The primary gene transcript for the adhesive extracellular matrix glycoprotein fibronectin (FN) is alternatively spliced in three regions (EIIIA, EIIIB and V). At least one of these regions (V) has been shown to encode cell-binding sites, suggesting that splicing represents a mechanism to create functionally different forms of FN at different times and places. In order to test this hypothesis, we have examined the extent of alternative splicing of fibronectin during embryonic development. The distribution of the different spliced forms of FN mRNA in developing chicken embryos was determined using probes specific for the spliced regions in ribonuclease protection and in situ hybridization experiments. At embryonic day 2–4 (E2–4), all three spliced regions were included wherever FN mRNA was detected. At E16, however, we found spatially distinct splicing differences within the embryo, with cell-type-specific splicing excluding EIIIA and/or EIIIB in some tissues. In contrast, we did not detect exclusion of the V region. In a more detailed developmental study of the simplest of these tissues, the chorioallantoic membrane, we found that EIIIB was preferentially excluded after the completion of growth. These results suggest that FN splicing is used during development as a mechanism to create different forms of FN within the extracellular matrix by the inclusion or exclusion of specific segments. The data are consistent with an essential role for one of these segments, E111B, in the migration and/or proliferation of embryonic cells prior to their terminal differentiation and also suggest possible roles for the EIIIA segment.

The interactions of cells with the adhesive extracellular matrix (ECM) glycoprotein fibronectin (FN), can alter many aspects of cell behaviour including adhesion, migration and differentiation (Hynes & Yamada, 1982; Hynes, 1986). These interactions can occur at multiple sites within the FN molecule; cells will bind either to a cell-binding domain that contains the RGDS sequence and is recognized by a cell-surface integrin receptor (Ruoslahti & Pierschbacher, 1987; Buck & Horwitz, 1987; Hynes, 1987) or to a heparin-binding domain that is bound by cell-surface heparan sulphate proteoglycan (Saunders & Bernfield, 1988; McCarthy et al. 1988). Moreover, the interaction of the cell with two or more sites simultaneously can dramatically alter the behaviour of the cell. For example, the binding of the cell to both cell- and heparin-binding domains together can induce stress fibres terminating in focal contacts that are not seen on binding to either domain alone (Woods et al. 1986; LeBaron et al. 1988) and a second site within the cell binding domain acts synergistically with the RGDS site, both sites being required for the formation of actin microfilament bundles (Obara et al. 1988). These and other studies suggest that cooperativity between different sites on the FN molecule may be an important general mechanism by which FN–cell interactions can alter cell behaviour.

The finding of multiple sites of cell interaction is in keeping with the modular structure of FN, with three types of homologous repeating units (reviewed in Hynes, 1985). Interestingly, however, the primary FN gene transcript is alternatively spliced so as to include or exclude three distinct regions of the transcript. Two exons, termed.EIIIA and EIIIB (or ED-A and ED-B), can be completely included or excluded (Kornblihtt et al. 1985; Schwarzbauer et al. 1987a; Norton & Hynes, 1987) while a third region, V (or IIICS), can be partially excluded (Schwarzbauer et al. 1983; Kornblihtt et al. 1985; Sekiguchi et al. 1986a; Norton & Hynes, 1987). The patterns of splicing are cell-type-specific; for example, hepatocytes synthesizing plasma FN (pFN) exclude EIIIA and EIIIB, while fibroblasts and astrocytes synthesizing insoluble cellular FN largely include EIIIA and, to a lesser extent, EIIIB (Komblihtt et al. 1984b; Paul et al. 1986; Schwarzbauer et al. 1987a; Norton & Hynes, 1987).

Taken together with evidence that the different FN mRNAs are translated to give distinct FN subunits (Schwarzbauer et al. 1985; Paul et al. 1986), these findings suggest the attractive hypothesis that alternative splicing might be a mechanism by which segments of FN could be included or excluded in the final protein so as to create functionally distinct forms of FN. Recent work on the V region supports this hypothesis; this region has been shown to contain sites recognized by certain cell types which provide adhesion additional to that of the RGDS-containing cell-binding fragment and so alter the behaviour of these cells (Humphries et al. 1986, 1987, 1988; Dufour et al. 1988). However, the extent to which splicing of the V region occurs in vivo has not been established. Moreover, the roles of the other two spliced regions, EIIIA and EIIIB, in the interactions of FN are unknown and their splicing pattern in vivo remains largely undefined, although studies on total RNA from embryos indicate that these two segments are prevalent in FN mRNA from early embryos (Norton & Hynes, 1987).

In order to estabfish the extent of alternative splicing in vivo, and as a further step towards understanding the role of the alternatively spliced segments in the control of cell behaviour, we have analysed the distribution of alternatively-spliced FN mRNAs in situ in developing embryos where the widespread presence of FN in the ECM suggests an important role in the molecular mechanisms involved in growth and development. Using in situ hybridization and ribonuclease protection experiments with probes specific to the different spliced regions, we show that splicing is developmentally regulated, as the patterns of FN mRNA produced are temporally and spatially distinct both within and between different tissues. These results suggest that alternative splicing is indeed used as a mechanism to create different forms of FN in the embryonic ECM.

Preparation of RNA

Embryonic day 16 tissues were removed after the embryo had been killed by decapitation, and these tissues were then frozen in liquid nitrogen. The chorioallantoic membrane (CAM) was obtained by removing the embryo and yolk sac from the egg and then stripping the CAM from the inside of the eggshell. Frozen tissues were pulverized in a pestle and mortar on dry ice and RNA was extracted in guanidinium thiocyanate and centrifuged through CsCl as described by Chirgwin et al. (1979).

Ribonuclease protection analysis

The probes used to examine the splicing pattern of EIIIA, EIIIB and V have been characterized previously as A, B and VI in Norton & Hynes (1987), and the templates used to generate these probes were a kind gift of Dr P. Norton. The protocol used was essentially that described by Norton & Hynes (1987). Uniformly labelled RNA probes were prepared by in vitro transcription using SP6 or T7 polymerase with α- [32P]-UTP (800Ci mmol−1) included in the reaction mixture, and these probes were subsequently purified on denaturing polyacrylamide gels. For hybridization a molar excess of probe was added to 5 μg of CAM RNA or 10 μg of embryonic tissue RNA, and yeast tRNA was added to a total of 30 μg. After denaturation at 80 °C, hybridization was performed in 80 % formamide, 40 mm-Pipes Ph 6·4, 0·4 m-NaCl and 1·0 mm-EDTA overnight at 37 °C. Unhybridized RNA was removed by digestion with 0·5 μg ml−1 RNase A (Sigma, type IIIA) and 0·5 μg ml−1 RNase T1 (Calbiochem) for 30 min at 30 °C. The digestion was terminated by adding proteinase K (0·2 mg ml−1) and SDS (1 %) and incubating for a further 30 min at 37°C. After phenol/chloroform extraction, the protected fragments were precipitated with ethanol and one third of each sample was analysed on a denaturing 6 % polyacrylamide gel.

Tissue preparation for in situ hybridization

Fertilized eggs were kept in a humidified incubator for appropriate times, after which the embryos were removed for tissue fixation. E2–4 embryos (stage 10–24, Hamburger & Hamilton, 1951) were fixed intact in ice-cold 5 % glacial acetic acid, 4 % formaldehyde and 85 % ethanol (AFE) for 30min. For E16 liver, heart and pectoral muscle, the embryos were decapitated and perfused through the ascending aorta, after which the tissue was removed and immersed in AFE for a further 2–3 h. E16 brain and chorioallantoic membrane (which was taken from the membrane in the air pocket) were fixed by immersion for 3h. All tissues were then washed, dehydrated, embedded in wax and sectioned at 7 μm onto poly-L-lysine-coated glass slides as previously described (ffrench-Constant & Hynes, 1988).

