L-type voltage-gated Ca²⁺ channel Ca_V1.2 regulates chondrogenesis during limb development

Significance

Membrane potential is the difference in electric potential between the inside and outside of a cell. Cells can alter their membrane potential through modulating ion channels and pumps. The role of membrane potential has been well studied in a variety of settings, including as a mechanism of coding and transmitting information in neurons. However, far less is known about the role of regulated membrane potential in other contexts, such as in cell fate decisions during embryonic morphogenesis. We show a critical role for membrane depolarization in triggering chondrogenesis during limb development, mediated through the opening of a voltage-gated Ca²⁺ channel, Ca_V1.2, which in turn activates expression of downstream genes, expanding our knowledge of the role bioelectric signals play in development.

Abstract

All cells, including nonexcitable cells, maintain a discrete transmembrane potential (V_mem), and have the capacity to modulate V_mem and respond to their own and neighbors’ changes in V_mem. Spatiotemporal variations have been described in developing embryonic tissues and in some cases have been implicated in influencing developmental processes. Yet, how such changes in V_mem are converted into intracellular inputs that in turn regulate developmental gene expression and coordinate patterned tissue formation, has remained elusive. Here we document that the V_mem of limb mesenchyme switches from a hyperpolarized to depolarized state during early chondrocyte differentiation. This change in V_mem increases intracellular Ca²⁺ signaling through Ca²⁺ influx, via Ca_V1.2, 1 of L-type voltage-gated Ca²⁺ channels (VGCCs). We find that Ca_V1.2 activity is essential for chondrogenesis in the developing limbs. Pharmacological inhibition by an L-type VGCC specific blocker, or limb-specific deletion of Ca_V1.2, down-regulates expression of genes essential for chondrocyte differentiation, including Sox9, Col2a1, and Agc1, and thus disturbs proper cartilage formation. The Ca²⁺-dependent transcription factor NFATc1, which is a known major transducer of intracellular Ca²⁺ signaling, partly rescues Sox9 expression. These data reveal instructive roles of Ca_V1.2 in limb development, and more generally expand our understanding of how modulation of membrane potential is used as a mechanism of developmental regulation.

Although most often thought of in the context of excitable cells such as neurons, all cells maintain, and respond to changes in, transmembrane potential (the difference in electric voltage between the interior and exterior of a cell, V_mem). Changes in V_mem are involved in a wide variety of cellular behaviors including cellular proliferation, survival, differentiation and motility (1). These V_mem changes are mainly determined by ion influx/efflux, mediated by channels, pumps, and gap-junctions embedded in the plasma membrane (2). As some ion channels are expressed with temporal and tissue specificity, endogenous spatiotemporal gradients of V_mem can be generated across tissues and organs (1). Consistent with this, changes in V_mem are known to influence embryonic development, among other processes (2–11).

The impacts of changes in V_mem (depolarization or hyperpolarization) are often mediated through the opening or closing of voltage-sensitive ion channels, such as the voltage-gated Ca²⁺ channel (VGCC) (12, 13). The extracellular Ca²⁺ concentration is thousands of times higher than the intracellular concentration under physiological conditions, hence the opening of VGCCs by membrane depolarization allows Ca²⁺ ions to flow into cells (14). In addition to mediating action potentials in excitable cells, cytoplasmic Ca²⁺ ions entering through VGCCs act as intracellular second messengers, activating several downstream signaling pathways, such as the calmodulin (CaM)/calcineurin and the PKC/MAPK pathways. The cellular impact of VGCC activation has been particularly well studied in the context of muscle differentiation (15, 16), and neuronal proliferation, migration, and differentiation (17, 18). However, VGCCs are also found in nonexcitable cells (19). For example, Ca_V2.1, a P/Q- type VGCC, is abundant on the membrane of human red blood cells (20), and Ca_V1.2, L-type, and Ca_V3.2 T-type VGCCs are expressed in murine chondrocytes and osteoblasts (21). At a functional level, Ca_V3.2 has been shown to play an essential role in tracheal cartilage formation (22). VGCCs are also correlated with congenital malformations in humans, including the presence of syndactyly in Timothy syndrome patients who carry a heterozygous mutation in the CACNA1C gene encoding the pore-forming subunit of Ca_V1.2 (23). Despite these intriguing findings, however, the elucidation of definitive roles for V_mem and voltage-gated Ca²⁺ channels in development, especially in amniotes, has remained elusive.

