IFN-gamma and transforming growth factor-beta 1 differently regulate fibronectin and laminin receptors of human differentiating monocytic cells.

The regulation of extracellular matrix (ECM) protein receptor expression was followed in the human promonocytic cell line U937 before and after stimulation either with PMA or various cytokines implicated in monocytopoiesis. On undifferentiated U937 cells, alpha-chains of very late Ag (VLA)-4, VLA-5, and VLA-6 were constitutively expressed whereas alpha-chains of VLA-2 (alpha 2) and vitronectin receptor (alpha V) were not. Maturation of U937 cells with PMA resulted in a marked decrease in alpha 4 expression (25% of control by day 5), and a small but significant increase in the expression of alpha 2 and alpha v over 4 days of stimulation. Unstimulated U937 cells attached to fibronectin (FN) but not to laminin (LM), collagens I/IV-coated surfaces. After PMA stimulation, U937 cells exhibited enhanced adherence on FN and expressed the ability to adhere to LM. PMA stimulation also promoted U937 spreading both on FN and LM. Adhesion on FN all along the maturation pathway was specifically and totally inhibited by anti-alpha 5 mAb but not by anti-alpha 4 mAb. Anti-beta 1, anti-alpha 6, anti-alpha 2, and anti-alpha v mAb, as well as Tyr-Ile-Gly-Ser-Arg and Arg-Gly-Asp synthetic peptides from LM, had no effect on adhesion of PMA-stimulated cells on LM, implying that U937 cell adherence to LM is mediated through hitherto distinct receptors. In the presence of rIFN-gamma, differentiating U937 cells did not adhere to LM and lost the capacity to bind to FN. Loss of adhesion to FN was correlated with the concomitant decrease in the expression of alpha 4 and alpha 5 integrin subunits. In contrast, TGF-beta 1 mimicked most of the effects of PMA by enhancing the attachment of maturating U937 cells on FN through alpha 5 receptors and by promoting adherence to LM. TGF-beta 1 stimulation also promoted U937 cell spreading on both FN- and LM-coated surfaces. The data suggest that inflammatory cytokines such as IFN-gamma and TGF-beta 1 may be critically important in the homing of monocytic cells at sites of inflammation by modulating cell-surface expression of ECM receptors.     

Activation mechanisms of platelet-activating factor in U937 cells: possible involvement of protein kinase C.

We have previously demonstrated that platelet-activating factor (PAF) binds specifically on cell membranes isolated from U937 cells. We now describe biological evidence showing that the effect of PAF on U937 cells is a receptor-mediated event. myo-[3H]Inositol-labeled U937 cells were used to investigate the possible role of phosphoinositide metabolism in these cells after binding of PAF. Formation of inositol phosphates (IP1, IP2, and IP3) in response to PAF was increased two- to threefold more than in vehicle control in U937 cells. The effect of PAF on endogenous protein phosphorylation was also studied by using 32PO4-labeled cells. PAF stimulates the phosphorylation of a 45-kDa protein in a time-dependent and dose-related fashion. Since the phospholipase C-generated diglyceride is an important activator of protein kinase C, the phosphorylated 45-kDa protein could be the substrate of protein kinase C. In this regard, we were able to demonstrate that phorbol ester enhances the phosphorylation of the same 45-kDa protein band. In addition, sphingosine, a protein kinase C inhibitor, inhibits the phosphorylation of the same 45-kDa protein band. Down-regulation of the protein kinase C also inhibits the 45-kDa protein phosphorylation. These results suggest that protein kinase C is involved in the PAF-U937 cell interaction.     

CD44 and annexin A2 mediate the C5a chemotactic cofactor function of the vitamin D binding protein

