The cochlear scalas are differentiated from a single tube with a lining by the tall epithelium, that is, the cochlear duct. However, we have no information about the mechanism involved in formation of the scalas. Transverse sectioning of each coil has shown that the membranous labyrinth of the inner ear cochlea is composed of three regions: the scala vestibuli at the apical side, the scala tympani at the basal side, and the scala media containing the organ of Corti. The scala media contains “endolymphatic fluid” (or endolymph), whereas the scala vestibuli and the scala tympani contain “perilymphatic fluid” (or perilymph). The difference in terms might be based on the morphology in the history of research, such as the “endo”-lymph attaching to the organ of Corti in contrast to the “peri”-lymph separated from the specific nerve terminal. The endolymphatic fluid is known to resemble the intracellular fluid with high potassium ion and relatively low sodium (Williams,1995). Hirt et al. (2010) reported the difference in osmotic pressure: 322 mOsm/kg H2O in the endolymphatic fluid; 289 mOsm/kg H2O in the perilymphatic fluid.
According to Streeter (1917), the scala tympani forms before the scala vestibuli, with scala formation occurring at 8–17 weeks of gestation. At these stages, the cochlea is composed of the basal and second coils, with the apical part added after 15 weeks (Arnold and Lang,2001; Jeffery and Spoor,2004; Yasuda et al.,2007). Although the cochlear nerve reaches the cochlear duct at 11 weeks (Yamashita et al.,1993), myelination starts much later than the stages (Ray et al.,2005). Before the differentiation of the organ of Corti in fetuses, and, in some cases, the established scala media, the fetal scala media is called the “cochlear duct” (Fig. 1), in which tall epithelial cells separate the potassium-rich endolymphatic fluid from the surrounding sodium-rich perilymphatic fluid. Thus, in the scala vestibuli and scala tympani, these spaces or subdivisions are called the “perilymphatic space,” in contrast to the “endolymphatic space,” which corresponds to the cochlear duct and its subdivisions, including the organ of Corti and inner tunnel. In the previous descriptions on the early development of the scalas in the cochlea (Hamilton and Mossman,1978; Moore and Persaud,1998), there seems to be no agreement as to whether the early scalas carry the definite mesothelial lining (Fig. 1A–C) or they are formed by the fusion or coalescence of irregularly shaped perilymphatic spaces or vacuoles without lining mesothelium (Fig. 1D–F). However, we considered this point as a clue to provide better understanding of how the cochlear scala forms.
In the inner ear immunohistochemistry, previous researchers have paid attention to connexin-26 in the organ of Corti because the molecule seems to be responsible for hearing loss (Kikuchi et al.,1994; Fish et al.,2001; Kammen-Jolly et al.,2001). However, there has been no or few immunohistochemical study on the cochlear mesenchymal cells around the scala. Therefore, we used immunohistochemistry to examine the human fetal membranous labyrinth, using frontal or sagittal sections, with special emphasis on the formation of the scalas. The immunohistochemistry would help us to demonstrate site-dependent differences in cell differentiation along the scala. The major question is what and how makes the fluid space scala without definite epithelia such as the cochlear duct.
MATERIALS AND METHODS
We used paraffin-embedded specimens of 20 fetuses, eight at 8–9 weeks of ovulational age [early stage; crown–rump length (CRL) 28–45 mm], eight at 11–12 weeks (middle stage; CRL 52–74 mm), and four at 14–15 weeks (late stage; CRL 90–110 mm). All specimens were part of the large collection kept at the Embryology Institute of the Universidad Complutense, Madrid and were the products of miscarriages and ectopic pregnancies at the Department of Obstetrics of the University. The study was approved by our university ethics committee and was performed in accordance with the provisions of the Declaration of Helsinki 1995 (as revised in Edinburgh, 2000). Because the specimens were taken from the products of miscarriages and ectopic pregnancies, they may have had abnormal pathology. However, we have attempted to describe the morphology commonly present in each group.
After routine procedures for paraffin-embedded histology, most of the specimens were cut frontally at a thickness of 5 μm and at intervals of 20 μm. However, sagittal sections were made for one specimen in each group of early, middle, and late stages. Depending on the size of each specimen, we needed to examine approximately 30–100 sections, including almost the entire cochlea, of each. Most sections were stained with hematoxylin and eosin, while some sections in all stage groups were used for immunohistochemistry (see below).
