THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 263, No. 21, Issue of July 25, pp. 10464-10469, 1988 Printed in U.S.A.

Factor XI11 Cross-linking of Fibronectin at Cellular Matrix Assembly Sites*

(Received for publication, March 10, 1988)

Elizabeth L. R. Barry and Deane F. Mosher From the Departments of Physiological Chemistry and Medicine, University of Wisconsin Medical School, Madison, Wisconsin 53706

We describe the effect of activated Factor XI11 (Fac- tor XIIIa, plasma transglutaminase) on the incorpora- tion 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 mul- timers of the extracellular matrix. When Factor XIIIa was included in the binding medium of skin fibroblasts, accumulation of ‘251-fibronectin in the deoxycholate- insoluble matrix was increased. Fibronectin accumu- lating in the cell layer was cross-linked into nonreduc- ible high molecular weight aggregates. The 70-kDa amino-terminal fragment of fibronectin inhibited the binding and cross-linking of ‘“I-fibronectin to cell layers, whereas fibrinogen had little effect. When ‘“I- 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 de- oxycholate-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 extra- cellular matrix.

Fibronectin is a high molecular weight glycoprotein found in plasma at approximately 300 pg/ml, on cell surfaces, and in the insoluble extracellular matrix (1-3). Protameric fibro- nectin is a disulfide-bonded dimer of 240-250-kilodalton sub- units. The subunits contain small regions of structural het- erogeneity due to alternative splicing of a single gene. Each subunit is composed of a series of structural domains that mediate interactions with other molecules including fibrin, collagen, heparin, DNA, actin, and cell surface receptors. The capacity of fibronectin to interact with many different mole- cules suggests its involvement in a large group of processes from cell adhesion and migration to hemostasis and wound healing.

Plasma fibronectin probably serves as a reservoir for tissue fibronectin since it is incorporated into the cell layer of

* These studies were supported by National Institutes of Health Grant HL21644 and by a grant from the Lucille P. Markey Charitable Trust, Miami, FL. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

cultured fibroblasts where it forms disulfide-bonded multi- mers by disulfide exchange and accumulates in extracellular fibrils (4-6). Fibronectin fibril assembly does not occur when fibronectin is incubated with isolated matrices (4) and is rapidly modulated by changes in the intracellular CAMP levels or treatment of cells with transforming growth factor B (7,8). These facts suggest that assembly of fibronectin fibrils is a cell-mediated process. Binding studies done on monolayers of human skin fibroblasts indicate the existence of specific sat- urable cell surface binding sites that recognize the amino- terminal region of fibronectin with a Kd of 3.6 X lo-* M (4, 6, 9). A human fibrosarcoma cell line, HT-1080, is unable to bind exogenous fibronectin or assemble fibronectin fibrils (4) but acquires the ability to do so following treatment with dexamethasone (10).

Factor XI11 is the zymogen form of plasma transglutami- nase and circulates at a concentration of 10-20 pg/ml. Upon cleavage by thrombin to activated Factor XI11 (Factor XIIIa), Factor XIIIa catalyzes the formation of t-(yglutamy1)lysyl cross-links between specific protein pairs (11, 12). The zy- mogen consists of two A subunits and two B subunits (A2B2) that are noncovalently complexed. Activation of Factor XI11 involves cleavage of the catalytic A subunits by thrombin, a calcium ion-dependent conformational change of the A sub- units, and dissociation of the A subunits from the B subunits. Factor XIIIa catalyzes the cross-linking of fibrin molecules in the final step of blood coagulation, thereby increasing the mechanical stability of the clot and its resistance to plasmin degradation. Deficiency in Factor XI11 is associated with a bleeding tendency, abnormal wound healing, and spontaneous abortions.

Some of the physiological effects of Factor XI11 may be mediated through its actions on fibronectin (13). Intact fibro- nectin contains 2 glutamine residues/subunit that are suscep- tible to Factor XIII. These are located at opposite ends of the molecule (14, 15). Fibronectin can be cross-linked by Factor XIIIa to fibrin, to collagen, and to itself if self-aggregation is induced (13, 16-19). Fibronectin in the extracellular matrix of cultured fibroblasts can be cross-linked into a large com- plex, but the protein pairs involved have not been determined (20, 21). The physiological importance of fibronectin-Factor XI11 interactions has been shown by studies demonstrating the fibronectin covalently incorporated into fibrin clots changes the mechanical properties of the clot and allows cell adherence (22, 23). Similarly, cross-linking of fibronectin to monomeric collagen enhances cell adherence to the collagen (24).

In this paper we provide further information about the interaction between fibronectin and Factor XI11 by exami- nation of the effect of Factor XIIIa on the incorporation of iodinated plasma fibronectin into the extracellular matrix of cultured cells.

