Competence of cranial ectoderm to respond to Fgf signaling suggests a two-step model of otic placode induction

The paired craniofacial sensory organs that mediate smell, vision, hearing,balance and pain reception derive from ectodermal placodes adjacent to the anterior neural plate. Despite the importance of these vertebrate innovations,the molecular basis for their embryonic origins has received attention only recently and the mechanisms by which ectoderm is singled out to become a particular cranial placode are only just beginning to be understood(Baker and Bronner-Fraser,2001; Streit,2004). Several lines of evidence support the idea that craniofacial placodes may derive from a common, `pre-placodal' field during early development. First, morphological studies in a variety of vertebrates suggest that some, or sometimes all placodes can derive from a single epithelial thickening on the head (Braun,1996; Knouff,1935; Miyake et al.,1997; O'Rahilly and Müller, 1985; Platt,1896; van Oostrom and Verwoerd, 1972; Verwoerd and van Oostrom, 1979). Second, embryological experiments in which cranial ectoderm adjacent to the anterior neural plate was rotated along the anteroposterior axis demonstrated that nasal, lens and otic placodes are not determined in the young embryo, and that presumptive or `pre-placodal' tissue is competent to generate a variety of cranial placodes if transplanted at an appropriately early age (Jacobson,1963c; Jacobson,1966). Third, fate-mapping experiments in chick and zebrafish show that cells destined to give rise to different cranial placodes are intermingled in young embryos, suggesting that different placodes arise from common ectodermal domains (Bhattacharyya et al., 2004; Kozlowski et al.,1997; Streit,2002; Whitlock and Westerfield, 2000). Finally, the expression of a variety of genes in a border region around the neural plate from which cranial placodes derive has provided molecular evidence for the existence of a genetically distinct pre-placodal domain (reviewed by Groves,2005; Streit,2004).

The functional significance of a genetically distinct pre-placodal identity is not clear at present. For example, it is not known whether the competence to form cranial placodes is contingent upon the assumption of pre-placodal identity, or whether naïve ectoderm can differentiate into particular placodal derivatives without first expressing pre-placodal genes. It has been suggested that placode induction proceeds in a series of stages, in which an initial induction is required to instill some form of common placodal identity in cranial ectoderm, followed by subsequent local inductive interactions which single out individual placodes (Baker and Bronner-Fraser, 2001; Streit,2004). The advent of molecular markers has now rendered this question experimentally tractable, and in the present study we have used the induction of the otic placode as a test of this sequential model.

The otic placode forms adjacent to the posterior half of the hindbrain, and will ultimately give rise to the entire inner ear and the vestibulo-acoustic ganglion (Barald and Kelley,2004; Brown et al.,2003; Riley and Phillips,2003; Torres and Giraldez,1998; Whitfield et al.,2002). Studies in a variety of vertebrates demonstrate the primacy of Fgf signaling in the induction of the otic placode. Several Fgf family members are expressed in tissues adjacent to the presumptive otic placode in different species, such as Fgf3, Fgf4, Fgf8, Fgf10 and Fgf19(Karabagli et al., 2002; Kil et al., 2005; Ladher et al., 2000; Mahmood et al., 1995; Maves et al., 2002; Shamim and Mason, 1999; Wright et al., 2003). Some of these family members are necessary for placode induction; for example, Fgf3 and Fgf8, and Fgf3 and Fgf10 are necessary for placode induction in zebrafish and mouse respectively(Leger and Brand, 2002; Liu et al., 2003b; Maroon et al., 2002; Phillips et al., 2001; Solomon et al., 2004; Wright and Mansour, 2003). Mis-expression of Fgf family members can also induce ectopic placodes in amphibians and chick (Lombardo et al.,1998; Vendrell et al.,2000), although it has yet to be determined whether this is a direct action of Fgf on presumptive placodal ectoderm, as opposed to an indirect upregulation of other inducing signals. Fgf19 has also been proposed to induce otic markers in co-operation with Wnt8c in chick embryos(Ladher et al., 2000).

To determine the relationship between the existence of a pre-placodal domain and the induction of the otic placode by Fgf family members, we undertook an analysis of Fgf-mediated induction of the otic placode in chick embryos. In the present study, we show that Fgf signaling is both necessary and sufficient for the induction of some, but not all markers of the otic placode, and that Fgf acts directly on presumptive placodal ectoderm to induce placodal markers. In a series of grafting experiments, we demonstrated that the competence of embryonic ectoderm to express otic placode markers in response to Fgf signaling correlates with the expression of genes that define the pre-placodal region. This work thus provides the first functional evidence for a sequential, two-step model of placode induction, and suggests a molecular basis for the competence of ectoderm to give rise to the otic placode.

