Eur. J. Biochem. 146,571 -579 (1985) 0 FEBS 1985

Elution of fibronectin proteolytic fragments from a hydroxyapatite chromatography column A simple procedure for the- purification of fibronectin domains

Lucian0 ZARDI, Barbara CARNEMOLLA, Enrica BALZA, Laura BORSI, Patrizia CASTELLANI, Mattia ROCCO, and Annalisa SIR1 Cell Biology Laboratory, Istituto Nazionale per la Ricerca sul Cancro, Genova

(Received July 9/0ctober 19, 1984) - EJB 840744

Human plasma fibronectin is composed of at least five distinct domains which we refer to as Hep-l/Fib-l, Gel, Cell, Hep-2 and Fib-2 depending on their affinity for heparin (Hep), gelatin (Gel), the cell surface (Cell) or fibrin (Fib). These domains are aligned from the NH2 to the COOH terminus in the above order and can be separated from each other by mild proteolytic digestion.

We have studied the elution of fibronectin thermolysin digest from a hydroxyapatite column using a linear gradient (0.5- 190 mM) of sodium phosphate buffer. The five major fibronectin domains were eluted from the hydroxyapatite chromatography column in the following order: Gel, Fib-2, Cell, Hep-l/Fib-1 and Hep-2. They were identified on the basis of their molecular mass, affinity to different macromolecules and reaction with domain-specific monoclonal antibodies. All domains except the Cell and Hep-2 domains eluted as single homoge- neous peaks. The Cell domain eluted as two different peaks and the Hep-2 domain eluted as four different peaks. This is the first time that heterogeneity of these two domains has been observed.

Since chromatography of a fibronectin thermolysin digest on a hydroxyapatite column provides a good separation of the five major fibronectin domains, we have elaborated a procedure in which each fibronectin domain is purified by no more than two steps: hydroxyapatite and molecular exclusion chromatography. Fractionation of fibronectin proteolytic digest on a hydroxyapatite chromatography column should be of great value in the comparative analysis of fibronectin from different sources and in the study of fibronectin heterogeneity. Its use in combination with molecular exclusion chromatography offers a simple and high-yield method for the purifica- tion of large amounts of fibronectin domains.

Fibronectins are multifunctional glycoproteins present in soluble form in plasma and other body fluids and in insoluble form in the extracellular matrices and basement membranes. The functions attributed to fibronectin, such as cell-cell and cell-substrate adhesion, cell spreading, opsonization of bac- teria and other particulate matter, ability to induce a more normal phenotype in transformed cells and wound healing, are all based on its affinity to cell surfaces and to many different macromolecules including collagens (or gelatin), fibrin, heparin, DNA, actin, complement component Clq and ganglioside (for reviews on distribution, structure and biologi- cal functions of fibronectin see [l - 71).

The human plasma fibronectin molecule is composed of two polypeptides (M and fi chains) with molecular masses of about 220 kDa each. These subunits may have slight differences in sequence and are linked as a dimer via interchain disulfide bonds near the C-terminus. The subunits are composed of a series of discrete structural domains which are connected by flexible peptide segments. These different domains mediate specific interactions with cell surfaces and the above-mentioned macromolecules. Since the interdomain polypeptides are very sensitive to proteolytic enzymes the different domains can be purified, after mild proteolytic diges- tion, by affinity chromatography using the appropriate ligands (see reviews above and references therein).

Recently Sekiguchi et al. [8, 91 have shown that digestion with thermolysin can cleave human plasma fibronectin into

Abbreviations. SDS, sodium dodecyl sulfate; ConA, concanavalin A; PAGE, polyacrylamide gel electrophoresis.

five domains distinct from each other in their biological functions. According to Sekiguchi and Hakomori [9] we refer to these domains as Hep-l/Fib-l (28 kDa), Gel (40 kDa), Cell (1 10 kDa), Hep-2 (38 - 29 kDa) and Fib-2 (20 kDa) depend- ing on their affinity for heparin (Hep), fibrin (Fib), gelatin (Gel) or the cell surface (Cell) respectively. These domains are aligned from the NH2 to the COOH terminus in the above order. Here we have studied the elution of fibronectin domains, obtained by thermolysin digestion, from a hydroxyapatite column. This seems to be a very useful step in the purification of large amounts of the different fibronectin domains.

