THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 263, No. 32, Issue of November 15, pp. 16536-16544,1988 Printed in U.S.A.

Laminin, a Multidomain Protein THE A CHAIN HAS A UNIQUE GLOBULAR DOMAIN AND HOMOLOGY WITH THE BASEMENT MEMBRANE PROTEOGLYCAN AND THE LAMININ B CHAINS*

(Received for publication, June 23, 1988)

Makoto Sasaki, Hynda K. Kleinman, Hans HuberS, Rainer DeutzmannSQ, and Yoshihiko Yamadall From the Laboratory of Developmental Biology and Anomalies, National Institute of Dental Research, Nationnl Institutes of Health, Bethesda, Maryland 20892 and the $Max-Planck-Institut fur Bwchemie, 0-8033, Martinsried, Munich, Federal Republic of Germany

Laminin (Mr = 800,000) is a glycoprotein consisting of three chains, A, B1, and B2, and has diverse biolog- ical activities. Previously we reported the complete primary structure of the B1 and B2 chains of mouse laminin deduced from cDNA sequence (Sasaki, M., Kohno, K., Kato, S., Martin, G. R., and Yamada, Y. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 935-939; Sasaki, M., and Yamada, Y. (1988) J. Biol. Chem. 262, 17111-17117). Here we describe the isolation, char- acterization, and sequence of cDNA clones spanning 9,520 bases which encode the entire A chain of mouse laminin. The nucleotide sequence of the clones contains an open reading frame of 3,084 amino acids including 24 amino acids of a signal peptide. The A chain con- tains some eight distinct domains including a-helices, cysteine-rich repeats and globules. There is consider- able sequence and structural homology between the A chain and the B1 and B2 chains. However, the A chain has a unique globular structure containing homologous repeats at the carboxyl terminus and constituting one third of the molecular mass of the chain. Furthermore, the A chain contains three globules and three cysteine- rich domains at the amino terminus, whereas the B1 and B2 chains have only two each of such domains. The A chain shows homology to the basement mem- brane heparan sulfate proteoglycan core protein and the extracellular domain of the Drosophila neurogenic protein Notch. There is an RGD (Arg-Gly-Asp) se- quence in one of the cysteine-rich domains of the A chain. This potential cell binding sequence could be active as another adhesion signal in addition to the previously identified cell binding sequence YIGSR (Tyr-Ile-Gly-Ser-Arg) of the B1 chain.

Laminin, a glycoprotein localized specifically in basement membranes, is a potent regulator of epithelial cell behavior. Numerous studies indicate that laminin has diverse biological activities which include stimulating adhesion, migration, growth, and differentiation of various cells (reviews in Refs. 1 and 2). Laminin also induces neurite outgrowth (3) and increases the metastatic ability of tumor cells (4). It binds cells through cell surface receptors and interacts with other

* 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.

The nucleotide sequence(s) reported in this paper hos been submitted to the GenBankTM/EMBL Data Bank with accession number($ 504064.

Supported by Grant De 349/1-1 from the Deutsche Forschungs- gemeinschaft.

405, NIDR, NIH, Bethesda, MD 20892. 7l To whom correspondence should be addressed Bldg. 30, Rm.

components of basement membranes including collagen IV, heparan sulfate proteoglycan, and nidogenlentactin.

Laminin is composed of three chains, B1 (Mr = 222,000), B2 (MI = 210,000), and A (M, = 400,000), which are joined by disulfide bonds into a cross-shaped molecule comprising one long and three short arms with globules at each end (5, 6). Studies with proteolytic fragments, domain-specific anti- bodies, and synthetic peptides have identified different re- gions of laminin with biological activity. A major cell binding site was initially identified on the proteolytic fragments, which encompass the portion of the molecule where the three short arms intersect (7). Subsequently, a pentapeptide, YIGSR (Tyr-Ile-Gly-Ser-Arg), from the cysteine-rich domain of the B1 chain was found to be active for cell attachment, receptor binding, and chemotaxis and was able to prevent the metastasis of tumor cells in experimental animals (8-10). Recently, a fragment from the carboxyl half of the long arm of laminin has been shown to have a high affinity cell binding site (11, 12). This region has also been shown to promote neurite outgrowth by various neuronal cells (13,14). Heparin binds to the globule at the end of the long arm as well as to the short arms of laminin (15, 16). Collagen IV binding sites have been assigned to the terminal regions of the short and long arms (17-19).

The complete primary structures of the B1 and B2 chains of both mouse (20-22) and human laminin (23,24) as well as of the B1 chain of Drosophila laminin (25) have been deduced from cDNA sequencing. These results indicate that the B1 and B2 chains have similar structural features with some six different domains. The carboxyl-terminal portions of the B chains are rich in a-helix and presumably form a double or triple coiled-coil structure in the rod-like portion of the long arm. Each short arm is formed by the NHn-terminal portion of one of the chains. In this region, the B1 and B2 chains have two globules separated by two cysteine-rich rod-like domains which consist of many homologous repeats each containing 8 cysteines and approximately 50 amino acids.

Here, we report cDNA clones coding for the entire A chain of mouse laminin. The deduced amino acid sequence reveals that the A chain is also a multi-domain protein containing many structural features in common with the B1 and B2 chains. The A chain, however, has a large globular structure at the carboxyl terminus which is distinct from the B1 and B2 chains.

EXPERIMENTAL PROCEDURES

Peptide Purification and Sequencing-Laminin from EHS’ tumors (26) was reduced with dithiothreitol at 37 ‘C and carboxymethylated

The abbreviations used are: EHS, Engelbreth-Holm-Swarm; PTH, phenylthiohydantoin; kb, kilobase(s1; EGF, epidermal growth factor; BPG, basement membrane heparan sulfate proteoglycan core protein.

16536

Laminin A Chain Sequence TABLE I

Amino-terminal amino acid sequences of tryptic peptides from the A chain

16537

Residues in parentheses indicate uncertain amino acids. Letters underneath the peptide sequences represent residues deduced from cDNA which differ from the protein sequence.