Fertilized chicken and quail eggs were obtained from a local commercial supplier (AA Labs, Westminter, CA) and incubated at 37.8°C in an humidified atmosphere. Chick embryos were staged according to the number of somites, or by the Hamburger and Hamilton (HH) staging system(Hamburger and Hamilton,1992).

Head chunks comprising the prospective otic ectoderm were dissected from the 0- to four-somite stage (HH stage 6 to stage 8) and five- to eight-somite stage embryos (HH stage 9) in Ringer's solution using 30-gauge hypodermic needles. Tissues were stored in complete medium on ice (10% fetal bovine serum in L15 medium) until required, then rinsed three times in Ringer's before transplanting into collagen gels. To isolate ectodermal tissue from 0- to four-somite stage chicks, embryos were treated with 0.1 mg/ml dispase in DMEM/F12 medium for 15 minutes on ice and 5 minutes at 37°C. The embryos were allowed to recover for at least 10 minutes on ice in complete medium (10%fetal bovine serum in L15 medium) and were then transferred to Ringer's solution. Using a 30-gauge hypodermic needle, ectodermal tissue was isolated either from prospective otic ectoderm, prospective trigeminal ectoderm or lateral tissue from the level of prospective trigeminal ectoderm. The dissected tissues were kept in complete medium on ice until required. Tissues were rinsed three times in Ringer's solution prior to transfer into collagen gels. Embryonic anterior epiblast was isolated from HH stage 3+ to 4 quail and chick embryos by transferring yolks to PBS and cutting the yolk circumferentially slightly above the equator, peeling the embryo from the yolk with forceps, and peeling the blastoderm off the vitelline membrane. Blastoderms were pooled in Ringer's solution containing 5 mM EDTA for 5 minutes at room temperature to enable easier dissection. An area comprising the anterior 25% of the region between Hensen's node and the anterior extent of the zona pellucida was isolated with 30-gauge hypodermic needles, and the epiblast freed from underlying tissue. Epiblasts were stored in Ringer's solution on ice until required, and then cut to appropriately sized pieces immediately before transfer.

Chick embryos were cultured using the filter paper carrier method developed by Chapman and colleagues (Chapman et al.,2001). Eggs were cracked into a glass Petri dish and the thick albumen covering the blastoderm removed using a piece of folded tissue paper. A filter paper with a central aperture of ∼0.5 inches was placed on top of the blastoderm. The vitelline membrane was cut around the paper and the embryo and filter paper carefully lifted from the yolk using forceps and pooled in Ringer's solution. The blastoderm was peeled off the vitelline membrane which was placed on a 35 mm agar albumen plate(Chapman et al., 2001) still attached to the filter paper. The area opaca was cut off the embryo and the embryo placed back on the vitelline membrane, dorsal side upwards. The graft site was prepared using a pulled glass needle. The graft was transferred from donor (quail) to host (chick) with a glass mouth pipette, and the graft positioned dorsal side upwards using a blunt pulled glass needle. The lid was placed on the agar albumen plate and immediately incubated at 37°C in a covered 150 mm Petri dish with moistened tissue paper lining the base. Some embryos were sacrificed after 4, 8 or 24 hours and fixed overnight at 4°C in 4% paraformaldehyde. Other embryos were transferred to Ringer's solution after 2 or 8 hours, and the graft removed with 30 gauge hypodermic needles and cultured in collagen gels for 12 hours. Grafts placed into chick hosts for up to 8 hours could be removed by mechanical dissection with little or no contamination of mesodermal or neural tissue, obviating the need for enzymatic treatment prior to dissection.

Collagen matrix gels were prepared as previously described(Groves and Bronner-Fraser,2000). Briefly, 900 μl of collagen solution, 100 μl of 10×MEM were combined. A few drops of 7.5% sodium bicarbonate were added to adjust the pH to 7.5. For head chunk cultures, 10 μl drops of the prepared collagen solution were plated and allowed to set. DMEM-BS medium (1 ml) [a modification of the chemically defined medium of Bottenstein and Sato(Bottenstein and Sato, 1979; Wolswijk and Noble, 1989] was added with either SU5402 or DMSO vehicle. Slits were cut in the gel, and the heads slipped in. Cultures were grown at 37°C and 5% CO₂ for 24 hours. Cultures were fixed in 4% paraformaldehyde overnight at 4°C. For ectoderm explants, 5 μl drops of the prepared collagen solution were plated and allowed to set. When set, 1 ml of DMEM-BS medium was added together with 50 ng/ml or 100 ng/ml Fgfs. The following Fgfs were used: Fgf1, Fgf2, Fgf4,Fgf5, Fgf6, Fgf7, Fgf8, Fgf9, Fgf10, Fgf16, Fgf17, Fgf18 and Fgf19. Explants were collected with a mouth pipette and placed in the collagen matrix. Each gel contained 6-10 explants. Cultures were grown at 37°C and 5%CO₂ for 4, 6, 7.5, 21 or 24 hours and fixed in 4% paraformaldehyde overnight at 4°C. Explants were processed for immunocytochemistry or dissected free of the gel for in situ hybridization.

