THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 257, No. 16, Issue of August 25, pp. 9740-9744, 1982 Printed in U S.A.

Isolation of a Subunit of Laminin and Its Role in Molecular Structure and Tumor Cell Attachment*

(Received for publication, March 15, 1982)

C. Nageswara RaoS, Inger M. K. MarguliesS, Tommie Sue TralkaS, Victor P. Terranova& Joseph A. Madrin, and Lance A. LiottaSII From the +Laboratories of Pathophysiology and Pathology, National Cancer Institute and the §Laboratory of Developmental Biology and Anomalies, National Institute of Dental Research, National Institutes of Health, Bethesda, Maryland 20205 and the TDeDartment of Pathology, Yale University School of Medicine, New Haven, Connecticut 0651 0

.. -

Laminin, the glycoprotein of basement membranes, migrates as two components of 200 kilodaltons (kDa) (a subunit) and 400 kDa (j3 subunit) after reduction on polyacrylamide gel electrophoresis. We have isolated the a subunit and studied its structure by electron microscopy and its function as an attachment factor for tumor cells. Using selective proteolysis of laminin by a-thrombin, the j 3 subunit was removed without any change in the quantity or size of the a subunit. Removal of the j 3 subunit caused a 35-4076 decrease in the total mass of the laminin molecule. The a and p subunits differed by 50-fold in the amount of reducing agent required for complete migration on polyacrylamide gels. By electron microscopy, the whole laminin mole- cule appeared as a “cross” with three identical short arms (37 nm) and one long arm (75 nm). The a subunit examined by electron microscopy was missing the long arm and had no change in the length of the three short arms. This subunit of laminin mediated the attachment of human squamous carcinoma cells to type IV colla- gen. Such attachment properties were lost after pepsin treatment which is known to remove the globular end regions of the short arms. We conclude that the j 3 subunit of laminin is embodied in the long arm of the molecule and that the a subunit consists of three similar chains of 200 kDa. The globular end regions of the laminin short arms are required for the attachment of tumor cells to type IV collagen.

Laminin is a glycoprotein of basement membranes which was first purified by Timpl et al. (1). By sodium dodecyl sulfate-polyacrylamide gel electrophoresis reduced and dena- tured laminin migrates as two major components of 200 kDa’ ( a subunit) and 400 kDa (p subunit). The a and p subunits cut out of the acrylamide gel were similar in amino acid composition ( 2 ) . However, Cooper et al. (3) have presented evidence that the two subunits of laminin are different gene products. Engel et al. (4) have studied the molecular config- uration of laminin using electron microscopy. The whole molecule appears as a cross-shaped structure with three iden- tical short arms and one long arm (4). It has not previously been possible to experimentally study which arms of the

* 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.

( 1 To whom correspondence should be addressed at Building 10, Room 8B19, National Institutes of Health, Bethesda, MD 20205.

’ The abbreviations used are: kDa, kilodalton; P1 fragment, pepsin fragment.

laminin cross contribute to the a and p subunits seen on gel electrophoresis. During our studies of laminin degradation, we noted that a-thrombin selectively digests the ,8 subunit ( 5 ) . We have now extended this finding to isolate the native a subunit of laminin. The purified a subunit was studied by electron microscopy and, as described in this report, is missing the long arm of the laminin “cross.”

Laminin has recently been shown to be an attachment factor for normal and neoplastic cells (6 , 7). We postulated that tumor cells which preferentially use laminin to attach to basement membranes may have a selective advantage in the formation of metastases ( 7 ) . In order to gain insight into the domains of the laminin molecule which participate in its adhesion function, the ability of the purified a subunit to mediate attachment of human squamous carcinoma cells to type IV collagen was compared with the whole laminin mol- ecule or with the P1 fragment of laminin produced by pepsin digestion (8). The pepsin fragment was chosen because its structure had been previously studied by electron microscopy (4) and has been shown to lack both the long arm and the globular end regions of the three short arms of the laminin cross. The isolated a subunit and the P1 fragment were markedly different in their tumor cell attachment activity.

MATERIALS AND METHODS

Laminin Purification-Laminin was extracted from the mouse Engelbreth-Holm-Swarm tumor in 50 mM Tris, 0.5 M NaCI, pH 7.6, purified by DEAE-cellulose and agarose A-5 m column chromatog- raphy as described previously (I) , and stored frozen in phosphate- buffered saline.

Iodination of Larninin-Laminin was iodinated by the lactoper- oxidase method as described previously (9). The reagents and the free label in the reaction mixture were removed by chromatography on Sephadex G-25.

