Inhibitory effect of phenyl N-terf-butyl nitrone on Kupffer cell phagocytosis

Summary

Light microscopy studies of rat liver were conducted after injection of India ink. The data indicated that Kupffer cell phagocytosis was inhibited by C-phenyi-N-tert-butyl nitrone (PBN), as well as by the Kupffer cell antagonist gadolinium chloride.     

Ultrastructural study of the endothelial cells in teleost liver sinusoids under normal and experimental conditions

Summary The ultrastructure of the endothelial cells of liver sinusoids was studied in the teleost, Pimelodus maculatus. These cells have the ability to form pinocytotic vacuoles, starting with the formation of marginal folds. The latter occur in many cells after stimulation by India ink injections and ink particles are ingested by pinocytosis and by micropinocytosis. Desmosomes, structures rarely described between liver sinusoidal endothelial cells, are present in this species.     

In vitro induction of tumor necrosis factor-α by ochratoxin A (OTA) from rat liver: role of Kupffer cells

Tumor necrosis factor-α (TNF-α) is released from blood-free perfused rat liver by the fungal metabolite ochratoxin A. Here we have identified Kupffer cells as the sole source of OTA-mediated cytokine release. If single cell preparation of Kupffer cells, hepatocytes, or sinusoidal endothelial cells were prepared from rat livers, only Kupffer cells released TNF-α upon incubation with 2.5 μmol/l OTA. OTA failed to induce TNF-α release in the blood-free perfused isolated rat liver when Kupffer cells were blockedin vitro by 15 μmol/l gadolinium chloride. When rats were pretreatedin vivo with the Kupffer cell depleting clodronate liposomes, OTA-mediated TNF-α release was abrogated in the isolated perfused liver model.     

Intrahepatic uptake and processing of intravenously injected small unilamellar phospholipid vesicles in rats

Small unilamellar vesicles consisting of sphingomyelin, cholesterol and phosphatidylserine in a molar ratio of 4:5:1 containing [³H]inulin as a marker of the aqueous space or [Me-¹⁴C]choline-labeled sphingomyelin as a marker of the lipid phase were injected intravenously into rats. After separation of the non-parenchymal cells into a Kupffer cell fraction and an endothelial cell fraction by elutriation centrifugation analysis of the radioactivity contents demonstrated that Kupffer cells were actively involved in the uptake of the vesicles whereas endothelial cells did not contribute at all. Uptake by total parenchymal cells was also substantial but, on a per cell base, significantly lower than that by the Kupffer cells. By comparising the fate of the [³H]inulin label and the [¹⁴C]sphingomyelin label it was concluded that release of liposomal lipid degradation products especially occurred from Kupffer cells rather than from parenchymal cells. In both cell types, however, substantial proportions of the ¹⁴C-label accumulated in the phosphatidylcholine fraction, indicating intracellular degradation of sphingomyelin and subsequent phosphatidylcholine synthesis. Treatment of the animals with the lysosomotropic agent chloroquine prior to liposome injection effectively blocked the conversion of the choline-labeled sphingomyelin into phosphatidylcholine in both cell types. This observation indicates that uptake of the vesicles occurred by way of an endocytic mechanism.     

Different fate in vivo of oxidatively modified low density lipoprotein and acetylated low density lipoprotein in rats. Recognition by various scavenger receptors on Kupffer and endothelial liver cells

Human low density lipoprotein was oxidized (Ox-LDL) by exposure to 5 microM Cu2+ and its fate in vivo was compared to acetylated low density lipoprotein (Ac-LDL). Ox-LDL, when injected into rats, is rapidly removed from the blood circulation by the liver, similarly as Ac-LDL. A separation of rat liver cells into parenchymal, endothelial, and Kupffer cells at 10 min after injection of Ox-LDL or Ac-LDL indicated that the Kupffer cell uptake of Ox-LDL is 6.8-fold higher than for Ac-LDL, leading to Kupffer cells as the main liver site for Ox-LDL uptake. In vitro studies with isolated liver cells indicated that saturable high affinity sites for Ox-LDL were present on both endothelial and Kupffer cells, whereby the capacity of Kupffer cells to degrade Ox-LDL is 6-fold higher than for endothelial cells. Competition studies showed that unlabeled Ox-LDL competed as efficiently (90%) as unlabeled Ac-LDL with the cell association and degradation of 125I-labeled Ac-LDL by endothelial and Kupffer cells. However, unlabeled Ac-LDL competed only partially (20-30%) with the cell association and degradation of 125I-labeled Ox-LDL by Kupffer cells, while unlabeled Ox-LDL or polyinosinic acid competed for 70-80%. It is concluded that the liver contains, in addition to the scavenger (Ac-LDL) receptor which interacts efficiently with both Ac-LDL and Ox-LDL and which is concentrated on endothelial cells, an additional specific Ox-LDL receptor which is highly concentrated on Kupffer cells. In vivo the specific Ox-LDL recognition site on Kupffer cells will form the major protection system against the occurrence of the atherogenic Ox-LDL particles in the blood.     

