Sites of lipoprotein particles in normal rat hepatocytes.

Very low density lipoprotein (VLDL) particles are packaged by the Golgi apparatus into vacuoles which move to the plasma membrane and empty the particles into the space of Disse, via exocytosis. Traditionally, all lipoprotein-containing cisternae and vacuoles are thought to be parts of this pathway. Observations reported here demonstrate that there is a second population of lipoprotein-containing cisternae and vacuoles. This population is part of GERL, an organelle we consider to be a specialized hydrolase-rich region of the endoplasmic reticulum (ER). To our knowledge, this is the first systematic study of GERL in normal rat hepatocytes.

Molecular trapping of a fluorescent ceramide analogue at the Golgi apparatus of fixed cells: interaction with endogenous lipids provides a trans-Golgi marker for both light and electron microscopy.

We have previously shown that a fluorescent derivative of ceramide, N-(epsilon-7-nitrobenz-2-oxa-1,3-diazol-4-yl-aminocaproyl)-D-eryth ro-sphingosin e (C6-NBD-Cer), vitally stains the Golgi apparatus of cells (Lipsky, N. G., and R. E. Pagano. 1985. Science (Wash. DC). 228:745-747). In the present paper we demonstrate that C6-NBD-Cer also accumulates at the Golgi apparatus of fixed cells and we explore the mechanism by which this occurs. When human skin fibroblasts were fixed with glutaraldehyde and then incubated with C6-NBD-Cer at 2 degrees C, the fluorescent lipid spontaneously transferred into the cells, labeling the Golgi apparatus as well as other intracellular membranes. Subsequent incubations with defatted BSA at 24 degrees C removed excess C6-NBD-Cer from the cells such that fluorescence was then detected only at the Golgi apparatus. Similar results were obtained using other cell types. A method for visualizing the fluorescent lipid at the electron microscopic level, based on the photoconversion of a fluorescent marker to a diaminobenzidine product (Sandell, J. H., and R. H. Masland, 1988. J. Histochem. Cytochem. 36:555-559), is described and evidence is presented that C6-NBD-Cer was localized to the trans cisternae of the Golgi apparatus. While accumulation occurred in cells fixed in various ways, it was inhibited when fixation protocols that extract or modify cellular lipids were used. In addition, Filipin, which forms complexes with cellular cholesterol, labeled the Golgi apparatus of fixed cells and inhibited accumulation of C6-NBD-Cer at the Golgi apparatus. These results are discussed in terms of a simple model based on the physical properties of C6-NBD-Cer and its interactions with endogenous lipids of the Golgi apparatus. Possible implications of these findings for metabolism and transport of (fluorescent) sphingolipids in vivo are also presented.

Studies of the secretory process in the mammalian exocrine pancreas. I. The condensing vacuoles.

Phosphatase cytochemistry was used to distinguish between the Golgi apparatus and GERL (considered as a specialized region of endoplasmic reticulum [ER] at the inner [trans] aspect of the Golgi stack) in pancreatic exocrine cells of guinea pig, rat, rabbit, and hamster. The trans element of the Golgi stack exhibits thiamine pyrophosphatase (TPPase) but no acid phosphatase (AcPase) activity. In contrast, GERL shows AcPase but no TPPase activity. The nascent secretory granules, or condensing vacuoles, are expanded cisternal portions of GERL. Continuities of condensing vacuoles with rough ER are suggested, and it is proposed that some secretory components may have direct access to the condensing vacuoles from ER. Connections of Golgi apparatus with GERL were not seen.

Subcellular location of stimulus-affected endogenous phosphoproteins in the rat parotid gland.