The vitamin D binding protein (DBP) is a plasma protein that significantly enhances the chemotactic activity of C5a and C5a(desArg) (cochemotactic activity). The objective of this study was to investigate how DBP mediates this process using neutrophils and U937 cells transfected with the C5a receptor (U937-C5aR cells) and comparing chemotaxis to C-activated serum (DBP dependent) vs purified C5a (DBP independent). Binding to the cell surface is essential for this protein to function as a chemotactic cofactor, and DBP binds to a chondroitin sulfate proteoglycan (CSPG) on neutrophil plasma membrane preparations. To determine whether a CSPG also functions to mediate cochemotactic activity, U937-C5aR cells were grown in chlorate to inhibit CSPG sulfation or treated with chondroitinase AC. Either treatment significantly inhibited chemotaxis only to C-activated serum. CD44 is a major cell surface CSPG on leukocytes, and functions to facilitate chemotaxis. Treatment of cells with anti-CD44 blocks chemotaxis of neutrophils and U937-C5aR cells to C-activated serum but not purified C5a. DBP binds to CD44 on the cell surface as evidenced by coimmunoprecipitation, confocal microscopy, and cell binding studies. Annexin A2 associates with CD44 in lipid rafts; therefore, its potential role in mediating cochemotactic activity was investigated. Results demonstrate that anti-A2 inhibits neutrophil and U937-C5aR chemotaxis specifically to C-activated serum, blocks DBP binding to cells, and colocalizes with anti-DBP on the cell surface. These results provide clear evidence that CD44 and annexin A2 mediate the C5a chemotactic cofactor function of DBP.     

Identification of the protein HC receptor

In the present study, we demonstrate for the first time the presence of a specific receptor for protein HC on the surface of human cells using the human histiocytic lymphoma cell line U937. Cells treated for 4 days with the maturation inducer phorbol 12-myristate 13-acetate, were found to increase both the number of cells binding protein HC (76% higher than for untreated cells) and the expression of protein HC receptors. Protein HC bound to these cells in a specific and saturable manner. Scatchard analysis at 4 degrees C, using radioiodinated protein HC, indicated a single class of low-affinity receptor (Ka = 2-5 x 10(7) M-1) and 20,000-30,000 receptors per cell. Monoclonal antibodies against protein HC abrogated specific binding of this protein to U937. In contrast, monoclonal antibodies that did not react with protein HC (anti-LFA-1 alpha, anti-MO1 alpha) were without effect on the binding reaction.     

Tumor necrosis factor signal transduction. Cell-type-specific activation and translocation of protein kinase C

We have investigated the changes in protein kinase C (PKC) activity after treatment of several cell lines with TNF. Binding studies with [3H]phorbol dibutyrate (PBt2) on whole cells revealed rapid and transient activation of PKC in Jurkat, K562, and U937 cells with a maximum of phorbol ester binding at 6 min after TNF treatment. As shown by Scatchard analysis, the TNF-induced increase of [3H]PBt2 binding reflected increments of phorbol ester binding site numbers rather than greater binding affinities. Upon subfractionation of TNF-treated U937 cells a transient increase of PBt2 binding in the membrane fraction was accompanied by a long term loss of PBt2-binding in the cytosol, indicating a TNF-induced translocation of PKC from the cytosol to the cell membrane. With histone III-S as a substrate, the determination of specific PKC activity revealed similar kinetics of PKC translocation in U937 cells. TNF also induced PKC translocation in K562 and Jurkat cells. However, although TNF caused long term down-regulation of cytosolic PKC activity in U937 cells, the cytosolic PKC activity only transiently decreased in both Jurkat and K562 cells and then recovered to near basal levels. In the human nonmalignant fibroblast cell line CCD18, PKC was not activated by TNF. Our data suggest that PKC activation may play a major role in TNF signal transduction in some, but not all target cells.     

Human immunodeficiency virus infection of T cells and monocytes proceeds via receptor-mediated endocytosis

The rates of internalization and uncoating of 32P-labelled human immunodeficiency virus (HIV) in the human T lymphoid cell line CEM are consonant with a receptor-mediated endocytosis mechanism of entry. This interpretation was affirmed by electron microscopic observation of virions within endosomes. Virus binding and infectivity were inhibited to the same extent by pretreatment with OKT4A antibody, therefore, the CD4 receptor-dependent pathway of internalization appears to be the infectious route of entry. The pattern of internalization by the human monoblastoid cell line U937 proved to be more complex, involving rapid and efficient CD4-independent internalization. Electron microscopy revealed the presence of large intracellular vesicles, each containing several virions. Antibody against the CD4 receptor for virus efficiently blocked infection, but did not reduce significantly HIV binding or internalization in the U937 cell line. Consequently, U937 cells have a CD4-independent pathway of virus internalization that does not coincide with the route of entry for infectious HIV.     

