Chu, M.-L., Mann, K., Deutzmann, R., Pribula-Conway, D., Hsu-. Chen, C. C., Bernard, M. P., and Timpl, R. (1987) Eur. J. Biochem. 1 6 8 , 309-317. Chu, M.-L.
THEJOURNAL OF BlOLOGrCAL
CHEMISTRY
Vol. 266, No. 13, Issue of May 5, pp. 8626-8633,1991 Printed in U.S.A.
0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.
Human a3(VI)Collagen Gene CHARACTERIZATION OF EXONS CODINGFOR THE AMINO-TERMINAL GLOBULARDOMAIN ALTERNATIVESPLICINGIN NORMAL AND TUMORCELLS*
AND
(Received for publication, November 2, 1990)
David G. Stokes$, Biagio SaittaSj, Rupert Timplll, and Mon-Li Chu$II** From the $Departmentof Biochemistry and Molecular Biology and the 11 Department of Dermatology, Jefferson Institute of Molecular Medicine, Thomas Jefferson university, JeffersonMedical College, Philadelphia, Pennsylvania 19107 and the IIMar-Planck-Institut fur Biochemie,0-8033 Martinsried, Federal Republic of Germany
We recently reported the isolation and sequencing of different chains, denoted al(VI), a2(VI), and a3(VI), which human cDNA clones corresponding to the a3 chain of come together toform a dumbbell-shapedmolecule consisting type VI collagen (Chu,M.-L., Zhang, R.-Z., Pan, T.-c., of twolargeglobular domains separated by a short triple Stokes, D., Conway, D., Kuo, H.-J., Glanville, R., helical domain (1, 4, 5 ) . The type VI collagen monomers can Mayer, U., Mann, K., Deutzmann, R., and Timpl, R. participate in well defined dimers and tetramers, the latter (1990)EMBO J. 9,385-393). The study indicates thatbeing the building block for the microfibrillar structure by theamino-terminalglobulardomain of thea3(VI) associating in an end-to-end fashion (see Ref. 1).The correchain consists of nine repetitive subdomains of approx- sponding cDNAs for all three chains of human andchick type imately 200 amino acid residues (Nl-Ng) and the gene VI collagen have been cloned and sequenced (6-14). The appeared to undergo alternative splicing since some clones lacked regions encoding the N9 and part of the al(V1) and a2(VI) chains have an apparent molecular mass of 140 kDa and arevery similar in structure, each consisting N3 subdomains. In the present study, we report the of an amino-terminal globular domain of approximately 230 exon structure for the region encoding the amino-terminal globular domainof the humana3(VI) chain. The amino acid residues, a short triple helical region, and a carnine repetitive subdomains are encoded by 10 exons boxyl-terminal globular domain of approximately 430 resithe spanning 2 6 kilobase pairs of genomic DNA. Eight of dues.Comparison of theamino acidsequencewithin the repetitive subdomains (N2-N9) were found to be chains andalso to other proteins demonstrated that the amino encoded by separate exonsof approximately 600 base and carboxyl globular domains were apparently the resultof pairs each. The only exception is the N1 subdomain an internal triplication of a -200-residue subdomain (8, 11, above structure, themuch largera3(VI) which is encoded by two exons of 417 and 146 base 12). In contrast to the pairs. Characterization of the exonfintron structure chain (340 kDa) was found to have at least eight more repetshowed that the cDNA variants were the result of itive 200-amino acid subdomains in its amino-terminalglobsplicing out of exon 9 (encoding the N9 subdomain) ular domain (see Fig. 1) (9, 13, 14). These subdomains, in all and partof exon 3 (encoding the N3 subdomain). Nu- three chains, were found to be similar to the typeA domains clease S1 analysis and the polymerase chain reaction of von Willebrand Factor (vWF)’ andalso to domains found demonstrated that exon 7 (N7 subdomain) was also inthe cartilage matrixprotein(CMP) alongwithseveral subject to alternative splicing in normal skin fibroother proteins (8,9, 11-14). The a3(VI) chain was also shown blasts. Examination of these splicing events by nuto contain three other unrelated subdomains in the carboxylclease S1 analysis in normal fibroblasts, three differ- terminal domain which showed similarity tofibronectin, salent human tumor cell lines, and several human tissues (9, 13). ivary proteins, and aprotinin type protease inhibitors showed that splicing out of exon 9 is much more effiThe type A domains of vWF have been shown to bind to cient in normalas compared to tumorcells. extracellular matrix componentssuch as type I collagen, and this has shed some lighton the function of the typeVI collagen amino-terminal subdomains(15, 16). Moreinsight was gained The extracellular matrix of a wide variety of tissues con- when a fusionprotein consistingof five of the amino-terminal tains a microfibrillar network composed of type VI collagen subdomains from the a3 chain of chicken type VI collagen molecules (see Ref. 1for a review). This network is interwoven was shown to bind to type I collagen which had been immowith large interstitial collagen fibers, located near some non- bilized on plastic surfaces (14). The binding appeared to be epithelial basement membranes, and also found close to cells saturable, reversible, and could be competed for by the tissue (2, 3). The type VI collagen monomer is composed of three form of type VI collagen. In support of this data, immunolocalization studies suggest that type VI collagen may serve to * This research was supported in part by the National Institutes anchor structures such as cells, large collagen fibers, nerves, of Health Grants AR-38912,AR-38923, AR-38188, and AR-39740 and blood vessels to the surrounding matrix (2).Indeed, cell and the Deutsch Forschungsgemeinshaft Grant SFB-266. The costs attachment assays have shown that the a2(VI) and a3(VI) of publication of this article were defrayed in part by the payment of chains exhibit cell binding activity (17). These studies have page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18U.S.C. Section 1734 solelyto indicate led to the hypothesis that RGD (Arg-Gly-Asp) sequences in the triple helical regions of all three chains, or yet undefined this fact.
5 On leave of absence from the Istituto de Biologia dello Sviluppo, Consiglio Nazionale Ricerche, Palermo, Italy. ** To whom correspondence should be addressed Dept. of Biochemistry and Molecular Biology, Thomas JeffersonUniversity, 1020 Locust St., Philadelphia, PA 19107.
The abbreviations used are: vWF, von Willebrand factor; bp, base pairs(s); kb, kilobase pair(s); nt, nucleotide(s); Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonicacid; PCR, polymerase chain reaction; SDS, sodium dodecyl sulfate; CMP, cartilage matrix protein.
