The inner ear in both mammals and birds comprises a dorsal vestibular compartment for balance and a ventrally positioned auditory part for sound wave detection. Inner ear sensory functions are carried out by specialized sensory organs located in restricted regions of the epithelial labyrinth. The six mammalian and eight avian sensory organs each contain in a precise order various types of supporting cells together with mechanosensory hair cells (reviewed in Wu and Oh,1996; Riley and Phillips,2003; Barald and Kelley,2004).
The sophisticated labyrinth, together with the neurons of the vestibulocochlear ganglion, develops from a simple ectodermal otic placode, which is induced adjacent to the caudal hindbrain. The otic placode first invaginates, forming an otic cup, and thereafter closes to form a vesicle. During subsequent stages, the development of the vestibular and cochlear compartments occurs rather independently of each other and is controlled by a partially overlapping set of genes (reviewed in Torres and Giraldez,1998; Anagnostopoulos,2002; Riley and Phillips,2003; Fekete,2004).
Members of the Gata-family zinc finger transcription factors are essential in numerous developmental processes such as cell-fate specification, cell proliferation, differentiation, as well as cellular movements (reviewed in Patient and McGhee,2002). Two mouse Gata-factor encoding genes, Gata3 and Gata2, are expressed in developing inner ear already at early stages (Karis et al.,2001; Lawoko-Kerali et al.,2002; Lilleväli et al.,2004,2006). Gata3 is known to regulate multiple events during inner ear development. It is required at the first morphogenetic steps during otic placode invagination to generate a closed vesicle (Lilleväli et al.,2006). Also the subsequent morphogenesis is dependent on Gata3 since no semicircular ducts form in Gata3−/− embryos, and the utricle as well as the cochlear and endolymphatic ducts are strongly reduced (Karis et al.,2001; Lilleväli et al.,2006). Furthermore, Gata3 is also involved in the formation of the cochlear ganglion and in the guidance of efferent neuron axons to the ear (Karis et al.,2001). In addition to morphogenesis and neurogenesis, the continuous expression of Gata3 during cochlear hair cell development suggests an important function for this factor also in auditory sensory development (Lawoko-Kerali et al.,2002; Lilleväli et al.,2004; van der Wees et al.,2004). A role for Gata3 in hair cell maintenance in adult mice has already been established (van der Wees et al.,2004).
Despite dramatic problems in epithelial morphogenesis, inactivation of Gata3 does not affect early marker gene expression or patterning of the otic vesicle. However, Gata2 is strongly down-regulated in Gata3-deficient otic epithelium showing that Gata2 is downstream of Gata3 during inner ear development (Lilleväli et al.,2004). Embryos lacking Gata2 die around embryonic day (E) 10.5 without any defects in inner ear, suggesting that Gata3 can compensate for the loss of Gata2 during early otic development (Tsai et al.,1994; Lilleväli et al.,2004).
Here we have analyzed the expression of chicken Gata3 and Gata2 during avian inner ear development and compared the observed patterns with those known for mouse. We were especially interested in getting insights into the conservation of the Gata3-regulated events in morphogenesis, neurogenesis, and sensory development and into the preservation of the regulatory hierarchy between the two Gata factors. In addition, because nothing was known of the roles of Gata2 in inner ear, we hoped to get more indications of its functions by comparing its expression in two distinct vertebrate classes. Mouse Gata3 is the first factor known to regulate Fgf-signaling in inner ear by activating Fgf10 expression (Lilleväli et al.,2006). Therefore, we compared also the expression patterns of Gata3 and Fgf10 during chicken otic development.
Our results showed that Gata3 expression was predominant in developing sensory organs and cochlear ganglion while Gata2 was expressed in nonsensory epithelium, suggesting divergent roles for these two related factors during ear development. When compared with mouse, the expression patterns became increasingly similar when morphogenesis progressed. However, interesting differences were observed during otic cup stage, where chicken Gata3 was restricted to the lateral part and the strongest expression of the mouse counterpart was confined to the medial part. This observation suggested that the molecular control of placode invagination might be different in mouse and chicken.
