Expression of the heparan sulfate 6‐O‐endosulfatases, Sulf1 and Sulf2, in the avian and mammalian inner ear suggests a role for sulfation during inner ear development
Abstract
Background: Inner ear morphogenesis is tightly regulated by the temporally and spatially coordinated action of signaling ligands and their receptors. Ligand–receptor interactions are influenced by heparan sulfate proteoglycans (HSPGs), cell surface molecules that consist of glycosaminoglycan chains bound to a protein core. Diversity in the sulfation pattern within glycosaminoglycan chains creates binding sites for numerous cell signaling factors, whose activities and distribution are modified by their association with HSPGs. Results: Here we describe the expression patterns of two extracellular 6‐O‐endosulfatases, Sulf1 and Sulf2, whose activity modifies the 6‐O‐sulfation pattern of HSPGs. We use in situ hybridization to determine the temporal and spatial distribution of transcripts during the development of the chick and mouse inner ear. We also use immunocytochemistry to determine the cellular localization of Sulf1 and Sulf2 within the sensory epithelia. Furthermore, we analyze the organ of Corti in Sulf1/Sulf2 double knockout mice and describe an increase in the number of mechanosensory hair cells. Conclusions: Our results suggest that the tuning of intracellular signaling, mediated by Sulf activity, plays an important role in the development of the inner ear. Developmental Dynamics 244:168–180, 2015. © 2014 Wiley Periodicals, Inc.
Introduction
The vertebrate inner ear comprises a complex system of fluid filled ducts and chambers, within which are at least six distinct sensory organs. These sensory organs house cells that act as mechanoreceptors, which are responsible for the perception of sound in the cochlea, and gravity and acceleration in the vestibular system (Groves and Fekete, 2012). In both birds and mammals, the inner ear's complex labyrinth of ducts and sensory epithelia develops from a simple sheet of placodal surface ectoderm located around the rostral neural plate. The embryo repeatedly uses multiple signaling pathways to regulate all stages of inner ear morphogenesis. Fibroblast growth factor (FGF) signals govern the induction of an initial ectodermal domain that gives rise to both the otic and the epibranchial placodes (Ladher et al., 2005; Ohyama et al., 2007). Subsequent high levels of Wnt signaling from the midline act to refine a subset of cells from this otic‐epibranchial progenitor domain (OEPD) to become the otic placode, in part by up‐regulating Notch signaling, and down‐regulating FGF signaling (Ladher et al., 2000; Chambers and Mason, 2000; Freter et al., 2008; Jayasena et al., 2008). After the otic placode invaginates and moves inside the head, its dorsal–ventral axis is patterned in a way similar to that observed in the spinal cord—ventral structures develop in response to sonic hedgehog (Shh) signals emanating from the notochord and floorplate, and dorsal structures develop in response to secreted Wnts from the midline (Riccomagno et al., 2002, 2005). The development of the sensory organs and their mechanosensitive cells is controlled by a combination of Notch, Wnt, FGF, and bone morphogenetic protein (BMP) signals. Notch mediated lateral inhibition acts to establish the salt‐and‐pepper cellular mosaic of alternating mechanosensory hair cells and nonsensory supporting cells in the sensory epithelia of the inner ear (Bermingham et al., 1999; Daudet and Lewis, 2005; Daudet et al., 2007). FGF, Wnt, and BMP signaling play important roles in maintaining the populations of sensory progenitors within, and defining the borders of, the sensory epithelia (Riccomagno et al., 2005; Ohyama, 2006; Hayashi et al., 2007, 2008; Chang et al., 2008; Ohta et al., 2010; Ohyama et al., 2010; Ono et al., 2014). A common feature of these signaling pathways is that they require the presence of heparan sulfate proteoglycans (HSPGs) on the cell surface to mediate receptor‐ligand interactions (Ornitz, 2000; Lin, 2004).
HSPGs are surface molecules that comprise much of the extracellular matrix surrounding all animal cells studied (Bernfield et al., 1999; Turnbull et al., 2001). Multiple, serine bound, HS glycosaminoglycan chains are attached to a core proteoglycan (Lindahl et al., 1998). HS chains themselves are synthesized in the Golgi by glycosyltransferases to create a basic glycosaminoglycan chain. The glycosaminoglycan chains undergo subsequent modifications, including partial deacetylation, N‐sulfation, epimerization, and O‐sulfation. This creates rich structural heterogeneity both within a single glycosaminoglycan chain, and between different glycosaminoglycan chains. This vast diversity in sulfation pattern allows HSPGs to interact with many different signaling ligands and receptors. These interactions are often key in facilitating growth factor signaling, activation and receptor binding (Turnbull et al., 2001; Kusche‐Gullberg and Kjellén, 2003; Ledin et al., 2004; Chapman et al., 2004).
