Nkcc1-null mutant mice used for this study were generated and genotyped as previously described (Flagella et al., 1999). One hemi-maxillary incisor of each mouse was used for immunohistological studies and the other one was freeze-dried for micro-CT analysis and western blotting. For each genotypic mouse strain, at least three wild-type mice and three null mutant mice were analyzed.

All experiments were approved by the Committee for Animal Care (Vrije Universiteit Amsterdam; ACTA-12-01) and by the Animal Care and Use Committee of the National Institute of Dental and Craniofacial Research, National Institutes of Health (ASP 13–686). The methods were carried out in accordance with the approved guidelines.

HAT-7 cells were grown in DMEM/F12 (Sigma-Aldrich, St. Louis, MO, USA) with 10% HyClone fetal bovine serum, 100 U/mL of penicillin, 10 μg/mL of streptomycin (Sigma-Aldrich) and 10⁻⁵ mM dexamethasone in humidified atmosphere containing 5% CO₂ at 37°C (Bori et al., 2016).

Mouse jaws were fixed by immersion in 5% paraformaldehyde in 0.1 M phosphate buffer pH 7.3 and embedded in paraffin. Calcified tissues from mice older than 2 weeks were first decalcified in 4% EDTA, pH 7.3 for 2–3 weeks at 4°C. Salivary glands were fixed in Bouin's fixative (75 ml of saturated picric acid, 25 ml of 40% formaldehyde, and 5 ml of glacial acetic acid) at room temperature for 48 h and processed into 5–7 μm thick paraffin sections. Dewaxed sections were stained with 1% hematoxylin (1 min) and eosin (5 min) (HE) or used for immunohistochemicals staining.

At 40x objective pictures of secretory and maturation stage of enamel organ were made and images selected based on the anatomical position in the tooth. The length of the long axis of secretory and maturation ameloblasts was measured in sagittal sections using imaging software (image J).

Total RNA was extracted from HAT-7 cells and various fresh mouse tissues using the NucleoSpin RNA/protein kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's instructions. First strand cDNA synthesis was performed in a 20 μl reverse transcription reaction containing 200 ng of total RNA using VILO kit (Invitrogen) according to the manufacturer‘s instructions. Real-time PCR analysis was performed to analyze expression of Slc12a2 (Nkcc1) with the primer sequences (FW:5' GAAGAAAGTACTCCAACCAGAGATG 3'; REV: 5' CTGAAGTAGACAATCCTGTGATA 3'; size: 232 bps) and the housekeeping protein tyrosine 3-monooxygenase (Ywhaz) with sequences (FW:5'GATGAAGCCATTGCTGAACTTG3'; REV:5'CTATTTGTGGGACAGCATGGA3'; size:229 bps) shown by using the LightCycler 480 system based on SYBR Green I dye (Roche Applied Science, Indianapolis, IN, USA). The LightCycler reactions were prepared in 20 μl total volume with 7 μl PCR-H₂O, 0.5 μl forward primer (0.2 μM), 0.5 μl reverse primer (0.2 μM), 10 μl LightCycler Mastermix (LightCycler 480 SYBR Green I Master; Roche Applied Science, IN, USA), to which 2 μl of 5 times diluted cDNA was added as PCR template. Controls in the real-time RT-PCR reaction included RT reactions without the reverse transcriptase (control for DNA carry over) and RT reactions without template (control for reagent contamination). With the Light Cycler software, the crossing points were assessed based on a standard curve of five serial dilutions ranging from 10 ng to 1.6 pg of cDNA. PCR efficiency (E) was automatically calculated using the fit point method (E = 10⁻¹/slope). Gene expression data were used only if the PCR efficiency was within a 1.85–2.0 range. For each gene the amount of measured DNA was normalized to that of YWHAZ housekeeping gene to calculate relative gene expression. The relative gene expression in different tissues was normalized to kidney levels for each gene in the graphs.