RESULTS AND DISCUSSION
Expression of Gata3 During Early Otic Development
In chicken embryos, expression of Gata3 was detected already at the preplacodal stage (Hamburger and Hamilton stage [HH] 9) covering a broad domain of the surface ectoderm, including the presumptive otic region (data not shown). At HH12, Gata3 was expressed over the entire otic placode extending also to the neighboring surface ectoderm (Fig. 1A). Thus the initiation of Gata3 expression in chicken occurred in an earlier morphological stage than in mouse, where Gata3 expression can be detected first in a distinguishable otic placode (Lilleväli et al.,2006). Nevertheless, in both vertebrate species, Gata3 is one of the earliest otic marker genes.
During otic placode invagination in chicken (HH14–HH16), Gata3 expression became restricted to the lateral-most compartment of the forming otic cup. At HH14, expression was detected only in a small anterolateral domain (data not shown). At HH15, Gata3 expression was detected as a rather narrow strip covering most of the rim lining the opening to the surface except for the dorsal parts (Fig. 1B,C). Strongest expression was detected in the anterior-most area (arrowhead in Fig. 1B). The strong lateral expression of Gata3 along the closing rim of the otic cup could indicate a potential role in the epithelial closure event in chicken to form a vesicle. In fact, closure defects are frequently observed also in Gata3-deficient mouse embryos, although the mouse counterpart is strongly expressed throughout the otic cup (Lilleväli et al.,2006). Especially strong expression of mouse Gata3 localizes to the ventromedial region close to the neural tube, and this seems important for otic cup morphology during invagination since inactivation of Gata3 leads to severe perturbations in cellular arrangement/adhesion in this region (Lilleväli et al.,2006). The lack of Gata3 in the ventromedial part of chicken otic cup suggests that the chicken factor is not directly involved in regulating epithelial morphology in the medial wall. This surprising result suggests that the molecular control and requirements for otic cup morphogenesis during invagination differ remarkably between birds and mammals.
After otic vesicle closure, expression of chicken Gata3 remained restricted to the lateral wall (Fig. 1D), also where strongest mouse Gata3 can be observed at the corresponding stage (Lilleväli et al.,2004). No expression was detected in surface ectoderm or in periotic mesenchyme during otic cup and early vesicle stages (Fig. 1B–D). The laterally restricted expression of chicken Gata3 in the otic vesicle stage suggests important roles in the subsequent development of the derivatives of the lateral wall such as the semicircular ducts, which are completely missing in Gata3-deficient mouse embryos (Karis et al.,2001; Lilleväli et al.,2006).
When otic vesicle started to grow out and elongate at HH19, chicken Gata3 was expressed ventromedially (arrowhead in Fig. 1N) in addition to a broad lateral domain. At HH21, Gata3 expression was detected also in two distinct domains of the periotic mesenchyme (asterisks in Fig. 1R) and in the posteroventral epithelium (arrow in Fig. 1R). Gata3 was expressed in a highly similar manner at corresponding stages in mouse (Lilleväli et al.,2004).
Expression of Gata2 and Fgf10 During Early Otic Development
Chicken Gata2 expression was first detected in a small ventrolateral area of the otic cup (Fig. 1F,G). This area was included in the broader Gata3 expression domain (Fig. 1B,F). In the closed otic vesicle, Gata2 continued to be expressed like Gata3 in the lateral wall (Fig. 1D,H). At HH19, expression of Gata2 remained ventrolateral (Fig. 1O), overlapping with the lateral domain of Gata3 expression (Fig. 1N). However, Gata2 was absent from the more medial Gata3-positive domain (arrowhead in Fig. 1N). At HH21, the ventrolateral Gata2 expression had extended anteriorly into the epithelium next to the forming otic ganglion (Fig. 1S) partially overlapping with Gata3 (Fig. 1R). In contrast to Gata3, no Gata2 expression was detected in the periotic mesenchyme (Fig. 1S). The observed expression of chicken Gata2 resembled closely that known for the mouse counterpart at these early stages except that, in mouse, Gata2 expression is initiated later, in the newly closed otic vesicle (Lilleväli et al.,2004).
