hand our morphological results clearly show that this increased secretory activity is accompanied by a dramatic structural reorganization and redistribution of the RER. Whereas the RER in untreated cells was organized mainly as tubules, vesicles and small cisternae (TV-RER) (Fig. 9A), 24 h after addition of dexamethasone, when the induction of secretion was almost maximal (Fig. 3A,B), both tubules and smaller cisternae had aligned to form the SC-RER (Fig. 9B), which became the predominant RER type at later times (Fig. 9C,D). At the whole-cell level, laser scanning confocal microscopy with antibodies against ER membrane proteins showed a redistribution of the ER in dexamethasone-treated cells (Fig. 10c,d). Whereas, in control cells the ER staining was concentrated mostly in basal regions of the cell, after drug treatment the ER staining appeared as discrete ‘patches’ at various cell levels, presumably corresponding to the stacks of SC-RER observed at the ultrastructural level. A similar ‘patchy’ distribution of the SC-RER has been reported in estrogen-treated Xenopus male livers (Bergink et al., 1977; Herbener et al., 1983; Rajasekaran et al., unpublished) and in the posterior silk gland of the silk worm during the fourth and fifth instars of the larval stage (Morimoto et al., 1968).

The dexamethasone-induced increase in amylase activity and reorganization of the RER were abolished by pretreatment of the cells with RU 38486, an antiglucocorticoid agent (Figs 1A,2 and 9E), suggesting that these processes are mediated by a glucocorticoid receptor. At the moment it is impossible to decide whether the increase in the synthesis of secretory proteins acts as a signal for the reorganization of the RER, or whether a common factor regulates both processes via the glucocorticoid receptor. The alignment of RER elements (Fig. 9B) suggests the participation of cytoskeletal elements that is regulated by dexamethasone and may be responsible in the generation and maintainance of the SC-RER (Rajasekaran et al., unpublished results).

In the vast majority of actively secreting cells, or when the secretory state is induced, the RER appears as SC-RER (Weiss, 1988; Bergink et al., 1977; Herbener et al., 1983; Pacifici and Iozzo, 1988; Wiest et al., 1990,Morimoto et al., 1968; Rajasekaran et al., unpublished results). On the basis of these and our results, we postulate that the SC-RER of secretory cells represents a more efficient arrangement of the components of the translation and translocation apparatus, thus allowing for an increase in the RER output of secretory products. This hypothesis is based on the observation that the amount of bound polysomes remains the same in spite of a dramatic increase in the synthesis of secretory proteins. Two possible explanations may account for this observation. It is possible that some proteins that are synthesized at significant levels prior to dexamethasone treatment are either not synthesized or synthesized at a much lower level after dexamethasone treatment; these ribosomes could then be utilized for the dexamethasone-induced synthesis of secretory proteins. Another possibility is that a sufficient synthetic capacity is present within the existing bound polysome population. Our preliminary results support the latter view. When the total membranes from control and dexamethasone-treated cells were briefly treated with RNase about 2-fold more ribosomes were released from the control membranes, suggesting that these ribosomes are bound to the membrane via their mRNA but not through the nacent chain and, hence, not active in protein synthesis (Rajasekaran et al., unpublished results). A structural reorganization of the RER to become SC-RER may result in a more ordered distribution of these ribosomes and thus increase the efficiency of protein synthesis.

Furthermore, reorganization of TV-RER into SC-RER may also represent a faster and energetically more efficient mechanism of responding to increased demands for secretory products. Sea urchin eggs reorganize their non-cortical RER from a cisternal type to a more finely divided RER and back to a cisternal RER within 5-8 min of fertilization (Terasaki and Jaffe, 1991). In AR42J cells the reorganization of the ER was evident already within 24 h of dexamethasone treatment (Fig. 9B). In cases where the synthesis of new RER is well documented, this process usually takes much longer, e.g. more than 2 days in B cells induced by LPS (Wiest et al., 1990) and 5-15 days in frog hepatocytes induced with estrogen (Bergink et al., 1977; Herbener et al., 1983; Rajasekaran et al., unpublished).

We have presented strong evidence that a dramatic reorganization of the RER from a tubulo-vesicular into a large stacked cisternal form, rather than an increase in the size of the RER, accounts for the increased protein synthetic and secretory activity of AR42J cells stimulated by a glucocorticoid. These results suggest an explanation for the pre-dominance of this form of RER in actively secreting cells. Additional work is needed to identify cytosolic factors that control this striking transformation of the RER.

We thank Dr David D. Sabatini for encouragement and support. We also thank Dr Craig Logsdon and Dr Horst Kern for providing us with AR42J cells, Dr Daniel Louvard for ER antibody and Dr Philbert for RU 38486. Special thanks to Iwona Gumper, who carried out the thin-section electron microscopy. The help of J. Culkin, F. Forcino and H. Plesken with the preparation of the illustration is gratefully acknowledged. The help of the Cornell University Medical Art and Photography facility in the preparation of photographs is acknowledged. We are grateful to Dr Diego Gravotta and members of the G.K. Lab. for helpful discussions.

This work was supported by National Institute of Health grants GM21971 (G.K.), GM20277 (Dr David D.Sabatini) and GM 34107 (E.R.B.) and by a grant from the American Cancer Society (CD-514; G.K.).