Fibronectin: a chromatin-associated protein?

Cell, Vol. 18. 649-657.

November

Fibronectin:

1979.

Copyright

0 1979

by MIT

a Chromatin-Associated

Lociano Zardi, Annalisa Siri, Barbara Carnemolla and Leonardo Santi lstituto di Oncologia Universita’ di Genova lstituto Scientific0 per lo Studio e la Cura dei Tumori di Genova Viale Benedetto XV, 10 16132-Genova, Italy William D. Gardner and Sallie 0. Hoch Department of Cellular Biology Scripps Clinic and Research Foundation La Jolla, California 92037

Summary We have previously reported that chromatin preparations from human cultured fibroblasts contain a single homologous serum protein. In this paper we present evidence, based on immunological identity and physicochemical properties, that this serum protein is fibronectin. Furthermore, using a radioimmunoassay system, we have estimated that fibronectin represents about 0.7% of the total protein in both chromatin preparations and whole fibroblasts. Using a nitrocellulose filter assay system, we also show that fibronectin is a DNA-binding protein having an equilibrium constant of 4.6 x lo-’ M. Equilibrium competition experiments have demonstrated that fibronectin has the ability to differentiate among nucleotides, indicating that fibronectinDNA interaction is at least partially specific, and that a minimum polymer length of 12-18 nucleotides is required for effective binding to occur. Fibronectin has been isolated readily from plasma using DNA-affinity chromatography. We do not have direct evidence that fibronectin is an actual nonhistone chromosomal protein, but fibronectin is a DNAbinding protein (at least under in vitro assay conditions) and appears to be a normal constituent of chromatin as chromatin is currently isolated from cell nuclei. Introduction Specific antibodies are extremely powerful tools for the study of heterogeneous complexes of molecules, so we began our study of chromatin-associated proteins using immunochemical techniques (Zardi, Lin and Baserga, 1973; Zardi et al., 1974, 1978; Zardi, 1975). We have recently reported that antibodies to chromatin proteins from human or mouse fibroblasts which have been cultured for more than 10 generations with heterologous serum show specificity for a single homologous serum protein (Zardi, Siri and Santi, 1976). This observation indicates that among

Protein?

the chromatin-associated proteins, there is one which has extensive structural similarities to a serum protein (Zardi et al., 1976). In this paper we present evidence that this protein is fibronectin, also known as coldinsoluble globulin, a major plasma protein (Mosesson and Umfleet, 1970; Mosher, 1975; Mosesson, Chen and Huseby, 1975). Fibronectin was observed for the first time 30 years ago by Morrison, Edsall and Miller (1948) in studies on the purification of human fibrinogen. Cell surface fibronectin (also known as LETS protein, CSP or galactoprotein “a”) is a protein with many immunological and physicochemical similarities to plasma fibronectin (Gahmberg, Kiehn and Hakomori, 1974; Ruoslahti and Vaheri, 1975; Burridge, 1976; Hynes, 1976; Vaheri and Mosher, 1978; Yamada and Olden, 1978). These proteins are known to share some biological functions (Pena and Hughes, 1978). The potential functions of fibronectin are the subject of intense investigation relative to the role of the protein in cell-cell and cell-substratum interactions. The relationship of this protein to the transformed cell is of particular interest. A correlation between an increase in tumorigenicity and a decrease in the percentage of cells expressing cell surface fibronectin has been postulated for several cell lines (Chen, Gallimore and McDougall, 1976), although recent evidence suggests that such a decrease may actually indicate the tendency of such cells to metastasize (Chen et al., 1978). As a corollary, reconstitution experiments involving the addition of purified cell surface fibronectin to transformed cells can partially restore the normal morphology, adhesion and cytoskeleton of these cells (Yamada, Yamada and Pastan, 1976; Ali et al., 1977). The functional reason for the binding of fibronectin to an array of macromolecules (collagen, heparin, hyaluronic acid, fibrin) is not clear (Vaheri and Mosher, 1978; Yamada and Olden, 1978). The recent identification of fibronectin as the opsonic a2 surface-binding glycoprotein of the reticuloendothelial system, however, may indicate a critical role in the immune response (Blumenstock et al., 1978). We present evidence that a relatively large amount of fibronectin is associated with purified chromatin preparations from cultured human fibroblasts. Furthermore, with a variety of in vitro assays that have been used to characterize the DNA-binding proteins in serum (Hoch, Longmire and Hoch, 1975; Parsons and Hoch, 1976; Hoch and McVey, 19771, fibronectin can be demonstrated to be a DNA-binding protein of some specificity, having an equilibrium constant in the range of that observed for the lac repressor for nonoperator DNA. The ability of fibronectin to bind DNA has been used to isolate fibronectin from plasma by DNA-affinity chromatography.

