The avian inner ear contains seven vestibular organs and one auditory organ. Each sensory organ contains mechanosensory hair cells that are innervated by the peripheral processes of statoacoustic ganglion (SAG) neurons. During development, SAG axons are presumed to be guided to their sensory targets by axon guidance cues present along their trajectory (Webber and Raz,2006; Fekete and Campero,2007; Appler and Goodrich,2011), and their survival requires interactions with their targets (Fritzsch,2003; Yang et al.,2011). Classic axon guidance molecules such as the semaphorins (Gu et al.,2003) and ephrins (Bianchi and Gray,2002), as well as neurotrophins (Tessarollo et al.,2004; Fritzsch et al.,2005) are involved. However, members of the bone morphogenetic protein (BMP), sonic hedgehog (Shh), Wnt, and fibroblast growth factor (FGF) families are also expressed within and surrounding the inner ear during stages of SAG axon outgrowth and pathfinding, in both birds (Oh et al.,1996; Wu and Oh,1996; Farkas et al.,2003; Alsina et al.,2004; Chang et al.,2004; Sanchez-Calderon et al.,2004; Bok et al.,2005; Sanchez-Calderon et al.,2007) and mammals (Morsli et al.,1998; Pirvola et al.,2002; Riccomagno etal.,2002; Pauley et al.,2003; Hayashi et al.,2007; Jacques et al.,2007; Driver et al.,2008; Jahan etal.,2010; Liu et al.,2010; Ohyama et al.,2010). The expression pattern of these molecules makes them candidate chemotropic factors for axons that are navigating toward their sensory targets and candidate chemotrophic factors for axons that have reached their targets.
Morphogens are secreted proteins that exist as concentration gradients and exert concentration-dependent effects on target cells at a distance (Charron and Tessier-Lavigne,2005). They initially participate in tissue patterning during development by specifying cell fate and then serve as axon guidance cues (Zou and Lyuksyutova,2007). For example, BMP and Shh are expressed in opposing gradients in the neural tube to specify dorsal and ventral cell fates, respectively. BMP7 is expressed in the roof plate to repel commissural axons ventrally toward the floor plate (Augsburger et al.,1999) and Shh, expressed in the floor plate, attracts commissural axons ventrally (Charron et al.,2003). To date, BMPs are considered repellents, with no reports of them being attractive (Zou and Lyuksyutova,2007). Shh serves as an attractant in some systems and serves as a repellent in others (Sanchez-Camacho and Bovolenta,2009). Due to the diverse roles that morphogens play during development, how a particular ligand influences developing neurons appears to be very context and system dependent.
We have recently shown that SAG neurons are unresponsive to Wnts-1, -4, -5a, -6, and -7b in vitro, and Wnts-4, and -5a in vivo (Fantetti et al.,2011). Therefore, we have focused on testing the responses of SAG neurons to other morphogens that are known to be present during ear development. Several morphogens (BMPs, FGFs, and Shh) promoted SAG neuron survival in a collagen gel assay and all except BMP7 promoted SAG neurite outgrowth in a bead assay. The results from this study suggest that SAG neurons may rely on morphogens as survival factors and possible guidance cues, during innervation of the auditory and vestibular systems.
Explant Dissection and Culture
Fertilized White Leghorn chicken eggs (Purdue University Farm) were incubated at 38°C and staged (Hamburger and Hamilton,1951) before dissection. HH20–25 SAG and HH24–25 spinal cord explants were dissected and cultured as described (Bianchi and Cohan,1993; Fantetti etal.,2011; Fantetti and Fekete, 2011). Explants were cultured in 1.5 mg/mL rat-tail Type I collagen (BD Biosciences) gels for 24 h, under serum-free conditions, supplemented with 10 ng/mL NT-3 (Sigma), 10 ng/mL CNTF (Sigma), and 10% ITS+ (Sigma) to enhance survival. To visualize neuron cell bodies and neurites, explants were fixed and stained with monoclonal anti-β-tubulin (Sigma) as described (Fantetti et al.,2011).
Recombinant proteins were purchased from R&D Systems. Purified human BMP4 (10–100 ng/mL), BMP7 (0.25–1.0 μg/mL), Shh-N (0.5–20 μg/mL), FGF2 (1–20 ng/mL), FGF10 (50–200 ng/mL), FGF19 (50–200ng/mL), and mouse FGF8 (50–500 ng/mL) were diluted in phosphate-buffered saline (PBS) and added to culture medium. PBS was added to control cultures. For inhibitor experiments, 1 μg/mL mouse Noggin (R&D Systems), 2 μg/mL mouse Follistatin (R&D Systems), 10 μM Cyclopamine (Calbiochem), or 0.1 μM SU5402 (Tocris Bioscience) was added to culture medium.
