Voltage-dependent Ca2+ channels promote branching morphogenesis of salivary glands by patterning differential growth

Branching morphogenesis is a crucial part of early developmental processes in diverse organs, but the detailed mechanism of this morphogenic event remains to be elucidated. Here we introduce an unknown mechanism leading to branching morphogenesis using mouse embryonic organotypic cultures with time-lapse live imaging. We found spatially expressed L-type voltage-dependent Ca2+ channels (VDCCs) in the peripheral layers of developing epithelial buds and identified the VDCCs as a core signaling mediator for patterning branching architecture. In this process, differential growth in peripheral layers by VDCC-induced ERK activity promoted cleft formation through an epithelial buckling-folding mechanism. Our findings reveal an unexpected role of VDCCs in developmental processes, and address a fundamental question regarding the initial process of branching morphogenesis.

] o and branching morphogenesis using a database of salivary gland gene expression (http://sgmap.nidcr.nih.gov/sgmap/sgexp.html) 18 . Among the molecules related to [Ca 2+ ] o transport, we identified that several types of VDCCs, transient receptor potential (TRP) channels, and stromal interaction molecule (STIM) 1 are highly expressed in the critical period for branching organization [embryonic day (E) [12][13][14][15][16]. We then blocked the action of each of these components in developing SMG cultures with various chemical antagonists and found that nifedipine (an L-type VDCC antagonist) strikingly diminished new bud formation (Fig. 1C,D). We then confirmed the dose dependency and L-type specificity of this inhibitory effect ( Fig. 1E-G). Nifedipine also clearly suppressed branching morphogenesis in mouse embryonic lung cultures, suggesting that this finding has broad implications for diverse organs (Supplementary Fig. S1A-C). To accurately evaluate the morphological consequences, we monitored developing SMGs for 18 h (from E13) by time-lapse live imaging. The serial images of the development pattern revealed that nifedipine-treated SMGs failed to progress a new cleft, resulting in no additional bud formation ( Fig. 1H and Supplementary Video 1). We next cultured isolated epithelial buds of SMGs (eSMGs) and verified the purity of the cultures ( Supplementary Fig. S1D,E) and the inhibitory effect of nifedipine on cleft formation (Fig. 1I). These results indicate that a major driving force of cleft formation is derived from the intrinsic physiological effect of VDCCs in the epithelial bud and not in the surrounding mesenchyme.
Localized expression of VDCCs in developing SMGs. This newly identified function of L-type VDCCs in epithelial bud development led us to verify the expression of these channels in SMG compartments ( Fig. 2A). Among the four subtypes of L-type VDCC (Ca V 1.1 to 1.4), three types (Ca V 1.1 to 1.3) were detected in both the mesenchyme and epithelial buds, but the epithelial portion had a mRNA expression level of approximately 1% compared to the mesenchyme (Fig. 2B). Instead, immunostaining revealed a localized expression pattern of VDCCs that was exclusively concentrated in the peripheral cell layers of the epithelial buds (Fig. 2C). Based on quantitative analysis, over 50% of the VDCCs were expressed within the three outermost layers of the epithelial buds ( Supplementary Fig. S2A). The same expression patterns were confirmed in eSMG ( Supplementary  Fig. S2B) and lung cultures ( Supplementary Fig. S2C) by immunostaining and fluorescence in situ hybridization ( Supplementary Fig. S2D). This characteristic localized expression pattern may explain the inconsistency between the obvious function of VDCCs in bud formation and the low expression of the channels in epithelial  (Figs 1F and 2B). Moreover, a higher Ca 2+ level was detected in the peripheral cell membranes of eSMGs by expression of a membrane-tethered Ca 2+ biosensor (GCaMP6s-CAAX), implying functional expression of the channels ( Supplementary Fig. S2E).
Next, we probed the molecular mechanism underlying localized expression of VDCCs. The growth factorreceptor tyrosine kinase (RTK) pathway is a representative signaling cascade that plays versatile roles in branching morphogenesis 3,19 . The growth factor signal exogenously guides spatial patterns of organ architecture through interaction with the extracellular matrix 20 . Therefore, we investigated RTK activity in epithelial buds by visualizing the spatial pattern of immunolabeled phosphorylation of tyrosine residues (pTyr) in eSMG cultures and a found striking pattern of pTyr concentrated in the peripheral epithelial layers (Fig. 2D). Based on this result, we determined that the RTK signal is essential for VDCC expression regardless of growth factor subtype specificity as demonstrated by the decrease in VDCC expression caused by removing epidermal growth factor (EGF) and/or fibroblast growth factor (FGF) from the eSMG culture media (see Methods section; Fig. 2E). The expression level of VDCCs was also significantly decreased by treatment with a pan-RTK inhibitor (AP24534) (Fig. 2F).
