Changes in fibrinogen and fibrin induced by a peptide analog of fibrinogen gamma365-380.

BACKGROUND: The effects of synthetic peptides with sequences derived from the gamma-chain of fibrinogen on the functional properties of fibrinogen and fibrin were investigated. METHODS: Methods included thrombelastography, clot turbidity measurement, clot elasticity measurement, platelet aggregation, and scanning transmission electron microscopy (STEM). RESULTS: Peptide gamma369-380 (NH(2)-WATWKTRWYSMK-COOH) showed the greatest impact on fibrin structure, compared with the 76 other overlapping dodecapeptides. Addition of this peptide, or peptide gamma365-380 (NH(2)-NGIIWATKTREWYSMK-COOH) to a mixture of fibrinogen and thrombin resulted a shorter clotting time, higher clot turbidity, lower clot elastic modulus, a higher degree of D-trimer and D-tetramer formation, and impaired plasmin proteolysis of the clot. In STEM, fibrin formed in the presence of peptide gamma369-380 consisted of a more extensive array of linear fibrils typically consisting of 20 or more molecules. Fibrils were better organized than those from non-peptide containing mixtures. CONCLUSIONS: Replacement of the tryptophan residue gamma376 massively reduced the effect of the peptide on fibrin structure. Binding of the peptide to fibrinogen induces conformational changes, which result in accelerated clotting and increased lateral association of fibrin protofibrils. The results imply a relevant functional role of sites interacting with peptide gamma369-380 region in the fibrinogen molecule.

Fibrinogen and fibrin structure and functions.

Fibrinogen molecules are comprised of two sets of disulfide-bridged Aalpha-, Bbeta-, and gamma-chains. Each molecule contains two outer D domains connected to a central E domain by a coiled-coil segment. Fibrin is formed after thrombin cleavage of fibrinopeptide A (FPA) from fibrinogen Aalpha-chains, thus initiating fibrin polymerization. Double-stranded fibrils form through end-to-middle domain (D:E) associations, and concomitant lateral fibril associations and branching create a clot network. Fibrin assembly facilitates intermolecular antiparallel C-terminal alignment of gamma-chain pairs, which are then covalently 'cross-linked' by factor XIII ('plasma protransglutaminase') or XIIIa to form 'gamma-dimers'. In addition to its primary role of providing scaffolding for the intravascular thrombus and also accounting for important clot viscoelastic properties, fibrin(ogen) participates in other biologic functions involving unique binding sites, some of which become exposed as a consequence of fibrin formation. This review provides details about fibrinogen and fibrin structure, and correlates this information with biological functions that include: (i) suppression of plasma factor XIII-mediated cross-linking activity in blood by binding the factor XIII A2B2 complex. (ii) Non-substrate thrombin binding to fibrin, termed antithrombin I (AT-I), which down-regulates thrombin generation in clotting blood. (iii) Tissue-type plasminogen activator (tPA)-stimulated plasminogen activation by fibrin that results from formation of a ternary tPA-plasminogen-fibrin complex. Binding of inhibitors such as alpha2-antiplasmin, plasminogen activator inhibitor-2, lipoprotein(a), or histidine-rich glycoprotein, impairs plasminogen activation. (iv) Enhanced interactions with the extracellular matrix by binding of fibronectin to fibrin(ogen). (v) Molecular and cellular interactions of fibrin beta15-42. This sequence binds to heparin and mediates platelet and endothelial cell spreading, fibroblast proliferation, and capillary tube formation. Interactions between beta15-42 and vascular endothelial (VE)-cadherin, an endothelial cell receptor, also promote capillary tube formation and angiogenesis. These activities are enhanced by binding of growth factors like fibroblast growth factor-2 (FGF-2) and vascular endothelial growth factor (VEGF), and cytokines like interleukin (IL)-1. (vi) Fibrinogen binding to the platelet alpha(IIb)beta3 receptor, which is important for incorporating platelets into a developing thrombus. (vii) Leukocyte binding to fibrin(ogen) via integrin alpha(M)beta2 (Mac-1), which is a high affinity receptor on stimulated monocytes and neutrophils.

The structure and biological features of fibrinogen and fibrin.

