Novel tenascin variants with a distinctive pattern of expression in the avian embryo

Tenascin is a multimeric glycoprotein found in the extracellular matrix of embryos and tumors (Chiquet and Fambrough, 1984; Chiquet-Ehrismann et al., 1986; Grumet et al., 1985). Tenascin is especially prominent in the developing central nervous system, in the matrix lining the pathways of migratory cells, in mesenchyme at sites of mesenchymal-epithelial interactions and in developing connective tissues (for reviews see Erickson and Bourdon, 1989; Chiquet et al., 1991b; Chiquet-Ehrismann, 1993). Unlike fibronectin, which has a more ubiquitous distribution in embryonic mesenchyme, tenascin acts as an anti-adhesive substratum for cells in vitro. Cells will attach and migrate on tenascin-coated tissue culture plastic, but they fail to spread and form strong adhesions (Spring et al., 1989; Halfter et al., 1989; Lotz et al., 1989). Moreover, when cells are first allowed to spread on fibronectin or native basal laminae and tenascin is then added to the culture medium, these cells will lose their strong attachments and become rounded (Chiquet-Ehrismann et al., 1988; Halfter et al., 1989; Lotz et al., 1989). If the cells used are from wing bud cultures, the cell rounding apparently promotes chondrogenesis (Mackie et al., 1987). For these reasons, tenascin is believed to play a role in regulating cell-matrix interactions in a way that may promote cell motility and/or differentiation.

Several forms of tenascin have been found in the extracellular matrix that are the result of differential mRNA splicing. Three splice variants have been identified in the chicken (Jones et al., 1989; Spring et al., 1989) with apparent relative molecular masses of 230×10³ (Tn230), 200×10³ (Tn200) and 190×10³ (Tn190). These forms differ from one another in their number of fibronectin type III repeats. Tn230 has eleven fibronectin type III repeats; two of these repeats are not found in Tn200 and exons encoding three of these repeats are spliced from transcripts encoding Tn190. This diversity may reflect different functional forms of tenascin, as splice variants have different affinities for other matrix molecules (Chiquet-Ehrismann et al., 1991) and cell surface glycoproteins (Zisch et al., 1992). Both immunohistochemistry with splice variant-specific antibodies (Matsuoka et al., 1990; Chiquet-Ehrismann et al., 1991; Kaplony et al., 1991) and in situ hybridization with splice variant-specific probes (Prieto et al., 1990; Mackie and Tucker, 1992; Tucker, 1993) have revealed distinct tissue distributions and cellular origins for the variants as well. Comparison of avian tenascin with tenascins from other species shows that the spliced, variable domain of the molecule has been conserved through evolution (Gulcher et al., 1989; Saga et al., 1991; Weller et al., 1991).

It was noted earlier that the primary sequences of fibronectin type III repeats from chicken and human tenascin correspond with one another colinearly. That is, the first repeat from human tenascin is more similar to the first repeat in chicken than to other human fibronectin type III repeats, and so on. Within the variable domain, sequence analysis revealed that one of the human fibronectin type III repeats, repeat ‘C’, did not have a counterpart in the chicken. For this reason, we have undertaken a study of genomic sequence and have found a homologous avian exon. Monospecific antibodies raised against avian repeat ‘C’ fusion proteins have identified this previously unknown splicing variant both in tenascin from chicken embryo fibroblast-conditioned medium and in tissue sections. Finally, the tissue distribution of transcripts encoding tenascin with repeat ‘C’ was examined by reverse transcriptase PCR and in situ hybridization. By investigating the fibronectin type III repeats surrounding repeat ‘C’-containing mRNA, two further novel fibronectin type III repeats were discovered. Whereas one of these repeats represents an entirely new sequence, the other one is the homologue of the recently published human tenascin ‘AD1’ repeat (Sriramarao and Bourdon, 1993). Thus our results provide further evidence that the structure of tenascin is phylogenetically conserved, and that tenascin may have multiple functions in development that are regulated via the expression of different forms of the molecule.