10464

Cross-linking of Assembling Fibronectin by Factor XIII 10465

EXPERIMENTAL PROCEDURES

Cell Cultures-Human foreskin fibroblasts (TJ-6F) were derived by Dr. Lynn Allen-Hoffmann (University of Wisconsin). The cells were cultured in a 1:l mixture of Ham's F-12 and Dulbecco's modified Eagle's medium (GIBCO) supplemented with 10% fetal bovine serum (Hazelton Research Products). Experiments were performed on cells up to 15 passages in culture. Human fibrosarcoma cells, HT-1080, were a gift from Dr. Noellyn Oliver (Salk Institute, La Jolla, CA). , HT-1080 cells were grown in modified Eagle's medium (GIBCO) supplemented with 10% fetal bovine serum.

Dexamethasone (Sigma) was dissolved at 5 mg/ml in ethanol and diluted into modified Eagle's medium containing 5% fetal bovine serum for a final concentration of 1 X M for use in HT-1080 cell culture. Appropriate cultures were incubated with dexamethasone for 48 h prior to a binding assay as well as during the assay. Cytochalasin B (Sigma) was dissolved at 1 mg/ml in dimethyl sulfoxide and diluted into Ham's F-l2/Dulbecco's modified Eagle's medium to a final concentration of 5 pg/ml for use in TJ-6F cell cultures. Appropriate cultures were incubated with cytochalasin B for 4 h prior to a binding assay as well as during the assay.

Purification of Proteins-Human plasma fibronectin was purified from a fibronectin- and fibrinogen-rich by-product of Factor VI11 production (18). The 70-kDa amino-terminal fragment of fibronectin was purified from cathepsin D digests (9). Fibrinogen was purified from human plasma by ammonium sulfate precipitation and chro- matography on DEAE-cellulose (16). Factor XI11 was purified from human plasma as follows. Barium sulfate was added to plasma at 100 g/liter to absorb vitamin K-dependent coagulation factors. Ammo- nium sulfate was added to the barium sulfate supernatant to make a 25% saturated solution. The pellet was resuspensed into 0.3 M NaCl, 0.01 M Tris (pH 7.4), 1 mM EDTA. Two additional precipitations in 25% saturated ammonium sulfate were performed before the resus- pended protein was dialyzed against 0.1 M NaCl, 0.01 M Tris, pH 7.4, 1 mM EDTA. Fibrinogen was precipitated by heating at 56 "C for 3 min. The supernatant was diluted 1:2 with 0.01 M Tris, pH 7.4, and applied to a DEAE-Sephacel (Pharmacia LKB Biotechnology Inc.) column. The bound protein was eluted with a linear gradient from 0.05 to 0.3 M NaCl. Fractions containing Factor XIII, as determined by SDS-PAGE,' were concentrated by dialysis against 50% polyeth- ylene glycol and dialyzed against TBS (10 mM Tris, pH 7.4, 150 mM NaC1). A final chromatography on a DEAE-5PW column (Waters Associates) was performed to remove minor contaminants.

Activation of Factor Xiii-Factor XIII, 100 pg, was activated with 1 unit (0.5 pg) of thrombin (a gift from Dr. John Fenton, 11, Albany, NY) for 30 min at room temperature just prior to its use in an experiment. The low final concentration of thrombin (50 ng/ml) present in an experiment had no effect on the binding of fibronectin to cell layers as previously shown (25) and as determined by controls (not shown) in the present experiments. In some experiments a thrombin inhibitor, hirudin (Sigma), was used.

Binding of 125i-Fibronectin to Cultured Cells-Fibronectin was io- dinated using chloramine T and purified from free iodine by gelatin- Sepharose chromatography (4). All binding assays were performed in 24-well culture dishes (Falcon) on confluent monolayers of TJ-6F skin fibroblasts and HT-1080 fibrosarcoma cells essentially as de- scribed previously (4). The binding medium was Ham's F-l2/Dulbec- co's modified Eagle's medium containing 0.2% bovine albumin for the TJ-6F fibroblasts and modified Eagle's medium containing 5% fetal bovine serum for the HT-1080 fibrosarcomas. 1251-Fibronectin was included in the binding mix at 800,000-1,500,000 cpm/ml. In some wells, the binding mix contained Factor XIIIa or other additions. After incubation with 0.5 ml of labeled medium at 37 "C for various times, the medium was removed, and the cell layers were rinsed four times in cold TBS and solubilized in 1.3 ml of 1% deoxycholate in 0.02 M Tris, pH 8.3, containing 2 mM phenylmethanesulfonyl fluoride, 2 mM EDTA, 2 mM N-ethylmaleimide, and 2 mM iodoacetic acid. The cell layer was removed with vigorous pipetting and centrifuged in a Microfuge for 15 min. The supernatant containing deoxycholate- soluble material (called pool I) and the pellet containing insoluble material (called pool 11) were analyzed for radioactivity. Nonspecific binding of '251-fibronectin was defined as the binding in the presence of excess unlabeled fibronectin (500 pg/ml). Specific binding was usually 75-80% of total binding (e.g. see Table I).