Fibrinogen and fibrin play important, overlapping roles in blood clotting, fibrinolysis, cellular and matrix interactions, inflammation, wound healing, and neoplasia. These events are regulated to a large extent by fibrin formation itself and by complementary interactions between specific binding sites on fibrin(ogen) and extrinsic molecules including proenzymes, clotting factors, enzyme inhibitors, and cell receptors. Fibrinogen is comprised of two sets of three polypeptide chains termed A alpha, B beta, and gamma, that are joined by disulfide bridging within the N-terminal E domain. The molecules are elongated 45-nm structures consisting of two outer D domains, each connected to a central E domain by a coiled-coil segment. These domains contain constitutive binding sites that participate in fibrinogen conversion to fibrin, fibrin assembly, crosslinking, and platelet interactions (e.g., thrombin substrate, Da, Db, gamma XL, D:D, alpha C, gamma A chain platelet receptor) as well as sites that are available after fibrinopeptide cleavage (e.g., E domain low affinity non-substrate thrombin binding site); or that become exposed as a consequence of the polymerization process (e.g., tPA-dependent plasminogen activation). A constitutive plasma factor XIII binding site and a high affinity non-substrate thrombin binding site are located on variant gamma' chains that comprise a minor proportion of the gamma chain population. Initiation of fibrin assembly by thrombin-mediated cleavage of fibrinopeptide A from A alpha chains exposes two EA polymerization sites, and subsequent fibrinopeptide B cleavage exposes two EB polymerization sites that can also interact with platelets, fibroblasts, and endothelial cells. Fibrin generation leads to end-to-middle intermolecular Da to EA associations, resulting in linear double-stranded fibrils and equilaterally branched trimolecular fibril junctions. Side-to-side fibril convergence results in bilateral network branches and multistranded thick fiber cables. Concomitantly, factor XIII or thrombin-activated factor XIIIa introduce intermolecular covalent epsilon-(gamma glutamyl)lysine bonds into these polymers, first creating gamma dimers between properly aligned C-terminal gamma XL sites, which are positioned transversely between the two strands of each fibrin fibril. Later, crosslinks form mainly between complementary sites on alpha chains (forming alpha-polymers), and even more slowly among gamma dimers to create higher order crosslinked gamma trimers and tetramers, to complete the mature network structure.

Fibrinogen and fibrin are anti-adhesive for keratinocytes: a mechanism for fibrin eschar slough during wound repair.

During cutaneous wound repair the epidermis avoids the fibrin-rich clot; rather it migrates down the collagen-rich dermal wound margin and over fibronectin-rich granulation tissue. The mechanism(s) underlying keratinocyte movement in this precise pathway has not been previously addressed. Here we demonstrate that cultured human keratinocytes do not express functional fibrinogen/fibrin receptors, specifically alpha v beta 3. Biologic modifiers known to induce integrin expression or activation did not induce adhesion to fibrin, fibrinogen, or its fragments. Epidermal explant outgrowth and single epidermal cell migration failed to occur on either fibrin or fibrinogen. Surprisingly, fibrin and fibrinogen mixed at physiologic molar ratios with fibronectin abrogated keratinocyte attachment to fibronectin. Keratinocytes transduced with the beta 3 integrin subunit cDNA, expressed alpha v beta 3 on their surface and attached to and spread on fibrinogen and fibrin. beta-gal cDNA-transduced keratinocytes did not demonstrate this activity. Furthermore, beta 3 cDNA-transduced keratinocyte adhesion to fibrin was inhibited by LM609 monoclonal antibody to alpha v beta 3 in a concentration-dependent fashion. From these data, we conclude that normal human keratinocytes cannot interact with fibrinogen and its derivatives due to the lack of alpha v beta 3. Thus, fibrinogen and fibrin are authentic anti-adhesive for keratinocytes. This may be a fundamental reason why the migrating epidermis dissects the fibrin eschar from wounds.

Paris I dysfibrinogenemia: a point mutation in intron 8 results in insertion of a 15 amino acid sequence in the fibrinogen gamma-chain.

