Real-time kinetic studies on the interaction of transforming growth factor alpha with the epidermal growth factor receptor extracellular domain reveal a conformational change model.

Transforming growth factor alpha (TGF-alpha), epidermal growth factor (EGF), and related factors mediate their biological effects by binding to the extracellular domain of the EGF receptor, which leads to activation of the receptor's cytoplasmic tyrosine kinase activity. Much remains to be determined, however, about the detailed molecular mechanism involved in this ligand-induced receptor activation. The determination of the binding mechanism and the related thermodynamic and kinetic parameters are of prime importance. To do so, we have used a surface plasmon resonance-based biosensor (the BIAcore) that allows the real-time recording of the interaction between TGF-alpha and the extracellular domain of the EGF receptor. By immobilizing different biotinylated derivatives of TGF-alpha on the sensor chip surface, we demonstrated that the N-terminus of TGF-alpha is not directly involved in receptor binding. By optimizing experimental conditions and interpreting the biosensor results by several data analysis methods, we were able to show that the data do not fit a simple binding model. Through global analysis of the data using a numerical integration method, we tested several binding mechanisms for the TGF-alpha/EGF receptor interaction and found that a conformational change model best fits the biosensor data. Our results, combined with other analyses, strongly support a receptor activation mechanism in which ligand binding results in a conformation-driven exposure of a dimerization site on the receptor.

NMR study of the transforming growth factor-alpha (TGF-alpha)-epidermal growth factor receptor complex. Visualization of human TGF-alpha binding determinants through nuclear Overhauser enhancement analysis.

The study of human transforming growth factor-alpha (TGF-alpha) in complex with the epidermal growth factor (EGF) receptor extracellular domain has been undertaken in order to generate information on the interactions of these molecules. Analysis of 1H NMR transferred nuclear Overhauser enhancement data for titration of the ligand with the receptor has yielded specific data on the residues of the growth factor involved in contact with the larger protein. Significant increases and decreases in nuclear Overhauser enhancement cross-peak intensity occur upon complexation, and interpretation of these changes indicates that residues of the A- and C-loops of TGF-alpha form the major binding interface, while the B-loop provides a structural scaffold for this site. These results corroborate the conclusions from NMR relaxation studies (Hoyt, D. W., Harkins, R. N., Debanne, M. T., O'Connor-McCourt, M., and Sykes, B. D. (1994) Biochemistry 33, 15283-15292), which suggest that the C-terminal residues of the polypeptide are immobilized upon receptor binding, while the N terminus of the molecule retains considerable flexibility, and are consistent with structure-function studies of the TGF-alpha/EGF system indicating a multidomain binding model. These results give a visualization, for the first time, of native TGF-alpha in complex with the EGF receptor and generate a picture of the ligand-binding site based upon the intact molecule. This will undoubtedly be of utility in the structure-based design of TGF-alpha/EGF agonists and/or antagonists.

Improved yields of the extracellular domain of the epidermal growth factor receptor produced using the baculovirus expression system by medium replacement following infection.

The extracellular domain of the epidermal growth factor receptor (EGFR) was expressed using the baculovirus expression vector system. The maximum level of the EGFR extracellular domain secreted into the medium in Sf-9 (Spodoptera frugiperda or fall army-worm) cell batch culture was approximately 2.5 micrograms ml-1. In order to increase this yield, a process was developed that included the following sequence of steps: batch growth to maximum cell density, infection of the cells with recombinant virus, and replacement of spent medium. By using this process, the specific yield of recombinant protein, which in batch culture drops when infection is carried out at densities greater than 3 x 10(6) cells ml-1, can be maintained at a maximum in cultures infected at densities of 10(7) cells ml-1 or greater. The process, when applied to 3-1 and 11-1 bioreactor cultures, allowed a maximum volumetric yield of triple the maximum value attainable in batch culture. Spent-medium analysis indicates that medium replacement provides certain nutrients that could otherwise be limiting for recombinant protein production.