The role of gangliosides in the interaction of a growth inhibitor with mouse LM cells

We have isolated and characterized glycopeptides, derived from mouse and bovine cerebral cortex cells, that inhibit protein synthesis and cell growth of normal but not transformed cells. The inhibitor binds to target cell surfaces, and gangliosides have previously been shown to influence cell sensitivity to the glycopeptides. Preincubation with 3.0 micrograms/ml ganglioside GM1 at 0 degrees C for 3 hr sensitized the mouse L-cell line to the inhibitor, as determined by protein synthesis assays. Preincubation of LM cells with ganglioside GM1 alone did not affect protein synthesis rates. In addition, the gangliosides GD1a and GM3 also sensitized the LM cells to the protein synthesis inhibitory effect of the glycopeptide inhibitor. Binding experiments were performed with 3T3 (sensitive) and LM (insensitive) cells to determine if sensitivity to the glycopeptide inhibitor was reflected in binding of the inhibitor to these cells. Binding of 125I-labeled inhibitor to 3T3 cells was maximal after 60 min at 0 degrees C and saturable at approximately 1 X 10(4) molecules/cell. Furthermore, binding of the inhibitor was dose-dependent, with half-maximal binding at 1.5-2.0 nM and saturation at 8.0-10.0 nM. Scatchard plot analysis indicated that the Kd was about 1 X 10(-9) M and that there are 1 X 10(4) receptors/cell. Binding of the inhibitor to LM cells was maximal after 30 min at 0 degrees C and saturation occurred at 5 X 10(3) molecules/cell. We then examined the possibility that gangliosides are the cellular receptor or co-receptor for the glycopeptide inhibitor. Binding of the inhibitor to ganglioside GM1 was first examined after the ganglioside had been preadsorbed to polystyrene tubes. These experiments indicated that the ganglioside did not bind the inhibitor. Ganglioside-containing liposomes from phosphatidylcholine or LM cell membrane components were also prepared; these artificial membranes did not bind appreciable amounts of the iodinated inhibitor. Competition experiments showed that the gangliosides GM1 and GD1a did not neutralize the protein synthesis inhibitory activity of the glycopeptides, indicating that gangliosides do not directly interact with the glycopeptide inhibitor. In addition, binding of the inhibitor to LM cells preincubated with ganglioside GM1 was studied. Although the binding of the inhibitor to LM cells was one-half that observed for 3T3 cells, incorporation of exogenous gangliosides into LM cells did not result in increased binding of the inhibitor.(ABSTRACT TRUNCATED AT 400 WORDS)     

Characterization of human monocytic cell line, U937, in taking up acetylated low-density lipoprotein and cholesteryl ester accumulation. A flow cytometric and HPLC study

The uptake of LDL and acetylated LDL and the ability of cholesteryl ester accumulation by cells of a human monocytic cell line, U937, has been characterized by flow cytometric assay using a fluorescent probe, DiI, and by high-performance liquid chromatography (HPLC). The increase of mean fluorescence intensity of U937 incubated with DiI-labeled lipoproteins demonstrates that this cell line could incorporate DiI-AcLDL, as well as DiI-labeled LDL. Competition and saturation studies indicate that the manner of taking up DiI-AcLDL is receptor-mediated. While differentiated U937 incubated with 16 nM phorbol myristate acetate for 24 h took up little DiI-AcLDL, HPLC analysis confirmed that intracellular free and esterified cholesterols significantly increase in the U937 cells incubated with AcLDL or LDL. The ability of mouse peritoneal macrophage to abundantly accumulate at least five kinds of cholesteryl ester were also shown in this analysis. In contrast, in U937 cells, free fatty acids are incorporated into various substances rather than into cholesteryl esters (as revealed by HPLC analysis), so that the cholesterol in AcLDL taken up by U937 cells is not synthesized into cholesteryl esters to any great extent.     

Phorbol esters inhibit the binding of low-density lipoproteins (LDL) to U-937 monocytelike cells

The present study demonstrates that U-937 monocytelike human cells possess specific LDL receptors. 125I-LDL binds at 4 degrees C on the cell surface. The bound molecules are releasable by heparin. The reaction requires Ca2+ and the binding sites are sensitive to proteolysis. Unlabeled LDL compete with 125I-LDL, whereas HDL are ineffective. At 37 degrees C, LDL are internalized and degraded by a chloroquine-sensitive pathway. Tumor-promoting phorbol esters inhibit the binding of 125I-LDL to its receptor on U-937 cells. This inhibition exhibits temperature, time, and concentration dependence. At 37 degrees C, inhibition is 50% at 5 X 10(-9) M of TPA. After removal of phorbol esters, treated cells recover their 125I-LDL-binding activity in 60 min. The inhibitory activities of various phorbol esters are proportional to their tumor-promoting activities. Inhibition appears to be due to a reduction in the number of available LDL receptors rather than a decrease in receptor affinity.     

