Expression pattern of the FGF-related proto-oncogene int-2 suggests multiple roles in fetal development

There is increasing evidence that growth factors and their receptors have important roles in development (Adamson, 1987; Mercola & Stiles, 1988). Although these roles may include the control of cell proliferation in certain tissues, a number of growth factors have also been implicated in other biological processes (Sporn & Roberts, 1988; Gospodarowicz et al. 1986). This is exemplified by the apparent involvement of the endothelial cell mitogen, basic fibroblast growth factor (FGF), in cell migration (Folkman & Klagsbrun, 1987), neuronal cell differentiation and survival (Morrison et al. 1986; Walicke et al. 1986; Anderson et al. 1988; Hatten et al. 1988) and inductive tissue interactions (Slack et al. 1987; Kimelman & Kirschner, 1987). Recently, we have presented evidence that an FGF-related gene, int-2 (Dickson & Peters, 1987), may have roles in the development of the gastrulating and early-somite-stage mouse embryo (Wilkinson et al. 1988).

int-2 was first identified as a locus associated with mammary tumour formation in the mouse (Peters et al. 1983). In many of these tumours, activation of int-2 transcription has occurred due to insertion of mouse mammary tumour virus in the vicinity of the int-2 gene (Dickson et al. 1984; Moore et al. 1986; Peters et al. 1986). However, normal expression of int-2 thus far reported is restricted to the postimplantation mouse embryo (Jakobovits et al. 1986; Wilkinson et al. 1988). Between 7·5 and 9·5 days of development, int-2 is expressed in parietal endoderm, early-migrating mesoderm, a portion of the pharyngeal pouches and the neuroepithelium of the hindbrain adjacent to the developing otocysts (Wilkinson et al. 1988). Based on these observations and the properties of other FGF-related proteins, we proposed that int-2 may have multiple roles, influencing cell migration and perhaps acting as a morphogen in the development of the otocyst.

In the present report, we describe the pattern of int-2 expression throughout the subsequent period of gestation. int-2 transcripts are restricted to sensory regions of the inner ear, the neuroblastic layer of the retina, Purkinje cells in the cerebellum and the mesenchyme of the developing teeth. We discuss the implications of these data for the possible functions of int-2 in mouse development.

Preparation of RNA from microdissected tissues, electrophoresis, blotting and hybridization were all performed as described previously (Wilkinson et al. 1987a,b). In situ hybridization was also performed as described (Wilkinson et al. 1987a,b).

For immunocytochemistry and in situ hybridization analysis of serial sections, the brain was dissected from 16·5 day post coitum (p.c.) embryos and fixed in 4% paraformaldehyde in PBS at 4°C overnight. Following this, the tissue was dehydrated in an increasing concentration of ethanol, embedded in polyethyleneglycol 400 distearate, 1% cetyl alcohol at 42°C, and then 6μm sections were cut and mounted on chrome-alum-subbed slides. After overnight drying of these slides, wax was removed with two changes of 100% ethanol for 10min each, and the sections air dried for at least 20 min. Adjacent sections were used for in situ hybridization or for immunocytochemistry with antiserum against chick intestinal calcium-binding protein (Jande et al. 1981), which specifically recognizes Purkinje cells in the developing cerebellum. For immunocytochemistry, endogenous peroxidase was blocked with 0·3 % hydrogen peroxide in methanol for 30 min, washed in PBS, and nonspecific binding blocked with 1% bovine serum albumin in PBS for 30min. Incubation with 1/500 diluted antiserum was carried out at 4°C overnight. Washing and visualization of antibody with horseradish peroxidase and diaminobenzidine was performed using the Vectastain kit (Vector Laboratories) as recommended by the manufacturers.

Camera-lucida drawings were made of otic tissue and the sites of int-2 expression in serial sections of the inner ear of a 17·5-day mouse embryo. Reconstruction of these data using the SSRCON system (Shepherd et al. 1984) was carried out as described previously (Wilkinson et al. 1987b).

