A 49-residue peptide from adhesin F1 of Streptococcus pyogenes inhibits fibronectin matrix assembly

F1 is an adhesin of Streptococcus pyogenes which binds the N-terminal 70-kDa region of fibronectin with high affinity. The fibronectin binding region of F1 is comprised of a 43-residue upstream domain and a repeat domain comprised of five tandem 37-residue sequences. We investigated the effects of these domains on the assembly of fibronectin matrix by human dermal fibroblasts, MG63 osteosarcoma cells, or fibroblasts derived from fibronectin-null stem cells. Subequimolar or equimolar concentrations of recombinant proteins containing both the upstream and repeat domains or just the repeat domain enhanced binding of fibronectin or its N-terminal 70-kDa fragment to cell layers; higher concentrations of these recombinant proteins inhibited binding. The enhanced binding did not result in greater matrix assembly and was caused by increased ligand binding to substratum. In contrast, recombinant or synthetic protein containing the 43 residues of the upstream domain and the first 6 residues from the repeat domain exhibited monophasic inhibition with an IC(50) of approximately 10 nm. Truncation of the 49-residue sequence at its N or C terminus caused loss of inhibitory activity. The 49-residue upstream sequence blocked incorporation of both endogenous cellular fibronectin and exogenous plasma fibronectin into extracellular matrix and inhibited binding of 70-kDa fragment to fibronectin-null cells in a fibronectin-free system. Inhibition of matrix assembly by the 49-mer had no effect on cell adhesion to substratum, cell growth, formation of focal contacts, or formation of stress fibers. These results indicate that the 49-residue upstream sequence of F1 binds in an inhibitory mode to N-terminal parts of exogenous and endogenous fibronectin which are critical for fibronectin fibrillogenesis.     

Identification of a novel recognition sequence for fibronectin within the NH2-terminal beta-sandwich domain of tissue transglutaminase

Tissue transglutaminase belongs to the multigene transglutaminase family of Ca2+-dependent protein cross-linking enzymes. Unlike other transglutaminases, it is involved in cell-matrix interactions and serves as an adhesion co-receptor for fibronectin. Previous work established that the fibronectin-binding motif(s) is located within the NH2-terminal proteolytic fragment of the protein consisting of residues 1-272. Here we identify a novel fibronectin recognition site within this sequence of tissue transglutaminase. Substitution of individual domains of tissue transglutaminase with those from homologous factor XIIIA showed that the major fibronectin-binding site is present within the first beta-sandwich domain of the protein. Experiments with deletion mutants of the first domain revealed that amino acids 81-140 of tissue transglutaminase are involved in fibronectin binding. Using synthetic peptides encompassing this region, we found that the peptide 88WTATVVDQQDCTLSLQLTT106 inhibited the interaction of tissue transglutaminase with fibronectin and decreased transglutaminase-dependent cell adhesion and spreading. In the three-dimensional structure of the first domain, amino acids 88-106 comprise an extended hairpin formed by antiparallel beta strands 5 and 6. Mutations of Asp94 and Asp97 within the beta5/beta6 hairpin to Ala significantly reduced the affinity of tissue transglutaminase for fibronectin, indicating that these residues are critical for fibronectin binding. Identification of the fibronectin-binding site on tissue transglutaminase will help to dissect the role of this protein in cell-matrix interactions.     

Production of Monoclonal Antibodies to Guinea Pig Liver Transglutaminase

Monoclonal antibodies are now a powerful tool in biology and medicine. Transglutaminase has been implicated in diverse biological functions, and the characteristics of its catalytic action are suitable for applied enzymology. In this study, we produced hybridoma cells which synthesize monoclonal antibodies against guinea pig liver transglutaminase by fusing mouse myeloma cells with spleen cells of mouse immunized with the enzyme protein. Eight hybridoma clones (coded 2F, 4B, 7C, 8B, 8D, 8E, 9F and 11C) were selected to produce monoclonal antibodies. The subclass of IgG produced by clone 9F was IgG_2a and those from the seven other clones were all IgG₁ The 9F antibody inhibited transglutaminase activity, but the other antibodies did not. A solid-phase antibody-binding assay showed that of these antibodies, 8D antibody has the highest affinity to the antigen. Transglutaminase protein in crude liver extract was identified with Western blotting analysis using 8D antibody as the probe.     