Paris I dysfibrinogenemia results in the production of a fibrinogen molecule containing a functionally abnormal gamma-chain. We determined the basis of the molecular defect using polymerase chain reaction (PCR) to amplify the gamma-chain region of the Paris I subject's genomic DNA. Comparative sequence analysis of cloned PCR segments of normal and Paris I genomic DNA revealed only an A-->G point mutation occurring at nucleotide position 6588 within intron 8 of the Paris I gamma-chain gene. We examined six normal individuals and found only normal sequence in this region, indicating that this change is not likely to represent a normal polymorphism. This nucleotide change leads to a 45 bp fragment being inserted between exons 8 and 9 in the mature gamma Paris I chain mRNA, and encodes a 15 amino acid insert after gamma 350 [M-C-G-E-A-L-P-M-L-K-D-P-C-Y]. Alternative splicing of this region from intron 8 into the mature Paris I gamma-chain mRNA also results after translation into a substitution of S for G at position gamma 351. Biochemical studies of 14C-iodoacetamide incorporation into disulfide-reduced Paris I and normal fibrinogen corroborated the molecular biologic predictions that two additional cysteine residues exist within the gamma Paris I chain. We conclude that the insertion of this amino acid sequence leads to a conformationally-altered, and dysfunctional gamma-chain in Paris I fibrinogen.

The polymerization of fibrinogen Dusart (A alpha 554 Arg-->Cys) after removal of carboxy terminal regions of the A alpha-chains.

The six polypeptide chains of normal fibrinogen are covalently linked by interchain disulphide bonds, and there are no free sulphydryl groups. Fibrinogen Dusart is a congenital fibrinogen variant in which A alpha 554 Arg is replaced by Cys; albumin is disulphide linked to these fibrinogen molecules, possibly at A alpha 554 Cys. Functionally, Dusart fibrinogen displays markedly abnormal fibrin polymerization, characterized by delayed lateral fibril association and matrix fibre bundles that are thinner than normal fibrin bundles. These observations are consistent with experiments suggesting that the carboxy terminal region of the A alpha-chain contains a polymerization domain that participates in lateral fibril associations. In order to investigate the location and the effect of albumin binding to Dusart fibrinogen, we examined the fibrinogen by electron microscopy, and compared the polymerization and ultrastructure of fibrin prepared from normal fibrinogen containing intact A alpha-chains (fraction I-2) or plasmin degraded fibrinogen molecules lacking carboxy terminal regions of A alpha-chains (fraction I-9D), with fibrin prepared from Dusart fraction I-2 and I-9D. Most bound albumin was released from Dusart fibrinogen by plasmin degradation involving the A alpha-chains. Nevertheless, we were able to visualize albumin molecules remaining covalently bound to Dusart I-9D as well as to Dusart I-2 fibrinogen, as distinct globular domains situated near the fibrinogen D domain. The presence of albumin in these fractions was confirmed by Western blotting using anti-albumin. Dusart fibrin polymerized much more slowly than normal I-2, as previously reported, whereas polymerization of Dusart I-9D fibrin was faster than Dusart I-2 and nearly the same as normal I-9D fibrin.(ABSTRACT TRUNCATED AT 250 WORDS)

Characterization of the gamma chain platelet binding site on fibrinogen fragment D.

Glycoprotein (GP) IIb/IIIa on adenosine diphosphate (ADP)-activated human platelets interacts with specific sites on the fibrinogen molecule leading to aggregation. We characterized the platelet-binding site on the gamma chains of fibrinogen using plasmic fragments D gamma A and D gamma'. Fragment D gamma A, which contains the carboxy terminal gamma A400-411 platelet-binding sequence (HHLGGAKQAGDV), was 70-fold more active than the synthetic gamma A400-411 peptide in inhibiting ADP-induced platelet aggregation. Fragment D gamma A inhibited fibrinogen binding and also bound directly to ADP-activated platelets. The Kd values determined for fibrinogen and fragment D gamma A binding were 0.55 mumol/L and 1.2 mumol/L, respectively. In contrast, fragment D gamma', which differs from fragment D gamma A with respect to its gamma chain sequence from position 408 to the COOH-terminus at position 427, did not inhibit platelet aggregation or fibrinogen binding, and did not bind directly to the platelet surface. Denaturation of fragment D gamma A with guanidine-HCl caused a loss of inhibitory activity in platelet aggregation assays. These data indicate that the native conformation of the gamma chain platelet-binding site on fibrinogen is important for optimal binding to GPIIb/IIIa.

Hemostatic evaluation of Sarns/3M-VAD implantation in calves.