Beads were prepared for co-cultures as described (Fantetti and Fekete, 2011). In brief, Affi-gel Blue gel beads (BioRad #153-7302) were soaked for 1–2 h at room temperature in either PBS (control), 10 μg/mL BMP4, 10 μg/mL BMP7, 2.5 μg/mL Shh-N, or 500 ng/mL FGF-8, -10, -19. Beads were positioned 50–500 μm from the edge of explants before collagen polymerization. Standard serum-free culture medium that is used to maintain SAG explants was added to cultures.
Quantitative Analysis of Neurite Outgrowth
Immunostained explant cultures were imaged with a 10× objective on a Nikon C1-plus confocal microscope. Cultured ganglia tend to flatten against the bottom of the culture dish to assume dimensions that are typically 200–400 μm wide by 100–150 μm deep. Furthermore, when imaged from the bottom of the gel, axons projecting vertically into the gel are often too deep to image. Thus, the imaged axons and corresponding measurements are biased in favor of those growing parallel to the bottom of the gels. Image stacks were collapsed into 2D projections and quantified using NIH ImageJ software. Pixel measurements were performed to quantitatively determine neurite responsiveness. For each image, the threshold was adjusted to convert the image into black (neurites) and white (background). The explant was excluded from the image by outlining the explant and cropping the image. The number of pixels occupied by neurites (black) was quantified and normalized to the length of the explant, to control for variation in explant size. SAG neurite length (not shown) and spinal cord neurite outgrowth measurements were performed as described (Fantetti et al.,2011).
For purified protein experiments, effects on neurite outgrowth were determined by quantifying pixels for the entire explant and the explant circumference was used as the length. For bead co-cultures, effects on neurite outgrowth were evaluated by comparing neurite outgrowth facing the bead with outgrowth opposing the bead. Images of co-cultures were divided into four quadrants. Neurite outgrowth (in pixels/explant length) for the proximal (facing the bead) and distal (opposing bead) quadrants was quantified. Explant length was measured along the explant edge between the boundaries of the measured quadrant. The pixel ratio of proximal/distal was calculated for each explant. The average of the normalized values was calculated for each treatment group. A ratio equal to 1 was obtained for control cultures and treated cultures with no effect on neurite outgrowth, because there was no difference between the proximal (P) and distal (D) quadrants. A value greater than 1 reflects more neurite outgrowth in the proximal quadrant (neurite promotion), and a value of less than 1 reflects less neurite outgrowth in the proximal quadrant.
Terminal dUTP Nick-End Labeling and Cell Death Quantification
After confocal imaging, collagen gels were processed again to evaluate cell death. A similar protocol has been described elsewhere (Fantetti et al.,2011). Gels were incubated overnight in 30% sucrose in PBS, embedded in tissue freezing medium (Triangle Biomedical Sciences), and cryosectioned at 15 μm. Two series of sections were collected to perform terminal dUTP nick-end labeling (TUNEL) labeling on alternate sections. To detect apoptotic cells, TUNEL was performed using a TUNEL kit (In situ Cell Death Detection kit, TMR Red; Roche Applied Science). The manufacturer's protocol was followed with one modification: enzyme and label solutions were diluted 1:1 with TUNEL dilution buffer (Roche Applied Science). Sections incubated with dUTP (label solution) alone or DNase I for 10 min before TdT were used as negative and positive controls, respectively. Cell nuclei were counterstained with DRAQ5 (Cell Signaling Technology), diluted 1:1000 in PBS. All sections were imaged with 40× objective on a Nikon-C1 confocal microscope. Manual cell counts were performed from 2D projections of the two middle sections of each explant [Fig. 1(A–C)], using NIH ImageJ software. The percentage of TUNEL-positive cells was calculated by dividing the number of TUNEL/DRAQ5 double-positive cells by the total number of DRAQ5 labeled cells.
Phosphohistone H3 Immunohistochemistry and Cell Proliferation Quantification
Immunostaining was performed on cryosections alternate to sections that were TUNEL labeled. Sections were fixed in 4% paraformaldehyde for 20 min at room temperature, rinsed in PBS, and incubated in blocking solution (10% calf serum, 0.1% TX-100, 0.1% sodium azide in PBS) for 1 h at room temperature. To visualize proliferating cells, sections were incubated for 1 h at room temperature with rabbit polyclonal PhosphoDetect Anti-Histone H3 (pH3; Calbiochem) diluted 1:200 in blocking solution, rinsed in PBS, and incubated for 1 h at room temperature with Alexa fluor 568 goat anti-rabbit (Molecular Probes) diluted 1:500 in blocking solution. After PBS rinses, sections were counterstained with Hoechst33258 (Invitrogen), diluted 1:10,000 in PBS, for 10 min at room temperature. Sections were rinsed for 5 min in PBS and coverslipped with VECTASHIELD HardSet (Vector Laboratories). Sections were imaged with a 20× objective on a Nikon Eclipse E800 microscope and quantified the same day as staining. Manual cell counts were performed from the three middle sections (alternate to TUNEL sections) of each explant. The percentage of proliferating cells was calculated by dividing the number of pH3/Hoescht double-positive cells by the total number of Hoescht labeled cells. See Figure 1 for representative staining.
