Apr 25, 2018 - From the Center for Human Genetics, University of Leuven, Campus Gasthuisberg 0 & N, .... primers for pGEM-3Z and SK and KS primers (Genofit, Geneva, ...... Powell, L. M., Wallis, S. C., Pease, R. J., Edwards, Y. H., Knott, T.
THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 264, No. 12, Issue of April 25, pp. 7017-7024,1989 Printed in U.S.A.
Partial Primary Structureof the 48- and 90-Kilodalton Core Proteins of Cell Surface-associated Heparan Sulfate Proteoglycans of Lung Fibroblasts PREDICTION OF AN INTEGRAL MEMBRANE DOMAIN AND EVIDENCE FOR MULTIPLE DISTINCT CORE PROTEINS AT THE CELL SURFACE OF HUMANLUNG FIBROBLASTS* (Received for publication, August 10, 1988)
Peter MarynenS, J i Zhang, Jean-Jacques Cassiman, Herman Van den Berghe, and Guido David5 From the Center for Human Genetics, University of Leuven, Campus Gasthuisberg 0 & N, Herestraat B-3000 Leuuen, Belgium
Cell surface-associated proteoglycans appear involved in a Heparitinasetreatment of cell surface-associated heparan sulfate proteoglycans (HSPG) of human lung variety of biological processes. These pertain to the mechafibroblasts revealscore proteins with apparent M , val- nisms of cell-matrix (Koda et al., 1985) and cell-cell (Cole et ues of 125,000, 90,000, 64,000, 48,000 and 35,000 al. 1985) adhesion, to the control of cell growth (Fritze et al., (Lories, V., De Boeck, H., David, G., Cassiman, J.-J., 1985; Ratner et al., 1985), the activation of proteinase inhiband Van den Berghe, H. (1987) J. Biol. Chern. 262, itors (Low et al., 1981), the regulation of receptor function 854-859). The 90- and 48-kDacore proteins share the (Fransson et al., 1984), antigenpresentation (Sant et al., epitope of the monoclonal antibody 6G12 which was 1985), and even the generation of biological rhythms (Jackson used to screen a human lung fibroblast expression et al., 1986). cDNA library. In the execution of these processes, different modesof Rescreening of the librariesyielded clone 48K5 with association of the proteoglycans with the cell surface seem to an insert of 3439 base pairs. Polyclonal antibodies exist. Some forms are periferally membrane-bound to recepwere raised in rabbits against a fragment of the protein tors for the glycosaminoglycan chains (Kjell6n et al., 1980) or encoded by the 48K5 cDNA different from the part for structures on the core protein (Glosll et al., 1983). In the carrying the 6G12 epitope. These antibodies specifi- latter instance, the membrane association may only be very cally recognize the 90- and 48-kDa core proteins on transient, representing proteoglycan destined for internaliWestern blots of total cellular extractsof human lung zation and furtherprocessing (Bienkowski and Conrad, 1984; fibroblast HSPG. The specific reactivity of the poly- Ishihara et al., 1986). Other cell surface proteoglycans, howclonal antiserum confirms the identity of the 4 8 3 5 ever, seem endowed with properties that allow for a more clone and further distinguishes the 48- and the 90-kDa direct insertion into the membrane; some of these may be core proteins,which do share the 6G12-defined epitope linked to theplasma membrane through an inositol phosphoand at least one additional antigenic determinant with lipid anchor (Ishihara et al., 1987), but others are presumed the 48K5 cDNA-encoded protein, from the 125-, 64-, to possess hydrophobic core protein segments that span the and 35-kDa core proteins of cell surface HSPG of membrane (Rapraegerand Bernfield, 1985). Indirect evidence human lung fibroblasts which do not react with either antibody preparation.Theprotein encodedby the for the existence of integral membraneforms also stems from 48K5 clone contains a stop-transfer sequence indica- the tentative identification of some core proteins as alternative of an integralmembrane protein and threepoten- tively processed membrane glycoprotein: the class I1 (Ia) histocompatibility antigen-associated invariant chain in lymtial glycosaminoglycan attachment sites. The 48K5 clone detects two major poly(A)* RNA phocytes (Giacoletto et al., 1986) and thrombomodulin in species in human lung fibroblasts presumably gener- endothelial cells (Jackman et al., 1986). For a large part, the functional properties of these proteoated by the use of alternative polyadenylation signals. The 48K5 gene was mapped to chromosome 8q23 by glycans seem to depend on the charge and the structure of in situ hybridization and hybridization to DNA of so- the glycosaminoglycan moieties. The latter comprise mostly heparan sulfate but also chondroitin sulfate/dermatan sulfate matic cell hybrids. chains (see Fransson, 1987). These seem to mediate or, to the * This investigation was supported inpart by Grants 3.0066.87 and contrary, interfere with the binding of a number of different 3.0088.88 of the Fonds voor Geneeskundig Wetenschappeliijk Onder- ligands: linking cells to collagen (Koda and Bernfield, 1984) zoek, Belgium, by United States Public Health Service Research and fibronectin (Laterra et al., 1983) in the matrix, catalytiGrant HI-31750 (to G. D.), and by the Inter-university Network for cally inactivatingproteinases of the coagulation cascade Fundamental Research sponsored by the Belgian Government (1987- (Marcum et al., 1986), or shielding receptor structure in 1991). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby confluent cells (Coster et al., 1986). Thetransformationmarked “aduertisement” in accordance with 18 U.S.C. Section 1734 associated changes of these complex carbohydrates (Hook et solely to indicate this fact. al., 1984; David and Van den Berghe, 1983) may therefore The nucleotide sequence(s)reported in thispaperhas been submitted have profound implications for proteoglycan function and to the GenBankTM/EMBL Data Bank withaccession numbeds) cellular behavior. Besides providing a means to concentrate 504621. $ Bevoegdverklaard Navorser of the Nationaal Fondsvoor Weten- and anchor glycosaminoglycan chains at the cell surface, the core proteins may also have a functional role of their own schappeliijk Onderzoek, Belgium. Onderzoeksleider of the Nationaal Fondsvoor Wetenschappelijk right, e.g. as cofactor for the activation of protein C (Bourin Onderzoek, Belgium. To whom correspondence should be addressed. et al., 1986) or by providing a transmembrane link between
7017
7018
Structure of an Integral Membrane Proteoglycan Core Protein
the cytoskeleton and theextracellular matrix (Rapraeger and Bernfield, 1982; Woods et al., 1985). The evidence, however, is indirect, and the information on core protein properties is too limited for an evaluation of these possibilities. Accumulating data suggest, perhaps not surprisingly, that this functional versatility of the cell surface proteoglycans is accompanied by an outspoken structural diversity (see Fransson, 1987). Moreover, we have obtained evidence that structural heterogeneity occurs among membrane proteoglycans of a single class within a single cell type. Indeed, unreduced preparations of hydrophobic heparan sulfate proteoglycan of confluent human lung fibroblasts harbor multiple core protein forms with apparent M , values of 125,000,90,000,68,000, 48,000, and 35,000 after heparitinase digestion (Lories et al., 1987). The origin of this heterogeneity is not clear, but it cannot simply be accounted for by the heparitinase treatment itself (Lories et al., 1987, 1989). Comparative peptide maps and the reactivity of a panel of monoclonal antibodies imply that more than one single proteoglycan species must exist (Lories et al., 1989). However, the two distinct epitopes which are recognized by the monoclonal antibodies F58-6G12 and F58-10H4 are common to both the 90- and the 48-kDa core proteins, suggesting thatthelatter may be related. The present report describes the molecular cloning of a partial cDNA sequence, "derived" from human lung fibroblasts, which seems to encode a peptide that is common to both these 48/90-kDa coreproteins. The sequence suggests the existence of a carboxyl-terminal intracytoplasmic domain, a stop-transfer sequence with a hydrophobic transmembrane segment, and an extracellular domain with several potential glycosaminoglycan chain acceptor sequences. The recombinant core protein cross-reacts immunologicallywith the 48/90-kDa core proteins but not with the other forms, which is further support for the existence of multiple cell surface heparansulfate proteoglycan species in these lung fibroblasts. MATERIALS AND METHODS
