Solubilization and hydrophobic immobilization assay of a cAMP binding protein from Dictyostelium discoideum plasma membranes

A cAMP binding site present on isolated plasma membranes of aggregation-competent $\text{D.}\text{discoideum}$ cells has been solubilized with the nonionic detergent Emulphogene BC-720. An assay has been developed based on the principle of hydrophobic chromatography, in which the detergent solubilized cAMP binding protein is immobilized on alkyl-agarose beads at low detergent concentration. This allows the necessary rapid separation of bound and free [³H]-cAMP by filtration of the beads. The kinetics and nucleotide specificity of the detergent solubilized cAMP binding protein are comparable to those of the cAMP chemotactic receptor on intact cells and plasma membranes. The alkyl-agarose bead assay may have general utility for the assay of detergent solubilized membrane receptors.     

Human C4-binding protein: N-terminal amino acid sequence analysis and limited proteolysis by trypsin

Fres isolated blood cells recombined with normal heparinized plasma and then incubated with endotoxin, induced a 100-fold increase in monocyte tissue thromboplastin synthesis. In contrast, recombination of these cells with heat inactivated plasma, cobra venom factor-treated plasma, Ca²⁺-free plasma, or BioRex 70-treated plasma (plasma free of Clq and D) before incubation with endotoxin, failed to induce monocyte synthesis of tissue thromboplastin. These results strongly support the hypothesis that complement is required for endotoxin stimulation of blood monocyte synthesis of tissue thromboplastin.     

Binding of thrombin to human platelet plasma membranes

Binding of ²⁵I-labeled thrombin to isolated human platelet plasma membranes was studied. Two classes of sites, one with high and one with low affinity for thrombin, were demonstrated. The apparent dissociation constants for the high and low affinity sites were 3.2 and 600 nM, respectively, similar to values obtained with intact platelets. Maximum binding was within 10 s, the shortest time measured, and then decreased with time to a constant level of binding within 45 s. When th equilibrium was perturbed by dilution, the system re-equilibrated with less thrombin bound than in a control that was diluted before mixing thrombin and membranes. Neither the time-dependent decrease nor the dilution effect were observed with phenylmethylsulfonyl-¹²⁵I-labelled thrombin, an irreversibly inhibited thrombin, suggesting that these phenomena may involve a thrombin-catalyzed modification of the membranes leading to decreased binding.     

Effect of Escherichia coli lipopolysaccharide on the glucagon and insulin binding to isolated rat hepatocytes

Summary Number and affinity constant of low affinity binding sites of insulin and glucagon to isolated hepatocytes decreased when the cells were incubated with Escherichia coli 0111:B4 lipopolysaccharide. This effect agrees with a non-specific binding of lipopolysaccharide to hepatocytes, similar to the well-recognized non-specific binding of albumin. Also, binding of different lectins to their glycoprotein receptors did not affect the [¹⁴C]lipopolysaccharide interaction with the cell membrane surface. Endotoxin depresses gluconeogenesis from lactate when the precursor was incubated with the cells for short time intervals. The longer the preincubation interval with lipopolysaccharide, the higher the inhibition of gluconeogenesis in the absence and in the presence of glucagon.The effect of endotoxin was also studied on the glucagon-induced synthesis of cyclic AMP and the glucagon binding. Levels of cyclic AMP and hormone binding decreased with increasing both endotoxin concentrations and preincubation intervals at which cells were in contact with endotoxin.     

A Thermodynamic Approach to Hormone-Receptor Interaction; Application to Insulin Binding to Adipocytes,Adipocyte Plasma Membranes and Liposomes Incorporating Adipocyte Insulin Receptors

Abstract

The binding of insulin to its receptor in rat adipocyte and isolated plasma membranes has been measured. The adipocyte insulin receptor has been reconstituted in lecithin liposomes and the binding of insulin investigated. A method of interpreting binding data presented as binding vs. the logarithm of free insulin concentration (binding isotherms) in terms of the binding potential concept of Wyman (1965) is described, and the results are compared with the commonly used Scatchard analysis of binding. The binding potential approach enables binding constants and Gibbs energies of formation of the insulin-receptor complex to be determined as a function of insulin bound. The limiting Gibbs energies of binding at 15°C to intact cells, membranes and liposomes were found to be -55, -52 and -49 kJ mol^?1 respectively. The affinity of the receptor for insulin decreases smoothly with increase in binding in all three systems. For intact adipocytes the number of insulin receptors per cell is found to be approximately 43,000.     

