Alternatively spliced tenascin-C mRNA isoforms in human fetal ...

Molecular Human Reproduction vol.5 no.11 pp. 1066–1076, 1999

Alternatively spliced tenascin-C mRNA isoforms in human fetal membranes

S.C.Bell1,4, J.H.Pringle2, D.J.Taylor1 and T.M.Malak1,3 Preterm Birth Research Group, 1Department of Obstetrics and Gynaecology, and 2Department of Pathology, Faculty of Medicine and Biological Sciences, University of Leicester, Leicester LE3 7LX, UK 3Current address Eastbourne General Hospital, Eastbourne, East Sussex, BN21 2UD, UK 4To

whom correspondence should be addressed

Tenascin-C is an extracellular matrix glycoprotein whose monomers include eight consecutive fibronectin type III-like repeats, encoded by exons 10–16, and which are subject to alternative splicing. Transcripts containing these exons are expressed during tissue wounding and active tissue remodelling. Human fetal membranes have been proposed to undergo active tissue remodelling as part of the mechanisms leading to their rupture and immunoreactive tenascin-C has been detected in this tissue. Employing reverse transcription– polymerase chain reaction (RT–PCR) and exon-specific primers, products corresponding to multiple splicing events in the alternatively spliced region have now been identified. The overall splicing pattern would indicate that the major transcripts correspond to complete exclusion of the alternatively spliced region; inclusion of only exon 16; and inclusion of exons 10–14 and 16, including or excluding exon 12. The sole site in tenascinC susceptible to cleavage by matrix metalloproteinases (MMP)-2 and MMP-3 is found within the exon 12 encoded repeat, therefore translation of isoforms which include or exclude exon 12 may produce ‘large’ tenascins mediating functions ascribed to this form but susceptible or resistant to these MMPs. The demonstration of expression of ‘large’ tenascin mRNA isoforms supports the concept that fetal membranes at term are a site of active tissue remodelling. Key words: amniochorion/fetal membranes/mRNA/splicing/tenascin-C

Introduction The fetal membranes which encapsulate the human fetus and amniotic fluid normally rupture spontaneously at term during labour. However, in ~60% of preterm births rupture of the fetal membranes occurs prior to labour and is a direct antecedent of preterm birth (Kelly 1995; French and McGregor, 1996). Although infection has been implicated in the aetiopathology of a proportion (French and McGregor, 1996) the mechanisms of the pre-labour rupture of the fetal membranes in its absence (Malak and Bell, 1993, 1994; Kelly 1995; French and McGregor, 1996; Parry and Strauss, 1998), and indeed the mechanisms underlying their spontaneous rupture during term labour, are unknown. According to a recent proposal rupture may result from ‘degradation’ of the fetal membrane in a restricted area of the membranes prior to labour, analogous to tissue remodelling observed in a ‘wound response,’ and which inadvertently results in a regional structural weakening (Malak and Bell, 1994, 1996; Bell and Malak, 1997). Changes associated with this ‘degradation’ include myofibroblast differentiation (McParland et al., 1999), typical of tissue wounding. Recently immunoreactive tenascin-C, an extracellular matrix protein, has also been detected within fetal membranes both prior to, and after, labour and delivery (McParland and Bell, 1998; McParland et al., 1998). Tenascin-C is expressed in wounds and during other tissue remodelling conditions (Mackie et al., 1988; Whitby et al., 1991), possessing properties consistent with a function during the early wound response 1066

(Chiquet-Ehrismann, 1990; Vrucinic-Filippi and ChiquetEhrismann, 1993). For example, in the presence of fibronectin, tenascin up-regulates gene expression for matrix metalloproteinases (MMP)-1, MMP-3 and MMP-9 (Tremble et al., 1994), enzymes considered to be the main physiologically relevant mediators of matrix component degradation (StetlerStevenson et al., 1993). Tenascin-C is an oligomeric glycoprotein in which six monomers are disulphide linked at their N-termini to extend from a central core. The subunits are composed of structural domains that include the globular amino terminal domain, 14.5 epidermal growth factor-like repeats, 8–16 fibronectin type IIIlike repeats, and a carboxy-terminal sequence with homology to the globular domain of β- and γ-chains of fibrinogen (Erickson, 1993). It is encoded by a single gene and its expression is regulated by a single promoter (Gherzi et al., 1995). In the human eight contiguous fibronectin type III-like repeats (repeats A1–A4 and B–D encoded by exons 10–16, and the subsequently identified ‘additional domain 1’ encoded by exon AD1) have been demonstrated to be subject to alternative splicing (Gulcher et al., 1991; Nies et al., 1991; Sriramarao and Bourdan, 1993). Several monomeric protein isoforms have been identified by gel electrophoresis, but most commonly two tenascin subunit isoforms are reported in cells and tissues i.e. ‘small’ and ‘large’ isoforms, with the expression of ‘large’ isoforms associated with the wound response and other situations of active tissue remodelling such as neoplasia © European Society of Human Reproduction and Embryology

Tenascin and human fetal membranes

and development (Chiquet-Ehrismann, 1993; Crossin, 1996). In human cultured fibroblasts, these two forms have been interpreted to represent either complete omission or inclusion of the alternatively spliced repeats (Borsi et al., 1995). Differences in biological functions and properties have been ascribed to these isoforms with the ‘large’ isoform associated with increased mitosis, cell migration and down regulation of focal adhesions mediated by binding to annexin II (Murphy-Ullrich et al., 1991; Chung and Erickson, 1994; Chung et al., 1996). However, techniques of direct sequencing of cDNA clones and characterization of reverse transcriptase–polymerase chain reaction products (RT–PCR) (Siri et al., 1991; Sriramarao et al., 1993; Wilson et al., 1996; Vollmer et al., 1997; Saghizadeh et al., 1998) has revealed a more complex picture of alternative splicing in this region with the possibility of the production of functionally diverse isoforms. Eight mRNA species with variable numbers of the repeats have been characterized employing RT–PCR techniques and primer sites within the alternatively spliced region i.e. exon 14/fibronectin type III-like repeat B (Sriramarao and Bourdan, 1993) and exon 16/fibronectin type III-like repeat D (Siri et al., 1991; Wilson et al., 1996). However, in two subsequent studies where primers were selected to encompass the whole alternatively spliced region products corresponding to all theoretical sizes have been identified (Vollmer et al., 1997) together with additional unique forms (Saghizadeh et al., 1998). These studies have been performed on cell lines and tumours and only rarely performed using normal human tissues, i.e. ovarian (Wilson et al., 1996) and corneal tissue (Saghizadeh et al., 1998), so little information is available concerning potential tissue-specific patterns of splicing in normal tissues. Given this potential complexity, the wide range of reported molecular weights of the small and large protein isoforms, the potential contribution of differential glycosylation and the possibility of inclusion or exclusion of single repeats, the nature of the primary structure cannot be inferred from the description of ‘small’ and ‘large’ isoforms on gel electrophoresis and indeed the relationship between these isoforms from different cellular and tissue must be now uncertain. The characterization of tenascin-C within the fetal membrane is important to the concept of the role of a ‘wound response’ in the genesis of fetal membrane rupture. Because of the problems of interpreting potential tissue-specific patterns of splicing from protein isoform examinations, we have investigated the pattern of its alternative splicing employing RT– PCR and a wide range of exon-specific primers.