Interaction of transforming growth factor alpha with the epidermal growth factor receptor: binding kinetics and differential mobility within the bound TGF-alpha.

The interaction of transforming growth factor alpha (TGF-alpha) with the complete extracellular domain of the epidermal growth factor receptor (EGFR-ED) was examined by nuclear magnetic resonance (NMR) spectroscopy. The 1H NMR resonances of the methyl groups of TGF-alpha were used as probes of the interaction of TGF-alpha with the EGF receptor to determine the binding kinetics and the differential mobility within the bound TGF-alpha. The methyl resonances were studied because there are 14 methyl containing residues well dispersed throughout the structure of TGF-alpha and the relaxation properties of methyl groups are well understood. Changes in the longitudinal and transverse 1H NMR relaxation rates of the methyl resonances of TGF-alpha caused by binding to the 85-kDa EGFR-ED were studied. From these measurements it was determined that the interaction was in the NMR fast exchange limit. A binding mechanism to rationalize the different rates determined by NMR and surface plasmon resonance techniques [Zhou, M., et al. (1993) Biochemistry 32, 8193-8198] is proposed. The transverse relaxation rate (R2) enhancements of the various methyl resonances displayed a regional dependence within the bound TGF-alpha molecule. Resonances from the C-terminus of TGF-alpha, which were flexible in the unbound molecule, revealed dramatic increases in their R2 upon binding to the EGFR-ED along with resonances from the interior of TGF-alpha. However, upon binding, the R2 enhancements of the methyl resonances from the N-terminus of TGF-alpha, which were also flexible in the unbound TGF-alpha, were slight; indicating a retention of mobility of this region for bound TGF-alpha. The implications of these data with respect to the mechanism of receptor activation and the design of antagonists are discussed.

The extracellular domain of the epidermal growth factor receptor. Studies on the affinity and stoichiometry of binding, receptor dimerization and a binding-domain mutant.

The binding of epidermal growth factor (EGF) or an EGF-like growth factor to the EGF receptor is the initial event which leads to receptor activation, and consequently the induction of cell growth. In order to study this binding interaction in detail, we produced the extracellular domain of the EGF receptor (EGFR) using the baculovirus expression system. Affinity-labeling and Western-blot analyses revealed that the baculovirus-infected insect cells secrete active EGFR extracellular domain relatively efficiently, however a significant amount of inactive EGFR extracellular domain is retained within the cells. The apparent dissociation constant (Kd) of the secreted EGFR extracellular domain for EGF and transforming growth factor alpha (TGF-alpha), as determined using an immobilized receptor binding assay, was approximately 200 nM. Interestingly, this Kd value is 30-40-fold lower than that of the full-length EGFR derived from detergent-solubilized A431 cell membranes. The stoichiometry of binding of the EGFR extracellular domain to EGF and TGF-alpha was examined by band-shift analysis on non-denaturing PAGE and was estimated to be 1:1. We have also shown, using sedimentation equilibrium analysis, that ligand binding induces significant dimerization of the EGFR extracellular domain. Finally, we carried out site-specific mutagenesis on the EGFR extracellular domain in order to define the ligand-binding region. We identified amino acid residues which are close to the binding site since they are common to the epitopes of several ligand-competitive monoclonal antibodies. However, these residues do not contribute directly to ligand binding since the affinity of the mutated EGFR extracellular domain for EGF and TGF-alpha was unaffected.

Action of insulin in rat adipocytes and membrane properties.

