Breeding pairs of Slc26a4^-/- and Slc26a4^+/+ mice were obtained from the colony of Dr Susan Wall (Emory University, Atlanta, GA, USA) to establish a new colony at KSU. Mice used for this study were anaesthetized either with 4% tribromoethanol (0.014 ml/g body weight i.p.) or pentobarbitol (0.1 mg/g body weight, i.p.) and sacrificed by decapitation or by transcardial perfusion. Transcardial perfusion consisted of Cl^-free solution (6 ml, 1 min) followed by Cl^- free solution containing 4% paraformaldehyde (24 ml, 4 min). Cl^- free solution contained mM 150 mM Na-gluconate, 1.6 mM K₂HPO₄, 0.4 mM KH₂PO₄, 4 mM Ca-gluconate₂, 1 mM MgSO₄ and 5 mM glucose, pH 7.4.

Mice that either express (Slc26a4^+/+ or Slc26a4^+/-) or lack (Slc26a4^-/-) a functional pendrin gene were studied at four developmental stages, before and after the age that marks the onset of hearing at postnatal day 10 (P10) and P15, respectively, after weaning (P30-41) and adult (P74-170). Genotypes were determined by PCR as described previously 20. All procedures involving animals were approved by the Institutional Animal Care and Use Committee of Kansas State University.

Temporal bones from age matched Slc26a4^-/- and Slc26a4^+/- mice were rendered blood free and fixed by transcardial and perilymphatic perfusion with Cl^- free solution containing 4% paraformaldehyde. Temporal bones were decalcified in EDTA, processed through a sucrose gradient and infiltrated with polyethylene glycol. Mid-modiolar cryosections (12 μm, CM3050S, Leica, Nussloch, Germany) were blocked in PBS-TX (137 mM NaCl, 10.1 mM Na₂HPO₄, 1.8 mM KH₂PO₄, 2.7 mM KCl, pH 7.4 with 0.2% Triton X 100) and 5% bovine serum albumin. Slides were incubated overnight at 4°C with Alexa-488 conjugated rat anti-mouse CD68 antibody (1:25, Serotec, Raleigh, NC, USA) in PBS-TX with 1–3% BSA. After incubation, slides were washed with PBS-TX, mounted with FluorSave (Calbiochem, La Jolla, CA, USA), and viewed by confocal and laser scanning brightfield microscopy (LSM 510 Meta, Carl Zeiss, Göttingen, Germany).

Temporal bones from age matched Slc26a4^-/- and Slc26a4^+/- mice were rendered blood free by transcardial perfusion with Cl^- free solution. Stria vascularis was obtained by microdissection and fixed for 2 hrs at 4°C in Cl^- free solution containing 4% paraformaldehyde, washed twice in Cl^- free solution and once in PBS-TX and then blocked with 5% BSA in PBS- TX for 45 min at RT and then washed three times in PBS-TX. Stria vascularis was then incubated overnight at 4°C either with Alexa-488 conjugated rat anti-mouse CD68 antibody (1:25, see above) or goat anti-Kcnq1 primary antibody (1:200, C20, Santa Cruz Biotechnology, Santa Cruz, CA, USA) in PBS-TX with 1–3% BSA. Tissues incubated with goat anti-Kcnq1 primary antibody were washed with PBS-TX and incubated for 1 h at 25°C with chicken anti-goat Alexa 594 secondary antibody, 1:1,000 (Molecular Probes, Eugene, OR, USA) in PBS-TX with 1–3% BSA.

After antibody incubation, stria vascularis was washed in PBS-TX (0.2% Triton-X) and stained with phalloidin conjugated to Alexa 594 (1:40; Invitrogen, Carlsbad, CA, USA), washed three times with PBS-TX, mounted with FluorSave (Calbiochem), and viewed by confocal and laser scanning brightfield microscopy (LSM 510 Meta, Carl Zeiss).

Temporal bones were removed and stria vascularis and spiral ligament were obtained by microdissection. Microdissection solutions were changed twice and isolated tissues were washed to minimize contamination between tissue fractions. In addition, liver and spleen were collected and rapidly frozen in liquid nitrogen. Total RNA was isolated and residual DNA contamination was removed by DNase treatment (RNeasy micro, Qiagen, Valencia, CA, USA). Frozen samples of liver and spleen were pulverized and homogenized. Total RNA was isolated and freed from DNA contamination (RNeasy mini, Qiagen). Isolated RNA was either used immediately or diluted and stored at -80°C for later analysis (RNA storage solution, Ambion, Austin, TX, USA).