Isolation and characterization of Kupffer and endothelial cells from the rat liver

Non-parenchymal cell suspensions were prepared from rat livers by three different methods based on a collagenase, a pronase and a combined collagenase-pronase treatment. The highest yield of Kupffer and endothelial cells was obtained with the pronase treatment. Attempts were made for a further purification of these cells by Metrizamide density gradient centrifugation after preferentially loading lysosomal structures in Kupffer cells with Triton WR 1339, Jectofer^®, Neosilvol^®, Zymosan or colloidal carbon. After loading with Triton WR 1339 or Jectofer^®, highly purified endothelial cell suspensions were obtained, but the final Kupffer cell preparations were contaminated with about 20% of endothelial cells. Kupffer and endothelial cells purified in this way showed an altered ultrastructure and contained increased activities of the lysosomal enzymes acid phosphatase, arylsulphatase B and cathepsin D. As an alternative procedure for the purification of Kupffer and endothelial cells, a method based on centrifugal elutriation was employed. With this procedure, highly purified preparations of Kupffer or endothelial cells with a well preserved ultrastructure were obtained. Compared with endothelial cells, purified Kupffer cells had a three times higher cathepsin D activity, whereas the arylsulphatase B activity was three times higher in endothelial cells. The high cathepsin D activity in Kupffer cells could be nearly completely inhibited by the specific cathepsin D inhibitor pepstatin, which excludes a possible contribution to this activity by proteases endocytosed during the isolation of the cells.     

Characterization of the simultaneous binding of Escherichia coli endotoxin to Kupffer and endothelial liver cells by flow cytometry.

BACKGROUND: The triggering of cellular responses during endotoxic shock is initiated for the binding of endotoxin (lipopolysaccharide; LPS) to the cell surface. Kupffer and endothelial liver cells, involved in the removal of endotoxin from blood circulation, show in vitro a rapid response to LPS in the absence of serum. METHODS: A double-labeling fluorescent assay was designed to evaluate the binding properties of Escherichia coli O111:B4 LPS to individual endothelial and Kupffer cells in suspension, where both populations occurred in the same relative proportion as in liver. After immunolabeling of the Kupffer cell population with the monoclonal antibody ED1 conjugated to R. phycoerythrin, the binding characteristics of LPS labeled with fluorescein to both endothelial and Kupffer cells were simultaneously studied by flow cytometry in serum-free conditions. RESULTS: Specific and saturable binding of endotoxin was observed with both populations, showing properties of a receptor-mediated process. The Kupffer cell population showed a faster capacity and a higher affinity for LPS binding. The Hill coefficients indicated positive cooperativity in the LPS interaction with both populations. CONCLUSIONS: Specific endotoxin binding to liver sinusoidal cells occurs in a serum-independent manner, particularly at high LPS concentrations. Flow cytometry is a fast, precise, and efficient technique to evaluate the simultaneous interaction of a ligand with two different cell types.     

Formaldehyde treated albumin contains monomeric and polymeric forms that are differently cleared by endothelial and Kupffer cells of the liver: evidence for scavenger receptor heterogeneity.