Induction of fibronectin matrix assembly in human fibrosarcoma cells by dexamethasone

Previous studies have suggested that the assembly of fibronectin into the extracellular matrix of cultured fibroblasts is mediated by specific matrix assembly receptors that recognize a binding site in the amino terminus of the fibronectin molecule (McKeown-Longo, P.J., and D.F. Mosher, 1985, J. Cell Biol., 100:364-374). In the presence of dexamethasone, human fibrosarcoma cells (HT-1080) acquired the ability to specifically bind exogenous plasma fibronectin and incorporate it into a detergent-insoluble extracellular matrix. Dexamethasone-induced fibronectin binding to HT-1080 cells was time dependent, dose dependent, and inhibited by cycloheximide. Saturation binding curves indicated that dexamethasone induced the appearance of 7.7 X 10(4) matrix assembly receptors per cell. The induced receptors exhibited a dissociation constant (KD) for soluble fibronectin of 5.0 X 10(-8) M. In parallel experiments, normal fibroblasts exhibited 4.1 X 10(5) receptors (KD = 5.3 X 10(-8) M) per cell. In the presence of cycloheximide, the induced fibronectin-binding activity on HT-1080 cells returned to uninduced levels within 12 h. In contrast, fibronectin-binding activity on normal fibroblasts was stable in the presence of cycloheximide for up to 54 h. The first-order rate constant (Kt = 2.07 X 10(-4) min-1) for the transfer of receptor-bound fibronectin to extracellular matrix was four- to fivefold less than that for normal fibroblasts (Kt = 1.32 X 10(-3) min-1). Lactoperoxidase-catalyzed iodination of HT-1080 monolayers indicated that a 48,000-mol-wt cell surface protein was enhanced with dexamethasone. The results from these experiments suggest that dexamethasone induces functional matrix assembly receptors on the surface of HT-1080 cells; however, the rate of incorporation of fibronectin into the matrix is much slower than that of normal fibroblasts.     

Transforming growth factor beta increases cell surface binding and assembly of exogenous (plasma) fibronectin by normal human fibroblasts.

Transforming growth factor beta (TGF-beta) enhances the cell surface binding of 125I-fibronectin by cultured human fibroblasts. The effect of TGF-beta on cell surface binding was maximal after 2 h of exposure to TFG-beta and did not require epidermal growth factor or protein synthesis. The enhancement was dose dependent and was found with the 125I-labeled 70-kilodalton amino-terminal fragment of fibronectin as well as with 125I-fibronectin. Treatment of cultures with TGF-beta for 6 h resulted in a threefold increase in the estimated number of fibronectin binding sites. The increase in number of binding sites was accompanied by an increased accumulation of labeled fibronectin in detergent-insoluble extracellular matrix. The effect of TGF-beta was biphasic; after 6 h of exposure, less labeled fibronectin bound to treated cultures than to control cultures. Exposure of cells to TGF-beta for greater than 6 h caused a two- to threefold increase in the accumulation of cellular fibronectin in culture medium as detected by a quantitative enzyme-linked immunosorbent assay. The second phase of the biphasic effect and the increase in soluble cellular fibronectin were blocked by cycloheximide. Immunofluorescence staining of fibroblast cultures with antifibronectin revealed that TGF-beta caused a striking increase in fibronectin fibrils. The 70-kilodalton amino-terminal fragment of fibronectin, which blocks incorporation of fibronectin into extracellular matrix, blocked anchorage-independent growth of NRK-49F cells in the presence of epidermal growth factor. Our results show that an increase in the binding and rate of assembly of exogenous fibronectin is an early event preceding the increase in expression of extracellular matrix proteins. Such an early increase in cell surface binding of exogenous fibronectin may be a mechanism whereby TGF-beta can modify extracellular matrix characteristics rapidly after tissue injury or during embryonic morphogenesis.     

Isolation of an amino-terminal fibronectin-binding protein on human U937 cells and rat peritoneal macrophages.