Fixed embryos or fixed gels containing the cultured heads were equilibrated in PBS containing 30% sucrose and mounted and frozen in OCT. Sections (12μm) were collected on Superfrost Plus slides (Fisher) and stored at-20°C. Slides were washed twice with PBS containing 0.1% Triton X-100 and blocked in PBS containing 0.1% Triton X-100 and 5% goat serum for 2-3 hours. Primary antibodies were diluted in blocking buffer and applied overnight at 4°C. The slides were washed twice in PBS/0.1% Triton X-100 and secondary antibodies were diluted in blocking solution and applied for 45 minutes at room temperature. Slides were washed twice in PBS/0.1% Triton X-100 and incubated in DAPI solution for 10 minutes, then washed in PBS before being mounted in Fluoromount G (Southern Biotechnology). The following antibodies were used in this study: a rabbit polyclonal antibody to Pax2 (Zymed; 1:500);a rabbit polyclonal antibody to Epha4 (a gift from Elena Pasquale; 1:1000); a rabbit polyclonal antibody to pan-Distalless proteins (a gift from Grace Boekhoff-Falk and Jhumku Kohtz, 1:1000)(Panganiban et al., 1997); and Pax3 and QCPN IgG1 monoclonal antibodies (DSHB, University of Iowa, USA; 1:1). For immunocytochemistry against Dlx3, the slides containing the explants were blocked in PBS containing 10% donkey serum and incubated in 1:1500 Dlx3 goat polyclonal antibody (a gift from Sujata Bhattacharyya) overnight at 4°C. Secondary antibodies (Molecular Probes) were diluted in blocking buffer and applied for 2 hours at room temperature. Slides were washed four times in PBS between each incubation.

We performed whole-mount in situ hybridization and cRNA probe synthesis according to the protocol of Stern (Stern,1998) with modifications as described by Kil et al.(Kil et al., 2005). Whole-mount in situ hybridization and QCPN antibody staining was performed as described previously (Stern,1998). The probes used in this study were obtained from the following sources: chick Pax2 (Domingos Henrique), Sip1(Claudio Stern), Bmp7 (Brian Huston), Krox20 (David Wilkinson), Wnt8c (Jane Dodd), Sax1 and tailless (Kate Storey), Brachyury (Susan Mackem), Eya2 (Guillermo Oliver), and Zic2 (Cliff Tabin).

Fgf signaling has been implicated in induction of the chick otic placode(Ladher et al., 2000; Ladher et al., 2005; Represa et al., 1991; Vendrell et al., 2000). To determine whether Fgf signaling is necessary for otic placode induction, we isolated chunks of embryonic head tissue containing presumptive otic placode ectoderm, neural tube, mesoderm and endoderm from a region corresponding to the hindbrain level of 0- to eight-somite stage (ss) embryos. We cultured the head chunks in a chemically defined medium together with SU5402, an inhibitor of Fgf receptors (Mohammadi et al.,1997) and examined the expression of otic placode markers in the cultures after 24 hours by antibody staining or in situ hybridization(Fig. 1A). Previous work from our laboratory showed that the otic placode becomes specified with respect to the otic placode marker Pax2 from the 4 ss onwards(Groves and Bronner-Fraser,2000). Unspecified 0-4 ss explants treated with DMSO vehicle alone expressed Pax2, Epha4, Bmp7 and Dlx3(Table 1). However, blocking Fgf signaling with SU5402 resulted in the loss or great reduction of Pax2 and Epha4 expression in the otic ectoderm (Fig. 1B). Interestingly, expression of Bmp7 or Dlx3 was unaffected by SU5402 (Table 1). These results suggest that Fgf signaling is necessary for the induction of some, but not all otic placode markers. To test if Fgf signaling was required for the maintenance of otic placode markers after their specification, we isolated head chunks from older embryos (5-8 ss), in which the otic placode is already specified (Groves and Bronner-Fraser, 2000). In these older embryos, SU5402 treatment failed to block the expression of all otic placode markers tested(Table 1; Fig. 1B), suggesting that although Fgf signaling is necessary for the induction of some otic placode markers, it is not required for their maintenance.