Formaldehyde treated albumin (F-HSA) was found to consist of a monomeric and a polymeric fraction. Both fractions were primarily endocytosed by rat liver sinusoidal cells. However, immunohistochemical staining of endocytosed material showed that the relative contribution of the endothelial and Kupffer cells in uptake of the monomer and the polymer differed significantly, with the monomer mainly having an endothelial cell- and the polymer predominantly having a Kupffer cell pattern of distribution. To directly confirm these heterogeneous patterns, we injected in vivo the 125I-labeled F-HSA fractions and isolated the endothelial and Kupffer cells by centrifugal elutriation. 73.7% of the monomeric F-HSA was found in endothelial cells and only 14.9% was found in Kupffer cells. In contrast, the polymeric F-HSA (1500 kD) was mainly endocytosed by Kupffer cells (71%), whereas the endothelial cells contributed only for 24% in hepatic uptake. In vivo studies and isolated perfused rat liver experiments showed that endocytosis of both monomer and polymer was inhibited by co-administration of polyinosinic acid, a well known inhibitor for scavenger receptors, indicating that these receptors on endothelial and Kupffer cells are mainly involved in this uptake process.     

High activity of glucose-6-phosphate dehydrogenase in Kupffer cells isolated from rat liver

Summary Sinusoidal cells in the rat liver react intensively for G6DPH activity after appropriate incubation (Rieder et al. 1978). After isolation and purification of the sinusoidal Kupffer and endothelial cells, it was demonstrated that Kupffer cells exhibit a 5–8 times higher G6PDH activity on a per cell basis by comparison with endothelial cells, while the specific G6PDH activity was 3–4 times higher in Kupffer cells. The Kupffer cells can be divided into two groups which differ significantly in G6PDH activity calculated on a per cell basis. In histochemical studies, G6PDH can be used as a marker for Kupffer cell identification.     

Hormonal control of glycogenolysis in parenchymal liver cells by Kupffer and endothelial liver cells

Conditioned media of isolated Kupffer and endothelial liver cells were added to incubations of parenchymal liver cells, in order to test whether secretory products of Kupffer and endothelial liver cells could influence parenchymal liver cell metabolism. With Kupffer cell medium an average stimulation of glucose production by parenchymal liver cells of 140% was obtained, while endothelial liver cell medium stimulated with an average of 127%. The separation of the secretory products of Kupffer and endothelial liver cells in a low and a high molecular weight fraction indicated that the active factor(s) had a low molecular weight. Media, obtained from aspirin-pretreated Kupffer and endothelial liver cells, had no effect on the glucose production by parenchymal liver cells. Because aspirin blocks prostaglandin synthesis, it was tested if prostaglandins could be responsible for the effect of media on parenchymal liver cells. It was found that prostaglandin (PG) E1, E2, and D2 all stimulated the glucose production by parenchymal liver cells, PGD2 being the most potent. Kupffer and endothelial liver cell media as well as prostaglandins E1, E2, and D2 stimulated the activity of phosphorylase, the regulatory enzyme in glycogenolysis. The data indicate that prostaglandins, present in media from Kupffer and endothelial liver cells, may stimulate glycogenolysis in parenchymal liver cells. This implies that products of Kupffer and endothelial liver cells may play a role in the regulation of glucose homeostasis by the liver.     

Interaction in vivo and in vitro of apolipoprotein E-free high-density lipoprotein with parenchymal, endothelial and Kupffer cells from rat liver.

The interaction of apolipoprotein (apo) E-free high-density lipoprotein (HDL) with parenchymal, endothelial and Kupffer cells from liver was characterized. At 10 min after injection of radiolabelled HDL into rats, 1.0 +/- 0.1% of the radioactivity was associated with the liver. Subfractionation of the liver into parenchymal, endothelial and Kupffer cells, by a low-temperature cell-isolation procedure, indicated that 77.8 +/- 2.4% of the total liver-associated radioactivity was recovered with parenchymal cells, 10.8 +/- 0.8% with endothelial cells and 11.3 +/- 1.7% with Kupffer cells. It can be concluded that inside the liver a substantial part of HDL becomes associated with endothelial and Kupffer cells in addition to parenchymal cells. With freshly isolated parenchymal, endothelial and Kupffer cells the binding properties for apo E-free HDL were determined. For parenchymal, endothelial and Kupffer cells, evidence was obtained for a saturable, specific, high-affinity binding site with Kd and Bmax. values respectively in the ranges 10-20 micrograms of HDL/ml and 25-50 ng of HDL/mg of cell protein. In all three cell types nitrosylated HDL and low-density lipoproteins did not compete for the binding of native HDL, indicating that lipids and apo B are not involved in specific apo E-free HDL binding. Very-low-density lipoproteins (VLDL), however, did compete for HDL binding. The competition of VLDL with apo E-free HDL could not be explained by label exchange or by transfer of radioactive lipids or apolipoproteins between HDL and VLDL, and it is therefore suggested that competition is exerted by the presence of apo Cs in VLDL. The results presented here provide evidence for a high-affinity recognition site for HDL on parenchymal, liver endothelial and Kupffer cells, with identical recognition properties on the three cell types. HDL is expected to deliver cholesterol from peripheral cells, including endothelial and Kupffer cells, to the liver hepatocytes, where cholesterol can be converted into bile acids and thereby irreversibly removed from the circulation. The observed identical recognition properties of the HDL high-affinity site on liver parenchymal, endothelial and Kupffer cells suggest that one receptor may mediate both cholesterol efflux and cholesterol influx, and that the regulation of this bidirectional cholesterol (ester) flux lies beyond the initial binding of HDL to the receptor.     