Enzymatic Digestion-Purified human a-thrombin was kindly sup- plied by Dr. John W. Fenton. Digestion with a-thrombin was per- formed at pH 7.6, 25 “C using an enzyme/substrate ratio of 1:100 as described previously (5). The thrombin digestion was arrested by adding a 2-fold excess of hirudin (Sigma). Human plasmin was kindly supplied by Dr. G. Murano, National Institutes of Health. The purity of the thrombin and plasmin was verified as described previously (5). Chymotrypsin and trypsin were obtained from Worthington. Diges- tion with trypsin, chymotrypsin, and plasmin was performed at 37 “C for 150 min. Digestion with plasmin and trypsin was stopped by the addition of a 5-fold excess of soybean trypsin inhibitor. Digestion with chymotrypsin was arrested by adding aprotinin (Sigma). The laminin digestion products were analyzed by slab gel electrophoresis. The P1 fragment of laminin was isolated after pepsin digestion as described previously (8).

Sodium Dodecyl Sulfate-Polyacrylamide Slab Gel Electrophore- sis-This was performed according to the method of Laemmli (10) except that 0.5 M urea was included for band resolution (5). Proteins were directly dissolved in sample buffer, boiled for 3 min, and sepa- rated on 4 or 510 polyacrylamide slab gels. Protein components in the

9740

n L I - I -'":-I 4 I.. .. a Laminin Subunit 9741 LrLul-ucLcr u u L c u r L UI

gel were visualized by Coomassie brilliant blue (H) staining. Radio- active components were visualized by autoradiography with the aid of intensifier screens (Ilul'ont Cronex).

T o examine the effect of reducing conditions on the 400-klla and 200-kDa subunits, 40-pg aliquots of laminin were subjected to sodium dodecyl sulfate-polyacrylamide (47 ) slab gel electrophoresis at differ- ent concentrations of dithiothreitol (Rio-Had). Ilithiothreitol was made up fresh in electrophoresis sample buffer, diluted with the buffer containing laminin, and electrophoresed a t 25 mA/slab. l'he density of components identified by gel electrophoresis was quantified by scanning the gel fluorogram or the negative of a photo of the gel stained with Coomassie blue in a density scanner (Quick Scan, Helena Laboratories) previously calibrated with different known amounts of laminin.

Column Chrornafography-One to two milligrams of tr-thrombin digests of laminin were dialped overnight at 4 "C against 0.2 M ammonium bicarbonate, pH 7.8, and then applied to a 60-cm Toyo Soda Korp. SW 3 O O O high performance liquid chromatography gel filtration column (Altex Scientific Inc., Berkeley, CA) previously equilibrated with 0.2 M ammonium bicarbonate buffer. The column was eluted at a flow rate of 30 ml/h at 600 p.s.i. using a Waters pump (Waters Associates, Milford, MA) equipped with a Model 440 absor- bance detector, and fractions of 1 ml were collected. (b subunit was eluted in the void volume. The protein was assayed by the Bio-Had microassay method using Bradford's reagent ( I 1).

Elecfron Microscopy-This was performed according to the method of Engel et al. (4). Laminin or the purified subunit of laminin derived from ct-thrombin digestion was made to 30 pg/ml in a buffer containing 60"; glycerol and 0.2 M ammonium acetate. pH 7.4, and was sprayed onto freshly cleaved mica discs by using a nebulizer (Ladti Research Industries, Inc.). The samples were shadowed with platinum/palladium followed by carbon in an evacuated chamber (Denton 1)V-502, 5 X 10 ' torr) at angles of 8:l and t+l:j, respectively, with rotation speed at medium setting. The replicas were floated onto distilled water, picked up on 150-mesh grids. and viewed in a I'hilips electron microscope 201 or 400 at 60 kV. The length of the individual molecules was measured with a Hewlett-Packard 9874A Digitizer (Fort Collins. CO).

Cell Affnchrnenf Assa-v-Attachment of A431 human squamous carcinoma cells to type IV collagen was studied using various attach- ment factors with an assay described previously (6, 7). l'he origin of this tumor line has been reported by Ciard et nl. (12).

RESULTS

The density ratio of the two components of laminin identi- fied by slab gel electrophoresis was compared under a series of different reducing conditions (Fig. 1). At very low concen- trations of dithiothreitol (0.1 mM), part of the laminin failed to enter the gel, and the remainder migrated at a 400-kDa (/3 subunit) position with no 200 kDa (a subunit) identified. As the concentration of dithiothreitol was increased from 1.0 to 6.0 mM, the density of the a subunit increased. At concentra- tions of dithiothreitol in excess of 6 mM, the ratio of [ j to a was constant a t a value of approximately 0.55. Further in- creases in dithiothreitol (greater than 10.0 mM) did not cause any change in the ratio. Thus, the a subunit requires at least a 50-fold greater concentration of the reducing agent than the /3 subunit for complete migration in the gel.