DF-1 Survival Assay for Bioactivity of Purified Human FGF2
UMNSAH/DF-1 chicken embryo fibroblasts (DF-1 cells; ATCC) were seeded into collagen gels at 6.5 × 105 cells per well and cultured for 24 h with SAG culture medium supplemented with PBS (control) or 1–10 ng/mL FGF2. Gels were immunostained as described for explant cultures (Fantetti et al.,2011). Apoptotic cells were labeled with rabbit polyclonal anti-caspase-3 (R&D Systems) and all cells were counterstained with DRAQ5 (Cell Signaling Technology). Images were taken with a confocal microscope and collapsed into 2D projections. The number of apoptotic cells (caspase-3/DRAQ5 double-positive) and DRAQ5-positive cells were manually counted, and the percentage of apoptotic cells was calculated by dividing the number of double-positive cells by the total number of cells.
Measurements are given in mean ± standard error (SE). Means represent data collected from two to three independent experiments for each condition. To determine statistically significant effects, a one-way analysis of variance (ANOVA; followed by a Tukey's multiple comparisons test) or a student's t-test was performed using GraphPad Software (GraphPad Software, Inc.). Results were considered statistically significant with a p value of less than 0.05.
BMP4 Promotes SAG Neurite Outgrowth and Neuron Survival
Bmp4 transcripts are expressed by prosensory regions of the inner ear [Fig. 2(A); Wu and Oh,1996] that will become innervated, suggesting that SAG neurons may be BMP4-responsive. After 24 h in culture, SAG explants displayed longer and denser neurite outgrowth in the presence of BMP4 [Fig. 2(C)] compared with controls [Fig. 2(B)]. Image analysis revealed that 10–100 ng/mL BMP4 treatment produced average pixel numbers that were significantly greater than controls [Fig. 2(H); ANOVA, p < 0.0001]. Noggin inhibits BMP signaling by binding with high affinity to BMP4 ligands (Zimmerman et al.,1996). When explants were cultured in the presence of Noggin alone or Noggin and BMP4, neurite outgrowth resembled controls [Fig. 2(E,F,H)]. These results suggested that BMP4 promotes SAG neurite outgrowth.
An enhancement of neurite outgrowth could result from increased cell numbers due to trophic effects on proliferation or survival rather than, or in addition to, any neurotropic effects. To evaluate effects on cell survival, TUNEL assays were performed to detect apoptotic cells with double-stranded DNA breaks (Gavrieli et al.,1992). If BMP4 promoted SAG neuron survival, we expected there to be fewer TUNEL-positive cells for BMP4-treated explants than for untreated explants. There were significantly fewer TUNEL-positive neurons in the presence of 50 ng/mL BMP4 compared with controls [Fig. 2(I); t-test, p < 0.0001]. Proliferating cells were immunolabeled with anti-phosphohistone H3 (pH3). Under these culture conditions, most SAG neurons are postmitotic on E4–5, as shown by very low percentages of pH3/Hoescht double-positive cells. BMP4 did not affect cell proliferation [Fig. 2(J); t-test, p = 0.4928]. From these results, we concluded that BMP4 promotes the survival of SAG neurons under these culture conditions.
Bath application of proteins does not create a morphogen point source, which is necessary to assay for directional effects on neurites. Placing a bead at a distance may permit proteins to diffuse from the beads and may create a protein gradient through the collagen gel. To test the possibility that BMP4 may serve as a directional cue for growing neurites, in addition to promoting neuron survival, we tested the responses of SAG neurites to a point source of BMP4. Beads were incubated in PBS [Fig. 2(D); control] or 10 μg/mL BMP4 [Fig. 2(G)] and placed into collagen gels so that they were usually a few hundred microns from the explant edge. If BMP4 was neurite-promoting or attractive, we expected neurites to preferentially grow toward a BMP4-soaked bead. After 24 h in culture, SAG neurites were short and did not orient toward or away from control-treated beads [Fig. 2(D)]. On the other hand, neurites were denser and longer in the presence of BMP4-treated beads and displayed more robust outgrowth on the side of the explant facing the bead [Fig. 2(G)]. Quantitatively, there was significantly greater neurite outgrowth in the proximal (facing the bead) quadrant compared with the distal (opposite) quadrant [Fig. 2(K); ANOVA, p < 0.0001]. From these results we concluded that BMP4 promotes SAG neuron survival and neurite outgrowth under these culture conditions.