Northern blot analysis of total RNA detects roughly constant levels of int-2 transcripts throughout development of the mouse fetus (data not shown). The in situ hybridization studies reported here reveal that this reflects complex and dynamic patterns of int-2 expression in specific regions of the developing cerebellum, retina, teeth and inner ear. Each of these sites is described in detail below.

int-2 transcripts were first detected in the developing cerebellum at 14·5 days p.c. The cerebellum originates from a population of proliferating neuroepithelial cells, termed the ventricular zone, located at the roof of the hindbrain. Postmitotic cells migrate away from the ventricular zone along routes distinctive for each of the terminal cell types (Altman & Bayer, 1985a,b) and eventually settle in specific sites to form the stratified structure of the mature cerebellum. These events occur initially in lateral regions and then in progressively more medial parts of the cerebellum. At 14·5 days p.c. int-2 expression was detected in cells adjacent to the ventricular zone in lateral sections (Fig. 1A,B) but not in more medial sections (data not shown). Two days later, int-2 transcripts were detected in cells distal from the ventricular zone in lateral regions (Fig. 1C), but in more medial sections int-2 expression occurred in cells adjacent to the ventricular zone (data not shown). This pattern suggests that int-2 expression is occurring in Purkinje cells which are known to originate and migrate in this fashion (Altman & Bayer, 1985a,b). To confirm this, serial sections were used for in situ hybridization with int-2 probe (Fig. 1C) or for immunocytochemistry with an antibody to a calcium-binding protein that is restricted to Purkinje cells in the cerebellum (Jande et al. 1981) (Fig. 1D). At 16·5 daysp.c., int-2 expression was confined to regions stained by the Purkinje cellspecific antibody. In the cerebellum of the newborn mouse, int-2 RNA was detected in a narrow zone of cells located distal from the hindbrain ventricle, and convoluted in this lateral section (Fig. 1E,F). This is the site of the Purkinje cell layer. Northern blot analysis (Fig. 1G) showed four int-2 transcripts (for details of multiple transcripts see Smith et al. 1988 and Mansour & Martin, 1988) to be present in the cerebellum at 17·5 days p.c. and in the newborn mouse, but barely detectable in the mature cerebellum of a 2-week postnatal mouse. Consistent with this, int-2 transcripts were not detected by in situ hybridization in the 2-week postnatal cerebellum (data not shown). Thus, high level int-2 expression appears to be restricted to early stages of Purkinje cell differentiation.

The retina initially consists of a one-cell-thick neuroepithelium which forms a thickened neuroblastic layer following the migration of proliferating cells. Postmitotic differentiating cells first arise in the posterior of the retina, then subsequently in more anterior regions. At·5 days p.c., int-2 expression was detected in a small group of cells in the posterior region of the retina (Fig. 2A,B). Two days later int-2 transcripts are detected throughout the entire neuroblastic layer (Fig. 2C,D). Thus, int-2 expression does not correlate with cell proliferation which is occurring throughout the neuroblastic layer during this period, but rather with the onset of appearance of postmitotic cells in early stages of their differentiation (Hinds & Hinds, 1978, 1979). While the neuroblastic layer remains apparently unstructured in the fetal mouse, ganglion cells segregate to form an inner layer, int-2 transcripts are not detected in this ganglion cell layer at 16·5 days p.c. (Fig. 2E,F) or in the newborn mouse retina (Fig. 2G,H). In the newborn mouse, int-2 transcripts are restricted to the anterior periphery of the retina (Fig. 2G,H). This region is the site of mitotic and early-stage postmitotic cells in the retina at this stage, whilst more posterior regions consist only of later-stage differentiating cells, int-2 transcripts were not detected by in situ hybridization in the mature retina (data not shown) and thus it seems that high-level int-2 expression is associated with early stages of retinal cell differentiation. Consistent with this. Northern blot analysis reveals int-2 transcripts to be down-regulated between 17·5 days p.c. and birth, and barely detectable 2 weeks postnatally (Fig. 2I).

The major structures of teeth are derived from cells of two distinct origins: ectodermal cells of the jaw epithelium that form the enamel, and neural-crest-derived mesenchymal cells that form the inner dentine and pulp tissue (Lumsden, 1987). Prospective tooth ectoderm is apparent at 12·5 days p.c.. but int-2 transcripts are not detected at this stage (Fig. 3A,B). By 14·5 days p.c.. this thickened ectodermal layer has invaginated and tooth bud mesenchyme has been recruited, int-2 expression is detected only in the tooth mesenchyme (Fig. 3C,D). Subsequently, progressive differentiation occurs concurrent with the ectodermal layer forming a morphology characteristic of incisor (Fig. 3E,F) or molar (Fig. 3G,H) teeth. At 17·5 days of development. int-2 transcripts are detected in mesenchymal cells destined to form the dentine and pulp of the incisor and first molar teeth until at least 18·5 daysp.c. (Fig. 3E-H, and data not shown), but not in the second molar which is at an earlier morphological stage (Fig. 3G,H).