Modulation of glucose uptake in animal cells. Studies using plasma membrane vesicles isolated from nontransformed and simian virus 40-transformed mouse fibroblast cultures.

Plasma membrane vesicles isolated from nontransformed and Simian virus 40-transformed mouse fibroblast cultures catalyzed carrier-mediated D-glucose transport without detectable metabolic conversion to glucose 6-phosphate. Glucose transport activity was stereospecific, temperature-dependent, sensitive to inactivation by p-chloromercuriphenylsulfonate, and accompanied plasma membrane material during subcellular fractionation. D-Glucose efflux from vesicles was inhibited by phloretin, an inhibitor of glucose uptake in intact cells. Cytochalasin B, a potent inhibitor of glucose uptake when tested with the intact cells used for vesicle isolation did not inhibit glucose transport in vesicles despite the presence of high affinity cytochalasin binding sites in isolated membranes. The enhanced glucose uptake observed in intact cells after viral transformation was not expressed in vesicles: no significant differences in glucose transport specific activity could be detected in vesicle preparations from nontransformed and transformed mouse fibroblast cultures. These findings indicate that cellular components distinct from glucose carriers can mediate changes in glucose uptake in mouse fibroblast cultures in at least two cases: sensitivity to inhibition by cytochalasin B and the enhanced cellular sugar uptake observed after viral transformation.     

Photoaffinity labelling of the nucleoside transporter of cultured mouse lymphoma cells

Nitrobenzylthioniosine (NBMPR), a potent and specific inhibitor of nucleoside transport, is bound reversibly by high affinity sites on nucleoside transporter proteins of erythrocyte membranes and, upon photoactivation, NBMPR molecules become covalently bonded to the sites. This study showed that [³H]NBMPR molecules reversibly bound to intact S49 and L5178Y mouse lymphoma cells became covalently bound upon exposure to UV light. Electrophoretic analysis of plasma membrane fractions from the labelled cells showed that ³H was present in polypeptides which migrated as a major band with an apparent M_r of 45000–65000.     

Sialoadhesin expressed on IFN-induced monocytes binds HIV-1 and enhances infectivity

Background

HIV-1 infection dysregulates the immune system and alters gene expression in circulating monocytes. Differential gene expression analysis of CD14⁺ monocytes from subjects infected with HIV-1 revealed increased expression of sialoadhesin (Sn, CD169, Siglec 1), a cell adhesion molecule first described in a subset of macrophages activated in chronic inflammatory diseases.

Methodology/Principal Findings

We analyzed sialoadhesin expression on CD14⁺ monocytes by flow cytometry and found significantly higher expression in subjects with elevated viral loads compared to subjects with undetectable viral loads. In cultured CD14⁺ monocytes isolated from healthy individuals, sialoadhesin expression was induced by interferon-α and interferon-γ but not tumor necrosis factor-α. Using a stringent binding assay, sialoadhesin-expressing monocytes adsorbed HIV-1 through interaction with the sialic acid residues on the viral envelope glycoprotein gp120. Furthermore, monocytes expressing sialoadhesin facilitated HIV-1 trans infection of permissive cells, which occurred in the absence of monocyte self-infection.

Conclusions/Significance

Increased sialoadhesin expression on CD14⁺ monocytes occurred in response to HIV-1 infection with maximum expression associated with high viral load. We show that interferons induce sialoadhesin in primary CD14⁺ monocytes, which is consistent with an antiviral response during viremia. Our findings suggest that circulating sialoadhesin-expressing monocytes are capable of binding HIV-1 and effectively delivering virus to target cells thereby enhancing the distribution of HIV-1. Sialoadhesin could disseminate HIV-1 to viral reservoirs during monocyte immunosurveillance or migration to sites of inflammation and then facilitate HIV-1 infection of permissive cells.     