Ectopic expression of Fgf family members by Fgf-coated beads or adenoviral transduction suggests a role for Fgf signaling in otic placode induction(Adamska et al., 2001; Lombardo et al., 1998; Vendrell et al., 2000). However, as these studies were performed in living embryos, it was not clear whether Fgf signaling alone was sufficient to induce some or all otic placode markers, or whether it was acting directly or indirectly. We therefore determined if Fgf signaling was sufficient to induce otic markers in the absence of other tissues. Presumptive ectoderm from the trigeminal level of 0-4 ss embryos was cultured for 24 hours in chemically defined medium containing a variety of different Fgf family members(Fig. 2A). Trigeminal level ectoderm (taken from adjacent to the presumptive midbrain)(Baker et al., 1999) was used in these experiments as it is highly competent to express otic placode markers(Groves and Bronner-Fraser,2000), but does not normally do so when cultured in vitro.

Pax2 was induced in 0-4 ss presumptive trigeminal ectoderm in the presence of between 5 and 50 ng/ml Fgf2 (Fig. 2A). To determine the time course of the Fgf response in ectoderm,we examined Pax2 expression in Fgf2-treated presumptive trigeminal ectoderm exposed to Fgf2 for 4, 6, 7.5 or 21 hours(Fig. 2B). No Pax2 expression was found after 4 hours (n=4). Pax2 expression appeared as early as 6 hours, after which time 60.6% of explants were positive for Pax2(n=33). The number of positive explants increased with time of exposure: 91% after 7.5 hours (n=11) and 100% after 21 hours(n=8). Explants cultured for 6 hours were then tested for the expression of other otic markers. Most of the explants expressed Epha4 (75%),whereas only a few expressed Dlx3 (30%) and none expressed Bmp7. These data along with those obtained after a 24 hour incubation(Table 2; Fig. 2C) suggest that Fgf2 is sufficient to induce and maintain Pax2, Epha4 and Dlx3 but not Bmp7. To test whether other Fgf family members were also sufficient to induce the otic placode, we screened for Pax2 expression in explants treated with Fgf1,Fgf4, Fgf5, Fgf6, Fgf7, Fgf8, Fgf9, Fgf10, Fgf16, Fgf17, Fgf18 and Fgf19. All Fgf family members tested apart from Fgf2 failed to induce Pax2 expression at 50 ng/ml, although Fgf1 and Fgf4 were able to induce Pax2 in 40%(n=10) and 37.5% (n=48) of explants at a concentration of 100 ng/ml. As Fgf2 has been reported to activate isoforms of all four Fgf receptors and gave robust induction of several otic markers, we used it in all subsequent assays in our study.

We routinely observed that only a subset of cells in our Fgf2-treated cultures expressed otic markers, and that these cells were always found together in a restricted patch. This observation, together with the fact that upregulation of otic markers took at least 6 hours raised the possibility that Fgf2 might be acting indirectly on the ectoderm in our experiments, possibly by first inducing mesodermal or neural tissue, which are reported to have otic inducing activity (Kil et al.,2005; Ladher et al.,2000; Ladher et al.,2005; Woo and Fraser,1998; Wright and Mansour,2003). To test this possibility, we assayed for the expression of neural and mesodermal markers in our Fgf2-treated cultures. We first examined the expression of the pan-neural markers Sip1 and Zic2. Sip1is first expressed in the chick neural plate at stage 4+ and remains throughout the neural plate at least until the 10 ss(Sheng et al., 2003) and Zic2 is expressed throughout the neural plate at similar stages (K.M. and A.K.G., unpublished). We did not observe expression of either gene in presumptive trigeminal explants treated with Fgf2 after 3, 6 or 24 hours(Fig. 3; Table 3). We also examined a panel of regionally expressed neural genes in our Fgf2-treated explants. Tailless is expressed in the forebrain (Yu et al., 1994), Krox20 in rhombomeres 3 and 5(Nieto et al., 1995), Wnt8c in rhombomere 4 and the caudal neural plate(Hume and Dodd, 1993; Kil et al., 2005), and Sax1 in the caudal neural plate(Spann et al., 1994). None of these regionally expressed markers was expressed in Fgf2-treated explants at any time point examined (Table 3; Fig. 3). Finally, none of our Fgf2-treated explants expressed the brachyury gene (a marker of axial and primitive streak mesoderm)(Liu et al., 2003a) at any time points examined (Fig. 3). Together, these results suggest that Fgf2 is sufficient to induce several otic placode markers in competent ectoderm without first inducing neural or mesodermal tissue.

