We first determined the transcript expression level of 24 claudin isoforms in the whole cochlea of normal (Slc26a4+/−) mice at three different ages after birth: P2 and P6 before the onset of endocochlear potential generation and hearing in mice, and at P15, after acquisition of hearing. Transcripts of 21 claudin isoforms were detected at all ages, while 3 isoforms (Cldn-16, − 17 and − 18) were not detected (Fig. 2). The permeability properties of several isoforms have been unambiguously determined [17, 18] and are shown in (Fig. 2), as described in the figure legend. Cldn-10  and Cldn-19 were each determined for two splice variants, v1 and v2 (Table 1).
Six cochlear claudin isoforms increase with development at P6 and/or P15: Cldn-9, − 13, − 14, − 15, and -19v2. By contrast, Cldn-6 expression decreases with development. Cochlear claudins that do not change significantly with development include: Cldn-1, − 2, − 3, − 4, − 5, − 7, − 8, −10v1, −10v2, − 11, − 12, −19v1, − 20, − 22, and − 23. As described above, endocochlear potential normally develops between P6 and P15. So genes that change their expression in this period might be involved in establishment of the special properties of the paracellular barrier of the epithelial cells that border the endolymph, and thereby provide the resistive barrier that supports the large endocochlear potential. Cldn-19v2 appeared to increase expression only transiently during this period. The post-natal changes in expression of multiple claudin isoforms are consistent with the likely presence of factors that regulate claudin expression during development. Most striking of all, Cldn-13 shows a remarkably large increase in cochlear expression compared to the others. Previously, Abuazza et al.  reported maturational decrease of Cldn-6, − 9 and − 13 transcripts and of paracellular protein in several segments of the mouse kidney. They suggested these changes may contribute to developmental changes in the paracellular permeability of kidney tubules. In our study of the cochlea, Cldn-6 undergoes developmental decrease in transcript expression from P2 to P6, and further from P6 to P15, as in the kidney (Fig. 2). By contrast to the kidney, Cldn-9 and -13 transcripts increased from P2 to P6, and further from P6 to P15 (Fig. 2).
The cochlear tissues expressing these claudins were resolved in two subsequent experimental series. In the first series (Fig. 3a), cochleae of adult (P18-P32, mean P22.6) Slc26a4+/− mice were subdivided into three fractions: 1) lateral wall (exclusive of outer bone), 2) medial region, and 3) outer bone. These fractions were assayed for 12 claudin isoforms: Cldn-5, − 6, − 7, −10v1, −10v2, − 13, − 15, −19v1, −19v2, − 20, − 22, and − 23. All of these claudins were detected both in the lateral wall and the medial region fraction. Cldn-19v1 and Cldn-19v2 were expressed most strongly in the medial region.
Interestingly, Cldn-13 was expressed virtually exclusively in the outer bone fraction, in spite of the statistically significant difference in the minimal expression in the two soft tissues. Wongdee et al. examined claudin expression in skull and tibia bone  and determined localization of Cldn-5, − 11, − 14, − 15 and − 16. The expression was limited to the cells lining the bone (periostieum), suggesting a function of claudin other than tight junction formation. They, however, did not test bone for the presence of Cldn-13. Johnson et al. reported Cldn-13 expression in G1E cells, a proerythroblastic cell line  and Cldn-13 was identified in a stress induced erythropoiesis pathway that is mainly expressed in tissues associated with haematopoietic function . It is therefore likely that expression of Cldn-13 in cochlear outer bone might originate from the associated bone marrow, which develops during the early postnatal period [15, 24]. In support of this proposition, it was found that Slc26a4−/− mice exhibit delayed bone marrow maturation between P6 and P15 . Mouse Cldn-13 does not have a human homolog .
In the second series, cochleae of adult (P19-P28, mean P22.0) Slc26a4+/− mice were subdivided further into four micro-dissected fractions: 1) the stria vascularis and 2) spiral ligament fractions were separated from the lateral wall; 3) the organ of Corti and the 4) modiolus were separated from the medial structures. Claudins in the lateral wall and medial fractions that gave high expression signals (claudin mRNA/18S > 4.5) in the first experimental series (Fig. 3a) were analyzed in the more-finely separated tissues of the second series (Fig. 3b). The epithelial fractions (stria vascularis and organ of Corti) were found to express Cldn-7 more strongly than their respective primarily non-epithelial fractions, spiral ligament (fibrocytes) and modiolus (neurons). By contrast, the other six isoforms did not show statistically significant differences between the epithelial fractions and their respective adjacent non-epithelial fractions. Non-significant comparisons are not shown in Fig. 3b and comparisons other than stria vascularis – spiral ligament and organ of Corti – modiolus are given in the Additional file 1.
Three claudins were selected to investigate the possible effect of Slc26a4 gene deletion on inner ear developmental expression of claudins. Developmental expression of the three isoforms demonstrated a dramatic postnatal increase in Cldn-13 that was not characteristic of the other two claudins, consistent with the notion that Cldn-13 is not regulated by a mechanism common to the claudins highly expressed in the epithelial tissues. We examined RNA from whole cochleae from age- and sex-matched littermates of Slc26a4+/− and Slc26a4−/− and analyzed by two-way ANOVA 1) Cldn-11, which is expressed in basal cells of stria vascularis , and whose deletion in mice causes hearing loss, 2) Cldn-13, which is expressed in cochlear outer bone (this report), and 3) Cldn-14, which is expressed in organ of Corti and is responsible for human hereditary deafness DFNB29. The results of analysis (Fig. 4) showed no statistically significant interaction between age and genotype in all three genes and no further comparisons of individual paired genotypes were made.
“Cldn-21” was not included in this study. The nomenclature has varied and developed since 2001 and was not identified in mouse at the time of this study . The mouse gene currently accepted as Cldn-21 [25, 26] has been heterologously expressed in MDCK epithelial cell cultures, immunolocalized to sites that also express the tight-junction protein occludin, and was shown to participate in a Na+-selective paracellular transport pathway .
Some of our data differ from previous observations: expression of Cldn-5, − 6 and − 15 was not detected by Kitajiri et al. , but were observed in our experiments. In kidney, Cldn-5 and -15 are expressed in endothelial cells, not epithelial cells . By contrast, Kitajiri et al.  reported that there had been no expression of Cldn-5 and Cldn-15 in stria vascularis nor spiral ligament, both highly vascularized tissues. We found Cldn-6 expression, but it gradually decreased during early development. Consistent with our observation, Kitajiri et al.  did not see any expression of Cldn-6 in the adult cochlea.