Isolation and Analysis of Poly(A)+RNA-The isolation of total RNA from human lung fibroblasts was modified from Maniatis et al. (1982). Typically, 2 X 10' fibroblasts were solubilized with 25 ml of 5 M guanidine isothiocyanate, 5 mM sodium citrate, 0.2 M 2-mercaptoethanol, 0.5% N-lauroylsarcosine. DNAwas sheared by repeated passage through a 20-gauge needle, 10 g of CsCl was added, and 7-ml fractions of the extracts were layered on 5.5 ml of a cesium trifluoroacetate solution (Pharmacia LKB Biotechnology Inc.) with a density of 1.51 and spun at 30,000 rpm for 24 hinaSW 41 rotor (Beckman Instruments). The RNA pellet was solubilized in 10 mM Tris, 1 mM EDTA, 5% phenol, 5% N-lauroylsarcosine, phenol-extracted once, and ethanol-precipitated twice. Poly(A)+ RNAwas isolated by two rounds of chromatography on oligo(dT)-cellulose (Bethesda Research Laboratories) according to Aviv and Leder (1972) and stored as an ethanol precipitate until further use. The absence of degradation of the RNA in the preparations was checked by Northern blotting. For Northern analysis, poly(A)+ RNA was denatured and separated on 1.2% agarose gels containing 0.2 M 3 - ( N morpho1ino)propanesulfonic acid, 0.05 M sodium acetate, pH 7.0, 10 mM EDTA, and 6% formaldehyde as described in Maniatis et al. (1982). RNA was transferred to Nytran membranes (Schleicher and Schuell) following the specifications of the manufacturers and crosslinked to its support by exposure to UV light (Church and Gilbert, 1984). Preparation and Screening of Libraries-The initial (48K1) cDNA clone was isolated from a human lung fibroblast cDNA library cloned into the expression vector XGTll (Young and Davis, 1983) obtained from commercial sources (Clontech, Palo Alto, CAI. The phages were grown in Y1090 r- host cells, induced with isopropyl 0-D-thiogalactopyranoside, and selected by blotting and immunostaining of the nitrocellulose filters with conditioned culture medium of the F586G12 hybridoma and alkaline phosphatase-conjugated affinity-purified goat anti-mouse antibodies (Promega Biotec). Additional clones were obtained by screening the fibroblast cDNA library and a com-
mercially available placenta XGTll library (Clontech) using random primer-labeled (Feinberg and Vogelstein, 1983) cDNA inserts and DNA hybridization. The cDNA for the XZAP library was prepared from 2 pg of lung fibroblast poly(A)+ RNA with a cDNA synthesis kit (Bethesda Research Laboratories). After methylation of internal EcoRI sites, ligation of EcoRI linkers, and digestion with EcoRI, the cDNA was separated by gel permeation chromatography on a BioGel A-50m (Bio-Rad) column cast in a siliconized glass 1-ml pipette (Huynh et al., 1985). The fractions containing the largest cDNAs werepooled (25% of thetotal cDNA), ethanol-precipitated, and ligated into dephosphorylated XZAP arms (Promega Biotec). The ligated DNA was packaged, plated onto BB4 host cells, and screened without further amplification, Analysis of the cDNA Clones-The inserts of XGTll clones were subcloned into pGEM-3Z (Promega Biotec). Bluescript SK- plasmids were obtained from the XZAP clones after superinfection with R408 helper phage as described by the manufacturer. Sequences of the inserts were obtained using the dideoxy chain termination method (Sanger et al.,1977) by direct sequencing of supercoiled plasmid (Chen and Seeburg, 1985) with SP6 and T7 primers for pGEM-3Z and SK and KS primers (Genofit, Geneva, Switzerland) for Bluescript SK- plasmids. To obtain the full sequence of both strands, exonuclease III/mung bean nuclease deletion subclones were prepared as described by Henikoff (1984). To alleviate the severe compression problems resulting from the GC-rich regions at the5' endof clones 48K5 and 48K3, sequencing of these fragments was done using 7"deaza-GTP (Misusawa et al., 1986) and Klenow enzyme and, after subcloning into M13, with dITP and Sequenase (United States Biochemical Co.). Gene Localization-The selection and karyotyping of human mouse somatic hybrids has been described elsewhere (Zhang et al., 1988). Somatic hybrids were rekaryotyped at the moment of harvest for DNA purification. The isolation, digestion, and blotting of DNA were according to established procedures (Maniatis et al., 1982). Human DNA, digested with EcoRI, yields two fragments of 23and 3.2 kilobase pairs when hybridized with the 48K3 probe. Under similar hybridization circumstances, murine DNA digested with EcoRI reveals an easily distinguishable pattern of DNA fragments of, respectively, 12.5, 5.1, 4.4, and 3.1 kilobase pairs. In situ hybridization of metaphase chromosome spreads of human white blood cells with [3H]dCTP-labeled48K3 was as described by Harper and Saunders (1981). Construction of Expression Plasmids and Isolation of the Hybrid Proteins-The coding EcoRI-PstI fragment of cDNA clone 48K1 (the equivalent of base 954-1385of clone 48K5) and the BamHI-PstI fragment of clone 48K3 (base 602-1385 of 48K5) were ligated into the plasmid expression vector pEX2(Genofit) digested with the corresponding restriction enzymes. Transformed POP 2136 cells were grown overnight at 34 "C on ampicillin plates. Colonies containing recombinant plasmids were detected by screening temperature-induced replica cultures for the production of 0-galactosidase-core protein fusion products by immunostaining with Mab' F58-6G12 and by DNA hybridization with oligolabeled cDNA insert. Fusion proteins for immunization and for use in affinity chromatography were isolated from exponentially growing cultures in LB medium, started from single colonies of transformed POP 2136 cells and induced by shifting the culture temperature for 2 h from 34 to 42 "C. Purification of the fusion proteins encoded by both recombinant plasmids was facilitated by the poor solubility of these components. After pelleting (10,000 X gav,15 min at 4 "C) and rinsing in 0.15 M NaCl, 50 mM Tris-HCI, pH 8.0, by resuspension and centrifugation, the cells were sonicated for 2 min in ice-cold rinse buffer containing 1 mg/ml lysozyme. The sonicated pellet was diluted in 10 volumes of ice-cold 0.5% Triton X-100, 6 M urea, 50 mM Tris-HC1, pH 8.0, supplemented with 1 pg/ml pepstatin A, 25 pg/ml leupeptin, 5 mM EDTA, 25 mM 6-aminohexanoic acid, 1 mM phenylmethylsulfonyl fluoride and extracted for 10 min at 4 "C. After centrifugation (25,000 X gav,30 min, 4 "C), the residual pellet was further extracted for 48 h at 4 "C in 4 M guanidine hydrochloride, 10 mM Tris-HC1, pH 8.0. This 4 M guanidine hydrochloride extract was cleared by centrifugation (40,000 X gav,30 min, 4"C) and was shown by Western blotting and immunostaining with Mah F58-6G12 to contain large amounts of fusion proteinsand, in comparison with the ureum extract, low amounts of bacterial proteins. The fusion proteins were further purified from these 4 M guanidine hydrochloride extracts by The abbreviations used are: Mab(s), monoclonal antibody(ies); bp, base pair(s); HSPG, heparan sulfate proteoglycan.