O‐sulfation of HS chains has been identified as a particularly important regulator of cell signaling. O‐sulfation within the Golgi is controlled by sulfotransferases (OSTs). However, a further, final modification of HS sulfation can be made on the cell surface by the secreted sulfatase enzymes Sulf1 and Sulf2. The Sulfs remodel HSPG structure by selectively removing sulfate groups from the 6‐O position of glucosamine residues in heparan sulfate chains (Ai et al., 2006). This alters the binding affinity of HSPGs to ligand–receptor complexes, and can positively or negatively influence their signaling (Lai et al., 2004; Ai et al., 2006; Freeman et al., 2008). Sulfs were first identified in Quail, with the description of QSulf1 (Dhoot et al., 2001), and mouse with the identification of mSulf1 and mSulf2 (Morimoto‐Tomita et al., 2002). Orthologues have since been identified in chick, human, rat, zebrafish, and amphibian (Ohto et al., 2002; Braquart‐Varnier et al., 2004; Nagamine et al., 2005; Freeman et al., 2008; Gorsi et al., 2010). Modification of HSPG sulfation by the Sulfs has been shown to regulate Shh, Wnt, FGF, BMP, GDNF, and VEGF signaling during embryonic development (Dhoot et al., 2001; Ai et al., 2003, 2007; Wang et al., 2004; Danesin et al., 2006; Freeman et al., 2008; Meyers et al., 2013; Gorsi et al., 2014).
Here, we present a detailed description of the expression patterns of the Sulf1 and Sulf2 genes during the development of the avian and the mammalian inner ear. Furthermore, we analyze the cochleae of single knockout mice and Sulf1−/− Sulf2−/− double knockout mice and describe a phenotype that suggests that these enzymes play a role in the development of mechanosensory hair cells.
Results
Expression of cSulf1 in the Chick Inner Ear
We used in situ hybridization to investigate cSulf1 and cSulf2 expression at different stages of chicken inner ear development. Embryonic day (E) 3 whole embryos, and E5, E8, and E11 dissected inner ears were analyzed (Fig. 1). These stages show some of the major developmental events that occur during inner ear morphogenesis—the transformation of the otocyst to the complex structures of all of the vestibular apparatus, the cochlea, and the differentiation of the mechanosensory cells within their sensory patches.
Cartoon view of the approximate cutting planes used to analyze the cellular localization of Sulf transcripts in the chick and mouse inner. For the chick inner ear: lc, lateral crista; sc, superior crista; pc, posterior crista; ut, utricular macula; s, sacular macula; bp, basilar papilla; lag, lagena. For the mouse inner ear: lc, lateral crista; ac, anterior crista; pc, posterior crista; ut, utricular macula; s, sacular macula.
At E3, cSulf1 transcripts are detected in derma‐myotome, the neural tube, the branchial arches, the midbrain and the hindbrain (Fig. 2A), in a pattern similar to the one previously reported in Xenopus (Freeman et al., 2008). Weak expression is also seen within the otocyst (Fig. 2A). cSulf1 is first detected in the inner ear at E5, in the presumptive lateral, posterior and superior cristae (Fig. 2B). MyosinVIIa immunolabeled sections of the cristae revealed that this expression is in the hair cell layer of the developing sensory epithelium (Fig. 2B′,B″,B″′; lateral crista shown). At E8, cSulf1 is also detected in the utricular macula (Fig. 2C). In all of the vestibular sensory epithelia, cSulf1 transcripts are found exclusively in the hair cell layer (Fig. 2C′,C″,C″′,D,D′,D″). At E11, the signal intensity is increased in the cristae whilst in the utricular macula it is decreased (Fig. 2E,E′,E″,F,F′,F″).
cSulf1 expression pattern in the developing vestibular organs of the avian inner ear. A: At E3, expression is present in the midbrain (mb), hindbrain (hb), branchial arches (ba), neural tube (nt), and the derma‐myotome (dm). Weak expression is also present around the otic vesicle (ov). B: At E5 expression is present in the developing lateral (lc), posterior (pc), and superior crista (sc). B′–B″′: Myosin VIIa labeling reveals cSulf1 is expressed in the hair cell layer of the developing cristae. C: At E8, cSulf1 transcripts are present in the posterior (pc), lateral (lc), superior (sc) crista, and the utricular macula (ut). C′–C″′: Myosin VIIa labeling shows cSulf1 remains restricted to the hair cell layer in the cristae. D–D″: Myosin VIIa labeling shows cSulf1 is also restricted to the hair cell layer in the utricular macula. E–E″: At E11, strong levels of cSulf1 transcripts are present in the hair cells of the cristae (superior crista shown). F–F″: cSulf1 remains present in the utricular hair cell layer of the E11 inner ear. HC: hair cell layer.
In the cochlea, cSulf1 transcripts are first observed at E8. Strong expression is present in the hair cells of the lagena (Fig. 3A,A′,A″). In the basilar papilla, expression of cSulf1 is graded along the basal to apical axis. At the apex, cSulf1 is strongly expressed in the hair cell layer across the entire width of the sensory epithelium (Figs. 3B, 2B′,B″). Expression is also present at the location of the spindle‐shaped cells (Fig. 3B″). In the mid‐section of the basilar papilla, cSulf1 is restricted to the tall hair cells located on the neural side of the sensory epithelium (Fig. 3C,C′,C″). Expression at the location of the spindle‐shaped cells (ssc) is also seen (Fig. 3C″). At the basal end, cSulf1 is absent from the sensory epithelium, but expression in the region of the spindle‐shaped cells remains (Fig. 3D,D′,D″). At E11, cSulf1 expression remains in the hair cells of the lagena (Fig. 3E,E′,E″). However, in the basilar papilla, the expression pattern has changed: cSulf1 is now absent from the apical region of the sensory epithelium (Fig. 3F,F′,F″). In the mid‐section of the basilar papilla, cSulf1 is restricted to the hair cells on the neural side, and is also strongly expressed in the tegmentum vascularis (Fig. 3G,G′,G″). At the base of the E11 basilar papilla, cSulf1 is expressed in the hair cell layer across the width of the sensory epithelia, although expression appears strongest on the neural side. Strong expression is also present in the tegmentum vascularis (Fig. 3H,H′,H″).