Aggregation in Dictyostelium discoideum

Cultured peritoneal macrophages have previously been shown to release a potent mitogen for mesenchymal cells. Peritoneal macrophages are derived from peripheral blood monocytes, one of the principal inflammatory cells associated with numerous tissue responses to injury. Cultured human monocytes can be activated by endotoxin or concanavalin A to secrete a potent growth factor(s) that is active on human smooth muscle cells, human fibroblasts and 3T3 cells. The optimal conditions for activation of monocyte release of this monocyte-derived growth factor(s) (MDGF) were to expose 5-day-old monocyte cultures (initially plated at 6.8 × 10⁵ cells/ml medium) to 10 μg/ml endotoxin or 6 μg/ml concanavalin A for approximately 20 hr. Monocytes can secrete MDGF into serum-free medium supplemented with 0.15% bovine serum albumin. MDGF stimulates both DNA synthesis and increase in cell number and is trypsin-sensitive, heat labile and nondialyzable. The relationship of MDGF to other monocyte products and its potential importance in wound repair and atherogenesis are discussed.     

Arsenic Exposure Increases Monocyte Adhesion to the Vascular Endothelium,a Pro-Atherogenic Mechanism

Epidemiological studies have shown that arsenic exposure increases atherosclerosis, but the mechanisms underlying this relationship are unknown. Monocytes, macrophages and platelets play an important role in the initiation of atherosclerosis. Circulating monocytes and macrophages bind to the activated vascular endothelium and migrate into the sub-endothelium, where they become lipid-laden foam cells. This process can be facilitated by platelets, which favour monocyte recruitment to the lesion. Thus, we assessed the effects of low-to-moderate arsenic exposure on monocyte adhesion to endothelial cells, platelet activation and platelet-monocyte interactions. We observed that arsenic induces human monocyte adhesion to endothelial cells in vitro. These findings were confirmed ex vivo using a murine organ culture system at concentrations as low as 10 ppb. We found that both cell types need to be exposed to arsenic to maximize monocyte adhesion to the endothelium. This adhesion process is specific to monocyte/endothelium interactions. Hence, no effect of arsenic on platelet activation or platelet/leukocyte interaction was observed. We found that arsenic increases adhesion of mononuclear cells via increased CD29 binding to VCAM-1, an adhesion molecule found on activated endothelial cells. Similar results were observed in vivo, where arsenic-exposed mice exhibit increased VCAM-1 expression on endothelial cells and increased CD29 on circulating monocytes. Interestingly, expression of adhesion molecules and increased binding can be inhibited by antioxidants in vitro and in vivo. Together, these data suggest that arsenic might enhance atherosclerosis by increasing monocyte adhesion to endothelial cells, a process that is inhibited by antioxidants.     

Characteristics of the binding of high-density lipoprotein3 by intact cells and membrane preparations of rat intestinal mucosa

We have investigated the binding of high-density lipoprotein (HDL3, d = 1.12-1.21 g/ml), and apolipoprotein E-deficient human and rat HDL, obtained by heparin-Sepharose affinity chromatography, to intact cells and membrane preparations of rat intestinal mucosal cells. Binding of 125I-labeled HDL3 to the basolateral plasma membranes was characterised by a saturable, specific process (Kd = 21 micrograms of HDL3 protein/ml, Bmax = 660 ng HDL3 protein/mg membrane protein) and E-deficient human HDL demonstrated a similar affinity for the binding site. The basolateral plasma membranes isolated from proximal and distal portion of rat small intestine showed similar binding affinities for HDL3, whereas the interaction of HDL with brush-border membranes was characterised by mainly nonspecific and nonsaturable binding. The binding of 125I-labeled HDL3 to basolateral plasma membranes was competitively inhibited by unlabeled HDL3 but less efficiently by unlabeled human LDL. The putative HDL receptor was not dependent on the presence of divalent cations but was markedly influenced by temperature and sensitive to pronase treatment. We have also demonstrated, using whole intestinal mucosal cells, that lysine and arginine-modified HDL3 inhibited binding of normal 125I-labeled HDL3 to the same extent as normal excess HDL3. These data suggest that basolateral plasma membranes of rat intestinal mucosal cells possess a specific receptor for HDL3 which contains mainly apolipoprotein A-I and A-II, and the mechanisms of recognition of HDL3 differ from those involved in binding to the B/E receptor.     