Structure of an Integral Memblrane Proteoglycan Core Protein gel filtration chromatography on Sepharose CL-4B or -6B in 4 M guanidine hydrochloride and by ion exchange chromatography on Mono Q (Pharmacia) in 0.5% Triton X-100, 6 M urea, 50 mM TrisHCI, pH 8.0. Elution was monitored by immunodot spotting using Hybond-N membranes (Amersham Corp.), Mab F58-6G12 (20 pg/ ml), and peroxidase-conjugated rabbit anti-mouse immunoglobulins (diluted 1:50, Dakopatts). Preparation of Polyclonal Antirecombinant Core Protein Antibodies-Purified pEX2-48K3-encoded fusion protein was suspended in phosphate-buffered saline, mixed with an equal volume of Freund's complete adjuvant, and injected subcutaneously into rabbits. After three additional injections with fusion protein-Freund's incomplete adjuvant mixtures, immune sera were obtained from ear bleedings. Pooled sera were incubated overnight with purified pEX2-48K3 fusion protein coupled to CNBr-activated Sepharose. After rinsing the beads, bound immunoglobulins were eluted with 0.15 M NaC1, 0.1 M glycine HCl, pH 2.0, and collected in tubes containing 0.1 ml of 2 M Tris-HC1, pH 8.0. To remove antibacterial specificities, the affinitypurified immunoglobulins were mixedwith 0.2 volume of 0.5% casein in phosphate-buffered saline and incubated overnight with CNBractivated Sepharose beads substituted with extracts from POP cells transformed with nonrecombinant pEX2 plasmids. Nonbound immunoglobulins were further absorbed on pEX2-48K1 fusion protein coupled to CNBr-activated Sepharose to obtain an antiserum specific for (core) protein epitopes other than the F58-6G12 epitope used for the initial selection.
gene and therefore coding for the 6G12 epitope, was found. A probe generated from the 5' end of 48K1 was used to screen both the original fibroblast cDNA library and a placental cDNA library. One clone, 48K3, with a 3373-bp insert, did contain the complete sequence of clone 48K1 and other 48K clones selected as determined by restriction mapping and Southern hybridizations. Sequencing did not reveal an initiator AUG codon at the 5' end of clone 48K3, and the open reading frame did code for a proteinwith a molecular mass of 41 kDa. Therefore, a size-selected cDNA library was constructed from poly(A)+ RNA isolated from human lung fibroblasts. The cDNA was size-selected by gel permeation chromatography and cloned into XZAP vector using EcoRI linkers. About lo6 independent plaques of the unamplified library were screened with a 500-bp probe isolated from the 5' end of clone 48K3. Of the 13 clones isolated, one clone 48K5,with a 3439-bp insert, contained additional information at the 5' end. The sequence of this clone, obtained by sequencing both strands, is shown in Fig. 2. 48K5 has a poly(A) tail preceded by the AATAAA polyadenylation signal 20 bp upstream. Two GC-rich regions were found extending from base 50 to 120 and base 260 to 560. No particular homologies were detected by searching GenBank release 54. Characteristics of the 48K5 Gene Product-Clone 48K5 also lacks an initiator AUG codon. Translation of the 1191-bp open reading frame at the5' end (Fig. 2) predicts a protein of 43 kDa with properties compatible with its tentative identification as part of the 48/90-kDa core proteins. Indeed, near the carboxyl terminus thereis one long stretch of24 hydrophobic amino acids followed by a short stretch rich in basic residues (4 out of 6) (see Fig. 2). This domain has the structure of a stop-transfer signal (Sabatini et al., 1982) and could thus constitute the membrane-spanning do-
RESULTS
Molecular Cloningof a Presumptive 48/90-kDa Core Protein cDNA-A human lung fibroblast cDNA library, cloned into the expression vector XGT11, was screened with the monoclonal antibody 6G12. One positive clone, 48K1, was plaquepurified, and the 1317-bp insert was subcloned into pGEM32 andsequenced (Figs. 1and 2). The orientation of the insert in the original XGTll-48K1 clone was determined from the restriction map, and an open reading frame of 265 base pairs, continuous with the open reading frame of the Xgtll lac Z l
I
0
1000
3000 bp
2000 I
FIG. 1. Analysis of the 48K clones. The position of48K clones is shown relative tothe 48K5 sequence. The position of the TAA stop codon is indicated by A. AATAAA sequences, possibly used as alternative polyadenylation signals, are indicated by A. The recognition sites for selected restriction enzymes are shown. (A)n indicates the presence of a poly(A) sequence. The thick lines delimit the GC-rich regions in the 48K sequence. The cross-hatched boxes show the 48K1 and 48K3 fragments used for the production of fusion proteins. The open boxes delimit the probes used for Northern analysis.
7019
I
48-7A4
4
4
I
"-
48 K 1
-
48-2E4
( A ) n 48-564
,
1
I
I
-- -
-
48-764
48-lA4
48-iA4
Structure of an Integral Membrane Proteoglycan Core Protein
7020 GGC AGG AGG G R R E
GAG G
GCC
CCA
A
GCG A
E
CCC P
CTA L
N
GGA G
GGA G
AGT S
GAG E
GAG E
TCC
CCC
AAC F
GTG
K
AGG A
S
L
TCG
S
GTG
V
A
TTC
A
P
R
R
CTG S
E
A
GCA
GAG
TCG R
AGA
A
S
N
K
D
M
GAG R
TAC
CTT
L
N
AAC
GCG
L
AGC
TCC
I
ATT
E
E
CCT P
ATT I
CCA P
AAG K
GAA E
D
D
ATA I
GAA E
GAT
T
L
ACT D
GAC
GAT
D
CTG
D
TTG
L
T
E
TAC
A
ACT A
A
GTT
K
V
K
L
AGT L
CTG AAA TTTCGG ACA GAA F K R T E V
CTG
GTG
TAT
CTG L
L
TTG V
CCT ACT AAG P T K E
Y
R
CGC M
R
ATG K
GAG TTT TAT GCG F Y A *
G
AGA K
TAA
E
TCT S
CAC TCA GAC H S D S
L
TCT
S
V
CTC
H
GCG
G
GTC L
AAG D
AAC
S
G
TCC
GAT S
AAC
GGA
E
ACG
T
TCA E
TCG D
D
ACG
T
K
GAA Y
TTA
GTC A
D
V
N
D
GTG
AAT
CCA
N
E
TAT G
TCT
E
CTA
E
CAG K
GCC
A
GAT
P
ATA
Q
P
GAG
S
ATT GCT GGT G G V I
AGC L
GAT E
I
GAC
M
GGA
GCT
CTG
L
GAA AAA AGG ATG R
CTA GCA GCT A A V I
AAG E
ACC T
GAC
D
GGC A
GCT AAA GTG CCA GAA
P
CAC
TCT S
GCT
A
AAA
.