Sequences of tryptic peptides Length Residues

T 1 : A Y Q P Q T S S T N Y N T L I 15 2113-2129 T 2 : A D N E V I C T ( S ) Y Y S K 13 158-171 T 3 : Y T V S Y D I P V E T V 12 572-583 T 4 : A E G L S L Q P Y E E Y F N V V R 17 606-624 T 5 : E A S A A E N P P V R T S K 14 2203-2218 T 6 D L I Y V G G L P H S K 12 2416-2429 T 7 : N S Y G V R 6 2457-2464 T E k Q H I Y A E Y P F W R 11 1193-1205

T 9 Y T V S Y D I P V E T V D S D L M X H A D ( I E M K ) 22 572-595 T l O ( L ) T A T Y Y W A A P ( E ) A Y L G N K 17 546-564

T 1 1 : T I D I S N L Y I G G L P E D K 16 2600-2617 T 1 2 : C S A G Y H G N P R 10 1474-1485 T 1 3 : A E S P H T H S T S A D T N D P I Y V G G Y P A H V K 27 2985-3013

T l 4 Q G L L A V F D A Y D T S D K 14 2382-2397 T 1 5 : T Q E P D N L L F Y L G S S S S S D F L A V E 23 2132-2156 T 1 6 : Y F N S V S E K 8 1279-1288 T 1 7 : A F M T V D G Q E S P S V T V V G N A T T L D V E R 26 2794-2821 T l E k T L E D V L A L X L R 11 2009-2020

P F Y

S

I

0 1 2 3 4 5 6 7 8 9 1 O k b

E E E E E I I I I I

A 4 j * A1171

TAL' ' ATTAAA - JAE3 I

I 1 AAffi I I ' "09

t I I I I I I I

I "01 ' "02 I &a3 : Moi - mix

"12 I "11

FIG. 1. Overlapping cDNA clones for the A chain of laminin. The lines indicate the size and order of the clones. E indicates internal EcoRI sites.

with sodium iodoacetate following standard procedures. The polypep- tide chains were separated on a 3.5% sodium dodecyl sulfate gel (5 mm thick, 200 mm wide), typically loading 20-40 mg. Protein bands were visualized by soaking the gel in 4 M sodium acetate for 15 min. The 400-kDa band of the A chain was excised and electroeluted. After dialysis against 0.4 M sodium acetate, the isolated polypeptide chain was precipitated by the addition of 9 volumes of ethanol. For cleavage with trypsin or endoproteinase Lys-C, the A chain dissolved in 0.2 M sodium bicarbonate was incubated overnight at 37 "C with the re- spective enzyme (enzyme to substrate ratio 1:30). Molecular sieve column chromatography was done on two tandemly arranged TSK 2000SW columns (1 X 60 cm, Toyo Soda, Tokyo, supplied by Beck- man), using 0.2 M ammonium acetate buffer containing 0.1% triflu- oroacetic acid. For reverse-phase separations, Vydac C I ~ columns (201TP54, Separations Group) were used at 55 "C with gradients of 0.1% trifluoroacetic acid (solvent A) against 70% acetonitrile in 0.1% trifluoroacetic acid (solvent B). Amino acid sequences were deter- mined with 0.1-2-nmol samples on a gas-phase (Model 470A, Applied Biosystems) or a spinning cup liquid-phase Sequenator (Model 890, Beckman). The gas-phase sequenator was equipped with an on-line PTH-derivative analysis system built from a high performance liquid chromatography micro pump (Model G, Brownlee) and a Spectroflow 773-UV detector (Kratos). The separation of PTH-derivatives was carried out by gradient elution on a PTH Cla microbore column (2.1 X 220 mm, Applied Biosystems) following the manufacturer's instruc- tions. PTH-derivatives from the spinning cup sequenator were iden- tified on a Supershere RP-8 column (4 X 250 mm, Merck) using an isocratic high performance liquid chromatography system.

Construction and Screening of cDNA Libraries-Three different types of cDNA libraries in a Xgtll vector were constructed from two RNA sources. Poly(A)+ RNA prepared from the EHS tumor was fractionated on a sucrose density gradient, and fractions larger than

28 S were pooled as described (22). Differentiated F9 cell poly(A)+ RNA was prepared from F9 cells after treatment with retinoic acid and &butyryl-CAMP for 5 days (21) and used as a template without size fractionation. Oligo(dT)-primed cDNA synthesis was carried out as described (22). Double-stranded cDNA was blunt-ended with T4 DNA polymerase, treated with EcoRI methylase, ligated to EcoRI linkers, and the linkers were digested with EcoRI (21). The cDNA was size-fractionated by 1% agarose gel electrophoresis, and fractions larger than 1 kb were used to ligate into EcoRI-cut Xgtll DNA (Vector Cloning System). Three different specific primed cDNA libraries were constructed from differentiated F9 cell poly(A)+ RNA as described (22). These oligonucleotides were 5'-dCTGCCCTGTGGACCAG- GTCACGGGC, 5'-dCTCGACAGCTTTCCGGCCAACCTCC, and 5'-dGGTCACTGTCCACCGTCTCCACTGG, which are complemen- tary to the sequences at nucleotide positions from 5662 to 5686, from 4089 to 4113, and from 1885 to 1879 as underlined in Fig. 2.

The EHS cDNA library was screened with antibodies to denatured laminin as described (22). The oligo(dT)-primed F9 cDNA library was screened with a 32P-labeled cDNA insert from XA-E3 (Fig. 1). The specific oligonucleotide cDNA libraries were screened with 32P- labeled probes from the most 5' portion of the previous cDNA clones in each screening step.

DNA Sequence and Analysis-DNA sequence was determined by the dideoxy chain termination method of Sanger et al. (27) with a "shotgun" random fragment strategy. Direct sequencing of DNA in Xgtll was performed using forward and reverse primers as described by the supplier (New England Biolabs). The nucleotide sequences were analyzed by the programs of Staden (28), IDEAS (29), and Protein Sequence Query (PSQ, Protein Identification Resource, Na- tional Biomedical Research Foundation). Internal homology of the A chain and homology to the B1 and B2 chains were analyzed by DOTMATRIX (PSQ), SEQHP, and SEQDP (IDEAS). FASTA de- veloped by Pearson and Lipman (30) was used for a homology search of the National Biomedical Research Foundation (NBRF) protein data base. Protein structure was analyzed by DELPHI, ALOM, CHOFAS, and HPLOT of IDEAS.