Both chicken and mouse Fgf10 genes are markers for the early sensory–neural otic epithelium (Pirvola et al.,2000; Alsina et al.,2004). We observed Fgf10 in the anterior part of the chicken otic placode at HH11–HH12 (Fig. 1I, and data not shown) inside the broader Gata3 domain (Fig. 1A, and data not shown). Similarly to mouse, Fgf10 expression was confined to a small anterior area of the chicken otic cup (Fig. 1J,K), where its expression extended more medially than that of Gata3 or Gata2 (Fig. 1C,G). In the newly closed otic vesicle at HH17 (Fig. 1L), and a day later at HH19 (Fig. 1P), Fgf10 was expressed in the small anteroventral sensory–neural competent area (arrowhead in Fig. 1P), but was lost in the more lateral parts of the epithelium where Gata3 expression was strongest (Fig. 1D,N). At HH21, Fgf10 was also detected on the posteroventral side of the otic vesicle (arrow in Fig. 1T) partially overlapping with Gata3 (arrow in Fig. 1R).
In mouse, Gata3 expression starts before Gata2 and Fgf10 during otic development and both genes are dramatically down-regulated in the early otic region of Gata3−/− embryos (Lilleväli et al.,2004,2006). Furthermore, Gata3 can transactivate Fgf10 expression in cell culture assays (Lilleväli et al.,2006), and the presence of several conserved GATA DNA binding sites in mouse, chicken, and human Fgf10 upstream regions (Ohuchi et al.,2005; Lilleväli et al.,2006) suggests that Gata3 can directly regulate Fgf10 expression in many species. We show here that the order of expression initiation is conserved in chicken and that the initial expression domains of Gata2 and Fgf10 are completely included in the broader Gata3 domain. Thus, the regulatory hierarchy of Gata3 being upstream of Gata2 and Fgf10 in early otic development could be conserved among vertebrates.
Expression of Gata3 and Fgf10 During Vestibular Development
In chicken, the period between HH23 and HH30 is a time of rapid morphogenesis, when all the different inner ear structures develop from a simple elongated otic vesicle (Bissonnette and Fekete,1996). In the vestibule, the utricle and saccule become separated from each other by deepening constrictions in the ventral part, while the three semicircular ducts are formed by a multistep process in the dorsal portion (Bissonnette and Fekete,1996). During this period, also the hair cells in the different sensory epithelia start to differentiate (Goodyear et al.,1995).
During chicken vestibular morphogenesis and growth at HH24–HH38, Gata3 showed robust expression in the lateral parts of the periotic mesenchyme, whereas weaker expression was detectable in the distal parts of the outgrowing and newly formed semicircular ducts (Fig. 2A,E,G,I,L, and data not shown). The observed expression in chicken vestibule resembles highly that observed earlier in mouse at corresponding stages (Lilleväli et al.,2004).
In addition to the periotic mesenchyme and semicircular ducts, a rather strong expression of Gata3 was observed at HH28–HH38 in a small region of thin nonsensory epithelium between two thicker sensory compartments (arrowheads in Fig. 2I,L), corresponding to the developing lateral crista and macula of utricle. The center of this region expressed also Gata2 (arrowhead in Fig. 2J). Interestingly, this small domain is the only vestibular nonsensory region where mouse Gata3 expression can be detected at E14.5 (data not shown). This region seems special because it is the only nonsensory domain that is clonally related to the neurons in the cochlear ganglion (Satoh and Fekete,2005).
Mouse Fgf10 is expressed in all sensory epithelia during inner ear development as well as in the vestibulocochlear ganglion (Alvarez et al.,2003; Pauley et al.,2003; Sánchez-Calderón et al.,2005). Expression of chicken Fgf10 in otic sensory areas has not been analyzed before. We detected Fgf10 in all vestibular sensory epithelia through chicken ear development at HH24–HH38 (Fig. 2C,H,K,N,Q, and data not shown) in a very similar way to Bmp4, a known inner ear sensory marker (Wu and Oh,1996; Cole et al.,2000; Fig. 2D,S, and data not shown).