Several small peptides inhibit insulin-promoted glucose uptake in rat adipocytes. At 10 microM peptide concentration, the extent of their inhibition of the insulin effect is related to the ability of these peptides to raise the bilayer- to hexagonal-phase transition temperature in model membranes. Hexane and DL-threo-dihydrosphingosine lower this phase transition temperature in model membranes, and they promote glucose uptake in adipocytes. There is thus an empirical relationship between the action of membrane additives on glucose uptake in adipocytes and their effect on the hexagonal-phase-forming tendency in model membranes. The most potent of the bilayer-stabilizing peptides tested in this work is carbobenzoxy-D-Phe-L-Phe-Gly. This peptide also inhibits insulin-stimulated protein synthesis in adipocytes. In contrast, DL-threo-dihydrosphingosine stimulates protein synthesis. The uptake of [125I]iodoinsulin by adipocytes is inhibited by carbobenzoxy-D-Phe-L-Phe-Gly. The mechanism of action of the bilayer-stabilizing peptides includes inhibition of insulin-dependent protein phosphorylation in adipocytes. The peptides are not specific inhibitors of a single function but are suggested to cause their effects by altering the physical properties of the membrane in a nonspecific manner. These results demonstrate that insulin-dependent functions of rat adipocytes can be modified by membrane additives in a manner predictable from the properties of these additives in model membranes.

Dimerization of the P-glycoprotein in membranes.

Plasma membranes from a CHO cell line, CHRC5, which exhibits multidrug resistance was studied using radiation inactivation analysis. The P-glycoprotein content of the membrane was determined by Western blots. Irradiation resulted in the loss of P-glycoprotein. The dependence of this loss on radiation dose corresponded to a target size of 250 kDa which is the molecular mass of a dimer of the P-glycoprotein. This is strong evidence to indicate that the P-glycoprotein self associates in the membrane.

Preferential hepatic uptake of iron from rat asialotransferrin: possible engagement of two receptors.

Hepatic iron uptake from and degradation of rat asialotransferrin prepared from the least anionic (major) component of rat transferrin were studied in intact rats. In experiments lasting 60-90 min, rat asialotransferrin delivered a three to four times larger fraction of the Fe dose to the liver than rat transferrin. Variations in the concentration of endogenous circulating rat 2Fe-transferrin by up to 300% failed to affect the enhanced hepatic delivery of Fe from rat asialotransferrin. However, pretreating the animals with a large dose of asialomucin, or fully sialylated human transferrin, or a combination of both did affect the delivery. In all cases, rat asialotransferrin delivered Fe to the liver at rates comparable with those seen with rat transferrin. The reason for the efficacy of human transferrin was clarified in competitive binding studies on rat hepatocytes and reticulocytes, which showed that human transferrin possessed an approximately sevenfold higher affinity for rat transferrin receptors than the homologous protein. These findings suggest that the enhanced hepatic uptake of Fe from rat asialotransferrin is mediated by simultaneous binding of the ligand both through its glycan and transferrin receptor affinity site. Pretreatment with asialomucin and human transferrin had no suppressing effect on basal hepatic delivery of iron from rat 2Fe-transferrin. The data suggest that deposition of a significant fraction of Fe in rat liver from rat transferrin is likely to take place by a mechanism not involving transferrin receptors. Desialylation shortened the metabolic half-life of rat transferrin from 33 to 24 h.(ABSTRACT TRUNCATED AT 250 WORDS)

Interaction of human lactoferrin with the rat liver.

Binding of human lactoferrin (hLf) by purified rat liver plasma membranes was studied to clarify whether the liver possesses specific hLf receptors. The binding was rapid between 4 degrees and 37 degrees C, with a pH optimum close to 5.0. At 22 degrees C and in glycine-NaOH (5 mM, pH 7.4) containing 150 mM NaCl and 0.5% albumin, 1 microgram of membrane bound a maximum of 11.8 ng hLf. The dissociation constant of the interaction was 1.6 X 10(-7) M. Other proteins of high isoelectric points (lactoperoxidase, lysozyme, and particularly salmine sulfate) and a piperazine derivative inhibited hLf binding in a concentration-dependent manner. In contrast, monosaccharides (galactose, N-acetylgalactosamine, mannose, and fucose) were ineffective. By omitting NaCl from the incubation buffer, binding was increased 3.6-fold. Erythrocyte ghosts bound hLf less firmly and alveolar macrophages more firmly than hepatic plasma membranes. Liver cell fractionations performed after the intravenous injection of labeled hLf showed that approximately 88% of the hepatic radioligand was associated with parenchymal cells. When binding was expressed per unit of cell volume, however, more hLf was present in nonparenchymal than in parenchymal cells, implying that the above value was determined by the relative cell masses rather than affinities alone. It is concluded that the binding of hLf by hepatic plasma membranes is electrostatic, i.e., is mediated by the cationic nature of the ligand, and that it is explicable in terms of a "specific nonreceptor interaction" of the generalized type proposed by Cuatrecasas and Hollenberg (Adv. Protein Chem. 30: 251-451, 1976).