Receptor-mediated internalization of Pseudomonas toxin by mouse fibroblasts

Pseudomonas exotoxin (PE) was used as a probe to study the mechanism by which protein ligands are internalized by mammalian cells. Both biochemical and electron microscopic methods were used to look at the internalization of PE by mouse LM cell fibroblasts. Our data suggest that PE enters cells by receptor-mediated endocytosis, a process previously thought to be restricted to the entry of biologically significant molecules such as lysosomal enzymes and peptide hormones. Biochemical studies showed that methylamine (20 mM) and chloroquine (10 microM) protected LM cells from the action of PE. Full protection was observed if methylamine or chloroquine was added to the monolayers simultaneously with toxin or if they were added up to 10 min after toxin binding. Later addition of amine or chloroquine afforded partial protection to the monolayers. With immunoelectron microscopy we observed that in the cold toxin bound diffusely to the cell surface but was rapidly internalized when cells were warmed to 37 degrees C. In the presence of methylamine, chloroquine or ammonium chloride, internalization did not occur. We propose that PE enters mouse fibroblasts by receptor-mediated endocytosis and that chloroquine and methylamine, agents which are known to block this process, prevent expression of toxicity.     

Binding and crosslinking of 125I-labeled recombinant human tumor necrosis factor to cell surface receptors

Highly purified recombinant human tumor necrosis factor (TNF) (molecular mass determined as 17 kilodaltons (kDa) by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and as 36 kDa by Sephadex G-100 gel chromatography) was labeled with 125I to a specific activity of 5 microCi/micrograms without appreciable loss of activity. The binding of 125I-TNF to eighteen human and twelve animal cell lines was examined. The binding varied considerably among different cell lines. In most cell lines, the binding was inhibited up to greater than 90% by the addition of a 100-fold excess of unlabeled TNF. Some human and mouse cell lines showed no significant binding above background levels, suggesting that these cell lines had no receptors for TNF. Among the TNF receptor-positive cell lines, there was no direct correlation between the level of specific TNF binding and the level of sensitivity to the cytotoxic or cytostatic effect of TNF. Some cell lines were sensitive to TNF, whereas others were not affected at all by TNF. The TNF receptor-negative cell lines were also resistant to TNF. Therefore, although the existence of TNF receptor seems to be necessary, it does not alone determine cellular sensitivity to TNF. Scatchard analysis of the binding data revealed that human HeLa S3 and THP-1 had about 50,000 and 10,000 receptors/cell with a dissociation constant (KD) of 0.3-0.5 nM, respectively. Similarly, mouse L-929 and L-M cells had about 5,000 receptors/cell with KD of 3-5 nM. 125I-TNF bound to HeLa S3 cells was rapidly internalized at 37 degrees C, presumably by receptor-mediated endocytosis, and degraded to acid-soluble products. The turnover of TNF receptors on HeLA S3 cells seemed to be rapid, since the level of specific binding quickly decreased after treatment with 100 micrograms/ml of cycloheximide at 37 degrees C with a half-life of about 1.5 h. The crosslinking of the cell-bound 125I-TNF with the use of disuccinimidyl suberate yielded a complex of 105 kDa for HeLa S3 and THP-1 cells, and a complex of 100 kDa for U937 cells. The crosslinking was completely inhibited by the addition of a 100-fold excess of unlabeled TNF. Assuming that the complex was due to a one-to-one association of the dimeric form of TNF (34 kDa) with the receptor, we estimated the molecular size of the human TNF receptor to be 71 kDa for HeLa S3 and THP-1, and 66 kDa for U937.     