Genomic clones of the tenascin gene were isolated from a library in EMBL3 (Clontech CL1004D) using a probe encompassing tenascin repeats ‘A’ and ‘B’. One of the clones isolated ended in repeat ‘D’. This DNA was used as a template for the PCR reaction using degenerate primers developed after the sequence of human tenascin repeat ‘C’. The following oligonucleotides were used: 5′-GATGAATTC-GAYATHAAYCCNTAYGG-3′ and 5′-CTAGTCGACSWNAC-CATNACYTCRTA-3′ (corresponding to the regions marked in Fig. 1). The resulting PCR product was subcloned and sequenced. Oligonucleotides corresponding to the obtained internal sequence were used to sequence the entire chicken repeat ‘C’ plus the adjacent introns of the original genomic clone. Oligonucleotides were then synthesized using the terminal sequences of this repeat to amplify the entire repeat for the subcloning into the bacterial expression vectors and to screen a chicken embryo cDNA library (Clontech) to isolate a cDNA clone containing repeat ‘C’. PCR products and cDNAs were subcloned into the Bluescript vector (Stratagene). Both strands were sequenced using the Sequenase Kit (U.S. Biochemicals).

For the production of fusion proteins containing selected fibronectin type III repeats of tenascin, the commercially available plasmid pEX-1 (Genofit) was used. Since it was shown that after the deletion of the C-terminal half of the β-galactosidase the stability and efficiency of the production of fusion proteins increased, the plasmid was modified according to Wingender et al. (1989). The plasmid pEX-1 was cleaved at the unique EcoRV and XbaI sites to delete part of the β-galactosidase coding region and the two partially complementary oligonucleotides 5′-GGTAATGGTGACCCGGGTAATTAAT-TAAT-3′ and 5′-CTAGATTAATTAATTACCCGGGTCACCAT-TACC-3′ were inserted at this site. With this insertion, the modified plasmid pEX-1 contains 3861 bp. After the truncated β-galactosidasecoding region, the aminoacids NGDP precede the restriction enzyme recognition site SmaI. Stop codons in all three reading frames follow after the SmaI site. Blunt end cloning into the SmaI site links the insert to the third position of the reading frame of the β-galactosidase part. Thus the fusion proteins produced contain the N-terminal part of the β-galactosidase migrating at 54×10³M_r on SDS-PAGE plus the added molecular weight of the attached tenascin fragments. They were purified from inclusion bodies according to the procedure of Wingender et al. (1989). Alternatively tenascin fragments were expressed as fusion proteins with glutathione-S-transferase using the pGEX series of vectors (Pharmacia). Since also these fusion proteins were recovered in inclusion bodies, they were purified using the same procedure as for the β-galactosidase fusion proteins.

The monoclonal antibodies against tenascin and the location of their epitopes have been described previously (cf. Chiquet-Ehrismann et al., 1991). The antiserum against the glutathione-S-transferase repeat ‘C’ fusion protein has been raised in a rabbit using standard procedures. The antiserum was affinity purified using β-galactosidase repeat ‘C’ fusion protein coupled to CNBr-activated Sepharose (Pharmacia). The monospecific antibodies were eluted using 0.1 M glycine/HCl (pH 2.6) and the eluate was immediately neutralized. Their specificity was tested using an ELISA procedure as described in Chiquet-Ehrismann et al. (1988).

The tenascin for the immunoblotting experiment was purified from chicken embryo fibroblast conditioned medium according to Chiquet et al. (1991a). SDS-PAGE and immunoblotting was performed as described earlier (Hofer et al., 1990), with the exception that a chemiluminescence detection kit (ECL, Amersham) was used to develop the filter. Immunofluorescence of cryosections was performed as described by Mackie et al. (1987).