Here we take advantage of the well-studied developing chick and mouse limb systems to decipher how spatiotemporal changes of V_mem and subsequent VGCC’s activation regulate organogenesis. We find that the mesenchymal cells of the early limb bud are initially hyperpolarized. Strikingly, however, cells in the core of the limb bud became depolarized during chondrogenic differentiation. This depolarization stimulates Ca²⁺ transients controlled by L-type VGCCs. We show that intracellular Ca²⁺ signals are sufficient to regulate chondrogenesis of cultured limb cells in vitro. Moreover, the voltage-gated calcium channel Ca_V1.2 is essential for proper chondrocyte development in the developing limb in vivo, at least in part, mediated by the Ca²⁺ signaling-dependent transcription factor, NFATc1. Taken together, our results establish a critical role for V_mem in controlling differentiation and patterning during embryonic development.

Results

Limb Mesenchyme Is Depolarized during Chondrogenic Differentiation.

The developing limb bud is 1 of the most intensely studied and best understood models for vertebrate pattern formation and morphogenesis. To see whether significant changes in V_mem take place during limb development, we stained slice cultures of mouse limb buds at embryonic day (E)10.5 (when the early limb bud mesenchyme is morphologically uniform and undifferentiated), E11.5 (when differentiation has started in the proximal limb), and E12.5 (when the distal limb is undergoing differentiation), using the potential sensitive fluorescent dye DiBAC₄(3) (Fig. 1 A and B). DiBAC is a charged molecule and hence is excluded from hyperpolarized cells, while it is taken up and concentrated in depolarized cells, allowing its use as a fluorescent voltage reporter in vivo (24). DiBAC staining was not observed in limb mesenchyme at E10.5, indicating that early limb bud progenitors have polarized V_mem. The polarized state of the early limb mesenchyme was confirmed by staining slice cultures with the low voltage-sensitive dye CC2-DMPE (SI Appendix, Fig. S1A). Moreover, DiBAC staining of the early limb bud mesenchyme could be induced by treating slice cultures with 40 mM potassium (K)-gluconate solution, which forces their depolarization (25) (SI Appendix, Fig. S1B). In contrast, at E11.5 and E12.5, areas of DiBAC staining became evident in the core of the limb buds in domains that correlated with regions of chondrogenesis, identified by Col2 staining of parallel slice cultures (Fig. 1C). Similar results were obtained with developing chick limb buds of an equivalent stage (SI Appendix, Fig. S1C).

Fig. 1.

Depolarization during chondrogenic differentiation of limb mesenchyme cells. (A and B) Transverse (A) and sagittal sections (B) of mouse hindlimbs at E10.5, E11.5, and E12.5. DiBAC₄(3) (DiBAC) signals are indicated by arrows. (C) Sagittal sections of mouse hindlimbs labeled with immunostaining against Col2a1. (D and E) Limb mesenchyme from Col2a1-tdTomato reporter mice was micromass-cultured in the presence of 4-OHT for 2 d, and stained with DiBAC. (Scale bars: 200 μm in A–C, 1 mm in D, and 100 μm in E.)

To verify that the depolarization we observed in later-stage limb buds was associated with chondrogenic differentiation, we turned to “micromass culture,” a well-established in vitro system for studying chondrogenesis of limb morphogenesis. When early limb bud cells are dissociated and placed into culture at high density, they spontaneously undergo differentiation, most prominently into chondrocytes (26). Micromass cultures were established from transgenic mice carrying a Col2a1-tdTomato reporter that produces a red fluorescent signal in chondrocytes, the cell type that expresses the endogenous Col2al gene. Although the micromass cultures are confluent, with no space between cells, focal tdTomato staining becomes evident after 2 d of culture, representing groups of cells undergoing chondrogenesis (Fig. 1 D and E). DiBAC produces a similar pattern of positively staining puncta, and merged images demonstrate that the DiBAC and tdTomato signals are indeed colocalized.

In principle, the depolarization we observe in the developing limb mesenchyme could play a key role in regulating chondrogenesis, or alternatively, it could be a consequence of chondrogenic differentiation. To differentiate between these possibilities, we turned to the signals potentially upstream of the change we observed in V_mem. In many cells where changes in membrane potential are observed, it is the opening of Na⁺ channels that is responsible for membrane depolarization. Moreover, at least 2 Na⁺ channels are known to be expressed in the limb membrane, Na_V1.4 and the epithelial Na⁺ channel (ENaC) (27, 28). As tetrodotoxin, a specific inhibitor of Na_V channels, does not have an effect on cartilage formation in micromass-cultured chick cells (28), we focused on the potential role of ENaC. When we treated chick micromass-cultured cells with Benzamil (a selective ENaC blocker), we observed a noted decrease in DiBAC staining after 2 d in culture, indicating that ENaC plays a significant role in the depolarization of chondrogenic cells in the limb (SI Appendix, Fig. S2A). Moreover, when chondrogenesis was assessed in these cultures via Alcian blue staining (a dye that binds to the acidic proteoglycan-rich cartridge matrix), we saw a dose-dependent decrease in chondrogenic differentiation (SI Appendix, Fig. S2 B and C), indicating that membrane depolarization acts upstream of chondrogenic differentiation in the developing limb bud.