Total RNA was isolated from stria vascularis of adult Slc26a4^+/+ and Slc26a4^-/- mice (P148 ± 3 and P153 ± 1, respectively). The minimal amount of blood present within capillaries was deemed insignificant; no cardiac perfusion was performed. Isolated RNA was concentrated and frozen for shipment to the Biotechnology Support Facility at University of Kansas Medical Center. A total of six chips were run; each chip was used to analyze RNA pooled from two animals. Three chips each were used to analyze RNA expression in stria vascularis of Slc26a4^+/+ and Slc26a4^-/- mice. RNA was amplified by two rounds of amplification, and cRNA was hybridized to high-density oligonucleotide gene chips (Small sample protocol, version II; mouse 430 2.0 gene chip, Affymetrix, Santa Clara, CA, USA). Gene array data were analyzed using commercial software (GCOS, Affymetrix; Genespring, Silicon Genetics, Redwood City, CA, USA) as well as custom-written macros (Excel, Microsoft, Redmond, WA, USA). Quality metrics conformed with MIAME standards (Table 1). Present/absent calls and averaged signal intensities (average of data obtained from three chips) were used to determine expression and changes in expression levels, respectively.

In the tabulated data summaries, 'Intensity' for Slc26a4^+/+ and for Slc26a4^-/- samples represents averages of data from one or more probes. For example, the gene Slc12a2 is represented on the chip by four probes. Present calls (P) were summarized for all three chips, e.g. 12/12 indicates that this gene was called present by all 12 probes (4 × 3 = 12); 9/15 indicates that the gene is represented by 5 probes on the 3 chips (5 × 3 = 15) and that the gene was called present by 9 of the 15 probes.

Ratios of intensity values (Slc26a4^-/- to Slc26a4^+/+) were calculated for each probe and averaged. Average ratios > 1.000 were reported as Fold with the Direction 'up'. Average ratios < 1.000 were inverted (1/average ratio) and reported as Fold with the Direction 'down'. Fold values are given in the tables only when the gene was called Present (P) in Slc26a4^+/- or Slc26a4^-/- samples in at least half of the available probes. The direction of the fold change is only given when it exceeded 1.30. Fold changes lower than 1.30 were not considered significant.

Age and sex matched Slc26a4^-/- and Slc26a4^+/- mice were rendered blood free by transcardiac perfusion with Cl^- free solution. Total RNA was isolated from microdissected stria vascularis and spiral ligaments, as well as from liver and spleen. In each 96-well plate, the expression of seven different genes as well as 18S rRNA was analyzed. Total RNA from stria vascularis and spiral ligaments or from spleen and liver of one Slc26a4^-/- and one matched Slc26a4^+/- mouse was analyzed in duplicate reactions in parallel to allow paired comparisons (paired t-test). qRT-PCR was performed in the presence of 0.5× SYBR green I on total RNA isolated from individual animals using gene specific primers (One step RT-PCR kit, Qiagen; iCycler, BioRad, Hercules, CA, USA; SYBR green I, Molecular Probes; Table 2). RT was performed for 30 min at 50°C and 15 min at 95°C. PCR consisted of 40 cycles of 1 min at 60°C, 1 min at 72°C, 20 s hot measurement, and 1 min at 94°C. Specificity of primers was verified by sequencing. The generation of a single product of the appropriate size was verified by agarose gel electrophoresis.

Template molecules were quantified according to T = 10^log (P_Ct)/(E_avg^C_t), where P_Ct is product molecules at C_t, E_avg is the average efficiency and C_t is cycle at which the fluorescence of the product molecules reached a set threshold. Efficiencies for individual reactions was obtained from the slope of the log-linear phase of the growth curve using an Excel-based program (LinRegPCR) 21.