The authors evaluated the potential for thrombotic complications arising from implantation of a ventricular assist device (Sarns/3M-VAD) in four calves. Coagulation screening tests (prothrombin time [PT], partial thromboplastin time [PTT], thrombin time [TT]), fibrinogen levels, and antithrombin III functional activity were found to be of little value as predictors of the degree of activation of the hemostatic system. However, platelet counts, adenosine diphosphate (ADP)- and collagen-induced platelet aggregation, and thromboxane (TXB2) levels were good indicators of changes in platelet reactivity. Platelet counts (initial value 6 x 10(5) rose, and were associated with increased rate and extent of ADP- and collagen-induced platelet aggregation, which remained elevated during the entire 25 day postimplantation period. The first 5 days postimplantation revealed a typical acute inflammatory response, with increased platelet levels, but with TXB2 levels significantly decreased during this period. A monoclonal antibody based bovine D-dimer assay and Western blot studies indicated a small but significant increase in circulating bovine D-dimer, indicating localized fibrin formation and its dissolution.

The role of fibrinogen A alpha chains in ADP-induced platelet aggregation in the presence of fibrinogen molecules containing gamma' chains.

Human plasma fibrinogen (Fgn) is heterogenous with respect to the size of its gamma chains, which differ in that residues 408 to 411 of gammaA chains (93% of total) are replaced in gamma' chains by a unique 20 amino acid sequence (gamma408 to gamma427). In this study, we compared the contribution to adenosine diphosphate (ADP)-induced platelet aggregation of the A alpha chains in Fgn molecules containing predominantly (fraction 1-2) or exclusively (peak 1 Fgn) gammaA chains with that of molecules containing approximately 50% gamma' chains (peak 2 Fgn). Using washed human platelets, we confirmed that the number of peak 2 Fgn molecules binding to platelets in the presence of ADP was about half the number of peak 1 Fgn molecules (18,962 +/- 2,298 v 44,366 +/- 16,096 molecules per platelet), and that isolated S-carboxymethylated (SCM) gammaA chains supported ADP-induced platelet aggregation nearly as well as peak 1 Fgn. In contrast, SCM-gamma' chains alone supported aggregation poorly, whereas a mixture of SCM-gammaA and gamma' chains (1:1 ratio) gave intermediate results. Despite the findings with isolated SCM-gamma' chains, we found that peak 2 Fgn supported platelet aggregation nearly as well as peak 1 Fgn. However, peak 2 Fgn from which carboxy (COOH)-terminal A alpha chain segments had been removed by digestion with plasmin showed a markedly decreased platelet aggregation potential. Peak 1 Fgn core fraction from an 88% to 90% coagulable plasmin digest, or Fgn fraction 1-9, which has a high gammaA/gamma' chain ratio (93:7), but lacks COOH-terminal regions of A alpha chains, supported platelet aggregation to the same extent as did intact peak 2 Fgn. These findings indicate that Fgn molecules containing gamma' chains can approach the aggregation potential of Fgn molecules containing predominantly or exclusively gammaA chains only if intact A alpha chains are also present.

Essential thrombocythemia in pregnancy.

A 32-year-old women with essential thrombocythemia and a history of bleeding completed a full-term pregnancy. During pregnancy, the platelet count declined markedly to near-normal levels, but returned to prepregnancy levels within two weeks after delivery. The clinical course was uneventful, except for two brief episodes of vaginal bleeding during the second trimester. Possible etiologic mechanisms for pregnancy-related reduction of platelet levels in thrombocythemia include increased platelet consumption or down-regulation of platelet production by a placental or fetal factor.

ADP-induced platelet aggregation depends on the conformation or availability of the terminal gamma chain sequence of fibrinogen. Study of the reactivity of fibrinogen Paris 1.

Fibrinogen Paris I contains a mutant gamma chain that is longer than the normal chain, resulting in altered fibrin polymerization and cross-linking. Because these functions involve the carboxy-terminal region of the gamma chain, we decided to determine whether fibrinogen Paris I or the isolated Paris I gamma chain supports normal ADP-induced platelet aggregation, a function that requires the ultimate 12 residues of the normal gamma chain (400 through 411). Aggregation of ADP-stimulated normal platelets was defective with fibrinogen Paris I and markedly depressed with the gamma Paris I chain. These findings prompted us to characterize the carboxy-terminal structure of the region of the gamma Paris I chain responsible for this activity. The carboxy-terminal cyanogen bromide (CNBr) peptide of the normal gamma chain (385 through 411) or that from gamma Paris I was isolated by differential adsorption to triethylene-tetramine resin or by reverse-phase high-performance liquid chromatography (HPLC). The CNBr peptide from the Paris I gamma chain was identical to that of the normal gamma chain in its retention time on HPLC, its amino acid composition, and its sequence. Thus, the primary structure of the gamma Paris I chain from residue 384 through 411 is normal, indicating that a peptide insertion has occurred upstream from residue 384, resulting in an impairment of those physiologic functions attributable to the carboxy-terminal end of the gamma chain from position 384 (ie, cross-linking, ADP-induced platelet aggregation, and at least a portion of the gamma chain polymerization site). These observations demonstrate that the gamma chain platelet recognition site in the fibrinogen molecule is necessary but not alone sufficient to support normal ADP-induced platelet aggregation. There appears to be an additional requirement for normal conformation of the gamma chain or availability of its terminal sequence during the interaction of fibrinogen with platelets.