Depolarization of Limb Mesenchyme Increases Ca²⁺ Transients in L-Type Voltage-Gated Ca²⁺ Channel-Dependent Manner.

To determine the mechanism by which the V_mem regulates chondrogenesis, we next investigated the immediate consequences of changes in membrane potential in the limb bud mesenchyme. One of the main ways cells respond to membrane potential is through changes in activity of voltage-gated channels, including the VGCCs. Thus, we asked if the depolarization would cause an increase in Ca²⁺ influx in differentiating micromass cultures of limb mesenchyme taking advantage of GCaMP, a genetically encoded Ca²⁺ sensor (29, 30). An RCAS-viral vector transducing GCaMP6s-2A-mCherry was transfected into early chick limb buds, and intracellular Ca²⁺ dynamics was monitored at a single-cell level (Fig. 2 and Movies S1–S6). GCaMP transients, indicative of Ca²⁺ flux, occurred in 3.6% (average 3.56%) of the transfected (mCherry⁺) cells (Fig. 2 A, B, and E). We then artificially drove depolarization of these cells by treatment with K-gluconate. This resulted in an increase in frequency of the Ca²⁺ transients to 5.27% (P < 0.01 vs. control), implying that VGCCs are activated in response to depolarization (Fig. 2 C–E).

Fig. 2.

Ca²⁺ transients in limb mesenchyme are regulated by voltage-gated L-type Ca²⁺ channels. (A–O) Live-imaging analyses of Ca²⁺ transients visualized with GCaMP6s in micromass-cultured chicken limb bud cells. mCherry was co-overexpressed as a control signal. The micromass-cultured cells were treated with K-gluconate (40 mM; C and D), Nifedipine (250 μM; H and I) or both (M and N). At each time point, images were thresholded to visualize GCaMP transients and newly emerged GCaMP signals were counted as GCaMP transients, as marked by yellow arrowheads in (B′, D′, G′, I′, L′, and N′). (E, J, and O) Temporal dynamics of GCaMP transients (n = 3 each). Time-lapse analyses started 5 h after plating. GCaMP flashes were counted every minute for 60 min. (Scale bars: 100 μm in A, F, and K.)

There are a number of classes of VGCCs mediating the entry of Ca²⁺ ions upon depolarization. Prominent among these are the L-type VGCCs. To test whether L-type VGCCs were responsible for the observed elevation of Ca²⁺ signals in depolarized limb mesenchyme, the GCaMP-expressing cells were cultured with Nifedipine, a drug that specifically blocks the function of L-type VGCCs (Fig. 2 F–J). Before Nifedipine administration, about 3.9% of the cells exhibited Ca²⁺ transients (Fig. 2 F, G, and J). In the presence of Nifedipine, Ca²⁺ fluctuation was significantly attenuated, with only 1.23% of the cells displaying transients (Fig. 2 H–J) (P < 0.01). Moreover, even when depolarization was artificially driven in the cultures by exposure to K-gluconate, Nifedipine prevented the cells from elevating their rate of Ca²⁺ transients (Fig. 2 K–O) (3.61% before treatment, 1.2% in treated groups, P < 0.01).

Interestingly, the effect of Nifedipine is only observed when cultures are treated during the first day of micromass culture. When similar cultures were compared following treatment with or without Nifedipine 2 d later, we observed no difference in percent of Ca²⁺ transients (SI Appendix, Fig. S3) (2.21% before treatment, 2.1% in treated groups, P = 0.056) (Movies S7 and S8). At day 2 of micromass cultures, we observed elongated-shaped myoblast-like cells, Ca²⁺ transients of which were completely suppressed by Nifedipine, and we disregarded these cells since the cells were not derived from limb mesenchyme. This unchanged frequency of Ca²⁺ transients in differentiating limb mesenchyme suggests that the role of L-type VGCCs in controlling Ca²⁺ flux is limited to the undifferentiated limb mesenchymal cells, and that they do not play such a role once chondrogenic differentiation has commenced.

L-Type Ca²⁺ Channels Regulate Cartilage Formation In Vitro.

