Matrix Vol. 1111991, pp.151-160
Original Papers
© 1991 by Gustav Fischer Verlag, Stuttgart
Molecular Cloning of the cDNA Encoding Human Laminin A Chain TAPIO HAAPARANTA 1 , JOUNI UITT02, ERKKI RUOSLAHTI 1 and EVA ENGVALL 1 La Jolla Cancer Research Foundation, 10901 North Torrey Pines Rd., La Jolla, CA 92037 and Department of Dermatology, Jefferson Medical College, Thomas Jefferson University, 1020 Locust St., Philadelphia, PA 19107, USA. 1
2
Abstract
Laminin is a large basement membrane glycoprotein composed of three subunits designated the A, B1, and B2. We report here the isolation and nucleotide sequence of human laminin A chain cDNA. The nucleotide sequence spans 9505 bases and has an open reading frame encoding 3075-amino acids. The sequence covers a 77-nucleotide long 5' untranslated region and a 190nucleotide long 3' sequence in front of the poly (A)+ tail. In analogy with the mouse A chain sequence, the deduced human amino acid sequence contains eight-distinct domains of fourglobular regions, three-cysteine-rich domains and an a-helical region, which is thought to interact with the B chains of laminin. The deduced amino acid sequence is 14-amino acids shorter than the mouse A chain sequence. Seven of these amino acids are located in the putative signal sequence. The overall identity between the sequences from the two species is 78%. The carboxylterminal globular (G) domain contains five homologous subdomains characterized by a conserved seven-amino acid repeat within each subdomain. Both human and mouse A chain are about 39% identical to the G domain of merosin, a recently discovered A chain homologue. Unlike the mouse A chain, the human A chain contains a potential cell binding sequence (RGD) in this domain. The RGD sequence that is thought to be a cryptic cell attachment site in the amino-terminal domain IIIb of mouse laminin is not conserved in the human sequence. Key words: cDNA, laminin A chain, sequence.
Introduction
Laminin is a large glycoprotein composed of three polypeptide chains, designated A, B1 und B2, which are disulfide-linked to each other. Rotary shadowing electron microscopy of laminin reveals an asymmetric cross-shaped molecule of three short arms and one long arm with globular domains at the ends of each arm. The A chain (M r 400,000) forms one of the short arms, joins the B chains to form part of the long arm and continues as the large globular domain at the carboxyl-terminal end. The B1 (M r 220,000) and B2 (M r - 210,000) chains each account for 1
Abbreviations: kb, kilobase(s); bp, basepair(s).
one short arm and are disulfide-linked to the A chain to form part of the long arm. Laminin is produced by a variety of cell types and is in tissue specifically localized to and is a major component in basement membranes (Timpl, 1989; Beck et aI., 1990; Yurchenco and Schittny, 1990). As can be expected from such a complex molecule, laminin has been shown to participate in a variety of biological functions. Laminin associates with other basement membrane components, such as nidogenientactin, type IV collagen and heparan sulfate proteoglycan, and it also self assembles. It affects cell behavior, including morphology, migration, growth, and differentiation, and it also promotes neurite outgrowth. The cellular interactions involve a family of cell surface receptors, the integrins; to date six-
152
T. Haaparanta et ai.
different integrins have been shown to bind laminin (Gehlsen et aI., 1988; Takada et aI., 1988; Languino et aI., 1989; Sonnenberg et aI., 1988; Kramer et aI., 1989; Turner et aI., 1989; Lotz et aI., 1990). Studies with proteolytic fragments of intact laminin, synthetic peptides corresponding to various parts of the molecule, or domain-specific antibodies (Engvall et aI., 1986; Goodman et aI., 1987; Aumailley et aI., 1987; Tashiro et aI., 1989; Gehlsen et aI., 1989) have shown that different regions of the laminin molecule display different biological activities. The end of the long arm, including the large carboxyl-terminal globule of the A chain, plays a major role in promoting neurite outgrowth and cell attachment, and this region also binds heparin (Y urchenco and Schittny, 1990). Nidogen/entactin, an - 150 kDa glycoprotein, binds to the center of the laminin-cross (probably domain III of the B1 chain) and interacts with type IV collagen, thus forming a bridge between these two macromolecules. Type IV collagen binds also directly to the ends of the laminin arms, although this interaction appears to be of lower affinity (Yurchenco and Schittny, 1990). Cell attachment sites have been located to the short arms of mouse laminin. Molecular cloning of mouse, human and Drosophila Bl and B2 chain, mouse A chain and part of human A chain cDNAs have provided detailed information about the primary structure of laminin (Sasaki et aI., 1987; 1989; Sasaki and Yamada, 1987; Durkin et aI., 1988; Pikkarainen et aI., 1987; 1988; Montell and Goodman, 1988; 1989; Chi and Hui, 1989; Olsen et aI., 1989). The predicted secondary structure of the mouse A chain suggest, that it consists of 8distinct domains including globular, a-helical and cysteine rich structures. We present in this paper the complete cDNA of the human laminin A chain. Sequence analysis of the cDNA clones and the deduced amino acid sequence reveals that the human and mouse homologues are highly conserved and share structural features. However, they differ in the location of their RGD sequences. A comparison with the carboxyl-terminal sequence of merosin (Ehrig et aI., 1990), a human laminin A chain variant, also reveals structural similarities between these two basement membrane components.