Rat parotid minces were labeled with [32P]Pi, stimulated with isoproterenol, homogenized in sucrose, and fractionated on continuous sucrose density gradients. We analyzed the resulting fractions by sodium dodecyl sulfate polyacrylamide gel electrophoresis and autoradiograms were made from the gels. Comparison of fractions from control and isoproterenol-stimulated minces revealed seven phosphoproteins that were affected by isoproterenol. The subcellular location of these proteins was determined by comparing their distribution in the sucrose gradients with that of a number of enzymes that are characteristic of specific organelles. Isoproterenol decreased the phosphorylation of two cytoplasmic proteins (Mr 16,000 and 18,000) and increased the phosphorylation of a third (Mr 14,000). The phosphorylation of two endoplasmic reticulum proteins was increased by isoproterenol (Mr 20,500 and 22,500), as was an Mr 31,000 protein which was probably the S6 ribosomal protein. The phosphorylation of a secretory granule protein (Mr 24,000) was decreased by isoproterenol. We then developed a purification scheme for parotid secretory granules. By using this method, we demonstrated that the phosphorylation of the Mr 24,000 was also decreased by carbamylcholine. Granules purified by this method also contained a small number of other phosphoproteins whose phosphorylation was increased only by isoproterenol. Secretory granule-associated stimulus-affected phosphoproteins were found in the particulate fraction when the granules were hypotonically lysed, and were not extracted from the particulate fraction by washing with 0.6 M KCl.

Topography of cerebroside sulfotransferase in Golgi-enriched vesicles from rat brain.

Cerebroside sulfotransferase (CST) catalyzes the final step in the synthesis of sulfatide (sulfogalactocerebroside) by transferring the sulfate from 3'-phosphoadenosine 5'-phosphosulfate (PAPS) to galactocerebroside. Orientation of CST was studied in vesicles enriched in this enzyme obtained from 21-d-old rat brain. Several lines of evidence indicate that CST is located on the luminal side of these vesicles. (a) Sulfation of endogenous galactocerebroside occurred in vesicles only in the presence of a detergent to render the membranes permeable to exogenous PAPS. (b) There is a pool of latent enzyme within the vesicle, which is released by Triton X-100. (c) CST is not destroyed by trypsin unless the vesicle membranes are first made permeable by Triton X-100. (d) Glycolipid substrate, when covalently attached to agarose beads, was not sulfated unless the enzyme was solubilized. These results are similar to those obtained with thiamine pyrophosphatase, which is known to be located within the lumen of the vesicles. This study establishes that an enzyme synthesizing a complex glycolipid is localized within Golgi-enriched vesicles. Since the product of the CST reaction must also be localized to the luminal side of the vesicles, it is most likely that sulfatide is located at the intraperiod line (outer layer) of myelin. The orientation of CST within the vesicle provides a mechanism for the asymmetrical assembly of glycolipids in bilayers.

The corpus luteum of the guinea pig. II. Cytochemical studies on the Golgi complex, GERL, and lysosomes in luteal cells during maximal progesterone secretion.

This study characterizes the cytochemical properties of the Golgi complex, the structure which corresponds to Golgi complex-endoplasmic reticulum-lysosomes (GERL), and the granule population in luteal cells of guinea pigs at the time of maximum progesterone secretion, in material fixed by vascular perfusion, a method particularly suited for preserving both fine structure and enzyme activity. The distribution of several marker enzymes was determined by electron microscope cytochemistry. Acid phosphatase (ACPase) and arylsulfatase were used to identify structures containing lysosomal proteins. To resolve specific problems, additional cytochemical markers were employed: localization of thiamine pyrophosphatase (TPPase) (in the Golgi complex) and alkaline phosphatase (ALPase) (a plasma membrane marker), and prolonged osmication (a generally accepted method of marking the outer cisterna of the Golgi complex). The results demonstrate that at the time of peak steroid secretion the Golgi complex in luteal cells, in marked contrast to that of most other cell types, typically displays intense ACPase activity in all of its cisternae. Similarly, all Golgi cisternae stain after prolonged osmication and may show TPPase activity. On the other hand, GERL in luteal cells of this age, unlike that in most cells, commonly shows low levels of, or lacks, ACPase activity. However, GERL resembles that of other cell types in being TPPase-negative and in being unstained by treatment with aqueous OsO4. GERL and some Golgi cisternae are reactive for ALPase. The granule population in luteal cells of this stage consists of lysosomes, multivesicular bodies, electrontransparent vacuoles, and microperoxisome-like bodies. These results form a base line with which luteolytic changes described in the companion study (Paavola, L.G. 1978. The corpus luteum of the guinea pig. III. Cytochemical studies on the Golgi complex and GERL during normal postpartum regression of luteal cells, emphasizing the origin of lysosomes and autophagic vacuoles. J. Cell. Biol. 79:59--73.) can be compared.

Cytochemical and stereological analysis of rat cortical astrocytes during development in primary culture. Effect of prenatal exposure to ethanol.

