To correlate the expression pattern of Nckx4 with ameloblast functions, ameloblasts of enamel organs at the various developmental stages were collected for semi-quantitative PCR. We found that as compared to secretory ameloblasts (P5), Nckx4 expression started to increase in the post-secretory ameloblasts (P7) and reached a peak at the early maturation stage (P8), followed by a reduction at the late maturation stage (see Figure 1A). Overall, the expression levels of Nckx4 in maturation ameloblasts were significantly greater than that in secretory ameloblasts. The pattern of Nckx4 expression in the maturation stage ameloblasts, in particular in the early maturation ameloblasts, was confirmed by in situ RNA hybridization (RNAscope) (see Figure 1B).

Immunostaining on the full spectrum of ameloblast lineage on incisal sagittal sections showed that the NCKX4 signal apparently increased in the transition ameloblasts compared to secretory ameloblasts (Figure 1D), peaked in the early maturation ameloblast (MAB), and slightly reduced at the end of maturation stage (Figure 1C). In maturation stage ameloblasts, NCKX4 was primarily concentrated along the apical end of ruffle-ended ameloblasts (RAB) (see Figure 1E), but scattered in the cytosol of smooth-ended ameloblasts (SMAB) with reduced intensity (see Figure 1F). A recapture of the temporal and spatial expression pattern of NCKX4 aids us to understand its stage- and phase-specific functions in the enamel maturation.

As compared to the smooth and translucent enamel of Nckx4 ^+/+ incisor (see Figure 2A) and molars (see Figure 2B), the enamel layers on both Nckx4 ^−/− incisor (see Figure 2C) and molars (see Figure 2D) were worn, chipped away and heavily discolored. These appearances are also found on the teeth of patients diagnosed with amelogenesis imperfecta.

MicroCT scanning showed that the enamel layer in both sagittal (see Figure 2E) and cross sections (see Figures 2E1, E2) of Nckx4 ^+/+ hemimandibles was radiopaque (indicated by red triangles in Figures 2E1, E2 for incisal enamel, orange triangles for molar enamel), and was well contrasted to the dentin and surrounding alveolar bone. Both enamel on incisors (indicated by red triangles in Figures 2F1, F2) and on molar (indicated by orange triangles in Figure 2F1) of Nckx4 ^−/− mice had a significantly reduced radiopacity. The mineral density in the Nckx4 ^−/− mouse enamel was reduced to 53% of that in the Nckx4 ^+/+ enamel in P1, and 57% in P2, n = 4, *p < 0.05, determined by unpaired two-tailed Student’s t-test (see Figure 2G).

Cross sections of mouse hemimandibles at the planes indicated by P2 in Figures 2E, F showed that there was no significant alteration in the thickness of Nckx4 ^−/− late maturation stage enamel as compared to that in Nckx4 ^+/+ matrix (see Figures 2H, J). SEM analysis revealed a normal decussating pattern of enamel prisms in both Nckx4 ^+/+ and Nckx4 ^−/− developing enamel. However, compared to the well-developed independent enamel rod in Nckx4 ^+/+ matrix (see Figure 2I), no individualized enamel rod was formed in Nckx4 ^−/− enamel matrix (See Figure 2K). Therefore, NCKX4 is critical for the formation of enamel ECM with normal composition and microstructure.

H&E stained sagittal sections of Nckx4 ^+/+ hemimandibles showed a pink colored enamel matrix that gradually grew to its full thickness, then disappeared at the late development stage. The remaining EDTA demineralized enamel space is indicated by black triangles (see Figure 3A). H&E stained Nckx4 ^−/− sections showed a similar secretory stage enamel matrix, but when advancing to its late maturation stage, the enamel matrix in Nckx4 ^−/− mice remained present (indicated by the black triangles in Figure 3B).