Isolated fibroblast matrices were obtained by incubating confluent

The abbreviations used are: SDS-PAGE, sodium dodecyl sulfate- polyacrylamide gel electrophoresis; TBS, Tris-buffered saline.

cell layers with 1 ml of 1% deoxycholate in 0.02 M Tris (pH 8.3) buffer containing 2 mM phenylmethanesulfonyl fluoride, 2 mM EDTA, 2 mM N-ethylmaleimide, and 2 mM iodoacetic acid for 10 min at room temperature. The matrices were then rinsed three times with TBS before use in a binding assay. Trypsin-treated fibroblast cell layers were obtained by incubating cell layers with 1 ml of TBS containing 1 pg/ml trypsin (Cooper Biomedical) for 2-8 min at 37 "C. This treatment caused the cells to begin to separate from each other even at the earliest time points examined. The cell layers were then rinsed twice with TBS before use in a binding assay. Freshly plated cell layers were obtained by harvesting cell layers in 0.5 mg/ml trypsin, 0.3 mM EDTA for 5 min at 37 "C. The cells were resuspended in medium containing 0.2% bovine serum albumin and '"I-fibronectin and seeded into 35-mm culture dishes for immediate assaying as described above.

Gel Electrophoresis-SDS-PAGE was performed on slabs of 8% separating and 3.3% stacking gels using a discontinuous buffer system (26). Gels were stained with Coomassie Brilliant Blue to visualize protein, dried, and exposed to XAR-2 film (Eastman Kodak).

Incorporation of Fluoresceinated Fibronectin into Cell Cultures- Fluoresceinated fibronectin was prepared as previously described (4). Cells were seeded onto glass coverslips in 35-mm culture dishes. TJ- 6F fibroblasts and HT-1080 fibrosarcomas were incubated with 20 pg/ml fluoresceinated fibronectin for 4 and 24 h, respectively. In some cases, the incubation mixtures also contained 10 pg/ml Factor XIIIa or lo" M dexamethasone. Coverslips were rinsed with TBS, fixed in 3.5% paraformaldehyde for 30 min, rinsed, mounted on glass slides in glycerol-gelatin (Sigma), and photographed with a Nikon micro- scope equipped with epifluorescence and phase contrast.

Antibodies to Factor XZZZa-Rabbit antihuman Factor XIII-AZ (the catalytic subunits) was purchased from Calbiochem and purified on protein A-Sepharose (Sigma). Antibodies that bound to the column equilibrated in TBS were eluted with 0.1 M glycine, pH 3, neutralized with 1.0 M Tris, pH 11, and dialyzed against TBS. For some binding assays, Factor XI11 was preincubated with the antibodies (0-200 pg/ ml) before activation with thrombin.

RESULTS

To examine the effect of Factor XIIIa on the incorporation of plasma fibronectin into the extracellular matrix, confluent cultures of human skin fibroblasts were incubated with "'1- fibronectin for 1 h in the presence of increasing concentra- tions of Factor XIIIa (Fig. 1). Binding of fibronectin to the deoxycholate-soluble pool (pool I) or deoxycholate-insoluble pool (pool 11) was determined. At low Factor XIIIa concentra- tions, between 0 and 10 pg/ml, binding to pool I1 increased in a dose-dependent manner, whereas pool I binding was un- changed. At concentrations of 10 pg/ml and above, fibronectin binding in pool I1 was more than double the binding in the absence of Factor XIIIa. Pool I and Pool I1 binding respec-

I

X "1

I I 1 I 1 1

FX I11 (pg/rnl) 20 40 60

FIG. 1. Dose response for the effect of Factor XIIIa on the binding of fibronectin (FN) to cell layers. Confluent cultures of human fibroblasts were incubated with medium containing '"I-fibro- nectin (1 pg or 800,000 cpm/ml), and the indicated concentration of Factor XIIIa. After 1 h at 37 "C, labeled media was removed and the monolayers were processed as described under "Experimental Pro- cedures" to determine specific binding in pool I and pool 11. Total binding was calculated as the sum of pool I and pool I1 binding. The points are averages of duplicate wells.

10466 Cross-linking of Assembling Fibronectin by Factor XIII tively decreased and increased slightly at Factor XIIIa con- centrations greater than 10 pg/ml.