Cell 650

Results Identification of the Serum Protein Cross-Reactive with Antiserum to Chromatin and Its Ouantitation in Chromatin Preparations We have previously reported that antibodies to chromatin from cultured human fibroblasts react with a single homologous serum protein (Zardi et al., 1976). In the present paper this protein is identified as plasma fibronectin by the following criteria. lmmunoassays Antichromatin antibodies in the immunodiffusion assay react with human serum, showing a pattern of complete identity with the fibronectin precipitin line (Figure I). A complete inhibition of the precipitation reaction between human serum and antichromatin antibodies was observed after human serum absorption by antibodies to fibronectin. Similarly, inhibition of the precipitation reaction between human serum and antibodies to fibronectin was observed after human serum absorption by antichromatin antibodies. In immunoelectrophoresis experiments, fibronectin and the serum protein reacting with antichromatin antibodies have identical electrophoretic mobilities, migrating as /3, globulins (Figure 2). Ammonium Sulfate Fractionation An electroimmunoassay using antibodies to human fibronectin and antichromatin antibodies was used to quantitate the two antigens precipitated from human serum at different concentrations of ammonium sul-

fate. Plasma fibronectin and the serum protein reacting with antichromatin antibodies were precipitated in parallel at concentrations of ammonium sulfate ranging from 15 to 35% saturation as follows: at 15-239/o saturation, neither protein precipitated; at 26% saturation, 50% of the two proteins precipitated; at 29% saturation, 75% of the proteins precipitated; at 32% saturation, both proteins precipitated completely. Gel Filtration Chromatography Figure 3 shows that human fibronectin and the chromatin-like serum protein elute at identical positions during human serum fractionation on an Ultrogel AcA 22 chromatography column, indicating that the two proteins have similar molecular weights. Gelatin Affinity Chromatography Engvall and Ruoslahti have reported (1977) that fibronectin is one of the few plasma proteins with a high affinity for collagen. We therefore studied the gelatin affinity of the human fibronectin and serum protein reacting with antichromatin antibodies. Serum proteins bound to a gelatin-coupled CNBr-activated Sepharose 4B column were eluted using a linear urea gradient (O-5 M). Figure 4 shows that the human serum protein reacting with antichromatin antibodies and the human fibronectin are both gelatin-binding proteins and elute identically at a urea concentration of 1.3-2-l M. To quantitate the amount of fibronectin in whole fibroblasts and in our chromatin preparation we have used the radioimmunoassay system described in Experimental Procedures. Using this system we have compared the inhibition of binding of antifibronectin antibody to fibronectin absorbing the antiserum (diluted l/270) with various amounts of purified fibronectin, sonicated whole fibroblasts or sonicated chromatin preparations (Figure 5). Our results indicate that in both chromatin preparations and whole human fibroblasts, fibronectin represents about 0.7% of the total protein. DNA-Binding Properties of Fibronectin Using the nitrocellulose filter assay, the equilibrium constant for the binding of double-stranded Wilp lymphocyte DNA to fibronectin was determined to be 4.6 x 10e6 M (Figure 6). The protein did not exhibit any preference for double- versus single-stranded DNA as

Figure 1. lmmunodiffusion Assay Using Antibodies to Human nectin and to Chromatin from Cultured Human Fibroblasts

Fibro-

The samples in each well are as follows: (a) antibodies to fibronectin; (b) human serum absorbed by antibodies to fibronectin; (c and f) antichromatin antibodies; (d and g) human serum; (e) human serum absorbed by antichromatin antibodies. Absorption of serum was carried out by adding 3 vol of fibronectin or chromatin antibodies, followed by incubation at 37°C for 4 hr and centrifugation at 2000 X g for 30 min. The supernatant was used for immunodiffusion assay. The precipitation reaction between g and b. d and e. is due to the excess of antibodies used for the absorption.

L-r

_

._..

_ .

-+..a--

Figure

-

-

2. lmmunoelectrophoresis

_I.

-

of Human

(A) Rabbit antiserum to human plasma rum to human fibroblast chromatin.

-

Serum

fibronectin;

I3

--

in Agarose (B) rabbit

antise-

I

Fibronectin: 651

Figure Serum

a Chromatin-Associated

3. Gel Proteins

Filtration

Protein?

Chromatography

of

6.0 t

5 ml of human serum were applied to an Ultrogel AcA 22 column (2.6 x 100 cm) equilibrated in 0.1 M Tris-HCI (pH 8.0) containing 0.4 M NaCl and 0.05% NaN3. The buffer was pumped at a rate of 5.5 ml/hr and 4.4 ml fractions were collected. The profile was monitored by measuring the absorbance at 280 nm. Column fractions were assayed by immunodiffusion: (+) fractions having a posdive precipitation reaction with antichromatin antibodies; (0) fractions having a positive precipitation reaction with antifibronectin antibodies.

Fraction Figure 4. Gelatin-Sepharose matography of Fibronectin

Affinity

number

Chro-

10 ml of human serum were recycled through a column of gelatin-Sepharose (1 X 10 cm) at a flow rate of 30 ml/hr for 24 hr. The column was then washed extensively with phosphatebuffered saline before eluting the gelatin-binding proteins with a urea gradient (O-5 M). The protein profile was monitored by measuring the absorbance at 280 nm. Column fractions were assayed by immunodiffusion: (+) fractions having a positive reaction with antichromatin antibodies; (0) fractions having a positive reaction with antibodies to fibronectin

tested by competition assays using human placental DNA (Table 1). To examine the ability of fibronectin to differentiate among the nucleotides commonly found in DNA and RNA, we tested the binding of this protein to a series of synthetic homopolyqrers. Table 2 summarizes the results and suggests that the binding is at least partially specific rather than the result of nonspecific interactions with the phosphate backbone. Moreover, a minimum length nucleotide fragment is necessary for effective binding to occur. No binding was observed with the polynucleotides dGn, dGs and dGIo. Using the polymer dG,2-18, we were able to calculate a relative affinity of 2.13, which coincides with the results obtained with poly(dG) (as shown in Table 2) using the base of 1 .OO for Wilp lymphocyte DNA. Finally, saturation binding studies of fibronectin and the Wiln lymphocyte DNA indicate an average of one molecule bound per 280 nucleotide bp.