cSulf1 expression in the developing cochlea of the avian inner ear. A–A″: cSulf1 transcripts are first detected in the E8 cochlea, and are strongly expressed in the hair cells of the lagena, as confirmed by immunostaining with anti‐myosin VIIa. B–B″: Coronal sections reveal an apical to basal gradient of cSulf1 expression. At the apex of the basilar papilla, cSulf1 is expressed in hair cells across the width of the sensory epithelium, and in the location of the spindle‐shaped cells (ssc). C–C″: In the mid‐section, expression is restricted to a subset of hair cells on the neural side (arrowheads denote margins of expression), and in the location of the spindle‐shaped cells (ssc). D–D″: At the basal end, cSulf1 expression is absent from the sensory epithelium but remains in the location of the spindle‐shaped cells (ssc). E–E″: cSulf1 expression remains present in the hair cells of the E11 lagena. F–F″: cSulf1 transcripts are absent in the apical hair cells of the E11 basilar papilla. Expression is present in the tegmentum vascularis (tv). G–G″: In the mid‐section of the E11 basilar papilla, transcripts are present in the hair cells on the neural side of the sensory epithelium, and also in the tegmentum vascularis (tv). H–H″: At the basal end, transcripts are present in all hair cells, with stronger expression on the neural side (arrow heads). Strong expression in the tegmentum vascularis is present (tv). HC, hair cell layer.
Expression of cSulf2 in the Chick Inner Ear
Sulf1 and Sulf2 are expressed in similar regions of the developing embryo in several model systems (Freeman et al., 2008; Winterbottom and Pownall, 2009; Gorsi et al., 2010; Guiral et al., 2010; Ratzka et al., 2010). Both Sulf1 and Sulf2 share substrate specificity, and as such are considered functionally redundant (Ai et al., 2006; Ratzka et al., 2008). Thus we analyzed the expression pattern of Sulf2 to determine its inner ear expression.
At E3, cSulf2 is expressed in the neural tube, midbrain, forebrain, and the eye, but not in the otic vesicle (Fig. 4A). At E5, cSulf2 is expressed in the developing lateral, superior and posterior cristae, and also in the vestibular ganglion and inferior ridge (Fig. 4B). Cryosections through the lateral crista show that cSulf2 is present in the hair cell layer, and colocalizes with the expression of MyosinVIIa protein, but also extends beyond the region of MyosinVIIa expression (Fig. 4C′,C″,C″′; black arrowhead denotes the boundary of cSulf2 expression). At E8, cSulf2 is strongly expressed in the hair cell layer of the utricle (Fig. 4D,D′,D″) in addition to the hair cell layers of all three cristae (Fig. 4E,E′,E″; posterior crista is shown). At E11, this pattern of strong expression in the hair cells of the utricle, and weaker expression in the hair cells of the cristae is maintained (Fig. 4F,F′,F″).
cSulf2 expression in the developing vestibular organs of the chick inner ear. A: At E3, expression is present in the forebrain (fb), midbrain (mb), and the neural tube (nt). There is no expression in the otic vesicle at this stage (ov). B: At E5, cSulf2 is expressed in the posterior (pc) and the superior (sc) cristae. Transcripts are also present in the vestibular ganglion (vg) and in the inferior ridge of the developing basilar papilla (ir). C–C″: Cryosections and immunostaining for myosin VIIa in the posterior crista shows expression in the hair cell layer (hc). cSulf2 expression extends beyond the boundary of myosinVIIa expression (arrowhead). D–D″: At E8, cSulf2 is expressed in the hair cell layer of the utricula macula. E–E″: cSulf2 is weakly expressed in the hair cells of all three cristae at E8 (posterior crista shown). F–F″: At E11, cSulf2 remains strongly expressed in the hair cells of the utricular macula (ut), and weaker expression is still seen in the hair cells of the cristae (superior crista, sc, shown). HC, hair cell layer.