Materials and Methods Materials
Restriction endonucleases and modifying enzymes were purchased from either New England Biolabs (Beverly, MA), Bethesda Research Laboratories (Gaithersburg, MD) or Stratagene (La Jolla, CAl. Hybond nylon membranes and the cDNA synthesis kit were from Amersham (Arlington Heights, IL). Avian myeloblastosis virus reverse transcriptase was obtaineo from Life Sciences (St. Petersburg, FL). The lambda ZAP II vector and Gigapack Gold lambda
packaging extract were from Stratagene. Radioactive nucleotides were purchased from DuPont-New England Nuclear (Boston, MA) and reagents for sequencing with the modified T7 DNA polymerase (Sequenase) were from United States Biochemical Corp. (Cleveland, OH). Oligonucleotides were synthesized using an Applied Biosystems DNA synthesizer (Foster City, CAl. Construction ofJAR cell eDNA libraries
The human choriocarcinoma JAR cell line (ATCC HTB 144) was obtained from American Type Culture Collection (Rockville, MD). Cells were cultured in Dulbecco's modified Eagle's minimum essential medium supplemented with 10% fetal calf serum. Cytoplasmic RNA was isolated from semiconfluent cultures of JAR cells and poly (A) + RNA was selected twice by oligo (dT) cellulose column chromatography (Aviv and Leder, 1972). Two-cDNA libraries were constructed using either random hexanucleotides or oligo{dT) as primers for the first strand cDNA synthesis (Gubler and Hoffman, 1983). Synthesis of the cDNAs were accomplished using a commercial kit (cDNA Synthesis Plus, Amersham) according to the manufacturers instructions, except that avian myeloblastosis virus reverse transcriptase (Life Sciences, St. Petersburg, FL) was used and methyl mercuric hydroxide was used to denature the mRNA, prior to first strand synthesis. The blunted cDNA was ligated with EcoRIlNot 1 linker-adapters (Invitrogen, San Diego, CAl and phosphorylated with T4 polynucleotide kinase. The cDNA was fractionated on a 1 % agarose gel and the cDNA species larger than 700 bp was electroeluted and ligated into EcoRI-cut "ZAPII vector arms (Short et aI., 1988). The ligated cDNA was in vitro-packaged employing the Gigapack Gold packaging extract and plated on XL-1 Blue cells (Stratagene). Over 1 X 10 6 independent clones were obtained, of which 98% were recombinant, from both the oligo{dT) and random hexanucleotide-primed cDNAs; each library was amplified once. Both libraries contained - 0.29% actin positive plaques and a Southern blot of total phage DNA digested with Notl and probed with a y-actin cDNA fragment (Gunning et aI., 1987) yielded bands at about 2 kb, the approximate size of a full length actin mRNA, and tailing down to about 1.3 kb indicating a good representation of full length mRNA species in the libraries (Hagen, 1988). Isolation of eDNA clones
Phage were screened by plaque hybridization (Benton and Davis, 1977) using cDNA fragments radiolabeled by the random primer method of Feinberg and Vogel stein (1984) as probes. Duplicate plaque lifts were prepared and hybridization was conducted at 42°C for 16 h in 5 x SSPE (I x SSPE: 0.15 MNaCl, 10mMNaH2 P0 4 , 1 mMEDTA),
Human Laminin A Chain Sequence 1 x Denhardt's solution (0.02% Ficoll, 0.02% bovine serum albumin, 0.02% polyvinylpyrrolidone), 0.1% sodium dodecyl sulfate, 40% formamide and 150llglml sonicated and denatured salmon sperm DNA. The filters were washed in 0.5 X SSPE, 0.1 % sodium dodecyl sulfate at 55°C, and autoradiographed overnight at - 80°C using intensifier screens. Positive plaques were purified to homogeneity and the Bluescript plasmids containing the cDNA inserts were excised in vivo by coinfection of XL-1 cells with the recombinant "-ZAP II phage and R408 helper phage according to protocols provided by Stratagene. Plasmid DNA was isolated for subsequent restriction enzyme mapping and DNA sequence analysis. DNA sequencing
Sequencing was performed on double-stranded plasmid DNA in both directions as described by Kraft et al. (1988) or Toneguzzo et al. (1988) using the Sequenase enzyme and oligonucleotide primers. Regions exhibiting gel compressions and/or pause sites were sequenced using dITP and single strand binding protein (United States Biochemical Corp.). The labeling step was generally performed on ice for 2 min and the extension reaction at 50°C for 5 min. Ambiguous regions were resolved using one or a combination of these modifications of the standard Sequenase protocol. Nucleic acid and amino acid sequences were analyzed with the Microgenie (Beckman, Palo Alto, CAl and PC Gene (Intelligenetics, Mountain View, CAl programs. Northern blot analysis
Denatured total (50 Ilg per lane) or poly(A) + RNA (5 Ilg per lane) was fractionated on a 1.2 % agarose gel containing 0.66 M formaldehyde in 1 x MOPS buffer and then transferred onto Hybond-N filters. Blots were probed with random prime-labeled (Feinberg and Vogelstein, 1984) - 1Kb long restriction fragments of the A chain cDNA or a yactin fragment (Gunning et aI., 1987). The filters were prehybridized, hybridized and washed as described for the plaque lift hybridization procedure using random primelabeled cDNA probes. RNA integrity was monitored by ethidium bromide staining, and the size of the hybridizing RNA was determined by the positions of the 28S and 18S ribosomal RNA bands and RNA molecules of known size (high MW RNA ladder, BRL) run in parallel.