Isolation of Plasma Fibronectin by DNA-Affinity Chromatography The ability of fibronectin to bind DNA has been utilized to isolate fibronectin from plasma using affinity chromatography on DNA-cellulose, as described in Experimental Procedures. Samples from individual fractions obtained from affinity chromatography on DNA-cellulose were applied to a 10% sodium dodecylsulfate-polyacrylamide slab gel to determine the protein distribution (Figure 7). The fibronectin is localized primarily in the 0.4 M NaCl eluate fractions, as represented by the heavy band in the high molecular weight range,at the top of the gel. At a lower concentration of acrylamide this band, as isolated from DNA-cellulose, can be resolved into a closely spaced triplet. The identity of this protein as both fibronectin and the protein recognized by the chromatin antiserum

Cl?ll

652

Table 1. Affinity Competing

Polynucleotide

5. Inhibition of Binding of Mouse Antibodies to Human Fibronectin

Anti-human

Plasma

Fibro-

Antibody absorption was carried out with various amounts Q.rg) of human fibronectin (M). sonicated cultured human fibroblasts (A-A) and sonicated chromatin from cultured fibroblasts (O---O), as detected using the solid phase double-antibody radioimmunoassay described in Experimental Procedures.

In

-3.0

-2.0

‘g

-1.0

5.0c

0

1.0

(l/tDNA

I

I

I

2.0

3.0

40

ADDEOI)X~O-~

Figure 6. Determination of the Equilibrium Constant for Binding of 3H-Labeled Double-Stranded Wilz Lymphocyte DNA by Human Plasma Fibronectin The DNA concentrations are expressed as moles of nucleotides per liter. Linear regression analysis data: intercept = 1.36 x 1 O5 liters per mole; slope = 0.65; r = 0.9994.

was verified by immunodiffusion, as shown in Figure 8. We obtained a complete line of identity using antisera to fibronectin obtained from three separate laboratories and using the antiserum to human fibroblast chromatin. Discussion In this paper we show that the human serum protein reacting with antibodies to chromatin from cultured fibroblasts is fibronectin. This identification is based on the criteria of immunological identity, electrophoretie mobility, precipitation by equal concentrations of ammonium sulfate, molecular weight as determined by gel filtration and gelatin affinity. We obtained a good immunological response from each of ten rabbits by using a total of 1.5 mg of chromatin proteins for each animal. Considering the amount of antigen that is generally required to induce

for Natural

Polynucleotides’ Relative

Human Wiln lymphocyte Double strandb

1 .oo

Human placental Double strand Single strand

1.35 1.39

CU. perfringens Single strand Figure nectin

of Fibronectin

DNA

DNA

0.48

B. subtilis DNA Single strand M. lysodeikticus Single strand

Affinity

0.87 DNA 0.19

’ All relative affinities were obtained by dividing the amount of labeled Wib lymphocyte DNA present by the initial concentration of competitor necessary to reduce the amount of the label bound to the filter by 50%. This quantity is called the [Iso]. ’ Since double-stranded Wilz lymphocyte DNA was the standard from which all relative affinities were calculated, it is defined as having a relative affinity of 1 .OO.

antibodies in rabbits, we can conclude that a considerable amount of fibronectin is consistently present in chromatin preparations from cultured fibroblasts. Furthermore, by studying the inhibition of the binding of antifibronectin antibodies to human fibronectin with various amounts of sonicated whole fibroblast and sonicated chromatin preparations, we have estimated that fibronectin represents about 0.7% of the total protein of both chromatin and whole fibroblasts. These figures should be considered carefully, Since after sonication we were absorbing antibodies with a suspension rather than a solution and at least part of the fibronectin determinants could have been masked by other cellular components. [Vaheri and Ruoslahti (1975) and Yamada, Yamada and Pastan (19771, however, have reported that in normal human fibroblasts, fibronectin represents from 0.5 to 3% of the total cellular protein.] These observations suggest that fibronectin is an in vivo chromatin-associated protein and that it is constantly present in chromatin preparations as a contaminant. Even though it has been reported that contamination of chromatin prepared as described is negligible (Augenlicht and Baserga, 19731, we have no direct evidence for either of these possibilities. Furthermore, due to the large number of proteins present in both the cytoplasm and nucleus (Peterson and McConkey, 19761, it is almost impossible to establish the origin of certain proteins. The possibility of such a high level of contamination must be a consideration in studies on chromatin proteins. The reduced level of fibronectin in transformed fibroblasts could be the cause of artifacts in comparative experiments of chromatin protein from normal and transformed cells. The following considerations favor the nuclear lo-

Fibronectin: 653

a Chromatin-Associated

Protein?