BMP7 Promotes SAG Neuron Survival
On E4, Bmp7 is broadly expressed within the ventral otocyst, including the mesenchyme, and overlaps with Bmp4 expression in the prosensory domains [Fig. 3(A); Oh et al.,1996]. SAG explants cultured for 24 h displayed significantly greater neurite outgrowth in the presence of BMP7 compared with controls [Fig. 3(B,C,J); ANOVA, p < 0.0001]. Follistatin inhibits BMP7 function by binding with high affinity to BMP7 ligands (Lemura et al.,1998). When explants were cultured in the presence of Follistatin alone or Follistatin and BMP7, there was no significant difference in outgrowth compared with controls [Fig. 3(F,G,J)], suggesting that the effects we observed were due to the addition of BMP7. Furthermore, there was a 3-fold reduction in the percentage of TUNEL-positive neurons in the presence of 0.5 μg/mL BMP7 compared with controls [Fig. 3(K); t-test, p < 0.0001], with no change in the mitotic index [Fig. 3(L); t-test, p = 0.2668].
BMP7 was also expected to promote outgrowth as a point source, after observing enhanced neurite outgrowth in the presence of purified BMP7. However, SAG neurites appeared unresponsive to beads soaked in 10 μg/mL BMP7 [Fig. 3(D,H)]. There was no significant difference in outgrowth between control and BMP7-treated cultures [Fig. 3(M); ANOVA, p = 0.5441] and neurites did not orient toward or away from BMP7 beads [Fig. 3(D,H)]. Before accepting these negative data, positive control tissues, known to be BMP7-responsive, were used to confirm bioactivity of the BMP7-treated beads. In the mouse, BMP7 repels neurite outgrowth from mouse dorsal spinal cord explants in vitro (Augsburger et al.,1999). This study replicated the outgrowth inhibiting effects of BMP7 seen in the mouse by culturing E4 chick dorsal spinal cord explants under the same conditions as E4 SAG explants, on the same day. We observed reduced neurite outgrowth from BMP7-treated explants compared with controls [Fig. 3(E,I)]. Quantification of neurite outgrowth showed significantly less outgrowth in pixels per unit length (ANOVA, p < 0.0001) for explants cultured with beads soaked in 1 μg/mL (10.74 ± 1.03, n = 10) and 10 μg/mL BMP7 (3.26 ± 1.16, n = 7) compared with controls (23.63 ± 1.78, n = 10). This confirmed that the beads soaked in purified BMP7 were bioactive, thus strengthening the conclusion that BMP7 does not influence SAG neurite outgrowth.
Responses to Shh Are Concentration- Dependent
Notochord and floorplate sources of Shh are known to directly influence ear development in the chick (Bok et al.,2005). We hypothesized that a ventral source of Shh may attract a subset of SAG neurites to grow ventrally, toward the cochlear duct, or may repel neurites to prevent them from projecting too far ventrally (past the cochlear duct). To test this hypothesis, responsiveness to 0.5–20 μg/mL of Shh was tested and we observed concentration-dependent responses (Fig. 4). The lowest Shh concentration, 0.5 μg/mL, produced significantly greater neurite outgrowth compared with controls [Fig. 4A,B,I); ANOVA, p < 0.0001]. Cyclopamine specifically inhibits Shh signaling by binding to smoothened, the transmembrane protein required for pathway activation (Incardona et al.,1998; Chen et al.,2002; Frank-Kamenetsky et al.,2002). Adding Cyclopamine to culture medium abrogated the neurite-promoting effects of Shh [Fig. 4(E,F,I)], as expected if Shh exerts a direct effect. We also observed significantly fewer TUNEL-positive neurons in the presence of 0.5 μg/mL Shh [Fig. 4(J); ANOVA, p < 0.0001].
In contrast, higher concentrations of Shh, in the range of 5–20 μg/mL, inhibited neurite outgrowth. Explants displayed significantly fewer and shorter neurites compared with controls [Fig 4(A,C,I); ANOVA, p < 0.0001]. Interestingly, the amount of neurite outgrowth at 5 μg/mL is not significantly different from that observed at 10 and 20 μg/mL [Fig. 4(I)]. As well, cyclopamine did not rescue the reduction in neurite outgrowth seen at 5 μg/mL Shh [Fig. 4(I)]. This suggested that high Shh concentrations may inhibit neurite outgrowth without binding to smoothened, possibly acting as a toxin. TUNEL assays were performed to determine if Shh addition was associated with higher levels of cell death. The percentage of TUNEL-positive neurons increased with Shh concentration, but only the 20 μg/mL concentration reached a statistically significant difference from untreated controls [Fig. 4(J); ANOVA, p < 0.0001]. Overall, it appears that Shh becomes increasingly toxic to SAG neurons at high concentrations and this may explain the reductions in neurite outgrowth [Fig. 4(I)].