8626
Alternative Splicing
ofHuman the
a3(VI) Collagen Gene
8627
sequences,couldserve as cell attachmentsites, while the 10 mM MgCI2,and 5 units of the Klenow fragment of DNA polymerase After primer extension the prodrepetitive globular subdomains could function to anchor the I (Bethesda Research Laboratories). ucts were digested with the appropriate restriction enzyme and free cells to the surrounding matrix(2, 3, 7, 8, 17). nucleotides were removedby running overa Sephadex G-50 spin Previous studies have shown heterogeneity in size of both column. Single-stranded probeswere isolated on denaturing 5%polya3(VI) transcripts andpolypeptide chains, indicatingpossible acrylamide gels and recovered from excised bands by the gel-crush proteolytic processing and/or alternative splicing events (6, method (28). 18,19). Isolationof cDNA clonesfor the human a3(VI) chain Between 1 X IO4 and 2 X lo4 cpm of probe was hybridized with either 1-5 pg of poly(A)+ RNAor 10-20 pg of total RNA in 20-40 p1 whichlacked theN9domain or part of theN3domain of 80% formamide, 25 mM Hepes (pH 7.0), 300 mM NaC1, 0.25 mM supports the notion of alternative splicing (9). In order to EDTA. Yeast tRNA was added to give a final concentration of 0.6 further investigate thesepossible differential splicing events, mg/ml. The hybridizations were carried out for 3-16 h at 48 "C. S1 we have isolated cosmid clones containing the region of the nuclease digestion was accomplished by adjusting the volume to 300 human gene encoding the nine amino-terminalglobular sub- p1 and a final concentration of 50 mM NaCl, 30 mM sodium acetate domains of thea3(VI)chain.Restrictionmapping of the (pH 4.5), 1mM ZnSO,, 1.25 pg/pl denatured salmon sperm DNA, and cosmids and sequencing of the exons revealed that subdo- 0.4 unitlpl S1 nuclease (U. S. Biochemical), followed by incubation a t 37 "C for 40 min. The products were recovered by ethanol precipmains two through nine (N2-N9) are encodedby separate itation and run on denaturing 5% polyacrylamide gels followed by encoded is exons of -600 bp each, and that the N1 subdomain autoradiography of the dried gels. Scanning densitometry was perby two exons of 417 bp and 146 bp. Analysis of the a3(VI) formed using a LKB scanning laser densitometer (LKB Ultrascan transcripts in different cell lines and tissues by SI nuclease XL). Percentagesof spliced uersus unspliced products were computed of cytidine protection and the polymerase chain reaction (PCR) indicates by normalizing the peak areas with respect to the number that the exonsencoding subdomains N7 and N9were subject residues in the protected portionof the probe. Polymerase Chain Reaction Amplification (PCR) of (u3(VI) Tranto alternative splicing. These findings introduce some inter- scripts-PCR was performed by a modification of the procedure of esting evolutionary and biological implications with respect Saiki et al. (29). Single-stranded cDNA was synthesized from 15 pg to typeVI collagen. of total RNA (3349 fibroblasts) by using random hexamers as primers (Boehringer Mannheim) andMoloney murine leukemia virus reverse transcriptase (Bethesda Research Laboratories) under conditions recommended by the manufacturer. The first-strand cDNA was phenolIsolation andCharacterization of Genomic Clones-Two human cosmid libraries were screened with a3(VI) cDNA probes in order to chloroform-extracted and ethanol-precipitated.Approximately 15 ng of first-strand cDNA or 10 ng of control a3(VI) plasmid cDNA plus isolate the gene. One library was constructed with human leukocyte 0.1 p~ primer was used in a polymerase chain reaction as recomDNA and the pHV4 vector as described previously (20). The other mended by Cetus Corp. (Emeryville, CA). Triton X-100 was added to library was obtained from Stratagene (La Jolla,CA) and was a human a final concentration of 0.25% as recommended by Promega. Tetraplacental DNA cosmid library constructed with the pWE15 vector. methylammonium chloridewas added to a final concentrationof 0.5In each case approximately 4 X lo5colonies were plated and screened 1 X lo-' M to help suppress nonspecific priming (30). In all cases 2.25 using previously characterized (u3(VI) cDNA inserts and subfragunits of Taq DNA polymerase (Promega) was used. 30-35 cycles of ments by the procedure of Dillela and Wu (21). Cosmid clones were amplification were performed in a thermal cycler from COY Labofurther characterized by Southern blot analysis using cDNA frag- ratories (Ann Arbor, MI). PCR fragments were run on 1.2% agarose ments, cosmid fragments, and synthetic oligonucleotides derived from gels and Southern-blotted to nylon filters (Hybond-N, Amersham the cDNA sequence as probes. Cosmid fragments containing exons Corp.), and the filterswere hybridized to 5"end-labeled oligonucleoweresubcloned into pBluescript vectors (Stratagene), and introntides. Oligonucleotides used in this study were synthesized on a Du exon sequences were obtained by the dideoxy chaintermination Pont DNA synthesizer (Coder 300). method (22) using a sequencing kit from U. S. Biochemical (Cleveland, OH) and [a-"Slthio-dATP from Du Pont-New England Nuclear. RESULTS Cell Lines, Tissues, and RNA Isolation-Established human skin fibroblasts lines (1520, 3349) were obtained from theCoriell Institute IntronlExon Structureof the HumanGenomic Region Codfor Medical Research, Camden, NJ. 1520 fibroblasts were originally ing for the Amino-terminal Globular Domain of the a3(VI) established from the foreskin of a 3-day-oldmale and 3349 fibroblasts werefroma 10-year-old male. The H4, HT1080, and Hs913T cell Chain-Five overlapping cosmids were isolated from screenlines were obtained from the American Type Culture Collection. H4 ing two different human cosmid libraries with two different a3(VI) cDNA fragments (Fig. 1). Cosmids 3A, 4B, and 7B cells were established from a neuroglioma removed from a 37-yearold male. HT1080 and Hs913Tcells were both established from adult were isolated by screening a human placental DNA cosmid fibrosarcoma tumors,All cell lines were grown in Dulhecco's modified library with an842-bp EcoRI fragment containing part of the Eagle'smedium supplemented with 10% fetal bovine serum. Fetal 5"untranslated region and the first494 bp of the N9 subdoskin, muscle, and nerve tissues were obtained from a male fetus of main and were all found tobe identical (-33 kb). Cosmid 2B 22-week gestationaftertherapeuticabortion. Adult fibrosarcoma