calization of some fibronectin. There is precedent for the identification of one or more nonhistone chromosomal proteins with serum proteins; Ruoslahti et al. (1977) identified the mouse urinary protein (MUP) as such a protein. Chromatin proteins have also been found in the cellular cytoplasm (Peterson and McConkey, 1976). More recently, Bustin and Neihart (1379) using specific antibodies to chromosomal HMG proteins, have demonstrated the presence of these proteins in the cytoplasm. Comings and Harris (1976) suggested that practically all nonhistone chromosomal proteins could have cytoplasmic counter-

Table 2. Affinity Polynucleotrdes” Competing

of Plasma

Fibronectin

for Synthetic

Relative

Polynucleotide

Human Wrl? lymphocyte Double strand’

Affinity

DNA 1 .oo

Poly(rG)

9.31

Poly(rl)

1.52

Poly(dG)

2.04

Poly(dT)

0.1

P0ly(rU)

0

Poly(rA)

0

Poly(rC) Poly(dG-dC)

parts, and they postulated that some cytoplasmic and nuclear protein could be in equilibrium. The data presented here identify fibronectin as a DNA-binding protein of some specificity. The facts that only a limited number of the synthetic polynucleotides tested are bound by the fibronectin and that a sizeable fragment is necessary if any binding is to occur suggest a specific site capable of recognizing differences between nucleotides. The specificity of nucleic acid binding shown by fibronectin lends support to the possibility that it may have an actual in vivo function. The equilibrium constant of fibronectin is also similar to those of the lac repressor for non-operator DNA. Reported values for the equilibrium constant of the lac repressor of 1 O-5 to 1 O-a M, depending upon the base composition of the DNA used, compare favorably with values of 1 Oe6 M for double-stranded Wila lymphocyte DNA with fibronectin (Richmond and Steitz, 1976). There has been no direct demonstration of fibronectin in the cell nucleus by the usual immunofluorescence techniques, but these results may parallel the results of similar studies with actin. In an early demonstration of actin filaments in mouse fibroblasts (using fluorescent antibodies), the observed nuclear fluorescence was believed to represent only nonspecific antibody binding. Yet in a study of cellular protein distribution, actin was demonstrated to be the major constituent of the nonhistone chromosomal proteins isolated from HeLa cells (Peterson and McConkey, 1976). Furthermore, several observations also suggest the possibility that fibronectin could be associated with the cellular actin in vivo (Kuusela, Ruoslahti and Vaheri, 1975; Ali and Hynes, 1977; Hynes and Destree, 1978).

0 - (dG-dC)

0

a All relative affinities were obtained by dividing the amount of labeled Wilz lymphocyte DNA present by the initial concentration of competitor necessary to reduce the amount of label bound to the filter by 50%. This quantity is called the [kO]. ’ Since double-stranded Wilp lymphocyte DNA was the standard from which all relative affinities were calculated, it is defined as having relabve affinity of 1 .OO.

Figure 7. DNA-Cellulose Fibronectin

Chromatography

of

Approximately 670 mg of protein were applied to a DNA-cellulose column equilibrated in 0.1 M Tris-phosphate (pH 6.6) containing 1 mM 2-mercaptoethanol. 1 mM EDTA and 0.1 mM phenylmethylsulfonyl fluoride. The column was washed in succession with this buffer containing 0. 0.1 and 0.4 M NaCI. (Inset) Sodium dodecylsulfate-polyacrylamide slab gel electrophoresis of individual column fractions. (1) Wash fraction pool of tubes 14-30; (2-4) 0.1 M NaCl eluate, tubes 55. 56 and 57, respectively; (5-7) 0.4 M NaCl eluate. tubes 77, 76 and 79. respectively.

3.0

2.0 g 2 I 1.0

L

I

IO

20

30

40

50

Fraction Number

60

70

80

Cell 654

Our ability to isolate fibronectin easily as a DNAbinding protein from sera from normal subjects may be attributed to two features of our protocol: starting with plasma that has a concentration of fibronectin approximately twice that of serum, due to fibronectin binding to the fibrin clot (MOSeSsOn and Umfleet, 1970; Ruoslahti and Vaheri, 1975); and the maintenance of physiological salt concentrations during all procedures so that the fibronectin did not precipitate out of solution. Plasma fibronectin is a DNA-binding protein at least under in vitro assay conditions. Whether it is an actual chromatin protein remains to be determined, but in any event it appears to be a normal constituent of chromatin as chromatin is presently isolated from the cell nucleus. Experimental

Figure

8.

Immunoassay

of Plasma

DNA-Binding

Proteins

(a) 0.4 M NaCl eluate from the DNA-cellulose column described in the legend to Figure 5. Protein concentration is 0.49 mg/ml; (b. d and e) antifibronectin antibodies from three separate laboratories: (c) antibodies to human fibroblast chromatin.

l/30

l/90

l/270

l/S10 l/1620 113240 l/6480 ANTISERA DILUTION

l/12960

Figure 9. Titration Curve of the Mouse Anti-Human Plasma Fibronectin Antiserum (O--O) Using the Solid Phase Double-Antibody Radioimmunoassay Described in Experimental Procedures (M)

Mouse

preimmune

serum.

As far as our results are concerned, there has recently been a report of a 200,000 dalton serum protein that binds tightly to DNA-cellulose eluting with 600 mM salt. This protein is absent or reduced in normal individuals but is present in significant amounts in the sera of individuals with various malignancies (Parsons et al., 1979). Using this protocol, the DNAbinding protein fraction eluted with a high salt solution was isolated from 5 ml of cancer sera. When concentrated, it produced a weak reaction with antifibronectin antisera (S. 0. Hoch, unpublished data). Fibronectin also shows many physicochemical similarities to a previously described DNA-binding serum protein (Thoburn, Hurvitz and Kunkel, 1973).