Results
A cDNA fragment covering 2232 nucleotides of the 3'end of human laminin A chain cDNA was previously isolated from a placental "-gt11 expression library (Olsen et aI.,
153
1989). Attempts to obtain clones extending further upstream from the placental library or an endothelial cell library (Ginsberg et aI., 1985) by screening with anti-laminin antibodies (Engvall et aI., 1986) or the carboxyl-terminal cDNA fragment (Olsen et aI., 1989) were not successful. The libraries were also screened for the presence of extended clones by the polymerase chain reaction (Friedman et aI., 1988) by using a "-gt11 sequencing primer and oligonucleotide primers derived from the 5' end of the laminin A chain cDNA, but no extended cDNAs were obtained. We therefore decided to test different human cell lines known to express laminin (Alitalo et aI., 1981) for the presence of A chain transcripts by Northern blot analysis and then construct a cDNA library from a positive cell source. A human placental choriocarcinoma line, JAR, expressed a - 10-kb transcript when lOllg of poly(A)+ RNA was analyzed by Northern blotting using a 700-bp fragment from the 5' end of the 2.2-kb human A chain cDNA clone (Olsen et aI., 1989) as the probe (not shown). Two other less prominent transcripts of about 7.5 kb and 5 kb were also detected. These bands may represent transcripts coding for variant or truncated laminin A chains. Northern blotting of HT29, T24, HTl080 or MG63 cell RNA (Alitalo et aI., 1981) showed very low or undetectable A chain message levels. We also analyzed RNA isolated from rabbit tissues, including liver, lung, tongue, kidney, and skeletal muscle, human placental and brain RNA (Clontech, Palo Alto, CAl but failed to detect any A chain transcripts, even at relatively low stringencies. We thus concluded that the JAR cells were the best source of A chain mRNA and the best candidate for library construction. Random oligonucleotide or oligo(dT) primers were used to construct two separate cDNA libraries which were screened with a 700-bp cDNA fragment from the 5' end of the 2.2-kb human A chain cDNA. Fifty-two positive clones were isolated from the random-primed and 20 from the oligo(dT)-primed library. This process was repeated four times by using a fragment from the 5' end of the most 5' sequence from each round until the entire sequence was obtained. Clone 123 was found to contain a 179-bp deletion between nucleotides 1045 to 1225 (indicated by a break in the line representing this clone in Fig. 1). This deletion was most probably a cloning artifact because it created a change in the reading frame and it was not present in clones 126 and 129. A partial restriction endonuclease map of the cDNA and the positions of the clones which were sequenced in their entirety are shown in Figure 1. DNA sequence analysis
A total of 72-overlapping cDNA clones were isolated and their ends were sequenced. Clones 7,67,123 and 129 were sequenced in their entirety and found to have a considerable sequence homology to the mouse A chain cDNA (Sasaki et
154
T. Haaparanta et al. 3'
5' EBK !