Tenascin variant transcripts were identified in frozen sections of chicken embryos at a variety of developmental stages using in situ hybridization. To identify the mRNA encoding repeat ‘C’, an upstream anti-sense primer corresponding to the first 22 nucleic acids of repeat ‘C’ (5′-AGGCCATGCCCCCACTGGAAAA-3′) and a downstream sense primer corresponding to the final 21 nucleic acids of repeat ‘C’ (5′-CTGTAACAGCCACAGTGCTTA-3′) were used to generate a 272 bp reverse transcriptase PCR product (TNC). TNC was identified as a probe to the sequence encoded in the ‘C’ repeat by a restriction digest with HindIII. The in situ hybridization pattern with TNC was compared to the pattern obtained with two other probes: TN230 (Tucker, 1993), a 545 bp PCR product generated from primers corresponding to the beginning of repeat ‘A’ and the end of the repeat ‘B’; and cTn8, an 1899 bp chicken tenascin cDNA that corresponds to the portion of the tenascin mRNA encoding the heptad repeats, the EGF-like repeats and the first one and one-half fibronectin type III repeats (Pearson et al., 1988). Thus, TN230 hybridizes to tenascin transcripts encoding variants with repeats ‘A’ and ‘B’, and cTn8 hybridizes to all known variant mRNAs. All probes were labelled with ³⁵S-dCTP using random primers (Promega), and unincorporated nucleotides were removed with a Sephadex G-50 mini-spin column (Worthington). To control for spurious hybridization, a 955 bp TaqI/Sau3AI pUC19 restriction fragment was similarly purified and labelled.

Chicken embryos (Hubbard Farms) at embryonic day 7 (E7), E10 and E14 were fixed by immersion overnight in ice-cold 4% paraformaldehyde in phosphate-buffered saline (PBS), cryoprotected overnight in 25% sucrose in PBS and embedded in OCT compound. Serial 12–14 μm sections were cut in a cryostat and collected on poly-L-lysine-subbed slides. Details of the in situ hybridization protocol have been published previously (Tucker and McKay, 1991). In brief, air-dried sections were prehybridized for 1 hour at room temperature in 5× SSC and 5× Denhardt’s solution, with 100 μg/ml salmon sperm DNA and 20 mM β-mercaptoethanol. Slides were then dehydrated in ethanol, and 5×10⁶ cts/minute of the appropriate probe in hybridization buffer (prehybridization buffer with 50% formamide, 20 mM Tris, 0.1% lauryl sulfate and 1% dextran sulfate; all reagents from Sigma) was added and allowed to hybridize overnight at 42 °C. The next day the slides were rinsed with 1× SSC at room temperature and at 42 °C. Sections were exposed to high-resolution X-ray film (Hyperfilmβ, Amersham), then dipped in Kodak NTB2 nuclear track emulsion.

Chicken embryo fibroblast cultures were prepared from skin of 11-day embryos and RNA was prepared after the procedure of Chomczynski and Sacchi (1987). Poly(A) RNA was isolated using the mRNA kit from Pharmacia. For the RT-PCR (reverse transcriptase and Taq-polymerase, Boehringer-Mannheim) of chicken embryo fibroblast mRNA, the following primers were used:

repeat ‘C’ forward (5′-TAGAATTCTAGCCAT-GCCCCCACTGGA-3′),

repeat ‘C’ reverse (5′-TAGAATTCTATGTAA-CAGCCACAGTGC-3′),

repeat ‘D’ reverse (5′-ATGTCGACGTCCAC-CTCGGGTTCTGC-3′),

repeat ‘5’ forward (5′-ATGTCGACCAACCAT-CAAGGGTTCGA-3′),

repeat ‘A’ forward (5′-ATGTCGACGGTGT-GATTCGAGGCTAC-3′),

repeat ‘6’ reverse (5′-TAGAATTCTATG-TTTTCAGAATTCCAG-3′),

repeat ‘B’ forward (5′-ATGAATTCATAAGT-GCTGCGGCTACA-3′),

repeat ‘AD1’ reverse (5′-ATGTCGACTGAGAT-GTGAGCGCTGCG-3′).

All products of the reactions shown in Fig. 9.II were subcloned and sequenced.

When the fibronectin type III repeats of chicken and human tenascin are aligned with each other, it is evident that the order of the fibronectin III repeats is conserved. Thus, we assigned numbers to identify the homologous constant repeats and letters to identify the corresponding repeats known to be subject to alternative splicing (cf. Vručinic’-Filippi and Chiquet-Ehrismann, 1993; Aukhil et al., 1993; Fig. 9). Repeat ‘A’ occurs in 4 versions in human tenascin, but only once in the chicken. Human repeat ‘C’ shares 35% identity with repeat ‘B’ of chicken tenascin, but this did not seem to be enough similarity to qualify as an authentic homologue, since all other homologous repeats have a counterpart in the chicken with over 50% amino acid identity. Therefore, we postulated that in the chicken a previously unidentified ‘C’ repeat could exist between the known repeats ‘B’ and ‘D’.