Procedures

Cell Cultures Human skin fibroblasts were obtained from skin explants as reported by Paul (1972). and were grown in Minimal Essential Medium (Flow Laboratories; Irvine, Scotland) supplemented with 10% fetal calf serum. Cells grown for lo-20 passages were used after forming confluent monolayers in plastic flasks (Falcon. 75 cm2). Preparation of Chromatin Confluent human skin fibroblasts were rinsed twice with 10 ml of icecold calcium- and magnesium-free Hank’s balanced salt solution. The cells were then scraped off with a rubber policeman into calcium/ magnesium-free Hank’s solution and pelleted at 800 x g for 6 min at 4°C. The cells were homogenized in 10 vol of 80 mM NaCl and 20 mM EDTA (pH 7.4). The homogenate was centrifuged at 800 x g for 6 min. The crude nuclear pellet was washed with 10 vol of the above mentioned saline-EDTA solution containing 1% Triton X-i 00 and then with 10 vol of 0.05 M Tris-HCI (pH 8) at 4°C and was recovered each time by centrifugation at 800 X g for 6 min. The washed nuclear pellet was resuspended in 5 vol of distilled water, disrupted with 30 strokes of a tight-fitting Dounce homogenizer and sedimented at 18,000 x g for 15 min at 4°C. The homogenization and centrifugation were then repeated. The final sediment was resuspended by gentle homogenization in 6 ml of distilled water, mixed with 18 ml of 1.7 M sucrose, layered onto 6 ml of 1.7 M sucrose in an SW25 Spinco tube and then centrifuged for 3 hr at 24,000 rpm. Approximately 80% of the total cellular DNA was recovered in the chromatin pellet. Antisera Preparation Chromatin prepared as described above was used to immunize New Zealand white rabbits as previously reported (Zardi et al.. 1973, 1974, 1978; Zardi. 1975). The chromatin was suspended in 10 mM Tris-HCI (pH 8.0) and homogenized in the same volume of complete Freund’s adjuvant (Difco; Detroit, Michigan). Four injections, each containing 300 pg of total proteins, were given intradermally to rabbits on day 1,8. 16 and 24. A subcutaneous booster injection was given 7 days later. The antisera were collected 7 days afler the booster. Fibronectin prepared as described below was used to immunize BALB/c mice and New Zealand white rabbits. In mice, four injections each containing 20 pg of purified fibronectin in PBS homogenized in an equal volume of complete Freund’s adjuvant (Difco) were given intradermally on day 1, 8. 16 and 24. Antisera were collected 7 days later and absorbed with Sepharose conjugated to fibronectin-free human plasma proteins obtained by passing plasma through gelatinSepharose 48 as described by Zardi. Carnemolla and Croce (1979: manuscript in preparation). The antiserum obtained was monospecific as determined by immunoelectrophoresis. The rabbit immunization procedure was similar to that followed for mice, except that 100 pg

Fibronectin: 655

a Chromatin-Associated

Protein?

of fibronectin were used for each injection and a subcutaneous booster was given on day 31. Samples of rabbit antisera to purified fibronectin were donated by A. Vaheri. D. F. Mosher and M. W. Mosesson. lodinated rabbit antibodies against mouse Fab were a gifl from Ft. Accolla. Immunoassay The Ouchterlony and Nilsson procedures were used for immunodiffusion and immunoelectrophoresis (Ouchterlony and Nilsson. 1973). The electroimmunoassay was carried out using the Laurel1 technique (LaurelI, 1972). Radioimmunoassay To estimate the amount of fibronectin in whole fibroblast and chromatin preparations we have used a modification of the solid phase double-antibody radioimmunoassay (RIA) previously described (Klinman et al., 1976; Accolla and Celada. 1978). Purified fibronectin in PBS at a concentration of 100 pg/ml was incubated in wells of polyvinyl plates (Cooke Labs; Alexandria, Virginia). Antigen was removed after 2 hr and wells were filled with PBS containing BSA. After another 2 hr period, the protein solution was removed: wells were washed with PBS and 100 ~1 of an appropriate dilution of mouse antiserum to human plasma fibronectin were added to each well. 2 hr later the antiserum was removed and wells were rewashed with PBS. The amount of antibody bound was detected by adding specific I*?labeled rabbit anti-mouse Fab and incubating for 2 hr at room temperature (100 ~1 per well containing about 25,000 dpm) (Figure 9). The amount of fibronectin present in the chromatin preparation or whole fibroblasts was estimated by comparing the inhibition of antifibronectin antibody binding to fibronectin by absorbing the antiserum (diluted l/270) with various amounts of purified fibronectin, sonicated whole fibroblasts or sonicated chromatin preparations. Absorption of antibodies was carried out by mixing equal volumes of antifibronectin antibodies and the appropriate absorbent in PBS containing 0.2% bovine serum albumin and 0.2 mM phenylmethylsulfonyl fluoride. The mixtures were incubated for 16 hr at 4°C. All dilutions were made in PBS containing 0.2% bovine serum albumin. Fibronectin Purification Fibronectin was isolated using gelatin-Sepharose affinity chromatography essentially according to the procedure of Engvall and Ruoslahti (1977). The protein was more than 95% homogeneous at this stage, but was purified further by passage over a DEAE-cellulose column and then through an Ultrogel AcA 22 (LKB) filtration column. Nitrocellulose Filter Assay For the nucleic acid binding reaction, 2.0 pg of labeled human Wib lymphocyte DNA and 1 pg of fibronectin were brought to a volume of 0.54 ml using 10 mM potassium phosphate (pH 6.0) containing 1 mM 2-mercaptoethanol. The mixture was incubated at room temperature for 20 min. The reaction was stopped by filtration onto nitrocellulose filters (Riggs. Suzuki and Bourgeois, 1970) on a Millipore sampling manifold. The filters were washed three times with 2.0 ml aliquots of the incubation buffer, dried and counted. All other substrates were unlabeled and tested in a competition assay using the labeled doublestranded Wiln lymphocyte DNA as the standard from which all relative binding affinities were calculated. The relative affinities were obtained by dividing the amount of labeled Wil? lymphocyte DNA present by the initial concentration (in moles of nucleotide per liter) of competitor necessary to reduce the amount of labeled Wilp lymphocyte DNA bound to the filter by 50%. This quantity has been called the [Iso]. Human lymphocyte DNA was prepared from Wib lymphocytes grown in the presence of 3H-methyl-thymidine. The DNA in the cell lysate was purified through successive hydroxyapatite columns at room temperature and at 60°C (Meinke and Goldstein, 1974; Meinke. Goldstein and Hall, 1974). Human placental DNA was isolated by the Sevag method (Chuang and Saunders. 1974). The synthetic homopolymers were purchased from Miles Laboratories, and the poly(dG) series from Collaborative Research.