In the E8 cochlea, cSulf2 is expressed weakly in the hair cells of the lagena (Fig. 5A,A′,A″), but is not expressed in hair cells at the apex of the basilar papilla. Instead, expression is observed in the supporting cell layer (Fig. 5B,B′,B″). In the mid‐section of the E8 basilar papilla, cSulf2 is expressed in the neural side of both the hair cell and support cell layer of the sensory epithelium (Fig. 5C,C′,C″). At the basal end, cSulf2 is strongly expressed across the width of the hair cell layer and the supporting cell layer (Fig. 5D,D′,D″). Robust expression is present in the inferior ridge (ir), adjacent to the abneural side of the sensory epithelium, throughout the length of the basilar papilla (Fig. 5B–D). At E11, cSulf2 expression is stronger in the hair cells of the lagena (Fig. 5E,E′,E″). At the apex of the basilar papilla, cSulf2 expression is down‐regulated from both the supporting cells and the hair cells, but expression is observed in the region of the spindle‐shaped cells, and the inferior ridge (Fig. 5F,F′,F″). In the mid section, cSulf2 is expressed in supporting cells, the inferior ridge and the spindle‐shaped cells (Fig. 5G,G′,G″). At the base of the basilar papilla, cSulf2 is expressed strongly in both the hair cell layer and the support cell layer, as well as in the inferior ridge (Fig. 5H,H′,H″).
cSulf2 expression in the developing cochlea of the chick inner ear. A–A″: In the E8 cochlea, cSulf2 is expressed weakly in both the hair cells (identified by myosin VIIa immunostaining) and supporting cells of the lagena, and strongly in the inferior ridge (ir). B–B″: At the apex of the basilar papilla, expression is weak in the supporting cell layer, absent from the hair cell layer, and strong in the inferior ridge (ir). C–C″: In the mid section of the basilar papilla, expression is weak in the supporting cells and the hair cells on the neural side of the sensory epithelium (identified by myosin VIIa immunostaining, arrowhead indicates boundary of cSulf2 expression). Strong expression is present in the inferior ridge (ir). D–D″: At the base of the E8 basilar papilla, expression is strong in the hair cells and the inferior ridge (ir), and weak in the supporting cell layer. E–E″: In the E11 cochlea, cSulf2 is strongly expressed in the hair cells of the lagena, and weakly expressed in the supporting cell layer. F–F″: At the apex of the basilar papilla, cSulf2 is absent from the hair cell and supporting cell layers, but is present at the location of the spindle shaped cells (ssc) and the inferior ridge (ir). G–G″: In the mid‐section of the E11 basilar papilla, cSulf2 is absent from the hair cell layer but is expressed in the supporting cell layer and the inferior ridge (ir). There is also some expression present in the region of the spindle‐shaped cells (ssc). H–H″: At the basal end, cSulf2 is expressed strongly in the hair cell and the supporting cell layers, as well as the inferior ridge (ir). HC, hair cell layer; SC, supporting cell layer.
Sulf1 Expression in the Mouse Inner Ear
We compared the expression pattern of Sulf1 and Sulf2 in the avian inner with that of the mammalian inner ear. The presence of Sulf1 and Sulf2 transcripts in the mammalian inner ear has been previously reported (Ratzka et al., 2010), but a detailed temporal and spatial analysis has never been performed. Using in situ hybridization, we found that mSulf1 expression in the inner ear begins at E12.5, in the ventrolateral region of the developing otocyst (Fig. 7A). At E14.5, mSulf1 is expressed strongly in the roof of the cochlea duct (Fig. 7B). Expression is also present in the spiral ganglion cells (Fig. 7B; dotted lines). The expression in the cochlea duct encompasses the antero–medial side of the prosensory domain (Fig. 7B′, line denotes antero–medial expression). At this stage, mSulf1 is not expressed in the vestibular region (data not shown). However, by E17.5, mSulf1 is expressed in the hair cells of the developing cristae and the utricle (Fig. 7C,C′,C″). In the cochlea, mSulf1 expression is strong in the lateral wall, which will develop into the stria vascularis and Reissner's membrane, and expression is present in the outer hair cells, but absent from the inner hair cells (Fig. 7D,D′,D″). In the post‐natal day (P) 0 inner ear, expression is present in the spiral ganglion and the hair cell layer of the vestibular sensory organs (Fig. 7E). In the cochlea, mSulf1 is expressed in the epithelial cell layer of Reissner's membrane, the marginal cells of the stria vascularis, and in both the inner and outer hair cells (Fig. 7E′,E″,E″′,E″″).
Sulf2 Expression in the Mouse Inner Ear
mSulf2 expression in the inner ear is first detected at E12.5. Like mSulf1, mSulf2 expression is restricted ventrally. However, its expression is located in the medial wall (Fig. 8A). At E14.5, mSulf2 is expressed in the spiral ganglion cells (Fig. 8B; dotted lines), and also in a posterior region of the roof of the cochlea duct (Fig. 8B′). The expression in the cochlea duct also encompasses the posterior side of the medial wall (Fig. 8B′). In the vestibular organs mSulf2 is expressed in the hair cell layers of the developing cristae and the utricle (Fig. 8C,C′,C″). At E17.5, expression remains in the hair cell layer of the cristae and the utricle (Fig. 8D,D′,D″). In the cochlea, mSulf2 expression is observed in the spiral ganglion cells (dotted lines), and the entirety of the lateral wall (Fig. 8E,E″). Expression is also detected in both the inner and outer hair cells, although is more robust in the outer hair cells (Fig. 8E″). At P0, expression remains strong in the hair cells of the cristae and the utricle, as seen in earlier stages (Fig. 8F). In the cochlea, mSulf2 is expressed in the epithelial cell layer of Reissner's membrane, the marginal cells of the stria vascularis, and in both inner and outer hair cells of the cochlea (Fig. 8F′).