The number of product molecules at C_t (P_Ct) was calculated by amplifying known numbers of 18S rRNA (T_18S) molecules according to P_Ct = T_18S × E_18S^C_t, where E_18S is the average efficiency of all the 18S rRNA reactions. The mass of 18S rRNA in 1 μg of total RNA was estimated to calculate the number of 18S rRNA molecules in a given amount of RNA sample. Calculations were based on three assumptions: firstly, total RNA consists to 100% of 18S and 28S rRNA. Secondly, 18S and 28S rRNA occur in a 1:1 ratio. Thirdly, murine 18S and 28S rRNA contains 1869 and 4712 nucleotides [Genbank: X00686] and [Genbank: X00525], respectively. The mass of 18S rRNA per 1 μg of total RNA was estimated to be 0.284 μg, equivalent to 4.7 × 10^-13 mol or 2.8 × 10¹¹ molecules of 18S rRNA (molecular weight of 18S rRNA estimated to be 598,080).

Temporal bones from age matched Slc26a4^-/- and Slc26a4^+/- mice were rendered blood free by transcardial perfusion with Cl^- free solution. Stria vascularis and spiral ligament were isolated by microdissection. Proteins in stria vascularis and spiral ligament from one animal were isolated by heating (10 min, 95°C) in 20 μl of a diluent (Compound B, NanoOrange, Invitrogen). After cooling (20 min, RT), the isolated protein was quantified (NanoOrange, Invitrogen) and either used immediately in Western blots or stored at -80°C.

Proteins (15 μg per well) were denatured in the presence of Laemmli sample buffer (62.5 mM Tris-HCl, pH 6.8, 25% glycerol, 2% SDS, 0.01% bromophenol blue, 5% 2-mercaptoethanol) at 80°C for 5 min and resolved by electrophoresis (150 V for 45 min) in 4–15% polyacrymalide Tris-HCl gradient gels (10 well mini-gel, BioRad laboratories, Inc., Hercules, CA, USA). Molecular weight standards were used to estimate molecular weights (Precision Markers, BioRad). Proteins were transferred in transblot buffer (25 mM Tris, 192 mM glycine, 20% v/v methanol, pH 8.3). (25 V, 1 h; XCell II blot module, Invitrogen) onto a nitrocellulose membrane (0.2 μm, BioRad).

Nitrocellulose membranes were first evaluated for lysozyme expression, then stripped (Restore Western Blot Stripping Buffer, Pierce, Rockford, IL, USA) and evaluated for actin expression. Membranes were blocked in TBS-Tween (20 mM Tris-Cl, pH 7.6 with 0.1% Tween) containing 5% non-fat dry milk powder (BioRad) and incubated (1 hr RT) with rabbit anti-human lysozyme primary antibody (1:500; Dakocytomation, Carpinteria, CA) or rabbit anti-Actin primary antibody (1:1,000; Sigma) in TBS-Tween containing 5% non-fat dry milk. Membranes were washed three times with TBS-Tween and then incubated with horseradish peroxidase conjugated donkey anti-rabbit antibody (1:25,000, Amersham Biosciences) in TBS-Tween containing 5% non-fat dry milk. The membranes were washed again three times with TBS-Tween and then treated with chemiluminescent detection agents (SuperSignal West Femto Maximum Sensitivity Substrate; Pierce), exposed to film (CL-XPosure Film, Pierce), which was developed immediately (XOmat 2000A, Kodak, Rochester, NY, USA).

Numeric data are presented at average ± sem, unless specified otherwise. The number (n) of animals, blots or cells is given. Differences were determined by paired t-tests. Significance was assumed at p < 0.05.

Previously, we reported that stria vascularis of adult Slc26a4^-/- mice is hyperpigmented and that marginal cells appear disorganized 1517. Stria vascularis was obtained from age matched Slc26a4^+/- and Slc26a4^-/- mice and hyperpigmentation was evaluated by laser-scanning light microscopy. Hyperpigmentation was absent at P10 and P15 but observed after weaning (P33-41) and in adult (P84-96) Slc26a4^-/- mice (Fig. 1).