Materials and methods Tissue collection and processing Tissues were obtained from the Leicester Royal Infirmary Maternity Hospital and ethical approval was granted by the ethical committee of the Leicester Royal Infirmary National Health Authority Trust. Placentae with attached fetal membranes and umbilicus were collected from term elective Caesarean sections, the criteria being previous Caesarean section or breech presentation. Whole specimens of dissected fetal membranes as well as a scrape from the maternal aspect of the membranes, i.e. the decidua, were obtained. Specimens of

Wharton’s Jelly and artery were dissected from the umbilical cord. Specimens representing the chorionic plate and chorionic villi were dissected from the placenta. Frozen biopsies of normal skin and tonsil were obtained from the Department of Pathology. The specimens were immediately placed into Solution D (4 mol/l guanidinium thiocyanate, 0.025 mol/l sodium citrate pH 7.0, 0.5% N-laurylsarcosine NaCl, 1 mol/l 2-β-mercaptoethanol) on ice. Adjacent specimens were placed in formal saline and wax-embedded for histological examination to confirm the site of biopsy. A small quantity of antifoam C (Sigma, Poole, Dorset, UK) was added to Solution D to prevent foaming. Tissue was homogenized using a Janke and Kunkel Ultra Turrax T25 homogenizer fitted with an S25N–10 G probe for ~6320 s pulses at 20 000 rpm. In between pulses the homogenate was cooled on ice. The homogenized tissue was separated into 500 µl aliquots and stored at –20°C until required.

Cell line and tissue culture Established human melanoma cell line SK-MEL-28 was obtained from the American Type Culture Collection (Rockville, MD, USA). Cells were grown in α-minimal essential medium (αMEM; 10% v/v fetal calf serum (Life Technologies Ltd, Paisley, UK) either to 50 or 90% confluence, before being harvested with Solution D. Cell lysates were stored in 500 µl aliquots at –20°C. RNA purification RNA was purified according to a previously described method (Chomczynski and Sacchi, 1987). RNA was extracted from 500 µl aliquots of the Solution D homogenates to allow the whole protocol to be performed in 1.5 ml Eppendorf tubes. The following solutions were added sequentially: 50 µl of 2 mol/l sodium acetate, 500 µl water-saturated phenol, 100 µl chloroform:isoamyl alcohol mixture (24:1), with mixing after each addition. The final solution was cooled on ice for 15 min and centrifuged at 4°C for 20 min at 2000 g. The aqueous phase was removed and the total RNA precipitated in isopropanol for 1 h at –20°C. The RNA was sedimented, resuspended in Solution D, and re-precipitated in isopropanol. The final pellet was washed in 75% ethanol and dissolved in diethyl pyrocarbonate (DEPC)-treated water. The efficiency of the extraction and the integrity of the RNA was assessed by gel electrophoresis of RNA on 1% agarose and by spectrophotometrical analysis at 260 and 280 nm. RT–PCR The sequence for the oligonucleotide primers are shown in Table I. These include oligonucleotides employed for RT–PCR, Southern blotting and sequencing. The positions of these primers in relation to the exonic structure of the tenascin gene are shown in Figure 1. Forward strand primers were synthesized with a 59 biotin group for subsequent direct sequencing reactions. cDNA reactions was prepared using 1 µg total RNA in RT buffer (50 mmol/l Tris–HCl, 40 mmol/ l KCl, 5 mmol/l MgCl2, 0.5% Tween 20 v/v, pH 8.3), 10 mmol/l dithiothreitol (DTT), 1 mmol/l dNTP’s, 24 IU of Rnasin (Promega, UK), 100 pmol oligo d(T)12–18 (Amersham Pharmacia), 50 IU AMV reverse transcriptase (Promega) in a volume of 25 µl. The reaction was incubated at 42°C for 1 h. Controls were prepared from duplicate reactions incubated without AMV reverse transcriptase. PCR was performed in a Hybaid Omnigene thermocycler for 30–40 cycles. The PCR was carried with 1 µl of cDNA produced by the RT reaction in the following reagents: – AJ buffer (45 mmol/l Tris pH 8.8, 11 mmol/l (NH4)2SO4, 4.5 mmol/l MgCl2, 200 mmol/l dNTP’s, 110 µg/ml bovine serum albumin (BSA), 6.7 mmol/l β-mercaptoethanol and 4.4 mmol/l EDTA, pH 8.8), 10 pmoles of forward/reverse primers in a total volume of 50 µl. The DNA was denatured at 98°C for 5 min, held at 58°C during the addition of

1067

S.C.Bell et al.

Figure 1. Schematic diagram of fibronectin type III-like repeats of tenascin-C indicating the alternatively spliced region (empty boxes) and with the adjacent conserved repeats (shaded boxes) Within the boxes the upper numbers and letters correspond to the fibronectin type IIIlike repeat assignment and the lower numbers to the exon number assignment. The external numbers, employed to indicate the position of the polymerase chain reaction (PCR) primers T8F, T9F, T11F, T11P, T12P, T13P, T14F, T14P, AD1, T15P, T16P, T17R and T18R, refer to the nucleotide number of the full length cDNA of tenascin-C (Gulcher et al., 1991; Nies et al., 1991) The triangle indicated the position of the subsequently described ‘additional domain 1’, AD1 (Sriramarao and Bourdon, 1993). The P series were also employed as probes.