Specific binding sites for corticosterone in isolated cells and plasma membrane from rat liver

Summary The specific binding of [³H]corticosterone to hepatocytes is a nonsaturable, reversible and temperature-dependent process. The binding to liver purified plasma membrane fraction is also specific, reversible and temperature dependent but it is saturable. Two types of independent and equivalent binding sites have been determined from hepatocytes. One of them has high affinity and low binding capacity (K _D=8.8nm andB _max=1477 fmol/mg protein) and the other one has low affinity and high binding capacity (K _D=91nm andB _max=9015 fmol/mg). In plasma membrane only one type of binding site has been characterized (K _D=11.2nm andB _max=1982 fmol/mg). As it can be deduced from displacement data obtained in hepatocytes and plasma membrane the high affinity binding sites are different from the glucocorticoid, progesterone nuclear receptors and the Na⁺,K⁺-ATPase digitalis receptor. Probably it is of the same nature that the one determinate for [³H]cortisol and [³H]corticosterone in mouse liver plasma membrane. Beta-and alpha-adrenergic antagonists as propranolol and phentolamine did not affect [³H]corticosterone binding to hepatocytes and plasma membranes; therefore, these binding sites are independent of adrenergic receptors. The binding sites in hepatocytes and plasma membranes are not exclusive for corticosterone but other steroids are also bound with very different affinities.     

Interactions of Vesicular Stomatitis Virus Structural Proteins with HeLa Plasma Membranes

THE processes whereby nucleoprotein core particles of certain animal viruses become enveloped by and bud off from host cell membranes can be studied by preparing membrane^1,2 or “sedimentable”³ fractions from infected cells and examining them for the presence of virus proteins. We find that similar experiments designed to monitor assembly of vesicular stoma-titus virus (VSV) at sites along the plasma membranes of HeLa cells are best interpreted after first investigating the possibility that virus proteins adsorb to plasma membranes during cell fractionation and membrane isolation. In this report, we show that at 0° C the membrane protein of VSV, among other virus proteins, adsorbs to plasma membranes isolated from uninfected HeLa cells. With appropriate pulse-chase experiments, however, we are able to demonstrate the progressive association, in vivo, of VSV core protein with plasma membranes of infected HeLa cells.     

Calcium Binding to Plasma Membranes from Normal and SV40 Transformed Hamster Lymphocytes

Abstract

The calcium binding properties of isolated plasma membranes from normal and SV40 transformed hamster lymphocytes were compared over the Ca²⁺ concentration range of 10^?5M to 5 × 10^?3M and at physiological ionic strength. At all Ca²⁺ concentrations, normal membranes bound more Ca²⁺ than tumor membranes; at blood Ca²⁺ levels (1–2 mM) plasma membranes of normal cells bind twice as much as membranes from tumor cells. Normal plasma membranes demonstrated positive cooperative Ca²⁺ binding whereas tumor membranes displayed non-interacting Ca²⁺-binding sites. Ca²⁺ binding to both membranes was insensitive to Mg²⁺ (0.1 to 2.5 mM). A pH shift from 7 to 6 resulted in a 70% decrease of normal membrane-bound Ca²⁺ compared to a 40% decrease observed with tumor membranes. Extracellular surface Ca²⁺ binding to intact cells was also studied after a 72-hour equilibration of cells with ⁴⁵Ca²⁺ and with ethylene-glycol-bis-(β-amino-ethyl ether) N, N′-tetraacetate chelation as marker for surface Ca²⁺. Tumor cell surface Ca²⁺ binding was only 10% of that observed with quiescent lymphocytes. Normal lymphocytes stimulated to divide with phytohemagglutinin also showed a decreased level of surface Ca²⁺ (50%). However, plasma membranes isolated from non-dividing and phytohemagglutinin-stimulated lymphocytes exhibited equivalent Ca²⁺ binding.     

Studies of albumin binding to rat liver plasma membranes: Implications for the albumin receptor hypothesis