At the borders of the above-mentioned vestibular nonsensory region, Gata3 expression extended to overlap with Fgf10 (Fig. 2I,K,L,N). In fact, the observed Gata3 expression in utricular macula at HH24–HH28 (Fig. 2G,I) became clearly confined to the striolar region at HH38 (Fig. 2L) where it has also been observed in young posthatch chicken (Hawkins et al.,2003). Weak expression in the striola of the saccule was also observed at HH38 (data not shown). Restricted Gata3 expression in striolas of embryonic and postnatal mice has also been observed (Karis et al.,2001, and data not shown). In contrast to Gata3, Fgf10 was expressed all through the maculae, including the striolar regions (Fig. 2N, and data not shown). No Gata3 expression was detected in the macula of the lagena at any of the stages analyzed (arrowhead in Fig. 3A, and data not shown). At all developmental stages analyzed, we used Pax2 as a control gene (Fig. 2T, and data not shown) to identify especially the developing endolymphatic duct and saccule (Sánchez-Calderón et al.,2005).
Weak Gata3 expression appeared also in the edges of the cristae at HH28 (arrows in Fig. 2I), preceding a strong expression localizing in the lingula regions of these sensory structures at HH38 (Fig. 2O). Fgf10 expression was excluded from these parts, which were devoid of hair cells (Cole et al.,2000), and was instead strongly expressed in more centrally located hair cells (Fig. 2Q) with Bmp4 (Fig. 2S). Gata3 expression extended also further down on the slopes of the cristae (arrows in Fig. 2R) to the dark cells that secrete K+ to the endolymph (Wangemann,2002). As in chicken, mouse Gata3 expression is restricted to the nonsensory compartment of the cristae, the cruciate eminence (Karis et al.,2001, and data not shown). No Gata3 expression could be detected in morphologically distinguishable hair cells in either mouse (Lilleväli et al.,2004, and data not shown) or chicken vestibule, suggesting that this factor plays no direct roles in their maturation or maintenance.
In contrast to the hair cell-specific chicken Fgf10, mouse Fgf10 is expressed also in the nonsensory cruciate eminence (Pauley et al.,2003, and data not shown). Inactivation of mouse Fgf10 results in the absence of cruciate eminence and abnormal orientation and morphology of the cristae (Pauley et al.,2003). The overlapping expression of Gata3 and Fgf10 in cruciate eminence suggests that, in mouse, Gata3 could influence the formation of this structure by controlling Fgf10 expression. In chicken, however, the expression of Fgf10 and Gata3 overlaps partially only in the developing cristae, and a strictly complementary expression was observed in more mature sensory organs. This finding suggests that Gata3 cannot be directly required for the maintenance of Fgf10 expression in more mature cristae.
Expression of Gata2 in Vestibule
Gata2 was expressed broadly in the nonsensory parts of the chicken vestibular epithelium at HH24–HH38, except for the saccule, and no expression in the developing sensory domains was observed (Fig. 2B,F,J,M,P). An especially strong expression was detected in the proximal parts of the semicircular ducts in the epithelium that was forming the fusion plate (Fig. 2B, and data not shown). This special epithelium is destined to disappear through programmed cell death in chicken to generate the duct structures (Fekete et al.,1997). After fusion plate clearance, Gata2 expression was detected in the inner edges of the newly formed semicircular ducts being complementary to Gata3 there (data not shown). Clear Gata2 expression was detected also in the periotic mesenchyme surrounding the developing semicircular ducts at HH24–HH28 (Fig. 2B,F). These observations indicated a specific role for chicken Gata2 related to the fusion plate in contrast to a more general function of mouse Gata2, which is strongly expressed throughout the vestibular nonsensory epithelium (Lilleväli et al.,2004).
At HH38 in chicken, the mesenchymal Gata2 expression was restricted to the fibrocytes located immediately beneath the vestibular sensory organs except the lagenar macula (Fig. 2M,P, and data not shown). The same kind of mesenchymal expression was also detected beneath the mouse vestibular organs at E14.5–18.5 (data not shown). Gata2 was not expressed in any of the vestibular sensory epithelia in either mouse or chicken, strongly suggesting that is not directly involved in vestibular hair cell development. However, the conserved Gata2 expression in mesenchymal fibrocytes during hair cell maturation in both chicken and mouse may have indirect effects on hair cells, because fibrocytes participate in the regulation of ionic balance, which in turn, is essential for proper hair cell functioning (Delprat et al.,2002, and references therein).