Nckx4 ^+/+ secretory ameloblasts (SAB) were elongated and polarized with the characteristic Tomes’ processes (TPs) developed at their apical surface (see Figure 3C), which are responsible for the EMP secretion and decussating pattern of enamel prisms (Zhang et al., 2019). There were no obvious morphological differences between Nckx4 ^+/+ and Nckx4 ^−/− secretory ameloblasts (SAB) (see Figure 3G). When Nckx4 ^+/+ ameloblasts differentiated to maturation stage, as expected, their height reduced to the half of SAB, and there was a well-defined boundary from papillary layer cells at their basal surface (see Figures 3D, E). Ruffle-ended ameloblasts were identifiable as a classic appearance of a foamy vacuolization was present in their apical cytoplasm (Josephsen and Fejerskov, 1977) (see Figure 3D), while smooth-ended ameloblasts had increased intercellular spaces (see Figure 3E). However, no distinguishable ruffle-ended and smooth-ended ameloblasts were found in the Nckx4 ^−/− maturation stage development. In addition, MAB lost their single-cell layer arrangement, and boundary from papillary layer cells was unclear.

The enamel matrix layer was retained throughout Nckx4 ^−/− maturation stage (see Figures 3H, I). Epithelial cell polarity biomarker protein kinase C (PKC) alpha was immunolocalized at the apical of Nckx4 ^+/+ MAB (see Figure 3F), but not in the Nckx4 ^−/− MAB (see Figure 3J).

Immunostaining on Nckx4 ^+/+ enamel matrix showed a reduced amelogenin immunoreactivity that finally disappeared as protein was removed and replaced by mineral in the mature enamel matrix (see white triangles in Figure 4A). In contrast, in Nckx4 ^−/− mouse enamel, amelogenin immunostaining signal continued through the secretory stage, and remained even at the end of maturation stage (see Figure 4B).

SDS-PAGE showed no difference in the secretory stage enamel matrix proteins of Nckx4 ^+/+ mice as compared to Nckx4 ^−/− mice (see Figure 4C), presenting mostly full-length amelogenin with a mass of 25 kDa, and amelogenin without the C-terminus (removed by MMP20) with a mass of 20 kDa (so called amelogenin 20). However, a considerable amount of matrix proteins remained in the maturation stage of the Nckx4 ^−/− enamel matrix. Western blot analysis further showed that amelogenin variants in the secretory stage of Nckx4 ^−/− enamel were similar to those in the Nckx4 ^+/+ enamel, but more amelogenin 20 was detected in the maturation stage of Nckx4 ^−/− enamel as compared to controls (see Figure 4D). The density analyses of the SDS-PAGE and WB data confirmed significantly more enamel matrix proteins, particularly amelogenins, retained in the Nckx4 ^−/− maturation stage of enamel matrix compared to Nckx4 ^+/+ enamel (see Supplementary Figure S1). These results suggest that NCKX4 is necessary for processing and removing protein components as the enamel ECM models from a protein matrix to a mineralized matrix, a novel function of NCKX4 that has not been previously described.

During the maturation phase, KLK4 is the primary enzyme to hydrolyze amelogenin and other EMPs, allowing proteins to be removed and replaced by mineral (Smith, 1998; Ryu et al., 2002). The retention of enamel matrix proteins, in particular amelogenins in Nckx4 ^−/− maturation enamel, led us to investigate the effect of NCKX4 deletion on the KLK4 expression and activity.

Semi-quantitative PCR analyses on mRNA extracted from enamel organ at the transition (P9), early maturation (P11) and middle maturation stage (P13) showed that an increase in Nckx4 ^−/− expression compared to Nckx4 ^+/+ ameloblasts at all three development stages. There was an average of 2.3-fold increase in P9 cells, 1.7-fold in P11 cells, and 1.28-fold in P13 cells (see Figure 5A). The increase of Klk4 mRNA in Nckx4 ^−/− maturation ameloblasts was confirmed by in situ hybridization (see Figures 5B, C).

Western blot showed no effects of Nckx4 deletion on the protein abundance of MMP20 in the secretory stage enamel matrix (see Figure 5D). However, consistent with increased Klk4 mRNA expression levels, there was relatively more KLK4 present in the maturation stage of Nckx4 ^−/− enamel matrix as compared to that in the Nckx4 ^+/+ enamel (see Figure 5E). A quantitative analysis on the density of WB immunoreactive bands confirmed that NCKX4 deletion did not affect MMP20 abundance in the secretory stage of enamel matrix, but resulted in a greater amount of KLK4 presence compared to the Nckx4 ^+/+ enamel matrix (see Figure 5F).