To characterize a previously proposed hepatocyte albumin receptor, we examined the binding of native and defatted ¹²⁵I-labeled rat albumin to rat liver plasma membranes. After incubation for 30 min, binding was determined from the distribution of radioactivity between membrane pellet and supernatant following initial centrifugation (15 000 × g for 15 min), after repeated cycles of washing with buffer and re-centrifugation. ¹²⁵I-labeled albumin recovered in the initial membrane pellet averaged only 4% of that incubated. Moreover, this albumin was only loosely associated with the membrane, as indicated by recovery in the pellet of under 0.5% of the counts after three washes. Binding of ¹²⁵I-labeled albumin to the plasma membranes was no greater than to erythrocyte ghosts, was not inhibited by excess unlabeled albumin, and was not decreased by heat denaturation of the membranes, all suggestive of a lack of specific binding. Failure to observe albumin binding to the membranes was not due to a rapid dissociation rate or ‘off-time’, as incubations in the presence of sufficient ultraviolet light to promote covalent binding of ligands to receptors did not increase ¹²⁵I counts bound to the membrane. Finally, affinity chromatography over albumin/agarose gel of solubilized membrane proteins provided no evidence of a membrane protein with a high affinity for albumin. These studies, therefore, do not support the hypothesis that liver cell plasma membranes contain a specific albumin receptor.     

ASSEMBLY OF LIPIDS INTO MEMBRANES IN ACANTHAMOEBA PALESTINENSIS : II. The Origin and Fate of Glycerol3H-Labeled Phospholipids of Cellular Membranes

The membranes of Acanthamoeba palestinensis were studied by examination in fixed cells, and then by following the movements of glycerol-³H-labeled phospholipids by cell fractionation. Two previously undescribed structures were observed: collapsed cytoplasmic vesicles of cup shape, and plaques in food vacuole and plasma membrane similar in size to the collapsed vesicles. It appeared that the plaques formed by insertion of collapsed vesicles into membranes and/or that collapsed vesicles formed by pinching off of plaques. Fractions were isolated, enriched with nuclei, rough endoplasmic reticulum (RER), plasma membrane, Golgi-like membranes, and collapsed vesicles. The changes in specific activity of glycerol-³H-labeled phospholipids in these membranes during incorporation, turnover, and after pulse-labeling indicated an ordered sequence of appearances of newly synthesized phospholipids, first in nuclei and RER, then successively in Golgi membranes, collapsed vesicles, and finally, plasma membrane. In previous work we had found no large nonmembranous phospholipid pool in A. palestinensis. These observations are consistent with the hypothesis that membrane phospholipids are synthesized, perhaps as integral parts of membranes, in RER and nuclei. Subsequently, some of the newly synthesized phospholipids are transported to the Golgi complex to become integrated into the membranes of collapsed vesicles, which are precursors of the plasma membrane. Collapsed vesicles from the plasma membrane by inserting into it as plaques. When portions of the plasmalemma from food vacuoles, collapsed vesicles pinch off from their membranes and are recycled back to the cell surface.     

Separation of dense, polysaccharide-containing vesicles from secreting, cultured oat cells. Characterization of a putative secretory vesicle fracton

Suspension cultured oat (Avena sativa L. cv. Garry) cells, which secrete polysaccharides into the medium, were used as starting material for analyses of Golgi-derived vesicle membranes and plasma membranes isolated during cell fractionation. Vesicles collected by a procedure employing ultrafiltration followed by ultracentrifugation into a sucrose step gradient exhibited an equilibrium density of 1.27 g cm^?3 when run on continuous sucrose gradients, a feature which is most likely attributable to the high concentration of enclosed polysaccharides. Brief sonication lowered the density of these vesicles to about 1.15 g cm^?3, as judged from the coincidence of the protein peak and the marker enzymes for Golgi [Triton-stimulated UDPase, cold-storage IDPase (EC 3.6.1.6)] and plasma membrane [vanadate-inhibited K⁺, Mg²⁺-ATPase (EC 3.6.1.3)]. Sonication of these vesicles also greatly diminished the amount of detectable polysaccharide observed in a colorimetric assay for sugars. Fractionation of a plasma membrane-enriched preparation from these cells on continuous sucrose gradients showed the major protein peak and the peak activity for the plasma membrane marker at 1.17 g cm^?3, however, there was also significant overlap with a mitochondrial [cytochrome c oxidase (EC 1.9.3.1)] peak at 1.18 g cm^?3, Smaller peaks of the Golgi markers were seen at 1.14 g cm^?3. Analyses of marker enzymes for ER and mitochondria (EC 1.6.99.3) showed little contamination of the membranes of presumptive secretory vesicles from these sources, and the lack of significant vanadate-insensitive ATPase activity in the density range from 1.13–1.18 g cm^?3 in either fractionation scheme suggests that these membranes do not include material from the tonoplast. The coincidence of markers for Golgi and plasma membrane with from the tonoplast. The coincidence of markers for Golgi and plasma membrane with the membranes of sonicated, dense vesicles, at a density slightly lower than that of plasma membranes prepared from the same cells, supports the possibility that membranes en route to the plasma membrane are incompletely differentiated.     