Pool I1 extracts were analyzed by SDS-PAGE and autora- diography to determine whether they contained covalently cross-linked '251-fibronectin (Fig. 2). Fibronectin accumulat- ing in the absence of activated Factor XIIIa migrated as approximately equal amounts of high molecular weight disul- fide-bonded aggregates, which remained at the top of the stacking gel, and a closely spaced doublet of fibronectin dimer. On reducing gels, this fibronectin migrated only as monomer, demonstrating the absence of nonreducible covalent cross- links. Fibronectin accumulating in the presence of increasing concentrations of Factor XIIIa contained increasing amounts of high molecular weight aggregates. On reducing gels, some of these aggregates failed to migrate as fibronectin monomer, indicative of the presence of nonreducible covalent cross- links. Some of the fibronectin aggregates formed by Factor XIIIa remained at the top of the stacking gel when analyzed under reducing conditions. Others, especially at lower Factor XIIIa concentrations, migrated as apparent dimers of fibro- nectin after reduction, indicating that they contained nonre- ducible cross-links and had also been part of a disulfide- bonded complex. Even at the highest concentration of Factor XIIIa used (64 pglml), a proportion of pool I1 fibronectin migrated as dimer under nonreducing conditions and mon- omer under reducing conditions. Fibronectin that accumu- lated in the presence of thrombin alone or in the presence of Factor XI11 that had not been activated with thrombin did not contain nonreducible cross-links. Pool I fibronectin and fibronectin in the medium were also analyzed by SDS-PAGE

A Flla: - + + + + + + + - FXIII: 0 0 2 4 8 l6 32 64 16

n FIG. 2. Autoradiograph of "%fibronectin bound in pool 11.

Deoxycholate-insoluble (pool 11) extracts from the experiment shown in Fig. 1 were solubilized in SDS and divided in half for polyacryl- amide gel electrophoresis analysis under nonreducing ( A ) and reduc- ing ( E ) conditions as described under "Experimental Procedures." The concentration of Factor XIIIa (pglml) and the presence or absence of thrombin (Factor XIIa) are indicated above each lane. Arrows show the positions of monomeric (M) and dimeric (D) fibro- nectin.

(not shown) and migrated as dimer under nonreducing and monomer under reducing conditions regardless of whether Factor XIIIa was present or not.

The effect of Factor XIIIa, 10 pg/ml, on '%I-fibronectin binding to human fibroblasts over a 24-h time course is shown in Fig. 3. The binding of fibronectin into Pool I was the same regardless of the presence of Factor XIIIa. The rate of accu- mulation of fibronectin in pool I1 was doubled in the presence of 10 pg/ml Factor XIIIa. The increased accumulation of fibronectin in pool I1 was not seen if the Factor XI11 was preincubated with antibodies to the catalytic subunit of Factor XI11 (not shown).

Pool I1 extracts from the time course experiment were analyzed by SDS-PAGE and autoradiography (Fig. 4). In the presence of Factor XIIIa, larger amounts of high molecular weight aggregates, which could not be reduced to fibronectin monomers, were formed at each time point. Interestingly, even in the absence of Factor XIIIa, nonreducible dimer-like fibronectin aggregates were evident at the longer time points. Fibronectin in pool I and from the medium migrated as dimer under nonreducing conditions and monomer under reducing conditions (not shown).

A number of experiments was done to examine whether fibronectin was cross-linked by Factor XIIIa directly to the extracellular matrix. Deoxycholate-insoluble matrices were made from confluent human fibroblast cell layers. No signif- icant binding or cross-linking of fibronectin was detected when these isolated matrices were used in a time course binding assay (Fig. 3). Previous studies have shown that treatment with cytochalasin B causes cells to round up, dis- rupts the actin filament component of cytoskeletal networks, and causes the separation of the fibronectin fibrillar network from the cell body (27). There was no significant binding or cross-linking of fibronectin to cytochalasin B-treated fibro- blast cell layers (Table I). In the converse experiment, cell layers were either lightly trypsin-treated or completely tryp- sin-digested and freshly plated to remove their extracellular matrix prior to their use in a binding assay (not shown). After 2-4 h at 37 "C, the amounts of lZ5I-fibronectin bound in pool I of both types of trypsin-treated cells were the same in the presence and absence of Factor XIIIa while pool I1 accumu-

0-0 POOL I- 0-0 POOL I+ - D-a POOL 11- w POOL XI+ w ISOL. MATRIC- / -

:2ot A-A ISOL. MATRIC+ // 1

TIME (HOURS)

FIG. 3. Time course of the binding of human plasma f'ibro- nectin (FN) to cell layers. Confluent cultures of human fibroblasts or isolated extracellular matrices (ISOL. MATRIC) were incubated with medium containing '*'I-fibronectin (1 pg or 900,000 cpm/ml) in the presence (+) or absence (-) of 10 pg/ml Factor XIIIa. At the indicated times, labeled media were removed, and the monolayers were processed as described under "Experimental Procedures" to determine specific binding in pool I and pool 11. The points represent the average of duplicate wells. Only total binding, the sum of pool I and pool I1 binding, is shown for the isolated matrices.