Several cell-mediated activities for the amino terminus of fibronectin have been documented. In the present study we describe a macrophage surface protein with binding activity directed to the amino terminus of the fibronectin molecule. The binding of a 29-kDa amino-terminal fibronectin fragment to macrophages reached steady state by 30 min and was half-maximal at approximately 2 x 10(-8) M. This binding was specifically inhibited by excess unlabeled 29-kDa fragment or intact fibronectin but not by a 180-kDa fibronectin fragment which lacks the amino terminus. Competitive binding studies of the 70-kDa amino-terminal fibronectin fragment to macrophages revealed a single binding site with KD = 7.14 x 10(-8) M and approximately 8 x 10(4) binding sites/cell. Radiolabeled surface proteins extracted from rat peritoneal macrophages and from the human U937 cell line were applied to an affinity column comprised of the 70-kDa amino-terminal fragment of fibronectin coupled to a solid support. A single trypsin-sensitive radiolabeled protein of 67 kDa, from either cell type, was eluted from this column with urea. This protein showed no immunologic identity with fibronectin, fibrin(ogen), or albumin. The 67-kDa protein exhibited identical apparent molecular weight under reducing and nonreducing conditions, as assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis. We have localized the fibronectin binding activity of this protein to within the 29-kDa amino-terminal domain of fibronectin. The 67-kDa protein eluted from the 70-kDa column failed to bind to a column comprised of the 45-kDa gelatin-binding fragment of fibronectin. Additionally, the 67-kDa protein was specifically eluted from the 70-kDa column by the 29-kDa amino-terminal fragment but not by the 45-kDa gelatin-binding fragment. These data suggest that this 67-kDa protein is a macrophage cell surface binding protein for the amino terminus of fibronectin.     

Dissociation of fibrinogen and fibronectin binding from transglutaminase-mediated cross-linking at the hepatocyte surface

The interaction of fibrinogen and fibronectin with hepatocytes has been dissociated into distinct binding and cross-linking steps. Binding and cross-linking of 125I-labeled ligands were both decreased by transglutaminase inhibitors, but not by heparin or hirudin. Transglutaminase activity was manifest by Ca2+-dependent incorporation of [14C]putrescine into cells. Preferential cross-linking of fibrinogen A alpha over gamma chains, and lack of inhibition by heparin or hirudin indicates the involvement of tissue transglutaminase, and not Factor XIIIa. Hepatic transglutaminase activity, as well as binding and cross-linking of fibrinogen and fibronectin, were maximally supported by Ca2+, partially supported by Mn2+ and Sr2+, and markedly decreased by Mg2+ and Ba2+. In contrast, Co2+ supported binding but not cross-linking or transglutaminase activity, indicating that binding and cross-linking are dissociable events. This conclusion was corroborated by the finding that fibrinogen fragments D95 and D78 both inhibited Ca2+-dependent fibrinogen binding without being cross-linked themselves. Ligand binding in the presence of either cation was localized to the cell surface as evidenced by its trypsin sensitivity. Thus, fibrinogen and fibronectin binding to hepatocytes is independent of transglutaminase activity, whereas cross-linking of these adhesive macromolecules requires an enzymatically active cellular transglutaminase. In addition, fibrinogen binding appears to be mediated by molecular determinants present in fragment D78.     

Interaction of the 70,000-mol-wt amino-terminal fragment of fibronectin with the matrix-assembly receptor of fibroblasts

Plasma fibronectin binds saturably and reversibly to substrate-attached fibroblasts and is subsequently incorporated into the extracellular matrix (McKeown-Longo, P.J., and D. F. Mosher, 1983, J. Cell Biol., 97:466-472). We examined whether fragments of fibronectin are processed in a similar way. The amino-terminal 70,000-mol-wt catheptic D fragment of fibronectin bound reversibly to cell surfaces with the same affinity as intact fibronectin but did not become incorporated into extracellular matrix. The 70,000-mol-wt fragment blocked binding of intact fibronectin to cell surfaces and incorporation of intact fibronectin into extracellular matrix. Binding of the 70,000-mol-wt fragment to cells was partially abolished by cleavage into 27,000-mol-wt heparin-binding and 40,000-mol-wt gelatin-binding fragments and more completely abolished by reduction and alkylation of disulfide bonds. Binding of the 70,000-mol-wt fragment to cells was not blocked by gelatin or heparin. When coated onto plastic, the 70,000-mol-wt fragment did not mediate attachment and spreading of suspended fibroblasts. Conversely, fibronectin fragments that had attachment and spreading activity did not block binding of exogenous fibronectin to substrate-attached cells. These results indicate that there is a cell binding site in the 70,000-mol-wt fragment that is distinct from the previously described cell attachment site and is required for assembly of exogenous fibronectin into extracellular matrix.     