Uptake of high-density lipoprotein by Y-organs of the crab,cancer antennarius. II. formal characterization of receptor-mediation with isolated membranes

Y-organs are the ecdysial glands of crustaceans, responsible for synthesis and secretion of ecdysteroid hormones. For this purpose, the glands acquire cholesterol as obligate precursor entirely from circulating high-density lipoprotein (HDL). A preceding study provided evidence for the mechanism of acquisition: Y-organs take up cholesterol bound to HDL by an energy-requiring process, receptor-mediated absorptive endocytosis. The present study characterized the receptors involved utilizing isolated Y-organ membranes. HDL binding was saturable and specific; a dissociation constant (K_d) of 1.08 × 10^?7 M and a binding maximum at equilibrium (B_max) of 70 μg HDL protein/mg membrane protein, were obtained. Binding was decreased by protease and was dependent upon calcium. Y-organs are regulated negatively by a peptide hormone from the eystalks, molt-inhibiting hormone (MIH). Y-organ membranes from de-eyestalked crabs (MIH absent) exhibited the same K_d value as membranes from intact crabs, but a B_max 17% higher. Thus, MIH activity apparently does not change the binding affinity of HDL, but decreases the number of binding sites. These results agree with our previous findings that MIH depresses ecdysteroid synthesis in part by inhibiting cholesterol uptake. Generally, Y-organ cells appear to contain receptors for HDL that are of high affinity and high binding capacity, similar to the characteristics reported for the binding of insect HDL (vitellogenin) to fat bodies and oocytes. © 1995 Wiley-Liss, Inc.     

The role of the plasma membrane in the development of Dictyostelium discoideum. II. Developmental and topographic analysis of polypeptide and glycoprotein composition

Previous workers have shown in a variety of ways that cell contact is required for the differentiation of Dictyostelium discoideum. Because interactions between cells are probably mediated by molecules on their plasma membranes, we have characterized the polypeptide composition of the membrane of cells at different stages of development. At least 55 polypeptides are found in the plasma membrane of vegetative cells. The polypeptide composition of the plasma membranes changes considerably during development. Treatment of intact cells with pronase indicated that many of the altered components appear to be located on the external surface of the plasma membrane where they could participate in interactions between cells. Similar digestion of the isolated membranes destroys most of their polypeptides, indicating that the bulk of the proteins of the plasma membrane are not completely embedded in the membrane. Several polypeptides appear to change in sensitivity to pronase during development. There are several changes in glycoprotein composition which occur between log phase and aggregation phase. An almost complete change in glycoprotein species occurs between aggregation and pre-culmination. Unlike the polypeptides, the glycoproteins are very resistant to pronase treatment in intact cells. However, some are pronase sensitive in isolated membranes.     

Identification of the insulin receptor in undifferentiated and differentiated NB-15 mouse neuroblastoma cells by affinity labeling

Plasma membranes prepared from clonal NB-15 mouse neuroblastoma cells were sequentially incubated with ¹²⁵I-labeled insulin (10 nM) and the bifunctional cross-linking agent disuccinimidyl suberate. This treatment resulted in the cross-linking of ¹²⁵I-labeled insulin to a polypeptide that gave an apparent M_r of 135 000 on a sodium dodecyl sulfate-polyacrylamide gel electrophoresed in the presence of 10% β-mercaptoethanol. Affinity labeling of this polypeptide was inhibited by the presence of 5 μM unlabeled insulin, but not by 1 μM unlabeled nerve growth factor. Using the same affinity labeling technique, ¹²⁵I-labeled nerve growth factor (1 nM) did not label any polypeptide appreciably in the plasma membranes of NB-15 cells but labeled an M_r 145 000 and an M_r 115 000 species in PC-12 rat pheochromocytoma cells. The number of insulin binding sites per cell in the intact differentiated NB-15 mouse neuroblastoma cells was approx. 6-fold greater than that in the undifferentiated NB-15 mouse neuroblastoma cells as measured by specific binding assay, suggesting an increase of the number of insulin receptors in NB-15 mouse neuroblastoma cells during differentiation.