Sulf1 and Sulf2 Knockout Mice Suggest a Role for the Sulfs in Hair Cell Development
Given the similar expression patterns of the Sulfs in the mammalian inner ear, we decided to analyze the cochleae of Sulf1 and Sulf2 mutant mice (Nagamine et al., 2012) to establish if the expression of the Sulfs was functionally significant during the development of the mammalian inner ear. We used F‐actin staining to assess the architecture of the P0 organ of Corti, as this allowed clear visualization of the cells that comprise the sensory epithelia. We found a mild but reproducible increase in the number of supernumerary inner hair cells and outer hair cells in the Sulf1−/−Sulf2−/− cochleae in comparison with wild‐type controls. Analysis of variance (ANOVA) of four separate animals per experimental sample (8 cochleae per sample) revealed that wild‐type P0 organ of Corti have an average of 3.2 supernumerary inner hair cells, and 10.8 supernumerary outer hair cells. In contrast, Sulf1−/−Sulf2−/− double knockout mice have a higher average of 22.75 supernumerary inner hair cells (ANOVA P < 0.0001; Tukey test P < 0.01), and 34.75 supernumerary outer hair cells (ANOVA P = 0.002; Tukey test P < 0.01; Fig. 9A,D,E). To ascertain whether these increases in inner and outer hair cell numbers were linked to the loss of a particular Sulf, we analyzed the number of supernumerary inner and outer hair cells in single knockout Sulf mutant mice. Sulf1−/− mice exhibit an increase of 12.75 supernumerary inner hair cells (ANOVA P < 0.0001; Tukey test P < 0.05; Fig. 8B,E). However, the average number of supernumerary outer hair cells is not significantly increased (average 10.5 supernumerary outer hair cells; Tukey test: nonsignificant; Fig. 9B,E). In Sulf2−/− mice, an average increase of 10.75 supernumerary inner hair cells (ANOVA P < 0.0001; Tukey test P < 0.05) and 30 supernumerary outer hair cells is observed (ANOVA P = 0.002; Tukey test P < 0.05; Fig. 9C,G). Analysis of Sulf1−/−Sulf2+/− and the Sulf1+/−Sulf2−/− mice revealed the phenotype is not dose dependent. Sulf1−/−Sulf2+/− cochleae exhibit an average increase of 13.29 inner hair cells and 13.57 outer hair cells, and Sulf1+/−Sulf2−/− cochleae exhibit an average increase of 11.25 inner hair cells and 27.25 outer hair cells, neither of which differ significantly from the values we observed in the sulf1−/− cochleae or Sulf2−/− cochleae (Tukey test: nonsignificant). We compared the number of supernumerary hair cells observed in single knockout mice with those observed in the Sulf1−/−Sulf2−/− double knockout mice. The number of supernumerary inner hair cells observed in the double knockout cochleae is significantly greater than that observed in the Sulf1−/− knockout (Tukey test: P < 0.01), and in the Sulf2−/− knockout (Tukey test: P < 0.01), suggesting that both Sulf1 and Sulf2 contribute to inner hair cell numbers. The number of supernumerary outer hair cells in the double knockout cochleae only differs significantly from Sulf1−/− cochleae (Tukey test: P < 0.01), but does not differ from Sulf2−/− cochleae, suggesting that only Sulf2 expression affects outer hair cell numbers. Finally, we compared the gross morphology and cochlea lengths of the different genotypes, and found no significant differences between them (Fig. 9F,F′,G; cochlea length analysis: ANOVA P = 0.8). These data, along with the expression pattern of the Sulfs in several important cell types in the developing avian and mammalian inner ear, suggest that both Sulf1 and Sulf2 play important roles in the morphogenesis of the inner ear.
Discussion
Sulf Expression in the Hair Cells of the Inner Ear
Differentiation of the inner ear is precisely coordinated so that the specialized cell types that underpin its function are correctly arranged. Therefore, it is unsurprising that the Sulfs, with their ability to fine‐tune cellular responses to several signaling pathways, are expressed during avian and mammalian inner ear morphogenesis. In both the chicken and mouse inner ear, we have shown that Sulf1 and Sulf2 are expressed in complex, dynamic spatial and temporal patterns. Both Sulfs are expressed in various regions of the developing inner ear, suggesting they may function to regulate multiple signaling events during inner ear morphogenesis.