Lactoferrin catabolism in the rat liver.

The hepatic uptake and degradation of human diferric 125I-lactoferrin by the liver of the intact rat were studied. After intravenous injection, the tracer was rapidly cleared by the liver, probably by adsorptive pinocytosis, as inferred from observations with a 3,470-fold dose range. Endocytosed lactoferrin was transferred, with a delay, from a light-density subcellular particle to an organelle that had a density similar to lysosomes. The loss of protein bound 125I from the liver was very slow (half-life 2.7 h), and its rate matched closely that of human asialotransferrin type 3. Lactoferrin was found to be a poor substrate for lysosomal hydrolases in vitro. Fucoidin effected the release of a portion of lactoferrin from the liver back into the plasma. By using this agent, indirect evidence was obtained suggesting that a fraction of lactoferrin is being repeatedly endo- and exocytosed (diacytosed) by the liver over prolonged periods of time. Fucosylation failed to impart lactoferrinlike properties on human asialotransferrin type 1, although the derivatized protein exhibited a less than or equal to 10-fold increase in affinity for the liver relative to the parent molecule.

Free-flow electrophoresis of the low-density structures that contain asialoglycoproteins in the rat liver.

Rats were given an intravenous dose (1-2 micrograms/100 g) of iodine-labelled asialotransferrin, asialofetuin, or asialoorosomucoid either alone or in combinations, and the distribution of the radioactivity in the liver, removed 10-20 min after the injection, was analyzed by free-flow electrophoresis in an Elphor VaP 11 apparatus. Liver homogenates were prepared for electrophoresis according to an elaborate ultracentrifugation scheme that is outlined in detail with respect to conditions and yields. The scheme involved differential centrifugation, followed by density gradient centrifugation in a linear sucrose gradient and gel filtration using Sepharose 2B. Two ligand-containing fractions were obtained during differential centrifugation, each associated with a different complement of subcellular marker enzymes. On free-flow electrophoresis, the ligand present in either fraction exhibited a major and a minor peak. They were incompletely separated, the minor peak shouldering on the major one. The major peak had a higher electrophoretic mobility than the peaks of the acid phosphatase and phosphodiesterase I activities, but it had the same mobility as the sialyltransferase activity. The minor, less electronegative peak comigrated with the peaks of acid phosphatase and phosphodiesterase I activities and also with the major protein component of the subcellular fraction. It is concluded that the asialoglycoprotein-transporting subcellular vesicles are heterogeneous in regard to charge and that their complete separation from subcellular marker enzymes cannot be accomplished by free-flow electrophoresis.

Partial resialylation of human asialotransferrin types 1 and 2 in the rat.

125I-labeled asialotransferrin types 1 and 2 were administered in small doses to rats. The protein still in the plasma after 1-12 h was partially repurified and electrophoresed at pH 8.1, together with a transferrin standard that is composed of all six forms of the protein with respect to sialic acid content. The electrophoretic mobility of both asialotransferrins increased with time, type 2 being affected sooner than type 1. The changed mobility was due to increased electronegativity that was fully reversible by treatment of the samples with neuraminidase, thus identifying the underlying cause as partial resialylation. Asialotransferrin incubated in vitro with serum, plasma, or whole blood for 16 h exhibited no change in electrophoretic mobility. In conjunction with an earlier study on asialotransferrin type 3, it was found that the apparent speeds of resialylation of the three asialotransferrins were in the same order as their affinities for the asialoglycoprotein-binding hepatic lectin. This suggests the involvement of an endo- rather than of an ecto-transferase. Transfer of 59Fe from asialotransferrins to the liver was used to monitor the frequency of hepatocyte-asialotransferrin interactions. Iron deposition in the liver took place much more rapidly than the appearance of detectable quantities of partially resialylated asialotransferrin molecules in the circulation. It is concluded that each asialotransferrin molecule probably undergoes several passages through the hepatocyte before its glycans become modified.