Table I. Sequence data for primers and probes Primer/probe

Sequence

Exon position

Repeat number

T8F T9F T11F T11P T12P T13P T14F T14P AD1 T15P T16P T17R T18R T25F T26P T27R

59 59 59 59 59 59 59 59 59 59 59 59 59 59 59 59

Exon Exon Exon Exon Exon Exon Exon Exon Exon Exon Exon Exon Exon Exon Exon Exon

4 5 A2 (7) A2 (7) A3 (8) A4 (9) B (10) B (10) AD1 C (11) D (12) 6 (13) 6 (13) F3 F4 F5

CAA TCC AGC GAC CAT CAA CG 3’ AGA AAG GCA GAC ACA AGA GC 3’ CAA TTT GGG AGA GGT CGT GG 3’ ACG ACC TCT CCC AAA TTG GG 3’ GCC TGG GAT TTC CAT GGA AC 3’ TCT CCC AGC TGT GGG AGA TC 3’ TCT GGT GCT GAA CGA ACT GC 3’ GTT CGT TCA GCA CCA GAG AT 3’ GCA GTG AGT GAG CGT CAC TC 3’ TTA GGT TTT CCA GAA GGG GC 3’ GTT GTC AAC TTC CGG TTC GG 3’ AGA AAA TGA CTT CCT TTG GG 3’ CGT CCA CAG TTA CCA TGG AG 3’ TGA ACA AAA TCA CAG CCC AG 3’ TGA TGG CTG AAT CTG TGT CC 3’ CAG TGG AAC CAG TTA ACG CC 3’

8 (3057–3076) 9 (3302–3321) 11 (3633–3652) 11 (3650–3631) 12 (4080–4061) 13 (4190–4171) 14 (4594–4613) 14 (4610–4591) AD1 (130–111) 15 (4741–4722) 16 (5010–4991) 17 (5290–5271) 18 (5410–5391) 25 (6323–6342) 26 (6541–6522) 27 (6656–6637)

The two terminologies employed to describe the fibronectin type III – like repeats in the alternatively spliced region are listed but the alphanumeric system is employed in this study (see Figure 1).

1 IU of Taq polymerase (Promega) and heated to 72°C for 1 min. The following cycle profile was used: –95°C for 1 min, 58°C for 45 s, 72°C for 1min, then finally held at 72°C for 10 min. PCR amplification products were loaded onto a 1% agarose gel containing 15 µg/100 ml ethidium bromide, and gels were run at 180–130 V for ~2 h. Detection was performed on a UV transilluminator and photographed using video capture equipment.

Southern blotting Southern analysis of RT–PCR generated fragments was carried out based on the method described by Southern (1975) but modified to detect filter bound hybrids using digoxigenin-labelled oligonucleotide probes, anti-digoxigenin alkaline phosphatase conjugates, and chemiluminenscent visualisation. RT–PCR product gels were pre-treated prior to transfer in denaturing and neutralizing solutions each for 1 h. DNA was transferred to positively charged nylon membrane (Boehringer Mannheim, Mannheim, Germany) by capillary transfer in 203 SSC (3 mol/l NaCl, 0.3 mol/l trisodium citrate) buffer. DNA was immobilized by UV cross-linking for 30 s on a transilluminator. Efficiency of the transfer was assessed by re-staining the gel with ethidium bromide and visualization under UV. Oligonucleotide probes representing each exon were labelled with

1068

digoxigenin-11-dUTP using terminal deoxynucleotidyl transferase (Promega). Labelling reactions were prepared in a sterile Eppendorf on ice in the following order: reaction buffer (0.2 mmol/l potassium cacodylate, 25 mmol/l Tris–HCl, 0.25 mg/ml bovine serum albumin pH 6.6), 5 mmol/l CoCl2, 0.08 mmol/l digoxigenin-11-dUTP, 100 pmol oligonucleotide, 0.4 mmol/l dATP, 50 IU of terminal transferase. The reaction was incubated at 37°C for 15 min and then placed on ice. The reaction was terminated by adding EDTA to a final concentration of 25 mmol/l and labelled probes were stored at –20°C. Nylon filters were pre-wetted in 63 SSC and placed between two nylon meshes in a rotary hybridization tube. Pre-hybridization was carried out in 50 ml hybridization buffer (53 SSC DEPC, 30 µg/ml denatured salmon sperm DNA, 0.1% N-laurylsarcosine NaCl, 0.02% sodium dodecyl sulphate, 30% deionized formamide, 2% w/v blocking reagent (Boehringer Mannheim) in wash buffer (100 mmol/l maleic acid, 150 mmol/l NaCl, pH7.5) at 37°C for 1 h in a Hybaid hybridization rotary oven. Membranes were probed with single oligonucleotides at 5 ng/ml and hybridized overnight at 37°C. Posthybridizations were carried at 37°C with 23 SSC/0.1% SDS/30% formamide twice for 10 min and subsequently with 23 SSC/0.1% SDS/40% formamide twice for 10 min. Immunological detection of the digoxigenin-labelled hybrids by

Tenascin and human fetal membranes washing the filter in wash buffer for 1 min and blocking the filter in 2% blocking reagent for 30 min. Incubation of the filter with 1:10 000 dilution of anti-digoxigenin-AP conjugate (Boehringer Mannheim) was followed by thorough washing and equilibration in chemiluminescent detection buffer (100 mmol/l Tris–HCl pH 9.5, 100 mmol/l NaCl, 50 mmol/l MgCl2). The chemiluminescent signal was produced by incubating the filter in sealed plastic bag with CDP-Star™ (Boehringer Mannheim) diluted 1:100 in the detection buffer for 5 min followed by exposure of the filter to X-ray film for 15 s – 5 min at room temperature.

Sequencing A direct method of sequencing was applied to isolated agarose gel electrophoresis purified PCR products. Briefly, 2–3 µg PCR products were purified on 1% NuSieve agarose gel following electrophoresis. Forward strands were immobilized by the addition of Dynabeads M-280 Streptavidin (Dynal, UK) and subsequently denatured with 0.1 mmol/l NaOH. The biotinylated strand was isolated using the Dynal magnetic particle concentrator and then washed to remove the other strand. The resulting beads were then used as a template for sequencing using [35S]-labelled dATP and the Sequenase® version 2.0 Kit (Promega) or BigDye terminator cycle ready reaction kit with AmpliTaq® DNA polymerase, FS (Perkin Elmer). Radioactive sequencing reactions were size fractionated on a 6% denaturing polyacrylamide gel. Gels were dried and exposed to autoradiography for 1–5 days. BigDye terminator reactions were run on a ABI Prism 377 DNA sequencer and the resulting sequence profiles were analysed using the Chromas software.