RESULTS

Isolation and Sequencing of Peptides from the A Chain of Laminin

Laminin prepared from EHS tumor was reduced and sub- jected to preparative slab gel electrophoresis. The A chain protein (Mr = 400,000) was excised, extracted, and subjected to proteolysis. For preparative isolation of peptides, the A chain was first cleaved with the lysine-specific protease en- doproteinase Lys-C, and the resulting fragments were par-

Laminin A Chain Sequence 16539 6000

6 1 2 0

6 2 4 0

6 3 6 0

S l 8 0

6 6 0 0

6 7 2 0

6 8 4 0

6960

7 0 8 0

7 2 0 0

7 3 2 0

7 4 4 0

7 5 6 0

7 6 8 0

7 8 0 0

7 9 2 0

8 0 4 0

8 1 6 0

8 2 8 0

8 4 0 0

8 5 2 0

8 6 4 0

8 7 6 0

8 8 8 0

9 0 0 0

9 1 2 0

9 2 1 0

9 3 6 0

9 4 8 0 9 5 2 0

tially separated by reverse-phase chromatography on a Vydac CIS column. The collected pools were then further digested with trypsin and separated by molecular sieve and reverse- phase chromatography. Although a large number of peptides were obtained, only 18 have been sequenced (Table I). As described below, the occurrences of these peptides in the cDNA sequence allowed the definitive identification of A chain cDNA clones.

Isolation and Characterization of Overlapping cDNA Clones Three different types of cDNA libraries were screened to

obtain a series of overlapping clones coding for the entire A chain of mouse laminin (Fig. 1). Initially, an oligo(dT)-primed cDNA library from EHS tumor cell poly(A)+ RNA was pre- pared in a Xgtll expression vector and was screened with polyclonal antibody to denatured laminin. A dozen clones out of several hundred positives were picked randomly and puri- fied. These clones were characterized first by Northern analy- sis to RNA prepared from tetratocarcinoma F9 cells grown under standard conditions following treatment with retinoic acid and dibutyryl-CAMP, which induces a 20-fold increase in the synthesis of laminin and collagen IV. All clones hybridized strongly to RNA from differentiated F9 cells but not from undifferentiated F9 cells. Northern analysis indicated that there were three different classes of cDNA. Class I and I1

clones hybridized to 6.5 and 8.0-kb RNA transcripts, respec- tively (data not shown). The third group hybridized to an mRNA transcript of 10 kb. DNA sequence analysis revealed that the class I and I1 clones coded for the B1 and B2 chains. The amino acid sequence deduced from one of the class I11 clones (XA-E3) matched the partial amino acid sequence of the proteolytic peptide T15 (Table I) of the mouse A chain, as described above. This clone was subsequently used as a probe to screen a newly prepared oligo(dT)-primed cDNA from differentiated F9 cell poly(A)+ RNA. Twelve positive clones were purified and characterized, including XA-06 and "09 (Fig. 1).

The remaining sequence of the A chain was obtained by isolating overlapping cDNA clones from libraries constructed by specific primer extension. Oligonucleotides 25 bases long were synthesized complementary to the 5' region of each successive clone and were used as primers to prepare new cDNA libraries from differentiated F9 cell poly(A)+ RNA. The libraries were screened by plaque hybridization using a restriction fragment of the 5' portion of the previous clone. XA-01, -02, -03, -04, -08, -11, and -12 are representative of multiple isolates obtained at each successive cloning step (Fig. 1).

The clones were first examined by sequencing both ends, and then their authenticity was checked by S1-protection

16540 Laminin A Chain Sequence

analysis. Using these approaches, we obtained a series of overlapping clones which extend over 9.5 kb of contiguous DNA. Primer extension analysis revealed that the most 5’ clone, XA-12, was missing 80 nucleotides at the 5‘ end of mRNA (data not shown). The largest clones obtained at each screening step were sequenced by the random fragment method.

Multi-domain Structure of the A Chain

cDNA Sequence-The complete 9520-base pair sequence of the A chain cDNA and the deduced amino acid sequence were determined (Fig. 2). There is an open reading frame of 3084 residues starting with a methionine followed by a stretch of 24 mostly hydrophobic amino acids whose sequence is char- acteristic of a signal peptide. Although the amino-terminal sequence of the A chain is not known, a signal peptidase cleavage site was presumed according to the “-3, -1” rule of von Heijne (31), and the most amino-terminal residue, glu- tamine, was numbered +l. Following the large open reading frame, there is about 200 base pairs of a 3”untranslated region which contains a polyadenylation signal, ATTAAA, followed by the poly(A) sequence. The A chain contains 163 cysteines and 46 Asn-X-Thr and Am-X-Ser sequences as potential N-linked oligosaccharide attachment sites.

Analysis of the deduced amino acid sequence suggests that the A chain contains some eight structural domains which

A

a$ IVb

FIG. 3. Structural model of laminin. The Roman numbers des- ignate the general regions of the domains in each chain. G designates the A chain terminal domain.

show sequence homology to the B1 and B2 chains. The A chain, however, has an additional carboxyl-terminal domain which is largely globular. The domain structure of the A chain together with those of the B chains is schematically illustrated in Fig. 3. The A chain domains are numbered according to the format used previously for the B1 and B2 chains and the large carboxyl-terminal domain is designated G.

The Carboxyl-terminal Globular Domain-Domain G (resi- dues 2,110-3,060) has 951 residues (Mr = 103,986). Computer- assisted structural analysis suggests that the domain forms a globular structure agreeing with the image of the rotary- shadowed laminin observed in the electron microscope. Com- puter analysis by the program SEQHP indicates that domain G has an internal repeating structure made up of five tandem repeats (Fig. 4). Each basic repeat contains approximately 180 amino acids. Each repeat contains glycines at regular intervals, in which 25 out of 86 glycines are completely con- served. The location of hydrophobic and charged residues is also highly conserved. There are 15 cysteines in the domain of which 10 are aligned at the same positions in the repeat. Analysis with the computer program ALOM indicated a po- tential transmembrane segment at residues 2,313-2,329, but there is no evidence that it is functional.

a-Helical Domains-Domains I and I1 (residues 1,538- 2,109) have 571 amino acids (M, = 64,025) and are predomi- nantly a-helical. Both domains have almost the same struc- tural feature, but were numbered I and I1 so that the relative positioning of domains is consistent among the three chains. Similar to the corresponding domains of the B1 and B2 chains, domains I and I1 of the A chain contain a series of heptad repeats in which hydrophobic amino acids are located at regular intervals as shown in Fig. 5. These heptad repeats are characteristic of proteins containing coiled-coil structure such as fibrinogen (32) and myosin (33). The size of domains I and I1 of the A chain is very similar to those in the B1 and B2 chains. The domains of the A chain, however, lack the cluster of polar amino acids present in domain I of the B1 and B2 chains. The A chain also contains seven prolines which could interrupt the a-helix.