tissue was obtained after surgical removal from an adult female. All (-32 kb) was isolated froma human lymphocyte DNA cosmid library screened with a 1-kb EcoRI-Hind111 fragment that tissues were snap frozen and stored inliquid nitrogen. cosmid Poly(A)+-enriched RNA was obtained from cultured cellsby selec- contained coding sequences for the N2 subdomain, and tion onoligo(dT)-cellulose (23). Cells in confluent 175-cm2 flasks D3 (-40 kb) was isolated by screening the same librarywith were lysedin buffer containing SDS-proteinaseK (Boehringer Mann- the a3(VI) cDNA clone P24, which contains coding sequences heim), and then the extract was homogenized in a Dounce homoge- for the triple helical region and part of the carboxyl globular nizer followed by incubation for 2 h a t 45 "C (24). The extract was were foundtocontain codingsethen passed throughtheoligo(dT)column,andtheRNA was domain.Thesecosmids recovered by ethanolprecipitation.TotalRNA was extracted quences for subdomains NlLN9 and partof the triple helical from tissues by the guanidinium-isothiocyanate-phenol-chloroform domain by Southern blot analysis with a3(VI) cDNA fragmethod (25). mentsand oligonucleotides basedoncDNA sequences as S1 Nuclease Protection Assay-The S1 nuclease protection assays probes. A restriction map of the area encoding subdomains were performed according to standard procedures (26-28). Briefly, Nl-N9 was generated by hybridizing Southern blots of didoubleuniformly labeled S1 probes were synthesized using denatured stranded cDNA clones as templates andspecific 17-mer oligonucleo- gested cosmids separated on agarose gels with a3(VI) cDNA 1 pmol of fragments, oligonucleotides, and cosmid fragments (Fig. 1). tides as primers. The primer extension reaction contained Appropriate restriction fragments containing exons were template, 5.0 pmol of primer, 1 p~ ["PIdCTP (3000 Ci/mmol, Du Pont-New England Nuclear), 60 p~ dCTP, 600 p~ dATP, T T P , subcloned into pBluescript and the exon/intron boundaries dGTP, 7 mM dithiothreitol, 100 mM NaC1, 50 mM Tris-HC1 (pH7.5), were determined by sequencing withspecific oligonucleotides MATERIALS ANDMETHODS
8628
Alternative Splicingof the Humana3(VI,, Collagen Gene cDNA
-
B
A
*
I E
4
"
sp352 sp34.5
I
I
E
E
I
I
E
E
1 kb
Triple C-terminal subdomains
N-terminal subdomains
helix
3 A . 4B. 78
2B
FIG. 1. cDNA map (top) and protein subdomain map (middle)of the human a3(VI)collagen chain as delineated by the cDNA and the restriction and exon map of the human cosmid clone 3A (bottom) encoding the nine amino-terminal subdomains Nl-N9 and the overlapping cosmid clones spanning this region (bottom).The cDNA map for the a3(VI)chain is a t top with the EcoRI sites indicated by E. Arrows A and B indicate position of primers used for PCR. sp345 and sp352 are exon 7- and 8-specific oligonucleotides, respectively, that were used as probes. The nine amino-terminal subdomains depicted in the protein map (shaded bones, middle) are encoded by 10 exons (filled bones, bottom). The exons are denoted la-9, starting at the3' end, (above, in parentheses) and are spread out over an area of -26 kb. The locations of the five overlapping cosmid clones encoding this region are shown below. Restriction enzyme sites: B, BamHI; E, EcoRI; H , HindIII. The complete restriction maps for BamHI and HindIII areshown.
TABLE I Nucleotide sequenceof the intronlexon junctionsand exon sizefor the region encoding the amino-terminal domain of the human a3(VI) collagen gene The column on the right shows the size of each exon in base pairs. Donor
Acceptor
cacattttttgcatttctttcttttaaagcACAk-EXON (A)laGln
sire (bp)
9--CAAGgtatgtatctct' GlnV(al)+
597
t a l t t g c c t t ~ t ~ . l t g t t g c c ~ c c t ~ = = ~ C A ~ -a--GAAGgtstggtgqtga -EXON (v)alxle GlnV(a1)
585
c t c a c t t t t t t t t c t g t g c t t = ~ = t g = ~ g ~ C A C - - E X O~ H-
(V)alHis
- C C A G ~ ~ ~ ~ C C ~ C C 600 C ~ ~ PTOG( 1U)
6--CCAGgtatggtaqaga ProV(a1)
573
gtgtaqttaccaggctgtcctcctcccagmCA--EXON 5--CCAGgtacgcagggaa (VI alser ProG(1y)
609
tCatqtatCCtaatCt~cEttC.ttgC.gAGAGC--EXON
(G)1US-r
gtgaagccatgctCaCatttgttt~g~~qGTGTT"EX0N r--CCAGgtgagatgqggg (GjlyVal ProA ( la)
606
3--CCAGgtatcacogtgt ProG (1" j
615
ctccttattttcactttsatcacttctagAGAAG--EXON 2--AAAGgtaactcccatt (GI luLys LysA(1a)
600
aacacatcaCCtgCatttctcctttccagcTTGT--EXON Ib-GAAGgtaggtgccaag (A) lacy6 GluA(1a)
417
aqctgtattaaatgtgtctttctttgcaqGAGTC--EXON (G)lyVal
14 6
tgtttccctttttcttctttgatctgcagcAGTF-EXON (A)laVal
la-GCTTgtgagttttttt LBU
* The intron sequences are denoted by lower case letters and the beginning and ends of the exon sequences are shown in upper case. t Underneath the exon nucleotide sequences are thecorresponding amino acid residues. Parentheses indicate amino acids which are encoded by split codons. generated from the cDNA sequence. Each of the subdomains N2-N9 were found to be encoded by a separate exon of -600 bp, with introns ranging in size from 1 to 7 kb (Table I and Fig. 1). Subdomain N1, the only exception, is encoded by two separate exons of 417 bp and 146 bp (Table I and Fig. 1).The exons were found to be spaced over a distance of -26 kb of
genomic DNA (Fig. 1).The nucleotide sequences of the intron/exon boundaries are shown in Table I. The boundaries were determined by comparison of genomic and cDNA sequences. All of the donor and acceptor splice sequences are in good agreement with the consensus sequence for these regions (31, 32). Upon closer examination of the exon/intron structure of the region encoding subdomains N2-N9 it was revealed that each exon (Fig. 1) defines the subdomains to within 4 to 7 amino acid residues of what was predicted by computer analysis of the cDNA-based protein sequence (9). This remarkable cassette type exon structure was further reinforced by the observation that exons lb-9 start with the second base and end with the first base of a codon (Table I). This type of splice site arrangement allows for the removal of any exon(s) without a disruption in the translational reading frame. Indeed, previous work had resulted in the isolation of a3(VI) cDNA clones which either contained or lacked the regions encoding the N9 or partof the N3 subdomains (9).The intron/ exon structure for this region revealed that these cDNA variants were the result of splicing out of exon 9 and thefirst 501 bp of exon 3. Nuclease S1 Analysis of Exons 9 and 3 in Human Skin Fibroblasts-S1 nuclease analysis was employed in order to ascertain the extent of these differential splicing events in two different human skin fibroblast lines. A uniformly labeled single-stranded DNA probe was synthesized which covered the last 102 nucleotides (nt) of exon 9 and the first 310 nt of exon 8 (Fig. 2, bottom). Digestion with S1 nuclease would yield three different protected fragments; a fully protected fragment of 412 nt, a fragment of 310 nt corresponding to protection of exon 8, and a fragment of 102 nt corresponding to protection of the endof exon 9. This probe was hybridized