Dewaxed paraffin sections were rinsed in phosphate buffered saline (PBS) and subjected to antigen retrieval in 10 mM citrate buffer (pH 6.0) either at 60°C overnight or for 20 min in microwave at 95°C. Endogenous peroxidase was blocked with a peroxidase block solution (Envision kit, Dakocytomation) for 5 min. Sections were washed 3x in tris-buffered saline (TBS). Non-specific staining was blocked for 30 min with 2% BSA after which sections were incubated overnight at 4°C with primary antibodies. These were (1) goat anti-NKCC1 (Santa Cruz, affinity purified, catalog number SC-21545), raised against the N-terminal end of human NKCC1. (2) Mouse anti-NCKX4 monoclonal antibodies (IgG2b isotype) from NeuroMab (UC Davis/NIH NeuroMab Facility, catalog # N414/25). (3) Matched non-immune IgG (1:200–1:300) or normal serum (same concentration as primary antibodies) served as controls. After overnight incubation at 4°C with primary antibodies, sections were washed three times in TBS and incubated with rabbit anti-goat secondary antibody conjugated to peroxidase (Thermo Scientific) for 1 h at room temperature. After washing staining was visualized using DAB (EnVision kit), counterstained with hematoxylin. For immunofluorescent staining, goat anti mouse–IgG conjugated to Alexa Fluor 488 (5 μg/mL; Invitrogen) was used and counterstained with propidium iodine (Vector Laboratories, Burlingame, CA, USA). Immunohistochemistry images were acquired with a Leica EL6000 or Axio Zoom V16 microscope.

To determine the degree of mineral content, hemi-maxillae were scanned at a resolution of 8 μm voxels in a microCT-40 high-resolution scanner (Scanco Medical, AG, Bassersdorf, Switzerland) to measure mineral density in enamel. An internal standard made of solid-sintered apatite (5-mm diameter, 1.5–2.0 mm thick, solid sintered) with density of 2.9 ± 0.2 g/mL (a gift from Himed; http://www.himed.com) was used as high-density standard. Beginning at the apical part of the incisor and moving toward the tip, cross-sectioned images through the incisors were collected at sequential intervals of 300 μm in maturation-stage and 60 μm in secretory-stage enamel. In each slice, the mineral density of enamel was measured halfway through the enamel layer at three sites within a circular area, with a diameter of 7 μm at the mesial, lateral, and central sides. Mean values and standard error of mean (SEM) of the mineral density were calculated and presented as mean ±SEM. Independent Student's t-test was used to compare the groups. Statistical significance was set at p < 0.05 level.

From freeze-dried upper incisors obtained from wild-type and Nkcc1^−/− mice early maturation stage enamel organs were micro dissected. The apical half of the enamel organ was dissected, dissolved in non-reducing condition in SDS loading buffer (from Nucleospin Triprep kit, Macherey-Nagel, supplied by Bioke, Leiden, NL) and protein was measured using the BCA protein assay (Bio-Rad, Hercules, CA). Twenty micro gram of enamel organ denatured protein and 10 μg of molecular weight markers [Novex® Sharp Pre-stained Protein Standard (# LC5800) or SeeBlue® Plus2 Pre-stained Protein Standard (#LC5925)] were subjected to electrophoresis in a 3–8% Tris acetate Nupage gel with Tris acetate running buffer for 60 min at 150 V or 4–12% Bis-Tris Protein Gels with MOPS buffer for 35 min at 200 V. and subsequently electroblotted by an iBlot device (Invitrogen) on nitrocellulose membrane according to the manufacturer's instructions. Membranes were blocked with BSA 2% for 1 h at room temperature and incubated overnight (4°C) with the primary antibodies. Blots were washed three times in PBS and incubated with IRDye secondary antibodies (LI-COR). Visualization and quantification was carried out with the LI-COR Odyssey scanner and software (LI-COR Biosciences). Actin was detected at 680 nm wavelength (shown as red) and other primary antibodies and tubulin were detected at 800 nm wavelength (shown as green). Quantification was performed using Odyssey software. Intensity values of the bands were normalized for actin or tubulin and expressed as percentage of wild type (100%). For western blots the following primary antibodies were used: rabbit anti-SLC26A3/Dra (Research Genetics, Huntsville, AL, USA) (Jalali et al., 2015), rabbit anti-SLC26A6/Pat1 (donated by Dr. P. Aronson, Yale University, New Haven, CT, USA) (Jalali et al., 2015), rabbit anti-NBCe1 (Jalali et al., 2014) (donated by Dr. W.F. Boron, Case Western Reserve University, Cleveland, Ohio, USA), rabbit anti-NCKX4 (NeuroMab, UC Davis/NIH, # N414/25), rabbit anti-connexin (Abcam, #ab11370), mouse anti-β-actin antibody (Sigma, A2228) and rabbit anti-tubulin antibody (Abcam, ab59680). Secondary antibodies: IRDye 800CW conjugated goat anti-rabbit IgG (H+L) highly cross-adsorbed (LI-COR; Product number: 926–32,211) and IRDye 680CW conjugated goat anti-mouse IgG (H+L) highly-cross adsorbed (LI-COR; Product number: 926–32,220). Dilutions: anti-β-actin and anti-tubulin (1:1,000); other primary antibodies (1:250); secondary antibodies (1:10,000).