In our earlier studies (Spring et al., 1989), none of the cDNA clones analyzed revealed a potential repeat ‘C’. Therefore, we decided to determine if the chicken tenascin gene contains an exon encoding repeat ‘C’. Genomic clones were isolated that hybridized with repeats ‘A’ and ‘B’. One of these clones ended within repeat ‘D’ and should, if repeat ‘C’ exists in chicken, contain the putative exon. Using repeat ‘C’-specific degenerate primers (derived from the human sequence corresponding to the two regions marked by arrows in Fig. 1), we were able to amplify by PCR a product of the predicted length using the genomic clone as a template. The sequence of the amplified fragment and of the corresponding exon in the genomic clone (including flanking intron sequences) confirmed that it was indeed derived from an exon encoding a chicken repeat ‘C’ (Fig. 2). This repeat has an identity of 80% with human repeat ‘C’.

An antiserum was raised against a recombinant glutathione-S-transferase repeat ‘C’ fusion protein (GST-C, Fig. 2A). In order to purify the antibodies from this serum specific for the ‘C’ repeat, a second fusion protein between β-galactosidase and repeat ‘C’ was prepared (β-gal-C). β-gal-C was coupled to CnBr-activated Sepharose 4B and the antiserum was applied to the column. Bound antibodies were eluted by low pH and the eluate was immediately neutralized. The specificity of the antiserum and the purified monospecific antibodies were then tested by ELISA (Fig. 2A). The antiserum loaded onto the column (L) reacted both with GST-C and β-gal-C, but not with a different β-galactosidase fusion protein (β-gal-D). The flow through (FT) still reacted with the GST fusion protein, but not with β-gal-C. The eluate (E), as expected, reacts with both types of repeat ‘C’ fusion proteins.

To determine whether chicken embryo fibroblasts produce mRNA and tenascin protein containing repeat ‘C’, mRNA was isolated from chicken embryo fibroblast cultures and used for RT-PCR. Primers to amplify the constant repeats ‘1’ and ‘2’ were used in a control reaction (Fig. 2B, lane a). Primers of sequences from either end of repeat ‘C’ (Fig. 2B, lane b) or a sense primer from the beginning of repeat ‘C’ and an anti-sense primer from the beginning of repeat ‘D’ (Fig. 2B, lane c) were used to test for the presence of repeat ‘C’. In all cases, we were able to amplify a PCR product of the appropriate size. The products also contained the predicted internal restriction sites (not shown). Thus, repeat ‘C’ is present in fibroblast tenascin mRNA adjacent to repeat ‘D’. Furthermore, tenascin isolated from the conditioned medium of chicken embryo fibroblasts was analyzed by immunoblotting using the ‘C’ repeat monospecific antibody. The ‘C’ repeat antibody recognized exclusively a high-molecular-weight tenascin variant that migrated just above the larger of the three major tenascin bands recognized by an antibody that recognizes all tenascin forms (Fig. 2B, lanes e-h). Thus, both the mRNA encoding the ‘C’ repeat, and a tenascin variant containing the ‘C’ repeat, are made in primary cultures of chicken embryo fibroblasts. However, the repeat ‘C’-containing tenascin is only a minor fraction of the tenascin produced in these cells, since it is not visible in the lane containing the purified tenascin stained by Coomassie Blue and it could only be detected with the general antitenascin antiserum in a lane loaded with a ten-fold higher tenascin concentration than was used to reveal the three major variants (Fig. 2B).

The distribution of tenascin mRNA encoding repeat ‘C’ was determined by in situ hybridization in sections of chicken embryos at several stages of development. The distribution of repeat ‘C’ transcripts was compared with the distribution of mRNAs containing the variable ‘AB’ repeats as well as the distribution of total tenascin transcript.