Probe synthesis for in situ hybridization

The construction of the templates used to synthesize the single-stranded RNA probes complementary to an invariant region of FN mRNA or to each of the spliced segments has been previously described (ffrench-Constant & Hynes, 1988). Probes, labelled with 3SS-UTP to specific activity of 108 cts min−1μg−1, were synthesized using SP6 or T7 polymerase and then purified on denaturing polyacrylamide gels. The FN-C, FN-EIIIA, FN-EIIIB and FN-V probes were 160nt, 160nt, 230nt and 144nt in length, respectively, and all were used without further reduction in length by alkaline hydrolysis, as this was found to improve the signal/noise ratio.

In situ hybridization

Prehybridization and hybridization were performed as previously described (ffrench-Constant & Hynes, 1988). Slides were dewaxed and passed sequentially through 0·2 M-HCl, 1 μg ml−1 proteinase K, 0·2 % glycine, 4 % paraformaldehyde and 1/200 (v/v) acetic anhydride in 0·1 M-triethanolamine pH 8·0 before addition of the hybridization mixture. Hybridization buffer (50 % deionized formamide, 10 % dextran sulphate, 0·3m-NaCl, 10 mm-Tris pH 7·6, 5 mm-EDTA, 0·02 % (wt/v) Ficoll 400, 0·02 % (wt/v) polyvinylpyrolidone, 0·02% (wt/v) bovine serum albumin, 10 mm-DTT, 100 μg ml−1 yeast tRNA and 500 μM non-radiolabelled thio-UTP) containing the appropriate probe at a concentration of 0·3 μg ml−1 per kilobase probe complexity was adjusted to a pH of 6·0 and placed on the sections. After hybridization overnight at 50 °C, slides were rinsed in 50 % formamide, 2 × SSC, 10mm-DTT at 50 °C, digested with 10 μg ml−1 RNase A (Sigma type IIIA) for 30 min at 37 °C and then washed in 50 % formamide, 2 × SSC, 10 mm-DTT at 65 °C for one hour. Slides were then dried, dipped in Kodak NTB-2 emulsion (diluted 1:1 in water) and exposed for 7 days at −20°C after which time they were developed in Kodak D19, fixed, stained with 0·02 % toluidine blue and mounted with DPX mountant. Sections were viewed on a Zeiss universal or axiophot microscope equipped with bright-field and dark-field optics, and photographed on Tech Pan film at 85ASA.

We have used two techniques, ribonuclease protection and in situ hybridization, to determine the splicing pattern of FN mRNA in chicken embryos, as these two methods provide complementary information. Ribonuclease protection allows accurate quantification of the splicing pattern of each segment in FN mRNA, but provides no information as to the localization of the differently spliced forms. In situ hybridization, in contrast, is only semiquantitative but allows spatial localization of the different forms. Together, these techniques determine the pattern of splicing of each segment within and between different tissues, although they do not demonstrate directly whether included segments are present on the same or different FN mRNA molecules.

FN splicing during early embryogenesis

We began this study by using in situ hybridization to examine the pattern of FN mRNA splicing in E2–4 (stage 10–24) chickens, at which time ribonuclease protection experiments have shown that FN mRNA largely includes EIIIA, EIIIB and the single spliced portion of the chicken V region (A+B+V+) (Norton & Hynes, 1987). [35S]-labelled single-stranded RNA probes complementary either to a region of the FN transcript always included in the final mRNA (FN-C) or to the three spliced regions (FN-EIIIA, FN-EIIIB and FN-V) were synthesized and shown to be specific for the appropriate spliced regions by their pattern of hybridization to E17 liver. In these control experiments, liver parenchyma (synthesizing pFN lacking EIIIA and EIIIB but including the V region in about 50 % of the mRNA) was labelled by FN-C, less well labelled by FN-V and unlabelled by FN-EIIIA and FN-EIIIB (ffrench-Constant & Hynes, 1988, also shown in Fig. 4 of this study).

When the FN-C probe was hybridized to sections of E2–4 chick, localized patterns of FN mRNA expression were observed. For example, in E4 embryos such as that shown in Fig. 1, intense labelling was seen over ectoderm (especially in the more caudal regions of the embryo), endocardium, extraembryonic membranes and the mesonephros. Labelling was also seen between the somites, and more weakly in the somites themselves, in developing limb buds and also over mesodermal cells within the loose matrix surrounding developing organs. By contrast, labelling was absent over the notochord and most of the neural tube although, interestingly, labelling of the ventral part of the neural tube adjacent to the notochord was observed at the caudal end of the embryo (Fig. 2). The presence of FN mRNA in the ventral neural tube has potential relevance for neuronal development in this region (cf. Dodd & Jessell, 1988).

Fig. 1.

A low magnification view of a longitudinal section through an entire E4 (stage 24) chicken embryo hybridized with the FN-C probe as described in Materials and methods, and viewed with bright-held (A) or dark-held (B) optics. The dark-held view shows the intense labelling seen with this probe over the ectoderm (EC), endocardium of the different chambers of the heart (El), extraembryonic membranes (EM) and mesonephros (M). Weaker labelling is seen in the connective tissue throughout the embryo, particularly in between those somites that are seen in this section (S-indicated by arrows). Note that the cranial regions of the neural tube, represented in this section by the developing diencephalon (NT), are unlabelled. Scale bar = 1 mm.

Fig. 1.

A low magnification view of a longitudinal section through an entire E4 (stage 24) chicken embryo hybridized with the FN-C probe as described in Materials and methods, and viewed with bright-held (A) or dark-held (B) optics. The dark-held view shows the intense labelling seen with this probe over the ectoderm (EC), endocardium of the different chambers of the heart (El), extraembryonic membranes (EM) and mesonephros (M). Weaker labelling is seen in the connective tissue throughout the embryo, particularly in between those somites that are seen in this section (S-indicated by arrows). Note that the cranial regions of the neural tube, represented in this section by the developing diencephalon (NT), are unlabelled. Scale bar = 1 mm.

Fig. 2.

A slightly oblique transverse section through the caudal regions of an E4 embryo similar to that shown in Fig. 1, hybridized with the FN-C probe and viewed with bright-held (A) and dark-held (B) optics. Labelling is seen over the ventral part of the neural tube (NT-outlined by arrowheads) in this more caudal region of the embryo. Note that the notochord is unlabelled (NC) while the ectoderm (EC) overlying the neural tube is intensely labelled. Scale bar= 100 μm.

Fig. 2.

A slightly oblique transverse section through the caudal regions of an E4 embryo similar to that shown in Fig. 1, hybridized with the FN-C probe and viewed with bright-held (A) and dark-held (B) optics. Labelling is seen over the ventral part of the neural tube (NT-outlined by arrowheads) in this more caudal region of the embryo. Note that the notochord is unlabelled (NC) while the ectoderm (EC) overlying the neural tube is intensely labelled. Scale bar= 100 μm.

In order to determine the pattern of alternative splicing in these different regions, adjacent sections were hybridized with the FN-EIIIA, FN-EIIIB and FN-V probes, which are of similar length to the FN-C probe (see Materials and methods). In these experiments, all those regions labelled by FN-C were also labelled by FN-EIIIA, FN-EIIIB or FN-V (not shown). These results are in keeping with ribonuclease protection experiments showing that A+B+V+ FN mRNAs predominate at early developmental stages (Norton & Hynes, 1987), and show this splicing pattern to be present throughout all regions of the embryo.