Cysteine-rich Domains-Domains IIIa (Mr = 20,329, resi- dues 1,345-1,537), IIIb ( A I r = 47,091, residues 692-1,142), and V (Mr = 27,020, residues 252-495) are rich in cysteines and glycines which create many turns and contain many homol- ogous repeats. The alignment of these sequences is shown in Fig. 6. Each repeat is composed of a 50-amino acid unit with eight cysteines. The distance between the first, second, and eighth cysteines, and between the fifth, sixth, and seventh cysteines is completely conserved. These repeats are of the same type as found in the corresponding domains I11 and V of the B1 and B2 chains. Domain IIIb at 1,123-1,125 contains an RGD (Arg-Gly-Asp) sequence which is a potential cell attachment site. Twenty-two out of about 50 amino acid positions are highly conserved, allowing a consensus sequence to be deduced (Fig. 6).

Amino-terminal Globular Domains-Domains IVa (Mr =

Laminin A Chain Sequence 16541

FIG. 5. Heptad repeats in domains I and I1 of the A chain. The sequence (residues 1563-2106) of domains I and I1 of the A chain are aligned so that hydro- phobic residues (shaded) occur predom- inantly in positions a and d and form a hydrophobic interaction edge.

FIG. 6. Alignment of cysteine- rich repeats in domains IIIa, IIIb, and V of the A chain. Sequences, in- cluding residues 251-495 in domain V, residues 692-1142 in domain IIIb, and residues 1345-1537 in domain IIIa, are aligned by the program SEQHP, with some gaps introduced by visual inspec- tion. Cysteines are shaded. The numbers at the right end in each line represent residue numbers. A consensus sequence is shown on the bottom line.

1 2 3 4 5 6 7 s

1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 s 1 9

1 2 3 4 5

i 5

7 8 9

1 0 1 1

2 3 4

d e f g a b c d e f g a b c d e f g a b c d e f g a b c

L T G V S @ A @ Y G l L E N L E N T T K Y F Q E N A K K I R I E I Q t E G I A E Q T E N L Q V L A R H Q K V N A E Y E R T S N G T Q A L A L H A N I K E I T E K V A T L N Q T A R K D F A L Q S U H Q N I S S L L C L I K E R N F T E

A N S L L S N H S E K L Q A A E E L L K E A G E L K A A K D L L S R I Q K R F Q K @ Q E K L

S N L L L L L V K A N L K E E F Q E K K L R V T S E L I A K G R E W V O A A G T H T A A A Q L E H H R D E L L L W A R K I R S H V D D L V A R D L V H R A E Q H A S E L Q S R A G A L D V S L N A T S A A H V H S N I Q T L T E E A E A H K T A N K T D L I S E S L A S R G K A V L r L K E S V G T R R K Q Q G l T M K L D E t K F Q E S V O N I T K Q A N D S L A M L R E S @ G R K A R E L A A A A N E S A V K T L E D V L V F N T S E D L S R V N A T V Q E T N D L C H A G R K M K D M E M Q A N L L L D R L K @ L K L S R N L S E I K L L i S R A R K O A A S l K

R K T Q M K S Q D M R M Q N G A N T V

Y E 8 K E T Q D L R L G L S L A

L L t T E L M L A S T u S T

V E

T I

E V N K Q Q T S E A

K

Q R

S A E E N a K N D R Q E R T N A

d e f g a b c d e f g a b c d e f g a b c d e f g a b c

T

V

R V

K

T

1

R R

1

1 5 9 0 1 6 1 8 1 6 4 6 1 6 7 4 1 7 0 4 1 7 3 2 1 7 6 0 1 7 8 9 1 8 1 7 1 8 4 7 1 8 7 8 1 9 0 6 1 9 3 4 1 9 6 2

2 0 1 9 1 9 9 1

2 0 5 0 2 0 7 8 2 1 0 6

Y C I C Y G H A S S E C N C H N K A K D P C N C D P V C S L S S V C I K D D R H A D L A N C K W P C Q C P C R U C Y A C D K C D R C Q F C Y R C F P N P C D C R T V G S L N E D P C I E C F C G V S G V

C P C Y Y D S S V A K E R R S L N T A G Q Y S C C G V C V N C S ~ N T T C I N C E T C I D ~ Y Y R P H K V S P Y D D M P

W D E E A K Q L Q C Q C E H N T C C E S C D R C C P G Y H Q Q P W R P G T I S S C N E C E 3 0 8 C R 3 7 8

E P C L C K K N V E G U N C D R C K P G F Y N L K E R N P E C 8: ::: C 4 9 5

P C E C H G M A S E C D P C A C P L S I D S N N F S P T C H

A C D C H E N G S L S G V C H P C N C S G N V D P L E A C H C D

P C N C S V E G S V S D N C T P C D C A M T Q N N C D A C N C S A V G S T S A Q C D P C C C D L R G T L P D T C D

P C F C F G L S Q L c s C

P C N C N N H S D V C D P C T C P H H P P F S F S P T C Y T C D C N P Q C S V H S D C D

ca 7 2 3 ca 1 7 2 c v 130 C R 113 cv 932 C T 9 7 9

C V 1 0 7 1

cs 1 1 3 1 1 1 0 2

1 1 4 2

ca 1 0 2 5

P E T C K C K S C R D H T S C D H C E L C A S C Y Y G K V T G L P G D V E G D S D F R C N A C L P G Y E C Q Y C E R C S A G Y H G N P R A A C G S

R A S C Q C V C K P C A T G L H C E K C L P R H I L Y E S O

C V C P P C T A G H S C P D C A P C Y Y R E K L P E S G C R C P R P L L A P C V 1 3 1 4 C T 1 4 3 3

C V 1 5 3 7 c a 1 4 9 0

A [ l V b l V T D L Y U T - N - K I R S O P D V L G - G H R O l S l N N T A V U O R L - T S T Y V W A A P E A Y L C N K L T A F C 4 F L K V T V S - Y D l P V 5 8 0 A [ I V o I L L H V V S ~ S N L K G T I E G - V H F O P P D T L L D A - E A V R O H l Y A E P F Y W R L P K O F O ~ D Q L L A V 4 9 K L ~ Y S V A F Y S T L G 1 2 3 1 8 2 1 I V 1 D I - - A V I S D S V F - P R Y F I A P V K F L ~ N O V L S Y C O N L S F S F R V - D R R D 5 5 2