The most striking similarity in the bird and mouse inner ear is the expression of the Sulfs in the mechanosensory hair cells. In the vestibular sensory epithelia of both species, Sulf1 and Sulf2 are expressed continuously in the hair cell layer. In the developing cochleae, Sulf expression differs spatially and temporally. In the avian cochlea, there is a temporal wave of Sulf expression along the apical–basal axis (Fig. 6). cSulf1 is initially expressed in hair cells located at the apex of the basilar papilla, and over time its expression becomes activated in basally located hair cells whilst being down‐regulated in apical hair cells. cSulf2 exhibits a similar apical to basal temporal expression pattern, in both hair cells and supporting cells. This apex to base expression is similar to the wave of hair cell differentiation that occurs in the basilar papilla (Goodyear et al., 1995), and it is possible that the Sulfs play a role in hair cell differentiation. What signaling pathways are involved in the differentiation of mechanosensory hair cells in the avian and mammalian inner ear? Hair cells and supporting cells arise from prosensory progenitors as a result of notch signaling from putative supporting cells acting upon neighboring Delta1 or Jag2 expressed in potential hair cells (Adam et al., 1998; Lanford et al., 1999; Daudet and Lewis, 2005). This ensures that only some prosensory cells retain the expression of the transcription factor Atoh1, which is both necessary and sufficient for hair cell differentiation (Bermingham et al., 1999). There is some evidence that HSPGs play a role in Notch signaling in Drosophila (Kamimura et al., 2004), and also during skeletal muscle satellite cell homeostasis (Pisconti et al., 2010). However, to date, the Sulfs have not been implicated in the Notch signaling pathway.
Schematic of Sulf1 and Sulf2 expression in the developing sensory epithelium of the avian basilar papilla. Sulf1 and Sulf2 expression temporally shifts from the apex to the base of the basilar papilla. After transient expression in the hair cells at the apex of the basilar papilla at E8, Sulf1 is up‐regulated in the hair cells at the base and down‐regulated at the apex by E11. At E8, Sulf2 is expressed in supporting cells across the length of the basilar papilla, but is restricted to hair cells located in the middle and the base. By E11, Sulf2 is down‐regulated in the apical supporting cells and the hair cells in the mid‐section of the cochlea. HC, hair cell layer; SC, supporting cell layer.
mSulf1 expression in the developing mouse inner ear. A: Expression first appears in the E12.5 inner ear on the ventro‐lateral wall of the developing otocyst. Expression in the floorplate of the neural tube is also seen (fp). B: At E14.5, mSulf1 is expressed throughout the length of the cochlea and also weakly in the spiral ganglion (dotted outlines). B′: Magnified view of the middle turn of the cochlea shows mSulf1 expression across the roof of the cochlea duct (cr), and an antero–medial portion of the medial wall (mw; expression marked by line). C–C″: mSulf1 is expressed in the hair cells of the cristae (superior crista shown, sc) and the utricle (u) of the E17.5 inner ear. D: At E17.5, mSulf1 expression remains in the spiral ganglion (dotted outlines) and in the lateral wall throughout the length of the cochlea. D′–D″: High magnification view of the E17.5 cochlea shows mSulf1 is also expressed in the outer hair cells but not the inner hair cells (OHC, IHC, labeled with myosin VIIa in D′). E: In the P0 inner ear, mSulf1 expression remains in the spiral ganglion (dotted outlines), and is expressed throughout the length of the cochlea. E′–E″″: mSulf1 is expressed in the hair cell layers of the cristae (E′, lateral crista shown), saccule (E″), utricle (E″′) and the inner and outer hair cells of the cochlea (E″″). Expression is also present in the epithelial cell layer of Reissner's membrane (rm) and the marginal cells of the stria vascularis (sv). Apical, middle, and basal turns of the cochlea are shown as (a), (m), and (b), respectively.
mSulf2 expression in the developing mouse inner ear. A: Expression is first seen at E12.5 in the ventro–medial wall of the developing otocyst. B: At E14.5, mSulf2 is expressed in the cochlea ganglion (dotted outlines) and throughout the length of the developing cochlea. B′: In the cochlea, mSulf2 is expressed in a posterior part of the roof of the cochlea duct (cr; expression domain marked by line) and also a posterior section of the medial wall of the cochlea (mw; expression domain marked by line). C–C″: In the vestibular apparatus, mSulf2 is expressed in the hair cell layer of the utricle (ut) and the crista (superior crista shown, sc; myosinVIIa labels hair cells). D–D″: mSulf2 expression continues in the vestibular hair cells at E17.5. (E) In the E17.5 cochlea, mSulf2 is expressed in the cochlea ganglion cells (dotted lines) and along the length of the cochlea. E′–E″: High magnification views show mSulf2 is expressed in both the outer and the inner hair cells (OHC; IHC), and the lateral wall (lw). F: At P0, mSulf2 expression remains localized to the hair cell layers of the vestibular organs (crista, sc, and utricle, ut, shown). F′: In the cochlea, mSulf2 is expressed in the epithelial cell layer of Reissner's membrane (rm), in the inner hair cells (IHC) and outer hair cells (OHC), and weakly in the marginal cells of the stria vascularis (sv). Apical, middle and basal turns of the cochlea are labeled (a), (m), and (b), respectively.