Spatial relationship between VDCCs and the MAPK pathway. Next, we searched for the signaling mediator of branching morphogenesis induced by localized VDCC activity. It has been reported that mitogen-activated protein kinase (MAPK) also shows localized activity confined to the peripheral region of the epithelial bud 21 , suggesting its correlation with VDCC expression patterns. Moreover, VDCCs are known to activate Ras, an upstream component of the MAPK pathway, through localized Ca 2+ -calmodulin (CaM) interaction 22,23 . Immunostaining results confirmed higher phosphorylated ERK (pERK) signals in the peripheral region of eSMG cultures (Fig. 3A,B), which were highly spatially correlated with VDCC expression patterns (Fig. 3C,D; R 2 = 0.8573). To verify the signaling hierarchy between VDCCs and ERK, we treated SMG cultures with either U0126 (a MEK inhibitor) or nifedipine, and compared the resulting changes in respective signaling activity (Fig. 3E). While U0126 did not affect the expression level of VDCCs, nifedipine reduced ERK phosphorylation (−28.61%, Fig. 3F and Supplementary Fig. S3A), indicating that VDCCs are an upstream mediator of ERK. This hierarchy was additionally confirmed by simultaneous monitoring of intracellular Ca 2+ (G-CaMP6s) and ERK activity (ERK-dTomato) in rat submandibular gland epithelial cells (SMG-C6) upon KCl depolarization. Application of KCl immediately increased G-CaMP6s signals, and subsequent nuclear translocation of ERK-dTomato was detected ( Fig. 3G and Supplementary Video 2). This effect was significantly blocked by nifedipine treatment (Fig. 3H). We also dissected the signaling pathway that couples VDCCs to ERK, seeking to identify pathway intermediates. To this end, we conducted an in-depth study of Ras activity using fluorescence resonance energy transfer (FRET) probes (RaichuEV-HRas) 24 in SMG-C6 cells (Fig. 3I). The activation of VDCCs induced a rapid and sustained increase in Ras activity, and this increase was completely abolished by preincubation with the Ca 2+ -CaM binding inhibitor, trifluoperazine (Fig. 3J). Taken together, these results clearly establish a connection between VDCC activity and ERK phosphorylation, demonstrating an intermediary role for Ca 2+ / CaM-dependent Ras activation. Because the Ras-MAPK pathway is also known as a downstream of RTKs, we next compared ERK activity in response to VDCC and growth factor signaling inputs through immunoblotting. KCl treatment yielded a higher pERK level in SMG-C6 cells than EGF treatment, and combined EGF-KCl treatment resulted in a synergistic increase in the phosphorylation level ( Supplementary Fig. S3B). We then evaluated SMG morphology following U0126 application and confirmed a similar inhibitory effect with nifedipine treatment ( Supplementary Fig. S3C,D). These data indicate that the VDCC-ERK cascade promotes branching morphogenesis in developing SMGs. Differential growth promotes cleft formation. How can VDCC-ERK signals trigger the branching process? We focused on the concept of differential growth, in which localized (or patterned) proliferation organizes epithelial architecture during the initial developmental process 25 . Given this background, we hypothesized that ERK-induced localized proliferation in the peripheral layers governs both bud outgrowth (increasing organ size) and cleft formation (increasing bud number), and that the fate of the developing pattern is determined by the mitosis orientation (Fig. 4A). In particular, an increase in peripheral cell density by differential growth with horizontally-directed mitosis was assumed to be a major driving force in cleft formation through epithelial buckling-folding mechanisms 26 . We initially quantified the local distribution of mitotic cells in branching epithelial buds to characterize the differential growth patterns (see Methods section; Supplementary Fig. S4A). As expected, 73.5% of mitoses occurred in the peripheral layers (defined as the outermost three layers) (Fig. 4B), and mitotic cell density in the developing buds was significantly reduced by nifedipine or U0126 treatment [84.00 (control), 36.28 (nifedipine), and 22.28 (U0126) mitotic cells; Fig. 4C,D). Next, mitosis orientation was measured based on the angle between the mitotic axis and the bud surface ( Fig. 4E; see Methods section). The measured mitosis angle (θ) in the peripheral layers showed a higher distribution in the horizontal direction (0° < θ < 45°) than in the vertical direction (45° < θ < 90°) with an approximately 2:1 ratio [62.8% (horizontal) versus 37.2% (vertical); Fig. 4F,G]. However, inhibition of VDCCs did not result in a notable change in the mitosis orientation (Fig. 4G). In the U0126-treated buds, it was difficult to measure the mitotic angle due to lower mitotic cell density than in the nifedipine-treated group (Fig. 4D). These data indicate that the VDCC-ERK cascade is involved in inducing mitotic signals rather than in regulating mitotic orientation.