Two populations of prelysosomal structures transporting asialoglycoproteins in rat liver.

Analyses by differential centrifugation of liver homogenates from rats that had received 131I-labeled asialoorosomucoid showed that, 1 min after injection, most of the intracellular ligand was associated with a particle that did not sediment at 2.5 X 10(5) g-min. However, by 10 min, undigested ligand became associated with a particle that did sediment at this speed. On analytical ultracentrifugation in sucrose gradients, both kinds of particles exhibited low densities (1.11-1.13 g X ml-1). In contrast to asialoorosomucoid, 125I-labeled asialotransferrin type 3, under noncatabolic conditions, remained largely confined to the nonsedimenting particle regardless of the duration of the study. Induction of catabolism of asialotransferrin was accompanied by the appearance of the ligand in the sedimentable particle. The nonsedimentable particle was separated by immunoadsorption from other subcellular particles contained in the low-density subcellular fraction. The adsorbant , prepared by immobilizing purified antibodies to the Gal/GalN-specific lectin from rat liver on coated polyacrylamide beads, removed 75-80% of the asialoorosomucoid and transferrin binding capacities present, together with a similar portion of the radioligands tested (asialoorosomucoid, asialotransferrin type 3, and human diferric transferrin). Significantly, the sialytransferase activity remained unadsorbed. From these findings, the nonsedimentable particle appears to be involved in the transport of ligands destined to such diverse fates as exocytosis or lysosomal degradation. The sedimentable particle, on the other hand, seems to represent a link between the first particle and the lysosome.

Transferrin glycans: a possible link between alcoholism and hepatic siderosis.

The hepatic uptake of 59Fe from diferric rat and rabbit asialotransferrins and from human transferrin lacking two sialyl residues was investigated in rats in experiments lasting for 1 hr. The 59Fe attached to either of these preparations disappeared from the plasma more rapidly than the 59Fe introduced with the unmodified respective parent proteins. Most of the 59Fe activity that had disappeared from the circulation could be recovered with the liver. Studies with double-labeled (125I, 59Fe) preparations showed that the enhanced 59Fe clearance was not associated with increased catabolism of the modified transferrins. Prolonged, heavy alcohol consumption, as shown by others, results in the appearance of sialic acid-deficient transferrin (two residues missing) in human serum. We suggest that the increased capacity of transferrin deficient in sialic acid to selectively deposit iron in the hepatocyte may be of significance for the development of the hepatic siderosis observed in alcoholism.

Receptor-rich intracellular membrane vesicles transporting asialotransferrin and insulin in liver.

A wide range of receptors are located at the blood sinusoidal aspect of the hepatocyte plasma membrane. Many circulating ligands that bind to receptors on the cell surfaces are interiorized along two pathways. Asialoglycoproteins are transferred from the plasma membrane to lysosomes and degraded, whereas immunoglobulin A and bile acids are transported across the hepatocyte interior and released into bile. Asialotransferrin type 3 (ref. 6) follows a further pathway termed diacytosis. After binding to the asialoglycoprotein receptor, asialotransferrin is endocytosed and then returned to blood with a proportion of its carbohydrate side chains resialylated. We now describe in liver the properties of intracellular asialotransferrin-enclosing vesicles (diacytosomes) and show that they differ from Golgi, lysosome and plasma membrane fractions. Furthermore, we show that the asialoglycoprotein binding sites are located on the cytoplasmic (outer) surface of diacytosomes.