Much of the early embryonic ectoderm is competent to form an inner ear when grafted to an appropriate location(Gallagher et al., 1996; Jacobson, 1963a; Jacobson, 1963b; Kaan, 1926; Yntema, 1933). In birds, young quail embryonic ectoderm can express otic placode markers, form an otocyst and generate vestibulo-acoustic neurons when grafted adjacent to the posterior hindbrain (Groves and Bronner-Fraser,2000). To test whether different populations of cranial ectoderm were equally competent to respond to Fgf2 in vitro, we isolated unspecified otic and trigeminal level ectoderm from 0-4 ss embryos, and anterior epiblast from HH stage 4 embryos and cultured the ectoderm explants in the presence of 50 ng/ml Fgf2 (Fig. 4A). Although both unspecified trigeminal and otic ectoderm expressed Pax2 in response to Fgf2 (73.8% and 80% explants, respectively; n=84 and 45),no anterior epiblast explants expressed Pax2 in the presence of Fgf2 (0%, n=40; Fig. 4B,C). A small number of explants of presumptive otic and presumptive trigeminal ectoderm expressed Pax2 in the absence of Fgf2. This is probably due to some of the otic explants being already specified at the time of isolation(Groves and Bronner-Fraser,2000), and possible contamination of some trigeminal explants with otic tissue or Pax2-expressing midbrain tissue. As both the trigeminal and otic level ectoderm explants were taken from within the domain of pre-placodal gene expression lying adjacent to the neural plate(McLarren et al., 2003; Streit, 2002), we tested whether ectoderm from this pre-placodal region was uniquely competent to respond to Fgf2. To do this, we again isolated presumptive trigeminal level ectoderm from stage 0-4 ss embryos, but this time took the ectoderm from a position lateral to the neural plate, outside the pre-placodal region defined by genes such as Dlx5, Dlx6, Eya2 and Six1(McLarren et al., 2003; Streit, 2002). No explants taken from regions lateral to presumptive trigeminal ectoderm expressed Pax2 following Fgf2 treatment (0/13; Fig. 4B,C).

The above results show that two populations of ectoderm taken from within the pre-placodal domain (presumptive otic and trigeminal) were competent to express otic markers in response to Fgf2, whereas two populations of ectoderm from outside the pre-placodal domain (anterior epiblast and lateral trigeminal level ectoderm) were not competent to respond to Fgf2. These results were surprising, as we had previously shown that all four populations of ectoderm were competent to express otic placode markers when grafted adjacent to the posterior hindbrain in vivo (Groves and Bronner-Fraser, 2000) (A.K.G., unpublished). One explanation for these results is that both presumptive otic and trigeminal ectoderm fall within a region in which pre-placodal genes (such as Dlx5, Dlx6, Eya2and Six1) are expressed, whereas lateral ectoderm and anterior epiblast do not (Fig. 4D). We hypothesized that naïve ectoderm such as anterior epiblast must first assume a pre-placodal cell identity in order for it to be competent to respond subsequently to Fgf2 by expressing otic placode markers. To test this, we grafted anterior epiblast tissue from stage 4 quail embryos adjacent to the neural plate of 0-4 ss chick hosts at the level of the presumptive trigeminal ganglion. This region lies within the pre-placodal region as defined by the expression of genes such as Dlx5, Dlx6, Eya2 and Six1(McLarren et al., 2003; Streit, 2002). We fixed the grafted embryos after either 4 or 8 hours and examined the QCPN-expressing quail grafts for expression of Eya2 by in situ hybridization, or Dlx genes by immunostaining with a pan Distalless antibody (pan-Dll)(Panganiban et al., 1997). Stage 4 anterior epiblast tissue did not express Eya2 or pan-Dll at the time of isolation (data not shown). When embryos were examined 4 hours after receiving grafts, only a small number of grafts expressed pre-placodal markers (pan-Dll: 20%, n=10; Eya2: 8.7%, n=23; Table 4; Fig. 5). However, after 8 hours, the majority of quail grafts expressed both markers(Table 4; Fig. 5). Grafts of epiblast to the trigeminal region will express the trigeminal marker Pax3 after 24 hours(Baker et al., 1999); however,none of the grafts expressed Pax3 protein at 4 or 8 hours after grafting(Table 4 and data not shown),suggesting that more than 8 hours is required to induce this marker. In contrast to grafts made into the pre-placodal region, grafts placed lateral to the pre-placodal region at the level of the trigeminal ganglion did not label with pan-Dll or probes to Eya2 after either 4 or 8 hours(Fig. 5). We also grafted epiblast to the most anterior region of the pre-placodal region (at the level of the future nasal and lens placodes; Fig. 6A) and also adjacent to the neural plate posterior the pre-placodal region (in the trunk at the level of somites 5-7; Fig. 6A). After 8 hours, grafts adjacent to the anterior neural plate expressed both Eya2 and labeled with pan-Dll, but grafts to the trunk did not(Fig. 6A). We have previously shown that young trunk ectoderm (0-6 ss) is competent to give rise to the otic placode after grafting, but that older ectoderm (7 ss and older) is not(Groves and Bronner-Fraser,2000). To see whether this differential response correlates with the ability to upregulate pre-placodal genes, we grafted old (7-9 ss) or young(0-2 ss) trunk ectoderm to the level of the presumptive otic placode, and examined the expression of Eya2 and labeling with pan-Dll after 8 hours (Fig. 6B). Young trunk ectoderm expressed both markers after grafting (pan-Dll: 62%, n=26;Eya2: 71%, n=7). Older trunk ectoderm grafts did not express Eya2(0%, n=15), and fewer old grafts expressed pan-Dll (30%, n=20).