MATERIALS AND METHODS

Materials

Thermolysin (protease type X), peroxidase, human fibrinogen, thrombin, hirudin and concanavalin A (ConA) were obtained from Sigma (St Louis, MO, USA). Peroxidase- conjugated antibodies were from Dako (Copenhagen, Den- mark). Gelatin was obtained from Difco (Detroit, MI, USA). Hydroxyapatite DNA-grade Bio-Gel, Bio-Gel P 30, Bio-Gel P 60, acrylamide, bisacrylamide, sodium dodecyl sulfate (SDS), and other electrophoresis products were from Bio-Rad (Richmond, CA, USA). AcA44 Ultrogel and Ampholines were purchased from LKB (Bromma, Sweden). CnBr-activat- ed Sepharose 4 B and heparin-Sepharose CL-6 B were from Pharmacia (Uppsala, Sweden). Nitrocellulose was from Schleicher & Schiill (Dassel, FRG). For molecular mass

572

standards in SDS/polyacrylamide gel electrophoresis, we used the low and high-molecular-mass calibration kits from either Pharmacia (Uppsala, Sweden) or Bio-Rad (Richmond, CA, USA). Ultrapure urea was purchased from Serva (Heidelberg, FRG) and was used without further purification. Other chemicals were ‘reagent grade’ reagents from Merck (Darm- stadt, FRG).

Purification of plasma fibronectin and its digestion by thermolysin

Human plasma fibronectin was purified on gelatin- Sepharose from citrated fresh plasma as previously described [lo, 111. Thermolysin digestion of purified fibronectin was performed according to Sekiguchi et al. [El. Purified fibronectin (1 mg/ml) in 25 mM Tris/HCl pH 7.6, containing 0.5 mM EDTA, 50 mM NaCl and 2.5 mM CaClz was digested with 5 pg/ml of thermolysin for 4 h at 22°C. The digestion was terminated by adding EDTA ( 5 mM final concentration) to inactive thermolysin.

Fractionation offibronectin proteolytic fragments by a hydroxyapatite chromatography column

Fibronectin thermolysin digest (30 mg) in about 30 ml of 0.5 mM sodium phosphate buffer pH 6.8 containing 0.1% NaN3 were loaded onto a chromatography column (20 x 1.5 cm) of hydroxyapatite prepared according the manufacturer’s instructions (Bio-Rad) and previously equilibrated with 0.5 mM sodium phosphate buffer pH 6.8 containing 0.1 % NaN3. After washing with 50 ml of the same buffer, elution of fibronectin digest was carried out with a linear gradient of 0.5 - 190 mM sodium phosphate buffer (pH 6.8) (550ml total volume) at room temperature with a flow rate of 15 ml/h. The buffer concentration of eluents was checked by refractive index measurements; 5-ml fractions were collected. After the gradient, the column was washed with a solution of 1.2 M potassium phosphate buffer, pH 6.8 containing 1 M NaCl. In these conditions no other poly- peptides eluted.

Sodium dodecyl suljiate/polyacrylamide gel electrophoresis and isoelectrofocusing

SDS-PAGE gradients (4 - 18%) were performed accord- ing to Laemmli [12]. Samples were reduced with 2% (v/v) 2-mercaptoethanol. The apparent molecular mass was estimated using the following proteins as standars: skeletal muscle myosin, 200 kDa; 1-galactosidase, 116 kDa; phosphorylase b, 94 kDa; bovine serum albumin, 68 kDa; ovalbumin, 43 kDa; carbonic anhydrase, 31 kDa; soybean trypsin inhibitor, 21 kDa; lysozyme, 14 kDa. Isoelectrofocus- ing of fibronectin proteolytic fragments was performed in 6% polyacrylamide according to the LKB Application Note 250, using Multiphor LKB equipment.