GCT
S
AGT
S
GAT AAA GAG K
GAC Y
AAC I
GAA E
GGA G
GTA L
AAG
P
GAG D
GAG T
GAT
T
N
ACA V
TTT
K
P
GTT ATT GGC F L F A
ATG
S
AGA
S P@A TCA
A
T
A
R
AGC
CCC
G
GCA 540 GGA
GGC
T
TTG
G
GCT Y
TCA
GGA 720 GTG
CTG
GTG
CAG
ACA
900 AAG
1 0 8 0TTT ATT
170 AAG 1GCA 390
GAA
CTT
ATC
360
A
TAA CTG*C AAC TTT TTT TGC TCA ATA
CCT
990 GAG
CAG K
TCA
330
K
TAT
CG
300
P
CTC TTT GCA F L I
Q
ACT
270
TAT AAAACT
E
TAT
810ACA ACA
L
S
GGC
240
Y
GCT
T
630 TTG
ACC
210
GAG
K
AAT
TTT I
V
CCG 450
180
A
P
CCT
TGG
CTC
GCT
CCA
T
Y
GAC AAA CTT CCA GGA TCC GAA AGT CGC GCT R
G
AGT
Q
G
150
V
V
GAA
CCC
CGC
CTG
L
-
ATA
A
GCC
270 90
P
F
ATC
S
T
GCT 180 CCC
120
GCG
GAG
G
GAA A
CGC
T
E
GCC
G
“ ” “ ” _
CGC
R
CGA
P
G
L
GCA
60
GTC
P
AAG
TGG
T
GGC
30
GTG AAA TCG CCA GGA ACA GTG 360 CAG
S
A
AGG 90 AGG
L
V
K
CTC
W
CGC
L
S
GGG
R
CGG
P
TCG
A
G
Q
AGG
P
CGC
E
I
S
A
GTG A
W
GAC
D
R
K
TCC
V
AGG
GGA
A
CAC AAAGGG GGA G
AAC
G
AGG
P
CGG
ACC
CGC
A
H
S
CGC
ATG
A
R
K
L
AAT
CCT
P
CGC GGG AGG R G A
CCG
R
GAA
V
CGT
G
CAG
S
AGC
S
R
ATT
R
CGG
AGG
R
CTC
G
E
N
AGC
GGG
R
T
TCC T
R
CTG
Y
CCT
CGG
V
ACA AAATCT GAC GAT ATG
D
P
E
R
CTC
M
P
GGG
R
R
TCC
G
AGG GAG GCG CGG R R G R R
CCG
Q
S
GGC R
I
CGG
L
R
CTG
R
AGC
S
CTG
T
ACT
L
CAG
S
S
CCC
L
K
GAG
GGC
G
GGT
L
R
GAG
L
G
AGC
S
GCA
E
T
GGG
S
G
GCG
ATA
E
L
P
GAC CCG GGG G R E A
ATT AAA TAT CCA TCC GAA TTA ACT
S
AGA Q
GCT
TGC
G
GCT
G
A
CGC
C
GCG
A
GCA
A
AAT
R
E
P
CGA
Y
TTC
R
GGC
A
CGC
AGC
F
CAG
P
CAG
I
GAG P
CTC
I
AGC
R
AAG
GCT
Q
TCA Q
S
CGG
GCT
L
CTT S
K
G
N
A
AAG
K
TCC
AGC
A
S
AGC
R
R
CAA
Q
GCG
S
S
GAG AAG GAG K E E D
CGC
A
TCC
L
S
GGA
R
AGC
CAG
CGT
CTT
G
S
Q
CCG
L
TTG
CGT
CAG
P
CCG
L
S
Q
GCG
AGC
GCC P
TGC
C
GCA AAA CCA
TGG
W
V
P
CCG
P
P
TGC C
R
S
TTG L
A
GAG
GGA GCC AGAAAAGGA GAA GAG GAG A R G K E E E E
1260 397
TAA TGA AGA TCT TTG TTT TTT GTT AAA TTC GGA ATT CCA TTC TGG CAC TTT AAT AAA GAT ATC CCA TTG TAT AAA TTA CAT TTC ATG TAT 1 3 5 0 TTC TTT AGA ACA ACA TAA AAT TAA AAT TTA ACA TCT TGT GCA GAA GTG TAG TTC CAG TGG CAA AAT ATT ATG AAA TTACCC TGATCG ATG 1 4 4 0 1530 AAA TAC AAA CTC GAT GAA AGC TGT TTC ATT TGT GTC TTC ATG GAA TTG GTT TAA ACT TTT ATG CGCAAA ATG ATT GTC TTT TTC CTA TGA 1 6 2 0TGC AGC ATG TCT CAG ATT GAC CTT ACC AAG TTG GTC TTA CTT TGT TGTTAA TGT CCC TTT CTT ATC CCT CTC CTC TGC CCT CCC TTC TTG 1710TTA GTG CCT TAA AAC CAA ACC CTA TGC CTT TTG TAG CTG TCA TGG TGC AAT TTG TCT TTG GAA AAT TCA GAT TAT ATG AAT TGA GGT AAT TTT TCA AAT ATG TAA ACT TTA ACT TCC ACT AAATTG TTT TAT TTA AGT GTC AGA CTA TCC ATT TTA CAC TTG CTT TAT TTT 1800 TCA TTA CCT 90 AAATGT TTG GTA GAT TAT TAC AAA TAC ACC GTT TAG TCA TAT CTA TCT AAT CAG 1 8ATC GTA GCT TTG GGC AGA TTT GCA ACA GCA AAT TAA 1 9 8 0TTG CTG CA TTC TTT TGG GAG GAT TTG ATG TAA GTT ACT GAC AAG CCT CAG CAA ACC CAA AGA TGT TAA CAG TAT TTT AAG AAG 2070 TTG GCC ACT GTA TTT GTT AAT TTC TTG CAA TTT GAA GGT ACG AGT AAA AGAAAT GGTCAG TTATTT AAGTTG TTC AAA TTAATG CAT TTA 2160 AGT TGT AAA CGT CTT TTT AAG CCT TTG AAG TGC CTC TGA TTC TAT GTA ACT TGT TGC AGA CTG GTG TTA ATG AGT ATA TGT AAC AGT TTA 2250 AAA AAA AAG TTG GTA TTT TAT AAG CAC AGA CAA TTC TAA TGG TAA CTT TTG TAG TCT TAT GAA TAG ACA TAA ATT GTA ATT TGG GAA CAT AAA AAC TACT q m CAT GTG GCC TAA TAT AAA TGATGT CACTGT TAT AAA TTT TGT ACA TTT TTG ATC AAA TGT ACA TCT CCC CTT 2340 2430 TGC TAA CGG CCG TCT GCT CTC AGG TTG ACG TGG GTT TGA TTT CTA AGT GTT TCA CAG TGT AAA CTGGAGTAACCTATCGTCAAGGATACC 2520AAG AAA TCT TGT ATT ATC CTG TGT GTG TCT AGG TAG AGA TAT TGG GAG ACT GTT TAT TAC CAG ATT CAC TTC TGA ATT GGC CAG GAA AGG GCT GCC AGG GGA TTT CGA AGT TTG CAA CCT TTA TAG GAT AAC TAT TGA TAA TGG GAC CAA AGA CGC CTG CTT TTG CAA ATA ACT 2610 TAC AAG 2700 ACT GTA AAT TCC AAA GAT CTG AAT GGG GCT TTC CTG ATG TTG GTA TCT AAG GCT TAG GCC TAT AGA TTG ATT TAC CTT TGG AAT TGT 2790 CCA AAT GTC TAC TGA AGC TTA ACC GAA GAA CTA ATA AAT GGA CTA CAG TAG CTC ACG TTA CAG GGA ACC ACC CTA GGC AGG GAG G GAG TGG TGA TGC CTG GGG AAG