Analytical Procedures Protein concentration was determined by the procedure of Lowry et al. (1951) using bovine albumin as standard. DNA concentrations of chromatin samples were determined by ultraviolet absorption at 260 nm in 0.2% sodium dodecylsulfate or by the diphenylamine technique (Burton. 1956). Electrophoresis Sodium dodecylsulfate-polyacrylamide gel electrophoresis was carried out as previously described using the system of Laemmli (Laemmli. 1970: Hoch and McVey. 1977). Isolation of Plasma Fibronectin by DNA-Affinity Chromatography The first two steps of this isolation procedure were modified from the protocols of Chen and Mosesson (1977) and Mosher (1975) for fibronectin isolation. A 50 ml sample of fresh platelet-deficient plasma was adsorbed by the addition of solid BaCI, (15 g/l) to remove the plasma thrombin. After centrifugation, ammonium sulfate was added to the supernatant to remove the barium ions remaining in solution. The suspension was recentrifuged. The supernatant was brought to 33% ammonium sulfate concentration using a saturated solution of ammonium sulfate to precipitate the fibronectin. The precipitate was resuspended in 20 ml of 0.25 M Tris-phosphate buffer (pH 7.0) containing 1 mM 2-mercaptoethanol, 1 mM EDTA and 0.1 mfvl phenylmethylsulfonyl fluoride (buffer A). The solution was dialyzed overnight at room temperature against 50 vol of buffer A. The nondialyzable fraction was diluted 2.5 fold with distilled water, and the pH was adjusted to 6.8 with 6 N acetic acid. The nondialyzable fraction was then applied to a 20 g DNA-cellulose column (containing 260 mg DNA). The column had been equilibrated in 0.1 M Tris-phosphate buffer (pH 6.8) containing 1 mM 2-mercaptoethanol, 1 mM EDTA and 0.1 mM phenylmethylsulfonyl fluoride (buffer B). The chromatography was carried out at room temperature. After addition of the sample the column was washed with buffer B until no further protein was eluted. The DNA-binding proteins were than eluted with successive washes with buffer B containing 0.1 and 0.4 M NaCI. DNAcellulose was prepared by the method of Litman (1968) using acidwashed cellulose (Whatman CFl 1) and native calf thymus DNA (Worthington). Acknowledgments We thank Drs. Ft. Accolla. M. W. Mosesson, D. F. Mosher and A. Vaheri for their generous gifls of antisera. We are indebted to Mrs. Laura Zardi and Mrs. P. Murphy for their help in the preparation of this manuscript and to Mr. Steve Raffanti for the revision of the manuscript. This study was funded by CNR grant Progetto finalizzato “Controllo crescita Neoplastica” (to L.Z.). and an NIH research grant and Research Career Development Award (to S.O.H.). B. C. was supported by a fellowship from “Lega ltaliana per la lotta contra i tumori.” The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received

April 18, 1979;

revised

July 24, 1979

References Accolla, R. S. and Celada. F. (1978). Immune response against the fl-galactosidase enzyme of E. coli at precursor cell level. I. Analysis of the secondary repertoire in Balb/c mice. Eur. J. Immunol. 8, 688692. Ali, I. U. and Hynes, R. 0. (1977). Effects colchicine on attachment of a major surface Biochim. Biophys. Acta 471. 16-24. Ali. I. U.. Mautner.

V., Lanza,

R. and Hynes,

of cytochalasin 8 and protein of fibroblasts.

