Retinoic acid modulation of the early development of the inner ear is associated with the control of c-fos expression

Retinoic acid exerts a variety of biological effects. The teratogenic consequences of hypervitaminosis A and the administration of retinoids have been known for a long time and include alterations of growth and morphogenesis of the nervous system (Morriss, 1972; Geelen, 1979; Durston et al. 1989). A specific role for retinoic acid during normal development has been supported by experiments on the developing limb where it is thought to specify position across the limb anlage (Tickle et al. 1982; Thaller and Echielle, 1987). In vitro experiments, using cultured cells, have shown that retinoic acid inhibits cell proliferation and induces differentiation in transformed cells from embryonic origin (Jetten, 1986; Sidell et al. 1983). The mechanism by which retinoic acid exerts its actions is being extensively studied and the role of RA receptors as transcriptional factors is beginning to be understood (Petkovich et al. 1987; Evans, 1988). Modulation of gene expression by retinoic acid extends to a variety of genes including homeobox and proto-oncogenes (Adamson, 1987; Astigiano et al. 1989).

The development of the inner ear is an interesting example of organogenesis in the nervous system. At early developmental stages, it involves the formation of a transient structure, the otic vesicle, which undergoes a distinct period of cell proliferation that precedes the differentiation and histogenesis of various sensory, secretory and supporting elements (Van de Water, 1984; Giraldez et al. 1987; Swanson et al. 1990). This process can be reproduced in vitro and it can also be arrested, to then be reactivated by growth factors (Represa et al. 1988; Miner et al. 1988; Represa and Bernd, 1989). We exploit here this preparation to examine the effects of retinoic acid on the early development of the inner ear and its relation to the expression of the nuclear proto-oncogene c-fos. A possible association between RA and c-fos is relevant because the c-fos gene product acts also as a transcriptional modulator whose expression is regulated by growth factors (see Curran, 1988). More generally, the possibility of regulatory roles for proto-oncogenes in normal growth during embryonic development is very attractive (Adamson, 1987).

The results show that physiological concentrations of RA (1–20 nw) produced a strong inhibition of cell proliferation along with a selective induction of cell tissue differentiation. These effects were associated with a rapid inhibition of mitogen-induced c-fos mRNA levels. They suggest a role for RA in controlling the early development of the inner ear and that this control is effected through the regulation of the protooncogene c-fos.

Experiments were done on chick embryo otic vesicles that were isolated by microdissection at stage 18 and cultured according to procedures described previously (Represa et al. 1988; Represa and Bernd, 1989) but with the following modifications. The culture chamber contained 250 μl of antibiotic-free culture medium, vesicles were incubated at 37 °C in an atmosphere of 5 % CO₂ and the standard culture medium consisted of serum-free M-199 medium with Earle’s salts (Flow laboratories).

For morphometry, otic vesicles were processed as previously described (Represa et al. 1986; Giraldez et al. 1987). Briefly, isolated otic vesicles were fixed with Bouin’s solution, dehydrated and embedded in paraffin. Serial frontal sections (8 μm) were then stained with haematoxylin-eosin. Morphometric measurements were made using graphical reconstructions in camera lucida. Some specimens (Fig. 5) were fixed in 2 % glutaraldehyde in 0.1 M cacodylate buffer, dehydrated in a graded series of acetones and propylene oxide and embedded in Epon resin. Frontal or sagittal sections (1 μm) were cut and stained with 1 % toluidine blue. Light microscopy radioautography was performed as described elsewhere (Bernd and Represa, 1989).

DNA synthesis was measured as acid-precipitable [³H]thy-midine incorporation. Vesicles were placed in standard incubation medium containing 0.4 μM (10 μCi ml^-1) [³H]thy-midine for periods of 24 h. Vesicles were then washed in cold Ringer, extracted with 10 % trichloroacetic acid and the radioactivity incorporated into acid-precipitable material was counted in Triton X-100-toluene using a scintillation counter.

For Northern Blot Analysis, total RNA was extracted by lysing isolated otic vesicles in 3 M guanidinium isothiocyanate and centrifugation over a CsCl gradient. RNA was quantitated by measuring OD₂₆₀ and analysed by electrophoresis in formaldehyde 1.3% agarose gels. After transfer to nylon membranes (Nytran, Schleicher and Schuell), the Northern Blots were hybridyzed with a random-priming labelled avian c-fos probe (2×10⁹cts min^-1μg^-1) in 5×SSPE, 50% formamide, 0.1% SDS for 20h at 55°C, washed and autoradiographed. The c-fos probe (Molders et al. 1987) was generously provided by Drs Rolf Muller and Martin Zenke.