Alternative Splicing of the Human cu3(VI) Collagen Gene M
1 2 3 4 5 6
nt 615492369-
J)
-0
* -
246-
EXON 9
,
EXON 8
I
Probe 412 n t
FIG. 2. S1 nuclease protection analysis of exon 9 in human skin fibroblast mRNA. The uniformly labeled single-stranded probe of 412 nt is shown at the bottom. This probe extends from nucleotide position 1253 (following the numbering described previously for the cDNA in Ref. 9) in exon 8 to an EcoRI site a t position 841 in exon 9. Splicing out of exon 9 would be expected to yield a protected fragment of 310 nt (bottom arrow). Full protection of the 412-nt probe is denoted by the top arrow. Lane M,123-bp DNA size ladder; lane I , probe alone; lane 2, probe plus S1 nuclease; lanes 3 and 4 , l and 4 pg of 3349 fibroblast poly(A)+ RNA plus probe and SI nuclease; lanes 5 and 6, 1 and 4 pg of 1520 fibroblast poly(A)+ RNA plus probe and S1 nuclease. The protected fragments were run on 5% polyacrylamide sequencing gels and autoradiography was performed a t -70 "C for 24 h. Note that the size markers used in these and subsequent S1 nuclease experiments were end-labeled doublestranded DNA fragments and, therefore,some strand separation has occurred a t the lower sizes (246-123 nt) during the denaturing electrophoresis. For thesefragments we have taken the center point between the two separated strands asbeing the correct length.
with poly(A)+ RNA from two different established human skin fibroblast lines (1520, 3349) and showed that both the full protected fragment and the 310-nt fragment are present and are of similar intensities (Fig. 2, arrows, lanes 3-6). The 102-nt fragment representing splicing out of exon 8 was not observed, even after prolonged exposure (data not shown). Scanning densitometry of these and otherrepresentative experiments indicate that the percentage of spliced variants is approximately 60%for 3349 fibroblasts (Fig. 2, lanes 3 and 4 ) and 75% for 1520 fibroblasts (Fig. 2, lanes 5 and 6). The heterogeneity in size of the 310-nt band representing exon 8 is consistently observed in numerous experiments with this probe and is probably due to an A-T-rich area of the splice junction where strand breathingcould occur during digestion. This problem was subsequently abated by use of a longer probe (see below). Elucidation of the exonlintron structureof the region encoding the N3 subdomain revealed that the cDNA clones lacking part of this region were the result of removal of the first 501 bp of exon 3, which is 615 bp inlength. Aweak splice acceptorconsensus sequence, CCCCAGACTGGTCTTCACAG/TG, was found at the point where the deletion ended within exon 3. When probes covering part of exons 3 and 4 or part of exons 2 and 3 were used in a S1 analysis on 3349 fibroblast mRNA, splicing out of the 501-nt segment was detectable only at very long exposure times (data notshown). Exon 7 Also Undergoes Alternative Splicing in Human Skin Fibroblasts-Probes were then constructed to investigate whether any other exons were subject to alternative splicing events. Several probes covering parts of exons 2 through 6
8629
were used in S1 nuclease protection experiments on mRNA from 3349 human skin fibroblasts. It was found that exon 7 also appears to undergo alternative splicing in fibroblasts. Fig. 3 shows the resultsof an S1 experiment in which a probe covering the last246 bp of exon 8 and thefirst 338 bp of exon 7 was hybridized with 1 and 4 pg of poly(A)+ RNAfrom 3349 fibroblasts and thendigested with S1 nuclease. A 246-ntband can be seen which would correspond to the correct size for protection of exon 8 and splicing out of exon 7 (lanes 3 and 4, bottom arrow). Faint bands can also be seen migrating at the 300-nt range in lanes 3 and 4. It is possible that these bands represent a very low level of splicing out of exon 8. This splicing event was not detected inthe S1 nuclease analysis of exon 9 (data not shown), most likely due to the smaller size of the protected fragment in these experiments (102 nt). Scanning densitometry of this film was performed and showed that exon 7 was spliced out approximately 3035% of the time. Since cDNA clones which lacked exon 7 had not been isolated before, PCR was employed in order to confirm the existence of the smaller species of mRNA. Primers (17-mers) located in exons 6 and 8 (Fig. 1,primers A and B ) were used in a PCR amplification which contained first-strand cDNA made from 3349 total RNA or an (u3(VI) cDNA clone containing exon 7sequences as acontrol. These primers are expected to give rise to two fragments; a 992-bp fragment corresponding to transcripts containingexon 7 and a 392-bp fragment representing transcripts in which exon 7 has been spliced out. Fig. 4A shows the expected bands were generated along with a faint band of -700 bp present in the RNA lane (lane 2). The amplified fragments were then sequentially hybridized with an oligonucleotide (Fig. 1, sp345) specific for exon 7 and then an oligonucleotide specific for the 3' end of exon 8 (Fig. 1, sp352). As can be seenin Fig. 4B, sp345 hybridizes with the 992-bp fragment, while sp352 hybridizes to both the 992-bp fragment and also the 392-bp fragment, M I 2 3 4
615492369-
avv1
246-
c
-
EXON 8
I
EXON 7
Probe
584 n t
FIG. 3. S1 nuclease protection analysis of exon 7 in human skin fibroblast mRNA. The uniformly labeled single-stranded DNA probe of 584 nt is depicted at the bottom. This probe extends from position 1866 in exon 7 to a Hind111 site a t position 1282 in exon 8. Splicing out of exon 7 would be expected to yield a protected fragment of 246 nt (bottom arrow). Full protection of the 584-nt probe is indicated by the top arrow. Lane M ,123 bp DNA size ladder; lane 1 , probe alone; lane 2, probe plus S1 nuclease; lanes 3 and 4, probe plus 1 and 4 pgof 3349 fibroblast poly(A)+ RNA and S1 nuclease. The products were run on 5% sequencing gels and exposed for 24 h a t -70 "C.