We also investigated the spatial rearrangement of the peripheral epithelium of developing buds by live staining with Hoechst dye for a short period (<1 h), enabling selective staining of the peripheral nuclei (see  Fig. S4B), indicating that increased peripheral cell density was a major triggering factor for spatial rearrangement of the epithelial structures. The increase in peripheral cell density was accelerated by outward migration of the inner cells, as characterized by pseudostratified-like epithelial regions in the outermost cell layer observed in fixed eSMG cultures ( Supplementary Fig. S4C,D). These results support our hypothesis explaining the triggering mechanism of branching morphogenesis: localized epithelial proliferation is a crucial mechanism for epithelial infolding and the resultant cleft formation (Fig. 5).

Discussion
In this study, we demonstrated that spatial rearrangement of epithelial layers by VDCC-induced localized proliferation is a key inducer of branching morphogenesis (Fig. 5). Epithelial proliferation is generally thought to play a central role in the overall developmental process by providing new cells to occupy the enlarging organ space. However, the functional relationship between epithelial proliferation and the branching process (including budding and clefting) has been a contentious issue in various epithelial organs 25,27 . Nakanishi et al. reported that proliferation of epithelial buds is not required for early-cleft formation in developing SMG cultures, a result at odds with our findings 28 . There are a number of possible explanations for these different outcomes. The first is limitations in the methods applied in both studies. The previous report used a DNA-based approach with X-ray irradiation and radioactive nucleotides. In our study, we blocked proliferation through chemical perturbation of MEK signaling and analyzed the proliferation rate by counting the mitotic cell number. Unfortunately, no method can ensure complete inhibition or accurate detection of cell proliferation, both of which are essential prerequisites for addressing questions relating to such a complex process. Second, the selection criteria for clefts in our study were different. As a practical matter, static images of developing buds are a somewhat ambiguous tool for detecting actual clefts, and thus we used a real-time monitoring system to accurately detect the whole process of the cleft formation (Fig. 1H,I). Using this method, we could exclude dimple-like structures, which occur through transient flexion of the outer epithelial layers. Overall, we suggest that these conflicts primarily reflect the different experimental approaches and interpretation of the data. Although previous reports have tended to regard epithelial bud proliferation as a phenomenon distinct from cleft formation, our work compels the conclusion that these two events are reciprocally related during early branching morphogenesis.
The effects of VDCC on branching morphogenesis seen in SMG cultures were experimentally reproduced in lung cultures ( Supplementary Fig. S1A-C), enhancing the biological relevance of our findings. The ERK signal, which we determined acts as a core downstream effector of the branching process, was previously reported to regulate the length and thickness of developing lung branches by affecting mitosis orientation 8 . The mitosis angle was typically arranged toward the elongating direction of the airway tubes, and enhanced ERK activity perturbed this orientation, resulting in the alteration of branching patterns in developing lungs (reduced length and increased thickness). In SMG cultures, mitosis orientation was horizontally arranged in relation to the outer surface of epithelial buds, which might be the reason for the spherical shape of SMG buds rather than an elongated morphology. In this context, we found that ERK activity was preferentially involved in localized induction of mitosis rather than affecting orientation and that the spatial distribution of epithelial proliferation is crucial for patterning differential growth. Given this set of results, ERK activity and related mitotic characteristics-orientation and spatial distribution-can be regarded as crucial factors for determining branching patterns among different epithelial organs. Moreover, we suggested the growth factor signal as a determinant factor of VDCC expression patterns. To date, diverse growth factors and related feedback systems have been introduced to account for the patterning of branching structures by computational modeling 29 . Recently reported model based on FGF-SHH feedback signals (ligand-receptor-based Turing mechanism) could explain a general mechanism for the regulation of stereotyped branching in diverse organs 30 . Through this study, we revealed that the growth factor signals patterning branching structures are also involved in patterning VDCC expression (Fig. 2D,F). Given signaling connectivity proposes that VDCC is a pivotal mediator in the ligand-receptor-based developmental program by providing supporting proliferation signals.