GAG ATG GAG TTA TGA TGTGGG 2880 TAC TGT GTTAAA ATG AGG GTC TCA CTG CTT TAG GAT TGA AGTAAAGGC 2970AGT CAG AA GGC TGG TAC TTT CTG TAC TAA ACA TTT CCT TTT TCT ATT TTA CCA CTA ATT TTC TTT TAA ACT GTG AGC CGT CCA 3060 GCA AAA AAA GCA ACT TTT CCA ACA TAC AAT TTA CTT TTA ATA AAG TAT GAA TAT TTC ATT TTG AGA ACA TTC CCT GGA ATT GCC 3150CCA TTC ATTAAA AAC ATT TTT TTA AGC AAC ACT TGG AAC AGT GTT AAATAC TCC TTT TTA ATG GCC TTA ATT AAT TCT CAG ATT CCT GCC 3240 AAA CGA TAG CCT AGC TTT CTA AAG CCA CGC TGT GTC CCT CAA TTA CAG AGG TCA CTT ACA GAA CCA ATT CAC TTT AGA AAA GTGAGGACT AAA AGA TTC GTG AAT TCA TAG AAT AAC AAC TGC TAT TGG CCG GAA TGC CAG GAA AAG 3330 GTA GGA ATG GNT ATA CCT CTA ACT GTG CAA AGC 3420 AAA AAT TTC TGG CAA ATA TGT TTTCAC TGC TGT AAA GCA AAA TAT TTG TGA AAG CTGC A A l n ] G T C TGT CAT GCC AAA AGT AAA AAA AAAAAAAAAAAAAAAAAA
FIG. 2. Sequence and derived primary structure of clone 48K5.Clone 48K5 was sequenced as indicated under “Materials and Methods.” Canonical polyadenylation signals are boxed. The presumptive transmembrane domain is underlined. Thicklines underscore the potential glycosaminoglycanattachment sites. An N-glycosylation site is indicated with a broken line.
mainofanintegralmembrane protein. The stop-transfer domain is followed by 33 amino acids which then would form the cytoplasmic domain upto the carboxyl terminus. Assuming similar requirements as for the synthesis of chondroitin sulfate, two Ser-Gly dipeptidesand one Ser-Gly-Ser-Gly tetrapeptide in the extracellular NHn-terminaldomain form three potential attachment sites for the heparan sulfate side chains of the HSPG. This attachment site has been defined as consisting of a few acidic residues closely followed by a Ser-Gly-X-Gly motif with X denoting any amino acid (Bourdon et al., 1987). This is indeed the case for the Ser-Gly-SerGly at position 251 which is closely precededby 4 aspartic
acids. According to the same model, the Ser-Gly at position 114 is unlikely to be an efficient glycosaminoglycan attachment site.The sequencearound the Ser-Gly dipeptide at position 237,however,displays a striking similarity to the sequencefoundaround the glycosaminoglycan attachment site of PG40, the coreproteinof a chondroitin/dermatan sulfate proteoglycanof human fibroblasts(Krusiusand Ruoslahti, 1986) (Fig. 3). Although the second glycine of the SerGly-X-Gly motif is lackinghere, it shouldbe noted that peptides lacking this second glycine still act as a substrate, albeit less efficiently, of xylosyl transferases in uitro (Bourdon et al., 1987). On the other hand, studies using model peptides
Structure of a n Integral Membrane Proteoglycan Core Protein 1
234
D
-
E
A
S
C
I
C
-
E
E
S
A
C
V
P-8
Y
48K5
48K5. Identical aminoacidsandconservativesubstitutionsare indicated. The sequence numbering of PG40 is from Krusius and Ruoslahti (1986). The numbering of 48K5 is as in Fig. 2.
2
3
0
4
. " I "
"I.
P-241
FIG. 3. Sequence similarity between the glycosaminoglycan attachment site of PG40 and a potential attachment site of
1
.. ..". .".... ............................
PC40
I
I I1 I1 I1 II I
7021
100
I " " " .
. " " " . . . . . . . . . . " " . . " I " "
80
. . I "
4
8
40
5
4 8 K I fusion protein
.
9 8 K 3 fusion protein
0
I
POP 2136 pEX2
f4
FIG. 4. Specificity of the rabbit anti-48K3antibodies. Total cell extract of E. coli (POP 2136) carrying the pEX2 plasmid and purified 48K1 and 48K3 fusion proteins (see "Results") were spotted as indicated. The dot blot strips were then challenged with ( 1 ) the monoclonal F58-10H4 (negative control): ( 2 ) F58-6G12Mab's; (3) rabbit anti-48K3 fusion protein; ( 4 ) absorbed rabbit anti-48K3 specificities minus 48K1 specificities, and (5)preimmune antisera. The blots were then developed with the appropriate horseradish peroxidase-conjugated second antibodies and substrate.