Results Isoform expression in the fetal membranes Total RNA prepared from fetal membranes was reverse transcribed to cDNA and amplified using 35–40 cycles PCR employing exon specific primers to the fibronectin type IIIlike repeat domains. These repeats are divided into conserved and alternatively spliced regions with the spliced region (exons 10–16) being situated between two conserved regions (exons 3–9 and 17–22). Initial experiments indicated that the primers based on the conserved exons immediately adjacent to the alternatively spliced domain, i.e. exons 9 and 17, were not as effective as primers based on exons 8 and 18. A series of primers sets were designed and employed using exon 8 in combination with all exons within the alternatively spliced region (Figure 1). Employing the exon primer sets T8F–T18R, covering the whole alternatively spliced domain, and T8F– T14P, the optimal PCR cycle number was selected as 40 for subsequent experiments (Figure 2). The dominant isoform amplified from fetal membrane cDNA with T8F–T18R primers was the smallest with exons 10–16 deleted (Figure 2). Several larger sized isoforms were detected in cDNA produced from the melanoma cell line SK-MEL-28 using the T8F–T18R primers, with the major isoform representing mRNA with exons 10–16 included. Figure 2 also shows that very little of the dominant fetal membrane isoform is detected in this cell line mRNA after 35 and 40 cycles. However, using T8F–T14P primers, similar sized PCR fragments whose size indicated inclusion of alternatively spliced exons in this region, were produced from the SK-MEL-28 and fetal membrane cDNAs. In order to further characterize the exon structure of tenascin-

C mRNA expressed in fetal membranes, cDNA was amplified using T8F as the forward primer and reverse primers from exons 11–18. These PCR reactions were analysed by agarose gel electrophoresis and the presence of exons confirmed by Southern blotting using a series of specific oligonucleotide probes designed to distinguish between the different exons (T11P, T12P, T13P, T14P, AD1, T15P and T16P, Table I). The results of the study are shown for one fetal membrane cDNA sample in Figure 3 and represented diagramatically for five fetal membranes in Figure 4. As shown in Figure 3, employing the exon primer sets T8F–T11P and T8F–T12P, single major bands were obtained whose size and Southern blotting results were consistent with mRNA species with the inclusion of exons 10–11 and of 10–12. However, the exon primer set T8F–T13P produced 3 major isoforms which corresponded to species resulting from the complete excision of exons 10–12 (314 bp), inclusion of exons 10, 11 and 13 (860 bp) and of exons 10–13 (1133 bp). This was confirmed by probing the blots and also by direct nucleotide sequencing of the eluted bands. A very minor band represented the inclusion of only exon 10 (587 bp). In a second minor band of 980 bp, between the two larger isoform bands, exons 8–11 were present but this band was not detectable with the exon 12 probe and could not be assigned to any combination of previously published exons boundaries. Employing the exon primer set T8F–T14P major bands corresponding to excision of exons 10–13 (461 bp), the inclusion of exon 13 (734 bp), exons 10, 11 and 13 (1280 bp), and the complete inclusion of four exons 10–13 (1553 bp) were detected, although the relative intensity of the smallest and largest varied between specimens (see Figures 2 and 3). A second minor band of 1400 bp present between the two larger isoform bands showed a similar Southern blotting profile to the 980 bp band detected with the T8F–T13P primer set. These bands were therefore isolated from the agarose gels, reamplified using T11F and either T13P or T14P and sequenced using both 59 and 39 primers. The 980 bp band comprised a ‘partial’ exon 12 sequence with a novel donor splice site 165 bases 59 into the exon (position 4060). This novel splice site was in frame with exon 13 and removed the priming site for T12P from exon 12. The 1400 bp band from T8F–T14P amplification, however, was comprised of mixed bands containing two novel splicing patterns. One pattern showed the ‘partial’ exon 12 sequence described above together with exon 13, but the other contained a normal exon 12 together with an additional ‘partial’ exon 13 sequence. The ‘partial’ exon 13 sequence showed a novel donor splice site 157 bases 59 into the exon (position 4325). This novel splice site is out of frame with exon 14, produces an amplified product only seven bases smaller than the ‘partial’ exon 12 sequence and would produce truncated protein after the first three amino acids coded by exon 14. No bands were obtained with the exon primer set T8F-AD1 indicating the absence of AD1-containing mRNA species. Amplification with the exon primer set T8F–T15P produced bands corresponding to total excision (319 bp), inclusion of exon 14 (592 bp), of exons 10, 11, 13, 14 (1411 bp) and of exons 10–14 (1684 bp). A minor band of 1530 bp between the two larger isoform bands detected in some specimens 1069

S.C.Bell et al.

Figure 2. Agarose gel electrophoresis of reverse transcription–polymerase chain reaction (RT–PCR) products from fetal membrane specimens and the cell line SK-MEL-28 mRNA using primers to T8F and T18R or T8F and T14P for 35 and 40 cycles. Unlabelled lanes show results of PCR amplification of mRNA samples without reverse transcription using the same primer set as the adjacent right-hand lane. The exon structure of the major isoforms as derived by subsequent blotting and sequencing are shown diagrammatically to the right of the gel. Oval shapes represent the exons employed for primer sites. Rectangular shapes show the included exons. The product with shaded rectangular boxes indicate the presence in the band of two products either containing ‘partial’ exons 12 with exon 13, or exon 12 with ‘partial’ exon 13. Lane L is a 100 bp DNA ladder.

Figure 3. Representative agarose gel electrophoresis of reverse transcription–polymerase chain reaction (RT–PCR) products from a fetal membrane mRNA preparation using a common forward primer T8F and the range of reverse primers T11P–T17R. Lanes show results for each primer set and are labelled with the exon number for the reverse primer. Bands whose size was consistent with predicted exon inclusions and which was confirmed by blotting and/or sequencing are labelled by their theoretical size in base pairs. *Products with bp were not consistent with the presence of complete exons and were subsequently characterized as products also containing ‘partial’ exons. Lane L 5 100 bp DNA ladder with major bands at 600 and 2072 bp. Summary data for multiple fetal membrane specimens is shown in Figure 4.

corresponded to forms containing exons 10, 11, 12 and 14 and the ‘partial’ exons 12 and 13. The exon structure for the additional bands of 865 and 1138 bp could not be determined. With the exon primer set T8F–T16P major bands, corresponded to complete excision (315 bp), inclusion of exon 14 (588 bp), inclusion of exons 10 and 14 (861 bp) and inclusion of 10–14 and 16 (1680 bp). The minor band of 1407 bp corre1070

sponded to inclusion of exons 10, 11, 13, 14 and 16. The band of 1530 bp corresponded to forms containing exons 10, 11, 12 and 14 and the ‘partial’ exons 12 and 13. With exon primer sets T8F–T17R the single major band corresponded to a completely excised isoform (322 bp). Minor bands corresponded to inclusion of exon 16 (595 bp), inclusion of exons 14 and 16 (868 bp), of exons 10, 14 and 16 (1141 bp),