FN splicing during later embryogenesis

We next examined the pattern of splicing in the E16 embryo to ask whether splicing of FN mRNA is altered after embryogenesis and organogenesis are largely completed. Previous work has established that chicken hepatocytes producing pFN have switched their splicing pattern to ABV+/− by this stage (Norton & Hynes, 1987), although the splicing pattern of the FN produced elsewhere has not been determined. Initially, therefore, we prepared RNA from five different tissues (heart, pectoral muscle, brain, liver and chorioallantoic membrane (CAM)) and examined the composition of the FN mRNA in these tissues using ribonuclease protection experiments. These results are illustrated in Fig. 3. FN mRNA was present in all tissues, although the levels in the brain were low. Apart from the expected V form in pFN mRNA from the liver, all FN mRNA appeared to be entirely V+. EIIIA and EIIIB, both virtually absent from liver as found previously, were present in the FN mRNA from the other organs. However, the splicing of these regions was very different from that observed during early embryogenesis; while EIIIA+ forms predominated, EIIIB+ forms were now the minority and, furthermore, appeared to be largely absent from the E16 CAM.

Fig. 3.

Ribonuclease protection experiments examining the inclusion of EII1A, EIHB and V in FN mRNA prepared from E16 chorioallantoic membrane (C), heart (H), pectoral muscle (M), brain (B) and liver (L). This RNA was hybridized with appropriate 32P-labelled probes, digested with RNase and analyzed on a 5 % denaturing acrylamide gel as described in Materials and methods. Three gels are shown, each labelled with the segment-specific probe used for hybridization in that experiment. The predicted sizes of the protected fragments that correspond to the inclusion or exclusion of each spliced segment in the RNA are indicated in the appropriate panel with a + or −, respectively. Each experiment is accompanied by a control experiment showing the absence of any protected fragments after hybridization of the probes with yeast tRNA (Y), and by a lane on the gel indicating the size of the probe prior to RNase digestion (P). The different lanes within each experiment were exposed for the same length of time except for brain and liver, for which longer exposures were required. This presumably reflects a lower level of FN mRNA in these tissues.

Fig. 3.

Ribonuclease protection experiments examining the inclusion of EII1A, EIHB and V in FN mRNA prepared from E16 chorioallantoic membrane (C), heart (H), pectoral muscle (M), brain (B) and liver (L). This RNA was hybridized with appropriate 32P-labelled probes, digested with RNase and analyzed on a 5 % denaturing acrylamide gel as described in Materials and methods. Three gels are shown, each labelled with the segment-specific probe used for hybridization in that experiment. The predicted sizes of the protected fragments that correspond to the inclusion or exclusion of each spliced segment in the RNA are indicated in the appropriate panel with a + or −, respectively. Each experiment is accompanied by a control experiment showing the absence of any protected fragments after hybridization of the probes with yeast tRNA (Y), and by a lane on the gel indicating the size of the probe prior to RNase digestion (P). The different lanes within each experiment were exposed for the same length of time except for brain and liver, for which longer exposures were required. This presumably reflects a lower level of FN mRNA in these tissues.

In order to determine the pattern of splicing within these tissues in more detail, we performed in situ hybridization studies using the four different probes on sections prepared from each region. In contrast with the results on the E2–4 chick, these experiments revealed spatial heterogeneity of FN mRNA splicing within some of the different regions, showing the splicing pattern to be more complex than suggested by the ribonuclease protection experiments.

Three different combinations of FN mRNA splicing were seen in the E16 liver. In the large vessels within the liver, the FN mRNA in the vessel wall was A+BV+, while in immediately adjacent smaller vessels with thicker walls (presumably arterioles and/or bile ducts) it was A+B+V+ (Fig. 4). As expected, hepatocytes synthesizing pFN contained ABV+ FN mRNA (Fig. 4).

A similar degree of complexity was present in the sternum and pectoral muscle, although different splicing combinations were observed. The perichondrial cells surrounding the developing sternal cartilage were strongly labelled by all probes, showing that they contained a high level of A+B+V+ FN mRNA. The chondrocytes within the cartilage contained a lower level of FN mRNA that was AB+V+ while the mRNA in labelled cells within the muscle was strongly A+V+ but only very weakly B+ (Fig. 5). Splicing in the outflow tract of the E16 heart, in contrast, was more simple; cells within the walls of the large vessels of the outflow tract contained A+B+V+ FN mRNA, while the mRNA-in the fibroblast-like cells surrounding the outflow tract was A+BV+ (not shown).

Fig. 4.

Four closely adjacent sections of E16 liver hybridized with FN-C or the segment-specific probes FN-EIIIA, FN-EIIIB or FN-V as indicated and then viewed with dark-field optics (B–E). A representative section viewed with bright-field optics is also shown (A). Two large vessels can be seen within the parenchyma of the liver, and smaller vessels, which are presumably arterioles or bile ducts (arrows), can be seen surrounding the upper large vessel. Labelling by FN-C is seen over parenchyma and all the vessel walls. The segment-specific probes show the FN mRNA within the large vessel walls to be EIIIA+EIIIBV+ (clearly seen in the lower large vessel) while the walls of the smaller vessels are EIIIA+EIIIB+V+ (arrows). Note that the parenchyma of the liver is EIIIA EIIIB V+. Scale bar = 200 μm.

Fig. 4.

Four closely adjacent sections of E16 liver hybridized with FN-C or the segment-specific probes FN-EIIIA, FN-EIIIB or FN-V as indicated and then viewed with dark-field optics (B–E). A representative section viewed with bright-field optics is also shown (A). Two large vessels can be seen within the parenchyma of the liver, and smaller vessels, which are presumably arterioles or bile ducts (arrows), can be seen surrounding the upper large vessel. Labelling by FN-C is seen over parenchyma and all the vessel walls. The segment-specific probes show the FN mRNA within the large vessel walls to be EIIIA+EIIIBV+ (clearly seen in the lower large vessel) while the walls of the smaller vessels are EIIIA+EIIIB+V+ (arrows). Note that the parenchyma of the liver is EIIIA EIIIB V+. Scale bar = 200 μm.

Fig. 5.

Four closely adjacent sections of E16 pectoral muscle and sternum hybridized and viewed as in Fig. 4. The developing sternum, consisting of chondrocytes (C) surrounded by perichondrial cells (P), and attached pectoral muscle (M) are shown in this section. Intense labelling by FN-C is seen in the perichondrial tissue and in scattered cells within the muscle, while the chondrocytes are more weakly labelled. The segment-specific probes show that the FN mRNA in the perichondrial cells is EIIIA+EIIIB+V+ in contrast with the chondrocytes in which it is EIllA EIIIB+V+. Scattered cells within the muscle are EIIIA+V+ but largely EIIIB. Note that the few grains over the chondrocytes after hybridization with FN-EIIIA correspond to the background level seen in the autoradiographic emulsion (C). Scale bar = 200 μm.

Fig. 5.