40
1
20
CHROMOSOMES A
B
C
D
Mr x
FIG. 7. In situ hybridization with the 48K3
probe. Metaphase chromosome spreads from human white blood cell cultures were hybridized with tritium-labeled 48K3 insert and subjected to autoradiography. 14% of all grains were present on chromosome 8q, which is 4.2 times the amount expected on the basis of a random distribution of the grains.
do not take into account the potential effects of the secondary structure of the protein. A potential N-glycosylation site (Asn-X-Ser) is found a t 97 . position 230. Resistance toward N-glycanase, however, suggests that no N-glycosylation of 48/90-kDa core proteins 68 would occur.' 42 48K5 Codes for Part of the 48190-kDa Core Protein-To confirm the identity of clone 48K5 as coding for part of the 48/90-kDa core protein, the coding sequence of 48K1 (corre2s . 18 . sponding to bases 954-1385 from 48K5, see Fig. 1) and part 15 of the coding sequence of clone 48K3 (correspondingto bases FIG. 5. Specific binding of the absorbed anti-48K3 antibod- 602-1385 from 48K5) were subcloned intothe expression ies to the 48- and 90-kDacore proteins. Cell surface-associated plasmid pEX2 (Stanley and Luzio, 1984). Synthesis of the pheparan sulfate proteoglycans, heparitinase-treated (lanes A, B, and galactosidase fusion proteins was induced in the Escherichia C ) or undigested (lane D),were separated ona 6-26% polyacrylamide coli strain POP 2136, and the appropriate fusion proteins were gel under denaturing circumstances and transferred onto nylon mempurified. A polyclonal antiserum was obtained by immunizing branes. Different slices of the blot were developed with the monoclonal antibody F69-3G10 binding all core proteins (A), the mono- rabbits with the larger 48K3 fusion protein. Anti-48K3 anticlonal F58-6G12 ( B ) ,and the absorbed rabbit anti-48K3 minus 48K1bodies were affinity-purified on a 48K3 fusion protein column antibodies (C and D). and absorbed with E. coli extract coupled to CNBr-activated Sepharose. The anti-48K3 antibodies were then absorbed with the shorter 48K1 fusion protein which carries the 6G12 epitope. 285 Complete out adsorptionof 48K1 specificitieswas monitored with dot blotswith both 48K1 and 48K3 fusionproteins (Fig. 4). These absorbed polyclonal antibodies do react with 48K3 fusion proteins but not with 48K1 fusion protein andtherefore recognize at least one epitope different from the 6G12 epitope 185 which is located on the 48K1 fusion protein. The absorbed antibodies (48K3 specificities minus 48K1 specificities) do react with diffuse high molecular weight species on a Western blot of detergent-extracted lung fibroblast heparan sulfate (Fig. 5, lune D).After heparitinase treatment of the samples, 48K5 I II Ill IV Western blots developed with the absorbed polyclonal antiFIG.6. Northern analysis with 48K5-derived probes. Human lung fibroblast poly(A)' RNA was separated on a denaturing serum reveal the same 90- and 48-kDa species as detected by formaldehyde agarose gel and blotted onto nylon membranes. The 6G12 (Fig. 5, lunes B and C). This demonstrates that the90'
-
membrane stripswere then hybridized with probes derived from 48K5 as indicated (see also Fig. 1).
* G. David, unpublished
results.
7022
Structure of an Integral Membrane Proteoglycan Core Protein TABLE I Chromosome mapping of 48K5 DNA isolated from human/mouse hybrids was digested with EcoRI. The fragments were separated by size on an agarose gel, blotted onto a nylon membrane, and hybridized with a 32P-labeled48K5 cDNA probe. A c-myc probe, mapping to 8q24 (HGM9) was used as a control. denotes the presence of a chromosome in at least 2 out of 10 metaphase spreads or the Dresence of human specific fragments revealed with the c-mvc and the 48K5 probe.
+
Chromsome
Hybrids 1
F49D5S1 LN5S2F43 LN3S3F31 M37 LN5F31C4 LN3BF49S1 F49D3S1 LN3F31D2 V27 F49D3S2 F49C4 FLSN9 LN5G5F31 FL5N35
2
3
4
5
6
7
8
9
10
11
12 17 13 16 15 14
18
19
20
21
22
X
+ + + - + + + + - + + + + + + - + + + + + + + - - - + + + + + + + - + + + + + + + + + - + - + + + + - + + - - + + - - + + - + + " + - + + - " + + + + - + - " + + " - + + - + - - - - + + + + + - - - - + + + - - - - - - - - - + -+ + - " " " " - " + + - " " " + - - + ""_ + - + " + + + - + - - + + - - + + - + + - "
-
+ - "
- -
"
-
-
-
+
""_
- - -
+
+ + - + " " _ - + + - + + + + - - + + + + + - + + + + " " "
"
"
- -
+ -
-
"
-
-
"
-
-
"
-
-
"
-
+ + - + + - - + + + + - - - - - - - - - - + + + + + + - + - + + + - + + + + + + + + +
and 48-kDa core proteins and the 48K5 protein do share, in addition to the 6G12 epitopes, other distinct antigenic determinants detected by the polyclonal antiserum and that in human lung fibroblasts these epitopes are unique for the 90/ 48-kDa core protein of membrane-associated HSPG. We therefore conclude that 48K5 is a partialcDNA clone for both the 90- and 48-kDa core proteins. Northern Analysis-Upon hybridization of Northern blots of human lung fibroblast poly(A)+RNA with 48135, two main RNA species of 4200 and 2900 bases, respectively, were highlighted (Fig. 6). Longer exposures sometimes revealed other minor species. When smaller probes were generated by restriction enzyme digestion of 48K5, all probes derived from upstream of the PstI site at base 1969 did light up both RNA species, while a probe reaching from the XmaIII site at base 2343 to the3' end of 48K5 hybridized exclusively to the4200base mRNA. This indicates that thedifferent mRNA species are generated by the presence of at least different noncoding 3' ends. The 48K mRNAs of Different Size Are Generated by Alternative Use of Polyadenylation Signals-To investigate possible heterogeneity of the 48K clones, the 5' and 3' ends of all clones isolated from the XZAP human lung fibroblast cDNA library in the last selection round were sequenced. No sequence divergences were found at the 3' end. The 3' ends were, however,clearly clustered (Fig. 1).48K3,48K5, and two more 48K clones possess the AATAAA signal at base 3390, followed 20 bases downstream by a poly(A) sequence. Clone 48K1 and four other 48K clones stop aroundbase 2250 of the 48K5 sequence, and another three 48 clones all end around base 1350 of the 48K5 sequence. Inspection of the 48K5 sequence reveals an AATAAA polyadenylation signal starting at base 1240 and one more starting at base 2262 of clone 48K5 (Fig. 2). In addition to this, one clone of each cluster carries a short poly(A) stretch not accounted for in the 48135 sequence. From this evidence and from the results of the Northern analysis reported above, it appears that alternative polyadenylation signals present in the 48K5 sequence are being recognized in fibroblasts to produce different sizes of mRNA. It should be noted that the size difference between mRNAs generating using the 3390 and the 2262 polyadenylation signals is large enough to account for the size difference of the two major mRNA species detected. The 5' ends of the 48K clones clearly cluster in the GC-