Reaction offibronectin proteolytic fragments with ConA and monoclonal antibodies IST-2 and IST-4

The two monoclonal antibodies directed against human plasma fibronectin IST-2 and IST-4 were prepared as reported [13]. IST-2 is specific for the Hep-2 domain whereas IST-4 is directed to the Cell domain [14]. The reaction of the fibronectin proteolytic fragment with IST-2, IST-4 and ConA was determined, after transfer of polypeptides from SDS-

PAGE to nitrocellulose sheets as previously described [ 151, using peroxidase-conjugated rabbit anti-(mouse Ig) for the two monoclonals [15, 161, and the procedure previously described by Hawkes for ConA [17].

Preparation of the Sepharose conjugates

Proteins were coupled to CNBr-activated Sepharose according to the instructions of the manufacturer (Pharmacia). Fibrin-Sepharose was prepared as previously described [18].

Interaction offibronectin proteolytic fragments with heparin, fibrin and gelatin

To study the interactions of the different fibronectin fragments with heparin, fibrin and gelatin, about 50 pg fibronectin fragments in 250 pl of 25 mM Tris buffer, pH 7.2, containing 0.135 M NaCl were added to 100 pl of Sepharose conjugated to heparin, gelatin or fibrin. After slow rotation in an end-over mixer for 3 h at room temperature the conjugated Sepharose was centrifuged, washed four times with 1 ml of 25 mM Tris buffer containing 0.135 M NaCl. The bound polypeptides were eluted with Laemmli’s sample buffer [ 121 and analyzed with polyacrylamide slab gel electrophoresis. Plain Sepharose 4B was used in the same way as the conjugat- ed Sepharose in control experiments.

Molecular exclusion chromatography

Three different kinds of molecular exclusion chromatogra- phy columns were used for the purification of the different fibronectin domains: Bio-Gel P30; Bio-Gel P60 and Ultrogel AcA 44. In all cases columns (100 x 2.5 cm) equilibrated with 10 mM sodium phosphate buffer, 0.15 M NaCl, 1 mM EDTA, 0.02% NaN3, pH 7.2, with a flow rate of 25 ml/h were used. Sample volumes were about 3 - 5 ml and 5-ml fractions were collected.

RESULTS

Fractionation offibronectin thermolysin digest on a hydroxyapatite chromatography column

Fibronectin is composed of at least five functionally distinct biologically active domains which can be separated from each other by mild thermolysin treatment [8,9]. Accord- ing to Sekiguchi and Hakomori [9] we refer to these domains as Hep-l/Fib-l, Gel, Cell, Hep-2 and Fib-2 depending on their affinity for heparin (Hep) fibrin (Fib) gelatin (Gel) or the cell surface (Cell). Fig. 1 shows the model for the domain structure of human plasma fibronectin.

Fig. 2 shows the polypeptide pattern on a 4-18% SDS- PAGE gradient of human plasma fibronectin digested by thermolysin using the conditions described in Materials and Methods. We have identified and numbered 20 major fragments in order of apparent molecular mass. These 20 fragments were constantly present in all the digestions performed. Some of these 20 bands have very similar molec- ular masses and were identified as distinct polypeptides only after fractionation on the hydroxyapatite chromatography column as described in this paper. Fig. 3 shows the elution profile of 30 mg fibronectin thermolysin digest from a

573

Fig. 1. Model of the domain structure of human plasma fibronectin. This model is mainly based on a previous report [9]; the molecular masses (given in kDa) of the different domains are those obtained by digestion of fibronectin by thermolysin as reported in the present paper

hydroxyapatite chromatography column using a 0.5 - 190 mM sodium phosphate linear gradient. All fractions were analyzed on a 4- 18% SDS-PAGE gradient (Fig. 4).