Less than 1% of SAG neurons were positively labeled with pH3 when fixed after 24h in culture. There was no significant difference in the mitotic index between control and Shh-treated cultures [Fig. 4(K); ANOVA, p = 0.4500]. Therefore, we expected that the effects on neurite outgrowth and neuron survival were not due to differences in the proportion of proliferating cells between treatment groups.
Explants were also co-cultured with beads that were incubated in either PBS [control; Fig. 4(D)] or 2.5 μg/mL Shh [Fig. 4(H)]. Both neurite promoting and inhibiting responses were considered based on the concentration-dependent responses obtained with purified proteins [Fig. 4(A–C,I)]. Significantly denser and longer neurite outgrowth was observed for neurites facing Shh-coated beads compared with the side of the explant opposing the bead [Fig. 4(H)] and compared with neurites facing control beads [Fig. 4(D,L); t-test, p < 0.05]. Beads were also incubated in higher concentrations of Shh (5, 10, 20, 40 μg/mL) to assay for a repulsive response, but neurite outgrowth resembled that observed for the 2.5 μg/mL condition (not shown). This suggested that either Shh is not repulsive to SAG neurites or that the beads are already saturated with proteins when incubated in 2.5 μg/mL Shh, and incubating beads in higher concentrations of Shh does not increase the amount released by the bead. To create a larger Shh point source, irrespective of the amount of protein on the beads, several Shh beads were placed adjacent to the explant. A repulsive response was still not observed (not shown), suggesting that in combination with the TUNEL results, high Shh concentrations are toxic to SAG neurons, which results in reduced neuron survival and reduced neurite outgrowth.
FGF8, FGF10, and FGF19 Promote SAG Neuron Survival and Neurite Outgrowth
FGF8 (Fig. 5), FGF10 (Fig. 6), and FGF19 (Fig. 7) also produced concentration-dependent neurite outgrowth responses. FGF8 and FGF19 significantly promoted neurite outgrowth at 50, 100, and 200 ng/mL, but at 500 ng/mL neurite outgrowth was comparable to control levels [Figs. 5(H), 7(H); ANOVA, p < 0.0001]. Only the 200 ng/mL FGF10 concentration significantly promoted SAG neurite outgrowth [Fig. 6(H), ANOVA; p < 0.01]. SU5402 is a tyrosine kinase activity inhibitor that inhibits FGF receptor activation (Mohammadi et al.,1997), thus interfering with FGF signaling. SU5402 abrogated the neurite-promoting effects of FGF8 [Fig. 5(E,F,H)] and FGF19 [Fig. 7(E,F,H)], but not FGF10 [Fig. 6(E,F,H)]. The increase in neurite outgrowth in cultures with FGFs was associated with a reduction in the percentage of TUNEL-positive neurons. FGF10- and FGF19-treated SAGs showed a 2-fold reduction in the percentage of TUNEL-positive cells compared with controls [Fig. 6(I), t-test, p < 0.01; 7I, p < 0.0001], whereas FGF8 had a weaker effect [Fig. 5(I); ANOVA, p < 0.01]. Neither FGF8 [Fig. 5(J); ANOVA, p = 0.1742], FGF10 [Fig. 6(J); t-test, p = 0.8366] nor FGF19 [Fig. 7(J); t-test, p = 0.8706] affected the mitotic index when compared with controls.