"
-
+ + + +
-
+ +
48K5
c-myc
+ + +-
+ + +-
-
-
+ + +
-
-
+
-
+ +
rich regions at the 5' end of clone 48K5. It appears that the problems encountered in obtaining full length clones, even from unamplified libraries, might stem from the apparent difficulty of the murine Moloney leukemia virus reverse transcriptase to copy these sequences. 48K clones containing the 5' end of 48K5 were even totally absent in an avian myeloblastosis virus reverse transcriptase-generated cDNA library (result not shown). Genomic Mapping of the 48K5 Gene-Zn situ hybridization with 3H-labeled 48K5 yielded a strong signal on chromosome 8 (Fig. 7). 896 grains were scored on 429 metaphases; 15% of these were present on chromosome 8. The majority of the grains (98 out of 136 grains on chromosome 8) mapped to 8q23. Southern hybridization with a panel of human/mouse somatic cell hybrids is concordant only with the unique localization of the 48K5 gene onchromosome 8 (Table I). Southern hybridization of DNA obtained from random individuals digested with several restriction enzymes with the 48K5 probe showed simple patterns with a limited number of fragments (not shown) consistent with the presence of a unique 48K5 gene in the human genome. DISCUSSION
The molecular cloning of a partialcDNA for the 48/90-kDa core protein of HSPG from human lung fibroblasts is presented here. The evidence is indirect but conclusive: we have shown that the proteins coded for by 48K5 and the 48/90kDa core proteins do share at least two independent antigenic sites, i.e. the epitope of 6G12, which is unique for the 48/90kDa core proteins in human lung fibroblasts, and the determinant(s) of the polyclonal antiserum raised against a different fragment of the 48K5 protein andwhich selectively stains the 48/90-kDa core protein on Western blots of total cellular extracts of HSPG of human lung fibroblasts. Present work further establishes the occurrence of multiple cell surface HSPG on human lung fibroblasts. It has been documented that theoccurrence of 125-, 90-, 64-, 48-, and 35kDa core proteins is not an artifact generated during purification of these core proteins (De Boeck et al., 1987; Lories et al., 1986, 1987). Monoclonal antibodies do detect different epitopes on each of these core proteins, and at least some of these epitopes seem to be uniquely defined by the protein moiety of the HSPG (De Boeck et al., 1987; Lories et al., 1989). The polyclonal antiserum generated against the 48133
Structure of an Integral Membrane Proteoglycan protein does recognize the 48- and 90-kDa core proteins but not the125-, 6 4 , and 34-kDa proteins. It is therefore unlikely that one gene product is processed to yield the 125,90,68,48, and 35-kDa core proteins. The 90- and 48-kDa core proteins, however, are clearly related although the exact nature of this relation remains to be elucidated. Several hypotheses are still possible. It is unlikely but not excluded that two different genes code for both proteins. The datapresented here indeed indicate that several epitopes detected on the 48-kDa protein are present on the 90-kDa protein, thereby suggesting that the 48-kDa protein sequences are contained in the 90-kDa protein. Sequencing of the 5’ end of all clones isolated in the last selection round and restriction mapping of all clones isolated did not detect any heterogeneity in the coding part of the 48K5 sequence. Furthermore, the in situ hybridization andthe Southern hybridization to human/mouse somatic cell hybrid do indicate that the 48K5 gene maps exclusively to chromosome 8q23, while the simple patterns obtained upon Southern hybridization suggest the presence of a single 48K5 gene. On the other hand, one gene could yield 48- and 90-kDa mRNAs by differential processing of RNAs, or one mRNA could yield bothproteins by posttranslational processing. Cloning of the 5’ end of the mRNA should allow us to elucidate this. To this aim, primer-extended cDNA libraries with primers designed to hybridize upstream of the GC-rich regions of 48K5 will beconstructed. The 48K5 probe detects two main and possibly some minor mRNA species upon Northern blotting of human lung fibroblast poly(A)+ RNA. Present experiments show that at least for the two main species of 4200 and 2900 bases, respectively, the difference could be generated by differential use of polyadenylation signals in the 48K5 sequence. Indeed, sequences 3‘ of the XrnaIII site at bp 2343 are exclusively present in the 4200 nucleotides, and one clone (48-6D4, see Fig. 1)was shown to possess a poly(A) tail downstream of the potential polyadenylation signal at bp 2262. The alternative use of different polyadenylation signals is a known mechanism whereby one gene can generate multiple mRNAs (for review, see Leff and Rosenfeld, 1986; Breitbart et al., 1987). In some cases, the alternative use of polyadenylation signals seems to be linked to differential splicing and thus to the synthesis of different gene products such as the generation of secreted or membrane-bound forms of immunoglobulins (Perry and Kelley, 1979) or the synthesis of calcitonin hormone in thyroid cells and of calcitonin-related peptide in the brain (Amara et al., 1982). It should be noted, however, that all potential polyadenylation signals occur downstream from the stop codon of 48K5 and thatno variability was detected in thecoding region. The alternative use of the different polyadenylation signals seems therefore not to be linked to protein heterogeneity, at least not for the carboxyl-terminal 43 kDa of the molecule. An intriguing example of differential polyadenylation was recently reported by Powell et al. (1987). According to these authors, a single gene generates the message for apoB-100 in the liver and for apoB-48 in the intestine by somenovel mechanism of RNA editing. In addition to this, in the intestine but not in the liver, the alternative use of polyadenylation signals leads to size variability of the mRNAs, and a possible relation between this differential use of polyadenylation signalsand the RNA-editing mechanism was hypothesized (Powell et aL, 1987). The most distinctive feature of the 48K5 protein is the presence of a stop-transfer sequence, suggesting that 48K5 is an integral membrane protein with the extracellular NH, terminus carrying the glycosaminoglycan side chain(s) and a
Core Protein
7023