Cross-linking of Assembling Fibronectin by Factor XIII 10467 +FXl l l

HRS: 1 2 4 8 24 1 2 4 8 24 "FXI I I

2 4 8 2 4 "FXI I I

D* M-

0

FIG. 4. Autoradiograph of '*W'ibronectin bound in pool II. Deoxycholate-insoluble (pool 11) extracts from the experiment shown in Fig. 3 were solubilized in SDS and divided in half for polyacryl- amide gel electrophoresis analysis under nonreducing ( A ) and reduc- ing ( E ) conditions. The time (HRS) of binding is indicated above each lane. Arrows show the positions of monomeric (M) and dimeric (D) fibronectin.

TABLE I Fibronectin binding and cross-linking to cytochalasin B-treated

cell layers Confluent cultures of human fibroblasts were preincubated with

fresh media with or without 5 pg/ml cytochalasin B (Cyt. B) for 4 h. The media were then removed, and media were added containing '%I- fibronectin (0.5 pg of 1,000,000 cpm/ml) in presence (+) or absence (-) of 10 pg/ml Factor XIIIa and/or 5 pg/ml cytochalasin B. After 2 h at 37 'C, the medium was removed, and the cell layers were processed as described under "Experimental Procedures" to deter- mine fibronectin binding. The data are the average of duplicate wells. Specific binding is also expressed as a percent of that found in the absence of Factor XIIIa and cvtochalasin B.

Conditions Pool

Binding

Factor XIIIa Cyt. B Total Nonspecific Specific

I I I + I + I1 I1 I1 + I1 +

- -

- -

w - 4.4 + 0.51 - 4.7 + 0.43

- 1.3 + 0.09

+ 0.17 - 4.8

w 0.89 0.29 0.60 0.28

0.15 0.07 0.29 0.10

w % 3.9 (100) 0.22 6 4.1 117 0.15 4

1.2 (100) 0.02 2 4.5 315 0.07 6

lation was more than twice as high when Factor XIIIa was present.

The effect of Factor XIIIa on fibronectin binding to cells at different concentrations of fibronectin was examined. Fi- broblasts were incubated with '251-fibronectin with increasing concentrations of unlabeled fibronectin in the presence or absence of 10 pg/ml Factor XIIIa (Fig. 5). At 4 h, when pool

I ' " " " " I

".oo[ ~ - - - - - - - - - - " j Y

n 5 H P O O L I -

e-+ POOL I+

L L loot/

A OO \

20 40 60 80 100 FN(pg/ml)

zz LL OO 20 40 60 80 t o 0

B FN (pg/rnl)

FIG. 5. Saturation binding of f'ibronectin (FN) to cell lay- ers. Confluent cultures of human fibroblasts were incubated at 37 "C with media containing '251-fibronectin (1 pg or 1,500,000 cpm/ml) in the presence (+) or absence (-) of Factor XIIIa. Binding mixes contained increasing amounts of unlabeled fibronectin to give the final total fibronectin concentration indicated. The labeled medium was removed, and the monolayers were processed as described under "Experimental Procedures." The fibronectin bound at each point was determined by multiplying the percent of labeled fibronectin bound by the total amount of fibronectin added. Specific binding in pool I was determined as described under "Experimental Procedures" after 4 h (A) . Accumulation in pool I1 over 3 h was determined by subtract- ing the amount in Pool I1 at 1 h from the amount in Pool I1 a t 4 h ( B ) .

I binding had reached an apparent steady state, the amount of fibronectin bound at fibronectin concentrations below 25 pg/ml was the same in the presence and absence of Factor XIIIa. Above 25 pg/ml, the amount of fibronectin bound in the presence of Factor XIIIa was slightly less. Fibronectin accumulation in pool I1 was greater in the presence of Factor XIIIa at all concentrations of fibronectin. Regardless of the presence of Factor XIIIa, the accumulation of fibronectin in pool I1 and the binding of fibronectin to pool I showed saturability.

No difference in the organization of the fibroblast fibronec- tin matrix formed from fluoresceinated fibronectin in the presence of Factor XIIIa compared to that formed in its absence was seen by fluorescent microscopy (not shown).