Fig. 3A-H shows low magnification overviews (X-ray film overlays) of cross sections through two axial levels of E7 chicken embryos hybridized with tenascin probes and a pUC control probe. cTn8, which hybridizes to the mRNAs of all known tenascin forms (Spring et al., 1989), hybridizes in the perichondrium, in endothelial cells of the aorta, in the connective tissue associated with forming vertebrae, as well as in lung buds and kidney mesenchyme (Fig. 3A). Some of these regions are shown at higher magnification in Figs 4 and 5, which show dark-field images of silver grains following emulsion autoradiography. The small, punctate signal seen near the center of the E7 spinal cord with cTn8 at low magnification is clearly present in a restricted portion of the ependymal layer just dorsal to the ventral floor plate (Fig. 4B; Tucker and McKay, 1991). This ‘spot’ corresponds to the strongest concentration of tenascin itself as seen with anti-tenascin staining of a nearby section (Fig. 4A). The strong cTn8 hybridization signal seen associated with the vertebrae at low magnification is actually present in connective tissue linking adjacent vertebral laminae, i.e. the developing ligamentum flavum (Fig. 5A). In lung, the cTn8 signal is present within the epithelial cells at the very tip of the bronchiole buds, as described previously (Koch et al., 1991). The signals seen with TN230, which recognizes mRNA containing sequences encoding variable repeats ‘A’ and ‘B’, are a subset of the regions where cTn8 hybridizes: kidney mesenchyme, bronchiole tips, ligamentum flavum, aorta endothelium and spinal cord ependyma are all ‘AB’ positive (Figs 3B,F, 5B,E, 4C). In turn, the hybridization pattern with TNC is a subset of the TN230 pattern. The TNC hybridization signal is strong in the ligamentum flavum and at the tips of bronchioles, and is also seen (albeit more weakly) in the kidney mesenchyme and aorta endothelium (Figs 3C,D, 5C,F). The TNC signal in the E7 spinal cord, in contrast to cTn8 and TN230, is indistinguishable from controls (Fig. 4D).

At later stages in development, the differences in the hybridization signals seen with the tenascin splice variant probes are even more distinct than at E7. Low magnification overviews of E10 chicken sections, for example, show a strong signal in perichondrium, tendons, the base of feather buds and in spinal cord with cTn8 (Fig. 3I). TN230 hybridizes in spinal cord, tendons and at the base of feather buds (Fig. 3K), whereas the TNC hybridization signal is limited to tendons and the base of feather buds (Fig. 3M). Higher magnifications emphasize the absence of a significant hybridization signal with TNC in the spinal cord (Fig. 4G), in contrast to the robust signal in glia seen with cTn8 and TN230 (Fig. 4E,F). At E14, TNC still fails to hybridize within the spinal cord white matter, in contrast to the other probes (Fig. 3J,L,N,P). The strongest signal seen with TNC is found in tendons and at the base of feather buds (Fig. 5G), in the same regions where anti-tenascin stains the extra-cellular matrix (Fig. 5H,I). Thus, tenascin mRNA encoding repeat ‘C’ is concentrated in a subset of the cells that synthesize tenascin with repeats ‘A’ and ‘B’. Repeat ‘C’ transcripts are not detectable by in situ hybridization in perichondrium or in the spinal cord, though other tenascin mRNAs are found in these regions.

The restricted distribution of repeat ‘C’-containing mRNA and the prominent expression in embryonic tendon was confirmed by immunostaining. We stained sections of chicken embryos with the ‘C’ repeat monospecific antibody. The only prominent staining was found in the flight tendon of a chicken wing in a 10-day embryo (Fig. 6). Fig. 6A shows an overview over the section stained with anti-TnM1, which stains all tenascin variants and therefore shows the most widespread distribution. This staining was compared to the staining pattern obtained by other monoclonal tenascin antibodies and the repeat ‘C’ antibody as shown Fig. 6B-E. Anti-Tn32 does not recognize the smallest tenascin variant without extra repeats and shows a subset of the anti-TnM1 staining pattern, reflecting the presence of tenascin variants containing extra repeats ‘A’ and ‘B’ (Fig. 6C). Even more restricted is the staining using anti-Tn26, which recognizes a large tenascin variant containing extra repeat ‘D’ (Fig. 6D). The ‘C’ repeat antibody consistently stained the tendon exclusively (Fig. 6E). There was no staining of tendons with control antiserum (not shown). Whereas the perichondrium was equally stained by the three monoclonal anti-tenascin antibodies (only shown for M1, Fig. 6A), another connective tissue structure (marked by a star) was brightly stained by anti-TnM1, moderately stained by anti-Tn32 and not stained at all by anti-Tn26 or repeat ‘C’ antibody.