small cytoplasmic domain with the carboxyl terminus. This feature clearly distinguishes 48K5 from the lung fibroblast chondroitin/dermatana sulfate core protein PG40 (Krusius and Ruoslahti, 1986) with which it seems to share only a glycosaminoglycan attachment site and the core protein of the large molecular weight lung fibroblast chondroitin sulfate proteoglycan (Krusius et al., 1987). Also, no sequence similarity was detected with the sequences of thrombomodulin (Jackman et al., 1986) andthe class I1 (Ia) histocompatibility antigen-associated invariant chain of HLA (Giacoletto et al., 1986), integral membrane proteins which have been defined as “part time” proteoglycans by Fransson (1987). Binding of integral membrane forms of proteoglycans to actin (Rapraeger and Bernfield, 1982) and the cytoskeleton (Woods et al., 1985) has been reported. Cell surface proteoglycans of rat fibroblasts codistribute with stress fibers in fully spread cells and with actin bundles during cell spreading and rounding (Woods et al., 1984). Proteoglycans with integral membrane core proteins would therefore act as link a between the extracellular matrix components and the cytoskeleton, thereby contributing to processes such ascell attachment, cell spreading, and the maintenance of cell shape. The availability of data on the primary structure of the extracellular, the transmembrane, and the cytoplasmic domain of a cell surface-associated proteoglycan will allow us to define further the structure-function relationshipswithin this class of integral membrane proteins. Acknowledgments-We thank Hilde Braeken, AnRayC,Magda Dehaen, and Marleen Willems for their expert technical assistance. REFERENCES Amara, S. G., Jonas, V., Rosenfeld, M. G., Ong, E. S., and Evans, R. M. (1982) Nature 2 9 8 , 240-244 Aviv, H., and Leder, P. (1972) Proc. Natl. Acad. Sci. U. S. A. 6 9 , 1408-1411 Bienkowski, M. J., and Conrad, H. E. (1984) J. Biol. Chem. 269, 12989-12996 Bourdon, M. A., Krusius, T., Campbell, S., Schwartz, N. B., and Ruoslahti, E. (1987) Proc. Natl. Acad. Sci. U. S. A . 8 4 , 3194-3198 Bourin, M. C., Boffa, M. C., Bjork, I., and Lindhal, U. (1986) Proc. Natl. Acad. Sci. U. S. A . 8 3 , 5924-5928 Breitbart, R. E., Andreadis, A., and Nadal-Ginard, B. (1987) Annu. Rev. Biochem. 56,467-495 Chen, E. J., and Seeburg, P. H. (1985) DNA ( N Y ) 4 , 165-170 Church, G . M., and Gilbert, W . (1984) Proc. Natl. Acad. Sci.U. S. A . 69,1408-1412 Cole, G. J., Schubert, D., and Glaser, L. (1985) J. Cell Biol. 100, 1192-1199 Coster, L., Carlstedt, I., Kendall, S., Malmstrom, A., Schmidtchen, A., and Fransson, L. A. (1986) J. Biol. Chern. 2 6 1 , 12079-12088 David, G., and Van den Berghe, H. (1983) J.Biol. Chem. 2 5 8 , 73387344 De Boeck, H., Lories, V., David, G., Cassiman, J.-J., and Van den Berghe, H. (1987) Biochem. J. 247, 765-771 and Vogelstein, P. (1983) Anal. Biochem. 1 3 2 , 6-13 Feinberg, A. Fransson, L. 4. (1987) Trends Biochem. Sci. 12,406-411 Fransson, L. A., Carlstedt, I., Coster, L., and Malmstrom, A. (1984) Proc. Natl. Acad. Sci. U. S. A . 8 1 , 5657-5661 Fritze, L. M. S., %illy, C. F., and Rosenberg, R. D. (1985) J. Cell Biol. 1 0 0 , 1041-1049 Giacoletto, K. S., Sant, A. J., Bono, C., Gorka, J., O’Sullivan, D. M., Quaranta, V., and Schwartz, B. D. (1986) J.Exp. Med. 164, 14221439 Glosll, J., Schubert-Prinz, R., Gregory, J. P., Damle, S., von Figura, K., and Kresse, M. (1983) Biochem. J. 2 1 5 , 295-301 Harper, M. E., and Saunders, G. G . (1981) Chromosome (Berl.)8 3 , 431-439 Henikoff, S. (1984) Gene (Amst.)28, 351-359 Hook, M., KjellBn, L., Johansson, S., and Robinson, J. (1984) Annu. Reu. Biochem. 53,847-869 Huynh, T. V., Young, R. A., and Davis, R. W . (1985) in DNA Cloning
c.,
7024
Structure of an Integral Membrane Proteoglycan Core Protein
(Glover, D. M., ed) Vol.I, pp. 49-78, IRL Press Ltd., Oxford, Great Britain Ishihara, M., Fedarko, N. S., and Conrad, H. E. (1986) J. Eiol. Chem. 261,13575-13580 Ishihara, M., Fedarko, N. S., and Conrad, H. E. (1987) J.Biol. Chem. 262,4708-4716 Jackman, R. W., Becher, D. L., Van De Water, J., and Rosenberg, R. D. (1986) Proc. Natl. Acad. Sci. U. S. A. 83,8834-8838 Jackson, F. R., Bargiello, T. A., Yun, S.-H., and Young, M. W.(1986) Nature 320,185-188 Kjellkn, L., Oldberg, A., and Hook, M. (1980) J. Biol. Chem. 2 5 5 , 10407-10413 Koda, J. E., and Bernfield, M. (1984) J. Biol.Chem. 259, 1176311770 Koda, J. E.,Rapraeger, A., and Bernfield, M. (1985) J. Biol. Chern. 260,8157-8162 Krusius, T., and Ruoslahti, E. (1986) Proc. Natl. Acad. Sci. U. S. A . 83,7683-7687 Krusius, T., Gehlsen, K. R., and Ruoslahti, E. (1987) J. Biol. Chem. 262,13120-13125 Laterra, I., Silbert, J. E., and Culp, L. (1983) J. Cell Biol. 1 0 6 , 112123 Leff, S. E.,Rosenfeld, M. G., and Evans, R. M. (1986) Annu. Reo. Biochem. 65,1091-1117 Lories, V., David, G., Cassiman, J.-J., andVan den Berghe, H. (1986) Eur. J. Biochem. 158,351-360 Lories, V., De Boeck, H., David, G., Cassiman, J.-J., and Van den Berghe, H. (1987) J. Biol. Chem. 262,854-859 Lories, V., Cassiman, J.-J., Van den Berghe, H., and David, G. (1989) J. Bwl. Chem. 264,7009-7016 Low,D.A., Baker, J. B., Koone, W. C., and Cunningham, D.D. (1981) Proc. Natl. Acad. Sci. U. S. A . 78, 2340-2344
Maniatis, T., Fritsch,E. F., and Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY Marcum, J. A., Atha, D. H., Fritze, L. M. S., Nawroth, P., Stern, D., and Rosenberg, R. D. (1986) J. Bwl. Chem. 2 6 1 , 7507-7517 Misusawa, S., Nishimura, S., and Sula, F. (1986) Nucleic Acids Res. 14,1319-1324 Perry, R. P., and Kelley, D. E. (1979) Cell 1 8 , 1333-1339 Powell, L. M., Wallis, S. C., Pease, R. J., Edwards, Y. H., Knott, T. J., and Scott, J. (1987) Cell 50,831-840 Rapraeger, A. C., and Bernfield, M. (1982) in Extracellular Matrix (Hawkes, S., and Way, J. L., eds) pp. 265-267, Academic Press, New York Rapraeger, A.C., and Bernfield, M. R. (1985) J. €501. Chem. 260, 4103-4109 Ratner, N., Bunge, R. P., and Glazer, L. (1985) J . Cell Biol. 1 0 1 , 744-754 Sabatini, D. D., Kreibich, G., Morimoto, T., and Adesnik, M. (1982) J. Cell Biol. 9 2 , 1-22 Sanger, F., Nicklem, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci. U. S. A . 74, 5463-5467 Sant, A. J., Cullen, S. E., and Schwartz, B.D. (1985) J . Zmmunol. 135,416-422 Stanley, K., and Luzio, K. (1984) EMEO J . 3 , 1429-1434 Woods, A., Hook, M., KjellBn, L.,Smith, C. G., and Rees, D. A. (1984) J. Cell Bwl. 99,1743-1753 Woods. A.. Couchman. J. P.. and Hook, M. (1985) J. Biol. Chem. 260,10872-10879 Young, R. A., and Davis, R. W. (1983) Proc. Natl. Acad. Sci.U. S. A. 80,1194-1198 Zhang, Y., Saison, M., Spaepen, M., De Strooper, B., Van Leuven, F., David, G.,Van den Berghe, H., and Cassiman, J.-J. (1988) Somatic dell Mol. Genet. 14,99-104 '