To correlate fibronectin fragments eluting from the hydroxyapatite chromatography column with the major fibronectin domains shown in Fig. 1, the fractions were first divided into 14 pools on the basis of the distribution of the polypeptides observed in Fig. 4. The polypeptides contained in the 14 pools were then analyzed (as described in Materials and Methods) for affinity to gelatin, fibrin, heparin, ConA, reaction with the monoclonal antibody IST-2 (specific for the Hep-2 domain) and reaction with the monoclonal antibody IST-4 (specific for the Cell domain). Fig. 5 shows a 4-18% SDS-PAGE gradient of the pools and Table 1 summarizes the results of the affinity studies. On the basis of the data shown in Table 1 and Fig. 5 we have been able to correlate the major part of the fragments to the different fibronectin domains (Table 2). Using a sodium phosphate linear gradient, fibronectin domains eluted from the hydroxyapatite chromatography column in the following order: Gel (frag- ment 6), Fib-2 (fragment 15), Cell (fragment 3), Hep-l/Fib-1 (fragment lo), Hep-2 (fragments 7 and 9) (Fig. 3). Cell and Hep-2 domains showed some heterogeneity in their elution from the hydroxyapatite chromatography column. The major part of the Cell domain eluted in fractions 53-59 while a minor component with identical immunological and binding characteristics and a slightly higher molecular mass eluted in fractions 61 - 66 (Fig. 3 and 4). We refer to these polypeptides as fragments 3a and 3b (Fig. 5 and Table 1). The Hep-2 domain eluted as four distinct fragments with identical immu- nological and binding characteristics and two different mo- lecular masses (Fig. 3 and 4, Table 1). The fragments eluting in fractions 95-97 (fragment 7a) and 101 -103 (fragment 7b) (Fig. 3, 4, and 5 ) have molecular masses of 38 kDa while the fragments eluting in fractions 101 - 103 (fragment 9a) and 112 - 1 15 (fragment 9 b) have molecular masses of 29 kDa. Since purification and further thermolysin digestion of frag- ment 7 (38 kDa) does not generate fragment 9 (29 kDa) and thermolysin digestion of purified fragments 1 (155 kDa) and 2 (145 kDa) generates fragments 7a and 7b (38 kDa) and fragment 9a and 9b (29 kDa) respectively (unpublished re- sults), we assume that fragments 7a and 7b originate from the heavier subunit (a) of the fibronectin molecule while the two 29-kDa fragments (fragments 9a and 9b) originate from the lighter subunit (p). We have referred to fragments 7a and 7b as domains Hep-21x1 and Hep-2a2, respectively, and to fragments 9a and 9b as domains Hep-2/31 and Hep-2/32, re- spectively (Fig. 1, Table 2).

Fig. 2. Digestion of human plasma fibronectin by thermolysin. Fibronectin (1 mg/ml) in 2.5 mM CaC12, 0.5 mM EDTA, 50 mM NaCl and 25 mM Tris/HCl (pH 7.2) was digested at 22°C with 5 pg/ ml of thermolysin for 4 h. 30 pg fibronectin fragments after reduction by 2% 2-mercaptoethanol were loaded onto a 4-18% SDS-PAGE gradient (T). We have identified and numbered 20 major fragments (f and numbers on the right indicate fragment numbers) in order of apparent molecular mass. Some of these 20 bands have very similar molecular masses and were identified as distinct polypeptides only after fractionation on a hydroxyapatite chromatography column as described in this paper. The values on the left are the molecular masses (in kDa) of the standards (S) as reported in Materials and Methods

Fig. 3. Elution profile of fibronectin thermolysin digest from the hydroxyapatite chromatography column. 30 mg of fibronectin digested as described in Fig. 1 were loaded after dialysis against 0.5 mM sodium phosphate buffer (pH 6.8) onto a chromatography column (1.5 x 20 cm) of hydroxyapatite. Elution of fibronectin fragments was carried out with a linear sodium phosphate gradient as described in Materials and Methods. The elution position of the five major fibronectin domains are also indicated

574

Fig. 4. Analysis on SDS-PAGE of fractions eluted from the hydroxyapatite chromatography column. Aliquots (50 pl) of all fractions of the fibronectin thermolysin digest eluted from the hydroxyapatite chromatography column reported in Fig. 3, after reduction by 2% 2-mercaptoethanol, were analyzed on a 4-18% SDS-PAGE gradient. Numbers above the lanes correspond to the fraction number of the hydroxyapatite chromatography column of Fig. 3. Values on the left are the molecular masses (in kDa) of the standards (S) as reported in Materials and Methods

Pur$ication protocol for fibronectin domains

Fig. 6 shows the protocol for purification of the major fibronectin domains. Each domain was purified with no more than two steps. The strategy for purification was identical for all domains : first hydroxyapatite chromatography, followed by molecular exclusion chromatography.