This study has investigated the effect of prenatal alcohol exposure on the qualitative and quantitative ultrastructure of proliferating and differentiated astrocytes in primary cultures as well as on the cytochemical activity of several subcellular phosphatase markers, including acid phosphatase, uridine diphosphatase, thiamine pyrophosphatase, 5'-nucleotidase and glucose-6-phosphatase. The astrocytes were obtained from 21-day-fetuses of both control and alcohol-fed rats. Our results show that several cell components, such as mitochondria, rough endoplasmic reticulum and lysosomes, exhibit qualitative and/or quantitative ultrastructural changes during the process of astrocyte maturation. In some cases these morphological changes are accompanied by variations in the cytochemical activity of enzymes located in these and other cell components, suggesting that these enzymes, and therefore the functional state of these organelles, are modulated during astrocyte development. When prenatally exposed to ethanol, both proliferating and differentiated astrocytes showed striking ultrastructural alterations compared with controls, including an increment of lysosomes as well as a decrease in the values of stereological parameters relative to mitochondria, rough endoplasmic reticulum and Golgi apparatus. Cytochemical analysis of these cells indicates that prenatal exposure to ethanol decreased the activities of all the enzymes tested, except for acid phosphatase, which was increased in both groups of treated astrocytes. These results suggest that prenatal exposure to ethanol could affect astrocytes during development in two different but probably complementary ways: a) by causing a delay in astrocyte maturation and, b) by inducing a direct toxic effect on these cells.

Effects of brefeldin A on autophagy in cultured rat fibroblasts.

Effects of brefeldin A on cellular autophagy were studied in cultured rat fibroblasts. Brefeldin A inhibits the activation and membrane-binding properties of most ADP-ribosylation factors and causes the redistribution of Golgi proteins into the endoplasmic reticulum. Immunofluorescence and enzyme cytochemical methods revealed the disappearance of the Golgi apparatus and trans-Golgi network during the brefeldin A incubation. The volume fractions of autophagic vacuoles increased about threefold in cells treated with brefeldin A for 4 h and about sixfold in serum-deprived cells as compared with controls. When cells were first treated with brefeldin A for 1 h and were then deprived of serum and treated with brefeldin A for 3 h, the volume fraction of autophagic vacuoles increased about 4.5-fold as compared with untreated cells. The results showed that brefeldin A is unable to prevent serum deprivation-induced accumulation of autophagic vacuoles and that brefeldin A even when acting alone increases the volume fraction of autophagic vacuoles. It was concluded that an intact Golgi apparatus and trans-Golgi network are not essential for the formation of autophagic vacuoles. It seems also probable that ADP-ribosylation factors are not needed when vacuoles are formed.

Presence of Golgi remnant membranes in the cytoplasm of brefeldin A-treated cells.

Brefeldin A (BFA) rapidly blocks secretion, induces disassembly of the Golgi complex and causes a redistribution of Golgi components into the endoplasmic reticulum (ER). In addition to these effects on the exocytotic pathway, BFA has been shown to induce fusion of endosomal membranes with the trans-Golgi network in some cell types. To better understand the mechanism through which BFA disrupts the exocytotic traffic, we have examined its effects on the ultrastructural organization of the Golgi complex. Within minutes of exposure to BFA, the Golgi cisternae were fragmented into a number of small tubules and vesicles, many of which had a non-clathrin coat on their cytosolic surface. In addition, a complex structure consisting of anastomosing tubules and associated vesicles appeared in the cytoplasm of cells incubated with BFA for 10 min or longer. These tubular networks were permanent, distinct structures separated from the ER cisternae. They contained cis, middle, and trans Golgi proteins as well as the lipid analogue C5-DMB-ceramide. Furthermore, secretory proteoglycans en route through the Golgi were retained in the lumen of the tubular networks. As judged by the endocytosis of cationized ferritin, endosomes do not contribute to the formation of these tubular networks. Reassembly of the Golgi complex after BFA incubation involved fragmentation and reorganization of the tubular networks as well as fusion with vesicles budded from the ER. We conclude that although in the presence of BFA the bulk of Golgi membranes are induced to fuse with the ER, as indicated by the detection of Golgi markers in this organelle, a fraction of these membranes remain in the cytoplasm organized as Golgi remnants.

Acrolein as a fixative for enzyme cytochemistry.