R. 0. (1977).

Restoration

Cell 656

of normal morphology, adhesion and cytoskeleton in transformed cells by addition of a transformation-sensitive surface protein. Cell 11, 115-126.

Litman, R. M. (1968). A deoxyribonucleic acid polymerase from Micrococcus luteus (Micrococcus lysodeikticus) isolated on deoxyribonucleic acid-cellulose. J. Biol. Chem. 243, 6222-6233.

Augenlicht, L. M. and Baserga, R. (1973). Preparation fractionation of non-histone chromosomal proteins from loid fibroblast. Arch. Biochem. Biophys. 758, 89-96.

and partial human dip-

Lowry, 0. H.. Rosebrough, N. J.. Farr. A. L. and Randall, R. J. (1951). Protein measurement with the folin phenol reagent. J. Biol. Chem. 193, 265-275.

Blumenstock, F. A., Saba, T. M., Weber, P. and Laffin. Ft. (1978). Biochemical and immunological characterization of human opsonic a2 SB glycoprotein: its identity with cold-insoluble globulin. J. Biol. Chem. 253, 4287-4291.

Meinke. W. and Goldstein, D. A. (1974). Reassociation and disassociation of cytoplasmatic membrane-associated DNA. J. Mol. Biol. 86, 757-773.

Burridge. K. (1976). Changes in cellular glycoproteins after transformation: identification of specific glycoproteins and antigens in sodium dodecyl sulfate gels. Proc. Nat. Acad. Sci. USA 73, 4457-4461. Burton, K. (1956). A study of the conditions and mechanism of the diphenilamine reaction for the calorimetric estimation of DNA. Biothem. J. 62, 315-323. Bustin. M. and Neihart. N. K. (1979). Antibodies against HMG proteins stain the cytoplasm of mammalian cells. 189. Chen, A. B. and Mosesson. M. W. (1977). purification of the cold-insoluble globulin Anal. Biochem. 79, 144-l 51.

chromosomal Cell 16, 181-

An improved method for of human plasma (Clg).

Chen, L. 8.. Gallimore, P. H. and McDougall, J. K. (1976). Correlation between tumor induction and large external transformation sensitive protein on the cell surface. Proc. Nat. Acad. Sci. USA 73, 35703574. Chen. L. 6.. Burridge. K., Bushnell, A., McDougall, J. of cell surface glycocalyx: sensitive protein. Ann. NY

Murray, A., Walsh, M. L., Copple, C. D.. K. and Gallimore, P. H. (1978). Modulation studies on large external transformationAcad. Sci. 312, 366-381,

Meinke. W., Goldstein, D. A. and Hall, M. R. (1974). Rapid of mouse DNA from cells in tissue culture. Anal. Biochem. 88.

isolation 58, 82-

Morrison, P. R., Edsall, J. T. and Miller, S. G. (1948). Preparation and properties of serum and plasma proteins, XVIII. The separation of purified fibrinogen from fraction I of human plasma. J. Am. Chem. sot. 70, 3103-3108. Mosesson, M. W. and Umfleet, R. A. (1970). The cold-insoluble globulin of human plasma. I. Purification, primary characterization, and relationship to fibrinogen and other cold-insoluble fraction components. J. Biol. Chem. 245, 5728-5736. Mosesson, M. W., Chen, A. B. and Huseby. R. M. (1975). insoluble globulin of human plasma: studies of its essential features. Biochim. Biophys. Acta 386, 509-524. Mosher. D. F. (1975). Cross-linking of cold-insoluble stabilizing factor. J. Biol. Chem. 250, 6814-6621.

The coldstructural

globulin

by fibrin

Ouchterlony. 0. and Nilsson. L. A. (1973). lmmunodiffusion and immunoelectrophoresis. In Handbook of Experimental Immunology, 2nd edition, D. M. Weir, ed. (Oxford: Blackwell Scientific Publications), pp. l-1 9, 39-49.

Chuang, C. and Saunders, G. F. (1974). Complexity of human satellite DNA. Biochem. Biophys. Res. Commun. 57, 1221-l 230.

Parsons, R. G. and Hoch, J. A. (1976). Purification and identification of a human-serum DNA binding protein associated with malignant diseases. Eur. J. Biochem. 71, l-8.

Comings, D. E. and Harris, D. C. (1976). Nuclear proteins II. Similarity of non-histone proteins in nuclear sap and chromatin, and essential absence of contractile proteins from mouse liver nuclei. J. Cell Biot. 70, 440-452.

Parsons, R. G., Hoch, J. A., Longmire, R. L. and Kowal, R. (1979). Serum DNA-binding proteins. In Biological Markers of Neoplasia: Basic and Applied Aspects, R. W. Ruddon, ed. (New York: Elsevier/ North-Holland), pp. 385-398.

Engvall. fibroblast l-5.

Paul, J. (1972). Special considerations for different tissues. brate tissues. In Cell and Tissue Culture (London: Churchill stone), pp. 252-263.

E. and Ruoslahti. E. (1977). Binding of soluble form of surface protein, fibronectin. to collagen. Int. J. Cancer 20,

Gahmberg, C. G., Kiehn, D. and Hakomori. S. (1974). Changes in a surface-labelled galactoprotein and glycolipid concentration in cells transformed by a temperature-sensitive polyoma virus mutant. Nature 248, 413-415. l-loch, S. 0. and McVey. E. (1977). of two major DNA-binding proteins 252, 1881-l 887.