The early development of the vertebrate inner ear involves the thickening and invagination of the ectoderm and the formation of the otic vesicle. At developmental stage 18 (Hamburger and Hamilton, 1951), it consists of a fluid-filled cavity lined by a transporting epithelium (Represa et al. 1986; Giraldez et al. 1987). In about 48 h the otocyst goes through a period of intense cell proliferation and evolves to form a more complex structure with signs of growth and morphogenesis. Cell division in the otic vesicle can be arrested in vitro by incubation in serum-free media and then reactivated by growth factors and mitogens (Represa et al. 1988; Miner et al. 1988). Fig. 1 shows a typical experiment where the effects of retinoic acid on this growth-factor-reactivated growth were assayed. The photograph on the left shows an otic vesicle that was isolated at stage 18 and arrested for 24h (OS). Two other vesicles reactivated to grow with 10% serum (10S) or with 100 nM bombesin plus 5 μgml^-1 insulin (B+I) are shown. They displayed the characteristic signs of growth and were taken as control conditions. The presence of RA in the incubation medium produced a concentration-dependent inhibition of vesicular growth whether induced by serum or bombesin. The numbers in Fig. 1 indicate RA concentrations (HM) and it can be seen that the inhibitory effect was detectable at InM RA and increased up to 50 nM. The incorporation of [³H]thymidine into acid-precipitable material in cultures corresponding to the above experiments are illustrated in Fig. 2. DNA synthesis was reduced in parallel with the inhibition of growth. Inhibition of [³H]thymidine incorporation was already detectable at InM RA concentration and a maximal effect occurred between 10 and 25 nM RA in vesicles reactivated by bombesin and serum, respectively. Fig. 3 compares autoradiographic sections of otic vesicles incubated with .serum (10%) in the absence (Fig. 3A, C) or in the presence (Fig. 3B, D, E) of retinoic acid. An example of an arrested vesicle, incubated in the normal medium but in the absence of additives (Fig. 3F), is shown for comparison. The reduced uptake of labelled thymidine with retinoic acid was associated with the loss of labelling of nuclei in the epithelium. It should be noted that the regional pattern of cell proliferation in the otic vesicle, ventral and medial, was preserved in the presence of RA.

The effect of retinoic acid on cell proliferation was also estimated by measuring the volume of the epithelial tissue of the otic vesicle. This was obtained morphometrically from serial reconstructions of otic vesicles that were incubated in the presence of 10 % serum and with increasing concentrations of retinoic acid (Fig. 4). These experiments were done in parallel with those used for DNA incorporation measurements. Cell density values remained fairly constant at different RA concentrations. For example, cell density values (× 10⁸cells cm^-3) were 95±26 for 10S in the absence of RA, 80±22 in 17 nM RA and 89±8 in OS (42 fields per vesicle, 2 vesicles per condition). This indicates that the measurements of epithelial tissue volume should faithfully reflect the absolute number of epithelial cells in each condition. For the cell densities given above and the corresponding values of epithelial volume from

Fig. 4, the number of cells (×10⁻⁴) was 13.5 in 10S, 2.4 in 17 nM RA and 2.5 in OS. The inhibition of RA on cell proliferation was, therefore, also demonstrated by direct estimates of the cell number.

Several morphological changes were also associated with the inhibition of cell proliferation by retinoic acid, and they were studied in 17 different vesicles cultured in serum plus different concentrations of retinoic acid 1 nM (4), 10 nM (9), and 25 nM (4). Another 8 vesicles incubated in serum with no added RA served as controls. First, low concentrations of RA (1–10 nM) in the presence of serum produced the induction of differentiated structures that are normally present in the otic vesicle at much more advanced developmental stages. Fig. 5A,B shows that incubation with serum and RA (10 nM) for 24 h induced the differentiation of dome-like epithelial structures, which strongly resembled the tegmentum vasculosum with characteristic large scale corrugations, and which normally do not appear until developmental stage 30–33 (5–6 days in culture). This is a secretory epithelium, which is the equivalent in birds of the mammalian stria vascularis. An example of a tegmentum vasculosum from a chick embryo otic vesicle after 5 days in culture is shown in Fig. 5C,D for comparison. No such structure was observed in 24–48 h cultures in the absence of RA in the present or previous work. Note that the RA-induced differentiation was located in the dorsal region of the otic vesicle which, during normal development, will also give rise to the physiological tegmentum vasculosum . The early induction of the tegmentum vasculosum was observed in 13 specimens (4, 5 and 4 in 1, 10 and 25nM RA, respectively). Another sign of differentiation of the secretory system was the formation of the endolymphatic sac (see below).