Selective Digestion of the /3 Subunit by a-Thrombin-a- Thrombin selectively digested the /3 subunit of laminin. The relative density of each subunit identified by slab gel electro- phoresis was compared after a-thrombin digestion for differ- ent lengths of time (Fig. 2). A total of nine time course digestions was performed. Gel electrophoresis was conducted under optimal reducing conditions. In the time course shown in Fig. 2, the density of the /3 subunit decreased to zero as the time of digestion increased. In contrast, the migration distance and density of the a subunit were not altered by a-thrombin. The densities for the two subunits of laminin following sepa- rate 240-min digestions with a-thrombin are compared on the right side of Fig. 2. There was no statistically significant difference in the mass of the a subunit before and after a- thrombin digestion of the [I subunit. This was confirmed by

5.6 -

5.0 - D Y

5 4.0 -

D

$ 3.0 - 0 a F

n 2.0 -

1.0

0.4

-

- - I .o

200K

_uIII1I1 3.0 6.0 8.0 10.0

mM DTT

FIG. I . Effect of dithiothreitol on the re la t ive dens i ty of t h e 400-kDa (p) and 200-kDa (a) subun i t s of laminin observed by gel electrophoresis. Aliquots of laminin (40 p g ) were heat dena- tured, diluted in sample buffer containing various concentrations of dithiothreitol (DTT). and subjected to sodium dodecyl sulfate-poly- acrylamide (45) gel electrophoresis. The gel was stained with Coo- massie brilliant blue (H) and the hands were scanned in a densitom- eter. The density ratio of the two subunits is compared in the hot groph for different dithiothreitol concentrations. At low dithiothreitol concentrations. the 2OO-kl>a ( ( 1 ) subunit failed t o enter the gel. At optimal dithiothreitol concentrations. the density ratio of [j t o (1

subunits was 0.50-0.55 and the ratio of the /I subunit density to the total densitv was 0.35-0.40.

0 L 5 10 30 0 240 T

lj!. 4MKD hxl 1 OKD

C TIME OF DIGESTION (minutes1

FIG. 2. T ime course digestion of laminin by human n- throm- bin. Aliquots of laminin (40 pg) in phosphate-buffered saline were incubated with n-thrombin at 25 "C at an enzyme/substrate ratio of 1:100 (w/w). At various time intervals, 10 units of hirudin were added to arrest the reaction. The digestion products were analyLed on 5c; sodium dodecyl sulfate-polyacrylamide gels under optimal reducing conditions as determined in Fig. 1. A representative time course and densitometric scan are shown. As time increases, the 400-kDa (p ) subunit is digested with no change in the density or position of the 200-kIla (a) subunit. A, density of the 400-kIla (/I) subunit of control laminin (40 p g ) , 20 separate tracks, mean f S.E: R. density of the 200- kDa (a) subunit of control laminin (40 pg), 20 separate tracks, mean f S.E; and C. density of the 200-kI)a ( ( 1 ) subunit of laminin after digestion of 40 pg of laminin with n-thrombin. 20 separate tracks. mean f S.E. KD, kilodalton.

9742 Characterization of

Coomassie blue staining (20 separate digestions) or by fluo- rography of laminin labeled with '"'I (five separate digestions).

Isolation and Characterization of the u Subunit-High pressure liquid chromatography was used to separate the 1t

subunit of laminin remaining after ct-thrombin digestion from other components in the digestion mixture, such as the en- zyme itself or small fragments of the /j subunit. Purity of the a subunit was verified by 3-127; gradient gel electrophoresis. Structural preservation of the isolated a subunit was studied by protease mapping using trypsin, chymotrypsin, and plas- min. In a method analogous to the V8 protease mapping of laminin performed by others ( 3 1 , the protease cleavage prod-

! 'I I

I . I

A B C D E F FIG. 3 , Protease cleavage pattern of purified a subunit of

laminin is compared with the a subunit of whole undigested laminin. I<nzyme/suhstrate ratio was maintained at 1 5 0 (w/w) . Digestion was performed for I50 min at 37 "C. Iligestion products were analyzed hy sodium dodecyl sulfate-l)olyacr?;lamide (5';) gel electrophoresis. All samples were reduced. A, (1 subunit of whole laminin + chvmotrvpsin; R, isolated (I subunit + chymotrypsin: P, (1

subunit of whole laminin + trypsin: 11. isolated (1 subunit + trypsin; b;, ( I subunit of whole laminin + plasmin; and F, isolated (I subunit + plasnlin.