K
S
S
I
i
i 4
3
2
H
ii
5
H
H
ii
i
6
7
i
KE S
!I'
i
i
8
9
(A)n kb
123
A
7
129
(A)n
67
126
11
f - - - - - - - - (A)n
19
1------(A)n
36
aI., 1988). The overlapping clones span 9505 nucleotides of sequence with an open reading frame of 9225 nucleotides that encodes a 3075-amino acid polypeptide of Mr 337,163. The sequence around the putative initiator methionine at position 78 is a likely site of initiation of translation. The sequence GGC GAG ATG C is in agreement with the Kozak consensus sequence of translation initiation by eukaryotic ribosomes (Kozak, 1987). This initiation codon is followed by a sequence that codes for an apparent signal sequence (von Heijne, 1986). Analysis of this sequence using the program SIGNAL (PC GENE) predicts two possible signal peptidase cleavage sites between amino acids 17 and 18 or 19 and 20 that conform to the (- 3, -1) rule. We have therefore chosen to number the deduced amino acid sequence beginning with the initiation methionine as the first amino acid in Figure 2, rather than predicting the beginning of the mature polypeptide. The 77-nucleotide cD NA sequence upstream of the ATG codon is characteristically GC rich (77% GC) and does not contain any ATG codons. The 3' noncoding region contains 190 nucleotides before the poly(A) tail. There is one polyadenylation signal (ATTAAA) 13-nucleotides upstream of the poly(A) tail. The protein coding sequence contains 34-potential Nglycosylation sites and 59-cysteine residues (excluding the signal peptide sequence). The putative cell-binding site, IKVAV (14), is present at amino acid position 2116 and an RGD sequence (Ruoslahti and Pierschbacher, 1987) is found at position2534. Recently, a tripeptide sequence, LRE, was identified in s-laminin, which is a Bl-related laminin polypeptide, as a motor neuron-selective attachment site (Hunter et aI., 1989a, b). This sequence occurs three times in the human A chain at positions 1789, 2050 and 2691. Olsen et al. (1989) has previously published a partial sequence of the human laminin A chain covering the carboxyl-terminal 2242 nucleotides. That sequence and the
Fig. 1. Map of human laminin A chain eDNA. A partial restriction enzyme map and schematic representation of the overlapping eDNA clones are shown. Cleavage sites for Eco Rl (E), Bam HI (B), Kpn 1 (K), Sma 1 (5), and Hind III (H) are indicated. The thick part of the upper line indicates the open reading frame. The break in the line representing clone 123 indicates the position of a deletion covering nucleotides 1045 -1225 in this clone.
sequence presented in this paper are identical except for differences at ll-nucleotide positions. Specifically, C at position 7263 (rather than T) results in an arginine instead of a tryptophan residue; C at position 8315 (rather than A) results in phenylalanine instead of leucine; G at position 9062 (rather than A) results in no amino acid change; Tat position 9237 (rather than C) results in phenylalanine instead of leucine. Nucleotides 9293-9300 are GACCGAGT in this paper rather than ACCGAGTC in Olsen et al. (1989) which changes the C-terminal amino acids to tyrosine-glutamic acid-serine instead of proline-serine-proline. Finally at position 9374 in the 3' non-coding region our sequence has an additional A (AGAATC rather than AGATC). Unless these differences are due to cloning artifacts, they may reflect differences between the JAR cell and placentallaminin A chains or represent allelic variants. Comparison of human and mouse laminin A chain and merosin amino acid sequences
The deduced amino acid sequences of human and mouse laminin A chains and the merosin heavy (M) chain were analyzed using the PALIGN program. The mouse laminin A chain polypeptide has been predicted to consist of 8 structurally distinct domains. Similar analysis of the human A chain predicts an identical domain organization. The overall amino acid sequence identity with the mouse A chain is 76%. Furthermore, all cysteine residues, except 2 in the putative signal peptide sequence, are conserved, suggesting a highly similar three-dimensional structure between the two molecules. The mouse sequence is 14amino acids longer than the human, but 7 of these extra residues are found in the putative signal peptide. Table I shows the degree of sequence similarity in the different domains between the human and mouse sequence. The a-helical rod portion (domains I and II) which is thought to interact with the B chains is the least conserved
Human Laminin A Chain Sequence
155
cxacTcn:G~OCOCGGCT~l\ICGCGliGGGCGIa:ta::1'(X;fCft'OCrocrorol'GlUiCCOCOCAG-I~ MRGGVLLVLLL@VAAQ@~~J\GLFPAILN
za
162
~'rGC'1'CACArcNrACX:AA~~TG'l'OC'CAOO'I'C'OOCCro~;.pa::roA~TGOCAACAOCOCAAN:CCCAGAGAN;OCCA'I'CCAAT L A S N A H I S T ~ (£) G E X G P E H F (£) K L V E H V P G R. P V R N P 0 (9 R I (9 0 G N SAN P R E R H P I
82
323
ATCACA.'l'OCCATAGA~TAACTOO'I'GOCMAGrc:cr::AGC7l'l'TCAGAA~TATCAC'l'GGGTCACAA~T"I'AAGACAGGTC'l'TTCMGTTGCATAro'I'CATCAlTAAAGC"l"GCCAA~TGGA.
__ !Gi\OOCCTGl"I"l'CC'l"'1"TC.'1'CAAT
.8.