Purification and characterization in human serum. J. Biol. Chem.

Hoch. S. 0.. Longmire, R. L. and Hoch, J. A. (1975). Unique DNAbinding protein in the serum of patients with various neoplasms. Nature 255, 560-562. Hynes, R. 0. (1976). Cell surface proteins mations. Biochim. Biophys. Acta 458, 73-l

and malignant 07.

Hynes, R. 0. and Destree, A. T. (1978). Relationships fibronectin (LETS protein) and actin. Cell 15, 875-886.

transforbetween

Klinman. N. R.. Pickard, A. R.. Sigal. N. H.. Gearhart, P. J.. Metcalf, E. S. and Pierce, S. K. (1976). Assessing B cell diversification by antigen receptor and precursor cell analysis. Ann. Immunol. Inst. Pasteur Cl 2 7, 489-494. Kuusela. P.. Ruoslahti. E. and Vaheri, A. (1975). Polypeptides of a glycoprotein antigen (SF) present in serum and surface of normal but not of transformed chicken fibroblasts. Biochlm. Biophys. Acta 379, 295-303. Laemmli, assembly

U. K. (1970). Cleavage of structural proteins during of the head of bacteriophage T4. Nature 227, 680-685.

Laurefl. C. B. (1972). Electroimmunoassay. Supplement 124, 21-33.

Stand.

J. Lab. Invest.

the 29,

Pena, S. B. J. and Hughes, R. C. (1978). Fibroblast contacts mediated by different forms of fibronectin. Reports 2, 339-344. Peterson, J. B. and McConkey, somal proteins from HeLa cells.

H. E. (1976). J. Biol. Chem.

Richmond, T. J. and Steitz, J. A. (1976). investigated by binding Escherichia coli poly(A.U-MgX). J. Mol. Biol. 703, 25-38.

VerteLiving-

to substratum Cell Biol. Int.

Non-histone chromo257, 548-554.

Protein-DNA lac repressor

interaction protein to

Riggs, A. D., Suzuki, H. and Bourgeois, S. (1970). Lac repressoroperator interaction. I. Equilibrium studies. J. Mol. Biol. 48, 67-83. Ruoslahti. E. and Vaheri, A. (1975). Interaction of soluble fibroblast surface (SF) antigen with fibrinogen and fibrin. Identity with coldinsoluble globulin of human plasma. J. Exp. Med. 747, 497-501. Ruoslahti. E.. Engvall, E., Jalanko, K. and Comings, D. E. (1977). Antigenic differences in nuclear proteins of normal liver and hepatoma. (Identification of nuclear protein present in hepatocytes but absent in hepatoma cells.) J. Exp. Med. 746, 1054-l 067. Thoburn, A. I., Hurvitz, C. and Kunkel. tein in the serum of certain mammalian USA 59,3327-3330. Vaheri. A. and Ruoslahti, duced but not retained Med. 142, 530-535.

H. G. (1973). DNA-binding species. Proc. Nat. Acad.

E. (1975). Fibroblast by viructransformed

pro Sci.

surface antigen prohuman cells. J. Exp.

Vaheri, A. and Mosher, D. F. (1978). High molecular weight, cell surface-associated glycoprotein (fibronectin) lost in malignant transformation. Biochim. Biophys. Acta 576, l-25.

Fibronectin: 657

Yamada. coproteins

a Chromatin-Associated

Protein?

K. M. and Olden, K. (1978). Fibronectins-adhesive of cell surface and blood. Nature 275, 179-184.

gly-

Yamada. K. M.. Yamada, S. S. and Pastan. I. (1976). Cell surface protein partially restores morphology, adhesiveness and contact inhibition of movement of transformed fibroblasts. Proc. Nat. Acad. Sci. USA 73, 1217-1221. Yamada. K. M.. Yamada. S. S. and Pastan. I. (1977). Quantitation of a transformation sensitive, adhesive cell surface glycoprotein. J. Cell Biol. 74, 649-654. Zardi, L. (1975). different chromatin

Chicken fractions.

antichromatin antibodies: specificity Eur. J. Biochem. 55, 231-238.

Zardi, L.. Lin, J. and Baserga, histone chromosomal proteins New Biol. 245, 21 l-21 3.

R. (1973). lmmunospecificity of antichromatin antibodies.

to

to nonNature

Zardi, L.. Siri, A. and Santi, L. (1976). A serum protein associated with chromatin of cultured fibroblasts. Science 791, 869-870. Zardi. L.. Carnemolla. B. and Croce. C. M. (1979). Expression of mouse and human fibronectin in mouse-human somatic cell hybrids. Exp. Cell Res. 118, 435-438. Zardi. L., Lin, J., Petersen, of antibodies to non-histone blasts. Cold Spring Harbor 741.

Ft. 0. and Baserga. R. (1974). Specificity chromosomal proteins of cultured fibroConferences on Cell Proliferation I, 729-

Zardi, L., Siri. A., Ferreri. L. and Santi. L. (1978). between neoplastic diseases and concentration with immunological identity to a DNA associated of Early Life in Man and Animals, Proc. Perugia Cancer Vlfh, 791-801.

Possible correlation of a serum protein protein. In Tumours Quad. Int. Conf. on