Through agar, fluorogenic quench peptide immobilized on the freshly microdissected maturation stage incisor segments was employed to visualize relative in situ KLK4 activity in the enamel matrix. The blue fluorescence signals from the hydrolyzed substrates (outlined by pink dotted line under the dentin-enamel junctions in Figure 5G), proportionally correspond to the KLK4 activity in the enamel matrix. The results demonstrated that KLK4 activity was decreased in Nckx4 ^−/− enamel matrix, thus unable to completely process matrix proteins for enamel ECM modeling (see Figure 5H).

As NCKX4 is a known ion transport, we first evaluated to what degree that Nckx4 deletion affected the ion composition in enamel matrix. EDX analysis, used to measure the elemental composition in incisor centrolabial enamel underlying the mesial angle of the first molars, showed reduced relative amounts of calcium and potassium by about 50%, while sodium increased about 50% in the Nckx4 ^−/− enamel matrix as compared to that in the Nckx4 ^+/+ enamel (see Figure 6A).

The EDX data led us to measure the KLK4 peptidase activity in the presence of various concentrations of calcium and sodium. Its results showed that while KLK4 activity was independent of calcium concentration (see Figure 6B) but it was significantly decreased in the presence of high (200 and 400 mM) NaCl, both in vitro (see Figure 6C) and in situ, where shows the suppressed KLK4 activity in the maturation stage of Nckx4 ^+/+ enamel matrix (see Figure 6D). KLK4 activity was maximized when NaCl concentration was between 100 and 150 mM. This data implies that the transitorily increased sodium concentration in confined enamel matrix caused by defective NCKX4 could adversely affect KLK4 activity.

We next asked if Nckx4 deletion would affect enamel matrix acidification as a consequence of the reduced calcium phosphate formation? A pH indicator stain of freshly microdissected mandibular incisors showed that on the surface of Nckx4 ^+/+ enamel, acidic pH bands (red) and neutral pH bands cyclically alternated throughout enamel development (see Figure 7A, top panel). However, in the Nckx4 ^−/− incisor enamel, the pH modulations failed to recur and the ECM remained neutral (see Figure 7A, lower panel).

To determine if this change in pH fluctuations affected KLK4 activity, we used an in vitro assay, in which the effect of pH on the ability of KLK4 to hydrolyze recombinant amelogenin 20 was measured. We used amelogenin 20 as the substrate since it is the major amelogenin species present in the early maturation stage of enamel matrix (Ryu et al., 1999). When the hydrolysis was performed in pH 5.25 or pH 6.0, there was less unhydrolyzed amelogenin 20 present on SDS-PAGE gel as compared to that in pH 6.75 or pH7.5 (see Figure 7B). The density analyses on the ratio of unhydrolyzed amelogenin20 vs the input protein confirmed that more amelogenin 20 was hydrolyzed in the acidic pH (see Figure 7C).

Furthermore, the ability of maturation stage Nckx4 ^+/+ incisor enamel matrix to hydrolyze fluorogenic peptide BOC-Val-Pro-Arg-AMC was also more effective in pH 6.2 than in pH 7.2 (see Figure 7D). These are the average pH found in enamel matrix under ruffle-ended ameloblasts and smooth-ended ameloblasts respectively. When there was no KLK4 present, the overall enzymatic activity of Klk4 ^−/− mouse enamel matrix to hydrolyze BOC-Val-Pro-Arg-AMC peptide reduced to 11% of that in wild-type controls (see Figures 7D, E), indicating that KLK4 is responsible for about 89% of enzymatic activity of wild-type maturation stage of enamel matrix. There was no difference in the enzymatic activity of Klk4 ^−/− enamel matrix in pH 6.2 and pH7.2 in term of hydrolyzing BOC-Val-Pro-Arg-AMC peptide, suggesting that other peptidases in Klk4 ^−/− maturation stage of enamel were independent of pH.

NCKX4 loss-of-function mouse colony (Nckx4 ^−/− ) with deletion of exon 6 and 7 (Li and Lytton, 2014) is a gift from Dr. Jonathan Lytton. KLK4 loss-of-function mouse colony (Klk4 ^−/− ) was purchased from The Jackson Laboratory. All animals were maintained in the UCSF animal care facility, which is a barrier facility, accredited by Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). All experimental procedures associated with these mouse models were approved by the Institutional Animal Care and Use Committee (IACUC) under the protocol AN183449-01.