a Laminin Subunit

ucts of t h e isolated (1 subunit were compared with the products derived from digestion of the 11 subunit of whole laminin (obtained from reduced preparative gel electrophoresis) not treated with ct-thrombin. As shown in Fig. 3 , each enzyme produced characteristic fragments which were virtually iden- tical when the isolated (I subunit was compared with the (k

subunit from whole untreated laminin. When the 1t subunit was subjected to 3.5% gel electrophoresis in the absence of dithiothreitol, it partially entered the gel a s a single band at about 600 kDa (data not shown). Since size estimation can be highly inaccurate for large disulfide-bonded glycoproteins, we proceeded directly to electron microscopic visualization of the u subunit. The intact laminin molecule exhibits a cross-shaped structure identical with that described by Engel et al. (4). T h e laminin cross has three identical short arms and one long arm. All four arms possess globular end regions (Fig. 4) . T h e long arm and the short arms measure 75 nm and 37 nm, respec- tively; in mean length (200 molecules measured) (Fig. 5). T h e isolated 01 subunit of laminin obtained after u-thrombin re- moval of the 13 subunit is missing the long arm of the cross (Fig. 4). The 11 subunit retains three arms which are the same mean length as the short arms of the whole laminin molecule (150 molecules measured) (Fig. 5). A short residual "stump" of the long arm could be recognized in some micrographs of the u subunit.

Tumor Celt Attachment Actiritv of the Subunit-In order to gain information about domains of the laminin molecule participating in its attachment function, whole laminin and the isolated u subunit, before and after pepsin treatment, were compared for their ability to mediate tumor cell attachment. Whole laminin and the 11 subunit both mediated attachment of human squamous carcinoma cells (Table I ) and of human MCF-7 breast carcinoma cells (data not shown) to type I\' collagen. Attachment was enhanced on type IV collagen, as compared with attachment on type I collagen or plastic. In contrast, the 1'1 fragment of whole laminin or of the n subunit inhibited attachment to type IV collagen.

A

B

FIG. 4. Representative examples of whole laminin molecules and the isolated a subunit of laminin. The molecules were rotary shadowed with platinum/palladium. The whole laminin (A ) appears as cross-shaped structure with one long arm (75 nm) and three similar short arms (37 nm). The arms possess globular end regions. The a sub- unit ( B ) is lacking the long arm. X 126, OOO.

Characterization of a Laminin Subunit 9743

"I 20 n A

3 4 3 8 4 2 4 6 5 0 5 4 5 9 6 2 6 6 7 0 7 4 7 8 8 2 8 6 9 0 LENGTH (nml

FIG. 5. A histogram of the lengths of the a rms of the whole laminin molecule or of the isolated a subunit. A , the length o f the long arm o f whole laminin. 200 molecules measured (examples shown in Fig. 4A): H. the length of the identical short arms of whole laminin. 'LOU molecules measured; and C, the length of the arms of isolated cb subunit of laminin, 150 molecules measured (examples shown in Fig. 4W).

1'AHI.E 1 P c v w n t of /urnor cc4ls crtlachrcl a / I20 min

The per cent (mean of quadruplicates range less than 105 of mean) of A431 human squamous carcinoma cells attached to various sub- strates in the presence of different attachment factors is shown. Attachment studies shown here were preceded by factor concentra- tion and by time curves. The plateau of the per cent of attached cells was reached before 120 min. Whole laminin or the o subunit mediated tumor cell attachment arc at virtually identical rates. The pepsin fragment inhibited attachment. Results shown were confirmed at different concentrations of attachment factors and in a separate human tumor cell line (MCF-7 breast carcinoma).

~ -~ ~ "" ~~~ ~~~~~

Suhstrate Attnrhnwnt fartor 'I'ype I\' 'I'ype I

ro~lagen roIIagen "'as''( - ~ ~ ~ .. ".

5 pg/??fl

Whole laminin" 7 3 3 4 48 (t Subunit of laminin" 67 29 43 1'1 fragment from whole laminin" 19 26 24 1'1 fragment from o suhunit" 20 24 18 Fihronectin 42 40 39 Media alone 39 24 24

~~

" Binds to type I\' collagen affinity column. '' Fails t o hind to type IV collagen affinity column.

DISCUSSION

tu-Thrombin selectively removes the p subunit of laminin without altering the size or amount of the u subunit (Fig. 2). Thus, a-thrombin is not simply converting a 400-kDa dimer to a 200-kDa monomer. If that were the case, the density of the u subunit would increase during the digestion shown in Fig. 2. The u subunit isolated after a-thrombin digestion is cleaved similarly by chymotrypsin, trypsin, and plasmin when compared with the CY subunit of untreated laminin (Fig. 3 ) .