Fgf8 is expressed in the sensory patches of the saccule and lagena (vestibular organs) but is absent from the auditory sensory region, the basilar papilla (Sanchez-Calderon et al.,2004). Fgf19 is strongly expressed in the sensory regions of the utricle and lagena (vestibular organs), the SAG neurons, and some of the prosensory region borders, but is also absent from the auditory basilar papilla (Sanchez-Calderon et al.,2007). On the basis of these expression patterns, we hypothesized that FGF8 and FGF19 may be important for guiding subsets of vestibular axons or for repelling auditory axons from innervating vestibular sensory regions during development. Explants were cultured with beads that were treated with PBS, FGF8 or FGF19. Effects on overall outgrowth (promotion and inhibition) as well as directional outgrowth toward (attraction) or away from (repulsion) the bead were all considered. FGF8-treated beads [Fig. 5(D,G)] and FGF19-treated beads [Fig. 7(D,G)] both promoted neurite outgrowth. At this developmental stage, the SAG contains vestibular and auditory neurons that will soon separate into two ganglia, so the neurites are probably somewhat segregated within the explants. If only a subset of neurites was responding to the bead, we expected there to be an asymmetric neurite outgrowth pattern in the purified protein cultures, but this pattern was not observed [Figs. 5(B,C) and 7(B,C)]. Quantitatively, there was significantly greater neurite outgrowth on the side of the explants that were facing beads soaked in FGF8 [Fig. 5(K); t-test, p < 0.01] or FGF19 [Fig. 7(K); t-test, p < 0.0001]. In the mouse, Fgf10 is expressed by all sensory organs and SAG neurons (Pirvola etal.,2000). However, there is limited FGF10 expression data for the chick. At HH26, FGF10 has been shown to be strongly expressed in the anterior and lateral cristae and the utricular macula, weakly expressed by the posterior crista [Fig. 6(A); Chang et al.,2004] and absent from SAG neurons on E2–3 (Alsina etal.,2004). To test the hypothesis that FGF10 may exert directional effects on neurite outgrowth, in addition to promoting neuron survival [Fig. 6(I)], we cultured explants with beads that were soaked in PBS [Fig. 6(D)] or 500 ng/mL FGF10 [Fig. 6(G)]. We observed significantly greater neurite outgrowth in the presence of FGF10-treated beads compared with control cultures [Fig. 6(K); t-test, p < 0.01]. These results suggest that in addition to providing trophic support for SAG neurons, FGFs -8, -19, and -10 may also provide outgrowth promoting cues for growing SAG neurites.
E4 SAG Neurons Are Unresponsive to Purified FGF2
FGF2 was tested because it promotes the survival of early (E2–3) and late (E8–16) stage chick SAG neurons in vitro (Hossain et al.,1996; Carnicero etal.,2001). We found no significant difference in neurite outgrowth between control and FGF2-treated explants [Fig. 8(A); ANOVA, p = 0.7874]. Bioactivity of FGF2 was confirmed by performing a cell survival assay under the same culture conditions that were used for SAG explants. DF-1 chicken embryo fibroblasts were cultured in collagen gels for 24 h with or without FGF2. Apoptotic cells were stained with anti-caspase-3 and all cell nuclei were stained with DRAQ5. There were significantly fewer caspase-3-positive cells in the presence of FGF2 compared with controls [Fig. 8(B); ANOVA, p < 0.0001]. From these results, we concluded that FGF2 promotes DF-1 survival and is bioactive under SAG explant culture conditions.
Combinations of Purified Proteins Do Not Produce Additive Effects
The expression patterns of some of the morphogens we tested either overlap or are mutually exclusive of one another. For example, the spatial distribution of Bmp4 and Bmp7 transcripts are largely similar [Figs. 2(A) and 3(A)], whereas Fgf8 and Fgf19 transcripts are mostly nonoverlapping [Figs. 5(A7) and (7A)]. If subpopulations of neurites are differentially responsive to single proteins added to culture medium, it might be possible to detect additive effects on neurite outgrowth with combinations of proteins. We tested the effects of BMP4 + 7, BMP4 + FGF8, and FGF8 + 19. In each case, neurite outgrowth was comparable to what was observed when one protein was added to the cultures (not shown). These results led us to conclude that additive effects, if present, must be insufficiently robust to detect with the quantification methods used here.
Morphogens are expressed within and surrounding the inner ear during stages of SAG neurite outgrowth and sensory organ innervation. On the basis of published expression data, we hypothesized that some morphogens may be important for guiding SAG axons toward their sensory targets and/or may be important for promoting SAG neuron survival. In an earlier study, we reported that E4 SAG neurites were unresponsive to some members of the Wnt morphogen family (Fantetti et al.,2011). In this study, we have expanded the Wnt results by testing the responsiveness of SAG neurons to some members of the BMP, Shh, and FGF morphogen families. Experiments were carried out in vitro to determine SAG responsiveness to individual molecules, to avoid potential confounding variables in a complex in vivo environment. We provide evidence that some of the molecules that we tested enhance survival and promote neurite outgrowth of SAG neurons.