Expression of Gata3, Gata2, and Fgf10 During Cochlear Development
The cochlear duct starts to grow out from the ventral part of the chicken otic vesicle, and a small outgrowth can be distinguished at HH23–HH24 (Bissonnette and Fekete,1996). The lateral cochlear duct epithelium is thinner and will give rise to nonsensory structures, while the thicker medial epithelium will contribute to the sensory organ, basilar papilla, where the first auditory hair cells can be detected around HH29 (Goodyear et al.,1995).
The strongest Gata3 expression was detected in the medial wall of the cochlear duct during HH24–HH28 (Fig. 3A,D,G, and data not shown). At HH38, Gata3 was strongly expressed in both hair and supporting cells of the basilar papilla as well as in the neighboring clear and cuboidal/hyaline cells (Fig. 3J). Similarly to Gata3 in chicken, mouse Gata3 has been detected in both hair and supporting cells of the developing cochlear sensory epithelium (Lilleväli et al.,2004; van der Wees et al.,2004). This expression is retained also in posthatch/postnatal animals (Hawkins et al.,2003; van der Wees et al.,2004). The strong and constant expression of Gata3 in mouse and chicken cochlear sensory organ suggests a conserved role in the development and maintenance of the auditory functions. As in mouse (Lilleväli et al.,2004), weak Gata3 expression could also be detected in the ventral part of the lateral wall of the cochlear duct at HH24–HH28 (arrowheads in Fig. 3G, and data not shown). Later at HH38, expression was detected in the tegmentum vasculosum (Fig. 3J), which is the chicken equivalent of mouse stria vascularis, where Gata3 is also present (Lilleväli et al.,2004).
At HH24–HH28, Gata2 was expressed weakly in the lateral and strongly in the medial wall of the outgrowing cochlear duct (Fig. 3B,E,H, and data not shown), partially overlapping with the domain expressing Gata3 (Fig. 3A,D,G, and data not shown). At HH38, Gata2 was expressed together with Gata3 in cuboidal/hyaline cells and weakly in part of the supporting cells of the basilar papilla (Fig. 3J,K). As in mouse (Lilleväli et al.,2004), no expression in sensory hair cells could be detected (Fig. 3K, and data not shown). However, because Gata2 was expressed in the presensory epithelium of early cochlear duct in both mouse and chicken, we cannot exclude a possible early role for Gata2 in cochlear sensory development.
We detected chicken Fgf10 only in the medial wall of the growing cochlear duct at HH24–HH28 (Fig. 3C,F,I, and data not shown) and at HH38 Fgf10 expression was confined to the hair cell precursors and to homogene and clear cells (Fig. 3L). In the same way, mouse Fgf10 has been detected in cochlear sensory cell precursors and supporting cells (Pauley et al.,2003). These results suggest a conserved role for Fgf10 signaling in the development of the cochlear sensory organ, although no defects were observed in Fgf10−/− cochleas (Pauley et al.,2003), suggesting that other Fgfs could compensate for Fgf10 in cochlear development.
Expression Analysis During Otic Ganglion Development
The neuroblasts forming the vestibulocochlear ganglion are specified already in otic placode in chicken (Adam et al.,1998) and in otic cup in mouse (reviewed in Rubel and Fritzsch,2002). The specified neuroblasts start to migrate out from otic epithelium during cup stage in chicken (Hemond and Morest,1991) and in early vesicle stage in mouse (reviewed in Fritzsch,2003). Although the vestibular and cochlear neuroblasts delaminate from slightly different regions of the ventral epithelium, they originally form a fused ganglion, which separates morphologically later (Lawoko-Kerali et al.,2004).
As shown above, Gata3 was originally expressed throughout the chicken otic placode, but its expression was rapidly restricted to the lateral-most domain of otic cup and vesicle, being only partially overlapping with the Fgf10-expressing neurogenic region (Fig. 1A–D,I–L; Alsina et al,2004). At HH19–HH21, Gata3 expression was detected in the anteroventral epithelium from which the neuroblasts migrate out (Fig. 1N,R). However, no expression in the migrating neuroblasts or in the ganglionic cells could be detected at this stage. This finding is in contrast to mouse, where Gata3 has been detected in a subset of delaminating neuroblasts at corresponding early developmental stages associating specifically with the auditory part of the developing ganglion (Karis et al.,2001; Lawoko-Kerali et al.,2004). In contrast to Gata3, Fgf10 could be detected already in migrating and/or ganglionic neuroblasts at HH21 (arrowhead in Fig. 1T). Also mouse Fgf10 is expressed in delaminating neuroblasts and in the newly formed ganglion (Pauley et al.,2003; Lilleväli et al.,2006).