of the Human or3(VI) Collagen Gene
Alternative Splicing
8630 0
A
M 1 2 3 4 5 6 7
C. I23
nt
13 24 bp
-984 -369 -123
7
1000-
ilarmn
bp
c"
8 9
984369123-
-984 0
-369 -123
EXON 9 1
EXON 8
-Probe 614nt
FIG. 4. PCR analysis of the mS(V1) collagen mRNA transcripts in human skin fibroblasts. A , PCR was performed after first-strand cDNA synthesis using total RNA isolated from 3349 skin fibroblasts and primers A and R (Fig. 1) located in exons 8 and 6 in 7. A control orderto amplify transcriptswithandwithoutexon consisted of using the same primers and a cDNA clone containing exon 7 sequences asa template. Amplified products were run on 1.2% agarose gels andthe DNA bands werevisualized withethidium bromide and ultraviolet light. Lune 1, 10 ng of control plasmid as template; lune 2, 10 ng of 3349 fibroblastfirst-strandcDNAas template; lune 3, DNA size ladder. R, Southern blot analysis of the gel from A . The filter was hybridized with oligonucleotides specific for exon 7 (sp345, Fig. 1) and exon 8 (sp352, Fig. 1).Lanes 1 and 2, same as in A hybridized with sp345; lunes 3 and 4, same as in A hybridized with sp352. C, BglI digestion of the PCR products. Lane 1 , 123-bp DNA size ladder; lune 2, digested products; lune 3, undigested products.
FIG. 5. S1 nuclease analysis of exon 9 in different tumor cell lines. The 614-nt single-stranded probe is depicted below. Full protection of this probe is indicated by the top arrow,while the bottom arrow denotes protection of exon 8 (512 nt). Lane M, 1-kb ladder; lune I , probe alone; lune 2, probe plus S1 nuclease; lune 3, 5 pg of 3349 fibroblastpoly(A)+RNA; lunes 4 and 5, 2 and 5 pg of H4 neuroglioma poly(A)+ RNA; lunes 6 and 7, 2 and 5 pg of HT1080 fibrosarcoma poly(A)+ RNA; lunes 8 and 9, 2 and 5 pg of Hs913T fibrosarcoma poly(A)+ RNA.The gel was exposed for 18 h at -70 "C. A nt
1 2 3 4
5 6 7 8 910
Human Fibrosarcoma and Neuroglioma Cell Lines Retain Exon 9 to a Much Greater Extent thun Fibroblasts-In order to extend our observations concerning the differential splicing of the a3(VI) transcripts, two different human fibrosarcoma cell lines (Hs913T and HT1080) and a neuroglioma cell line (H4)were studied by nuclease S1 analysis to determine if splicing out of exon 9 occurred to the same extent observed as in 3349 fibroblasts. Fig. 5 shows that much less splicing is observed for all three tumor cell lines. In each case the lower band (bottom arrow) corresponding to protection of exon 8 and splicing out of exon 9 is much less intense then the full protected band (upper arrow).Quantitation of the results by scanningdensitometry showed that exon 9 is spliced out approximately 25% of the time in H4 cells and 30-35% of the time in HT1080 and Hs913T cells. Different Patterns of Splicing for Exon 9 Are Observed in Human Fetal and Adult Tissues-S1 nuclease digestion was also used to characterize splicing of exon 9, in human fetal skin, skeletal muscle, sciatic nerve, and adult fibrosarcoma tumor tissue, in comparison with 3349 fibroblasts. Fig. 6 shows that the relative amounts of unspliced RNA variants were drastically reduced in all three fetal tissues. Only skeletal muscle RNA exhibited a small amount of the unspliced var-
I 2 3 4
I
615492-
1000I
369-
thus confirming the S1 analysis of exon 7. It canbe seen that the 700-bp fragment also hybridizes with sp345. This fragment is routinely generated along with the correctly sized fragments using different magnesium concentrations, annealing temperatures, and RNA samples. We believe that one explanation for this product could be the amplification of an incompletely spliced RNA since the initial reverse transcription step was performed with total cellular RNA. In order to further confirm the specificity of these bands we utilized the fact that both the 392-bp and 992-bp fragments should contain a BglI site located 100 bp from the 5' end, as predicted by the cDNA sequence (9). The results of a BglI digest of the PCR products are shown in Fig. 4C (lane 2 ) , and it can be seen that both products are cleaved to give fragments of sizes 892,292, and 100 bp.
B
1112
C
+
246-
c c
511394-
EXON 9
-
EXON 9
I
I
EXON8
-Probe 614nt
EXON 8
Probe 412 nt
FIG. 6. S1 nuclease analysis of exon 9 in human fetal and adult tissues. A , the probe depicted below is the same as described in Fig. 2 legend. Full protection of the probe is denoted by the top arrow (412 nt) and protection of exon 8 (310 nt) is indicated by the bottom arrow. Lunes 1-10 all contain probe plus S1 nuclease and 10 or 20 pg of total RNA as follows. Lunes I and 2, 3349 fibroblasts; lunes 3 and 4 , fetal skin; lunes 5 and 6, fetal muscle; lunes 7 and 8, fetal nerve; lunes 9 and 10, adult fibrosarcoma tumor; lune 11, probe plus S1 nuclease; lane 12, probe alone. R, the probe shown below the panel is a 614-nt single-stranded probe extending from position1455 to an EcoRIsite at position 841. Fullprotection of the probe is denoted by thetop arrow and protection of exon 8 is indicated by the bottom arrow. Lanes 1-4 contain 15 pg of total RNA as follows: lune I , fetal skin; lune 2, fetal nerve; lune 3, fetal muscle; lune 4 , fibrosarcoma tumor. Bothgels were exposed for 48 h a t -70 "C.