During development, a single morphogen can regulate neurogenesis, neuron survival, and axon guidance in the same tissue. For example, during spinal cord development, Shh first regulates neurogenesis and cell survival, and then is used as a chemotropic cue to guide commissural axons ventrally (Chiang et al.,1996; Briscoe and Ericson,1999; Jessell,2000; Litingtung and Chiang,2000; Charron et al.,2003). As a result, Shh is both a trophic (promotes cell survival) and a tropic (directs axon outgrowth) factor. Under culture conditions, neuronal survival will reflect a balance between cell proliferation (neurogenesis) and cell death, and both were evaluated in this study of SAG neurons. Assuming a relatively uniform responsiveness among the population of treated cells, more neurons would be expected to make more neurites. Indeed, SAG explants with enhanced neurite outgrowth contained significantly fewer TUNEL-positive cells, suggesting improved neuron survival. But there was no additional effect of morphogen treatment on neurogenesis. An isolated E4 chicken SAG should contain both progenitors and neurons: the first SAG neurons begin to differentiate on E3 and are all postmitotic by E7 (D'Amico-Martel,1982). In fact, overall neurogenesis was rather modest in control cultures (with added CNTF and NT-3), as reflected by a mitotic index of approximately 1%. There was no change in this mitotic index among cultures grown in the presence of purified morphogens, suggesting that neurogenesis was relatively constant across conditions. We confirmed that the proliferation assay we employed (phosphohistone H3 immunostaining) was sensitive enough to detect a significant change (2-fold) in the mitotic index. Specifically, SAGs grown in serum-free conditions and only supplemented with 20 or 100 ng/mL of insulin-like-growth factor-1 (IGF-1) had elevated mitotic indices when compared with untreated explants (i.e., those grown in serum-free media without additional supplements; data not shown). A similar response to IGF-1 was reported previously for E3 SAGs cultured either in isolation or in association with the otocyst: neurons displayed enhanced neurite outgrowth and increased cell proliferation when treated for 24 h with IGF-1 (Camarero et al.,2003). In summary, morphogen treatments that stimulated SAG neurite outgrowth may have done so by enhancing neuron survival, but apparently not by increasing neurogenesis.
In addition to the aforementioned trophic effects, we sought to address the question of whether enhanced neurite outgrowth was also induced by direct tropic effects on the neurites. This was examined by asking whether focal sources of morphogens could promote local outgrowth from the ganglion or turn the neurites toward the beads. Only BMP7 lacked the ability to promote neurite outgrowth on the side of the explant facing the beads. This led us to consider the degree to which differences in neuron survival could account for differing outgrowth responses. BMP7-treated cultures had the largest (3-fold) reduction in the percentage of TUNEL-positive neurons and Bmp7 transcripts display the broadest expression pattern across sensory and nonsensory regions (Oh etal.,1996). For BMP7, we presume that increases in neurite outgrowth are more likely to be explained by enhanced neuron survival than by directional outgrowth. Alternatively, FGF8 produced the weakest reduction in the proportion of TUNEL-positive neurons, but when presented as a point source significantly promoted neurite outgrowth. It is possible that for FGF8, enhanced neurite outgrowth is more accounted for by neurite responsiveness than it is by neuron survival. It appears that some of the molecules we tested may be more important for survival than they are for promoting outgrowth. Neurite length measurements (not shown) were performed for purified protein studies and produced similar results and statistical significance as pixel measurements. Similarly, bead co-cultures with an increase in outgrowth density also displayed longer neurites. Therefore, it is also possible that enhanced neurite outgrowth may have been due to changes in axon properties that we did not quantify, such as outgrowth rate, as observed for BMP chemorepellents and spinal cord commissural axons (Phan et al.,2010).
Studies from mammalian and zebrafish systems suggest that our results with BMP4, Shh, and FGF10 may extend to other species. In the postnatal mouse, BMP4 promotes dissociated spiral ganglion neuron survival in vitro (Whitlon et al.,2007) and cultures of dissociated rat spiral ganglion neurons show enhanced survival when co-cultured with olfactory ensheathing cells that are known to express BMP4 (Liu et al.,2010). Additionally, Bmp4 mutant mice display reduced cochlear innervation (Blauwkamp et al.,2007), suggesting that BMP4 may be important for survival or axon guidance in vivo. Deletion of Shh in the inner ear blocks the appearance of the cochlea, saccule, and spiral ganglion in mice (Riccomagno etal.,2005; Brown and Epstein,2011). Similarly, loss of hedgehog signaling in zebrafish results in reduced posterior macula innervation and a reduced SAG (Sapede and Pujades,2010). In these studies, the absence of Shh signaling may directly prevent neuronal specification. However, neuronal loss may also be secondary to the loss of sensory structures and the trophic support that they provide. FGF10 is expressed by spiral ganglion neurons and sensory epithelia (Pirvola et al.,2000) in the mouse, and may be important for neuron survival (Wright and Mansour,2003). Fgf10−/− mice lack a posterior crista (Pauley et al.,2003) and show an increase in the proportion of apoptotic neurons (Pirvola et al.,2000). The number of fibers contacting vestibular sensory epithelia is also reduced (Pirvola et al.,2000; Pauley etal.,2003). These results suggest that FGF10 is an important trophic factor in vivo for the mouse. Our results support the possibility that BMPs, Shh, and FGFs may serve as trophic factors once neurons begin to differentiate, not only in the chicken but also perhaps in other species.