HAT-7 cells were loaded with 5 μM Calcein-AM (Molecular Probes) for 20 min at 37°C. Changes in cell volume of single HAT-7 cells, plated on poly-lysine coated glass cover slips, were assessed by measuring calcein fluorescence using the calcein-quenching method (Ye et al., 2015). Cells were bathed in iso-osmotic solution (20–22°C) and transferred to a continuously perfused (5 ml/min) recording chamber, equipped with a microscope with 10X objective. An image was taken every 30 s. At the start images were obtained for 5 min in the iso-osmotic solution to establish the baseline. Cells were exposed to media supplemented with 10 μM bumetanide for 30 min, after which the iso-osmotic solution was re-introduced and images were taken for another 10 min. The cell surface and average fluorescence of each cell in the acquired images was calculated using Image-J software. Changes in cell surface and average fluorescence were expressed as S_t/S₀ and F_t/F₀ ratios, respectively, where S₀ and F₀ are the average cell surface area and fluorescence under iso-osmotic treatment at the beginning of the experiment.

Cover slips bearing HAT-7 cells were transferred to the recording chamber, containing 1.5 ml external solution. The external solution was changed at a rate of 1.5 ml/min using a gravity driven constant perfusion system. During the recordings, HAT-7 cells were perfused with standard artificial Cerebral Spinal Fluid (aCSF) containing (in mM): 126 NaCl, 3KCl, 10 D-glucose, 26 NaHCO₃, 1.2 NaH₂PO₄, 2 CaCl₂ and 1 MgSO₄, carboxygenated with 95% O₂ and 5% CO₂ to obtain pH 7.4 and an osmolality of 300 mOsm. For electrophysiological recordings, we used a EPC-8 amplifier and a Instrutech ICT-18 (all from Heka, D-67466 Lambrecht, Germany). Cells were identified and patched using an Olympus IX51 inverted microscope equipped with a LCAcn 40x 0.55nA ph2 objective (all Olympus corporation, Tokyo, Japan). Glass pipettes for whole-cell and cell attached recordings were made from borosilicate capillaries (OD 1.5 mm, ID 0.86 mm; Harvard Apparatus, Holliston, MA, USA) using a Sutter P-87 micro-electrode puller (Sutter instruments, USA) and displayed a resistance of 2.5–5 MΩ. Glass microelectrodes were filled with intracellular solution containing (in mM): 110 K-Gluconate, 10 KCl, 10 HEPES, 0.4 NaGTP, 4 Mg₂ATP and 10 K-Phosphocreatine (pH 7.3 adjusted with KOH, 290 mOsm). HAT-7 cells were gently lifted from the cover slip and placed in front of a piezo-driven theta-barrel electrode (TGC 200; Harvard Apparatus, Holliston, MA, USA), filled with standard aCSF on one side and standard aCSF supplemented with 10 μM bumetanide (B1158000 Sigma-Aldrich) on the other side. By changing the position of the barrel bumetanide was applied during the whole-cell recording. Voltage ramps from −70 to 80 mV (500 ms) were applied under control and in the presence of 10 μM bumetanide.