Tenascin and human fetal membranes

Figure 4. Summary of significant reverse transcription–polymerase chain reaction (RT–PCR) products identified with oligonucleotide primer sets involving the forward primer located within the conserved exon 8. Circles indicate conserved exons and boxes indicate exons within the alternatively spliced region. Open circles and boxes indicate the exon specific primer site. Closed circles and boxes indicate the presence of the exon within the RT–PCR product as evidenced by Southern blotting with exon specific probes. bp indicates the theoretical predicted size of the RT–PCR product. *Products also containing ‘partial’ exons 12 or 13. In the case of products with shaded exons 12 and 13 the band contained two products, either containing ‘partial’ exons 12 with exon 13, or exon 12 with ‘partial’ exon 13. ‘Intensity’ indicates the relative intensity of the corresponding RT–PCR product on gel electrophoresis by visually comparing multiple specimens on the same gel. The exonic structure of the RT–PCR products of 865 and 1138 bp obtained with the primer set T8F and T15R were not elucidated and are not shown.

of exons 10, 11, 13, 14 and 16 (1687 bp) and of exons 10–14 and 16 (1960 bp). The minor band at ~1810 bp corresponded to forms containing exons 10, 11, 12, 14 and 16 and the ‘partial’ exons 12 and 13. With exon primer set T8F–T18R a single major band was obtained which corresponded to complete excision (442 bp), and a minor band to the inclusion of exon 16 (715 bp). In the majority of fetal membrane RNA preparations the exon primer set T8F–T18R produced two minor bands corresponding to products containing exons 10, 11, 13, 14 and 16 (1807 bp) and exons 10–14 and 16 (2080 bp) (see Figure 5, lane 2 and Figure 8, lane 2). The minor band

at ~1930 bp corresponded to forms containing exons 10, 11, 12, 14 and 16 and the ‘partial’ exons 12 and 13. The minor bands of 500 and 620 bp detected with the primer sets T8F– T17R and T8F–T18R respectively, although not corresponding to any exon boundary, hybridized with the exon 16 specific probe (see Figure 5, lanes 2, 4 and 6 for results for T8F– T18R-derived products hybridized with exon 12, 14 and 16 specific probes) which indicates that they contained a ‘partial’ exon 16. To provide further information primer sets based on the forward primers T11F and T14F were employed. The results 1071

S.C.Bell et al.

Figure 5. Representative Southern blot of reverse transcription–polymerase chain reaction (RT–PCR) products from SK-MEL-28 and fetal membrane derived mRNA using primers T8F and T18R. The RT–PCR products for SK-MEL-28 mRNA are shown in lanes 1, 3 and 5 and fetal membrane mRNA in lanes 2, 4 and 6. The membrane was hybridized overnight with the following exon specific oligonucleotide probes: lanes 1 and 2 with T16P, lanes 3 and 4 with T14P and lanes 5 and 6 with T12P. The exon structure of the major isoforms are shown in the diagram to the right of the blot. Oval shapes represent the primer sites and the rectangular shapes show the included exons. In the case of product with shaded exon 12 and 13 the band contained two products, either containing ‘partial’ exons 12 with exon 13, or exon 12 with ‘partial’ exon 13.

Figure 6. Representative agarose gel electrophoresis of reverse transcription–polymerase chain reaction (RT–PCR) products from fetal membrane mRNA using a common forward primer T11F and the range of reverse primers T13P–T16R. Lanes show results for each primer set and are labelled with the exon number for the reverse primer. Bands whose size was consistent with predicted exon inclusions and which was confirmed by blotting and sequencing are labelled by their theoretical size in base pairs. *Products with bp were not consistent with the presence of complete exons and were subsequently characterized as products also containing ‘partial’ exons. Lane L 5 100 bp DNA ladder with major bands at 600 and 2072 bp. Summary data for multiple fetal membrane specimens is shown in Figure 7.

for forward primer TF11 are shown for one fetal membrane in Figure 6 and all are diagramatically represented for five fetal membranes in Figure 7. The anticipated single band at 450 bp was detected with the T11F–T12P set but with T11F– T13P primers a pair of strong bands corresponding to the exclusion (294 bp) and inclusion (567 bp) of exon 12 were amplified. Similarly with T11F–T14P the pair of bands corresponded to the inclusion of only exon 13 (704 bp) and of exon 12 and 13 (977 bp). With the primer set T11F–T16P two bands corresponded to isoforms with exons 11, 13, and 14 (831 bp) and with exons 10–14 (1104 bp). Two faint bands corresponded to forms with complete exon excision (285 bp) and inclusion of the single exon 15 (558 bp) respectively. 1072

When employing these primer sets a minor band was always detected between the two largest bands, which represented the isoforms excluding and including exon 12, and corresponded to forms containing the ‘partial’ exon 12 i.e. 410 bp (primer set T11F–T13P), or a mixture of forms containing exon 12 and ‘partial’ exon 13 and ‘partial’ exon 12 with exon 13, i.e. 820 bp (primer set T11F–T14P) and 950 bp (primer set T11F– T16P). The exon primer sets T14F–AD1 and T14F–T15P produced no significant bands. With T14F–T16P, a strong band corresponded to a form excluding exon 15 (143 bp) and a minor band to a form including exon 15 (426 bp). The exon primer sets T14F–T17R and T14F–T18R both produced a single major band corresponding to species with only exon 16

Tenascin and human fetal membranes

Figure 7. Summary of significant reverse transcription–polymerase chain reaction (RT–PCR) products identified with oligonucleotide primer sets involving forward primers based upon alternatively spliced exons 11 and 14. Circles indicate conserved exons and boxes exons within the alternatively spliced region. Open circles and boxes indicate the exon-specific primer site. Closed circles and boxes indicate the presence of the exon within the RT–PCR product as evidenced by Southern blotting with exon specific probes. bp indicates the theoretical predicted size of the RT–PCR product. *Products also containing ‘partial’ exons 12 or 13. In the case of product with shaded exons 12 and 13 the band contained two products, either containing ‘partial’ exons 12 with exon 13, or exon 12 with ‘partial’ exon 13. ‘Intensity’ indicates the relative intensity of the corresponding RT–PCR product on gel electrophoresis by visually comparing multiple specimens on the same gel.

included (450 and 543 bp respectively). They also amplified minor bands corresponding to the complete excision of exons AD1, 15 and 16 (160 and 270 bp respectively) and the inclusion of exons 15 and 16 (700 and 816 bp respectively).