The purification of the 40-kDa Gel domain (fragment 6) was obtained by Bio-Gel P60 chromatography of pool I1 obtained from the hydroxyapatite column. Fig. 7 shows the elution profile of the Bio-Gel P60 chromatography column; the hatched area indicates the fractions collected containing the purified 40-kDa Gel domain. The inset of Fig. 7 shows an SDS-PAGE of the polypeptides contained in the starting pool I1 and the purified Gel domain. Further thermolysin digestion of the 40-kDa Gel domain gives 30-kDa and 21-kDa major fragments (fragments 8 and 14 respectively) which still retain gelatin-binding capacity (inset of Fig. 7 and Table 1). The

Fig. 7 inset also shows the gelatin-binding fragments purified from the total thermolysin digest by affinity chromatography on gelatin-Sepharose.

The purification of the 20-kDa Fib-2 domain was achieved by Bio-Gel P30 chromatography column of pool IV from the hydroxyapatite column (Fig. 8). The inset of Fig. 8 shows an SDS-PAGE of the total fibronectin digest and of the purified 20-kDa Fib-2 domain (fragment 15).

Fig. 9 shows the elution profile of the hydroxyapatite pool VI from an AcA44 chromatography column for the purification of the 110-kDa Cell domain. The inset shows an SDS-PAGE of total fibronectin thermolysin digest and of the purified Cell domain (fragment 3).

Fig. 10 shows the elution profile of the hydroxyapatite pool X from a Bio-Gel P60 chromatography column for the separation of the 28-kDa Fib-1/Hep-1 domain from the 145 - 155-kDa Cell/Hep-2 domain. The inset of Fig. 10 shows the purified Hep-l/Fib-1 and Cell/Hep-2 domains. Further

575

Fig. 5. Analysis on SDS-PAGE of pooled fractions eluted from the hydroxyapatite chromatography column. Fractions eluted from the hydroxyapatite chromatography column were divided into 14 pools (Roman numerals) and, after reduction by 2% 2-mcrcaptoethanol, 50 p1 were analyzed on a 4- 18% SDS-PAGE gradient. Pool I, fractions 4- 15; pool 11, fractions 29 and 30; pool 111, fraction 31 ; pool IV, fractions 33 and 34; pool V, fractions 40-43; pool VI, fractions 53-57; pool VII, fractions 62-68; pool VIII, fractions 72-75; pool IX, fractions 76-78; pool X, fractions 79-86; pool XI, fractions 94-97; pool XII, fractions 101 - 103; pool XIII, fractions 105- 108; pool XIV, fractions 112- 115. f l , f2, f3 etc. represent the fragment number as in Fig. 2. Values on the left are the molecular masses (in kDa) of the standards (S) as reported in Materials and Methods. T, total thermolysin fibronectin digest

thermolysin digestion of the 28-kDa Hep-1 /Fib-1 domain shows a 23-kDa fragment (fragment 13) as a major cleavage product.

The Hep-2 domains eluted as pure polypeptides directly from the hydoxyapatite chromatography column as shown in Fig. 3, 4 and 5.

DISCUSSION

Chromatography of proteins on hydroxyapatite was origi- nally developed by Tiselius [I91 and extensively studied by Bernardi [20]. Here we have employed hydroxyapatite chromatography to fractionate fibronectin thermolysin fragments. We have utilized thermolysin because it cleaves

fibronectin almost quantitatively into biologically active domains [9].

Using a linear millimolar sodium phosphate gradient the major fibronectin domains elute in the following order (phos- phate concentration in parentheses): Gel (8 mM), Fib-2 (18 mM), Cell (52 mM), Hep-l/Fib-l (90 mM) and Hep-2 (120- 160 mM). Intact fibronectin elutes from the hydroxy- apatite chromatography column at 120 mM sodium phos- phate.