The specificities of fibronectin binding and Factor XIIIa- mediated cross-linking to fibroblast cell layers were examined (Fig. 6). Fibroblasts were incubated with '251-fibronectin in the presence of increasing concentrations of unlabeled fibrin- ogen or the 70-kDa amino-terminal fragment of fibronectin. Control values were measured in the absence of potential competitor; this value was more than twice as high for pool I1 formed in the presence of Factor XIIIa than in the absence of the transglutaminase. Fibrinogen, at concentrations up to 100 nM, caused a slight increase of 15-20% in the amount of fibronectin bound in pool I and a similar decrease in the pool I1 binding regardless of the presence of Factor XIIIa. The 70- kDa fragment of fibronectin, in contrast, blocked 80-90% of

10468 Cross-linking of Assembling Fibronectin by Factor XIII

120-

100 FIBRINOGEN z 3

m 0 SO-

-I - POOL I - 60- 7 0 - K O o FRAGMENT - POOL I+

t- - POOL II- g 4 0 -

- POOLII+ a-“ 20-

- FIG. 1 1 6. Specificity 20 of the 40 nM interaction COMPETITOR 60 of fibronectin SO 100 with 250 cell

layers. Confluent cultures of human fibroblasts were incubated with medium containing “‘I-fibronectin (0.5 pg or 1,400,000 cpm/ml) and the indicated concentration of the 70-kDa amino-terminal fragment of fibronectin or fibrinogen in the presence (+) or absence (-) of Factor XIIIa. After 4 h at 37 “C, labeled medium was removed, and the monolayers were processed as described under “Experimental Procedures” to determine specific binding in pool I and pool 11. The points represent the average of duplicate wells and are expressed as percentages of the specific binding found in the absence of potential competitor.

”_”” ””

r - r I I I I I I 2 4 6 a

” ” ””.-

TIME (HOURS)

FIG. 7. Binding of human plasma fibronectin (FN) to HT- 1080 cell layers. Confluent HT-1080 cell layers were pretreated with M dexamethasone for 48 h. Then the cells were incubated with medium containing 10” M dexamethasone and ”‘I-fibronectin (1 pg or 800,000 cpm/ml) at 37 “C. Binding was measured in the presence (+) and absence (-) of Factor XIIIa (10 pglml). At the indicated times, the labeled media were removed, and the monolayers were processed as described under “Experimental Procedures” to determine specific binding in pool I and pool 11. Points are the average of duplicate wells.

fibronectin binding in both pool I and pool I1 in the presence or absence of Factor XIIIa.

Previous studies have shown that HT-1080 fibrosarcoma cells bind fibronectin after pretreatment with dexamethasone (10). While the characteristics of binding were similar to that of normal fibroblasts, the amount of pool I binding was 40% of the pool I binding to normal fibroblasts, and fibronectin accumulated in pool I1 at a slower rate (10). The effects of 10 pg/ml Factor XIIIa on fibronectin binding by noninduced (not shown) and dexamethasone-induced (Fig. 7) HT-1080

cells were examined. Cells which had not been pretreated with dexamethasone did not bind fibronectin in either pool regard- less of the presence of Factor XIIIa. Cells pretreated with dexamethasone and incubated with ”‘1-fibronectin in the presence of Factor XIIIa bound 35% more fibronectin in pool I when apparent steady state was reached. The accumulation of fibronectin in pool I1 in the absence of Factor XIIIa was very little but increased dramatically, approximately 12-fold, in the presence of Factor XIIIa.

HT-1080 pool I1 cell extracts analyzed by SDS-PAGE and autoradiography showed the same characteristics of fibronec- tin cross-linking in the presence of Factor XIIIa as was described in Fig. 4 for normal fibroblasts (not shown). No difference in the organization of cell-bound fluoresceinated fibronectin formed in the presence of Factor XIIIa compared to in its absence was seen when dexamethasone-treated HT- 1080 cells were examined by fluorescent microscopy (not shown).

DISCUSSION

Fibronectin matrix assembly is thought to proceed in a stepwise manner with initial binding to cell surface sites, represented by the deoxycholate-soluble pool I, followed by multimerization by disulfide exchange and transfer to the extracellular matrix represented by the deoxycholate-insolu- ble pool I1 (4-10,28). In this paper we show that while Factor XIII does not alter the ability of fibronectin to bind to the cell surface, it increases the rate of transfer of bound fibro- nectin into the deoxycholate-insoluble extracellular matrix.