The inner ear arises from a thickened layer of ectodermal cells that invaginate to form the otic vesicle. At 9·5 days p.c. int-2 expression is not detected in the otic epithelium, but rather in cells of the hindbrain that are adjacent to this structure (Wilkinson et al. 1988). However, at 10·5 days int-2 expression is no longer apparent in the hindbrain (data not shown), but is detected in a thickened region of the otic vesicle, distal from the hindbrain, that is destined to form the sensory tissue of the vestibular organ (Carney & Silver, 1983) (Fig. 4A,B). int-2 transcripts continue to be detected during the subsequent morphogenesis of the inner ear. By 17·5 days p.c.. the major structures of the inner ear have formed, and we have examined the spatial pattern of int-2 expression in serial sections of the ear at this stage. Computer-aided reconstructions of these data reveal int-2 to be expressed exclusively in the developing sensory regions of the inner ear: in the ampullae of the semicircular canals, the utricle, saccule (Fig. 5A) and the cochlea (Fig. 5B). Detailed examination of the sections shows int-2 expression in both the sensory hair cells and the underlying support cells (Fig. 4C-H and data not shown).

The idea that growth factors and growth-factor-related genes may have multiple roles, in development as well as in tissues of the adult, is an attractive possibility. This view is consistent with the complex pattern of expression of the FGF-related proto-oncogene int-2. Based on the cell types expressing int-2 from 6·5 to 9·5 days of development, we have suggested that this gene may have roles in cell migration and in the induction of the otocyst (Wilkinson et al. 1988). The present report suggests that int-2 may have other, unrelated, roles in later fetal development.

Two of the fetal sites of int-2 expression are located in the central nervous system. There is already good evidence that basic FGF is expressed in many neural tissues, including the postnatal cerebellum and retina (Hatten et al. 1988; Baird et al. 1985) and that it may act as a neurotrophic factor promoting the differentiation and/or survival of specific neurones (Morrison et al. 1986; Walicke et al. 1986; Anderson et al. 1988; Hatten et al. 1988; for critical review, see Davies, 1988). By analogy, it seems possible that int-2 may act as a neurotrophic factor during early phases of Purkinje cell and retinal cell differentiation. The Purkinje cells are migrating during the period of int-2 expression, so in view of the apparent association of int-2 expression with early mesoderm cell migration (Wilkinson et al. 1988), it is also possible that it has a role in Purkinje cell migration.

The detection of int-2 transcripts in sensory cells and the underlying support cells in the inner ear is consistent with a role in sensory cell differentiation. These sensory cells form distinct structures in the ampullae, the utricle and saccule and the cochlea, but share the characteristic of being innervated by the vestibulocochlear ganglion. Thus, an alternative idea is that int-2 may act as a neurotrophic factor for the vestibular and cochlea nerves. Consistent with this, evidence exists for a trophic influence of the sensory epithelium on the vestibular and cochlea nerves (Sobkowicz et al. 1980; Ard et al. 1985). Overall, we suggest that int-2 may have two distinct roles in the formation of the inner ear; initially as a morphogenetic signal arising from the hindbrain between 8·5 and 9·5 days p.c. (Wilkinson et al. 1988), and subsequently as a factor involved in sensory cell differentiation and/or innervation. The location of the prospective sensory epithelium distal from the hindbrain at 10·5 days p.c. is consistent with this two-step model, rather than with a single role of int-2 in inner ear development.

There is at present no experimental evidence for a role of FGF-like proteins in tooth development. In view of the reciprocal inductive tissue interactions between the epithelial and mesenchymal components of the developing tooth (Thesleff & Hurmerinta, 1983; Kollar, 1983), it is possible that int-2 acts as a paracrine factor required for differentiation of the ectodermally derived ameloblasts. However, int-2 expression does not correlate specifically with the timing of mesenchymal induction of ectodermal differentiation (Lumsden, 1987), and thus it seems more likely that int-2 acts as an autocrine factor for the odontoblast pathway.

In summary, we envisage int-2 to have multiple roles in mouse development and we have made specific proposals based on correlations of int-2 RNA accumulation with developmental events. Rigorous testing of these ideas will require analysis of the biological effects of int-2 protein, and of neutralizing antibodies on in vitro culture systems, and the identification and characterization of its cellular receptor.

We wish to thank Eric Lawson for the gift of anti-calcium-binding protein antiserum and John Green for assistance with the computer-aided reconstructions. We are also grateful to Roger Morris, Andrew Lumsden, Peter Woodhams and John Seeley for useful advice and discussions, and thank Ena Heather for efficient typing.