iant (Fig. 6A, lane 6).However, the adult fibrosarcoma RNA showed, compared to 3349 fibroblasts, a relative increase of the unspliced variant. Thisincrease does not appear to be as high as was seen for the fibrosarcoma cell lines (see Fig. 5). DISCUSSION
a3(VI) Gene Structure Encoding the Amino-terminal Globular Domain Follows the One Exon-One Domain Rule-It has been speculated that proteinswhich contain repeated domains are formed from duplication events involving a single intron/ exon motif and that similar or identical repetitive sequences found within introns could serve as a site for recombination and internal duplication of such a segment (33-36). Proteins which fall into this context include immunoglobulins, albu-
Alternative Splicing of the 1Yuman a3(VI) Collagen Gene A3
144
AI
141
117
A2
1379
REP 1
REP 2
.................... 3G7
223
IS
CI
N1
NZ-N9
573-6
153
414
Ib
la
417
I46
453
153
FIG. 7. Comparison of exon structures encoding homologous 200-residue domains in human von Willebrand factor, and the a3 and a2 chains of chicken cartilage matrix protein, human typeVI collagen. The domains are denotedAI-A3 (vWF), REP1 and REP 2 (CMP), Nl-N9 in a3(VI),and CI in aB(V1).The exons encoding the domains are indicated by the shaded boxes and their size is given by the number below the box in base pairs. In the case of Nl-N9 for the a3(VI) chaintherange of exon sizes is indicated. Data for vWF (39) and CMP (38) were previously published.
min, von Willebrand Factor, fibronectin, the triple helix of most collagens and as shown here a large portion of t h e a 3 chain of type VI collagen (35, 37-39). The nine repetitive subdomains in the amino terminus of the a3(VI) chain are encoded by 10 exons spanning -26 kb of human genomic DNA. The most interesting feature of this genomic region is that each of thesubdomains N2-N9 isrepresented by a separate exon of -600 bp. The only exception to this is the N 1 subdomain which is encoded by two exons of 417 and 146 bp. A comparison of the gene structures encoding thehomologous subdomains in vWF, CMP, and the a2(VI) chain(Fig. 7) shows a somewhat different exon/intron arrangement as compared to the a3(VI) gene. The three repetitive type-A domains of human vWF (Al-A3) are encodedbyasingle exon (Al-A2) or four small exons (A3) (39). In contrast to this, both of the repetitive subdomains found in CMP are encoded by two exons of similar sizes to thatof exons l a a n d 1b encoding theN1subdomain,althoughtheintronsare located in different places (38). It is possible that the exons encoding the A1 and A2 subdomains have fused as the result of a loss of an intron, while the exon encoding the A3 subdomain has gained multiple introns. In the case of the CMP gene, and asdiscussed below for N1, onecould speculate that these exonsoriginally did not contain introns, but later gained them since the introns are located in different places. Two other proteins which contain homologous subdomains have exon/intron structures similar to thatof the vWF A3 subdomain. The a subunit of the leukocyte integrin p150,95 has a repetitive subdomain encoded by four small exons, while in human complement protein factor B the subdomain is encoded by five small exons (40,41). The C1 subdomain of the a2(V1) chain is also encoded by two exons (153 and453 bp, Fig. 7), whereas the a2(VI) N1 subdomain is encoded by a single exon (20).* These exon structures are consistent with a possible gain of intron event for the homologous domains. Multiple Duplications of a Primordial Exon as an Explanation for the Evolution of the cu3fVI)Gene-Since the primary sequence and structure of the al(V1) and a2(VI) chains are more similar to each other than to the a3(VI) chain and the respective genesare located in the same region of chromosome 21, it is thought that they arose from a gene duplication event involving a primordial type VI gene (9, 42). In this case, we
' B. Saitta,
unpublished observation.
8631
speculatethat a primordialtype VI collagen gene would encode the N1, C1, C2, and triplehelical regions. The similar organization of the region encoding the triple helical domains in all three genes is suggestive of the idea that the genes are derived from a common ancestor.' The a3(VI) chain also contains the corresponding domains found in the al(V1) and &(VI) chains (Nl,C1, C2, and the triplehelical segment, see subdoFig. 1) but differs in that it contains the additional mains N2-N9 and C3-C5 (9). In order to explain the formation of the a3(VI)gene, we propose that an earlier duplication of a primordial type VI collagen gene occurred thereby allowing more time for the divergence of the a3(VI)sequence (Fig. 8).This eventis followed by duplication of exon 1( N l ) , which later gains an intron. Exons 3-9 are then formed by sequential duplication of the newly formed exon 2 which would explain the fact that subdomains N2-N9 share more identity within themselves than to subdomains N1, C1, and C2 (Fig. 8 and Ref. 9). As stated above, the N1 subdomain of the a2(VI) chain isencoded by a single exon,* whichfurther supports the idea that the intron separating exons l a and Ibin the a3(VI) gene was acquired in a later event. Human a3(VI) Gene Exhibits Tissue and Cell Line-specific Alternative Splicing-Previously, it was reported that some a3(VI) cDNA clones isolated from human skin fibroblasts (3349) contained deleted areas which correspond to the N9 and part of the N3 subdomains (9). In this report we have demonstrated by nuclease S1 analysis that the cDNAclones which lacked the N9 subdomain were the result of an alternative splicing event involving exon 9, whichencodes the whole N9 subdomain. The data also show a different efficiency of exon 9 splicing in fibroblasts, tumor cell lines, and several human tissues. S1 analysis performed on mRNA from two different human skin fibroblast linesrevealed that exon 9 is spliced out about 60-75%of the time. In contrast to this, three different human tumor cell lines were much less efficient in splicing (20-35% spliced). Splicing out of exon9 was virtually complete in three different human fetal tissues but was notnearlyasevidentinhuman fibrosarcoma tissue. Whether this reflectsageneraldifference in efficiency of splicing of a3(VI) transcripts between normal and malignant cells remainsto be determined by amorecomprehensive comparison.Exon 7also exhibitedalternative splicing in human skin fibroblast mRNA with about30-35% efficiency. It remains to be determined whether exon 7 is subject to any cell- or tissue-specific alternative splicing as was found for exon 9. In the case of the N3 subdomain, the cDNA deletion corresponding to the first 501 bp of exon 3 (615bp) is apparently due to a weak splice acceptor consensus sequence within the exon. Nuclease S1 analysis showed that thissplicing pathway is a minor event in normal skin fibroblasts (data not shown). Again, more experiments need to be carried out in order to ascertain if this splicing event is prevalent in other cells or tissues.Untilthishasbeenshown we mayconsider this particular mRNA modification as a negligible event for modulating the a3(VI) structure. Alternative splicing of pre-mRNA appears to be a general way by which cells produce multiple forms of a protein from the same gene (43). Numerous genes have been reported to give rise toalternatively spliced transcripts,althoughthe exact functional significance in terms of the protein variants has not been elucidated in most cases (43). One gene which shows similarity to the cassette type structure reported here is the fast skeletal troponin T gene. The troponin T gene, which has 18 exons in all, contains five small homologous exons in its 5' end that arespliced in a cassette fashion (43-
8632
Alternative Splicing of the Human a3(VI) Collagen Gene ~
FIG. 8. Schematic representation of a possible scenario for the evolution of the human type VI collagen genes. The primordial type VI collagen is at top, left and is depicted as encoding the N1, triple helical, C1, and C2 subdomains. An earlier duplication event (right) gives rise tothe n3(VI) gene which is followed by duplication of the exon encoding the N1 domain. At this point the N1 exon gains an intron. Multiple duplication events then give rise to exons encoding subdomains N3-N9. The n3(VI) gene also gains coding sequences for subdomains C3-C5 by events not understood well yet. In a later event (left) the primordial gene is again duplicated to giverise to the al(V1) and a2(VI) genes.