Fibronectin domains were identified on the basis of mo- lecular mass, binding to different macromolecules and reac- tion with domain-specific monoclonal antibodies. Small dif- ferences in the apparent molecular mass of fibronectin domains were observed with respect to those reported by Sekiguchi and Hakomori [9]. These differences may be due to

576

Table 1. Analysis of fibronectin fragments contained in the different pools of fractions eluted from the hydroxyapatite chromatography column (Fig. 3 and 5 ) based on their binding to different macromolecules (heparin, fibrin, gelatin and concanavalin A ) and reaction with domain-specific monoclonal antibodies (IST-4 and IST-2) Affinity studies for macromolecules and antibodies were carried out as described in Materials and Methods. The IST-2 monoclonal antibody is specific for the Hep-2 domain; the IST-4 monoclonal antibody is specific for the Cell domain (see Materials and Methods)

Pool Fractions Fragment Mass Binding activity to number pooled number

HeP Fib Gel IST-4 IST-2 ConA

I

I1

111

IV

V

VI

VII

VIII

IX

X

XI

XI1

XI11

XIV

7-15

29 - 30

31

33 - 34

40 - 43

53 - 57

62 - 68

12 - 15

76 - 78

79-86

93 - 97

101 - 103

105-108

112-115

8 14 18 19

6 8

11 12 14 16 18

6 16

15 16

4 5

15

3a

3b 17

13

1 10 20

1 2

10

7a

7b

9a

9b

kDa

30 21

< 14 < 14

40 30 26 25 21 16

< 14

40 16

20 16

70 50 20

110

110 14

23

155 28

< 14

155 145 28

38

38

29

29

the different gels systems used for the estimation of the appar- ent molecular masses.

,All domains except the Cell and Hep-2 domains elute from the hydroxyapatite column as single homogeneous peaks. The Cell domain elutes as two different peaks with molecular masses of 1 10 kDa and 1 15 kDa. We cannot presently explain this heterogeneity but it may be due to different carbohydrate content, or differences in the polypeptide component. The Hep-2 domain elutes as four distinct peaks with two different molecular masses (38 kDa and 29 kDa). Two different Hep-2 domains obtained by thermolysin digestion and having dif- ferent molecular masses have already been reported [34]. Fur- ther thermolysin digestion of the 38-kDa fragment does not yield the 29-kDa fragment, excluding the possibility that these two fragments represent different stages of proteolytic pro- cessing of the same domain (data not shown). Differences

between a and p subunits of plasma fibronectin have already been reported to be located at the Hep-2/Fib-2 region [18, 211. These two Hep-2 domains (a and p) elute at different concentrations of sodium phosphate (Fig. 5, fragments 7 and 9). Furthermore Hep-2a and Hep-2P each elute as two dif- ferent peaks from the hydroxyapatite chromatography column (Fig. 4, 5 and Table 1; fragments 7a, 7b, 9a and 9b). At present we cannot explain this heterogeneity; however, it has been reported that domains close to the COOH terminus of fibronectin subunits are phosphorylated, probably on the Hep-2 domain 1221. Thus heterogeneity may reflect different phosphorylation levels.

Since hydroxyapatite chromatography of fibronectin thermolysin digests provides good separation of fibronectin fragments, we have set up a protocol for the purification of the different fibronectin domains (Fig. 6). This procedure is

577

Fig. 6. Protocol for purification offibronectin domains obtained by thermolysin digestion. Pools 11, IV, VI and VIII from the hydroxyapatite chromatography column (Fig. 3 and 5 ) were further fractionated by molecular exclusion chromatography as indicated, to obtain purified Gel, Fib-2, Cell and Hep-l/Fib-l domains (for details see Fig. 7-10). The different Hep-2 domains eluted pure from the hydroxyapatite chromatography column (pools XI, XII, XI11 and XIV; see Fig. 3 - 5). The numbers show the molecular mass in kDa