Effect of hepatocyte-stimulating factor and glucocorticoids on plasma fibronectin levels.

We evaluated the effects of hepatocyte-stimulating factor (HSF) and a glucocorticoid (dexamethasone) on changes in the levels, in vivo and in vitro, of plasma fibronectin (Fn), a glycoprotein that is synthesized and secreted by hepatocytes. In turpentine-treated chickens, plasma levels of Fn, which peaked at 48 h (whereas fibrinogen levels were maximum at 72 h) rose 200-250% over basal levels, whereas albumin levels decreased by 20-40%. Corticosterone levels in serum samples taken between 5 and 48 h after injection revealed a 124% increase in hormone levels at 24 h in turpentine-treated chickens. We also showed that circulating HSF levels were maximal 8 to 12 h after injection and that HSF activity, as assessed by molecular-exclusion chromatography, was eluted in the 30-45 kDa range. Addition of either serum-derived HSF or dexamethasone (2 nM) to chick hepatocyte cultures resulted in a 130-150% increase in secreted Fn as well as in fibrinogen. When HSF and dexamethasone were added together, a 360-489% increase in the secreted levels of both proteins was found. Chicken mononuclear phagocytic cells treated with lipopolysaccharide secreted an HSF activity that was eluted in two peaks, a minor peak at approximately 70 kDa and a major peak in the 25-40 kDa range. Addition of mononuclear-cell-derived HSF resulted in a greater increase in Fn levels than did the addition of serum HSF. These findings indicate that Fn, like fibrinogen, is an acute-phase protein, the production of which, at least in chickens, is stimulated by HSF and glucocorticoids in an additive manner.

Human platelet fibrinogen: purification and hemostatic properties.

Conditions were developed in which 80% to 90% of platelet fibrinogen could be routinely purified in nondegraded form from the fluid phase of platelet suspensions stimulated with the calcium ionophore, A23187, in the presence of calcium, leupeptin, and prostaglandin E1. Fibrinogen was separated from other released proteins by chromatography on diethylaminoethanol (DEAE)-cellulose using a continuous pH and ionic strength gradient. Purified platelet fibrinogen, greater than 98% homogeneous by immunoelectrophoresis and sodium-dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), consisted of intact A alpha, B beta and gamma A chains, but not gamma' chains, and was 95% to 96% clottable. Platelet fibrinogen was shown to compete for the binding of radiolabeled plasma fibrinogen to ADP-activated platelets in a manner identical to that of unlabeled plasma fibrinogen itself. Also, at equivalent protein concentrations, platelet and plasma fibrinogens supported platelet aggregation to an equivalent extent. Based upon these results, we conclude that there is no significant difference between platelet and plasma fibrinogen with respect to their size, their clottability, their affinity for the activated platelet fibrinogen receptor, or their capacity to support subsequent platelet aggregation.

The role of fibronectin in monocyte/macrophage function.

"Fibronectin" is a term describing a class of immuno-chemically related glycoproteins that are found in blood, in connective tissues, and in most basement membranes. All types of fibronectins have distinct binding domains that serve to mediate an adhesive function between collagen, cells (e.g., fibroblasts, macrophages), fibrin(ogen), and/or certain glycosaminoglycans (e.g., heparin). Monocytes and macrophages are phagocytic cells which collectively comprise the "mononuclear phagocytic" system. Plasma fibronectin (CIg) mediates: a) the attachment of monocytes to gelatin-coated surfaces; and b) the attachment of gelatin-coated RBCs or latex particles to surface-bound monocytes. This process was mediated only by surface-bound forms of fibronectin. Particle attachment to cells, per se, was not associated with augmented particle ingestion, although particle binding did result in increased expression of the monocyte C3 and Fc receptors. These data indicate that monocytes have surface receptors for fibronectin. It appears that the relatively strong affinity between fibrin and fibronectin may provide a mechanism for recognition and subsequent clearance of fibrin-fibrinogen complexes from the blood by attachment to fibronectin receptors on mononuclear phagocytes. One can also speculate that fibronectin binds to exposed or denatured collagen at sites of injury, leading to macrophage attachment and differentiation. Such events probably play an important role in wound organization and healing.