Partial resialylation of human asialotransferrin type 3 in the rat.

After the injection of a small dose (1 micrograms/100 g of body weight) of 125I-labeled human asialotransferrin type 3 in rats, the radioactivity became rapidly associated with the liver. However, during the ensuing 12 hr a significant fraction of the dose returned to the circulation as protein-bound 125I. The protein released by the liver was indistinguishable by gel filtration from the original preparation and was precipitable by an antiserum to human transferrin. Nevertheless, it no longer bound to the immobilized Gal/GalN-specific lectin from rabbit liver. However, binding could be restored to a large extent by treatment with neuraminidase, indicating that the loss of binding was due to resialylation. Changes in the electrophoretic mobility of asialotransferrin released by the liver showed that resialylation was partial--i.e., it involved the attachment of two or three sialyl residues. From analysis by deconvolution of the plasma curve of partially resialylated asialotransferrin it was calculated that the liver "repaired" this way approximately one asialotransferrin molecule out of four. Plasma clearance of partially resialylated asialotransferrin was similar to that of nondesialylated transferrin.

Subcellular distribution of human asialotransferrin type 3 in the rat liver.

A small quantity of 125I-labeled human asialotransferrin type 3 (2 to 4 microgram/100 g) was injected in intact rats and the distribution of the hepatic radioactivity analyzed by fractionation of liver homogenates on continuous sucrose density gradient. The ligand rapidly partitioned between plasma membrane and the interior of the cell at an approximate ratio of 1 to 4. The ratio remained constant between 3 min and 1 h. Intracellular 125I was encapsulated in a particle that was of a median equilibrium density of 1.11 (1.109 to 1.114) g/cm3 at 20 degrees C. The ligand recovered from the particles showed no sign of proteolytic digestion and was bound by the immobilized asialoglycoprotein-binding lectin from rabbit liver. The electron microscopic appearance of the subfractions containing of the entrapped ligand closely resembled that of an intermediate Golgi preparation. Various attempts were made to separate the ligand-containing particles from sialyltransferase and phosphodiesterase I activities, but complete separation could not be accomplished. 125I-Asialoorosomucoid studied in the same quantities and under the same conditions as asialotransferrin, yielded a subcellular distribution which was distinct from that of asialotransferrin type 3. Increasing the dose of asialotransferrin, to a level at which rapid catabolism of this asialoprotein occurs, profoundly changed the subcellular distribution of radioactivity. The subcellular distribution thus obtained was comparable with that found for asialoorosomucoid. These findings suggest that asialotransferrin type 3 is associated with different intracellular vehicles (different endosomes?) depending on whether the protein is simply diacytosed or is en route to lysosomes.

Binding of asialotransferrins by purified rat liver plasma membranes.

Interaction of four different asialotransferrins (human types 1, 2, and 3, and rabbit asialotransferrin) with purified plasma membranes from the rat liver was studied by using a direct binding assay. Binding of rabbit asialotransferrin, possessing a single biantenary glycan, was too weak to establish a complete binding curve, but the human asialotransferrins, possessing two glycan attachments, did yield binding data over a sufficiently wide range of concentrations for Scatchard plot analysis. At 22 degrees C, a quantity of plasma membrane equivalent to 1 mg of membrane protein bound comparable quantities (12.3-12.8 pmol) of the asialotransferrin types with an association constant of 1.5 X 10(6) M-1 for type 1, 1.4 X 10(7) M-1 for type 2, and 1.1 X 10(8) M-1 for type 3. At 4 degrees C, the number of binding sites and the association constants were reduced, more so for asialotransferrin types 2 and 3 than for type 1. At both temperatures, the shapes of the Scatchard plots for all three asialotransferrin types were similar: in the low range of bound asialoprotein (below 0.6-0.7 nM), each plot exhibited two to three convex peaks, tentatively identified as restricted domains of positive cooperativity; in the higher range, however, the plots were linear. The findings are consistent with the view that the binding sites involved in the binding of asialotransferrin are homogenous.