The next set of experiments was undertaken to investigate the nature of the tenascin variants containing repeat ‘C’. First we screened a chicken embryo cDNA library with a probe specific for repeat ‘C’. One of the clones isolated contained repeat ‘D’, preceded by repeat ‘C’, which surprisingly was not joined to a known repeat, but was preceded by a novel fibronectin type III sequence (Fig. 7). This new repeat revealed an identity of 59% to a recently discovered additional repeat in human tenascin and termed ‘AD1’ by Srimarao and Bourdon (1993). We thus decided to adopt this nomenclature and refer to this repeat as the chicken ‘AD1’ (cf Fig. 7; chAD1). We then performed three sets of RT-PCR reactions using chicken embryo fibroblast mRNA. In the first set using different combinations of primers from the published chicken tenascin repeats, bands corresponding to the previously described three tenascin variants could be amplified exclusively (Figs 8e-g, 9.I). To our surprise, when we performed RT-PCR with a primer from repeat ‘AD1’ together with primers from repeats ‘5’, ‘A’ and ‘B’, we obtained, in the latter case, a product of about 270 bp larger size than expected (Figs 8d, Fig. 9.II). The cloning and sequencing of this product revealed a further entirely novel fibronectin type III sequence, which we decided to term ‘AD2’ (Fig. 8; chAD2). A third set of RT-PCR reactions using primers of repeat ‘5’ together with repeat ‘C’ primers resulted in a number of bands spaced by about 270 bp (marked by white dots in Fig. 8h), which by restriction digests could be identified as the splicing variants shown in Fig. 9.III. Thus, chicken tenascin mRNA can contain up to six fibronectin type III repeats subject to alternative splicing as shown in the model of Fig. 9, and repeat ‘C’ occurs in multiple tenascin splice variants.

The importance of tenascin for development has recently been questioned by Saga et al. (1992), since they discovered that mice lacking tenascin developed apparently normally. This result led Erickson (1993a) to propose that tenascin could be a superfluous protein and that the sites of prominent expression of tenascin are not necessarily the sites where tenascin has a function. Tenascin is the product of a member of a multigene family consisting of three or four members (Erickson, 1993b; Chiquet-Ehrismann et al., 1993). Clearly the knocking out of genes belonging to multigene families often has no or only minor effects on the deficient mice and frequently the defects are a surprise and unpredictable. However, since in most of these cases the loss of a gene product is compensated for by other genes, the normal function of the deleted gene can not be found by this approach. Certainly redundant proteins are not superfluous, and if a protein has any function (which presumably every protein has cf. Brookfield, 1992), this function has to be performed where the protein is present. We therefore believe that it is still important to investigate the structure, function and expression pattern of tenascin in normal embryos as well as in tissue culture models to find possible functions of tenascin.

Evidence from both biochemical and morphological studies indicate that different splice variants of tenascin may have different functions. For example, low-molecular-weight tenascin has a higher affinity for both fibronectin (Chiquet-Ehrismann et al., 1991) and contactin/F11 (Zisch et al., 1992), than tenascin containing the variable domain. Moreover, when bovine endothelial cells are exposed to a fusion protein corresponding to the variable domain of human tenascin, they lose their focal adhesions (Murphy-Ullrich et al., 1991). In the embryo, individual motile cells like osteoblasts and glial cells make high-molecular-weight tenascin, whereas non-motile chondroblasts are a primary source of low-molecular-weight tenascin (Mackie and Tucker, 1992; Tucker, 1993). A possible role for high-molecular-weight tenascin in promoting invasive behavior is also provided by studies of tumors and in vitro transformation. For example, Borsi et al. (1992) have recently found that high-molecular-weight tenascin is preferentially expressed by invasive human breast carcinomas, whereas low-molecular-weight tenascin is expressed in normal breast tissue and by benign breast lesions. Moreover, when chicken embryo fibroblasts are transformed by Rous sarcoma virus middle T (Matsuoka et al., 1990), or when 3T3 cells are treated with bFGF (Tucker et al., 1993), there is a selective induction of tenascin containing the variable fibronectin type III repeats. Taken together, these data indicate that the key to understanding tenascin function may lie in our understanding the differences between alternatively spliced variants of the molecule.