R P G N W I
136
'lTl'~TCTCTGGA.'l'GOCACCACGT'l'CAoccx:cT'QX:AGTATTA'roC'.AGTCAOC~T'I'G'l'C'l'CGTTACAATATAAC't'CCMGACGAooa:cACCCACX:TACAooa:TGATGATGAAG'l'GA.TC~rccrATTATTCCAGATTG L E R S L O G T T F S P W 0 Y Y A V S 0 S E (£) L S R Y ~ P R R G P P T Y R ADD E V I (9 T S Y Y S R L
S H A lOG
T NNW W Q S P S I
Q N G R E Y H W V TIT
L D L R 0
V F 0
V A Y V I
I
K A A NAP
189
.. 5
G'I'GCX:AC'l"I'GP.OCAl'OOItGAGATTCArACATCACTCA'l'CM~l'CTn'C.ACXX:A1IGlTGTTOGAAT1CAC'I'1'CTOCN:XaTATA'l"I'CGO:'I'T
Original Papers
© 1991 by Gustav Fischer Verlag, Stuttgart
Molecular Cloning of the cDNA Encoding Human Laminin A Chain TAPIO HAAPARANTA 1 , JOUNI UITT02, ERKKI RUOSLAHTI 1 and EVA ENGVALL 1 La Jolla Cancer Research Foundation, 10901 North Torrey Pines Rd., La Jolla, CA 92037 and Department of Dermatology, Jefferson Medical College, Thomas Jefferson University, 1020 Locust St., Philadelphia, PA 19107, USA. 1
2
Abstract
Laminin is a large basement membrane glycoprotein composed of three subunits designated the A, B1, and B2. We report here the isolation and nucleotide sequence of human laminin A chain cDNA. The nucleotide sequence spans 9505 bases and has an open reading frame encoding 3075-amino acids. The sequence covers a 77-nucleotide long 5' untranslated region and a 190nucleotide long 3' sequence in front of the poly (A)+ tail. In analogy with the mouse A chain sequence, the deduced human amino acid sequence contains eight-distinct domains of fourglobular regions, three-cysteine-rich domains and an a-helical region, which is thought to interact with the B chains of laminin. The deduced amino acid sequence is 14-amino acids shorter than the mouse A chain sequence. Seven of these amino acids are located in the putative signal sequence. The overall identity between the sequences from the two species is 78%. The carboxylterminal globular (G) domain contains five homologous subdomains characterized by a conserved seven-amino acid repeat within each subdomain. Both human and mouse A chain are about 39% identical to the G domain of merosin, a recently discovered A chain homologue. Unlike the mouse A chain, the human A chain contains a potential cell binding sequence (RGD) in this domain. The RGD sequence that is thought to be a cryptic cell attachment site in the amino-terminal domain IIIb of mouse laminin is not conserved in the human sequence. Key words: cDNA, laminin A chain, sequence.
Introduction
Laminin is a large glycoprotein composed of three polypeptide chains, designated A, B1 und B2, which are disulfide-linked to each other. Rotary shadowing electron microscopy of laminin reveals an asymmetric cross-shaped molecule of three short arms and one long arm with globular domains at the ends of each arm. The A chain (M r 400,000) forms one of the short arms, joins the B chains to form part of the long arm and continues as the large globular domain at the carboxyl-terminal end. The B1 (M r 220,000) and B2 (M r - 210,000) chains each account for 1
Abbreviations: kb, kilobase(s); bp, basepair(s).
one short arm and are disulfide-linked to the A chain to form part of the long arm. Laminin is produced by a variety of cell types and is in tissue specifically localized to and is a major component in basement membranes (Timpl, 1989; Beck et aI., 1990; Yurchenco and Schittny, 1990). As can be expected from such a complex molecule, laminin has been shown to participate in a variety of biological functions. Laminin associates with other basement membrane components, such as nidogenientactin, type IV collagen and heparan sulfate proteoglycan, and it also self assembles. It affects cell behavior, including morphology, migration, growth, and differentiation, and it also promotes neurite outgrowth. The cellular interactions involve a family of cell surface receptors, the integrins; to date six-
152
T. Haaparanta et ai.
different integrins have been shown to bind laminin (Gehlsen et aI., 1988; Takada et aI., 1988; Languino et aI., 1989; Sonnenberg et aI., 1988; Kramer et aI., 1989; Turner et aI., 1989; Lotz et aI., 1990). Studies with proteolytic fragments of intact laminin, synthetic peptides corresponding to various parts of the molecule, or domain-specific antibodies (Engvall et aI., 1986; Goodman et aI., 1987; Aumailley et aI., 1987; Tashiro et aI., 1989; Gehlsen et aI., 1989) have shown that different regions of the laminin molecule display different biological activities. The end of the long arm, including the large carboxyl-terminal globule of the A chain, plays a major role in promoting neurite outgrowth and cell attachment, and this region also binds heparin (Y urchenco and Schittny, 1990). Nidogen/entactin, an - 150 kDa glycoprotein, binds to the center of the laminin-cross (probably domain III of the B1 chain) and interacts with type IV collagen, thus forming a bridge between these two macromolecules. Type IV collagen binds also directly to the ends of the laminin arms, although this interaction appears to be of lower affinity (Yurchenco and Schittny, 1990). Cell attachment sites have been located to the short arms of mouse laminin. Molecular cloning of mouse, human and Drosophila Bl and B2 chain, mouse A chain and part of human A chain cDNAs have provided detailed information about the primary structure of laminin (Sasaki et aI., 1987; 1989; Sasaki and Yamada, 1987; Durkin et aI., 1988; Pikkarainen et aI., 1987; 1988; Montell and Goodman, 1988; 1989; Chi and Hui, 1989; Olsen et aI., 1989). The predicted secondary structure of the mouse A chain suggest, that it consists of 8distinct domains including globular, a-helical and cysteine rich structures. We present in this paper the complete cDNA of the human laminin A chain. Sequence analysis of the cDNA clones and the deduced amino acid sequence reveals that the human and mouse homologues are highly conserved and share structural features. However, they differ in the location of their RGD sequences. A comparison with the carboxyl-terminal sequence of merosin (Ehrig et aI., 1990), a human laminin A chain variant, also reveals structural similarities between these two basement membrane components.