Earlier duDlication/ recombination event
~
Primordial type V I collagen gene
I
NI is duplicated
00 t
Later event u 3 . chromosome 2
N I gains intron
L I r - I I I H I -
t Recombination/duplication event giving rise to the a1 and a2 genes on chromosome 2 I .
46). Each exon, as in the a3(VI) gene, begins and ends in a split codon thereby preserving the translational reading frame (43, 46). Theoretically, the troponin T gene can give rise to 32 different variants by differential splicing of these exons (43,46). In thecase of the a3(VI)gene, eight different forms of thea3(VI) chain could be generated from differential splicing of exons 3,7, and9 (Fig. 9), although it is more likely that four variants are possible if one considers splicing of exon 3 to be a negligible contribution. It is in this context of interest that up to four mRNA and protein bands are observed for the a3(VI) chain (6, 18, 19). Their correlation to distinct splicing events has stillto be determined. Alternative splicing events have also been observed in other extracellular matrix proteins such as fibronectin, which is relevant here because it is also composed of repetitive subdomains. As with the a3(VI)gene, it has been postulated that the fibronectin gene has evolvedby acquisition of exons encoding these subdomains from different genes followed by duplication of the exons (37). Two exons encoding separate subject to type 111- repetitive modules (EIIIA,EIIIB)are tissue- and cell-specific alternative splicing. Both of these exons are virtually excluded from liver tissue and hepatocyte mRNA, while they are present in lung and embryonic fibroblast and astrocyte mRNA(47-49). Splicing of the EIIIA
N9
N8
N7
N6
N.5
N4
N3
N2
N1 TH C 1
C2
C3C4C5
N2 gives rise to N3-N9
region is also regulated differently with respect to malignant and normal tissues in that there is a significant increase in the fibronectin transcripts containing the EIIIA exon in human liver tumors as compared with normal liver (50). Possible Functional Consequences of Splicing-The alternative splicing observed for type VI collagen in this report may also play a role in defining the extracellular matrix characteristics of different tissues. The significance of the differential splicing events involving exons 9 and 7 of the a3(VI) chainis unknown, however, examination of the amino acid sequence encoded by the exons and drawing an analogy to the function of similar domains in other proteins yield some possible explanations. Both exons 7 and 9encode single cysteine residues and potential N-glycosylation sites which upon removal may change disulfide bonding patterns and protection of protein surfaces. Structural changes in theprotein could result from different patterns of disulfide bonding within this region of the a3(VI) chain and, in turn, could influence the organization of the type VI collagen microfibrils. Furthermore, type VI collagen is known to be a highly glycosylated protein, and changes involving the extentof glycosylation could also influence the microfibrillar structureand function. It is interesting tonote that the3' end of the a2(VI) chain mRNA also undergoes alternative splicing potentially giving rise to threeprotein variants. The differential splicing here involves the C2 subdomain which is homologous to the a3(VI) chain Nl-N9 subdomains, and an analogous C2 sub5"UTR' 9 8 7 6 5 4 3 2 domain is present in the a3(VI)chain (9,20). Splicing of this region of the a2(VI) chain generates two variants (a2C2a, a2C2a') that are smaller in size and lack a cysteine residue as compared to the thirdmore abundant variant (a2C2) (20). Production of one of the a2(VI)variants (a2C2a') also results FIG. 9. Schematic representation of possible different splic- in the loss of a potential N-glycosylation site. ing patterns for the region encoding the amino-terminal doFinally, the type A domains of vWF and a fusion protein main of the a3(VI)chain. Eight possible protein isoforms can he generated from five of the amino-terminal subdomains of the generated from splicing of exons 9, 7, and 3. Exon numbering is the chicken a3(VI) chainhave both been shown to bind to type I same as for Fig. 1. The structure of the exon(s) encoding the 5'untranslated region (5'-UTR)including the signal peptide are, so far, collagen in vitro (14). Theseresults, coupled with observations of a close association of type VI collagen with large interstitial unknown.
Alternative Splicing
ofHuman the
a3(VI) Collagen Gene
8633
18. Colombatti, A,, Ainger, K., and Colizzi, F. (1989) Matrix 9, 177185 19. Ayad, S., Marriott, A., Morgan,K., andGrant, M. E. (1989) Biochem. J. 2 6 2 , 753-761 20. Saitta, B., Stokes, D. G., Vissing, H., Timpl, R., and Chu, M.-L. (1990) J . Biol. Chern. 265,6473-6480 21. Dillela, A. G., and Woo, S. L. C. (1985) Focus 7 , 1-5 22. Sanger,F.,Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci. U. S. A. 7 4 , 4563-4567 23. Aviv, H., and Leder, P. (1972) Proc. Natl. Acad. Sci. U. S. A. 6 9 , 1408-1412 24. Burnett, W., and Rosenbloom, J. (1979) Biochem. Biophys. Res. Commun. 86,478-483 25. Chomczynski, P.,and Sacchi, N. (1987) Anal.Biochem. 