Separation and analysis of the major forms of plasma fibronectin.

Human plasma fibronectin exists in circulation in multiple molecular forms that are distinguishable by SDS-polyacrylamide gel electrophoresis (zone I, approx. 450 kDa dimers; zone II, 190-235 kDa; Zone III, 146-175 kDa). (Chen, A.B., Amrani, D.L. and Mosesson, M.W. (1977) Biochim. Biophys. Acta 493, 310-322). We report here on investigations of plasma fibronectin that had been purified from the 'heparin-precipitable fraction' of plasma by DEAE-cellulose chromatography using buffers containing a chaotropic salt (KSCN). Zone I fibronectin and zone II fibronectin were subsequently separated by Sepharose CL-6B chromatography in the presence of 0.3 M KSCN. Electrophoresis of reduced zone I fibronectin dimers showed the presence of three types of subunits (i.e., 220 kDa, 215 kDa, 207 kDa), evidently all having the same NH2-terminal sequence. Subunits of this size were also found in reduced zone II fibronectin, as well as another polypeptide of 190 kDa, the latter amounting to under 5% of the total. Unreduced zone I fibronectin was resolved by gel electrophoresis into a doublet. The upper component amounted to approx. 90% of the total and was comprised of 220 kDa and/or 215 kDa subunits; the lower component contained 207 kDa plus a 220 kDa or 215 kDa subunit. Scanning transmission electron microscopy indicated that under physiologic conditions zone II fibronectin molecules, like those in zone I, exist as pleiomorphic, loosely folded structures (approx. 16 X 8-12 nm) that are somewhat smaller than dimeric zone I molecules (approx. 24 X 16 nm). Circular dichroic spectral analyses suggests that both types have similarly folded local domains. Affinity chromatography experiments revealed a relative decrease in the binding of zone II fibronectin to gelatin but no difference from zone I fibronectin with respect to heparin or fibrin binding.

Hormonal regulation of fibrinogen synthesis in cultured hepatocytes.

Most of what was originally known of the effects of hormones on fibrinogen synthesis was based, as noted above, on experiments involving surgical removal of endocrine glands. Some caution should be exercised when using such in vivo experiments to derive the hormonal requirements of fibrinogen synthesis, however, since multiple hormonal alterations often occur in these animals. The development of a variety of ex vivo systems has allowed investigators to more carefully control the hepatocellular environment. The work of several laboratories, including our own, has now made it clear that hormones and other agents directly stimulate hepatocellular synthesis of fibrinogen. From the studies summarized here, using chick embryo hepatocytes as a model, several generalizations emerge: Fibrinogen synthesis may be considered to be a "constitutive" liver function, since hepatocytes cultured without serum, hormones or other macromolecular supplements synthesize this protein at a basal rate for several days. Addition of certain hormones (e.g. T3, dexamethasone, insulin), individually and in physiological concentrations, elicits an increase in fibrinogen production, varying with each agent in onset, dose, minimum exposure required and accompanying effects on the synthesis of other plasma proteins. Glucocorticoids and thyroid hormones are similar in the selectivity of their stimulation (neither affects albumin or transferrin synthesis) but differ in that thyroid hormones need to be present for just a short "triggering" period. The stimulation of fibrinogen synthesis by insulin occurs only following prolonged exposure to concentrations 10-times higher than the very low doses to which albumin synthesis responds rapidly.

Production of fibronectin by human epithelial cells in culture.

Human epithelial cell lines derived from both carcinomatous and nomalignant tissues were characterized with respect to the presence and distribution of fibronectin by immunofluorescence microscopy. In cell lines derived from nonmalignant tissues or from primary carcinomas, fibronectin was found predominantly in an extracellular matrix. In contrast, cell lines derived from metastatic carcinomas displayed very little or no fibronectin. Metabolic labeling studies indicated that a positive line synthesized fibronectin de novo rather than absorbing the protein from the media. Negative lines neither synthesized fibronectin nor secreted it into the culture fluid, suggesting that they were not producing fibronectin. Evidence is presented that cells in culture change their properties after extensive subculture since a small amount of fibronectin in an extracellular matrix was observed after extensive subculture of two metastatic lines that were originally negative.