Materials and Methods Materials
Restriction endonucleases and modifying enzymes were purchased from either New England Biolabs (Beverly, MA), Bethesda Research Laboratories (Gaithersburg, MD) or Stratagene (La Jolla, CAl. Hybond nylon membranes and the cDNA synthesis kit were from Amersham (Arlington Heights, IL). Avian myeloblastosis virus reverse transcriptase was obtaineo from Life Sciences (St. Petersburg, FL). The lambda ZAP II vector and Gigapack Gold lambda
packaging extract were from Stratagene. Radioactive nucleotides were purchased from DuPont-New England Nuclear (Boston, MA) and reagents for sequencing with the modified T7 DNA polymerase (Sequenase) were from United States Biochemical Corp. (Cleveland, OH). Oligonucleotides were synthesized using an Applied Biosystems DNA synthesizer (Foster City, CAl. Construction ofJAR cell eDNA libraries
The human choriocarcinoma JAR cell line (ATCC HTB 144) was obtained from American Type Culture Collection (Rockville, MD). Cells were cultured in Dulbecco's modified Eagle's minimum essential medium supplemented with 10% fetal calf serum. Cytoplasmic RNA was isolated from semiconfluent cultures of JAR cells and poly (A) + RNA was selected twice by oligo (dT) cellulose column chromatography (Aviv and Leder, 1972). Two-cDNA libraries were constructed using either random hexanucleotides or oligo{dT) as primers for the first strand cDNA synthesis (Gubler and Hoffman, 1983). Synthesis of the cDNAs were accomplished using a commercial kit (cDNA Synthesis Plus, Amersham) according to the manufacturers instructions, except that avian myeloblastosis virus reverse transcriptase (Life Sciences, St. Petersburg, FL) was used and methyl mercuric hydroxide was used to denature the mRNA, prior to first strand synthesis. The blunted cDNA was ligated with EcoRIlNot 1 linker-adapters (Invitrogen, San Diego, CAl and phosphorylated with T4 polynucleotide kinase. The cDNA was fractionated on a 1 % agarose gel and the cDNA species larger than 700 bp was electroeluted and ligated into EcoRI-cut "ZAPII vector arms (Short et aI., 1988). The ligated cDNA was in vitro-packaged employing the Gigapack Gold packaging extract and plated on XL-1 Blue cells (Stratagene). Over 1 X 10 6 independent clones were obtained, of which 98% were recombinant, from both the oligo{dT) and random hexanucleotide-primed cDNAs; each library was amplified once. Both libraries contained - 0.29% actin positive plaques and a Southern blot of total phage DNA digested with Notl and probed with a y-actin cDNA fragment (Gunning et aI., 1987) yielded bands at about 2 kb, the approximate size of a full length actin mRNA, and tailing down to about 1.3 kb indicating a good representation of full length mRNA species in the libraries (Hagen, 1988). Isolation of eDNA clones
Phage were screened by plaque hybridization (Benton and Davis, 1977) using cDNA fragments radiolabeled by the random primer method of Feinberg and Vogel stein (1984) as probes. Duplicate plaque lifts were prepared and hybridization was conducted at 42°C for 16 h in 5 x SSPE (I x SSPE: 0.15 MNaCl, 10mMNaH2 P0 4 , 1 mMEDTA),
Human Laminin A Chain Sequence 1 x Denhardt's solution (0.02% Ficoll, 0.02% bovine serum albumin, 0.02% polyvinylpyrrolidone), 0.1% sodium dodecyl sulfate, 40% formamide and 150llglml sonicated and denatured salmon sperm DNA. The filters were washed in 0.5 X SSPE, 0.1 % sodium dodecyl sulfate at 55°C, and autoradiographed overnight at - 80°C using intensifier screens. Positive plaques were purified to homogeneity and the Bluescript plasmids containing the cDNA inserts were excised in vivo by coinfection of XL-1 cells with the recombinant "-ZAP II phage and R408 helper phage according to protocols provided by Stratagene. Plasmid DNA was isolated for subsequent restriction enzyme mapping and DNA sequence analysis. DNA sequencing
Sequencing was performed on double-stranded plasmid DNA in both directions as described by Kraft et al. (1988) or Toneguzzo et al. (1988) using the Sequenase enzyme and oligonucleotide primers. Regions exhibiting gel compressions and/or pause sites were sequenced using dITP and single strand binding protein (United States Biochemical Corp.). The labeling step was generally performed on ice for 2 min and the extension reaction at 50°C for 5 min. Ambiguous regions were resolved using one or a combination of these modifications of the standard Sequenase protocol. Nucleic acid and amino acid sequences were analyzed with the Microgenie (Beckman, Palo Alto, CAl and PC Gene (Intelligenetics, Mountain View, CAl programs. Northern blot analysis
Denatured total (50 Ilg per lane) or poly(A) + RNA (5 Ilg per lane) was fractionated on a 1.2 % agarose gel containing 0.66 M formaldehyde in 1 x MOPS buffer and then transferred onto Hybond-N filters. Blots were probed with random prime-labeled (Feinberg and Vogelstein, 1984) - 1Kb long restriction fragments of the A chain cDNA or a yactin fragment (Gunning et aI., 1987). The filters were prehybridized, hybridized and washed as described for the plaque lift hybridization procedure using random primelabeled cDNA probes. RNA integrity was monitored by ethidium bromide staining, and the size of the hybridizing RNA was determined by the positions of the 28S and 18S ribosomal RNA bands and RNA molecules of known size (high MW RNA ladder, BRL) run in parallel.