Comparative tissue expression Total RNA was prepared from a number of pregnancy-related and adult tissues and reverse transcribed to DNA and amplified using PCR employing the exon specific primers sets T8F– T14P and T8F–T18R to determine the tissue distribution of the isoforms. Employing the T8F–T14P primers, bands corresponding to inclusion of the complete exon 10–14 domain, with and without exon 12, were detected in umbilical cord specimens, Wharton’s jelly and vessels and SK-MEL-28 cells as well as fetal membranes (data not shown). In contrast to SK-MEL-28 cells these tissues also showed a major band corresponding to the complete excised domain from 10 to 13. These forms were not detected in placental specimens. Figure 8 shows the results employing the exon primers T8F–T18R. Bands corresponding to the larger isoforms with complete inclusion except for exons 12, AD1 and 15, or except exons AD1 and 15, were only detected in the dissected Wharton’s jelly of the umbilical cord (Figure 8, Lane 5), some specimens of fetal membranes (Figure 8, Lanes 1 and 2) and SK-MEL28 cells (Figure 8, Lane 10) The major form in all tissues corresponded to the completely excised isoform.

Discussion We have employed RT–PCR to attempt to assess the actual composition and relative expression of alternatively spliced tenascin-C mRNA species present in human fetal membranes. However, particularly when the primer sets employed are based within the alternatively spliced region, the approach is problematic. Firstly, if alternative products are produced with different primer sets, suggesting alternative splice sites within the region, the actual combination represented in individual mRNA species is uncertain and can only be confirmed by cloning. Secondly, although the relative intensity of bands with a particular primer set may give a semi-quantitative estimation of the relative amounts of mRNA species containing the splicing pattern in that region, these all may represent minor mRNA species and reflect infidelity in the spicing mechanism. However, in the present study by selecting primer sets in a local manner both within, and encompassing the alternatively spliced region, a clearer picture of the splicing pattern may be deduced. According to previous studies involving the sequencing of clones, albeit primarily derived from tumour cell lines, conserved exon 9 acts as a donor to exons 10, 14 and 16 within the alternatively spliced region (Siri et al., 1991; Sriramarao et al., 1993). In the present study products can be identified that indicate that exon 9 may act as a donor to all exons excepting 11 and 12. However, using primers encompassing 1073

S.C.Bell et al.

Figure 8. Agarose gel electrophoresis of reverse transcription–polymerase chain reaction (RT–PCR) products from multiple tissues using primers T8F and T18R demonstrating the range of common isoforms expressed. Lanes 1–10 represent the following tissue samples: (lanes 1 and 2) fetal membrane; chorionic plate (3), placenta (4), Wharton’s jelly (5), umbilical artery (6), first trimester decidua (7), adult skin (8), tonsil (9) and the SK MEL-28 melanoma cell line (10). Adjacent unlabelled lanes show results of PCR amplification of mRNA samples without reverse transcription. The exon structure of the major isoforms are shown in the diagram at the right of the gel. Oval shapes represent the exons employed for primer sites and rectangular shapes show the included exons. In the case of product with shaded exons 12 and 13 the band contained two products, either containing ‘partial’ exons 12 with exon 13, or exon 12 with ‘partial’ exon 13.

the whole alternatively spliced region it was apparent that exon 9 was principally linked to either exon 10, 16, or exon 17, the latter representing the complete excision of the alternatively spliced region. Interestingly we also identified a product containing only a partial exon 16 suggesting that exon 9 may act as a donor to an internal acceptor site within exon 16. Therefore the isoform containing only exons 14 and 16, detected in malignant ovarian tissue (Wilson et al., 1996) and tumour-derived cell lines (Siri et al., 1991), is unlikely to be a major product in fetal membranes. Previous studies had concluded that if exon 10 is used as the acceptor, exons 10– 13 are also included, i.e. isoforms corresponding to inclusion or exclusion of the four exons 10–13 as a single cassette (Siri et al., 1991; Sriramarao and Bourdan, 1993; Wilson et al., 1996). However although our observations also indicate that the exons 10–14 are spliced as a cassette it is also apparent that a potentially important internal splice event occurs within this cassette. PCR product analysis supported the tight linking of exon 10 with exon 11, although we also identified minor forms suggesting splicing to exon 13 and 14. Splicing of exon 10 with exon 14 was also recently identified in specimens of normal, malignant and reactive oral mucosae (Mighell et al., 1997). Exon 11 appeared to act as a donor to a number of exons apart from exon 12, i.e. 13, 15 and 16, and this was supported by Mighell et al. (1997) who identified exon 11 linked to exon 13, 14 and 15. However, our studies would indicate that the principal acceptors to exon 11 are exons 12 and 13 and that the quantitatively most important splicing event within the exon 10–14 cassette is the alternative splicing of exon 12. It is very apparent that when employing a forward primer based upon the alternatively spliced exon 11 and a range of reverse primers up to exon 16, two amplified products were identified in each case corresponding to the inclusion and exclusion of exon 12. Indeed employing the whole 1074

range of primer sets in the present study pairs of products corresponding to the inclusion and exclusion of exon 12 were identified. That these pairs can appear of equal intensity suggests that this alternative splicing at exon 12 is important and not a minor event. Although not reported in the study of Siri et al. (Siri et al., 1991), we also identified exon 12 included and excluded forms of the largest PCR products from the tumour-derived cell line SK-MEL-28, raising the possibility that this may not be a feature specific to tenascin-C expression in fetal membranes. Within the cassette region, when exon 12 was present, it appeared to be exclusively linked to exon 13. Exon 13 was exclusively linked to exon 14. Another splicing event associated with the exon 10–14 cassette in this tissue was indicated by the consistent presence of products, albeit minor, with a ‘partial’ exon 12 linked with exon 13, or exon 12 linked with a ‘partial’ exon 13. Within exon 12 we identified a novel internal donor splice site in frame with exon 13 consistent with the possibility that the product could be translated into a novel isoform. Previous studies indicate that alternative splicing involving exon 14 as the donor may include all the possible remaining alternative exons AD1, 15, 16 and the conserved exon 17 (Siri et al., 1991; Sriramarao and Bourdan, 1993; Wilson et al., 1996; Mighell et al. 1997). Employing a forward primer based upon exon 14, we identified forms linking this exon with either exons 15, 16 or the conserved exon 17, however the most intense product corresponded to exon 14 linked with 16. In our study we found no evidence of inclusion of AD1 (Sriramarao and Bourdan, 1993) or other related ‘additional domains’ such as AD2 (Mighell et al. 1997) and, although employing exon 15-based primers, a range of products were identified these were of very low intensity. This was supported by the larger spanning primer sets. As shown diagramatically in Figure 9 the overall splicing pattern would indicate that the major alternatively spliced