On the other hand, expression data and results from functional studies suggest that some of the molecules we tested may have different functions on SAG neurons in different species. Although Bmp4 is expressed in all prosensory regions in the chick (Wu and Oh,1996), in the mouse it is restricted to the prosensory cristae (Morsli etal.,1998) and is not present in the auditory prosensory region (Ohyama et al.,2010). It is possible that in the mouse, BMP4 in vestibular prosensory regions serves as a chemoattractant and that BMP4 in nonsensory regions of the cochlea serves as a chemorepellent or has no effect. Shh is not expressed in the otocyst of the mouse (Riccomagno et al.,2002) nor the chick (Bok et al.,2005), but is expressed by mouse spiral ganglion neurons between E13.5 and P0 (Driver et al.,2008; Liu et al.,2010). To date, Shh expression has not been described for the chick SAG. The differences in Shh sources may result in differing responses of otic neurons to Shh across systems. FGF2 increases the survival of E2–3 chick SAG (Hossain et al.,1996) and E10.5–12 mouse spiral ganglion (Hossain and Morest,2000) in vitro. In contrast, we found E4 SAG neurons to be unresponsive to FGF2. The time-sensitive responses of chick SAG to FGF2 (Carnicero et al.,2001; Hossain et al.,2002) likely account for the lack of a response in this study. Fgf8 is expressed in the sensory patches of the saccule and lagena (vestibular organs), but is absent from the auditory sensory region of the chick (Sanchez-Calderon et al.,2004). In the mouse, FGF8 is expressed in the inner hair cells of the cochlea (auditory organ; Hayashi et al.,2007; Jacques et al.,2007). Therefore, vestibular and auditory neurons may be differentially responsive to FGF8 in the chick and mouse, respectively. In this study, FGF10 promoted SAG neurite outgrowth and neuron survival, but SU5402 failed to inhibit these effects. Thus, FGF10 may indirectly activate other signaling pathways in this system.
Although morphogen combination experiments yielded no additive effects on neurite outgrowth, we cannot exclude the possibility that combining neurotrophins (specifically CNTF and NT-3) with morphogens in our culture system altered the responsivity of SAG neurons. For example, BMP4 reduces the number of mouse sympathetic neurons in vitro, but when combined with NGF, the effects are reversed (Gomes and Kessler,2001). In this system, BMP4 increases the dependence on neurotrophins (Gomes and Kessler,2001). In another example, when mouse spiral ganglion neurons are cultured with FGF2, in combination with NT-3 or BDNF, neuroblast migration and neurite outgrowth are enhanced beyond levels of that observed with FGF2 alone (Hossain et al.,2008). It is possible that some of the morphogens we tested altered the dependence of SAG neurons on CNTF or NT-3 and thus acted indirectly on neuronal survival.
An in vitro approach provides a controlled and simplified environment to evaluate neurite outgrowth responses to individual molecules. This offers an alternative to in vivo gain- and loss-of-function approaches, where patterning defects can severely disrupt ear development and indirectly cause innervation defects as discussed earlier. However, there are limitations to using an in vitro system in the way that we have presented it here. First, on E4 the SAG consists of a heterogeneous cell population. There are two populations of neurons (vestibular and auditory), two populations of axons (peripheral and central), and non-neuronal cells (e.g., glial, endothelial). Subpopulations of neurons and axons may be differentially responsive. For example, cultures of dissociated mouse spiral ganglion neurons contain more monopolar neurons in the presence of BMP4 (Whitlon et al.,2007), suggesting that central and peripheral processes may respond differently to BMP4. Due to the technical limitations of our invitro system, we were unable to distinguish between these cell and neurite populations. Furthermore, non-neuronal cell types present in our cultures can serve as a target for the applied molecules. Schwann cells express neurotrophins that promote spiral ganglion neuron survival and consequently spiral ganglion neurites preferentially grow on Schwann cells in vitro (Hansen et al.,2001; Bostrom et al.,2009; Whitlon et al.,2009; Jeon et al.,2011). If morphogens influence Schwann cell development or survival, this could indirectly contribute to the responses we observed.
In summary, this is the first study to directly test the responses of E4 chick SAG neurons to members of the BMP, Shh, and FGF morphogen families. Our results suggest that morphogens, in addition to classic axon guidance molecules and neurotrophins, are also capable of regulating sensory neuron survival and sensory organ innervation. We provide evidence for how individual molecules regulate SAG neuron survival and neurite outgrowth, but leave open the possibility that there may be different responses to these molecules in vivo. Because these morphogen gene families play multiple roles in otic development (Groves and Fekete, 2012), it may be necessary to design targeted, conditional gene knockouts to evaluate their ability to serve as axon guidance cues in intact animals.
We thank Doris Wu and Wiese Chang for advice with experiments, and Rodney McPhail for help with figure illustrations.