Enhanced transglutaminase activity in transformed human lung fibroblast cells after exposure to sodium butyrate

The low level of transglutaminase activity in virus-transformed human embryonic lung fibroblasts (WI-38 VA13A) increased markedly when cells were exposed to sodium butyrate. The effect of sodium butyrate was time- and concentration-dependent and fully reversible. Transformed cells exposed for 5 days to 1 mM sodium butyrate had fewer cells, showed an 8–10-fold higher transglutaminase activity, and stained more abundantly for transglutaminase and pericellular fibronectin than control cells when examined by indirect immunofluorescence. Non-transformed cells (WI-38) showed only a 2–4-fold increase in transglutaminase activity when treated in a similar manner. Studies with metabolic inhibitors revealed the increase in activity was the result of synthesis of new protein. Kinetic studies showed the affinity of putrescine for the enzyme was essentially unchanged but the number of active sites increased 9-fold following exposure to sodium butyrate. Enhanced transglutaminase activity returned to control levels within 7 days after subculture and sodium butyrate removal. These findings suggest that sodium butyrate offers a potential model system to understand better the role of transglutaminase in cells in culture; particularly growth control in transformed cells.     

TSG-6 binds via its CUB_C domain to the cell-binding domain of fibronectin and increases fibronectin matrix assembly.

Human plasma fibronectin binds with high affinity to the inflammation-induced secreted protein TSG-6. Fibronectin binds to the CUB_C domain of TSG-6 but not to its Link module. TSG-6 can thus act as a bridging molecule to facilitate fibronectin association with the TSG-6 Link module ligand thrombospondin-1. Fibronectin binding to TSG-6 is divalent cation-independent and is conserved in cellular fibronectins. Based on competition binding studies using recombinant and proteolytic fragments of fibronectin, TSG-6 binding localizes to type III repeats 9-14 of fibronectin. This region of fibronectin contains the Arg-Gly-Asp sequence recognized by alpha5beta1 integrin, but deletion of that sequence does not prevent TSG-6 binding, and TSG-6 does not inhibit cell adhesion on fibronectin substrates mediated by this integrin. This region of fibronectin is also involved in fibronectin matrix assembly, and addition of TSG-6 enhances exogenous and endogenous fibronectin matrix assembly by human fibroblasts. Therefore, TSG-6 is a high affinity ligand that can mediate fibronectin interactions with other matrix components and modulate some interactions of fibronectin with cells.     

Catalytic utilization of eIF-2 and mRNA binding proteins are limiting in lysates from vesicular stomatitis virus infected L cells

Infection of mouse L cells by vesicular stomatitis virus results in the inhibition of cellular protein synthesis. Lysates prepared from these infected cells are impaired in their ability to translate endogenous or exogenous cellular and viral mRNAs. The ability of initiation factors from rabbit reticulocytes to stimulate protein synthesis in these lysates was examined. Preparations of eukaryotic initiation factor 2 (eIF-2) and the guanine nucleotide exchange factor (GEF) stimulated protein synthesis strongly in L cell lysates from infected cells but only slightly in lysates from mock-infected cells. Maximal stimulation was obtained when a fraction containing eukaryotic initiation factors 4B (eIF-4B) and 4F (eIF-4F) was also present. In lysates from infected cells, these initiation factors increased endogenous cellular mRNA translation on the average 2-fold. In contrast, endogenous viral mRNA translation was increased to a much greater extent: the M protein was stimulated 8-fold, NS 5-fold, N 2.5-fold, and G 12-fold. When fractions containing eIF-4B, eIF-4F, or eIF-4A were added to these lysates in the presence of eIF-2, all three stimulated translation. Fractions containing rabbit reticulocyte initiation factors eIF-3 and eIF-6 had no effect on translation in either lysate. The results suggest that lysates from infected L cells are defective in the catalytic utilization of eIF-2 and deficient in mRNA binding protein activity.     

Visualization of purified fibronectin-transglutaminase complexes.

It has been reported previously (Turner, P.M., and Lorand, L. (1989) Biochemistry 28, 628-635) that human erythrocyte transglutaminase forms a noncovalent complex with human plasma fibronectin near its collagen-binding domain. In the present study, we show by nondenaturing electrophoresis that guinea pig liver transglutaminase, similarly to the erythrocyte enzyme, forms a complex with human fibronectin. Studies of anisotropic shifts of fluorescein-labeled liver and erythrocyte transglutaminases, upon addition of fibronectin, indicated that both transglutaminases bind to fibronectin with a stoichiometry of about 2:1. Polymerization of fibrinogen by human erythrocyte transglutaminase was inhibited after complex formation with fibronectin. Complexes of fibronectin with either erythrocyte or liver transglutaminase were isolated by glycerol gradient zone sedimentation and examined by rotary shadowing electron microscopy. The globular transglutaminase could be readily identified binding to the thin fibronectin strand. The binding site for transglutaminase was within 5-10 nm of the N terminus of fibronectin, consistent with its proximity to the collagen-binding domain. Under some experimental conditions, the complex of fibronectin with erythrocyte transglutaminase appeared as a ring-shaped structure in which two transglutaminase molecules had probably dimerized. The molecular weight of the erythrocyte transglutaminase was determined by sedimentation equilibrium to be 71,440 +/- 830.     