The cell attached recordings were made by filling the recording pipet with aCSF for the control recordings and with aCSF supplemented 10 μM bumetanide for experimental recordings. All data was acquired using an internal 7-pole Bessel filter (5 kHz) and a sample frequency of 20 kHz. Recordings with an access resistance above 12 mΩ were excluded form analysis.

Transcripts for Nkcc1/Slc12a2 normalized for Ywhaz housekeeping gene were detectable in enamel organ and intestine (high), pulp and kidney moderate-(low); in the remaining tissues tested expression was very low or below detection limit (Figure 1A; Supplementary Figure 1). HAT-7 cells also expressed Nkcc1 transcripts (Figure 1B).

Bumetanide blocks activity of the NKCC's. To test whether this blocking agent also could affect Nkcc1 expression level in enamel epithelium, HAT-7 cells were exposed to various concentrations of this inhibitor for 45 min, washed, and incubated for 9 h in medium without inhibitor. Then total RNA was isolated and the amount of Nkcc1 transcripts measured by RT-qPCR (Figure 1B). Bumetanide did not change the number of transcripts of Nkcc1.

Anti-NKCC1 antibodies stained plasma membranes in HAT-7 cells as fine granular material (Figure 2A). Replacing primary (mouse) antibodies for normal non-immune mouse IgG failed to stain these membranes in HAT-7 cells (Figure 2B).

Strong immunostaining was detected in the basolateral plasma membranes of the acinar cells of salivary glands, a well-established site of NKCC1 expression (Evans et al., 2000) (Figure 2C).

In upper incisors, in presecretory stage and during the secretory and maturation stage, the plasma membrane of outer enamel epithelium cells was immunopositive for NKCC1 (Figures 2E,F; Supplementary Figure 2). No staining was seen in ameloblasts. Weaker staining was seen in dental epithelium between ameloblasts and outer enamel epithelium. Strong staining was apparent in the papillary layer (intracellular and membranes) during maturation (Figures 2E,G).

Sections from salivary glands (Figure 2D) and enamel organ from Nkcc1-null mice (Figure 2H) incubated with anti-NKCC1 failed to immunostain the tissues, validating the specificity of the antibodies. In developing molar tooth germs similar stainings were obtained: a strong staining for NKCC1 in outer enamel but no staining in ameloblasts (Supplementary Figure 3).

Slc12a2 (or Nkcc1) knockout mice exhibit growth retardation with a 30% incidence of death by the time of weaning. They are deaf, have less body fat, reduced mean arterial blood pressure, exhibit reduced cAMP-induced short circuit currents in jejunum, cecum, and trachea, unusual head postures, and engage in circling behaviors and rapid spinning, but have difficulty maintaining their balance (Flagella et al., 1999). Incisors of Nkcc1-null mice showed no obvious gross changes. Analysis of upper incisor enamel of Nkcc1-null mice showed no significant change in mineral density in enamel during secretory stage but a reduction at maturation stage, reaching a final density that was reduced about 10% compared to controls (P < 0.0001) (Figures 3E–I).

Next we examined expression of NCKX4 (sodium/calcium-potassium exchanger-4) a major transcellular calcium transporter in maturation ameloblasts, to determine whether calcium transport is affected in Nkcc1-null incisors. Nckx4 protein was strongly expressed in the apical membrane of maturation ameloblasts of wild type mice (Figure 3A), but its antibody-associated staining was weaker in Nkcc1-null mice (Figure 3B). Western blot analysis (Figure 3C) of protein extracts from wild type and Nkcc1-null maturation enamel organ, showed a small (but not significant) reduction of NCKX4 protein in Nkcc1-null enamel organ (Figure 3D).