Since acrolein can penetrate more quickly and deeply into tissue blocks than glutaraldehyde, the possibility of the use of this aldehyde as a prefixative in enzyme cytochemistry was reinvestigated. At low concentrations, acrolein preserves the activities of the enzymes investigated, including those of glucose-6-phosphatase, which is known as one of the most vulnerable to aldehyde fixation; thus, acrolein is usable in enzyme ultracytochemistry. Enzyme activities are also preserved in tissues fixed with acrolein and glutaraldehyde combined. The rapid penetration of acrolein enables fixation in larger tissue blocks and provides greater freedom in specimen selection, especially important advantages when encountering heterogeneous materials as in pathology.

Ultrastructural localization of phosphatase activity in the guinea pig pineal gland by the cerium technique.

Thiamine pyrophosphatase (TPPase), nucleoside diphosphatase (NDPase), and glucose-6-phosphatase (G-6-Pase) were localized by the cerium technique in guinea pig pinealocytes and compared with the corresponding lead technique. NDPase and TPPase were also compared at different pH values using the cerium technique. Vibratome sections of perfusion-fixed tissue were incubated with cerium chloride or lead nitrate. Substrates used were thiamine pyrophosphate (for TPPase), sodium inosine diphosphate (NDPase), and disodium glucose-6-phosphate (G-6-Pase). The 1-2 trans saccules of the Golgi apparatus showed TPPase and NDPase activity but none for G-6-Pase. The endoplasmic reticulum (ER) cisternae and perinuclear space had NDPase and G-6-Pase activity but not TPPase. The abluminal plasmalemma of endothelial cells and the plasmalemma of Schwann cells demonstrated TPPase and NDPase activity but the luminal plasmalemma of the endothelial cells and the plasmalemma of pinealocyte processes showed only NDPase activity. TPPase was active at all pH values tested, but NDPase was most active at pH values of 6.5 and 7.0. Lead phosphate precipitate was frequently seen in nuclei, perinuclear space, ER cisternae, and "synaptic" vesicles when lead was used as the capturing agent. These sites were usually not labeled when cerium was used.

Cytochemical localization of beta-NADPase in rat hepatocytes and Kupffer cells. Comparison with thiamine pyrophosphatase (TPPase).

The intracellular localization of beta-NADPase in rat hepatocytes and Kupffer cells has been studied and compared with the pattern of TPPase in these cells. The reaction product for beta-NADPase is present in some but not all hepatocytes in two cisternae on the trans aspect of the Golgi apparatus. It is absent from the trans-most lamella and the GERL of hepatocytes. TPPase, on the other hand, is limited to the first Golgi cisterna on the trans aspect with sprinkles of reaction product in the second lamella. Considering that TPPase is a marker of the trans Golgi lamella and hepatocyte Golgi stacks contain usually 2-4 lamellae, our observations suggest that beta-NADPase is localized in the trans as well as in the intermediate Golgi lamellae of liver parenchymal cells. In Kupffer cells, the reaction product for both beta-NADPase and TPPase was found in some but not in all cells. The enzyme beta-NADPase was localized in the rigid lamella and the tubulovacuolar system of GERL. This pattern differed significantly from that for TPPase, which was found in 2-3 cisternae at the trans aspect of the Golgi complex in Kupffer cells. These observations demonstrate the difference in the localization of beta-NADPase in hepatocytes and Kupffer cells. Such differences should be taken into consideration in studies of Golgi fractions, when phosphatase reactions are used as specific markers of Golgi components.

Enzyme modulation of the Golgi apparatus and GERL: a cytochemical study of parotid acinar cells.

The distribution of thiamine pyrophosphatase (TPPase) and acid phosphatase (AcPase) has been examined in resting parotid acinar cells as well as during decreased and increased secretory granule production. In resting acinar cells, TPPase activity was restricted to the trans Golgi saccules and AcPase activity was localized in GERL and immature secretory granules. Although secretory granule production is diminished during ethionine intoxication, no significant alteration in the distribution of either TPPase or AcPase was noted. However, marked changes in enzyme localization, especially of TPPase, occurred during accelerated secretory granule production. The alterations were essentially the same for all of the conditions studied (recovery from ethionine treatment, recovery from a protein depletion diet, secretory stimulation with isoproterenol, and postnatal maturation of the parotid gland). During maximal secretory granule production, TPPase activity was localized not only in the trans Golgi saccules, but also in GERL-like cisternae and immature secretory granules. The immature secretory granules were often in continuity with the GERL-like cisternae. At the same time that the TPPase activity was increased, the AcPase activity was frequently diminished. These modulations in enzyme activity provide evidence that GERL is derived from the trans Golgi saccule.