Four closely adjacent sections of E16 pectoral muscle and sternum hybridized and viewed as in Fig. 4. The developing sternum, consisting of chondrocytes (C) surrounded by perichondrial cells (P), and attached pectoral muscle (M) are shown in this section. Intense labelling by FN-C is seen in the perichondrial tissue and in scattered cells within the muscle, while the chondrocytes are more weakly labelled. The segment-specific probes show that the FN mRNA in the perichondrial cells is EIIIA+EIIIB+V+ in contrast with the chondrocytes in which it is EIllA EIIIB+V+. Scattered cells within the muscle are EIIIA+V+ but largely EIIIB. Note that the few grains over the chondrocytes after hybridization with FN-EIIIA correspond to the background level seen in the autoradiographic emulsion (C). Scale bar = 200 μm.

The results presented in Figs 4 and 5 therefore contain examples of cell types expressing all possible combinations of splicing of EIIIA and EIIIB at one developmental stage. In contrast, the remaining two tissues showed no clear heterogeneity of FN mRNA, although the splicing pattern was different in each. In the E16 CAM, FN mRNA within the tissue all appeared to be A+B V+ (as shown for E14 CAM in Fig. 9). In the E16 brain, scattered labelled cells were seen although, as expected in view of the low signal in the ribonuclease protection experiments, the majority of the tissue was unlabelled (Fig. 6). Similar results were obtained with all four probes (data not shown). However, the high cell density and lack of structural features in this tissue made it impossible to match labelled cells in adjacent sections hybridized with different probes. The identity of the labelled cells in the developing brain is not clear although rat astrocytes have been shown to synthesize FN in vitro (Price & Hynes, 1985) and this FN includes all three spliced segments (Schwarzbauer et al. 1987a) consistent with the pattern of in situ hybridization. The presence of FN mRNA in developing brain and in its precursor, the neural tube (Fig. 2), is of interest given the reports of transient expression of FN (Stewart & Pearlman, 1987; Chun & Shatz, 1988) which could play a role in neuronal development (cf, Dodd & Jessell, 1988).

Fig. 6.

A section of E16 cerebellum hybridized with the FN-C probe and viewed with bright-field (A) and dark-field (B) optics. Labelling of scattered cells within the cerebellum is seen. Scale bar= 100 μm.

Fig. 6.

A section of E16 cerebellum hybridized with the FN-C probe and viewed with bright-field (A) and dark-field (B) optics. Labelling of scattered cells within the cerebellum is seen. Scale bar= 100 μm.

FN splicing during development of the chorioallantoic membrane

These results show clearly that FN mRNA splicing is regulated at later developmental stages so as to produce different forms of FN mRNA in different locations. However, the complexity of many of these tissues makes it difficult to elucidate the possible function(s) of the splicing changes. We therefore chose the simplest of these tissues, the CAM, and examined the developmental sequence of splicing in more detail. The CAM, formed by the fusion of the chorion and the allantois at E4, consists of two cell layers (ectoderm and endoderm) separated by a matrix-filled space containing mesodermal cells and a vascular network that functions as the respiratory organ of the chick (Romanoff, 1960). Growth of the CAM continues until E10, with the membrane growing around the inside of the eggshell (Romanoff, 1960). During this period, the expansion of the vascular network is accompanied by capillary endothelial cell proliferation, and this proliferation diminishes when the growth of the membrane is complete (Ausprunk et al. 1974).

We examined the pattern of FN mRNA splicing during these different phases of growth in the CAM by performing ribonuclease protection experiments on RNA prepared from the CAM at E8,10,12,14 and 18. The results are shown in Fig. 7. As expected from our previous results, the V region was apparently included in all the mRNA at all the different ages. EIIIA was also included in the majority of FN mRNA, with no evidence for any change in the degree of inclusion at the different ages. The splicing of EIIIB, in contrast, changed considerably over the developmental stages examined. At E8, while the CAM is still growing, EIIIB+ and EIIIB forms of FN mRNA appear approximately equimolar. Between E10 and E18, after growth is completed, a progressive reduction was seen in EIIIB+ forms to a level of less than 10 % of total FN mRNA (as measured by densitometrie scanning of the autoradiograph).

Fig. 7.

Ribonuclease protection experiments similar to those shown in Fig. 3 examining the inclusion of EIIIA, EIIIB and V in RNA prepared from the chorioallantoic membrane (CAM) at E8,10,12,14 and 18. Each lane is labelled with the age of the CAM examined. Predicted fragment sizes and extra lanes showing the control experiment using yeast tRNA (Y) as well as the size of the probe used (P) are labelled as in Fig. 3. The V experiment has been overexposed to ensure that any V RNA would be detected.

Fig. 7.

Ribonuclease protection experiments similar to those shown in Fig. 3 examining the inclusion of EIIIA, EIIIB and V in RNA prepared from the chorioallantoic membrane (CAM) at E8,10,12,14 and 18. Each lane is labelled with the age of the CAM examined. Predicted fragment sizes and extra lanes showing the control experiment using yeast tRNA (Y) as well as the size of the probe used (P) are labelled as in Fig. 3. The V experiment has been overexposed to ensure that any V RNA would be detected.

In order to determine whether EIIIB splicing was identical in the different cell types within the CAM, we performed in situ hybridization experiments on sections of CAM at the different ages. These experiments showed that the loss of EIIIB in different cell types occurred at the same stage, and provided no evidence for heterogeneity of splicing changes in the CAM. At E6, FN-EIIIB labelled the mesodermal cells and smooth muscle and/or endothelial cells in the walls of large blood vessels within the CAM as intensely as did the FN-C and FN-EIIIA probes (Fig. 8). The same pattern of labelling was also observed at E9 (not shown). However, when sections of E14 CAM were exposed for autoradiography for the same length of time as these E6 and E9 sections, labelling of all these structures with FN-EIIIB was virtually absent while FN-C and FN-EIIIA labelling were still seen (Fig. 9). With longer exposure times, weak FN-EIIIB labelling could be seen at E14, but was absent by E18 while labelling with FN-C and FN-EIIIA remained (not shown), in keeping with the ribonuclease protection results. The pattern of labelling with FN-V was identical to that seen with FN-C at all the different ages (not shown).

Fig. 8.

Three closely adjacent sections of E6 chorioallantoic membrane hybridized with the FN-C probe or the segment-specific EIIIA or EIIIB probes as indicated and viewed with dark-field optics (B–D). A representative section viewed with bright-field optics is also shown (A). Intense labelling is seen by FN-C both of mesodermal cells under the ectoderm (EC) and within the matrix, and of the walls of larger blood vessels (BV). The endoderm, in contrast, is unlabelled. The segment-specific probes show the FN mRNA in both regions to be EIIIA+, EIIIB+ (panels C and D) and V+ (not shown). Scale bar = 100 μm.

Fig. 8.

Three closely adjacent sections of E6 chorioallantoic membrane hybridized with the FN-C probe or the segment-specific EIIIA or EIIIB probes as indicated and viewed with dark-field optics (B–D). A representative section viewed with bright-field optics is also shown (A). Intense labelling is seen by FN-C both of mesodermal cells under the ectoderm (EC) and within the matrix, and of the walls of larger blood vessels (BV). The endoderm, in contrast, is unlabelled. The segment-specific probes show the FN mRNA in both regions to be EIIIA+, EIIIB+ (panels C and D) and V+ (not shown). Scale bar = 100 μm.

Fig. 9.

Three adjacent sections of E14 chorioallantoic membrane hybridized and viewed as in Fig. 8. Labelling by FN-C is still present over the mesodermal cells and the walls of the large blood vessel (BV). However, in contrast to the E6 membrane, the segment-specific probes show this FN mRNA to be EIIIA+, EIIIB. Note that the ectoderm (EC) at this age is covered by the shell membrane (SM), which is visible in the dark-field views of some of these sections; this does not represent hybridization. Scale bar= 100 μm.