To examine if the expression of other Na⁺ and Cl⁻ ion transporters or/ and gap junction channels could compensate for the absence of NKCC1, we tested maturation stage enamel organs of Nkcc1-null mice for changes in protein levels of SLC26A6, DRA, NBCe1 and Cx43 by Western blotting (Figures 4A–H). Total protein extracted from Nkcc1-null enamel organs showed a 303% increase of SLC26A6/PAT1 (P = 0.003) (Figures 4A,E) compared with wild type a, 143% increase for SLC26A3/DRA (P = 0.04) (Figures 4B,F), 47% increase for NBCe1 (P = 0.04) (Figures 4C,G) and 78% increase for Cx43 (P = 0.01) (Figures 4D,H). The results are summarized in Figure 4I.

NKCC1 is one of the proteins involved in cell volume recovery after cell shrinkage in kidney (Walker et al., 2002). Analysis of histological sections of Nkcc1-null upper mouse incisors showed that late maturation ameloblasts lost their polarization and that the cells of the papillary layer were shorter compared to that seen in wild type controls. However, secretory stage ameloblasts and stratum intermedium seemed hardly or not affected in the null mutants in comparison to wild type mice (Figures 5F,G,I,J). To examine a possible role of NKCC1 in regulating cell volume in ameloblasts, we measured cell volume changes of single HAT-7 cells in vitro by measuring cell size (surface area measurements) and by using the calcein-quenching method before and after bumetanide (10 uM) treatment, (Figures 5A–E). Bumetanide decreased cell volume as measured directly (Figure 5D) and by loss of fluorescence (Figure 5E). In incisors of Nkcc1-null mice the values of the long axes of secretory ameloblasts were not different from those in wild types (Figures 5F–H) but were significantly shorter (~25%; P < 0.0001; Figures 5I–K) in maturation ameloblasts.

Bumetanide has been previously shown to induce potassium currents in canine kidney cells. To test whether bumetanide affects the membrane ionic conductance in HAT-7 cells, electrophysiological whole-cell voltage-clamp recordings were made from these cells. The membrane potential was ramped from −70 to +80 mV. In control conditions, the voltage ramp induced a small membrane current that reversed at −30 mV. Application of bumetanide (10 μM) increased membrane currents across the entire membrane potential range (Figure 6A). The outward current measured at +80 mV increased about 3-fold, whereas the inward current measured at −70 mV was 20-fold enlarged (Figures 6A,B). The increase of membrane currents induced by bumetanide hardly reversed upon washout (not shown). To our surprise, the bumetanide-induced currents reversed at −30 mV. Given that for the recording solutions used in these experiments the equilibrium potential for potassium was −95 mV, for chloride −65 mV and for sodium and calcium above +60 mV, this suggests that most likely a mixed ionic current was activated by bumetanide-block of NKCC transporters.

It is not known whether the effect of bumetanide on the membrane conductance required block of NKCC transporters across the entire cell membrane, or whether this effect could also be induced locally. To determine this, we made cell-attached recordings from HAT-7 cells and applied bumetanide only to the membrane patch inside the patch pipette. The pipette potential was kept at 0 mV. In control recordings, ion channel activity and occasional discrete single channel openings were observed that were alternated by silent periods (n = 10, Figure 7). At this pipette potential, only outward currents were observed. During recordings in which bumetanide (10 uM) was included in the recording pipette, ion channel activity conducting outward currents increased (n = 10, Figure 7). No discrete amplitude levels could be distinguished, which suggests that blocking NKCC transporters with bumetanide augments the activity of multiple ion channels in the membrane patch. These results suggest that the effects of bumetanide on membrane ionic conductance could be induced locally in a patch of the cell membrane.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer CC and handling Editor declared their shared affiliation.