The primary antibodies used were (1) rabbit monoclonal anti-human S100 protein (dilution 1:100, Dako Cytomation, Kyoto, Japan), (2) rabbit monoclonal anti-human aquaporin-4 (dilution 1:50, Santa Cruz Biotechnology, Santa Cruz, CA), (3) rabbit monoclonal anti-human CD68 (dilution 1:100, Dako, Glostrup, Denmark), (4) rabbit monoclonal anti-human alpha-1 smooth muscle actin (dilution 1:100, Dako); (5) mouse monoclonal anti-human CD34 (dilution, 1:100; Dako, Glostrup, Denmark); (6) mouse monoclonal anti-human podoplanin or D2-40 (dilution, 1:100; Nichirei, Tokyo), and (7) mouse monoclonal anti-human cytokeratin 19 (dilution, 1:100; Dako, Glostrup, Denmark). In most cases, paraffin sections were not pretreated, including microwave treatment. However, for D2-40, we used a ligand activator (Histofine SAB-PO Kit, Nichirei, Tokyo, Japan) with autoclave treatment (105°C, 10 min). Using Dako Envision ChemMate, the second antibody was labeled with horseradish peroxidase (HRP), and antigen–antibody reactions were detected using the HRP-catalyzed reaction with diaminobenzidine (with hematoxylin counterstaining). In addition, to identify cell death with DNA fragmentation, we conducted the deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick end labeling (TUNEL) method using Millipore ApopTag Plus Peroxidase in situ Apoptosis Kit (Merck, Darmstadt, Germany).
We assayed fetal specimens for expression of S100 protein, a marker of glial cells (Ludwin et al.,1976) and of developing cartilage and ossification (Chano et al.,1996; Duarte et al.,2003). Foster et al. (1994) and Yamashita et al. (1995) reported that S100 protein is a maturation marker of the fetal inner ear membranes. We also assayed the specimens for expression of the water channel protein aquaporin-4, expressed by specific cells in the organ of Corti (Takumi et al.,1998), CD-68, a macrophage marker, and smooth muscle actin, which is expressed by the endothelium of arteries and veins, smooth muscle cells, and cochlear pericytes, but not by lymphatic endothelium (Hayashi et al.,2008; Shi et al.,2008). CD34 is a famous marker for the vascular and mesenchymal progenitor cells (Xu et al.,2003; Zambidis et al.,2007). D2-40 is a limited but effective marker for the mesothelium and lymphatic epithelium (Ishikawa et al.,2006; Jin et al.,2010). Finally, cytokeratin 19 is generally expressed in epithelial or epithelium-like cells (Moll et al.,1982; Makin et al.,1984). To our knowledge, there was no previous data on immunohistochemistry of the membranous labyrinth using CD34, D2-40, or cytokeratin 19.
Early-Stage Group (CRL 28–45 mm)
Examination of the cochleae showed that the second loop had already formed, even in the smallest specimen (CRL 28 mm). The developing nerve elements were identified as mesenchymal cell condensations, reaching the cochlear duct before the formation of scalas (Fig. 2A). However, ganglion cells were not yet differentiated. In four of eight specimens in this group, we observed irregular perilymphatic spaces or vacuoles and their fusion or coalescence in the basal coil. All four of these specimens had a large space after fusion in the future scala tympani, and two of four had another smaller space in the future scala vestibuli (Fig. 2B). Notably, in two specimens of this group (CRL 29 mm and CRL 45 mm or Fig. 2B), we found a small gap in the cochlear duct epithelium at a site facing the future scala tympani (see also the description of the “gap” at later stages). This epithelial gap was attached to the mesenchymal condensation or nerve element. The tectorial membrane was not seen in this group.