Results
A cDNA fragment covering 2232 nucleotides of the 3'end of human laminin A chain cDNA was previously isolated from a placental "-gt11 expression library (Olsen et aI.,
153
1989). Attempts to obtain clones extending further upstream from the placental library or an endothelial cell library (Ginsberg et aI., 1985) by screening with anti-laminin antibodies (Engvall et aI., 1986) or the carboxyl-terminal cDNA fragment (Olsen et aI., 1989) were not successful. The libraries were also screened for the presence of extended clones by the polymerase chain reaction (Friedman et aI., 1988) by using a "-gt11 sequencing primer and oligonucleotide primers derived from the 5' end of the laminin A chain cDNA, but no extended cDNAs were obtained. We therefore decided to test different human cell lines known to express laminin (Alitalo et aI., 1981) for the presence of A chain transcripts by Northern blot analysis and then construct a cDNA library from a positive cell source. A human placental choriocarcinoma line, JAR, expressed a - 10-kb transcript when lOllg of poly(A)+ RNA was analyzed by Northern blotting using a 700-bp fragment from the 5' end of the 2.2-kb human A chain cDNA clone (Olsen et aI., 1989) as the probe (not shown). Two other less prominent transcripts of about 7.5 kb and 5 kb were also detected. These bands may represent transcripts coding for variant or truncated laminin A chains. Northern blotting of HT29, T24, HTl080 or MG63 cell RNA (Alitalo et aI., 1981) showed very low or undetectable A chain message levels. We also analyzed RNA isolated from rabbit tissues, including liver, lung, tongue, kidney, and skeletal muscle, human placental and brain RNA (Clontech, Palo Alto, CAl but failed to detect any A chain transcripts, even at relatively low stringencies. We thus concluded that the JAR cells were the best source of A chain mRNA and the best candidate for library construction. Random oligonucleotide or oligo(dT) primers were used to construct two separate cDNA libraries which were screened with a 700-bp cDNA fragment from the 5' end of the 2.2-kb human A chain cDNA. Fifty-two positive clones were isolated from the random-primed and 20 from the oligo(dT)-primed library. This process was repeated four times by using a fragment from the 5' end of the most 5' sequence from each round until the entire sequence was obtained. Clone 123 was found to contain a 179-bp deletion between nucleotides 1045 to 1225 (indicated by a break in the line representing this clone in Fig. 1). This deletion was most probably a cloning artifact because it created a change in the reading frame and it was not present in clones 126 and 129. A partial restriction endonuclease map of the cDNA and the positions of the clones which were sequenced in their entirety are shown in Figure 1. DNA sequence analysis
A total of 72-overlapping cDNA clones were isolated and their ends were sequenced. Clones 7,67,123 and 129 were sequenced in their entirety and found to have a considerable sequence homology to the mouse A chain cDNA (Sasaki et
154
T. Haaparanta et al. 3'
5' EBK !
K
S
S
I
i
i 4
3
2
H
ii
5
H
H
ii
i
6
7
i
KE S
!I'
i
i
8
9
(A)n kb
123
A
7
129
(A)n
67
126
11
f - - - - - - - - (A)n
19
1------(A)n
36
aI., 1988). The overlapping clones span 9505 nucleotides of sequence with an open reading frame of 9225 nucleotides that encodes a 3075-amino acid polypeptide of Mr 337,163. The sequence around the putative initiator methionine at position 78 is a likely site of initiation of translation. The sequence GGC GAG ATG C is in agreement with the Kozak consensus sequence of translation initiation by eukaryotic ribosomes (Kozak, 1987). This initiation codon is followed by a sequence that codes for an apparent signal sequence (von Heijne, 1986). Analysis of this sequence using the program SIGNAL (PC GENE) predicts two possible signal peptidase cleavage sites between amino acids 17 and 18 or 19 and 20 that conform to the (- 3, -1) rule. We have therefore chosen to number the deduced amino acid sequence beginning with the initiation methionine as the first amino acid in Figure 2, rather than predicting the beginning of the mature polypeptide. The 77-nucleotide cD NA sequence upstream of the ATG codon is characteristically GC rich (77% GC) and does not contain any ATG codons. The 3' noncoding region contains 190 nucleotides before the poly(A) tail. There is one polyadenylation signal (ATTAAA) 13-nucleotides upstream of the poly(A) tail. The protein coding sequence contains 34-potential Nglycosylation sites and 59-cysteine residues (excluding the signal peptide sequence). The putative cell-binding site, IKVAV (14), is present at amino acid position 2116 and an RGD sequence (Ruoslahti and Pierschbacher, 1987) is found at position2534. Recently, a tripeptide sequence, LRE, was identified in s-laminin, which is a Bl-related laminin polypeptide, as a motor neuron-selective attachment site (Hunter et aI., 1989a, b). This sequence occurs three times in the human A chain at positions 1789, 2050 and 2691. Olsen et al. (1989) has previously published a partial sequence of the human laminin A chain covering the carboxyl-terminal 2242 nucleotides. That sequence and the
Fig. 1. Map of human laminin A chain eDNA. A partial restriction enzyme map and schematic representation of the overlapping eDNA clones are shown. Cleavage sites for Eco Rl (E), Bam HI (B), Kpn 1 (K), Sma 1 (5), and Hind III (H) are indicated. The thick part of the upper line indicates the open reading frame. The break in the line representing clone 123 indicates the position of a deletion covering nucleotides 1045 -1225 in this clone.