Our results suggest that (1) naïve ectoderm can up-regulate pre-placodal genes after 4-8 hours when grafted into the pre-placodal region,but not when grafted lateral or posterior to this region; and (2) that the competence to respond to placode-inducing signals may correlate with the expression of pre-placodal genes. To test this hypothesis, we grafted HH stage 4 quail anterior epiblast into 0-4 ss chick hosts at the level of the presumptive trigeminal placode, either within the pre-placodal domain(adjacent to the neural plate), or outside the pre-placodal domain (far lateral from the neural plate; Fig. 7A). We then removed the grafts at either 2 hours of incubation(prior to the upregulation of pre-placodal genes) or at 8 hours of incubation(after the upregulation of pre-placodal genes). Isolated 2 hour and 8 hour grafts were then cultured in collagen gels for a further 12 hours in the presence or absence of 50 ng/ml Fgf2 and examined for Pax2 expression(Fig. 7B; Table 5). No grafts in any experimental group expressed Pax2 in the absence of Fgf2. Two hour or 8 hour grafts outside the pre-placodal domain, or 2 hour grafts inside the pre-placodal domain were also unresponsive to Fgf2(Fig. 7B; Table 5). However, epiblast that was grafted into the pre-placodal domain for 8 hours expressed Pax2 when subsequently cultured in the presence of Fgf2 (60%, n=25; Fig. 7B; Table 5). To confirm that grafts became responsive to Fgf2 when grafted into other locations within the pre-placodal region, we placed some grafts in the most anterior part of the pre-placodal region at the level of the presumptive lens and nasal placodes. Grafts placed in this region for 8 hours and then cultured in Fgf2 for 12 hours also upregulated Pax2 (55%, n=22 grafts in Fgf2 compared to 0%, n=20 grafts in the absence of Fgf2). To show that Fgf2 was not sufficient to upregulate pre-placodal genes in epiblast, we cultured isolated epiblast in Fgf2 for 20 hours without first grafting into chick embryos(Fig. 7A,C). No explants labeled with pan-Dll antibodies or an Eya2 probe under these conditions (Fig. 7C). These results show that competence of cranial ectoderm to respond to Fgf2 correlates with the expression of pre-placodal genes.

The existence of a molecularly distinct domain that gives rise to all cranial placodes is now well established. Genes that characterize this domain,including Six1, Six4, Eya2, Dlx5 and Dlx6, have been shown recently to be regulated by signals from the neural plate and underlying mesoderm, such as Fgf, Wnt and Bmp family members(Litsiou et al., 2005; McLarren et al., 2003). However, the significance of the pre-placodal region is still uncertain; for example, it is not clear whether the assumption of a pre-placodal cell state is necessary for the subsequent differentiation of particular placodes, and some authors have questioned the assumption that a pre-placodal state exists(Begbie and Graham, 2001). Moreover, the function of genes expressed in the pre-placodal region is only starting to be understood. For example, overexpression of Six1expands the pre-placodal region in Xenopus embryos, whereas loss of function expands neural crest and epidermal domains at the expense of pre-placodal tissue (Brugmann et al.,2004). Similar results have also been reported with a Xenopus Iroquois homologue, Xiro1(Glavic et al., 2004). In this study, we have shown that competence of chick cranial ectoderm to respond to otic placode-inducing signals correlates with the pre-placodal cell state. This suggests that induction of cranial placodes occurs in two steps - an initial step inducing pre-placodal cell identity adjacent to the anterior neural plate, and a subsequent set of steps in which particular placodes are induced by local signals.