Morphine inhibits indolactam V-induced U937 cell adhesion and gelatinase secretion

We demonstrate that indolactam V, a non-phorbol protein kinase C activator, promotes U937 cell attachment to fibronectin, type IV collagen and laminin. In the absence of indolactam V, 2-4% of U937 cells attach to all test substrates, however, in the presence of 100 nM indolactam V, 25, 16 and 11% of U937 cells attach to fibronectin, type IV collagen and laminin, respectively. When added concomitantly, 90 microM H-7, a protein kinase C inhibitor, reduces indolactam V-induced U937 cell adhesion to fibronectin by 91%. Monoclonal antibodies directed against both the beta1 and alpha 5 integrin subunits inhibit indolactam V-induced U937 cell adhesion to fibronectin by 62 and 52%, respectively. Indolactam V also promotes homotypic aggregation in U937 cells, which is blocked with either anti-ICAM or anti-LFA-1 antibodies. In addition, indolactam V promotes U937 cell secretion of a 92 kDa gelatinase as demonstrated by zymography. In the presence of low levels of morphine (10 nM-1.0 microM), the U937 cell attachment to matrix proteins was not significantly affected. However, in the presence of 10 microM morphine, the indolactam V treated cells exhibit a 71-74% reduction in cell adhesion to the matrix proteins. Further, 10 microM morphine also blocks indolactam V-induced homotypic aggregation and gelatinase secretion. The inhibitory effect of morphine on cell-matrix adhesion and gelatinase secretion was not inhibited by the opiate receptor antagonist naloxone (1 microM). While 10 microM naloxone did partially counteract the effect of 10 microM morphine on U937 cell attachment, this effect was likely non-specific since 10 microM naloxone alone increased cell adhesion. Supporting this conclusion, PCR analysis revealed that U937 cells do not express the mu high affinity morphine receptor. Also, indolactam V did not induce mu receptor expression, suggesting that morphine acts on U937 cells in a non-specific fashion.     

Complement lysis of U937, a nucleated mammalian cell line in the absence of C9: effect of C9 on C5b-8 mediated cell lysis

Previous studies have demonstrated that in general, nucleated cells are more resistant to killing by serum complement than are erythrocytes. During studies aimed at defining the mechanisms of nucleated cell resistance, we found that the human histiocytic cell line U937 was easily lysed by homologous serum. U937 cells were also killed by serum depleted of C9, but not by serum depleted of C8, implying that the C5b-8 complex was sufficient to cause lysis of these cells. Enumeration of complexes on the cell surface demonstrated that approximately 40-fold more complexes were required to lyse U937 cells in the absence of C9 than in the presence of an excess of C9. Examination of the effects of small amounts of C9 on lysis of U937 cells by the C5b-8 complex demonstrated that at very low doses, C9 inhibited C5b-8 mediated lysis. The use of radiolabeled anti-C8 antibody showed that C5b-8 complexes were eliminated from the surface of U937 cells at 37 degrees C, and C9 at the dose causing inhibition of lysis accelerated the elimination of complexes. These results suggest that the increased lytic potential resulting from binding of small amounts of C9 to C5b-8 complexes is outweighed by enhanced elimination of complexes resulting in decreased cell death.     

Plasma membrane and intracellular pools of transferrin receptors decline during in vitro cultivation of U937 cells

Transferrin receptor expression in the monocyte-like cell line U937 was investigated during in vitro cultivation. U937 cells expressed a single class of high affinity surface transferrin receptors (KD approximately 4 nM), with apparent subunit Mr of 90-95,000 Da as determined by SDS-reducing PAGE. [125I]-transferrin binding studies on detergent-solubilized cells revealed that half to two-thirds of the total functional binding sites were located intracellularly. Radioligand binding, immunofluorescence and flow cytometry studies were performed on intact, detergent-solubilized, or saponin-permeabilized cells, using either transferrin or the anti-transferrin receptor monoclonal antibody OKT9 IgG. These studies demonstrated that functional and antigenic transferrin receptor levels were maximal on cells 24 h after subculture at low density and declined during the culture period. Scatchard analysis of radioligand binding data suggested that the decline in functional transferrin binding sites resulted from a decline in the number of available receptors. These results demonstrate that in U937 cells there is a density-dependent regulation of transferrin receptor expression, resulting in a loss of functional and antigenic receptors from both plasma membrane and intracellular locations.     

N-acetyl-beta-endorphin1-31 antagonizes the suppressive effect of beta-endorphin1-31 on murine splenocyte proliferation via a naloxone-resistant receptor.