A C l V b l E T V D S D L Y S H A D I I l U O N ~ 1 " - - - - T i S T R A E C L S L O P Y E E Y F N V V ~ L V P S N F R O - F N T - R - R E l D R - O O L ~ l 6 4 4 A C I V a 1 - 7 C T ~ N Y ~ P P V - - L l ~ ~ - Q - R A R K H V I Y U D A P A ~ E N C V R Q D Y ~ - - V ~ Y - K ~ E F W K V F N S V S E K H V T H S D - F U S 1 2 9 5 E 2 [ I V 1 - T R L S A - ~ D L V - - L E - Q A O i ~ V S V P L ! A ~ C N S Y C I E T T V K - Y l - - F R ~ - H ~ A T - D Y - P W R P A - - L S P F E - F ~ K 6 1 1

A C I V b l Y L A ~ V T H L L l l t * N Y N S A K Y A L Y ~ i ~ S ~ S L D I I S C N A I D L - ~ V - - A A A C I V o 1 Y L S ~ I D Y i L f K A S Y C P G - L ~ O $ U l A N l S Y E V G ~ - K A V E L ~ A E C ~ ~ A

6 8 7

E 2 C I V l L L N R L T S I K 1 R C T Y - S E R T A C Y - L D D V T C P S f a * G P G - V ~ A T 1 3 3 9

6 5 0

FIG. 7. Homology in domains IVa and IVb of the A chain and domain IV of the B2 chain. Sequence of residues 513-687 of domain IVa of the A chain, residues 1161-1339 of domain 1% of the A chain, and residues 511-650 of domain IV of the B2 chains are aligned by the program SEQHP. Conserved residues are shaded.

8 2 8 1 ~ E P E F S Y G C A - E G - - - - S C Y - P A T G D L L I C R A ~ K L S V - T S T C ~ L H K - P ~ P Y C I - V S ~ L - - - O E - D K K - C F I C D S R D 6 2

A Y D E C A D E G C R P O R C ~ - P - E F V U A A F N - V T V - ~ A ~ N T C Q T P - - P E E Y C - - V Q l C V T C - - V T K S - C H L C D A G Q 6 1

Q O R G L F P A - I L N t A T N A - H I S - ~ N A T C ~ E - U G P E Y F C K L V E ~ - V P G R P V R H A Q C R V C ~ - C N 5 5 FIG. 8. Homology of domain VI in A

the A, B1, and B2 chains. Sequences 82 Q - H - - L Q H G A A F L - - - - - T D Y N - U O A D T T - Y * Q S O T Y L A G V ~ Y P N S I N L T L H L G K A - F D I T Y V R L K - - F H T S R P E S 1 2 4

ofresidues1-251oftheAchain,residues :' ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ i?? 1-248 of the B1 chain, and residues 1- 249 ofthe B2 chain are alignedby the E 2 F A I Y K f l T R E D - G P - - W I P Y O Y Y - - S G S - C E N T - Y S K A N - R Q F I R ~ G G - D - ~ O O A L C T D E F - S D I S P L T - G G N V A F - 1 8 7

program SEQHP. Conserved amino A W - I L E R S V - D - C V K F K - P W ~ Y Y ~ V S O T E C L - T R Y K l T P R R ~ P P - T Y R A O N E - - V l C T - S Y Y S K L V P L - E H G ~ - l H l 1 8 2 E 1 ~ L I - E ~ S S - D F G - ~ T W G V V R Y F I Y - D - " - ~ P U K - K V - D D - - - l l ~ - ~ ~ R Y S D l E P S T ~ - G E V l ~ R 1 8 7

8 1 - A L D - - - P - ~ F K I E D P Y S P R I ~ - N L L K I ~ N L R I K F V K ~ H T L O - ~ N L L D - ~ - R U - E I R E K - - - - - V ~ ~ A V Y D M V V R C 248 82 S T L E - E R P S A Y N F - D N - S P V L O E W V T A - T D ! R V T l N R L N T F 6 - D ~ V F N E P K - V l K S - - - - - - - - ~ Y ~ A I S D F A V C G 249

A S - L I N G R P S I D - - - D P - $ P O l L E F T S A R Y - t R L R L Q ~ l R T L N A ~ - L M l L $ H f l D L R D L D P l V T R R Y Y ~ S l K D l S V G G 2 5 1

acids are shaded.

22,742, residues 1,143-1,344), IVb (Mr = 21,968, residues 496- 691), and VI (Mr = 28,695, residues 1-251) contain mixtures of a-helix, p-sheet, and random coil and are likely to form the globules which are clearly evident in electron micrographs of laminin. There is a significant homology (about 25% identity) between IVa and IVb and between these domains and domain IV of the B2 chain. An alignment of these sequences is shown in Fig. 7. Domain VI does not share sequence homology to domains IVa and IVb. There are no cysteines in domains IVa and IVb, whereas the most amino-terminal domain, domain VI, contains 15 cysteines. Domain VI of the A chain is highly homologous to domain VI of the B1 and B2 chains and an alignment of these sequences is shown in Fig. 8.

Sequence Homology in the Three Chains Sequence homology between the A chain and the B chains

and within the A chain was analyzed by the DOTMATRIX program (Fig. 9). The internal homologies of the A chain shown by diagonal lines are apparent. Domain G contains 5 relatively large repeats and is unique to the A chain. The a-

helical domains I and I1 show numerous short homologies because of the heptad repeat. The cysteine-rich domains, IIIa, IIIb, and V, show homologous sequences within the domains as well as between them, and are very similar to domains I11 and V of the B1 and B2 chains. The globular domains IVa and IVb are homologous to each other and to domain IV of the B2 chain but not to that of the B1 chain. The sequence of the terminal domain VI is also significantly conserved in all three chains.

DISCUSSION

Laminin is one of the key extracellular matrix components which influence tissue development and function. As one might expect from an extraordinary large and complex mole- cule, different portions of laminin have been shown to be active in various biological functions. Together with the pre- viously determined sequence of the B1 and B2 chains, the primary structure of the A chain described in this paper allows one to predict for the first time the alignment of the three chains, as well as to provide insight into the detailed structure of the molecule.

16542 Laminin A Chain Sequence

FIG. 9. Dot-matrix analysis of the amino acid sequence of the laminin chains. Sequence homology was analyzed by the pro- gram DOTMATRIX (PSQ) with a window length of 25 and a score of 25 or greater. Points at which the two sequences are homologous are indicated by dots. At the top, the amino acid sequence of the A chain is compared to that of the B1 chain. In the center, the amino acid sequence of the A chain is compared to that of the B2 chain. At the bottom, the amino acid sequence of the A chain is compared to itself.