Sulf1−/− Sulf2−/− knockout mice exhibit increases in supernumerary inner and outer hair cells in the cochlea. A: P0 wild‐type cochlea stained for F‐actin with supernumerary inner hair cells (red arrowheads) and supernumerary outer hair cells (yellow arrowheads and lines) marked. A′: Magnified section of the wild‐type cochlea with supernumerary outer hair cells (yellow arrowheads) shown. B: P0 Sulf1−/− cochlea stained for F‐actin with supernumerary inner hair cells (red arrowheads) and supernumerary outer hair cells (yellow arrowheads and lines) marked. (B′) Magnified section of the Sulf1−/− cochlea with supernumerary inner hair cells (red arrowheads) shown. C: P0 Sulf2−/− cochlea stained for F‐actin with supernumerary inner hair cells (red arrowheads) and supernumerary outer hair cells (yellow arrowheads and lines) marked. C′: Magnified section of Sulf2−/− cochlea showing supernumerary inner hair cells (red arrowheads) and supernumerary outer hair cells (yellow arrowheads). D: P0 Sulf1−/−Sulf2−/− cochlea stained for F‐actin with supernumerary inner hair cells (red arrowheads) and supernumerary outer hair cells (yellow arrowheads and lines) marked. D′: Magnified section of the Sulf1−/−Sulf2−/− cochlea showing the supernumerary inner hair cells (red arrowheads) and supernumerary outer hair cells (yellow arrowheads) in more detail. Low magnification images are composites from multiple confocal images. Yellow lines denote 3 or more supernumerary outer hair cells in the marked region. E: Bar chart showing the average number of supernumerary inner and outer hair cells for wild‐type, Sulf1−/−, Sulf2−/− and Sulf1−/−Sulf2−/− cochleae. N= 8. Standard deviation bars are shown. Asterisks denote statistical significance between genotypes. F,F′: Paint filled inner ears from wild‐type (F) and Sulf double knockout (F′) mice. G: Bar chart showing the average cochlea length for wild‐type, Sulf1−/−, Sulf2−/−, and Sulf1−/−Sulf2−/− cochleae. N = 4, standard deviation bars are shown.
If the expression of the Sulfs in hair cells does indicate a role in hair cell differentiation, we might expect to observe a decrease in hair cell number in the absence of Sulf expression. However, in the Sulf1−/−Sulf2−/− mice, we observed an increase in the number of supernumerary hair cells, suggesting that, in the mouse at least, the Sulfs are not important regulators of hair cell differentiation. It is worth noting that Sulf expression in the auditory hair cells of the mouse cochlea does not appear to follow the transient expression pattern along the length of the cochlea that is observed in chick, despite the fact that a wave of hair cell differentiation also occurs in the mouse organ of Corti, albeit a basal to apical wave, in reverse of that observed in chick (Chen et al., 2002). This discrepancy in the expression pattern between the chick and mouse auditory sensory epithelia could suggest a difference in Sulf function in the two organs, and it will be interesting to perform loss‐of‐function experiments in the chick basilar papilla to see if they present a phenotype similar to that of the mouse.
Future, functional studies in the mouse should initially focus on the signaling pathways that govern the early stages of hair cell specification. The first step toward hair cell specification occurs with the induction of a population of “prosensory” cells, from which hair cells and supporting cells will differentiate. Sensory competence begins at E11.5, when the medial wall of the cochlea undergoes a series of subdivisions that are governed by several different signaling morphogens. As development progresses, signaling molecules progressively restrict prosensory identity. At E15.5, when the first inner hair cells begin to differentiate, only the most central portion of the medial cochlea wall is considered prosensory, and forms the cells of the organ of Corti. The cells that flank either side of this region develop into the greater epithelial ridge and the outer sulcus (Groves and Fekete, 2012). Altering the signaling pathways that control this restriction of prosensory identity causes changes in the number of hair cells in the mature cochlea. BMP concentration gradients, for example, influence cell fate decisions, with high levels favoring outer sulcus cell fates, intermediate levels favoring hair cells fates and low levels inducing markers of Kolliker's organ (Ohyama et al., 2010). FGF signaling is also important in establishing these domains within the medial wall of the cochlea. FGF receptor‐1 (FGFR1) activation regulates prosensory identity by maintaining levels of the prosensory marker, sox2, in the medial wall, and loss of FGFR1 expression results in a dramatic loss of hair cells (Pirvola et al., 2002; Ono et al., 2014). During the period of prosensory specification, Sulf1 is expressed in the anterior–medial region of the cochlea wall. Sulf2, conversely, is expressed in the posterior region of the medial wall, so that together, the expression of the Sulfs flanks either side of the prosensory domain. This is particularly interesting, as inner hair cells are derived from progenitors located toward the anterior of the medial wall, and outer hair cells from those toward the posterior of the medial wall. It is possible that a loss of Sulf1 or Sulf2 expression alters the efficacy of diffusible morphogens and shifts the anterior and posterior boundaries of the prosensory domain respectively. Such a shift could potentially explain the distinct increase in supernumerary inner or outer hair cells we observe in the single knock‐out mice, as well as the increase in supernumerary inner and outer hair cells in the double knock‐out mice.