Altered type I protein kinase in adhesion defective CHO cell variants

ADv cells are Chinese hamster ovary (CHO) cell variants which cannot adhere to fibronectin coated substrata (Harper & Juliano: J. Cell Biol., 1980; Nature 1981a,b). We have shown that the defect in some clones of ADv cells is distal to the initial interaction between fibronectin and its cell surface receptors (Cheung and Juliano: Exp. Cell Res., 1984), and that it extends to fibronectin mediated aggregation and endocytosis. The adhesion defect in some ADv clones can be corrected by raising intracellular cAMP levels (Cheung & Juliano: J. Cell Physiol., 1985). Here we examine the protein kinase activities and phosphorylation patterns in an adhesion defective variant clone ADv F11CA11. Analysis of the cAMP dependent protein kinase activity (cAdPK) in crude extracts of F11CA11 cells shows an apparent increase in K (activation) as compared to wild type (WT) CHO cell extracts. Further, the DE-52 cellulose chromatography profile of cAdPK in the F11CA11 variant is markedly different from WT in that the type I cAdPK peak elicited by 1 microM cAMP is essentially missing in F11CA11, while the type II cAdPK peak is similar to that in WT. Raising the cAMP level to 100 microM elicits a type I peak in F11CA11 with about 45% of the activity of the WT peak. Binding studies with 3H-cAMP reveal that the type I peak in F11CA11 has a Kd of 1.7 X 10(-8) M as compared to 2.0 X 10(-9) M for WT, whereas the type II peak Kd is approximately 1 X 10(-9) M for both WT and F11CA11. Two-dimensional polyacrylamide gel analysis of 32Pi labeled WT cells and F11CA11 cells with or without cAMP treatment reveals the presence of a protein(s) of 50 kilodaltons which is phosphorylated in WT cells and in cAMP treated F11CA11 cells but not in untreated F11CA11 cells. These findings, coupled with our previous observations, strongly indicate that the adhesion defect in ADvF11CA11 cells is associated with an altered type I cAdPK having lower affinity for cAMP. At normal cellular cAMP levels this enzyme fails to phosphorylate one or more critical protein substrates; however, by raising internal cAMP levels, the defect can be overcome. Thus type I cAdPK seems to play an important role in the regulation of fibronectin mediated cell adhesion, cell aggregation, and endocytosis.     

Factor XIII cross-linking of fibronectin at cellular matrix assembly sites

We describe the effect of activated Factor XIII (Factor XIIIa, plasma transglutaminase) on the incorporation of plasma fibronectin into extracellular matrix by cultured human fibroblasts. In the absence of added Factor XIIIa, fibronectin binds to cultured fibroblast cell layers and is assembled into disulfide-bonded multimers of the extracellular matrix. When Factor XIIIa was included in the binding medium of skin fibroblasts, accumulation of 125I-fibronectin in the deoxycholate-insoluble matrix was increased. Fibronectin accumulating in the cell layer was cross-linked into nonreducible high molecular weight aggregates. The 70-kDa amino-terminal fragment of fibronectin inhibited the binding and cross-linking of 125I-fibronectin to cell layers, whereas fibrinogen had little effect. When 125I-fibronectin was incubated with isolated matrices or with cell layers pretreated with cytochalasin B, it did not bind and could not be cross-linked by Factor XIIIa into the matrix. HT-1080 human fibrosarcoma cells bound exogenous fibronectin following treatment with dexamethasone; Factor XIIIa cross-linked the bound fibronectin and caused its efficient transfer to the deoxycholate-insoluble matrix. These results indicate that exogenous fibronectin is susceptible to Factor XIIIa-catalyzed cross-linking at cellular sites of matrix assembly. Thus, Factor XIIIa-mediated fibronectin cross-linking complements disulfide-bonded multimer formation in the stabilization of assembling fibronectin molecules and thus enhances the formation of extracellular matrix.     