Mice at postnatal seven-week-old (P7W) were collected by following standard IACUC protocols. Briefly, mice were anesthetized with 240 mg/kg tribromoethanol (Sigma-Aldrich), and perfuse-fixed with PBS and 4% paraformaldehyde (PFA). Hemimandibles were dissected and post-fixed with 4% PFA for 24 h at 4°C, then followed by decalcification in 8% EDTA at 4°C for 3 weeks with EDTA change every other day. The hemimandibles were then processed, embedded with wax and sectioned along their sagittal planes. Sections were stained with Hematoxylin & Eosin (H&E) for morphology assessment.

For immunostaining, the sagittal sections were boiled in 10 mM citrate buffer (pH 6.0) for 20 min to retrieve antigens. Next, the sections were incubated with GeneTex Universal Protein Blocking reagent for 1 h, followed by incubation with mouse anti-NCKX4 antibody (UC Davis/NIH NeuroMab Facility) or rabbit anti-amelogenin antibody for overnight at 4°C. Sections were then incubated with Alexa 594 fluorescein-conjugated anti-mouse IgG or FITC-conjugated anti-rabbit IgG (Life Technologies) for 1 h at RT, followed by counterstaining with 1 μg/mL Hoechst for 5 min (Life Technologies). The slides were imaged using a high-speed Leica TCS SP5 spectrum confocal microscope. To evaluate ameloblast polarity, after blocking with Universal Protein Blocking reagent, the sections were incubated with rabbit anti-PKC (protein kinase C) alpha antibody (Abcam) for overnight at 4°C. After washing, the sections were then incubated with HRP-conjugated goat anti-rabbit antibody for 1 h at RT, at last incubated with HRP substrate (Vector®) to reveal the positive immunoreactive signals.

RNAscope in situ hybridization 2.5 HD Red Detection Kit and mouse Nckx4 and Klk4 probes (Advanced Cell Diagnostics, ACDbio.) were used by following the manufacturer’s instructions. Briefly, sagittal sections of mouse hemimandibles at 5 µm of thickness were boiled in the target retrieval buffer for 15 min and then pre-treated with protease plus solution at 40°C for 15 min. The sections were next incubated with probes for 2 h. After incubating with a serial of amplification reagents, the sections were incubated with Fast Red complex to develop the red precipitates in which the intensity is proportional to the mRNA expression levels of target genes. The nuclei were lightly counterstained by hematoxylin.

P7W mouse hemimandibles were air-dried prior to imaging with a Leica M165C digital stereo microscope and a LAS V4.2 software. Hemimandibles were then subjected to X-ray micro-tomography scanning using a Micro XCT-200 system (Xradia). All scans were done at an operating voltage of 90 KVp and 66 μA of current, at an optical magnification ×2. A binning of 2 was used for 3D image reconstruction. Virtual sections were converted to bmp images using the Xradia TXM3DViewer 1.1.6. software. Mineral densities were determined following a detailed calibration protocol of the micro-XCT described previously (Djomehri et al., 2015). Ten orthogonal sections from each mouse (n = 3) to the long-axis of hemimandible at the plane perpendicularly across either the second buccal cusps of the first molar (P1 in Figures 2E, F) or the onset of mesial angle of the first molar (P2 in Figures 2E, F) were collected for mineral density analysis. The mineral density of calcified tissues from Nckx4 ^+/+ and Nckx4 ^−/− mice was compared using unpaired two-tailed Student’s t-test.