162, Acknowledgments-We would like to thank Drs. Sergio Jimenez 156-159 and Anthony Reginato for sharing the fetaltissues, Dr. Koulin Chou 26. Berk, A. J., and Sharp, P. A. (1978) Proc. Natl. Acad. Sci. U. S. for the fibrosarcoma tissue, Dr. Francesco Ramirez for help with the A. 75,1274-1278 initial screening of the cosmid library, Te-cheng Pan and Rui-Zu 27. Ley, T. J., Anagnou, N. P., Pepe, G., and Nienhuis, A. W. (1982) Zhang for all of the cDNA clones used in this study, and Loretta Proc. Natl. Acad. Sci. U. S. A. 7 9 , 4775-4779 Renkart for her expert technical assistance. 28. Maniatis, T., Fritsch, E.F., and Sambrook,J. (1989) in Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, REFERENCES Cold Spring Harbor, NY 1. Timpl, R., and Engel, J. (1987) in Structure and Function in 29. Saiki, R.K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R., Horn, G. T., Mullis, K. B., and Erlich, H.A. (1988) Science Collagen Types (Mayne, R., and Burgeson, R. E., eds) pp. 1052 3 9 , 487-491 143, Academic Press, Orlando, FL 2. Keene, D. R., Engvall, E,, and Glanville, R. W. (1988) J . Cell 30. Hung, T., Mak, K., and Fong, K. (1990) Nucleic Acids Res. 1 8 , 4953 Biol. 1 0 7 , 1995-2006 31. Brown, J. W. S. (1986) Nucleic Acid Res. 14,9549-9559 3. Bruns, R. R., Press, W., Engvall, E., Timpl, R., and Gross, J. 32. Shapiro, M. B., and Senapathy, P. (1987) Nucleic Acids Res. 15, (1986) J. Cell Biol. 103, 393-404 7155-7174 4. Trueb, B., and Winterhalter, K. H. (1986) E M B O J . 8 , 281533. Gilbert, W. (1978) Nature 271, 501 2819 34. Jeffreys, A. J., and Harris, S. (1982) Nature 2 9 6 , 9-10 5. Colombatti, A,, Bonaldo, A., Ainger, K., Bressan, G. M., and 35. Gilbert, W. (1985) Science 2 2 8 , 8 2 3 Volpin, D. (1987) J . Biol. Chem. 262, 14454-14460 36. Gilbert, W., Marchionni, M., and Mckight, G. (1986) Cell 46, 6. Chu, M.-L., Mann, K., Deutzmann, R., Pribula-Conway, D., Hsu151-154 Chen,C. C., Bernard, M. P., and Timpl, R. (1987) Eur. J. 37. Patel, R. S., Odermatt, E., Schwarzbauer, J. E., and Hynes, R. 0. (1987) EMBO J. 6,2565-2572 Biochem. 1 6 8 , 309-317 7. Chu, M.-L., Conway, D., Pan, T., Baldwin, C., Mann, K., Deutz- 38. Kiss, I., Deak, F., Holloway, R. G., Jr., Delius, H., Mebust, K. A., Frimberger, E., Argraves, W. S., Tsonis, P. A,, Winterbottom, mann, R., and Timpl, R. (1988) J . Biol. Chem. 2 6 3 , 18601N., and Goetinck, P. F. (1989) J. Biol. Chem. 264,8126-8134 18606 39. Mancuso, D. J., Tuley, E. A,, Westfield, L. A., Worrall, N. K., 8. Chu, M.-L., Pan, T.-C., Conway, D., Kuo, H.-J., Glanville, R.W., Shelton-Inoles, B. B., Sorace, J. M., Alevy, Y. G., and Sadler, Timpl, R., Mann, K., and Deutzmann, R. (1989) E M B O J. 8, J . E. (1989) J. Biol. Chem. 2 6 4 , 19514-19527 1939-1946 40. Corbi, A. L., Garcia-Aguilar, J., and Springer, T. A. (1990) J . 9. Chu, M.-L., Zhang, R.-Z., Pan, T.-C., Stokes, D., Conway, D., Biol. Chem. 265,2782-2788 Kuo, H.-J., Glanville, R., Mayer, U., Mann, K., Deutzmann, 41. Campbell, R. D., and Porter, R. R. (1983) Proc. Natl. Acad. Sci. R., and Timpl, R. (1990) E M B O J. 9 , 385-393 U. S. A. 80, 4464-4468 10. Triieb, B., Scharen-Wiemers, N., Schreier, T., and Winterhalter, 42. Weil, D., Mattei, M.-G., Passage, E., Van Cong, N’G., PribulaK. H. (1989) J. Biol. Chem. 2 6 4 , 136-140 Conway, D., Mann, K., Deutzmann, R., Timpl, R., and Chu, 11. Bonaldo, P., Russo, V., Bucciotti, F., Bressan, G. M., and ColM.-L. (1988) Am. J . Hum. Genet. 4 2 , 435-445 43. Breitbart, R. E., Andreadis, A., andNadal-Ginard, B. (1987) ombatti, A. (1989) J. Biol. Chem. 2 6 4 , 5575-5580 Annu. Rev. Biochem. 56,467-495 12. Koller, E., Winterhalter, K. H., and Trueb, B. (1989) E M B O J. 44. Medford, R. M., Nguyen, H. T., Destree, A. T., Summers, E., and 8,1073-1077 Nadal-Ginard, B. (1984) Cell 3 8 , 409-421 13. Bonaldo, P.,and Colombatti, A. (1989) J. Biol. Chem. 2 6 4 , 45. Breitbart, R. E., Nguyen, H. T., Medford, R. M., Destree, A. T., 20235-20239 Mahdavi, V., and Nadal-Ginard, B. (1985) Cell 4 1 , 6 7 - 8 2 14. Bonaldo, P., Russo, V., Bucciotti, F., Doliana, R., and Colombatti, 46. Breitbart, R. E., and Nadal-Ginard, B. (1987) Cell 49, 793-803 A. (1990) Biochemistry 2 9 , 1245-1254 “ I R. 0. 15. Pareti, F. I., Fujimura, Y., Dent, J. A., Holland, L. Z., Zimmerman, 47. Schwarzbauer. J. E.. Patel. R. S.. Fonda., D.., and Hvnes. (1987) EMBO J . 6 , 2573-2580 ’ T. S., and Ruggeri, Z. M. (1986) J. Bid. Chem. 2 6 1 , 15310- 48. Norton, P. A,, and Hvnes. R. 0. (1987) Mol. Cell. Biol. 8. 4297” , . . 15315 4307 16. Titani, K., and Walsh, K. A. (1988) Trends Biochem. Sci. 1 3 , 49. Oyama, F., Murata, Y., Suganuma, N., Kimura, T., Titani, K., 94-97 and Sekiguchi, K. (1989) Biochemistry 28, 1428-1434 17. Aumailley, M., Mann, K., von der Mark,H., and Timpl,R. (1989) 50. Oyama, F., Hirohashi, S., Shimosato, Y., Titani, K., and SekiguExp. Cell Res. 1 8 1 , 463-474 chi, K. (1989) J. Biol. Chem. 264, 10331-10334
fibrils by immunoelectron microscopy, point toward the idea that these subdomains are involved in binding to type I collagen or other matrix proteins(2). Removal of one or more of these subdomains from a3(VI)chain could cause a modulation in these interactions. It remains to be determined whether one subdomain is sufficient for binding or whether separate subdomains have measurable differences in binding. Further information concerning the function of the a3(VI) protein variantsneeds to be obtained before an understanding of the biological significance can be achieved.