This report not only provides a plausible explanation for the mechanism of branching morphogenesis, also expands the functional range of VDCCs beyond the previously well-known functions in excitable cells such as synaptic transmission and cardiac pacemaking 31 . We also introduced the synergistic role of VDCCs and growth factor signals in ERK activation, which is crucial for the proliferation of epithelial organs. The VDCC-ERK signaling cascade was firstly introduced in depolarization-induced ERK activation in neurons, which is crucial for synaptic plasticity and learning and memory 22,32 . In that context, Ca 2+ influx through VDCC transduces signals from plasma membrane to nucleus through CaM-dependent MAPK pathway 23 . In this study, we demonstrated the spatial relationship between VDCC expression and ERK activity, as well as the connectivity of the signals, using SMG culture models and cell lines expressing diverse genetically-encoded biosensors (Fig. 3G-J). However, the complete map of this pathway has not been established, and specifically, the identity of the guanine nucleotide exchange factor (GEF) responsible for CaM and Ras activation remains an important question. In light of our results, we expect to elucidate the additional roles of this membrane-to-nucleus signal in diverse biological systems. With regard to clinical applications, we expect that our findings will provide a fundamental basis for developing regenerative approaches in various organs.
Imaging equipment and procedures. SMG morphological evaluation was performed using a digital inverted fluorescence microscope (Nikon, Tokyo, Japan; Ti) equipped with a digital camera (Nikon, DS-Ri2) and a CFI Plan Fluor 4x objective (Nikon) or JuLI Br live cell movie analyzer (NanoEnTek, Seoul, Republic of Korea). Immunofluorescence images were taken by confocal laser scanning microscope (Carl Zeiss, Oberkochen, Germany;LSM700) equipped with Plan-Apochromat 10x, Plan-Apochromat 20x, and C-Apochromat 40x objectives (Carl Zeiss) and with 405, 488, and 555 nm wavelength excitation lasers. Live imaging of epithelial rudiments of SMG and SMG-C6 cells were conducted through a confocal microscope (Carl Zeiss) with a customized live cell chamber (Live Cell Instruments, Seoul, Republic of Korea) that maintained 5% CO 2 and 37 °C conditions. To visualize peripheral cell movement (Fig. 4I,J), the epithelial rudiments of SMGs were briefly stained with 1 μg/ml Hoechst 33342 (Thermo Fisher Scientific, Waltham, MA; H3570) -culture media solution for 1 h. After staining, cells were washed with culture medium two times.

Adeno-associated virus (AAV) production and transduction.
AAVs were produced and purified with a simplified polyethylene glycol (PEG)-based method 39 . For AAV plasmid transfection, human embryonic kidney (HEK)-293T cells were prepared with 70~80% confluence in Dulbecco's modified Eagle's medium (DMEM; WelGene, Daegu, Republic of Korea; LM-001-05) containing 10% fetal bovine serum (FBS). Lipofection was conducted using Lipofectamine 2000. AAV plasmids-AAV-CAG-GCaMP6s-CAAX, pHelper, and pAAV-RC1 were transfected at a 1:1:1 ratio. After 48 h, the transfected cells were detached by brief treatment of 0.5 M EDTA solution (pH 8), and collected by centrifugation at 1000 rpm for 10 min. The cell pellets were resuspended in phosphate buffered saline (PBS) and induced to release viral particles by repeated freeze-thaw cycles between −80 °C (deep freezer) and 37 °C (water bath). After centrifugation (13200 rpm, 10 min), the supernatants were mixed with 40% polyethylene glycol (Sigma-Aldrich, 89510) solution with 2.5 N NaCl at a 1:4 ratio. The mixture was incubated at 4 °C for 1 h, then centrifuged at 2000 rpm for 30 min. The supernatants were replaced with HEPES buffer-chloroform 1:1 solution, followed by vortexing (2 min) and centrifugation (400 rpm, 5 min). The upper solution in separated layers was collected and the chloroform was allowed to evaporate for 30 min. The collected AAV solution was dialyzed by two steps with sequential use of dialysis tubes with different pore sizes (3 KDa and 50 KDa nominal molecular weight limit; Millipore, UFC8003 and 4310). Dialyzed AAVs (1 × 10 11-12 copies/ml) were diluted in DMEM/F12 containing 1% penicillin-streptomycin. Epithelial rudiments of SMGs were incubated in the viral media for 1 h at room temperature. The rudiments were washed two times with DMEM/F12 containing 1% penicillin-streptomycin, and incubated in Matrigel.