Pups at postnatal day 5, 7, 8, 9, 11, and 13 (P5, P7,P8, P9, P11, and P13) were euthanized with carbon dioxide asphyxiation followed by cervical dislocation, mandibles were dissected and first molars from five mandibles were extracted and pooled as one sample. Enamel epithelium overlying these first molars is at the secretory, transition and maturation stage of development respectively (Ryu et al., 2002). The total RNA of these cells was purified using Qiagen RNeasy Micro Kit and mRNAs were reverse-transcribed into cDNA libraries using SuperScriptTM III First-Strand Synthesis System (Life Technologies). Semi-quantitative PCR was performed to quantify the relative expression levels of target genes. The sequences of primers used to amplify mouse endogenous Gapdh and target genes were listed as follows: Gapdh sense- 5′-TGG​CCT​TCC​GTG​TTC​CTA​C-3′, antisense- 5′-GAG​TTG​CTG​TTG​AAG​TCG​CA-3′; Nckx4 sense- 5′- CAC​GGA​GAT​GTC​GGT​GTA​GGA-3′, antisense- 5′- CAC​CAC​CTG​TCC​AGC​AAA​GAG-3′; amelogenin sense- 5′-GGG​ACC​TGG​ATT​TTG​TTT​GCC-3′, antisense- 5′-TTC​AAA​GGG​GTA​AGC​ACC​TCA;-3′; enamelin sense- 5′-GCT​TTG​GCT​CCA​ATT​CAA​A-3′, antisense- 5′-AGG​ACT​TTC​AGT​GGG​TGT-3′, ameloblastin sense- 5′-CTG​TTA​CCA​AAG​GCC​CTG​AA-3′, antisense- 5′-GCC​ATT​TGT​GAA​AGG​AGA​GC-3′; Mmp20 sense- 5′-TCC​AAG​CAT​TAT​ACG​GAC​CCC-3′, antisense- 5′-GTC​ACT​GCA​TCA​AAG​GAC​GAG-3′; Klk4 sense- 5′-CGG​GAG​TCT​TGG​TGC​ATC​C -3′, antisense- 5′- CTT​GGG​AGC​CTT​TCA​GGT​TAT​G -3′. To determine the relative expression levels of these target genes, the comparative threshold cycle method was used as previously published (Ryu et al., 1999).

Three sets of hemimandibles from either P7W Nckx4 ^+/+ or Nckx4 ^−/− mice were embedded in epoxy, and polished in an orientation crossing to the long-axis of incisor using a series of SiC paper and diamond polishing suspension. Subsequently specimens were etched with 10% HCl for 30 s. Specimens were sputter-coated with Au-Pd, and the thickness of incisal enamel matrix and enamel rods were imaged at 20 kV using a Scanning Electron Microscope, Quanta 3D FEG (FEI) as described previously (Habelitz, 2015).

Hemimandibles were dissected, alveolar bone at the lingual site of hemimandibles was carefully removed to expose incisors. Incisors were fragmented using molars as reference line as previous published (Hu et al., 2007). The incisor segments under the second and third molars were used to extract secretory stage enamel matrix. The incisor segments from the bottom of the first molar to the gingival margin were used to extract the maturation stage matrix. The stage-specific enamel matrix proteins were extracted as previously described (Yamakoshi et al., 2011). The incisor segments from 6 mice were pooled and incubated in 1 mL of 0.17 N HCl/0.95% formic acid for 2 h at 4°C with constant rotation. After the undissolved material was removed by centrifugation at 3500g in 4°C, the protein supernatant was subjected to buffer exchange with 0.01% formic acid using a centrifugal 3K-filter unit (EMD Millipore). The proteins retained by the filter were eluted with 250 μL of 0.01% formic acid and used for subsequent sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), Coomassie Brilliant Blue staining, and western blot immunoblotting.

Equal volumes of total enamel matrix proteins from six mice at either secretory stage or maturation stage were loaded to be separated on a 15% SDS-PAGE gel. Proteins on the SDS-PAGE were transferred to a PVDF membrane. After blocking, membrane was then incubated with rabbit anti-amelogenin IgG, rabbit anti-MMP20 IgG (Millipore-Sigma), or rabbit anti-KLK4 IgG (Abcam) for overnight at 4°C. Following washing, PVDF was secondly incubated with IRDye 680RD conjugated anti-rabbit IgG (Li-Cor) for 1 h at room temperature. The membranes were thoroughly washed and then scanned using an Odyssey Imaging System.