The cytochemical demonstration of GERL in rat hepatocytes during lipoprotein mobilization.

When a semisynthetic diet containing 1% orotic acid (OA) is fed to rats, the endoplasmic reticulum (ER) of hepatocytes vesiculates and lipoprotein (LP) droplets accumulate within the vesicles. When clofibrate (ethyl chlorophenoxyisobutyrate, CPIB) is added to the orotic acid-rich diet, the ER cisternae reform and the LP is mobilized through the reconstituted ER. A remarkable restoration of normal hepatocyte ultrastructure occurs except for a few organelles. From their morphological appearance it was suggested that cisternae which became dilated with small LP particles were part of GERL, abnormally enlarged. The present communication validates this interpretation through ultrastructural cytochemistry which can distinguish GERL from the adjacent Colgi apparatus. GERL shows acid phosphatase (AcPase) but not thiamine pyrophosphatase (TPPase) activity. In contrast, the adjacent Golgi element shows thiamine pyrophosphatase but not acid phosphatase activity. From such cytochemical studies we have recently proposed that GERL in normal rat hepatocytes may be involved in transforming LP particles, by enzymes like lipases that were presumed to be present in this hydrolase-rich portion of smooth ER. In the situation studied in this communication, the addition of ethyl chlorophenoxyisobutyrate to the diet causes the release from the ER of large amounts of LP to the Golgi apparatus and to GERL. Apparently the capacity of GERL to metabolize LP is exceeded and lipid accumulates in the residual bodies.

Phosphatase localization in the endomembrane system of the dinoflagellate Crypthecodinium cohnii.

The distribution of four enzymes within the endomembrane system of the protist Crypthecodinium cohnii has been determined using cytochemical localizations with lead as a capture agent. Nucleoside diphosphatase (NDPase) activity, using inosine diphosphate (IDP) and thiamine pyrophosphate (TPP) as substrates, was observed in the Golgi apparatus, with a gradient of increasing reaction product noted in some cells from the cis to trans cisternae. Tubules and vesicles associated with the trans cisternae also contained reaction product. The endoplasmic reticulum exhibited a high activity of glucose-6-phosphatase [with glucose-6-phosphate (G-6-P) as substrate]. Traces of reaction product were also observed in the cis-most and trans-most cisternae of the dictyosomes. Activity of acid phosphatase (AcPase) was observed in Golgi cisternae as well as in associated cytoplasmic vesicles. Heaviest deposition was localized in medial and trans dictyosome cisternae. The cytoplasmic system of flattened vesicles subtending the surface membranes in these cells did not exhibit reactivity with any of the substrates used. The distribution of these enzymes in this algal cell appears similar to that observed in animal cells and suggests that these enzymes may represent markers for algal cell endomembrane compartments.

Prenatal exposure to alcohol alters the Golgi apparatus of newborn rat hepatocytes: a cytochemical study.

The effect of prenatal exposure to ethanol on the Golgi apparatus of newborn rat hepatocytes has been studied cytochemically using several trans-Golgi markers (thiamine pyrophosphatase, uridine diphosphatase, inosine diphosphatase, acid phosphatase, and 5'-nucleotidase) as well as a cis-side marker (osmium impregnation). The amount of cerium phosphate formed in the cytochemical reactions was roughly quantitated by stereologic methods. The Golgi apparatus of about 40% of the hepatocytes appeared disorganized after alcohol treatment, and in the other 60%, the electron density of reaction product deposits for all phosphatases investigated was decreased. 5'-Nucleotidase was completely absent in cisternae of Golgi apparatus of treated cells. In control cells impregnated with osmium tetroxide, reduced osmium compounds were observed in most Golgi cisternae and in nearby vesicles. In contrast, only small vesicles appeared positive in treated hepatocytes. These results suggest that prenatal alcohol exposure alters some Golgi functions. Thus, the decrease in nucleoside diphosphatase and 5'-nucleotidase cytochemical activities after ethanol exposure strongly suggests that this treatment could affect glycosylation in the Golgi apparatus of newborn rat hepatocytes.