Tenascin and human fetal membranes

Figure 9. Diagrammatic representation of the pattern of alternative splicing of tenascin-C pre-mRNA in fetal membranes. Heavy lines indicate the preferred splicing pattern as deduced from the reverse transcription–polymerase chain reaction (RT–PCR) products. The products containing ‘partial’ exons 12, 13 or 16 are not shown.

isoforms in fetal membranes correspond to inclusion exon 16 alone, and the inclusion of the cassette 10–14, with and without the inclusion of exon 12, together with exon 16. In appropriate primer sets, major products corresponding to the total exclusion of the alternatively spliced region were always detected but, given potential difference in amplification efficiencies of isoforms in the non-quantitative PCR technique, we cannot infer the relative abundance of the excluded to the included forms. This is illustrated in Figure 2 where although the included forms would appear to be minor by employing the T8F–T18R primer set, employment of the T8F–T14P primer set indicates significant production. Additionally it must be considered that if relative translation efficiencies of these mRNA species are different their relative levels may not be reflected by the relative abundance of protein isoforms. However if such a pattern of splice variants is reflected by translated protein isoforms it may have to be considered that the ‘small’ and ‘large’ tenascin isoforms detected in gels by many investigators do not correspond to ‘complete’ inclusion or inclusion of the alternatively spliced region as previously suggested. Domains encoded by exons within the alternatively spliced region of tenascin must underlie specific properties of the large isoform, which have been linked to cellular processes such as cell proliferation, migration and induction of focal adhesion loss (Murphy-Ullrich et al., 1991; Chung et al., 1996). The inclusion of the exon 10–14 cassette may suggest a link between this region to these properties of the large isoforms and this may be related to the sites of potential glycosylation encoded within exons 10–13 (Nies et al., 1991). If the specific exon 12 included and excluded PCR products identified in this study are translated into their corresponding protein isoforms, i.e. inclusion and exclusion of the fibronectin type III-like repeat A3, then this may have functional significance in terms of the production tenascin isoforms exhibiting differential susceptibility to proteolytic degradation. Although tenascin-C has been reported to be degraded by the serine proteases cathepsin G (Imai et al., 1994) and plasmin (Gundersen et al., 1997) the ‘small’ and ‘large’ isoforms appear to exhibit differential susceptibility to the MMPs (Siri et al., 1995). MMP-7 can digest both ‘small’ and ‘large’ isoforms since it

can cleave adjacent to the globular amino terminal domain and within the alternatively spliced region over the fibronectin type III-like repeats A3-D (Siri et al., 1995). However the sole site susceptible to proteolytic cleavage by MMP-2 and MMP-3 is found within the exon 12-encoded fibronectin type III-like repeat A3, therefore ‘small’ tenascin is resistant to MMP-2 and MMP-3 digestion and ‘large’ susceptible to digestion (Siri et al., 1995). However our studies suggest the possibility that ‘large’ tenascin isoforms corresponding to the inclusion or exclusion of the single exon 12 could be produced which would therefore be either susceptible or resistant to MMP-2 and MMP-3-mediated digestion. Of interest the additional isoform representing inclusion of ‘partial’ exon 12 would also be translated into a ‘large’ isoform resistant to MMP-2 and MMP-3 mediated digestion, since based upon the Nterminal sequence of the resultant fragments (Siri et al., 1995) the coding sequence for this site would be absent. Tenascin-C in the presence of fibronectin up-regulates gene expression for MMP-1, MMP-3 and MMP-9 in rabbit fibroblasts (Tremble et al., 1994). Since all isoforms of tenascin appear resistant to MMP-9 (Imai et al., 1994; Siri et al., 1995), the relative expression of exon 12 included and excluded ‘large’ isoforms may provide a mechanism by which processes specifically associated with the ‘large’ isoform could be controlled by MMP-2/MMP-3 mediated digestion. In this context their relative expression in the fetal membrane will relevant since increased values of MMP-9 in the presence of constitutive values of MMP-2 have been reported in this tissue prior to their rupture at term (McLaren et al., 1999) and during labour (Vadillo-Ortega et al., 1995). The expression of a variety of splicing variants of tenascin mRNA in fetal membranes, including those that would encode ‘large’ isoforms, supports the concept that processes analogous to tissue remodelling in the ‘wound response’ are occurring in normal fetal membranes prior to labour and delivery. Their identity now enables the design of future studies to determine the relationship between these mRNA isoforms and the nature of tenascin at the protein level and whether alterations in their relative abundance or quantitative expression is associated with processes leading to fetal membrane rupture. 1075

S.C.Bell et al.

Acknowledgements The authors gratefully acknowledge the financial support provided by a grant awarded by Wellbeing and thank Mrs. S.Figgett for expert technical assistance.