Comparison of the amino acid sequences of the fibronectin type III repeats of human (Gulcher et al., 1989) and chicken (Spring et al., 1989; Jones et al., 1989) tenascin indicated that the ‘C’ repeat from the variable domain of human tenascin had yet to be identified in the chicken. Through a combination of genomic cloning with probes flanking the putative ‘C’ repeat and PCR amplification with degenerate oligonucleotides, we have found a ‘C’ repeat in chicken tenascin. Here we provide evidence that this previously unknown fibronectin type III repeat is actually present in avian tenascin: (1) monospecific antibodies to an avian repeat ‘C’ fusion protein recognize a band corresponding to a high-molecular-weight splicing variant of tenascin from medium conditioned by chicken embryo fibroblasts, (2) these antibodies stain a subset of the embryonic extracellular matrix stained by antibodies that recognize all tenascin forms, (3) appropriately sized PCR products with restriction sites characteristic for repeat ‘C’ can be amplified from chicken embryo fibroblast mRNA using primers designed from avian ‘C’ repeat sequence, and (4) a DNA probe specific for mRNA encoding repeat ‘C’ hybridizes within a subset of the embryonic tissues where other tenascin probes hybridize. Previously, we have shown that tenascin containing repeats ‘A’ and ‘B’ is made by migrating cells (osteoblasts and glia cells), at the sites of epithelial-mesenchymal interactions (feather buds, kidney and at the tips of growing lung bronchioles), and is found in embryonic tendons (Tucker, 1993). We now know that the high-molecular-weight tenascin found in these diverse regions is not all the same type. Within the CNS, most of the high-molecular-weight tenascin apparently lacks the ‘C’ repeat. This is evident from the absence of a significant hybridization signal in developing spinal cord with TNC. In contrast, the strong signal with TNC in tendon, lung and kidney and at the base of feather buds indicates that tenascin with repeat ‘C’ is a major component of high-molecular-weight tenascin concentrated at sites of epithelial-mesenchymal interactions as well as in embryonic tendon. These different sites of expression implicate repeat ‘C’ in functions that may be associated with inductive events, cell proliferation, or the binding of non-neuronal matrix components.

With the aim of determining in which type(s) of splicing variants the fibronectin type III repeat ‘C’ occurs, we discovered two novel repeats preceding repeat ‘C’. The homologue of one of them has also been identified recently in human tenascin and has been termed repeat ‘AD1’ for alternative domain 1 by Sriramarao and Bourdon (1993). We thus decided to adopt this nomenclature for the chicken tenascin repeats and termed our novel repeats ‘AD1’ preceded by ‘AD2’. In the case of the human ‘AD1’ repeat, it was suggested that its expression may be tumor associated (Sriramarao and Bourdon, 1993). It will be the subject of future studies to investigate the expression pattern of these novel repeats in the developing chicken. Furthermore, it remains to be seen whether mammalian tenascin also contains a homologue of the chicken ‘AD2’ repeat. Repeat ‘C’ does not always occur together with ‘AD1’, but appears to be present in a multitude of splicing variants in mRNA isolated from chicken embryo fibroblast cultures. These mRNAs are, however, a minority of the tenascin mRNA in these cultures, since they can only be detected using primers specific for either of the three novel repeats. In future studies, it will be most interesting to find out whether the novel repeats confer new activities to tenascin.

The sequences reported in this article can be obtained from the EMBL data base under the accession number X73833.

This work was supported in part by a grant from the National Science Foundation (BNS-9021124) to R. P. T.

We have recently sequenced the entire genomic region of about 9 kb of the chicken tenascin gene encompassing the exon encoding repeat “A” to the exon encoding repeat “C”. We confirmed the presence of the exons of all extra repeats described in the present manuscript, but did not find any other unknown exons encoding novel repeats. Therefore, we can conclude that the tenascin model presented in this paper most likely includes a complete set of all possible fibronectin type III repeats of chicken tenascin.