We conclude that cr-thrombin degrades the /3 subunit into small fragments with no discernible effect on the amount or on structure of the (1 subunit. The differential effect of dithi- othreitol on the electrophoresis of the two components of laminin (Fig. 1) suggests that the ck subunit has considerablv more disulfide bonds than the [j subunit. These data, there- fore, support the contention of previous investigators ( 3 ) that the two subunits of laminin are different.

The present electron microscopic studies (Figs. 4 and 6 ) indicate that the /I subunit (400 kDa) is embodied in the long a r m of the laminin cross and that the ct subunit is composed of the three short arms of the molecule. Thus, the ct subunit seen on gel electrophoresis with reduction apparently consists of three similar 200-kDa chains which co-migrate to form one hand. Estimation of the molecular weight of the long and short arms of laminin based on electron microscopic measure- ments by Engel et a/. (4) is consistent with this hypothesis.

One of the biologic functions of laminin is to mediate the attachment of cells to basement membranes (6). Our previous studies have indicated that at least some types of highly metastatic tumor cells utilize laminin to adhere to basement membrane type IV collagen (7). Tumor cell attachment stud- ies performed with the ( t subunit and the P1 fragment of laminin provide information concerning the domains of the laminin molecule which may be required for tumor cell at- tachment. The cr subunit of laminin mediates tumor cell attachment to type IV collagen at a rate equivalent to the whole laminin molecule (Table I). Therefore, the /j subunit is apparently not required for adhesion to type IV collagen. T h e P1 fragment of laminin fails to mediate and actually inhibits tumor cell attachment to type IV collagen (Table I) . Engel el al. (4) previously studied the structure of the P1 fragment. It is a three-armed disulfide-bonded cross composed of the three short arms (26 nm) of the laminin molecule with the globular end regions removed. Our gel electrophoresis studies showed that P1 fragment of whole laminin to he identical with the pepsin-derived fragment of the isolated n subunit (data not shown). Therefore, the 1'1 fragment is embodied in the central region of the a subunit. The finding (Table I) that the 11

subunit, hut not the 1'1 fragment, mediates tumor cell attach- ment to type IV collagen leads us to postulate that the globular end regions of the short arms are required for the attachment process. Both the labeled (Y subunit and the 1'1

laminin laminin + a-thrombin

FIG. 6. Model of a-thrombin digestion of laminin. Protease removes the /{ (400 kDa) subunit which is ernbodied in the long arm. The remaining three short arms (each 200 k l h ) are emhodied i n the (1 subunit.

9744 Characterization of a Laminin Subunit

fragment bind to the tumor cell surface (data not shown). However, since only the a subunit mediates attachment to type IV collagen, we can further speculate that the globular end regions of the laminin molecule short arms bind to type IV collagen.

REFERENCES

1. Timpl, H., Rohde, H., Robey, P. G., Rennard, S. I., Foidart, J.-M.,

2. Sakashita, S., and Ruoslahti, E. (1980) Arch. Biochem. Biophys.

3. Cooper, A. R., Kurkininen, M., Taylor, A., and Hogan, B. L. M.

4. Engel, J., Odermott, E., Engel, A., Madri, J . A,, Furthmayr, H.,

and Martin, G. R. (1979) J. Biol. Chem. 254,9933-9937

205, 285-290

(1981) Eur. J. Biochem. 119, 189-197

Rohde, H., and Timpl, R. (1981) J. Mol. Biol. 150,97-120 5. Liotta, L. A,, Goldfarb, R. H., and Terranova, V. P. (1981)

Thromb. Res. 21,663-673 6. Terranova, V. P., Rohrbach, D. H., and Martin, G. R. (1980) Cell

22, 719-726 7. Terranova, V. P., Liotta, L. A,, Russo, R., and Martin, G. R. (1982)

Cancer Res. 42, 2265-2269 8. Rohde, H., Bachinger, H. P., and Timpl, R. (1980) Hoppe-Seyler’s

2. Physiol. Chem. 361, 1651-1660 9. Thorell, J . I., and Johansson, B. G. (1971) Biochim. Biophys. Acta

251,363-369 10. Laemmli, U. K. (1970) Nature (Lond.) 227,680-685 11. Bradford, M. (1976) Anal. Biochem. 72, 248-252 12. Giard, D. J., Aaronson, S. A,, Todaro, G. J., Arnstein, P., Kersey,

J . H., Dosik, H., and Parks, W. P. (1973) J. Natl. Cancer Znst. 51, 1417-1423