Middle-Stage Group (CRL 52–74 mm)
In all the eight specimens, we found irregular perilymphatic spaces or vacuoles and their fusion or coalescence at the basal coil around at least one cut surface of the cochlear duct. In five of the eight specimens, the large space present after fusion was restricted in the basal coil and on one side of the second coil (Fig. 3A). However, in the remaining three specimens, there was a large space in all sections of the coils, from the basal site to the apical site (Fig. 3B). All these spaces were present in pairs, corresponding to the future scala tympani and scala vestibuli. The future scala tympani was not always larger than the future scala vestibuli (Fig. 3A). A gap in the cochlear duct epithelium was almost always (7/8) present at a site facing the scala tympani (Figs. 3 and 4). The gap was located at and opened to the margin of the mesenchymal condensation or nerve element. The cochlear duct epithelial cells appeared to disperse at the edges facing the gap. In contrast to cells in the epithelium, the dispersed cells appeared round and contained little cytoplasm (Fig. 4A). The tectorial membrane appeared at the basal coil in only two of these eight specimens possibly due to its formation following the complete differentiation of the organ of Corti (Fig. 4B), but it did not extend laterally (i.e., to the bony side) to reach the level of the epithelial gap. Along the scala wall, thin vessels containing red blood cells were present.
Late-Stage Group (CRL 90–110 mm)
The tectorial membrane was observed in all four of these specimens (Fig. 5A). We were able to identify the organ of Corti in all four: it was composed of (1) the tectorial membrane and (2) extremely tall supporting cells with the nuclei at the basal side, but (3) no inner tunnels or pillars (Fig. 5B). The supporting cell layer differed in cell height from that in other parts of the cochlear duct epithelium. Notably, the gap formation of the cochlear duct epitelium was consistently followed by the aforementioned differentiation of the organ of Corti. We found that at this stage, Reissner's and basilar membranes were differentiated from the cochlear duct epithelium: the former was thin but still contained two to three layers. However, mesothelial lining of the scalas tympani and vestibuli was not yet complete, with the surrounding mesenchymal cells appearing to face the scala at many sites. All these steps appeared to start in the basal coil, afterward extending to the apical side of the cochlea.
Among the three staged groups, the late-stage group specimens showed the best staining, with good preservation of tissues, although the histological procedure might usually make the tectorial membrane detached (Fig. 6). Cytokeratin 19 started expression in the early-stage group, and in the middle- and late-stage groups, it was positive in (1) the apical margin of the tall epithelial cells of the cochlear duct (Fig. 6A), (2) thin epithelia of the semicircular ducts, and (3) the epithelium of the middle ear and the skin. S100 protein was expressed in the basilar membrane, including the cochlear nerves, scala walls, and mesenchymal cells along and near the inner ear cartilage but not expressed in the organ of Corti, including in the scala wall facing the cartilage and containing thin vessels (Fig. 6B). In contrast, most of the supporting cells of the organ of Corti and in the future stria vascularis were positive for aquaporin-4 (Fig. 6C). Aquaporin was also expressed in the vascular endothelium along the scala wall near the cartilage. These thin vessels were negative for smooth muscle actin in contrast to vessels in and along the spiral ganglion. All these vessels were positive for CD34 (not shown), but loose mesenchymal cells were negative around the developing scala (data will be shown in the separated paper). Cells along the epithelial gap were negative for both S100 protein and aquaporin-4. Abundant CD68-positive macrophages were present in loose mesenchymal tissues around the scala (Fig. 6D) and along the cochlear nerve. D2-40 did not express in the membranous labyrinth including the loose mesenchyma in and around the scala, but it was positive in the cartilaginous cochlea (figure, not shown). We found no positive cells in the cochlea using TUNEL method although we tried twice.
Our findings indicate that, in the human fetal cochlea, scala formation occurs via fusion of small perilymphatic spaces. In contrast to the fetal cochlear duct, in which epithelial cells are tightly attached to each other, the primitive scala carries no definite mesothelial lining, even during later stages. A gap junction protein, connexin 26, which has been associated with many types of hearing loss, acts by maintaining high concentrations of potassium ions in the endolymph (Kikuchi et al.,1994). In human fetuses of 11–31 weeks gestation, connexin 26 is expressed in the mesothelial layer of the scala vestibuli, stria vascularis, and supporting cells of the organ of Corti (Kammen-Jolly et al.,2001). The mesothelial lining for scala perilymph seems to form much later than the epithelial lining system for endolymph although the lining was negative for a mesothelial marker D2-40. Mesothelial development likely follows the development of aquaporin-positive thin blood vessels along the scala walls starting at 11–12 weeks of gestation. In contrast to the scala wall, the epithelium of the cochlear duct expressed cytokeratin 19. This cytokerain seems to be available for a marker to identify a difference between the cochlear duct epithelium and the other scala walls. Conversely, until the middle stage, the scalas tympani and vestibuli are unlikely to carry the definite lining cells. Actually, Reissner's and basilar membranes were differentiated during the late stage.