Fig. 9.

Three adjacent sections of E14 chorioallantoic membrane hybridized and viewed as in Fig. 8. Labelling by FN-C is still present over the mesodermal cells and the walls of the large blood vessel (BV). However, in contrast to the E6 membrane, the segment-specific probes show this FN mRNA to be EIIIA+, EIIIB. Note that the ectoderm (EC) at this age is covered by the shell membrane (SM), which is visible in the dark-field views of some of these sections; this does not represent hybridization. Scale bar= 100 μm.

While these in situ hybridization experiments did not reveal any differences in splicing between the cell types of the CAM, they did show regulation at the level of overall FN mRNA levels during capillary formation. In addition to the mesodermal cells and large vessels examined above, the CAM also contains a network of capillaries adjacent to the ectoderm that cannot be seen at the magnification used in Figs 8 and 9. Examination of these capillaries at higher magnification showed that at E6, in the initial stages of the formation of the capillary network, all cells within this region of the CAM were labelled by FN-C (Fig. 10A,B). By E9, however, the capillary endothelial cells forming vessels with distinct lumens were often unlabelled (Fig. 10C,D) while other cells in the section were still labelled. This suggests that capillary endothelial cells downregulate FN mRNA levels after the formation of a lumen. In appropriate en face sections, these unlabelled capillaries could be seen to be continuous with labelled large vessel walls (such as that shown in Fig. 9) containing both endothelial and smooth muscle cells and expressing FN mRNA.

Fig. 10.

Two sections of E6 (A,B) or E9 (C,D) chorioallantoic membrane viewed at a higher magnification than in Figs 8 and 9 to show the formation of the capillary network under the ectoderm (EC). Both sections have been hybridized with the FN-C probe and viewed with bright-field (A,C) or dark-field (B,D) optics. Capillary endothelial cells, identified as the cells forming the walls of vessels containing blood cells (hollow arrows above vessels) are all labelled to differing degrees at E6. By E9, however, unlabelled endothelial cells can be seen forming a small vessel (hollow arrow) while adjacent mesodermal cells remain intensely labelled. Scale bar = 25 μm.

Fig. 10.

Two sections of E6 (A,B) or E9 (C,D) chorioallantoic membrane viewed at a higher magnification than in Figs 8 and 9 to show the formation of the capillary network under the ectoderm (EC). Both sections have been hybridized with the FN-C probe and viewed with bright-field (A,C) or dark-field (B,D) optics. Capillary endothelial cells, identified as the cells forming the walls of vessels containing blood cells (hollow arrows above vessels) are all labelled to differing degrees at E6. By E9, however, unlabelled endothelial cells can be seen forming a small vessel (hollow arrow) while adjacent mesodermal cells remain intensely labelled. Scale bar = 25 μm.

These results show the EIIIB segment to be excluded from FN mRNA in mesodermal and vessel wall cells once growth of the CAM is complete, and also demonstrate downregulation of total FN mRNA levels in endothelial cells after capillary formation.

Fibronectin splicing is developmentally regulated

Three major conclusions concerning the splicing of the FN primary gene transcript in the intact chicken embryo may be drawn from this study. First, inclusion of EIIIA and EIIIB is developmentally regulated. At E2-4, the great majority of the FN mRNA in all regions of the embryo contains the three spliced segments EIIIA, EIIIB and V. Later in development, at E16, ribonuclease protection and in situ hybridization experiments show EIIIB and/or EIIIA to be excluded from the FN mRNA in a number of tissues. This is particularly well illustrated by the CAM, in which EIIIB is largely excluded from FN mRNA as embryonic development is completed. Second, this developmental regulation occurs in a cell-type-specific manner, as shown by the in situ hybridization analysis at E16 showing inclusion of EIIIA and EIIIB to differ among the cell types within a single tissue. Third, the EIIIA and EIIIB exons are included or excluded independently during development to produce all possible combinations of these two exons in spatially distinct patterns within the embryo. When taken in conjunction with the finding that about 50 % of liver FN mRNA coding for pFN contains the V region but not EIIIA or EIIIB, this allows the further conclusion that the inclusion of all three spliced regions in FN mRNA is independently variable in vivo.

Alternative splicing of FN has been studied previously in cultured cells or isolated liver at both the RNA and the protein level. Some of the features we observe in the intact embryo have also been observed in these studies. Thus, the V region is variably included by hepatocytes (Tamkun & Hynes, 1983; Schwarzbauer et al. 1983, 1985; Paul & Hynes, 1984; Paul et al. 1986; Norton & Hynes, 1987) but almost completely included by other cell types. The EIIIA and, to a lesser extent, the EIIIB segments are partially included by most cultured cell types except primary hepatocytes (Kornblihtt et al. 1984a,b; Paul et al. 1986; Schwarzbauer et al. 1987a; Norton & Hynes, 1987; Zardi et al. 1987; Gutman & Kornblihtt, 1987) and all possible combinations of EIIIA and EIIIB can be found in cultured cells (Schwarzbauer et al. 1987a). Changes in the pattern of spliced forms of FN mRNA within a single cell type have also been reported in these studies of cultured cells. Thus, the levels of inclusion of the V and EIIIA regions both rise as hepatocytes are cultured (Tamkun & Hynes, 1983; Paul & Hynes, 1984; Paul, Schwarzbauer & Hynes, unpublished data) while Schwarzbauer et al. (1987a) showed a decreasing level of EIIIB inclusion with repeated passaging of freshly dissociated cells. In addition, increases in the inclusion of all three segments have been reported on oncogenic transformation in some systems (Castellani et al. 1986; Borsi et al. 1986; Schwarzbauer et al. 1987a; Zardi et al. 1987) but not others (Norton & Hynes (1987)).

Developmental regulation of the splicing pattern in FN mRNA has not previously been demonstrated. Sekiguchi et al. (1986b) analysed FN extracted from embryonic and adult lungs and found that a FN proteolytic fragment containing the V and EIIIA regions was shorter in the adult tissue, suggesting that one or both of these regions are excluded from FN synthesized in the adult lung. However, this shorter fragment corresponds with one derived from pFN. The difficulty of excluding the alternative explanation that the deposition of pFN is responsible for the observed differences in FN highlights the advantage of using in situ hybridization to determine patterns of FN mRNA splicing within different tissues.

Regulation of differently spliced FN mRNAs provides qualitative control of FN gene expression. This mechanism operates alongside quantitative changes in FN mRNA levels during development, as illustrated in the CAM. Endothelial cells in this membrane migrate to form a network of capillaries adjacent to the ectoderm between E4–-5 (Ausprunk, 1982, 1986), and contain abundant FN mRNA at this stage. However, once capillary formation is complete, the endothelial cells appear to downregulate FN mRNA. In addition to this quantitative change in endothelial cells, adjacent cell types show a qualitative change, as the FN mRNA in these adjacent cells remains at a high level while the pattern of splicing changes from A+B+V+ to A+BV+. Our results suggest that FN production in endothelial cells may only be required for initial migration and capillary formation.