References Bell, S.C. and Malak, T.M. (1997) Structural and Cellular Biology of the Fetal Membranes. In Romero, R., Elder, M.G. and Lamont, R.F. (eds), Preterm Labor. Churchill Livingstone, New York and London, pp. 401–428. Borsi, L., Balza, E. Gaggero, B. et al. (1995) The alternative splicing pattern of the tenascin-C pre-mRNA is controlled by the extracellular pH. J. Biol. Chem., 270, 6243–6245. Chiquet-Ehrismann, R. (1990) What distinguishes tenascin from fibronectin? FASEB J., 4, 2598–2604. Chiquet-Ehrismann, R. (1993) Tenascin and other adhesion-modulating proteins in cancer. Semin. Cancer Biol., 4, 301–310. Chomczynski, P. and Sacchi, N. (1987) Single step method of RNA isolation by acid guanidium thiocyanate–phenol–chloroform extraction. Anal. Biochem., 162, 156–159. Chung, C.Y. and Erickson, H.P. (1994) Cell, surface annexin II is a high affinity receptor for the alternatively spliced segment of tenascin-C. J. Cell. Biol., 126, 539–548. Chung, C.Y., Murphy-Ullrich, J.E. and Erickson, H.P. (1996) Mitogenesis, cell migration, and loss of focal adhesions induced by tenascin-C interacting with its cell surface receptor annexin II. Mol. Cell. Biol., 7, 883–892. Crossin, K.L. (1996) Tenascin : a multifunctional extracellular matrix protein with a restricted distribution in development and disease. J. Cell. Biochem., 61, 592–598. Erickson, H.P. (1993) Tenascin-C, tenascin-R and tenascin-X : a family of talented proteins in search of functions. Curr. Opin. Cell. Biol., 5, 869–876. French, J.I. and McGregor, J.A. (1996) The pathobiology of premature rupture of membranes. Sem. Perinatol., 20, 344–368. Gherzi, R., Carnemolla, B., Siri, A. et al. (1995) Human tenascin gene. Structure of the 59-region, identification, and characterization of the transcription regulatory sequences. J. Biol. Chem., 270, 3429–3434. Gundersen, D., Tran-Thang, C., Sordat, B. et al. (1997) Plasmin-induced proteolysis of tenascin-C: modulation by T lymphocyte-derived urokinasetype plasminogen activator and effect on T lymphocyte adhesion, activation, and cell clustering. J. Immunol., 158, 1051–1060. Gulcher J.R., Nies, D.E., Alexakos, M.J. et al. (1991) Structure of the human hexabrachion (tenascin) gene. Proc. Natl. Acad. Sci. USA, 88, 9438–9442. Imai, K., Kusakabe, M.J., Sakakura, T. et al. (1994) Susceptibility of tenascin to degradation by matrix metalloproteinases and serine proteinases. FEBS Letts., 352, 216. Kelly, T. (1995) The pathophysiology of premature rupture of the membranes. Curr. Opin. Obstet. Gynaecol., 7, 140–145. Mackie, E.J., Halfter, W. and Liverani, W. (1988) Induction of tenascin in healing wounds. J. Cell. Biol., 107, 2757–2767. Malak, T.M. and Bell, S.C. (1993) The structural characteristics of spontaneously ruptured fetal membranes: their relationship to membrane rupture and parturition. Cont. Rev. Obstet. Gynaecol., 5, 117–123. Malak, T.M. and Bell, S.C. (1994) Structural characteristics of term human fetal membranes (amniochorion and decidua): a novel zone of extreme morphological alteration within the rupture site. Br. J. Obstet. Gynaecol., 101, 375–386. Malak, T.M. and Bell, S.C. (1996) Fetal membranes structure and prelabour rupture. Fetal Maternal Med. Rev., 8, 143–164. Mighell, A.J., Thompson, J., Hume, W.J. et al. (1997) Human tenascin-C: Identification of a novel type III repeat in oral cancer and of novel splice variants in normal, malignant and reactive oral mucosae. Int. J. Cancer, 72, 236–240. McLaren, J., Taylor, D.J. and Bell, S.C. (1999) Increased pro-matrix metalloproteinase-9 in term fetal membranes overlying the cervix prior to labour: implications for membrane remodelling and rupture. Am. J. Obstet. Gynecol. in press. McParland, P.C. and Bell, S.C. (1998) Localization of the extracellular matrix protein tenascin-C at the materno–fetal interface during pregnancy. Br. J. Obstet. Gynaecol., 105 (Suppl. 17), 211. McParland, P.C., Pringle, J.H. and Bell, S.C. (1998) Tenascin and the fetal membrane wound hypothesis – programming for fetal membrane rupture? Br. J. Obstet. Gynaecol., 105, 1223–1224. McParland, P.C., Taylor, D.J. and Bell, S.C. (1999) Myofibroblast

1076

differentiation in the connective tissues of the amnion and chorion of term human fetal membranes : implications for fetal membrane rupture and labour. Placenta, in press. Murphy-Ullrich, J.E., Lightner, V.A., Aukhil, I. et al. (1991) Focal adhesion integrity is downregulated by the alternatively spliced domain of human tenascin. J. Cell. Biol., 115, 1127–1136. Nies, D.E., Hemesath, T.J., Kim, J. H. et al. (1991) The complete cDNA sequence of human hexabrachion (Tenascin) A multidomain protein containing unique epidermal growth factor repeats. J. Biol. Chem., 266, 2818–2823. Parry, S. and Strauss, J.F. (1998) Premature rupture of the fetal membranes. N. Engl. J. Med., 338, 663–670. Saghizadeh, M., Khin, H.L., Bourdan, M.A. et al. (1998) Novel splice variants of human tenascin-C in normal and bullous keratopathy corneas. Cornea, 17, 326–332. Siri, A., Carnemolla, B., Saginati, M. et al. (1991) Human tenascin: primary structure, pre-mRNA splicing patterns and localization of the epitopes recognised by two monoclonal antibodies. Nucleic Acids Res., 19, 525–531. Siri, A., Knauper, V., Veirana, N. et al. (1995) Different susceptibility of small and large human tenascin-C isoforms to degradation by matrix metalloproteinases. J. Biol. Chem., 270, 8650–8654. Southern, E.M. (1975) Detection of specific sequences among DNA fragments separated by gel electrophoresis. J. Mol. Biol., 98, 503–509. Sriramarao, P. and Bourdon, M.A. (1993) A novel tenascin type III repeat is part of a complex of tenascin mRNA alternative splices. Nucleic Acids Res., 21, 163–168. Stetler-Stevenson, W.G., Aznavoorian, S. and Liotta, L.A. (1993) Tumor cell interactions with extracellular matrix during invasion and metastasis. Ann. Rev. Cell. Biol., 9, 541–573. Tremble, P., Chiquet-Ehrismann, R. and Werb. Z. (1994) The extracellular matrix ligands fibronectin and tenascin collaborate in regulating collagenase gene expression in fibroblasts. Mol. Cell. Biol., 5, 439–453. Vadillo-Ortega, F., Gonzalez-Avila, G., Furth, E.E. et al. (1995) 92-kd Type IV collagenase (Matrix metalloproteinase-9) activity in human amniochorion increases with labor. Am. J. Pathol., 146, 148–156. Vollmer, G., Tan, M.I. Wunsche, W. and Frank, K. (1997) Expression of tenascin-C by human endometrial adenocarcinoma and stromal cells : heterogeneity of splice variants and induction by TGF-beta. Biochem. Cell. Biol., 75, 759–769. Vrucinic-Filippi, N. and Chiquet-Ehrismann, R. (1993) Tenascin function and regulation of expression. S.E.B. Symp., 47, 155–162. Whitby, D.J., Longaker, M.T., Harrison, M.R. et al. (1991) Rapid epithelialisation of fetal wounds is associated with the early deposition of tenascin. J. Cell. Sci., 99, 583–586. Wilson, K.E., Langdon, S.P., Lessells, A.M. and Miller, W.R. (1996) Expression of the extracellular matrix protein tenascin in malignant and benign ovarian tumours. Br. J. Cancer, 74, 999–1004. Received on January 21, 1999; accepted on August 6, 1999