Fig. 6 shows the thickening of defined areas of epithelium (Fig. 6A, area within arrows) which became stratified and contained both ‘clear’ and ‘dark’ cells with toluidine blue staining (Richardson blue). ‘Dark cells’ were flask-shaped and typically polarized towards the inner surface (Fig. 6B,C and inset). This epithelium was adjacent to the developing cochleo-vestibular ganglion (CVG in Fig. 6) and closely resembled intermediate stages of differentiation of the auditory and vestibular sensory epithelia, which again occurs much later during the course of normal development (days 5–6) (see Whitehead and Merest, 1985). This epithelium gives rise to hair cells by day 7 in vivo or the equivalent time in grafting experiments (Swanson et al. 1990). These features were observed in 6 vesicles incubated in 10nM RA, 3 in InM RA and 1 in 25 nM RA).

The second interesting morphological feature associated with retinoic acid incubation was the inversion of the direction of growth in the endolymphatic duct, which normally develops as a finger-like outpocketing of the dorsal vesicular wall but which in vesicles cultured with RA occurred as an invagination of the dorsal wall. Fig. 7 shows that the induction of the differentiation of the endolymphatic sac was also accelerated compared to normal development. The endolymphatic sac appears normally between days 6 and 7 of development in vivo. Fig. 7 also shows, however, that it was directed inwardly towards the vesicular cavity. The reversion of the endolymphatic duct was observed in 7 out of the 9 vesicles incubated in 10 nM RA. The fully developed endolymphatic sac could be detected in two of these experiments.

The tendency of the growing vesicle to polarize inwardly is also illustrated in Fig. 8 where the values of the internal and external vesicular surface areas (S_i, S_e) obtained morphometrically were plotted against RA concentration in the culture medium. During normal growth (serum-stimulated) S_i was always smaller than S_e, a relation that also applied for growth-arrested vesicles. Increasing concentrations of RA reversed this relationship S_i becoming larger than S_e. The inset of Fig. 8 shows the ratio S_i/S_e plotted versus RA concentration. It can be seen that the ratio S_i/S_e increased from about 0.7 in the absence of RA to 1.25 at 50 nM RA. Average values for S_i/S_e ratios were 0.77±0.03 and 1.33±0.07 for 10S and 10S+RA (50nM), respectively (n=3). That is, at high concentrations of RA the internal surface had increased almost two-fold relative to the external surface, and the normal polarity of growth was reversed.

Finally, we examined the possibility that the effects of RA were related to c-fos expression in the otic vesicle. Fig. 9A shows a Northern Blot analysis of otic vesicles at different developmental stages (indicated by the numbers). Levels of c-fos mRNA were undetectable in stage 12, they were just measurable in stage 18, increased in stage 21 and were again undetectable in stages 22 and 27. Values of optical density are plotted in Fig. 9B along with the cumulative cell number taken from Giraldez et al. (1987) and normalized to the value in stage 21. Two other experiments gave a similar stagedependent increase in the mRNA level of c-fos. The expression of c-fos appeared, therefore, transient and correlated to the proliferative phase of the development of the otic vesicle.

Incubation of growth-arrested vesicles with bombesin plus insulin for 20 min at concentrations that induced cell proliferation and vesicular growth, produced a strong induction of c-fos (Fig. 10A, lane B + I). RNA extracted from stage 21 otic vesicles was run in this particular experiment as a positive control (lane 21). In view of the anti-proliferative effect of retinoic acid on bombesin-induced cell proliferation, we examined the effect of RA on c-fos induction. Growth-arrested otic vesicles were incubated in the presence of bombesin plus insulin for 20min in the presence of RA (25n,M). The result is shown in Fig. 10A (lane B + I+RA) and demonstrates that the mitogen-induced increase in c-fos mRNA level was suppressed in the presence of retinoic acid even below the control value (0S). A semiquanti-tative analysis of these results is given in Fig. 10B, where normalized values of optical density are displayed.

The results reported here show that retinoic acid inhibits cell proliferation, induces differentiation and positional changes, and suppresses c-fos induction. A crucial point is that the observed effects of retinoic acid on the otic vesicle are produced at concentrations ranging between 1 and 25 nM. These can be considered as ‘physiological’ concentrations since (1) these figures are within the range of concentrations measured in chick embryos by HPLC, 10–50nM, (Thaller and Eichelle, 1987), and (2) they are close to the affinity of RA-receptors for RA, 0.6–30 nM (Giguere et al. 1987; Petkovich et al. 1987). This quantitative correspondence is important because it sets a constraint for any possible role of retinoic acid in normal development.