High affinity binding sites for beta-endorphin1-31 (beta-EP) have been observed on transformed mononuclear cells such as the human U937 monocyte-like cell line and the murine EL4-thymoma line, and on normal murine splenocytes. Binding of beta-EP at these sites is resistant to competition by naloxone and other opiate receptor ligands but sensitive to N-acetyl-beta-endorphin1-31 (N-Ac), cations and GTP-gamma-sulfate. Thus, the following studies were done to determine the functional significance of binding beta-EP and N-Ac. beta-EP suppressed phytohemagglutinin (PHA)-stimulated [3H]thymidine uptake in a dose-dependent, naloxone-insensitive fashion. beta-Endorphin1-27, (des)-tyrosine beta-endorphin2-31, or N-Ac failed to duplicate the suppressive effect of beta-EP. However, N-Ac, which is equipotent to beta-EP at displacing 125I-beta-EP bound to murine splenocytes or U937 cells, antagonized the suppressive effect of beta-EP. Taken together with previous binding studies, the present observations suggest that beta-EP effects receptor-mediated responses on normal immunocytes that do not depend on the activation of neuronal-like opiate receptors which are naloxone-sensitive. N-Ac, which shows minimal binding to such brain opiate receptors, is a potent functional antagonist of the naloxone-insensitive immunocyte receptor for beta-EP.     

Characterization of receptors for immune interferon in U937 cells with 32P-labeled human recombinant immune interferon

Recombinant human immune interferon (HuIFN-gamma) was labeled with [gamma-32P]ATP and cyclic-AMP-dependent protein kinase from bovine heart to a specific radioactivity of 11,000 Ci/mmol. At least two molecules of phosphate were incorporated per molecule of interferon. The binding of [32P]HuIFN-gamma to human U937 histiocytic lymphoma cells was time dependent, and displaceable by HuIFN-gamma but not by HuIFN-alpha A or HuIFN-beta. The specific binding was saturable with less than 10% nonspecific binding. The dissociation constant of [32P]HuIFN-gamma for U937 interferon receptors was calculated to be 1.5 X 10(-10) M with a total of 1,800 binding sites/cell. Dissociation of bound [32P]IFN-gamma at 24 degrees C exhibited two distinct rates. A fast dissociation with a specific rate constant of 0.141 min-1, and a slow dissociation with a specific rate constant of 0.0027 min-1. The Kd for [32P]HuIFN-gamma was calculated from kinetic constants to be 5.4 X 10(-10) M.     

Binding of platelet activating factor by isolated membranes from U937 cells

Platelet activating factor (PAF), 1-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine (1-O-alkyl-2-acetyl-GPC) has been known to have biological effect on cells. The mechanisms of the effect of the potent phospholipid on cells has not been established. We have used 1-O-[3H]alkyl-2-acetyl-GPC [( 3H]PAF) to study the interaction on the isolated membranes of U937 cells. The binding process was time, protein concentration, temperature dependent and reversible. The binding of [3H]PAF to the U937 cell membranes was slightly inhibited by the addition of PAF analogue, 3-O-Hexadecyl-2-acetyl-sn-glycerol-1-phosphorylcholine. U937 cell membranes showed high affinity binding sites for PAF with equilibrium dissociation constant (Kd) of 5 x 10(-9) M. The displacement of bound [3H]PAF with 500-fold excess of nonlabeled PAF was not altered suggesting that the bound [3H]PAF was not degraded during the binding. Binding of [3H]PAF on U937 cell membranes was inhibited by PAF antagonist, 59227RP. The kinetic of the inhibition by PAF antagonist is competitive suggesting that PAF and PAF antagonist bind at the same site.     

Binding sites for elastase on cultured human fibroblasts that do not mediate internalization

The proteolytic actions of elastases have been implicated in extracellular matrix damage, which is characteristic of a variety of pathological conditions including emphysema and rheumatoid arthritis. In order to elucidate the molecular events involved in elastase interaction with connective tissue cells, the present study was designed to investigate the association of elastase with human fibroblasts at 4 degrees C. Elastase bound saturably to binding sites that were present on the surface of these cells. Analysis of cell-bound elastase by sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed the presence of a high molecular weight complex (Mr 54,000) that was not formed with elastase whose catalytic site serine was derivatized with a diisopropylphosphate group. The complex did not represent elastase bound to either protease nexin or contaminating serum. The cellular component with which elastase formed a complex could not be detected in the cell culture medium. Unexpectedly, elastase that had been pre-bound at 4 degrees C was not internalized after cells were warmed to 37 degrees C. The elastase binding site described in this report is therefore distinct from high affinity binding sites involved in receptor-mediated endocytosis and intracellular degradation.