The first signs of Gata3 in the otic ganglion were detected at HH24, when weak expression was observed in the cochlear part (Fig. 3A). Also Fgf10 was weakly expressed in the cochlear ganglion at this stage (Figs. 2C, 3C). Expression of both Gata3 and Fgf10 in cochlear ganglion increased considerably at HH26 (Fig. 3D,F), coinciding with the formation of postmitotic neurons and the time when most of the afferent fibers have reached their destination (Lang and Fekete,2001). Similar up-regulation of Gata3 has also been observed in mouse (Lilleväli et al.,2004). Both Gata3 and Fgf10 continued to be expressed strongly in the auditory ganglion at least until HH38 (data not shown). Targeted mutation of Gata3 results in a severely reduced cochlear ganglion (Karis et al.,2001), demonstrating an important function for Gata3 in auditory neurogenesis. Our observations suggest that this function could be conserved in chicken.
In mouse, no Gata2 expression could be detected in the anteroventral epithelium that gives rise to the ganglionic neuroblasts at E9.5–E10.5 or in the early ganglionic cells. Instead, Gata2 was expressed in the vestibular compartment of both the fused and divided ganglion (Lilleväli et al.,2004, and data not shown).
In the same way, the early expression of chicken Gata2 did not overlap with the Fgf10-expressing neurogenic region in early otic epithelium at HH14–HH17 (Fig. 1G,H,K,L, and data not shown). At HH21, Gata2 appeared in the anteroventral otic epithelium from where the neuroblasts migrate out as well as in a subset of cells in the ganglion (arrowhead in Fig. 1S) together with Fgf10 (arrowhead Fig. 1T). At subsequent stages, Gata2 was detected in the vestibular ganglion (Figs. 2F, 3E). Taken together, the expression pattern of all three genes (Gata3, Gata2 and Fgf10) in the otic ganglion was highly conserved between mouse and chicken, except for the late onset of Gata3 expression in the migrating/ganglionic cells in chicken and the early onset of Gata2. These differences may be due to temporal differences in delamination and migration of vestibular and cochlear precursors between mouse and chicken.
Fertilized chicken eggs were incubated at 38°C in humid environment until the embryos reached desired stages according to Hamburger and Hamilton (described in Bellairs and Osmond,1998).
RNA In Situ Hybridization
Nonradioactive whole-mount RNA in situ hybridizations were performed using digoxigenin-UTP (Roche) -labeled RNA probes (Wilkinson,1993). The stained embryos were blocked in gelatin and 25- to 30-μm sections were cut with a Vibratome. Radioactive in situ hybridizations were performed on 10-μm serial paraffin sections as reported previously (Salminen et al.,2000). The probes for Fgf10 and Bmp4 have been described before (Francis et al.,1994; Mandler and Neubüser,2004). Gata2, Gata3, and Pax2 probes were generated with reverse transcriptase-polymerase chain reaction (RT-PCR) from total RNA isolated from HH26 chicken embryos with Trizol (Life Technologies). Gata gene-specific primers were designed according to Sheng and Stern (1999). The primers for Pax2 RT-PCR were as follows: 5′-GCTTTCGATTGCTTTGCTTT-3′ and 5′-GAACATGGTGGGGTTTTGTC-3′. The generated fragments were cloned into pGEM-TEasy vector (Promega). Labeling of the antisense and sense probes was performed as described in Salminen et al. (2000). Red pseudocolor images were reproduced as previously described (Lilleväli et al.,2004).
We thank Tanja Matilainen for critical reading of the manuscript, Annette Neubüser for the Fgf10 probe, and Philippa Francis-West for the Bmp4 probe. We also thank Raija Savolainen and Sanna Seuna for expert technical assistance and Sophie Bel-Vialar for skillful advice with chicken embryos. M.S. was funded by the Finnish Academy, M.S. and M.H. were funded by CIMO and EU Marie Curie Early Stage Training, and F.P. and K.L. were funded by EU Marie Curie Industry Host.