One of our most striking findings was the observation of a gap in the cochlear duct epithelium, which was present in more than half of the cut surfaces (in number) of the cochlear ducts at and around CRL 50–55 mm. The earliest such finding was in the basal coil of a fetus with CRL 29 mm, suggesting that the same shape and site are maintained throughout the stages examined. Composite cells appeared to disperse from both edges of the cochlear duct epithelium facing the gap, possibly due to the cell death process (see below). In fetuses at 14–15 weeks of gestation, the epithelial gap delimited a lateral or bony side of S100 protein-negative organ of Corti. Thus, due to the topographic relationship between the gap and organ, the epithelial gap most likely corresponds to a site containing Hensen's cells, a supporting cell population (Fawcett,1994). Although Hensen's cells have been reported to strongly express the water channel protein aquaporin-4 (Takumi et al.,1998), we found that cells along the gap were “selectively” negative for aquaporin, suggesting that Hensen's cells, which act as a plug in the epithelial gap, develop during the latest stages of cochlear development. Although, we originally regarded the epithelial gap as the primitive inner tunnel of the organ of Corti, we found that the tectorial membrane never extended laterally to reach the gap. Therefore, we hypothesize that this epithelial gap is the first sign of the formation of the organ of Corti, suggesting that this organ starts to develop at the same fetal stage as the start of scala formation.
To our knowledge, Streeter (1917) was the first to describe the appearance of small perilymphatic spaces or vacuoles around the fetal semicircular duct, as well as their fusion or coalescence in the loose mesenchymal tissue. In the developed inner ear, the duct is filled with endolymphatic fluid, while the mesenchymal cells are embedded in the perilymphatic fluid. Thus, due to the specific function of the developing duct epithelium, even at 8–15 weeks of gestation, there seems to be a difference in osmotic pressure and ion contents between endolymph and perilymph (Hirt et al.,2010). Due to the presence of an epithelial gap, starting during early steps of scala formation, it is likely that the endolymph leaks through this gap, causing mesenchymal cell death and resulting in the coalescence of vacuoles containing perilymph. Moreover, this gap is likely to be closed by cells strongly positive for aquaporin-4 (Hensen's cells). Because the cochlear nerve may induce the organ of Corti (Takebayashi et al.,2007) and the organ physiologically faces the endolymphatic space (i.e., the cochlear duct and inner tunnel), the developing nerve element passing through the mesenchymal tissue should be resistant to exposure to endolymph. There is therefore a question regarding the type of cell death occurring in the mesenchymal cells surrounding the cochlear duct.
In the inner ear, apoptosis has been shown to occur in epithelial cells and in epithelium-derived cells in the surrounding mesenchyme (Represa et al.,1990; Nishokori et al.,1999). The present examination of mesenchymal tissues in the cochlea showed no TUNEL-positive mesenchymal cells, although Yasuda et al. (2007) reported apoptotic body-like fragments in the cochlear duct epithelial cells as well as in the surrounding early-stage mesenchymal tissue. TUNEL-positive mesenchymal cells have been reported in the middle ear but not in the inner ear (Palva et al.,2003). Although acid phosphatase-positive cells were observed along the bony or cartilaginous wall of the cochlea (Anderson et al.,1969), they were later shown to be osteoclast-like cells used for remodeling of the hard tissue (Jin et al.,2010). However, we observed abundant macrophages in cochlear mesenchymal tissues along and around the scala. Overall, mesenchymal cell death during scala formation may occur according to a TUNEL-negative mechanism, for example, without DNA fragmentation (e.g., Castro-Obregón et al.,2002), due to exposure to the endolymph. We believe that, after the initial endolymph leaks through the epithelial gap into the scala tympani, a potassium flow can quickly reach the apex of the coil to extend into the scala vestibuli. All nerve elements near the scala walls should be safe becasue of no toxicity of high potassium against nerve cells.