The mechanisms controlling the expression of differently spliced forms of FN mRNA remain undefined. Regulation could occur at the level of transcription, splicing or differential stability of the various FN mRNAs. The increasing evidence for the importance of growth factors in tissue differentiation makes them attractive candidates for a role in control of these processes. In support of this hypothesis, Ignotz & Massague (1986) have reported that TGF-β regulates expression of extracellular matrix including fibronectin and Balza et al. (1988) have reported that TGF-β increased the inclusion of EIIIA in the FN synthesized by cultured human fibroblasts. TGF-β1 is expressed in distinct patterns in the developing embryo (Heine et al. 1987; Lehnert & Akhurst, 1988). In particular, Lehnert & Akhurst (1988) found that TGF-β1 mRNA was not expressed in chondrocytes, but was expressed at a high level in surrounding perichondrial cells; a distribution similar to EIIIA expression in the developing sternum (Fig. 5). Further studies examining the in vivo correlation between the expression of growth factors such as TGF-β and the patterns of FN splicing, together with experiments determining the modulation of FN splicing by these growth factors in vitro, are required to test this speculation.

Splicing provides a mechanism to produce functionally different forms of FN

The finding that inclusion of EIIIA, EIIIB and V is independent and developmentally regulated in the intact embryo provides support for the hypothesis that splicing is used during development to create functionally different forms of FN within the extracellular matrix by including or excluding segments of FN that could in turn modify the behaviour of interacting cells. Clearly this hypothesis assumes that the differently spliced variants of FN mRNA are expressed at the level of the protein. Two lines of evidence suggest that this assumption is valid. First, there is a good correlation between the distribution of the spliced forms of FN mRNA in cultured cells and the pattern of protein expression by these cells (Tamkun & Hynes, 1983; Price & Hynes, 1985; Paul et al. 1986; Schwarzbauer et al. 1987a). Second, our results showing widespread embryonic expression of EIIIA in FN mRNA are consistent with the widespread expression of EIIIA+ FN in embryonic tissue (Vartio et al. 1987).

What cellular functions could be affected by the different forms of FN expressed during development? The only alternatively spliced segment to which defined functions have yet been assigned is the V (or IIICS) segment, for which a role in cell adhesion has been suggested. Studies using synthetic peptides corresponding to parts of the human V region as culture substrates show two sites within this region that are adhesive for murine melanoma cells in vitro (Humphries et al. 1987). One of these sites lies within a region that has significant homology with part of the chicken V region (Norton & Hynes, 1987), and a peptide corresponding to this region is an adhesive substrate for chicken peripheral nervous system neurons (Humphries et al.1998). These observations raise the possibility that chicken V region splicing could create forms of FN of differing adhesiveness within the ECM and so guide cell migration in the early embryo. This is a particularly attractive hypothesis to explain guidance of migrating trunk neural crest cells, as both the RGD-containing cell-binding site and the V region appear to be involved in neural crest cell migration on FN in culture (Dufour et al. 1988). However, we have found no evidence for V region heterogeneity in the E2-4 chick, during which time this early stage of neural crest cell migration occurs (this study; ffrench-Constant & Hynes, 1988). Indeed, apart from the exclusion of V from about 50 % of the mRNA encoding pFN in hepatocytes we found no evidence for differential expression of V at any age examined. Clearly, though, we cannot exclude the possibility that changes in the V region undetectable by in situ hybridization create FNs with differences sufficient to guide migrating cells.

The major changes in FN splicing which we have observed concern EIIIA and EIIIB. These two segments have been considered to be characteristic of so-called ‘cellular FN’. However, this is clearly an oversimplified view since all possible combinations of these two alternatively spliced segments are observed at different times and places during development. It seems much better to define the forms of FN by their content of the alternatively spliced segments. Based on the currently available data, all forms of FN except pFN are V+ but vary in their content of EIIIA and EIIIB. Neither the presence nor the absence of either of these segments defines FN as ‘cellular FN’ or ‘plasma FN’, and each segment can be selectively spliced out during different developmental processes. The challenge is to determine the functional distinctions among AB, A+B, AB+ and A+B+ forms of FN. The descriptive study presented here does not define clearly discrete functions for these different forms but does offer some working hypotheses for future analysis.

One of the developmentally regulated changes in FN mRNA splicing observed in this study appears to be correlated with proliferation and/or migration. By E16, EIIIB is largely excluded from the FN mRNA of the CAM, which completes its growth by E10 (Romanoff, 1960). In contrast, EIIIB remains present in much higher proportions in the FN mRNAs in the E16 heart, skeletal muscle and CNS which have not yet completed their growth at this stage. We have also reported previously that EIIIB is included in the FN mRNA expressed in cells associated with several different early embryonic cell migrations (ffrench-Constant & Hynes, 1988). The correlation of EIIIB inclusion with early development leads us to suggest that the EIIIB segment has an essential role in the migration and/or proliferation of cells prior to their terminal differentiation. For example, inclusion of this segment may create a form of FN appropriate for the transient cell–FN interactions that will occur during cell migration, while exclusion might favour more stable cell–substrate adhesion suitable for differentiated tissue. In keeping with this, we have found that EIIIB is re-expressed in the FN mRNA present in healing wounds (ffrench-Constant et al. 1989).

Turning to EIIIA, it is more difficult to point to any obvious correlation of inclusion or exclusion of this segment. However, two cell types which appear to synthesize EIIIA FN are hepatocytes (Fig. 4) and chondrocytes (Fig. 5). Neither of these cell types assembles extensive FN-rich matrices whereas fibro-blasts, which synthesize predominantly EIIIA+ FN, readily assemble FN matrices on and around their surfaces. These observations suggest that one function for EIIIA could be in matrix assembly, either by serving as a binding site for FN or other matrix components, or by interacting with a cell-surface ‘receptor’ involved in matrix assembly. Another possible function, as a negative regulator of cellular differentiation, is suggested by observations that ‘cellular’ FN extracted from fibro-blasts can inhibit chondrogenesis and myogenesis in vitro (West et al. 1979; Pennypacker et al. 1979; Podleski et al. 1979) while pFN is reported to be a less effective inhibitor (West et al. 1984). TGF-β acts as an inhibitor of chondrogenesis (Rosen et al. 1988) and myogenesis (Massague et al. 1986) and is known to elevate synthesis of FN (Ignotz & Massague, 1986) and perhaps inclusion of EIIIA (Balza et al. 1988).

The possible roles of EIIIA and EIIIB suggested by their differential expression during development need to be tested directly in model systems. This has been difficult to achieve since the most readily available form of FN, plasma FN, lacks both these segments and because most other sources comprise an ill-defined mixture of the various FN forms. The use of retroviral vectors to express FN forms differing in their content of the alternatively spliced segments (Schwarzbauer et al. 1987b) offers a way to generate the necessary reagents, and experiments are now in progress to test the above hypotheses.

We are grateful to Pam Norton for her advice on the ribonuclease protection experiments, and to Dianna Ausprunk, Judah Folkman and Don Ingber for their help with the interpretation of the results on the chorioallantoic membrane. We also thank Jun-Lin Guan, Pam Norton, Gene Marcantonio, Doug DeSimone, Betsy George and Lisa Urry for their comments on the text and Colleen Mazzeo for her help in preparation of the manuscript. Charles ffrench-Constant is a Lucille P. Markey visiting fellow and Richard O. Hynes is an Investigator of the Howard Hughes Medical Institute. This work was supported in part by a grant from the Lucille P. Markey Charitable Trust and by a grant from the USPHS, National Cancer Institute (R01-CA17007).