Table 2. Identification of the thermolysin fragments The correlation between the fibronectin thermolysin fragments and the major fibronectin domains (see Fig. 1) was determined using the data shown in Table 1. n.d. = not determined

Fragment Mass Domain number

Iso- electric point

1 2 3ab 4 5 6 7ab 8 9ab

10 11 12 13 14 15 16 17 18 19 20

kDa 155 145 110 70 50 40 38 30 29 28 26 25 23 21 20 16 14

< 14 < 14 < 14

Cell/Hep-2 Cell/Hep-2 Cell unknown sub-component Cell Gel Hep-2a1, Hep-2a2 sub-component Gel Hep-2fl1, Hep-2fl2 Hep- l/Fib-l sub-component Gel sub-component Gel sub-component Hep-l/Fib-l sub-component Gel Fib-2 unknown unknown unknown unknown unknown

5.2-5.4 5.3 - 5.4 4.8 - 5.2 n.d. n.d. 5.2-5.5 8.7 - 8.9 n.d. 9.0-9.2 8.2-8.4 n.d. n.d.

n.d.

n. d.

n.d. n.d. n.d.

8.0-8.1

5.6-5.7

9.0-9.2

based on fractionation of fibronectin fragments by hydroxy- apatite and molecular exclusion chromatography. This pro- tocol seems to present some advantages with respect to pre- vious protocols [8, IS] based on multistep fractionation of fibronectin proteolytic digest on DEAE-cellulose and affinity chromatography. Using the protocol described here, each fibronectin domain can be purified by no more than two steps and the different Hep-2 domains elute as pure polypeptides directly from the hydroxyapatite column. The yield from this procedure is very high: the polypeptide recovery from both hydroxyapatite and molecular exclusion chromatography

Fig. 7. Purification of the 40-kDa Gel domain. Elution profile of hydroxyapatite pool I1 (see Fig. 5 ) from a Bio-Gel P60 chromatogra- phy column. The hatched area indicates the pooled fractions contain- ing the pure 40-kDa Gel domain. Inset: values on the left are the molecular masses (in kDa) of the standards (S) as reported in Ma- terials and Methods. Lane 1 , total fibronectin thermolysin digest; lane 2, hydroxyapatite pool I1 before fractionation on Bio-Gel P60; lane 3, 10 pg of purified 40-kDa Gel domain (hatched area); lane 4, 40-kDa Gel domain further digested by thermolysin, the arrow indicates the thermolysin band; lane 5, gelatin-binding fragments purified from total fibronectin thermolysin digest by affinity chromatography on gelatin-Sepharose

columns is greater than 90%. In contrast, the yield from affinity chromatography is usually low and the procedure often requires denaturing agents. Furthermore the purity of the fibronectin domains obtained using the protocol described here is greater than 90% as estimated by SDS-PAGE.

578

Fig. 10. Purfication of the 28-kDa Fib-l/Hep-1 domain. Elution profile of hydroxyapatite pool VIII (see Fig. 5) from a Bio-Gel P60 chromatography column. The hatched area corresponds to the region of elution of the 145 - 155-kDa Cell/Hep-2 and 28-kDa Fib-1/Hep-I domains. Inset: lane 1, 12 l g of purified 145-155-kDa Cell/Hep-2 domain (hatched area); lane 2, 10 pg of purified 28-kDa Fib-l/Hep-l domain; lane 3, further thermolysin digest of the 28-kDa Fib-1/Hep- 1 domain, the arrow indicates the thermolysin band

Fig. 8. PuriJcation of the 20-kDa Fib-2 domain. Elution profile of hydroxyapatite pool IV (see Fig. 5) from a Bio-Gel P30 chromatogra- phy column. The hatched area indicates the pooled fractions contain- ing the pure 20-kDa Fib-2 domain. Inset: lane 1, total fibronectin thermolysin digest; lane 2, 12 pg of purified 20-kDa Fib-2 domain (hatched area). Values on the right are the molecular masses (in kDa) of the standards (S) as reported in Materials and Methods

In conclusion, the data reported here indicate that fractionation of fibronectin proteolytic digests on hydroxy- apatite chromatography columns should be of great value in (a) the comparative analysis of fibronectin from different sources, (b) the study of fibronectin heterogeneity and (c) in combination with molecular exclusion chromatography, as a simple and high-yield method for the purification of large amounts of fibronectin domains.