Marginal cells can be identified by there expression of the K⁺ channel Kcnq1 in their apical membrane 17. To gain information on the identity of surface epithelial cells, expression of Kcnq1 was visualized by immunocytochemistry and F-actin, a marker for tight junctions, was visualized by phalloidin staining. Marginal cells at all ages of Slc26a4^+/- mice and at P10 and P15 of Slc26a4^-/- mice expressed the K⁺ channel Kcnq1 evenly. Little variation was observed in cell surface areas at all ages of Slc26a4^+/- mice and at P10 and P15 of Slc26a4^-/- mice, and variations appeared normally distributed (Figs 2 and 3). The average cell surface area of marginal cells in Slc26a4^+/- mice was 104 ± 2 μm² (n = 120). No significant differences were detected in Slc26a4^+/- mice of different ages. In contrast, the average cell surface areas of marginal cells in P10 and P15 Slc26a4^-/- mice were significantly larger, 236 ± 12 μm² (n = 30) and 232 ± 12 μm² (n = 30), respectively. This enlargement of the apical surface area of marginal cells is consistent with the understanding that Slc26a4^+/- and Slc26a4^-/- mice do not differ in the total number of marginal cells, and that cell surface areas are enlarged to cover the larger area that results from the enlarged cochlear diameter found in Slc26a4^-/- mice.

After weaning, marginal cells of Slc26a4^-/- mice began to reorganize, leading to a multimodal distribution of cell surface areas (Figs 2 and 3). Most interestingly, marginal cells that reduced their surface areas from ~ 240 to ~ 100 μm² retained dense expression of Kcnq1, whereas marginal cells that increased their surface area gradually lost Kcnq1 expression. The degree of Kcnq1 staining in stria vascularis of P37 Slc26a4^-/- mice correlated inversely with the size of the cell surface area. Stria vascularis of adult Slc26a4^-/- mice consisted of a mosaic of Kcnq1-expressing and degenerated marginal cells. Kcnq1-expressing marginal cells are likely to remain functional, given that Kcnq1 is essential for K⁺ secretion and the endolymphatic K⁺ concentration was found to be normal (~ 140 mM) in adult Slc26a4^-/- mice 172223.

Gene expression in stria vascularis was analyzed by gene array to gain data on the cause of marginal cells reorganization. The quality of the gene array was determined by quality metrics (Table 1), and the quality of the expression analysis was determined by an evaluation of genes that are known to be expressed in stria vascularis (Table 3) and by an evaluation of genes that are known to be expressed in neighboring tissues that could have served as sources of contamination (Table 4). As expected, genes known to be expressed in marginal, intermediate basal and spindle cells were present in Slc26a4^+/+ and Slc26a4^-/- mice (Table 3). Interestingly, genes known to be expressed in marginal cells were downregulated in Slc26a4^-/- mice (Table 3A). In contrast, no change in expression was observed for genes known to be expressed in intermediate and/or basal cells (Table 3B). These data are consistent with a partial degeneration of marginal cells.

Genes known to be expressed in neighboring tissues were not found in stria vascularis, with the exception of otospiralin (Otos). Otos in known to be expressed in spiral ligament 24, which is the tissue adjacent to stria vascularis (Table 4). Taken together, these data demonstrate that gene array analysis provided reliable data of expression in stria vascularis.

Gene array analysis revealed that stria vascularis of adult Slc26a4^-/- mice expresses markers specific and/or consistent with the presence of macrophages (Table 5). Expression of Mac2 antigen, Itgax, Cd45, Cd83 and Cd68 is limited to leucocytes and hematopoietic cells including macrophages and dendritic cells 25. Expression of major histocompatibility complex II (MHCII) proteins is limited to antigen-presenting cells including macrophages 25. The expression of Lysz, a lysosomal enzyme, has been shown to be a marker for monocytes and macrophages 26. Further, the increased expression of major histocompatibility complex I, of complement components and of cathepsins is consistent with the presence of macrophages although the expression of the gene is not limited to macrophages 27.