Evidence has accumulated during recent years that retinoic acid can operate as a morphogen during embryogenesis and that concentration gradients of RA modulate tissue pattern formation in the developing limb (see Summerbell and Maden, 1990 for review). The effects of retinoic acid on the otic vesicle can be taken as a more general extension of this property. Retinoic acid induced in the otic vesicle differentiated tissue organizations such as the tegmentum vasculosum, early sensory epithelium and endolymphatic sac, which normally appear much later in development. In view of the results presented here, the search for early molecular markers of differentiation for the specific cell types becomes imperative and current work is in progress in that direction. Our results show that, morphologically, the differentiation of secretory structures can be accelerated by retinoic acid to reach very advanced stages of differentiation in the 24 h incubation period of observation. This effect, although present, was less pronounced in the case of sensory epithelia where fully developed receptors were not detected. The origin of this apparent differential sensitivity to retinoic acid remains unknown. The induction of differentiation had not only the property of being physiological in terms of the appearance of the tissue, but also it was located in the ‘correct’ presumptive areas of the otic vesicle, i.e. those from which they would originate during normal development (Li et al. 1978). Whether this is related to the regional distribution of RA receptors or to the selective activation of cellular programme is unknown. The effect of RA reversing the normal polarity of the growth of the otic vesicle is perhaps more difficult to interpret, but may well reflect the ability of RA or RA-gradients to organize growth in three dimensions. It has to be considered that, although non-stationary, a gradient of RA was established in the cultures, which was normal to the plane of the epithelium and directed towards the vesicular cavity.

The effects of retinoic acid on the differentiation of vesicular tissue were associated with a strong inhibition of cell proliferation. That cells inhibited to proliferate were those of the vesicular epithelium with little contribution, if any, from the surrounding mesenchymal cells was confirmed by the parallel evolution of epithelial tissue volume and DNA synthesis measurements. Thus, as in normal development, RA-induced differentiation is associated with the interruption of cell division. The precocious appearance of differentiated structures could either be due to a direct activation of specific transcriptional programme by RA or to the induction of such activity as a result of the inhibition of cell proliferation. Differentiation may normally be incompatible with proliferative phases and the transition to differentiative periods could be then controlled by inhibition of cell proliferation by RA. These observations on the effects of RA on the early development of the inner ear, taken together, suggest a physiological role for this molecule in the regulation of the development of the otic vesicle. Retinoic acid could, therefore, represent a putative morphogen in the developing inner ear.

The c-fos proto-oncogene belongs to a group of rapidly induced genes encoding proteins that form complexes regulating transcription (see Herschman, 1989 and Curran, 1988). Several growth factors are known to induce c-fos gene and protein (Müller et al. 1984; Lau and Nathans, 1987; Almendral et al. 1988). Yet, the relationship between c-fos expression and the processes of cell proliferation and differentiation, as judged from studies on cultured cell lines, is still unclear. Expression of c-fos has been measured in mouse embryo extracts during days 1 to 10 of prenatal development (Müller et al. 1982) and, using in situ hybridization techniques, c-fos mRNA has been detected in certain areas of the nervous system in late stages of mouse development (Caubet, 1989). However, no functional relation to defined events during embryogenesis has been demonstrated. The results reported here show (1) a transient expression of c-fos throughout the development of the otic vesicle, (2) the rapid induction of c-fos by bombesin and (3) its inhibition by retinoic acid. The first point is of interest because it shows the constitutive expression of c-fos in the otic vesicle and that it is stage dependent. Moreover, high levels of c-fos expression coincide with the most active period of the proliferative phase in the otic vesicle suggesting that the physiological regulation of c-fos is associated with the control of cell proliferation. This is also supported by the fact that a mitogen like bombesin is able to induce the expression of c-fos and, additionally, by the association between the inhibitory effects of RA and the suppression of c-fos induction. This indicates that c-fos expression may be a critical element for regulation of cell proliferation in the otic vesicle. Moreover, c-fos appears to be a target for retinoic acid. The developmental effects of RA, therefore, might be in part exerted via the regulation of the expression of c-fos.

We wish to thank D. N. Sheppard (Cambridge) for reading the manuscript and Esther Vazquez for encouragement and help in RNA analysis. The work was partially funded by DGICYT PB86/0326, FIS89/0301 and Junta de Castilla y León/88.