Expression of Sulfs in Nonsensory Regions of the Inner Ear
The expression of the Sulfs in the spindle‐shaped cells, acoustic ganglia, stria vascularis, and Reissner's membrane also warrants attention in future studies. Spindle‐shaped cells mark the path of neurites through the cartilaginous plates located between the sensory epithelium and the acoustic ganglia, and have been implicated in axonal guidance (Heller et al., 1998). Correct axon guidance is critical to normal inner ear development, as it establishes a tonotopic connection between the inner ear and the brain. Previously, Sulf1 and Sulf2 have been shown to play a crucial role in neural crest migration in Xenopus tropicalis (Guiral et al., 2010), and thus there is the intriguing possibility that they play a role in another migratory process, axon guidance, in the inner ear. A careful study of the innervation patterns of the sensory patches, using markers of afferent and efferent neurons, will shed further light on this in the future. The stria vascularis and Reissner's membrane are both epithelial cell membranes that are involved in ion transfer between the various fluid cavities of the mouse cochlea. The stria vascularis is a layered epithelium that produces the endolymph that fills the scala media. Reissner's membrane separates the scala media from the scala vestibuli, and functions to maintain ion concentrations between the endolymph and perilymph of the two compartments. HSPGs on Reissner's membrane have been previously proposed as regulators of the perilymph–endolymph permeability barrier, by creating an anionic charge on the cells that contact the endolymph of the scala media (Torihara et al., 1995). In the stria vascularis, the presence of heparan sulfate has also been proposed as an important regulator of ion concentrations in the endolymph (Torihara et al., 1994; Tsuprun and Santi, 2001). The expression of the Sulfs in these structures suggests that their modification of HSPG structure may play a role in establishing these ionic concentration boundaries. In vivo assays of endolymph and perilymph ionic concentrations, as well as measurements of the endocochlear potential in Sulf knockout mice will be essential to elucidate if the Sulfs do play such a role. However, the double knock‐out mice used in this study do not survive beyond 24 hr after birth (Nagamine et al., 2012). Generation of an otic specific, conditional knockout mouse line is essential to perform a complete functional study.
Experimental Procedures
Animals
Fertilized White Leghorn (Gallus gallus) eggs were obtained from Shiroyama Farm, Kanagawa, Japan and incubated at 37°C and 50–70% humidity. Embryonic stages are given in embryonic days (E), with E1 corresponding to 24 hr of incubation. Embryos older than E5 were killed by decapitation.
Sulf‐deficient mice were generated as previously described in (Nagamine et al., 2012). Mice were housed in accordance with local and national guidelines for animal experiments.
In Situ Hybridizations
Whole‐mount in situ hybridization for chicken inner ear samples was performed as described in (Daudet and Lewis, 2005). For mouse inner ear samples, in situ hybridizations were performed on 18 micron cryosections (Ladher et al., 2000).
Riboprobes
The antisense chick Sulf1 probe was kindly provided by Dr. Cathy Soula (García‐López et al., 2009). Antisense chick Sulf2, mouse Sulf1 and mouse Sulf2 probes were synthesized from cDNA fragments generated using the following primers: cSulf2_for AGTGCAAGGGAATGGTGAAC, cSulf2_rev ATTTGGGGCTGCATACTCTG; mSulf1_for AAGAGCCTGGACATTGGAGC, mSulf1_rev CAGATGCAGGGTTTGGAGGT; mSulf2_for GGCTAAGCGGCCA
TAGAGAG, mSulf2_rev GGATGTCTGGTTCTCGGCTC. The cDNA fragments were sequenced for verification and were then cloned into a pCR® 2.1‐TOPO® TA vector.
Cryosections
Fixed specimens were rehydrated, and then cryopreserved by immersion in sequential graded sucrose solutions of 10%, 15%, and 20% in phosphate buffered saline (PBS). Samples were then washed in a 1:1 solution of 20% sucrose and TissueTek™, before being embedded in TissueTek™, frozen in liquid nitrogen and stored at −80°C. Frozen sections (18 μm) were collected using a Microm HM525 Cryostat, mounted on SuperFrost Plus™ slides (Microm) and images were taken using a Leica DM5000 B fluorescent light microscope.
Immunocytochemistry
Immunocytochemistry was performed using the following antibodies: rabbit polyclonal anti‐Myosin VIIa (Proteus Biosciences 25–6790) and Alexa fluor 546 goat anti‐rabbit (Invitrogen A11010). Samples were analyzed on a Leica DM5000B microscope and a Zeiss LSM780 confocal microscope. For mutant mouse cochlea experiments, P0 mouse heads were fixed in 4% paraformaldehyde overnight at 4°C. The cochleae were dissected and washed in PBS+0.3% Triton X‐100 for 30 min. The cochleae were then incubated in PBS+Alexa Fluor 568 phalloidin (Invitrogen) for 1 hr to visualize f‐actin. Samples were visualized using a Zeiss LSM780 confocal microscope. Images were processed using imageJ software.
Paint Filling
Paint filling was performed using the protocol described in Bissonnette and Fekete (1996).
Calculating Supernumerary Hair Cell Numbers and Cochleae Lengths
High‐resolution images were created by combining multiple, separate confocal images to create a composite image of the entire length of the sensory epithelium. Supernumerary inner and outer hair cells were identified along the length of the sensory epithelium. As the normal pattern of the murine sensory epithelium consists of a single row of inner hair cells, and three rows of outer hair cells, any instance in which inner and outer hair cells were present in addition to this pattern were counted as supernumerary hair cells. Cochleae lengths were measured using imageJ by tracing a line along the length of the sensory epithelium using the inner hair cells as a guide.
Acknowledgments
We thank Dr. Cathy Soula for providing the template for synthesis of the cSulf1 riboprobe. A JSPS Post‐Doctoral Fellowship for Overseas Researchers funded this work.