The amino two-thirds of the A chain has a significant sequence homology to that of the B1 and B2 chains. There are, however, some structural differences between this region of the A and the B chains. Whereas the B1 and B2 chains have only two globules and two cysteine-rich domains, this portion of the A chain has three globules, domain IVa, IVb, and VI, which are separated by three cysteine-rich domains: IIIa, IIIb, and V. This is in contrast to electron microscopic observations on laminin which revealed only two globular domains in each short arm. A possible explanation could be that the third globule is in a near central position and might therefore have been overlooked. The A chain globules, domain IVa and IVb, are homologous to each other and are also homologous to domain IV of the B2 chain. The region encom- passing domains IIIa to IVb of the A chain is likely to have evolved by duplication of a sequence containing a globule fused to a rod structure. The most amino-terminal globule, domain VI of the A chain, has a significant sequence homology

with the B1 and B2 chains (about 30% identity) as shown in Fig. 8 but is distinct from the other globular domains. The high degree of sequence conservation of this domain in the three chains suggests a common function. Indeed, the ends of short arm are thought to be involved in collagen IV binding (17).

The a-helical domains, domains I and I1 of the three chains, could form a triple-stranded coiled-coil structure since the total number of residues in those domains is almost the same in all chains. These regions likely constitute the rod-like portion of the long arm of the molecule because the length (85 nm) of domains I and I1 calculated assuming a distance of 0.15 nm/residue in a-helix corresponds well to the electron microscopic measurement (77 nm) of this part of rotary- shadowed laminin. The heptad repeats in domain I1 of the A chain are less periodic than in domain I, similar to the B1 and B2 chains. This region of the A chain also contains five prolines which break a-helix. These features suggest that the amino-terminal portion of the rod-like region of the long arm forms less rigid interactions between chains compared with the carboxyl-terminal part. The interaction between the three chains is likely to be further stabilized by disulfide bonds at both boundaries of the rod-like domain. Two cysteines from domain I of the B1 and B2 chains are involved in the forma- tion of intermolecular disulfide bonds (34). At the center of the cross, six cysteines, two donated from each chain, could also form disulfide bonds to stabilize the structure. It is of interest to note that there is an interruption in the rod-like region of the long arm by domain ct in the B1 chain. Domain a is about 30 residues long and contains six cysteines, eight glycines and three prolines which generate many turns (21, 23). The B2 and A chains do not have a corresponding domain. Domain a might be a site of interaction with other basement membrane proteins, or with other laminin molecules to form dimeric or multimeric proteins.

The A chain sequence data allowed an exact localization of laminin fragment E8 which has been shown to stimulate neurite outgrowth and cell attachment (13,l l) . Fragment E8 is a 35-nm rod-like structure with a globule derived from the carboxyl half of the long arm. In agreement with the model for the laminin structure, it contains segments of all three chains. Amino-terminal sequence analysis2 revealed that frag- ment E8 contains the 226 and 246 residues of the carboxyl- terminal portions of the B1 and B2 chains, respectively. The A chain starts at position 1886 (Fig. 2) and should therefore comprise the terminal 223 amino acids of the a-helical region, in good agreement with the length of the B chain segments.

The cysteine-rich domains show a strong homology in all three chains. Eight cysteines are positioned at regular inter- vals and the distances between the cysteines which do not have gaps in the repeats are strictly conserved in all chains. Furthermore, the consensus sequence of the cysteine-rich domains of the A chain is the same as that of the B1 and B2 chains. Glycines as well as cysteines, which generate many turns, are highly conserved suggesting their role in forming a rod-like structure. Interestingly, each cysteine-rich domain has a different number of homologous repeats. Such different spacing by these domains may indicate that each short arm has a different functional role.

The carboxyl-terminal globule, domain G, comprises one- third of the molecular mass of the A chain and has a unique structure different from other regions of the molecule. The globule has been identified as one of the binding sites for collagen IV (18), heparin (35), and basement membrane hep-

* R. Deutzmann, J. Huber, K. A. Schmetz, I. Oderbaumer, and L. Hartl, submitted for publication.

Laminin A Chain Sequence

TABLE I1 Statistical analysis of domain ZV-like regions

16543

Similarity of domains IVa and IVb of the laminin A chain compared with domain IV of the B2 chain and with globular domains from the basement membrane proteoglycan in a region similar to the short arm regions of the laminin chains. Residue numbers for each domain compared are for the A chain 1143-1344 (IVa) and 496-691 (IVb), for the B2 chain 503-687 (IV) and for BPG5 1-186 (Gl) and 394-590 (G2). Analysis was performed using the IDEAS program SEQDP with a gap penalty of 8 and 100 random alignments. The score is determined for the best alignment using the mutation data matrix. This comparison weights individual amino acids and accounts for conservative or radical changes and sums these to give a score. The difference of the optimal alignment score from the scores of 100 random alignments, given in standard deviation units from the mean. Comparisons with domain IV of the laminin B1 chain indicated no similarity of this domain with those above.

Laminin (score (S.D.)) BPG (score (S.D.))

IVa IVb G1 G2

Laminin A IVa -192 (14.5) -211 (20.3) -232 (20.0) IVb -192 (16.2) -194 (20.7) -209 (20.6)

IV -80 (5.7) -113 (9.7) -98 (10.5) -86 (5.6) Laminin B2

aran sulfate proteoglycan (36). Heparan sulfate proteoglycan binding is thought to be due largely to interactions with basic amino acids because a moderate concentration of salt disrupts its binding. Near the end of the carboxyl terminus, there is a 21-amino acid segment (residues 3010-3032) which contains six positively charged amino acids without interruption by negatively charged amino acids. This segment could partici- pate in heparin binding.