Localization of cellular transglutaminase on the extracellular matrix after wounding: characteristics of the matrix bound enzyme.

Extending our previous observation that tissue transglutaminase (TGase) binds to extracellular matrix (ECM) fibronectin, we report here that endogenous tissue TGase is localized on the adjacent ECM after puncture wounding embryonic human lung fibroblasts (WI-38). The bound TGase persisted at the wound site for many hours, demonstrated by immunofluorescence and by catalytic activity using an overlay assay. The binding characteristics of TGase with ECM were studied further by the addition of exogenous TGase to cell monolayers and monitoring by immunofluorescence or overlay catalytic activity assays. Binding occurred equally well at 4 degrees C or 37 degrees C. Prior incubation of exogenous TGase with guanosine 5'-triphosphate (GTP), guanosine 5'-diphosphate (GDP), or adenosine triphosphate (ATP) had little effect on the amount bound to matrix, but prior treatment with calcium, magnesium, strontium, or manganese ions enhanced binding 2- to 3-fold. The Ca(++)-dependent change was a concentration-dependent effect on soluble exogenous TGase, rather than an effect on ECM. Immunofluorescent techniques showed that binding of exogenous TGase to ECM was prevented by prior mixing with fibronectin or collagen, but not with several other ECM components, including laminin, elastin, chondroitin sulfate, heparan sulfate, and hyaluronic acid. ECM-bound TGase was released by 2 M potassium thiocyanate (KSCN) treatment but was not released by treatment with a variety of amino acids, salts, reducing agents, glycerol, or other chaotropic agents.     

Expression of A and B types of monoamine oxidase in neuroblastoma hybrid cells

Monoamine oxidase (MAO) activity towards kynuramine as substrate was measured in 6 hybrid cells derived by fusion of neuroblastoma and glioma, liver or brain cells, and was compared with that of parental or non-parental clones. Activities varied from the lowest level of less than 0.15 pmol/min/mg protein in a neuroblastoma clone NB2A to the highest level of 127 pmol/min/mg protein in NCB20 mouse neuroblastoma × Chinese hamster embryo brain hybrid cells. The relative proportions of A and B types of MAO activities were determined in homogenates of each cell line by inhibition curves with clorgyline and deprenyl. Although the A type activity was found in all cell lines measured, MAO A was predominant in 9 clones, except for NCB20 hybrid cells, N4G-B-a neuroblastoma × glioma hybrid cells, and G8-1 myoblast. The ratio of type A/type B activity in NCB20, N4G-B-a and G8-1 cells was 20/80, 75/25 and 95/5, respectively. The results suggest that NCB20 cells are highly enriched in MAO type B, and that the NCB20 cell is an excellent model for studying the type B activity found in the brain in vivo.     

Bovine aortic endothelial cell transglutaminase. Enzyme characterization and regulation of activity.

Bovine aortic endothelial cells contain Ca2+-dependent tissue-type transglutaminase. Its activity in these cells was high, with apparent Km and Vmax. values with respect to putrescine of 0.203 mM and 18.5 nmol/min per mg of protein, and its activity was inhibited by the three competitive inhibitors dansylcadaverine, spermine and methylamine. The molecular mass of endothelial cell transglutaminase estimated by gel filtration chromatography was 88 kDa and it was immunoprecipitated by rabbit monospecific antiserum raised against rat liver transglutaminase. Its enzymic activity rose when the cell cultures reached confluence, and was further increased when their proliferation was arrested (synchronized at G0/G1 phase). Most of the enzymic activity was found in the 15,000 g soluble fraction, with only 4-22% of the activity found in the particulate fraction, depending on the state of cell proliferation. Examination of these cellular fractions by SDS/polyacrylamide-gel electrophoresis and immunoblotting revealed that at confluence endothelial cells have accumulated transglutaminase antigen in their 15,000 g particulate fraction. A series of experiments demonstrated the existence of a latent transglutaminase form in non-proliferating cells, and suggested that this might involve the formation of an inhibitory complex. Treatment of cell lysates and the 15,000 g particulate fraction with high salt concentration showed a significant increase in transglutaminase activity. Mixing experiments using the 100,000 g particulate fraction or purified rat liver transglutaminase on one hand and the cytosolic fraction on the other showed dose-dependent inhibition of the transglutaminase activity of the latter. It is concluded that endothelial cells contain a particulate fraction-residing inhibitor of transglutaminase which interacts via ionic interaction with the enzyme.