Thiamine pyrophosphatase (nucleoside diphosphatase) in the Golgi apparatus is distinct from microsomal nucleoside diphosphatase.

The properties of thiamine pyrophosphatase in the Golgi apparatus of rat liver were studied. Thiamine pyrophosphatase in an extract of the Golgi apparatus was separated into 6 bands of between pH 5.4 and 6.3 by isoelectric focusing on polyacrylamide gel. On the gels all these subforms catalyzed the hydrolyses of GDP, IDP, UDP, and CDP as well as that of thiamine pyrophosphate. The characteristics resembled those of Type B nucleoside diphosphatase of rat brain, though the enzyme did not have 3 subforms of Type B nucleoside diphosphatase in the higher pH region on isoelectric focusing. Thiamine pyrophosphatase of the Golgi apparatus was separated from microsomal nucleoside diphosphatase by DEAE-cellulose column chromatography. The properties of the enzyme were quite similar to those of Type B nucleoside diphosphatase with respect to its substrate specificity, optimum pH for activity, and inhibition by ATP. These findings suggest that thiamine pyrophosphatase in the Golgi apparatus is different from microsomal nucleoside diphosphatase and that it might be basically the same enzyme as Type B nucleoside diphosphatase except for different extents of modification.

Localization of vasoactive intestinal polypeptide- and avian pancreatic polypeptide-like immunoreactivity in the Golgi apparatus of peripheral neurons.

Using the indirect immunofluorescence technique of Coons and collaborators, vasoactive intestinal polypeptide (VIP)-like immunoreactivity (VIP-LI) and avian pancreatic polypeptide (APP)-like immunoreactivity (APP-LI) was observed in certain neurons of the peripheral nervous system of the cat. In the cell bodies in the cat sympathetic ganglia a strong to very strong VIP-LI or APP-LI was observed with a distribution resembling that of the Golgi apparatus. In addition, a weaker immunoreactivity was seen diffusely in the cytoplasm. After photography, the sections processed for immunohistochemistry were stained with the thiamine pyrophosphatase technique of Novikoff and Goldfischer. The latter technique is assumed to be a specific marker for the Golgi complex. In all cases it was found that the strong peptide immunoreactivity and the thiamine pyrophosphatase activity had an identical distribution. Thus, one pool of these peptides appears to be localized to the Golgi apparatus. Yet another pool is localized to other components, such as vesicles.

Autophagy-related vacuoles in mouse gallbladder epithelium.

The mouse gallbladder epithelial cells contain very heterogeneous vacuolar population. In an attempt to classify these vacuoles we identified NADPase and TPPase activity as well as the location of HRP which is used as the endocytotic marker. The results of the present study show that the vacuoles can be classified into three categories: (1) the vacuoles predominantly containing loose membrane coils related to the nascent autophagic vacuoles, (2) vacuoles containing densely packed membranes and exhibiting a positive HRP reaction, indicating the convergence of endocytotic and autophagic pathway, and (3) vacuoles composed of degraded membrane structures and containing the reaction product of NADPase activity, showing that the fusion of the lysosomes with the autophagosome-endosome took place. The highly developed cis, medial and trans Golgi compartments reflect the biosynthetic and endocytotic activity of the gallbladder epithelium.

Distribution of adenosine triphosphatase and thamine pyrophosphatase in the digestive system of Heteropneustes fossils.

Histochemical localization of adenosine triphosphatase and thiamine pyrophosphatase in the digestive system of the teleost fish, Heteropneustes fossilis has been studied. In the stomach, ATPase activity is observed in the mucosa, gastric glands and muscularis. The activity is stronger in the muscularis. Very weak TPPase activity is localized only in the mucosa and gastric glands. In the intestinal mucosa ATPase activity is stronger especially, along the brush border. Mild activity is also found in the connective tissue network and their nuclei, muscularis and serosa. In the posterior portion of the intestine and rectum, the localization pattern is similar to that of intestine but the activity is weaker. TPPase activity in the intestine and rectum is restricted only to the goblet shaped mucus secreting cells. In the liver, strong activity of ATPase and moderate activity of TPPase are found in the cytoplasm as well as the nuclei of the hepatic cells.