The characteristics of fibronectin binding in pool I were similar in the presence and absence of Factor XIIIa. Binding in pool I initially increases linearly with time and then pla- teaus at about 4 h. Pool I1 binding, on the other hand, continues to increase linearly over 24 h. In the presence of Factor XIIIa, the rate of accumulation of fibronectin in pool I1 of normal fibroblasts approximately doubled. This effect was maximal at a low concentration of Factor XIIIa ( 5 pg/ ml) and was true for a variety of fibronectin concentrations. At high fibronectin concentrations, fibronectin binding and accumulation showed saturability in the presence of Factor XIIIa. Fibronectin did not bind or become cross-linked to isolated matrices or to cytochalasin B-treated cell layers. However, Factor XIIIa increased the accumulation of fibro- nectin in pool I1 of lightly trypsin-treated or freshly plated cell layers that lacked extensive matrices. The 70-kDa amino- terminal fragment of fibronectin, which has been shown to block the fibronectin matrix assembly process (9, 29, 30), blocked both the binding and Factor XIII-mediated cross- linking of fibronectin. These results suggest that fibronectin is not cross-linked by Factor XIIIa directly to the extracellular matrix. Rather, fibronectin binds initially at the cell surface where it becomes available for Factor XIIIa-mediated cross- linking.

The effect of Factor XIIIa on the binding of fibronectin by dexamethasone-treated HT-1080 fibrosarcomas had similar characteristics, but the results were more striking. The dexa- methasone-treated HT-1080 cells seem defective in their abil- ity to transfer fibronectin to pool I1 (10). Factor XIIIa caused a 12-fold increase in the accumulation of fibronectin in pool 11. Since very little disulfide exchange occurs in these cells, exchange is apparently not required for Factor XIIIa-me- diated cross-linking to occur. Rather, it appears that fibro- nectin can be transferred into the matrix either by a cell- mediated disulfide exchange mechanism or by a Factor XIII- mediated cross-linking mechanism. One distinguishing char- acteristic seen with the HT-1080 cells, that was not seen with

Cross-linking of Assembling Fibronectin by Factor XIII 10469

normal fibroblasts, was an increase in pool I binding when Factor XIIIa was present.

Once a fibronectin molecule has bound in the cell layer, it is apparently covalently cross-linked by Factor XIIIa to a specific neighboring molecule. That molecule may be a prop- erly aligned fibronectin molecule on the end of a growing fibril, or some other matrix molecule, such as collagen. Some of the cross-linked fibronectin formed in the presence of Factor XIIIa is reduced to dimer-sized molecules rather than fibronectin monomer on SDS-PAGE analysis. This observa- tion suggests that nonreducible cross-linking occurs between two fibronectin monomers and that fibronectin molecules that are cross-linked can also undergo disulfide exchange. A small amount of similar nonreducible dimers of fibronectin was observed after a longer period of time in the absence of added Factor XIIIa. This may be due to the activity of a cellular transglutaminase, possibly released by a small amount of cell breakage. Cellular transglutaminase has a similar specificity for fibronectin glutamines as Factor XIIIa (14).

We propose that Factor XIIIa-mediated cross-linking of fibronectin at the fibroblast cell surface is a specific process involving particular protein pairs. In our assay, fibronectin binding and Factor XIIIa-mediated cross-linking to the fibro- blast cell layer were not inhibited by fibrinogen, another Factor XI11 substrate. In contrast, fibrinogen did inhibit the calcium-dependent interaction of fibronectin with rabbit hep- atocytes (31). In addition, cross-linking of extracellular matrix proteins of human fibroblasts does not seem to be a ubiquitous event. We have attempted to demonstrate cross-linking of thrombospondin, a platelet cy-granule protein, to fibroblast cell layers. Thrombospondin is a Factor XIIIa substrate (32) and binds to the extracellular matrix of fibroblast cell layers (33). Bound ‘261-thrombospondin, however, was not cross- linked when the cell layers were incubated with Factor XIIIa.’

The increased rate of assembly of the fibroblast extracel- lular matrix in the presence of Factor XIIIa may be a critical event during the formation of a provisional matrix during re- epithelialization of a wound (34). Fibroblasts migrate into the wound bed within a few days of injury. Coincidently, fibro- nectin becomes one of the major constituents of the newly forming matrix. Increased vasopermeability of the blood ves- sels in the early wound would supply Factor XI11 and plasma fibronectin. Therefore, one mechanism by which Factor XI11 may promote wound healing is by increasing the rate of matrix assembly by fibroblasts in granulation tissue and by forming a cross-linked matrix important for increased tensile strength of the wound.

REFERENCES

1. Mosher, D. F. (1984) Annu. Rev. Med. 3 5 , 561-575 2. Hynes, R. 0. (1985) Annu. Rev. Cell Biol. 1, 67-90 3. Skorstengaard, K., Jensen, M. S., Sahl, P., Petersen, T. E., and

* E. L. R. Barry and D. F. Mosher, manuscript in preparation.