A computer search of the National Biomedical Research Foundation (NBRF) using the program FASTA revealed that the cysteine-rich domains of the A chain are homologous to proteins containing epidermal growth factor (EGF)-like re- peats such as neurogenic protein Notch in Drosophila (37), the lin-12 homeotic protein in a soil nematode (38), EGF precursor (39), nerve growth factor receptor precursor (40), von Willebrand factor precursor (41), and transforming growth factor a (42). The Notch and lin-12 proteins showed much stronger homology with the A chain than with the other proteins, with an optimized alignment score of 320 and 250, respectively. Similar homologies to the EGF-like repeat have also been found in the B1 and B2 chains (21-25). This homology could reflect some functional relatedness. Recently laminin has been found to exert EGF-like growth factor a~t iv i ty .~ Although the laminin cell binding sequence YIGSR is located in an EGF-homology region of the B1 chain, the pentapeptide does not show EGF-like a~t ivi ty .~ I t is conceiv- able that one of the EGF-like sequences in the B2 and A chains may be responsible for this activity. The Notch locus is involved in cell-cell interaction and is essential for proper differentiation of the ectoderm in Drosophila (43). The Notch protein contains 36 EGF-like nonidentical repeats in the amino-terminal half, each of about 40 amino acids. Analysis of mutations indicate that there are different functions among the EGF-like repeats and that the total number of repeats is important for wild-type Notch protein function (44). By anal- ogy to these results, it is reasonable to assume that each cysteine-rich domain and each repeat within the domains in the A chain have a different function. For example, some domains may function as additional cell attachment sites, complementing the cell binding sequence YIGSR contained in the B1 chain.

In contrast to most EGF-like repeats which contain six cysteines in each repeat, the cysteine-rich repeats of the laminin chains consistently exhibit eight cysteines. It is rea- sonable to assume that the strict conversion of two additional

J. Engel, personal communication. ' G. Carpenter, personal communication.

cysteine residues, which are very likely disulfide bonded, is an important feature. They could help stabilize the rod-like structure of the short arms or form loops with yet unknown biological function. In this respect, it is very interesting to note that a portion of the sequence of the basement membrane heparan sulfate proteoglycan core protein (BPG) shows re- markable structure and sequence similarity to the short arm of the A chain. A partial primary structure of BPG has been determined by cDNA sequencing (45) and shown to contain globules separated by cysteine-rich domains which also con- tain eight cysteines at similar locations as those in the repeats in the A chain and B chains of laminin. This might indicate a wider distribution of this motif. Furthermore, the BPG globules show significant sequence homology to the corre- sponding domains of the A chain. The statistical significance of the similarity calculated using SEQDP of the IDEAS program is shown in Table 11. The comparison of the two globules, G1 and G2, in BPG with domains IVa and IVb of the A chain reveals a high alignment score (approximately -200) with a difference of 20 standard deviation units from the mean as determined using the mutation data matrix. Similarity between the A chain and BPG is even higher than between domains IVa and IVb of the A chain. Domain IV in the B2 chain shows some similarity to both G1 and G2 of BPG, whereas domain IV in the B1 chain shows none. The strong conservation of these regions in the A chain and BPG suggests not only evolutional relatedness but also common biological function between the two. In this context, it is of interest to note that BPG promotes cell binding through a unique cell surface-associated protein5 and that BPG binds to type IV collagen6 similar to the well documented function of laminin.

A pentapeptide, YIGSR, from the cysteine repeats in the B1 chain has been identified as being active for cell binding, chemotaxis, and receptor binding (8,9). This sequence is not present in the A and B2 chains. Since the carboxyl-terminal portion of long arm has also been shown to be active for cell binding (11, 12, 46), a different sequence must be used for this second site which is capable of interacting with a laminin receptor probably different from the 67-kDa receptor.

It is of interest to note that domain IIIb has an RGD sequence, known to be a versatile cell binding signal (47). Whether the RGD sequence is functional as a cell binding signal is likely to depend on the surrounding sequences since

B. Clement, B. Segui-Real, J. Hassell, Y. Yamada, and G. R.

J. R. Hassell, personal communication. Martin, manuscript in preparation.

16544 Laminin A Chain Sequence

the RGD sequence is found in many proteins but only a few of them are active as cell attachment sites. It is possible that the RGD sequence in laminin is active for a cell binding since integrin (the fibronectin receptor) can bind to laminin (48) and an antibody to integrin inhibits the interaction of integrin with laminin (49). Furthermore, it has been shown that endo- thelial cell attachment to a laminin substrate is inhibited by an RGD-containing synthetic peptide7, suggesting that lami- nin-mediated cell binding of certain cells may involve inter- action with an integrin family of receptors.

Most studies with laminin have been carried out using laminin prepared from the EHS tumor because of the diffi- culty of isolation from tissues. Although laminin prepared from the EHS tumor is composed of the A, B1, and B2 chains, there are several studies which suggest that isoforms of lam- inin occur and function. For example, levels of mRNA for the A chain are found to be much lower in adult mouse tissues compared to mRNA levels of the B1 and B2 chains (50). Schwann cells produce laminin molecules which lack the A chain (51,52). Although a variety of different activities of the A chain have been implicated, the exact function of the protein is still not clear. Studies with synthetic peptides, domain specific antibodies, and expressing A chain cDNA in appropriate cells will clarify the functional role of the A chain.

Acknowledgments-We thank Leslie Bruggeman, Peter Burbelo, George R. Martin, Douglas Noonan, and Gregory Sephel for critical reading of the manuscript and Minoru Kanehisa and David Lipman for the computer programs. We thank Sabine Tolle for technical assistance.

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

REFERENCES Kleinman, H. K., Cannon, F. B., Laurie, G. W., Hassell, J. R.,

Aumailley, M., Terranova, V. P., Martin, G. R., and DuBois- Daleq, M. (1985) J. Cell Biochem. 217 , 317-325

Martin, G. R., and Timpl, R. (1987) Annu. Rev. Cell. Biol. 3,57- 85

Baron van Evercooren, A., Kleinman, H. K., Ohno, S., Marangos, P., Schwartz, J. P., and Dubois-Dalcq, M. E. (1982) J. Neurosci. Res. 8,179-194

Terranova, V. P., Williams, J. E., Liotta, L. A., and Martin, G. R. (1984) Science 226,982-985

Cooper, A. R., Kurkinen, M., Taylor, A., Hogan, B. L. M. (1981) Eur. J. Biochem. 119,189-197

Engel, J., Odermatt, E., Engel, A., Madri, J. A., Furthmayr, H., Rohde, H., and Timpl, R. (1981) J. Mol. Biol. 150,97-120

Terranova, V. P., Rao, C. N., Kalebic, T., Margulies, I. M., and

448 Liotta, L. A. (1983). Proc. Natl. Acad. Sci. U. S. A. 80, 444-

Graf, J., Iwamoto, Y., Sasaki, M., Martin, G. R., Kleinman, H. K., Robey, F. A., and Yamada, Y. (1987) Cell 48,989-996

Iwamoto, Y., Robey, F. A., Graf, J., Sasaki, M., Kleinman, H. K., Yamada, Y., and Martin, G. R. (1987) Science 238,1132-1134

Iwamoto, Y., Graf, J., Sasaki, M., Kleinman, H. K., Greatorex, D. R., Martin, G. R., Robey, F. A., and Yamada, Y. (1988) J. Cell Physiol. 134, 287-291

Aumailley, M., Nurcombe, V., Edgar, D., Paulsson, M., and Timpl, R. (1987) J. Bwl. Chem. 262, 11532-11538

Goodman, S. L., Deutzmann, R., von der Mark, K. (1987) J. Cell

Edgar, D., Timpl, R., and Thoenen, H. (1984) EMBO J. 3,1463-

Engvall, E., Davis, G. E., Dickerson, K., Ruoslahti, E., Varon, S.,

Sakashita, S., Engvall, E., and Ruoslahti, E. (1980) FEBS Lett.