sequence presented in this paper are identical except for differences at ll-nucleotide positions. Specifically, C at position 7263 (rather than T) results in an arginine instead of a tryptophan residue; C at position 8315 (rather than A) results in phenylalanine instead of leucine; G at position 9062 (rather than A) results in no amino acid change; Tat position 9237 (rather than C) results in phenylalanine instead of leucine. Nucleotides 9293-9300 are GACCGAGT in this paper rather than ACCGAGTC in Olsen et al. (1989) which changes the C-terminal amino acids to tyrosine-glutamic acid-serine instead of proline-serine-proline. Finally at position 9374 in the 3' non-coding region our sequence has an additional A (AGAATC rather than AGATC). Unless these differences are due to cloning artifacts, they may reflect differences between the JAR cell and placentallaminin A chains or represent allelic variants. Comparison of human and mouse laminin A chain and merosin amino acid sequences
The deduced amino acid sequences of human and mouse laminin A chains and the merosin heavy (M) chain were analyzed using the PALIGN program. The mouse laminin A chain polypeptide has been predicted to consist of 8 structurally distinct domains. Similar analysis of the human A chain predicts an identical domain organization. The overall amino acid sequence identity with the mouse A chain is 76%. Furthermore, all cysteine residues, except 2 in the putative signal peptide sequence, are conserved, suggesting a highly similar three-dimensional structure between the two molecules. The mouse sequence is 14amino acids longer than the human, but 7 of these extra residues are found in the putative signal peptide. Table I shows the degree of sequence similarity in the different domains between the human and mouse sequence. The a-helical rod portion (domains I and II) which is thought to interact with the B chains is the least conserved
Human Laminin A Chain Sequence
155
cxacTcn:G~OCOCGGCT~l\ICGCGliGGGCGIa:ta::1'(X;fCft'OCrocrorol'GlUiCCOCOCAG-I~ MRGGVLLVLLL@VAAQ@~~J\GLFPAILN
za
162
~'rGC'1'CACArcNrACX:AA~~TG'l'OC'CAOO'I'C'OOCCro~;.pa::roA~TGOCAACAOCOCAAN:CCCAGAGAN;OCCA'I'CCAAT L A S N A H I S T ~ (£) G E X G P E H F (£) K L V E H V P G R. P V R N P 0 (9 R I (9 0 G N SAN P R E R H P I
82
323
ATCACA.'l'OCCATAGA~TAACTOO'I'GOCMAGrc:cr::AGC7l'l'TCAGAA~TATCAC'l'GGGTCACAA~T"I'AAGACAGGTC'l'TTCMGTTGCATAro'I'CATCAlTAAAGC"l"GCCAA~TGGA.
__ !Gi\OOCCTGl"I"l'CC'l"'1"TC.'1'CAAT
.8.
R P G N W I
136
'lTl'~TCTCTGGA.'l'GOCACCACGT'l'CAoccx:cT'QX:AGTATTA'roC'.AGTCAOC~T'I'G'l'C'l'CGTTACAATATAAC't'CCMGACGAooa:cACCCACX:TACAooa:TGATGATGAAG'l'GA.TC~rccrATTATTCCAGATTG L E R S L O G T T F S P W 0 Y Y A V S 0 S E (£) L S R Y ~ P R R G P P T Y R ADD E V I (9 T S Y Y S R L
S H A lOG
T NNW W Q S P S I
Q N G R E Y H W V TIT
L D L R 0
V F 0
V A Y V I
I
K A A NAP
189
.. 5
G'I'GCX:AC'l"I'GP.OCAl'OOItGAGATTCArACATCACTCA'l'CM~l'CTn'C.ACXX:A1IGlTGTTOGAAT1CAC'I'1'CTOCN:XaTATA'l"I'CGO:'I'T