The interaction in vivo of transferrin and asialotransferrin with liver cells.

Rat transferrin or asialotransferrin doubly radiolabelled with 59Fe and 125I was injected into rats. A determination of extrahepatic and hepatic uptake indicated that asialotransferrin delivers a higher fraction of the injected 59Fe to the liver than does transferrin. In order to determine in vivo the intrahepatic recognition sites for transferrin and asialotransferrin, the liver was subfractionated into parenchymal, endothelial and Kupffer cells by a low-temperature cell isolation procedure. High-affinity recognition of transferrin (competed for by an excess of unlabelled transferrin) is exerted by parenchymal cells as well as endothelial and Kupffer cells with a 10-fold higher association (expressed per mg of cell protein) to the latter cell types. In all three cell types iron delivery occurs, as concluded from the increase in cellular 59Fe/125I ratio at prolonged circulation times of transferrin. It can be calculated that parenchymal cells are responsible for 50-60% of the interaction of transferrin with the liver, 20-30% is associated with endothelial cells and about 20% with Kupffer cells. For asialotransferrin a higher fraction of the injected dose becomes associated with parenchymal cells as well as with endothelial and Kupffer cells. Competition experiments in vivo with various sugars indicated that the increased interaction of asialotransferrin with parenchymal cells is specifically inhibited by N-acetylgalactosamine whereas mannan specifically inhibits the increased interaction of asialotransferrin with endothelial and Kupffer cells. Recognition of asialotransferrin by galactose receptors from parenchymal cells or mannose receptors from endothelial and Kupffer cells is coupled to active 59Fe delivery to the cells. It is concluded that, as well as parenchymal cells, liver endothelial and Kupffer cells are also quantitatively important intrahepatic sites for transferrin and asialotransferrin metabolism, an interaction exerted by multiple recognition sites on the various cell types.     

Uridine catabolism in Kupffer cells, endothelial cells, and hepatocytes

Kupffer cells, endothelial cells, and hepatocytes were separated by centrifugal elutriation. The rate of uracil formation from [2-14C]uridine, the first step in uridine catabolism, was monitored in suspensions of the three different liver cell types. Kupffer cells demonstrated the highest rate of uridine phosphorolysis. 15 min after the addition of the nucleoside the label in uracil amounted to 51%, 13%, and 19% of total radioactivity in the medium of Kupffer cells, endothelial cells, and hepatocytes, respectively. If corrected for Kupffer cell contamination, hepatocyte suspensions demonstrated similar activities as endothelial cells. In contrast to non-parenchymal cells, hepatocytes continuously cleared uracil from the incubation medium. The lack of uracil consumption by Kupffer cells and endothelial cells points to uracil as the end-product of uridine catabolism in these cells. Kupffer cells and endothelial cells did not produce radioactive CO2 upon incubation in the presence of [2-14C]uridine. Hepatocytes, however, were able to degrade uridine into CO2, beta-alanine, and ammonia as demonstrated by active formation of volatile radioactivity from the labeled nucleoside. There was almost no detectable formation of thymine from thymidine or of cytosine, uracil, or uridine from cytidine by any of the different cell types tested. These results are in line with low thymidine phosphorolysis and cytidine deamination in rat liver. Our studies suggest a co-operation of Kupffer cells, endothelial cells, and hepatocytes in the breakdown of uridine from portal vein blood with uridine phosphorolysis predominantly occurring in Kupffer cells and with uracil catabolism restricted to parenchymal liver cells.     