BWl. 105,589-598

1468

and Manthorpe, M. (1986) J. Cell Biol. 103 , 2457-2465

116.243-246

16. Skubitz, A. B., McCarthy, J. B., Charonis, A. S., and Furcht, L. T. (1988) J. Bwl. Chem. 263 , 4861-4868

17. Rao, C. N., Margulies, I. M. K., Tralka, T. S., Terranova, V. P., Madrie, J. A., and Liotta, L. A. (1982) J. Biol. Chem. 257 ,

18. Charonis, A. S., Tsilibary, E. C., Saku, T., and Furthmayr, H.

19. Laurie, G. W., Bing, J. T., Kleinman, H. K., Hassell, J. R., and

20. Barlow, D. P., Green, N. M., Kurkinen, M., and Hogan, B. L. M.

21. Sasaki, M., Kohno, K., Kato, S., Martin, G. R., and Yamada, Y. (1987) Proc. Natl. Acad. Sci. U. S. A. 8 4 , 935-939

22. Sasaki, M., and Yamada, Y. (1987) J. Biol. Chem. 262, 17111- 17117

23. Pikkarainen, T., Eddy, R., Fukushima, Y., Byers, M., Shours, T., Pihlajaniemi, T., Saraste, M., and Tryggvason, K. (1987) J. Biol. Chem. 262,10454-10462

24. Pikkarainen, T., Kallunki, T., and Tryggvason, K. (1988) J. Bwl. Chem. 263,6751-6758

25. Montell, D. J., and Goodman, C. S. (1988) Cell 53,463-473 26. Timpl, R., Rohde, H., Gehron Robey, P., Rennard, S. I., Foidart,

27. Sanger, F., Coulson, A. R., Barell, B. G., Smith, A. J. A., and Roe,

28. Staden, R. (1982) Nucleic Acids Res. 10,2951-2961 29. Kanehisa, M. (1984) Nucleic Acids Res. 11,417-428 30. Pearson, W. R., and Lipman, D. J. (1988) Proc. Natl. Acad. Sci.

31. von Heijne, G. (1986) Nucleic Acids Res. 10 , 183-196 32. Doolittle, R. F., Goldbaum D. M., and Doolittle, L. R. (1978) J.

33. Mclachlan, A. D., and Karn, J. (1983) J. Mol. Bwl. 15,605-625 34. Paulsson, M., Deutzmann, R., Timpl, R., Dalzoppo, D., Odermatt,

35. Ott, V., Odermatt, E., Engel, J., Furthmayr, H., and Timpl, R.

36. Fujiwara, S., Wiedemann, M., Timpl, R., Lustig, A., and Engel,

37. Wharton, K. A., Johansen, K. M., Xu, T., and Artavanis-Tsa-

38. Greenwald, I. (1985) Cell 4 3 , 583-590 39. Gray, A., Dull, T. J., and Ullrich, A. (1983). Nature 303 , 722-

725 40. Johnson, D., Lanaham, A., Buck, C. R., Sehgal, A., Morgan, C.,

Mercer, E., Bothwell, M., and Chao, M. (1986) Cell 4 7 , 545- 554

41. Titani, K., Kumar, S., Takio, K., Ericsson, L. H., Wade, R. D., Ashinda, K., Walsh, K. A., Chopek, M. W., Sadler, J. E., and Fujikawa, K. (1986) Biochemistry 26,3171-3184

42. Derynck, R., Roberts, A. B., Winkler, M. E., Chem, E. Y., and Goeddel, D. V. (1984) Cell 3 8 , 287-297

43. Lehman, R., Jimenez, F., Dietrich, V., and Campos-Ortega, J. A. (1983) Roux's Arch. Deu. Bwl. 192,62-74

44. Kelley, M. R., Kidd, S., Deutsch, W. A., and Young, M. W. (1987) Cell 51,539-548

45. Noonan, D. M., Horigan, E., Ledbetter, S., Vogeli, G., Sasaki, M., Yamada, Y., and Hassell, J. R. (1988) J. Bwl. Chem., in press

46. Edgar, D., Timpl, R., and Thoenen H. (1988) J. Cell Bid . 1 0 6 , 1299-1306

47. Ruoslahti, E., and Pierschbacher, M. D. (1987) Science 238,491- 497

48. Horwitz, A., Duggan, K., Greggs, R., Decker, C., and Buck, C. (1985) J. Cell Biol. 101,2134-2144

49. Tomaselli, K. J., Reichardt, L. F., and Bixby, J. L. (1986) J. Cell Biol. 103,2659-2672

50. Kleinman, H. K., Ebihara, I., Killen, P. D., Sasaki, M., Cannon, F. B., Yamada, Y., and Martin, G. R. (1987) Deu. Bwl. 122 ,

51. Palm, S. L., and Furcht, L. T. (1983) J. Cell Biol. 96,1218-1226 52. Cornbrooks. C. J.. Carev. D. J., McDonald, J. A., Timpl, R., and

9740-9744

(1986) J. Cell Bwl. 103, 1689-1697

Aumailley, M. (1986) J. Mol. Bwl. 189, 205-216

(1984) EMBO J.3,2355-2362

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

B. A. (1980) J. Mol. BWl. 143 , 161-178

U. S. A. 85,2444-2448

Mol. BWl. 120,311-325

E., and Engel, J. (1985) EMBO J. 4,309-316

(1982) Eur. J. Biochem. 123,63-72

J. (1984) Eur. J. Biochem. 143,145-157

konas, S. (1985) Cell 43 , 567-581

373-378

Bunge, R.' P. (1983) Proc. Nutl. Acad. Sci. U. S . A . 80, 3850- ' K. Tashiro and D. Grant, personal communication. 3854