In the present study we show that Fgf signaling is both necessary and sufficient for the induction of some, but significantly not all, markers of the otic placode. Fgf signaling was both necessary and sufficient for the induction of Pax2 and Epha4, sufficient but not necessary for the induction of Dlx3 and neither necessary nor sufficient for the induction of Bmp7. This suggests that some otic placode-specific genes such as Dlx3 may be induced by redundant factors in addition to Fgfs, and that other genes such as Bmp7 are induced entirely independently of Fgf signaling. Other studies have also suggested that not all otic markers require Fgf signaling for their induction. For example, one study in zebrafish suggested that Fgf signaling is not necessary for the induction of the otic marker Pax8(Maroon et al., 2002),although a second study performed in an apparently similar fashion came to opposite conclusions (Leger and Brand,2002). Other zebrafish studies also demonstrated that foxi1,dlx3b, dlx4 and sox9b can be induced in the absence of Fgf3 and Fgf8 (Liu et al., 2003b; Solomon et al.,2004), and that Fgf signaling is not sufficient to induce foxi1 (Phillips et al.,2004). These data suggest that at least two signaling pathways -both Fgf-dependent and Fgf-independent - converge to induce the inner ear.

Our data suggest that the inducing action of Fgf2 on unspecified ectoderm is likely to be direct, rather than indirect. First, the induction of Pax2 protein by Fgf2 is reasonably rapid, being observed after 6 hours. Second, we were unable to detect expression of the early mesodermal marker brachyury, or the pan-neural markers Sip1 and Zic2in our explants after 3, 6 or 24 hours exposure to Fgf2. We were also unable to detect regional markers of the nervous system in our explants. Together,these results suggest that Fgf2 is capable of inducing otic placode markers without first inducing either neuronal or mesodermal tissue, both of which have been suggested to have otic placode-inducing potential(Groves, 2005). It is formally possible that Fgf2 is acting exclusively on ectoderm to induce a second inducing molecule without inducing neural or mesodermal tissue. Experiments to demonstrate cell-intrinsic otic placode induction by Fgfs using activated Fgf receptors, and to establish the Fgf signaling pathways involved in this induction are currently in progress.

Our results show that ectoderm taken from a region adjacent to the border of the anterior neural plate is qualitatively different in its response to Fgf2 than ectoderm taken from more lateral regions. Since this region of Fgf-responsive ectoderm is characterized by the expression of a battery of`pre-placodal' genes, we sought to demonstrate a correlation between Fgf responsiveness and pre-placodal cell identity. We find that a number of pre-placodal genes such as Dlx genes and Eya2 are upregulated in anterior epiblast tissue between 4-8 hours after grafting to the pre-placodal region at the level of the otic, trigeminal or lens/nasal placodes. Fgf responsiveness correlates with the upregulation of these pre-placodal genes, as epiblast grafted into the pre-placodal region for 8 hours and then treated with Fgf2 expresses otic markers, while epiblast grafted into the pre-placodal region for only 2 hours does not. Our results suggest that adoption of a pre-placodal cell state is a necessary prerequisite to respond to otic placode-inducing signals.

We have shown previously that many regions of cranial ectoderm, and epiblast from the gastrulating chick embryo are competent to form an otic placode when grafted adjacent to the posterior hindbrain(Groves and Bronner-Fraser,2000). However, our present results suggest that some populations of competent ectoderm - those that lie within the pre-placodal region - can express otic placode markers in the presence of Fgf2, while other populations of competent ectoderm - those do not express pre-placodal region genes -cannot respond to Fgf2. In order to reconcile these results, we propose a two-step, sequential model of otic placode induction. In the first step,naïve ectoderm is induced to form the pre-placodal region. Recent work by Streit and colleagues suggests that these signals are derived from cranial mesoderm progenitors and the neural plate, and include both Wnt and BMP antagonists, together with activation of the Fgf pathway(Litsiou et al., 2005). In a second step, cells within the pre-placodal region are induced to form particular placodes by local inducing signals, which in the case of the otic placode include Fgfs. Under this two step model, competence to form an otic placode (see Gallagher et al.,1996; Groves and Bronner-Fraser, 2000) is resolved into two sequential competencies. Naïve ectoderm from many regions of the embryo is competent to respond to pre-placodal inducing signals (such as those described in Litsiou et al., 2005) by upregulating pre-placodal genes, but is not competent to express otic markers in the presence of Fgf2. Naïve ectoderm only becomes competent to respond to otic placode-inducing signals such as Fgfs when it has first adopted a pre-placodal identity. Our model predicts that the induction of other cranial sensory placodes will also occur in a sequential fashion, with induction of pre-placodal genes being followed by induction of a particular placode in response to local signals. We have no evidence at present to suggest that our model can be generalized to other placodes, although the identification of candidate inducing tissues or molecules for other placodes(Baker et al., 1999; Begbie et al., 1999) suggests tests of this hypothesis.