Magnusson, S. (1986) Eur. J. Biochem. 161,441-453 4. McKeown-Longo, P. J., and Mosher, D. F. (1983) J. Cell Biol.

5. McKeown-Lonpo. P. J., and Mosher, D. F. (1984) J. Biol. ChPm 97,466-472

259 , rmo- i2 i15 6. Peters, D. M. P., and Mosher, D. F. (1987) J. Cell Biol. 104,121-

130 7. Allen-Hoffmann, B. L., and Mosher, D. F. (1987) J. Biol. Chem.

8. Allen-Hoffmann, B. L., Crankshaw, C. L., and Mosher, D. F.

9. McKeown-Longo, P. J., and Mosher, D. F. (1985) J. Cell Biol.

10. McKeown-Longo, P. J., and Etzler, C. A. (1987) J. Cell Biol. 104 ,

11. Lorand, L., Losowsky, M. S., and Miloszewski, K. J. M. (1980) Prog. Hemostasis Thromb. 5, 245-290

12. McDonagh, J. (1987) in Hemostasis and Thrombosis (Colman, R. W., Hirsh, J., Marder, V. J., and Salzman, E. W., eds) pp. 289- 300, J. B. Lippincott Co., Philadelphia

13. Mosher, D. F. (1987) in Hemostasis and Thrombosis (Colman, R. W., Hirsh, J., Marder, V. J., and Salzman, E. W., eds) pp. 210- 218, J. B. Lippincott Co., Philadelphia

14. Fesus, L., Metsis, M. L., Muszbek, L., and Koteliansky, V. E. (1986) Eur. J. Biochem. 164,371-374

15. McDonagh, R. P., McDonagh, J., Petersen, T. E., Thplgersen, H. C., Skorstengaard, K., Sottrup-Jensen, L., and Magnusson, S.

262,14361-14365

(1988) J. Mol. Cell. Biol., in press

100,364-374

601-610

(1981) FEBS Lett. 127 , 174-178 16. Mosher, D. F. (1975) J. Biol. Chem. 260,6614-6621 17. Mosher, D. F., Schad, P. E., and Kleinman, H. K. (1979) J. Clin.

18. Mosher, D. F., and Johnson, R. B. (1983) J. Biol. Chem. 2 5 8 ,

19. Williams, E. C., Janmey, P. A., Johnson, R. B., and Mosher, D.

20. Keski-Oja, J., Mosher, D. F., and Vaheri, A. (1976) Cell 9 , 29-35 21. Mosher, D. F., and McKeown-Longo, P. J. (1983) in Factor XZZZ

and Fibronectin (Egbring, R., and Klingemann, H.-G., eds) pp. 223-227, Die Medizinische Verlagsgesellschaft mbH, Marburg, Federal Republic of Germany

Invest. 64,781-787

6595-6601

F. (1983) J. Biol. Chem. 258,5911-5914

22. Grinnell, F., Feld, M., and Minter, D. (1980) Cell 19, 517-525 23. Kamykowski, G. W., Mosher, D. F., Lorand, L., and Ferry J. D.

24. Paye, M., and LaPiere, C. M. (1986) J. Cell Sci. 86,95-107 25. Mosher, D. F., and Thompson, E. D. (1986) Ann. N . Y. Acad. Sci.

26. Laemmli, U. K. (1970) Nature 227,680-685 27. Mautner, V., and Hynes, R. 0. (1977) J. Cell Biol. 76,743-768 28. Choi, M. G., and Hynes, R. 0. (1979) J. Biol. Chem. 264,12050-

12055 29. Spiegel, S., Yamada, K. M., Hom, B. E., Moss, J., and Fishman,

P. H. (1986) J. Cell Biol. 102, 1898-1906 30. McDonald, J. A., Quade, B. J., Broekelman, T. J., LaChance, R.,

Forsman, K., Hasegawa, E., and Akiyama, S. (1987) J. Biol. Chem. 262,2957-2967

31. Fellin, F. M., Barsigian, C., Rich, E., and Martinez, J. (1988) J. Biol. Chem. 2 6 3 , 1791-1797

32. Bale, M. D., and Mosher, D. F. (1986) Biochemistry 2 6 , 5667- 5672

(1981) Biophys. Chem. 13, 25-28

485,264-272

33. McKeown-Longo, P. J., Hanning, R., and Mosher, D. F. (1984) J. Cell Biol. 9 8 , 22-28

34. Clark, R. A. F., and Colvin, R. B. (1985) in Plasma Fibronectin (McDonagh, J., ed) pp. 197-262, Marcel Dekker, Inc., New York