Ausprunk
,
D. H.
(
1982
).
Synthesis of glycoproteins by endothelial cells in embryonic blood vessels
.
Devl Biol
.
90
,
79
90
.
Ausprunk
,
D. H.
(
1986
).
Distribution of hyaluronic acid and sulfated glycosaminoglycans during blood-vessel development in the chick chorioallantoic membrane
.
Am. J. Anat
.
177
,
313
331
.
Ausprunk
,
D. H.
,
Knighton
,
D. R.
&
Folkman
,
J.
(
1974
).
Differentiation of vascular endothelium in the chick chorioallantois: A structural and autoradiographic study
.
Devl Biol
.
38
,
237
248
.
Balza
,
E.
,
Borsi
,
L.
,
Allemanni
,
G.
&
Zardi
,
L.
(
1988
).
Transforming growth factor beta regulates the levels of different fibronectin isoforms in normal human cultured fibroblasts
.
FEBS Lett
.
228
,
42
44
.
Borsi
,
L.
,
Carnemolla
,
B.
,
Castellani
,
P.
,
Rosellini
,
C.
,
Vecchio
,
D.
,
Allemanni
,
G.
,
Chang
,
S. E.
,
Taylor-Papadimitriou
,
J.
,
Pande
,
H.
&
Zardi
,
L.
(
1987
).
Monoclonal antibodies in the analysis of fibronectin isoforms generated by alternative splicing of mRNA precursors in normal and transformed human cells
.
J. Cell Biol
.
104
,
595
600
.
Buck
,
C. A.
&
Horwitz
,
A. F.
(
1987
).
Cell surface receptors for extracellular matrix molecules
.
A. Rev. Cell. Biol
.
3
,
179
205
.
Castellani
,
P.
,
Siri
,
A.
,
Rosellini
,
C.
,
Infusini
,
E.
,
Borsi
,
L.
&
Zardi
,
L.
(
1986
).
Transformed human cells release different fibronectin variants than do normal cells
.
J. Cell Biol
.
103
,
1671
1677
.
Chirgwin
,
J. M.
,
Przybyla
,
A. E.
,
MacDonald
,
R. J.
&
Rutter
,
W. J.
(
1979
).
Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease
.
Biochemistry
18
,
5294
5299
.
Chun
,
J. J. M.
&
Shatz
,
C. J.
(
1988
).
A fibronectin-like molecule is present in the developing cat cerebral cortex and is correlated with subplate neurons
.
J. Cell Biol
.
106
,
857
872
.
Dodd
,
J.
&
Jessell
,
T. M.
(
1988
).
Axon guidance and the patterning of neuronal projections in vertebrates
.
Science
242
,
692
699
.
Dufour
,
S.
,
Duband
,
J-L.
,
Humphries
,
M. J.
,
Obara
,
M.
,
Yamada
,
K. M.
&
Thiery
,
J. P.
(
1988
).
Attachment, spreading and locomotion of avian neural crest cells are mediated by multiple adhesion sites on fibronectin molecules
.
EMBO J
.
7
,
2661
2671
.
Ffrench-Constant
,
C.
&
Hynes
,
R. O.
(
1988
).
Patterns of fibronectin gene expression and splicing during cell migration in chicken embryos
.
Development
104
,
369
382
.
Ffrench-Constant
,
C.
,
Van De Water
,
L.
,
Dvorak
,
H. F.
&
Hynes
,
R. O.
(
1989
).
Reappearance of an embryonic pattern of fibronectin splicing during wound healing in the adult rat
.
J. Cell Biol, (in press)
.
Gutman
,
A.
&
Kornblihtt
,
A. R.
(
1987
).
Identification of a third region of cell-specific alternative splicing in human fibronectin mRNA
.
Proc. natn. Acad. Sci. U.S.A
.
84
,
7179
7182
.
Hamburger
,
V.
&
Hamilton
,
H. L.
(
1951
).
A series of normal stages in the development of the chick embryo
.
J. Morph
.
88
,
49
92
.
Heine
,
U. I.
,
Munoz
,
E. F.
,
Flonders
,
K. C.
,
Ellingsworth
,
L. R.
,
Lam
,
H.-Y. P.
,
Thompson
,
N. L.
,
Roberts
,
A. B.
&
Sporn
,
M. B.
(
1987
).
Role of transforming growth factor-β in the development of the mouse embryo
.
J. Cell Biol
.
105
,
2861
2876
.
Humphries
,
M. J.
,
Akiyama
,
S. K.
,
Komoriya
,
A.
,
Olden
,
K.
&
Yamada
,
K. M.
(
1986
).
Identification of an alternatively spliced site in human plasma fibronectin that mediates cell type-specific adhesion
.
J. Cell Biol
.
103
,
2637
2647
.
Humphries
,
M. J.
,
Akiyama
,
S. K.
,
Komoriya
,
A.
,
Olden
,
K.
&
Yamada
,
K. M.
(
1988
).
Neurite extension of chicken peripheral nervous system neurons on fibronectin: relative importance of specific adhesion sites in the central cell-binding domain and the alternatively spliced type III connecting segment
.
J. Cell Biol
.
106
,
1289
1297
.
Humphries
,
M. J.
,
Komoriya
,
A.
,
Akiyama
,
S. K.
,
Olden
,
K.
&
Yamada
,
K. M.
(
1987
).
Identification of two distinct regions of the type III connecting segment of human plasma fibronectin that promote cell type-specific adhesion
.
J. biol. Chem
.
262
,
6886
6892
.
Hynes
,
R. O.
(
1985
).
Molecular biology of fibronectin
.
A. Rev. Cell Biol
.
1
,
67
90
.
Hynes
,
R. O.
(
1986
).
Fibronectins
.
Sci. Am
.
254
,
42
51
.
Hynes
,
R. O.
(
1987
).
Integrins, a family of cell surface receptors
.
Cell
48
,
549
554
.
Hynes
,
R. O.
&
Yamada
,
K. M.
(
1982
).
Fibronectins: multifunctional modular glycoproteins
.
J. Cell Biol
.
95
,
369
377
.
Ignotz
,
R. A.
&
Massague
,
J.
(
1986
).
Transforming growth factor-beta stimulates the expression of fibronectin and collagen and their incorporation into the extracellular matrix
.
J. biol. Chem
.
261
,
4337
4345
.
Kornblihtt
,
A. R.
,
Umezawa
,
K.
,
Vibe-Pedersen
,
K.
&
Baralle
,
F. E.
(
1985
).
Primary structure of human fibronectin: differential splicing may generate at least 10 polypeptides from a single gene
.
EMBO J
.
4
,
1755
1759
.
Kornblihtt
,
A. R.
,
Vibe-Pedersen
,
K.
&
Baralle
,
F. E.
(
1984a
).
Human fibronectin: molecular cloning evidence for two mRNA species differing by an internal segment coding for a structural domain
.
EMBO J
.
3
,
221
226
.
Kornblihtt
,
A. R.
,
Vibe-Pedersen
,
K.
&
Baralle
,
F. E.
(
1984b
).
Human fibronectin: cell specific alternative mRNA splicing generates polypeptide chains differing in the number of internal repeats
.
Nucl. Acids Res
.
12
,
5853
5868
.
Lebaron
,
R. G.
,
Esko
,
J. D.
,
Woods
,
A.
,
Johansson
,
S.
&
Hook
,
M.
(
1988
).
Adhesion of glycosaminoglycan-deficient chinese hamster ovary cell mutants to fibronectin substrata
.
J. Cell Biol
.
106
,