The activity of KLK4 to hydrolyze a peptide substrate or recombinant protein amelogenin 20 (amelogenin without C-terminus) was analyzed in various buffer. Briefly, recombinant KLK4 (R&D Systems) was activated by thermolysin (R&D Systems) by following the manufacturer’s instruction and previous publications (Yoon et al., 2007; Tye et al., 2011). Active KLK4 was then incubated with a fluorogenic quench peptide substrate BOC-Val-Pro-Arg-AMC (R&D Systems) in the 50 mM Tris-HCl supplemented with various concentrations of CaCl₂ or NaCl. To test the effect of pH, we used buffer supplemented with 50 mM Tris-HCl, 150 mM NaCl and 10 mM CaCl₂. Once the Arg-AMC amide bond in the peptide is hydrolyzed by KLK4, highly fluorescent AMC is released. The intensity of fluorescence proportionally indicates the activity of KLK4. One hundred microliter reaction was loaded to each well of a 96-well black plate. Each condition was quadruplicate. The plate was read at excitation and emission wavelengths of 380 nm and 460 nm respectively. A kinetic mode was set to read each well every 100 s for a total of 2,700 s interval.

To synthesize amelogenin 20, also named as rH146 in previous publication (Bai et al., 2020), a site-directed mutagenesis kit (Takara Bio.) was employed to delete C-terminus 28bp DNA from human wild-type amelogenin cDNA (rH174). After the site-specific deletion was confirmed by DNA sequencing, amelogenin 20 was expressed in T7 Express Competent Escherichia coli (New England Biolabs) and purified using an acid/heat treatment method described previously (Svensson Bonde and Bulow, 2012). Amelogenin 20 recombinant protein was incubated with or without active KLK4 (at a concentration ratio of 100:1) in 50 mM Tris-HCl, 150 mM NaCl and 10 mM CaCl₂ at various pH for overnight at 37°C with rotation. Then protein was loaded to a 15% SDS-PAGE. The gel was scanned and protein band density was measured using NIH ImageJ version 2.0.0. The percentage of undigested protein normalized to the input amelogenin 20 was calculated. An average of percentage from triplicate analyses was calculated and the significant difference among the different conditions was determined by one-way ANOVA analysis using SPSS Statistics package 19 followed by Tukey Post Hoc Tests.

To analyze the activity of native KLK4 possessed in the maturation phase of enamel matrix, the incisor segments overlying the bottom of the first molar to the gingival margin (maturations stage) were microdissected and incubated with fluorogenic peptide BOC-Val-Pro-Arg-AMC. The fluorescence was recorded every 100 s for a total of 2,700 s interval by a plate reader. To assess KLK4 enamel in vivo activity, the freshly microdissected incisor segments at the maturation stage as described above were quickly dipped into 1% agar in 50 mM Tris-HCl (pH 7.5), 150 mM NaCl and 10 mM CaCl₂ supplemented with fluorogenic peptide BOC-Val-Pro-Arg-AMC. One microliter peptide was supplemented to each 100-μL of 1% agar. Next, the incisor segments embedded by agar were incubated in a black Eppendorf tube for 30 min at 37°C. After incubation was complete, the incisal segments were photographed immediately using a Leica fluorescence microdissection microscope.

Hemimandibles from three 7-week old mice were embedded in epoxy, and polished in a cross orientation using a series of SiC paper and diamond polishing suspension starting from incisor tip and stopping when mesial angle of the first molar was exposed. For each hemimandible, six randomly areas (as illustrated in Supplementary Figure S1) were subjected to an energy dispersive X-ray spectroscopy (Quantax EDS, Bruker Nano Inc.) to collect elemental composition of Ca, P, Na et al at 15 keV using a variable pressure chamber. Back scattered electron (BSE) images were also collected using a field emission scanning electron microscope (Sigma VP500 field emission SEM, Carl Zeiss Microscopy). The elemental composition collected from three Nckx4 ^+/+ or Nckx4 ^−/− mice was compared and the significance difference between two mouse models was determined by unpaired two-tailed Student’s t-test.

To prepare the staining solution, 100 mg Methyl Red (Sigma-Aldrich) was diluted in 45 mL of 95% methanol. Four Nckx4 ^+/+ or Nckx4 ^−/− mice at postnatal 7 weeks old (P7W) were anesthetized with 240 mg/kg tribromoethanol (Sigma-Aldrich). Hemimandibles were dissected and the surrounding alveolar bone was removed to completely expose the mandibular incisors. The enamel organ epithelium overlying the labial enamel surface was gently wiped off with ice-cold Kimwipe. The incisors were immediately soaked in 0.2% Methyl Red/95% methanol for 45 s at room temperature, then air-dried for 5 min prior to documenting the stained incisors under a Leica dissecting microscope.