This study was partially funded by the Italian Research Council Progetto Finalizzato Oncologia. We thank Mrs Patrizia Mazzini for skilled secretarial assistance.

REFERENCES 1. Akiyama, S. K. & Yamada, K. M. (1983) in Connective Tissue

Diseases (Wagner, B., Fleischmajer, R. & Kaufman, N., eds) pp. 55 - 96, Williams and Wilkens, Baltimore.

2. Yamada, K. M. (1983) Annu. Rev. Biochem. 99,761-799. 3. Hynes, R. 0. & Yamada, K. M. (1982) J. Cell Biol. 95,369- 377. 4. Ruoslahti, E., Engvall, E. & Hayman, E. G. (1981) Collagen Res.

5. Vaheri, A. & Alitalo, K. (1981) in Cellular Controls in Differentia- tion (Reeds, D. & Lloyd, C., eds) pp. 87- 104, Academic Press, New York.

I , 95-128.

6. Mosher, D. F. (1980) Progr. Hemostasis Thromb. 5, 131-151. 7. Mosesson, M. W. & Amrani, D. L. (1980) Blood56, 145-158. 8. Sekiguchi, K., Fukuda, M. & Hakomori, S. (1981) J . Biol. Chem.

9. Sekiguchi, K. & Hakomori, S. (1983) J. Biol. Chem. 258, 3967-

Fig. 9. Purification of 110-kDa Cell domain. Elution profile of hydroxyapatite pool VI (see Fig. 5) from a AcA44 Ultrogel 256,6452-6462. chromatography column. The hatched areacorresponds to the region of elution OF the 110-kDa Cell domain. Inset: values on thc lcft are the molecular masses (in kDa) of the standards (S) as reported in Materials and Methods. Lane 1, total fibronectin thermolysin digest; lane 2, 15 pg of purified 110-kDa Cell domain (hatched area)

3973. 10. Engvall, E. & Ruoslahti, E. (1977) Znt. J . Cancer 20, 3 -5. 11. Zardi, L., Siri, A,, Carnemolla, B., Cosulich, E., Viale, G. &

Santi, L. (1980) J . Zmmunol. Methods 34, 155- 165.

579

12. Laemmli, U. K. (1970) Nature (Lond.) 227, 680-685. 13. Zardi, L., Carnemolla, B., Siri, A., Santi, L. & Acolla, R. S.

14. Sekighuchi, K., Siri, A., Zardi, L. & Hakomori, S. (1983)

15. Towbin, H., Stachelin, T. & Gordon, J. (1979) Proc. Natl Acad.

16. Zardi, L., Cianfriglia, M., Balza, E., Carnemolla, B., Siri, A. &

17. Hawkes, R. (1982) Anal. Biochem. 123, 143-146.

(1980) Znt. J. Cancer 25, 325-329.

Biochem. Biophys. Res. Commun. 116, 534- 540.

Sci. USA 76,4350 - 4354.

Croce, C. M. (1982) EMB0J.-I, 929-933.

18. Hayashi, M. &Yamada, K. M. (1983)J. Biol. Chem. 258,3332-

19. Tiselius, A., Hjerten, S. & Levin, 0. (1956) Arch. Biochem.

20. Bernardi, G. (1971) Methods Enzymol. 22, 325. 21. Sekiguchi, K. & Hakomori, S. (1983) Biochemistry 22, 1415-

22. Ledger, P. W. & Tanzer, M. L. (1982) J . Biol. Chem. 257,3890-

3340.

Biophys. 65, 132-155.

1422.

3895.

L. Zardi, B. Carnemolla, E. Balza, L. Borsi, P. Castellani, M. Rocco, and A. Siri, Laboratorio die Biologia Cellulare, Istituto Nazionale per la Ricerca sul Cancro, Viale Benedetto-XV 10,I-16132 Genova, Italy