Modes of endocytosis of latex particles in sinusoidal endothelial and Kupffer cells of normal and perfused rat liver

The endocytosis of latex particles (0.33, 0.46 and 0.80 micron in diameter) in the sinusoidal endothelial and Kupffer cells of the rat liver was studied electron microscopically. When the liver was perfused with serum-free oxygenated Krebs Ringer bicarbonate, latex particles of all three sizes were taken up by the endothelial cells. After a 10-min perfusion, particles were incorporated by the luminal cell surface of the perikarya or of the thick portion of the endothelial cells. A large patch of bristle coat was surrounding the ingested particle. The number of ingested particles in the endothelial cells, however, was much less than in the Kupffer cells. In in vivo experiments, no endocytosis of the latex particles was observed in the endothelial cells. In the Kupffer cells, particles were engulfed by the ruffled membranes or sank into the cytoplasm without a large patch of the bristle coat both in the perfusion system and in vivo. These observations show that at least 0.80 micron latex particles are taken up by the bristle-coated membranes in the sinusoidal endothelial cells of the perfused liver. The endocytic mechanism for latex particles in the endothelial cells is different from that of the Kupffer cells.     

Gas diffusion pathway in nodules ofCasuarina cunninghamiana

The gas diffusion pathway in nodules was traced by vacuum infiltration with India ink or aniline blue and by electron microscopy. India ink infiltration was observed in the outermost and the innermost cortex in sliced nodules, but not in intact nodules. With aniline blue infiltration, it was observed that intercellular air spaces in the outermost and the innermost cortex were connected to those in nodule roots. No air spaces were in contact with walls of infected cells, although intercellular air spaces existed in some groups of uninfected cells within the infected zone. Infiltration with either India ink or aniline blue could not be observed in the infected zone in essentially all cases. Thus it is suggested that the discontinuity of the intercellular air spaces represents a major resistance to O₂ diffusion in nodules ofCasuarina cunninghamiana.     

BETA-glucuronidase and chloroacetate-esterase staining discriminates rat liver sinusoidal endothelial cells from Kupffer cells in primary culture

Beta-glucuronidase and N-AS-D-chloroacetate esterase cytochemistry have been applied to rat liver sinusoidal endothelial cells and Kupffer cells. Both staining procedures allowed a clear-cut differentiation of either cell type. Kupffer cells which had been stained with beta-glucuronidase showed a positive reaction, whereas sinusoidal endothelial cells were completely negative. If the chloroacetate reaction was used, the former stained diffusely while the latter showed a characteristic granular staining pattern. Identity and purity of sinusoidal endothelial cells and Kupffer cells was validated by transmission and scanning electron microscopy as well as by the pattern of released eicosanoids which is characteristic for either cell type. These two staining techniques are a valuable addition to the peroxidase reaction commonly applied for differentiation.     

Beta-glucuronidase and chloroacetate-esterase staining discriminates rat liver sinusoidal endothelial cells from Kupffer cells in primary culture

Beta-glucuronidase and N-AS-D-chloroacetate esterase cytochemistry have been applied to rat liver sinusoidal endothelial cells and Kupffer cells. Both staining procedures allowed a clear-cut differentiation of either cell type. Kupffer cells which had been stained with beta-glucuronidase showed a positive reaction, whereas sinusoidal endothelial cells were completely negative. If the chloroacetate reaction was used, the former stained diffusely while the latter showed a characteristic granular staining pattern. Identity and purity of sinusoidal endothelial cells and Kupffer cells was validated by transmission and scanning electron microscopy as well as by the pattern of released eicosanoids which is characteristic for either cell type. These two staining techniques are a valuable addition to the peroxidase reaction commonly applied for differentiation.     