PCR.
Total RNA of SMG tissues and cells was extracted by RNeasy Mini Kit (Qiagen, Hilden, Germany; 74140).
1 μg of total RNA was used for synthesizing cDNA through reverse transcriptase (SuperScript III First-Strand Synthesis System; Thermo Fischer Scientific, 18080-051) with oligo-dT and random hexamer primers. Nested PCR (Supplementary Fig. S1E) was conducted using Platinum Taq DNA Polymerase (Thermo Fischer Scientific, 10966-018). Real-time PCR was performed using SYBR PCR master mix (Applied Biosystems, Foster City, CA; 4309155) with a real-time PCR instrument (Applied Biosystems, 7200). The sequences of primers were as follows (5′ to 3′) 40    Immunoblotting. SMG-C6 cells were lysed in ice-cold RIPA buffer (GenDEPOT, Barker, TX; R4200-010) and protein concentrations were measured using s spectrophotometer (Nanodrop; Thermo Fischer Scientific, ND-1000). Protein samples were separated using 10% SDS-PAGE gels (Bio-Rad, Hercules, CA). After electrophoresis in a Power-Pac Basic system (Bio-Rad), proteins were transferred to nitrocellulose membranes using an iBLOT 2 Dry Blotting system (Thermo Fisher Scientific, IB21001). The membranes were blocked with 10% non-fat milk and incubated with anti-ERK antibodies (1:1000; Cell Signaling Technology, 9102) and anti-pERK antibodies (1:1000; Cell signaling, 9101) at 4 °C overnight. After washing, membranes were incubated with anti-rabbit IgG-HRP (1:5000; Santa Cruz Biotechnology, sc-2030). Immunoreactivity was visualized by ECL reagents (Thermo Fisher Scientific, 32106) and detected by the Chemidoc XRS+ system (Bio-Rad Laboratories). Data analysis. Images were analyzed using Fiji software (National Institutes of Health). Bud numbers of SMG cultures were manually counted based on phase contrast images. To measure VDCC expression (Fig. 2F), we calculated the average intensity of inmmunolabeled VDCC signals on epithelial membrane of whole eSMG culture. Cell movement in the peripheral layer of SMGs (Fig. 4J) was recorded by manual tracking based on confocal fluorescent images. To identify mitotic cells (Fig. 4B,F and G), we selected the cells showing centrally-arranged and condensed DAPI signals between two separated mitotic centers represented by condensed γ-tubulin signals ( Fig. 4E and Supplementary Fig. S4A). The mitotic angle (θ) was calculated from parameters in Z-stack images (step width: 1 µm) of mitotic cells taken by a confocal microscope (Carl Zeiss). The equation is as follows: a: Z-stack distance between two γ-tubulin signals; b: horizontal distance between two γ-tubulin signals when the signals were orthogonally projected to a single virtual plane; a 2 + b 2 : actual distance between two γ-tubulin signals; c: difference between distances of each γ-tubulin signal-to-acinar surface.

Statistical analysis.
All experiments were performed at least two times for biological replicates. The difference between the two groups was determined using a two-tailed t-test. Multiple significance tests were performed using one-way ANOVA, and post hoc analysis was conducted via two-tailed t-test with Bonferroni correction. The Analysis ToolPak-VBA in Excel was used for all statistical analyses.

Data availability statement.
All data from the current study that were generated or analyzed are available upon reasonable request from the corresponding author. All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.