945
952
.
Lehnert
,
S. A.
&
Akhurst
,
R. J.
(
1988
).
Embryonic expression pattern of TGF beta type-1 RNA suggests both paracrine and autocrine mechanisms of action
.
Development
104
,
263
273
.
Massague
,
J.
,
Cheifetz
,
S.
,
Endo
,
T.
&
Nadal-Ginard
,
B.
(
1986
).
Type beta transforming growth factor is an inhibitor of myogenic differentiation
.
Proc. natn. Acad. Sci. U.S.A
.
83
,
8206
8210
.
McCarthy
,
J. B.
,
Chelberg
,
M. K.
,
Mickelson
,
D. J.
&
Furcht
,
L. T.
(
1988
).
Localization and chemical synthesis of fibronectin peptides with melanoma adhesion and heparin binding activities
.
Biochemistry
27
,
1380
1388
.
Newgreen
,
D.
&
Thiery
,
J. P.
(
1980
).
Fibronectin in early avian embryos: synthesis and distribution along the migration pathways of neural crest cells
.
Cell Tissue Res
.
211
,
269
291
.
Norton
,
P. A.
&
Hynes
,
R. O.
(
1987
).
Alternative splicing of chicken fibronectin in embryos and in normal and transformed cells
.
Molec. Cell Biol
.
7
,
4297
4307
.
Obara
,
M.
,
Kang
,
M. S.
&
Yamada
,
K. M.
(
1988
).
Site-directed mutagenesis of the cell-binding domain of human fibronectin: separable, synergistic sites mediate adhesive function
.
Cell
53
,
649
657
.
Paul
,
J.
&
Hynes
,
R. O.
(
1984
).
Multiple fibronectin subunits and their posttranslational modifications
.
J. biol. Chem
.
259
,
13407
13487
.
Paul
,
J. I.
,
Schwarzbauer
,
J. E.
,
Tamkun
,
J. W.
&
Hynes
,
R. O.
(
1986
).
Cell-type-specific fibronectin subunits generated by alternative splicing
.
J. biol. Chem
.
261
,
12258
12265
.
Pennypacker
,
J. P.
,
Hassell
,
J. R.
,
Yamada
,
K. M.
&
Pratt
,
R. M.
(
1979
).
The influence of an adhesive cell surface protein on chondrogenic expression in vitro
.
Expl Cell Res
.
121
,
411
415
.
Podleski
,
T. R.
,
Greenberg
,
I.
,
Schlessinger
,
J.
&
Yamada
,
K. M.
(
1979
).
Fibronectin delays the fusion of L6 myoblasts
.
Expl Cell Res
.
122
,
317
326
.
Price
,
J.
&
Hynes
,
R. O.
(
1985
).
Astrocytes in culture synthesize and secrete a variant form of fibronectin
.
J. Neurosci.
5
,
2205
2211
.
Romanoff
,
A. L.
(
1960
).
The Avian Embryo. Structural and Functional Development
.
Macmillan, NY
, pp.
1041
1140
.
Rosen
,
D. M.
,
Stempien
,
S. A.
,
Thompson
,
A. Y.
&
Seyedin
,
S. M.
(
1988
).
Transforming growth factor-beta modulates the expression of osteoblast and chondroblast phenotypes in vitro
.
J. cell. Physiol
.
134
,
337
346
.
Ruoslahti
,
E.
&
Pierschbacher
,
M. D.
(
1987
).
New perspectives in cell adhesion: RGD and integrins
.
Science
238
,
491
497
.
Saunders
,
S.
&
Bernfield
,
M.
(
1988
).
Cell surface proteoglycan binds mouse mammary epithelial cells to fibronectin and behaves as a receptor for interstitial matrix
.
J. Cell Biol
.
106
,
423
430
.
Schwarzbauer
,
J. E.
,
Mulligan
,
R. C.
&
Hynes
,
R. O.
(
1987b
).
Efficient and stable expression of recombinant fibronectin polypeptides
.
Proc. natn. Acad. Sci. U.S.A
.
84
,
754
758
.
Schwarzbauer
,
J. E.
,
Patel
,
R. S.
,
Fonda
,
D.
&
Hynes
,
R. O.
(
1987a
).
Multiple sites of alternative splicing of the rat fibronectin gene transcript
.
EMBO J
.
6
,
2573
2580
.
Schwarzbauer
,
J. E.
,
Paul
,
J. I.
&
Hynes
,
R. O.
(
1985
).
On the origin of species of fibronectin
.
Proc. natn. Acad. Set. U.S.A
.
82
,
1424
1428
.
Schwarzbauer
,
J. E.
,
Tamkun
,
J. W.
,
Lemischka
,
I. R.
&
Hynes
,
R. O.
(
1983
).
Three different fibronectin mRNAs arise by alternative splicing within the coding region
.
Cell
35
,
421
431
.
Sekiguchi
,
K.
,
Klos
,
A. M.
,
Hirohashi
,
S.
&
Hakomori
,
S.
(
1986b
).
Human tissue fibronectin: expression of different isotypes in the adult and fetal tissues
.
Biochem. biophys. Res. Commun
.
141
,
1012
1017
.
Sekiguchi
,
K.
,
Klos
,
A. M.
,
Kurachi
,
K.
,
Yoshitake
,
S.
&
Hakomori
,
S.
(
1986a
).
Human liver fibronectin complementary DNAs: identification of two different messenger RNAs possibly encoding the α and β subunits of plasma fibronectin
.
Biochemistry
25
,
4936
4941
.
Sekiguchi
,
K.
,
Siri
,
A.
,
Zardi
,
L.
&
Hakomori
,
S. I.
(
1985
).
Differences in domain structure between human fibronectins isolated from plasma and from culture supernatants of normal and transformed fibroblasts
.
J. biol. Chem
.
260
,
5105
5114
.
Stewart
,
G. R.
&
Pearlman
,
A. L.
(
1987
).
Fibronectin-like immunoreactivity in the developing cerebral cortex
.
J. Neurosci
.
7
,
3325
3333
.
Tamkun
,
J. W.
&
Hynes
,
R. O.
(
1983
).
Plasma fibronectin is synthesized and secreted by hepatocytes
.
J. biol. Chem
.
258
,
4641
4647
.
Vartio
,
T.
,
Laitinen
,
L.
,
Narvanen
,
O.
,
Cutolo
,
M.
,
Thornell
,
L. E.
,
Zardi
,
L.
&
Virtanen
,
I.
(
1987
).
Differential expression of the ED sequence-containing form of cellular fibronectin in embryonic and adult human tissues
.
J. Cell Sci
.
88
,
419
430
.
West
,
C. M.
,
De Weerd
,
H.
,
Dowdy
,
K.
&
De La Paz
,
A.
(
1984
).
A specificity for cellular fibronectin in its effect on cultured chondroblasts
.
Differentiation
27
,
67
73
.
West
,
C. M.
,
Lanza
,
R.
,
Rosenbloom
,
J.
,
Lowe
,
M.
&
Holtzer
,
H.
(
1979
).
Fibronectin alters the phenotypic properties of cultured chick embryo chondroblasts
.
Cell
17
,
491
501
.
Woods
,
A.
,
Couchman
,
J. R.
,
Johansson
,
S.
&
Hook
,
M.
(
1986
).
Adhesion and cytoskeletal organization of fibroblasts in response to fibronectin fragments
.
EMBO J
.
5
,
665
670
.
Zardi
,
L.
,
Carnemolla
,
B.
,
Siri
,
A.
,
Petersen
,
T. E.
,
Paolella
,
G.
,
Sebastio
,
G.
&
Baralle
,
F. E.
(
1987
).
Transformed human cells produce a new fibronectin isoform by preferential alternative splicing of a previously unobserved exon
.
EMBO J
.
6
,
2337
2342
.