Elevated expression of hormone-regulated rat hepatocyte functions in a new serum-free hepatocyte-stromal cell coculture model

Summary The specific performance of the adult hepatic parenchymal cell is maintained and controlled by factors deriving from the stromal bed; the chemical nature of these factors is unknown. This study aimed to develop a serum-free hierarchical hepatocyte-nonparenchymal (stromal) cell coculture system. Hepatic stromal cells proliferated on crosslinked collagen in serum-free medium with epidermal growth factor, basic fibroblast growth factor, and hepatocyte-conditioned medium; cell type composition changed during the 2-wk culture period. During the first wk, the culture consisted of proliferating sinusoidal endothelial cells with well-preserved sieve plates, proliferating hepatic stellate cells, and partially activated Kupffer cells. The number of endothelial cells declined thereafter; stellate cells and Kupffer cells became the prominent cell types after 8 d. Hepatocytes were seeded onto stromal cells precultured for 4–14 d; they adhered to stellate and Kupffer cells, but spared the islands of endothelial cells. Stellate cells spread out on top of the hepatocytes; Kupffer cell extensions established multiple contacts to hepatocytes and stellate cells. Hepatocyte viability was maintained by coculture; the positive influence of stromal cell signals on hepatocyte differentiation became evident after 48 h; a strong improvement of cell responsiveness toward hormones could be observed in cocultured hepatocytes. Hierarchial hepatocyte coculture enhanced the glucagon-dependent increases in phosphoenolpyruvate carboxykinase activity and messenger ribonucleic acid (mRNA) content three- and twofold, respectively; glucagon-activated urea production was elevated twofold. Coculturing also stimulated glycogen deposition; basal synthesis was increased by 30% and the responsiveness toward insulin and glucose was elevated by 100 and 55%, respectively. The insulin-dependent rise in the glucokinase mRNA content was increased twofold in cocultured hepatocytes. It can be concluded that long-term signals from stromal cells maintain hepatocyte differentiation. This coculture model should, therefore, provide the technical basis for the investigation of stroma-derived differentiation factors.     

Glucagon receptors in endothelial and Kupffer cells of mouse liver

To determine whether hepatic sinusoidal cells contain glucagon receptors and, if so, to study the significance of the receptors in the cells, binding of [125I]-glucagon to nonparenchymal cells (mainly endothelial cells and Kupffer cells) isolated from mouse liver was examined by quantitative autoradiography and biochemical methods. Furthermore, the pathway of intracellular transport of colloidal gold-labeled glucagon (AuG) was examined in vivo. Autoradiographic and biochemical results demonstrated many glucagon receptors in both endothelial cells and Kupffer cells, and more receptors being present in endothelial cells than in Kupffer cells. In vivo, endothelial cells internalized AuG particles into coated vesicles via coated pits and transported the particles to endosomes, lysosomes, and abluminal plasma membrane. Therefore, receptor-mediated transcytosis of AuG occurs in endothelial cells. The number of particles present on the abluminal plasma membrane was constant if the amount of injected AuG increased. Therefore, the magnitude of receptor-mediated transcytosis of AuG appears to be regulated by endothelial cells. Kupffer cells internalized the ligand into cytoplasmic tubular structures via plasma membrane invaginations and transported the ligand exclusively to endosomes and lysosomes, suggesting that the ligand is degraded by Kupffer cells.     

Quantification and regulation of apolipoprotein E expression in rat Kupffer cells

Apolipoprotein E (apoE) is synthesized by a wide variety of cells including cells of the monocyte-macrophage lineage. In order to assess the quantitative significance of apoE synthesis in a mature tissue macrophage, apoE synthesis was compared in Kupffer cells and hepatocytes isolated from rat liver. Immunoreactive apoE synthesized by both cell types exhibited identical isoform patterns when examined by high-resolution two-dimensional gel analysis. ApoE synthesis was not detected in hepatic endothelial cells. Northern blot analysis using a rat apoE cDNA probe demonstrated a single mRNA species of approximately 1200 nucleotides in freshly isolated hepatocytes and Kupffer cells. The absolute content of apoE mRNA in each cell type was determined with a DNA-excess solution hybridization assay. The apoE mRNA content (pg/microgram RNA) for Kupffer cells and hepatocytes was 35.7 and 98.8, respectively. Accounting for cellular RNA content and the population size of each cell type in the liver, Kupffer cells were calculated to contain about 0.7% of liver apoE mRNA; hepatocytes account almost quantitatively for the remainder. These results suggest that Kupffer cells are not major contributors to the plasma apoE pool. After intravenous injection of bacterial endotoxin, apoE mRNA was decreased in freshly isolated Kupffer cells whereas whole liver showed no change in apoE mRNA. Endotoxin treatment had no effect on the apoE mRNA content in several peripheral tissues. These results indicate that apoE expression in vivo is differentially regulated by endotoxin in Kupffer cells as compared to hepatocytes or apoE-producing cells in peripheral tissues.