Taken together, the data presented in Table 5 suggest that stria vascularis of adult Slc26a4^-/- mice is invaded by macrophages. Markers for T-cells, NK-cells, B-cells, neutrophils and for acute inflammation were absent in stria vascularis of Slc26a4^+/+ and Slc26a4^-/- mice (Tables 6 and 7). The absence of markers of acute inflammation was verified by qRT-PCR of total RNA isolated from stria vascularis and spiral ligament of age and sex matched Slc26a4^-/- and Slc26a4^+/- mice. No change in expression was found for Il1a, Il6, Il12a, Nos2 and Nos3 at P15 or at P34 (each n = 4–5; data not shown). Thus, no evidence for acute inflammation was obtained. It is conceivable, however, that short periods of acute inflammation escaped detection. Alternatively, it is conceivable that macrophage invasion was not preceded by neutrophil invasion and that macrophage recruitment occurred neutrophil-independently 28. The source of chemoattractants for monocytes (MCPs) in neutrophil-independent recruitment are endogenously harbored tissue macrophages and mesothelial cells 28. Such a mechanism could be present in stria vascularis of Slc26a4^-/-mice as the cochlea endogenously harbors macrophages in the fibrocytes of the lateral wall 29. The mechanism of recruitment into stria vascularis, however, remains unclear.

The observation that stria vascularis of adult mice is invaded by macrophages raises the question whether macrophage invasion is systemic and hence found in other organs, such as spleen and liver, or whether macrophage invasion is restricted to stria vascularis and hence not seen in the adjacent tissue, spiral ligament. This is an important issue, as the inner ear is immunologically responsive to systemic infections 30.

Total RNA was isolated from stria vascularis, spiral ligament, liver and spleen from age and sex matched Slc26a4^-/- and Slc26a4^+/- mice. The expression of a select group of transcripts was evaluated by qRT-PCR. Possible contamination between stria vascularis and the adjacent spiral ligament was evaluated by quantifying the expression of Otos and tyrosinase (Tyr) under the assumption that Tyr is expressed in stria vascularis and not in spiral ligament, and Otos in spiral ligament and not in stria vascularis. The detected expression level of Tyr in stria vascularis of Slc26a4^+/- and Slc26a4^-/- mice was 120 ± 33 and 50 ± 13 times higher than in spiral ligament, respectively (n = 6) and the detected expression of Otos in spiral ligament of Slc26a4^+/- and Slc26a4^-/- mice was 108 ± 32 and 50 ± 28 times higher than in stria vascularis, respectively (n = 3). These data illustrate a small contamination between stria vascularis and spiral ligament. Contamination was larger in Slc26a4^-/- mice (2%) than in Slc26a4^+/- mice (1%), consistent with the observation during microdissection that stria vascularis from Slc26a4^-/- is more fragile.

Expression of complement components and macrophage markers including C1r, C3, Ptprc (= Cd45), Cd83 and Lyzs was increased in stria vascularis of Slc26a4^-/- mice compared to Slc26a4^+/- mice (Fig. 4). Expression of these genes was not upregulated in spiral ligament with the exception of Lyzs and Cd45. Further, the expression of these genes was not upregulated in liver or spleen (Fig. 4). Taken together, these data suggest that macrophage invasion in Slc26a4^-/- mice is restricted to stria vascularis.

The finding that macrophage invasion was restricted to stria vascularis raises the question of when in development macrophage invasion occurs. Total RNA was isolated from stria vascularis and spiral ligament from sex and age matched Slc26a4^-/- and Slc26a4^+/- mice and a select group of transcripts was quantified by qRT-PCR. At P15, no difference in the expression of Lyzs, C3 and Cd45 was found in stria vascularis or in spiral ligament between Slc26a4^-/- and Slc26a4^+/- mice (Fig. 5). After weaning, and persistent into adulthood, however, the expression of Lyzs, C3 and Cd45 was higher in stria vascularis of Slc26a4^-/- mice compared to Slc26a4^+/- mice. A similar trend was observed with spiral ligament.

The expression of Lyzs was determined based not only on the transcript level but also on the protein level, which may be less prone to contamination between neighboring tissues. Western blotting revealed that Lyzs protein is upregulated in stria vascularis but not in spiral ligament of post-weaning Slc26a4^-/- mice (Fig. 6).

Macrophage invasion was not only determined by gene expression analysis but also by immunohistochemistry, using CD68 as a marker. The specificity of the anti-CD68 antibody was verified by using bone marrow cells as a positive control and heavily pigmented cells of the vestibular labyrinth as a negative control (Fig. 7). Staining for CD68 was clearly associated with hyperpigmentation in stria vascularis in post-weaning and adult Slc26a4^-/- mice (Fig. 7).