At present, we do not know which genes in chick are necessary or sufficient to confer competence on naive ectoderm to respond to pre-placodal region inducing signals, or which pre-placodal region genes are necessary or sufficient to confer Fgf responsiveness on pre-placodal region ectoderm. Although Six1 is necessary for placode development and upregulation of other pre-placodal genes in Xenopus(Brugmann et al., 2004), no single transcription factor can confer pre-placodal properties on naïve ectoderm in chick embryos (Andrea Streit, personal communication). In zebrafish, the dlx3b gene is expressed in the pre-placodal region(and later upregulated in the otic placode) and appears necessary for the induction of Pax2a in the presence of Fgf signaling(Liu et al., 2003b; Solomon and Fritz, 2002). Dlx5 and Dlx6 are expressed in chick and mice in a similar pattern to Dlx3b in zebrafish(Brown et al., 2005; McLarren et al., 2003; Ohyama and Groves, 2004), but perturbation of these genes in chick and mice does not appear to affect induction of the otic placode (McLarren et al., 2003; Robledo et al.,2002). It is also possible that additional competence factors expressed in a restricted region of the pre-placodal region may be required to induce particular placodes. For example, the zebrafish foxi1 gene is expressed initially in a domain broader than the otic placode and is necessary for the induction of Pax8 in the presence of Fgf signaling(Hans et al., 2004; Nissen et al., 2003; Solomon et al., 2003), but is not itself induced by Fgf signals. Although loss-of-function experiments can reveal candidate competence factors, further gain-of-function experiments are required to establish whether any of the pre-placodal genes or foxi1are sufficient to confer competence on ectoderm to respond to placode inducing signals such as Fgfs.

The work of Grainger and colleagues has identified a stage in lens induction which they refer to as `lens-forming bias'(Grainger et al., 1997; Henry and Grainger, 1987)(reviewed by Grainger, 1992; Grainger, 1996). This concept arose from the observation that presumptive lens ectoderm from embryos of progressively older ages shows an increasing response to the very weak lens-inducing activity of the optic vesicle(Grainger et al., 1988; Henry and Grainger, 1987). Expression of the transcription factors Otx2 and Pax6correlate with the presence of lens-forming bias in ectoderm(Zygar et al., 1998), although this ectoderm probably contains progenitors of both lens and olfactory placodes (Bhattacharyya et al.,2004), both of which require Pax6 for their correct development(Grindley et al., 1995; Quinn et al., 1996). As the molecular basis for lens-forming bias is presently unknown, it is not clear if`bias' sensu Grainger is a different property from the pre-placodal state that we now show correlates with Fgf responsiveness. Pre-placodal ectoderm is characterized by the expression of genes such as Six1, Six4, Eya2, Dlx5,Dlx6 and Ern1 in a region surrounding the entire anterior neural plate, whereas lens-forming bias correlates with expression of Pax6and Otx2, which are expressed in the most anterior border of the neural plate or the neural plate proper, respectively(Bhattacharyya et al., 2004; Chapman et al., 2002; Fernandez-Garre et al., 2002). Lens-forming bias can be observed in presumptive lens ectoderm from the neural plate stage onwards, but is only observed in ectoderm surrounding the presumptive lens ectoderm at neural tube stages(Grainger et al., 1997). By contrast, pre-placodal markers such as Six1 are already expressed at the border of the neural plate shortly after gastrulation in amphibians and birds (Brugmann et al., 2004; Litsiou et al., 2005; McLarren et al., 2003; Pandur and Moody, 2000; Streit, 2002). Finally, we show that the upregulation of pre-placodal genes correlates with Fgf-unresponsive ectoderm becoming responsive, whereas the acquisition of bias appears to represent a more graded response to lens-inducing signals(Grainger et al., 1997; Henry and Grainger, 1987). A more detailed molecular description of biased lens ectoderm and pre-placodal ectoderm is required to understand the differences and similarities between these two precursor populations.

The most important point to be stressed from our work is how the concept of competence to form a particular tissue can change in the light of new molecular data. Our sequential model of otic placode induction suggests that competence to respond to pre-placodal gene inducing signals is both molecularly and functionally distinct from the competence to express otic placode genes in response to Fgf2. Changes in competence to respond to inducing or survival signals can occur by simply changing receptor expression(Birren et al., 1993), or at the other extreme by changing chromatin structure(Song and Ghosh, 2004). The next challenge in the study of cranial placode induction will be to understand the molecular basis of these two different competent cell states.

We thank Xiomara Padilla and Juemei Wang for excellent technical assistance and Domingos Henrique, Claudio Stern, Brian Huston, David Wilkinson, Jane Dodd, Kate Storey, Susan Mackem, Guillermo Oliver and Cliff Tabin for probes. This work was supported by the House Ear Institute, by an NIH grant (DC04675)and by a March of Dimes Basil O'Connor Starter Scholar award to A.K.G. We thank Andrea Streit for many helpful discussions, and for sharing results from her laboratory with us prior to publication.