Our results indicate that V_mem effects both Ca²⁺ flux (via L-type VGCCs) and chondrogenesis in limb mesenchyme. To see whether the induced Ca²⁺ transients are themselves responsible for the regulation of chondrogenesis, micromass-cultured chick limb cells were treated with either Nifedipine to block Ca²⁺ influx through L-type VGCCs, or a Ca²⁺ ionophore, A23187 (also known as calcimycin), to increase the level of Ca²⁺ entering the cells. Nifedipine treatment down-regulated chondrogenesis, as assessed by Alcian blue staining, whereas A23187 increased Alcian blue staining, each in a concentration-dependent manner (Fig. 3 A and B). To verify this finding, qPCR was performed for markers of early- (Sox9), mature- (Col2a1), and late-stage (Indian hedgehog, Ihh) chondrocytes. Consistent with the Alcian blue staining, all of the markers of chondrocytes were strongly repressed by exposure to Nifedipine, while the level of Sox9 was significantly increased following exposure to A23187 (Fig. 3C).

Fig. 3.

Ca²⁺ influx via L-type Ca²⁺ channels regulates chondrogenesis of chicken limb bud mesenchyme in vitro. (A) HH20 chicken limb bud cells were micromass-cultured in the presence of Nifedipine (50 μM or 250 μM) and A23187 (50 nM or 100 nM) for 5 d, and stained with Alcian blue. (B) Quantification of Alcian blue intensity (n = 8 for each group). (C) Sox9, Col2a1, and Ihh mRNA expression levels were measured by qPCR and normalized to β-actin expression (n = 8 each). (D and E) Micromass-cultured chicken cells were treated with DMSO, Nifedipine (250 μM) from days 0 to 2, days 1 to 3, or days 2 to 4, for 5 d, and stained with Alcian blue (n = 5 each). (F) Relative expression levels of Sox9, Col2a1, and Ihh were quantified by qPCR (n = 5 each). Error bars represent SEM. *P < 0.05, **P < 0.01; 1-way ANOVA multicomparison with DMSO as control. (Scale bars: 2 mm in A and D.)

The fact that Nifedipine blocks expression of the early cartilage marker Sox9 suggests that the L-type VGCCs act to promote the earliest stages of chondrogenesis. To further test this, we exposed micromass cultures to Nifedipine for various time windows. When the cells were treated with the drug from day 0 through day 2 of culture (Fig. 3 D–F), there was a near total suppression of chondrogenic differentiation, as assessed by Alcian blue staining and Sox9, Col2a1, and Ihh expression. In contrast, treatment from day 1 to day 3, or from day 2 to day 4, showed a much weaker effect, with a mild decrease in Alcian blue staining and Col2al expression, and Sox9 levels comparable to those in control cultures. Nifedipine treatment appeared to affect only chondrocyte differentiation rather than cell proliferation because the size of the micromass cultures was similar between the cultures (SI Appendix, Fig. S4). Taken together, these results strongly implicate a role for L-type VGCCs in modulating Ca²⁺ levels to promote the first steps in chondrogenesis of limb mesenchyme.

Ca_V1.2, an L-Type Voltage-Gated Ca²⁺ Channel, Is Essential for Skeletal Development in the Mouse Limb.

Based on these results, we next interrogated the role of L-type VGCCs during limb development. Ca_V1.2 (encoded by the Cacnalc gene) is known to be broadly expressed in developing mouse limbs at E11.5 (31). We confirmed broad Ca_V1.2 protein expression in developing mouse hindlimbs by immunohistochemistry (SI Appendix, Fig. S5). This expression pattern made Ca_V1.2 a strong candidate for mediating the L-type VGCC-dependent Ca²⁺ influx that we observed during chondrogenesis. The Cacnalc gene has previously been deleted in mice, but the mutants exhibit an early embryonic lethality (32), precluding an analysis of limb-stage chondrogenesis. We therefore derived limb-specific Cacnalc conditional mutant mice by crossing the limb-specific Prx1-CreER lines with mice harboring a floxed allele. Control mice carrying either the Cacnalc^fl/fl allele, or the Prx1-CreER alone, did not show limb malformations at E18.5 following tamoxifen treatment (Fig. 4A and SI Appendix, Fig. S6). In contrast mice carrying both the Prx1-CreER and the Cacna1c^fl/fl allele (hereafter referred to as “mutant” mice) had notable limb abnormalities (Fig. 4B). (Note: For this analysis we focused on transgenic mice that were homozygous for the Prx1-CreER allele, as they achieve a faster deletion of the floxed allele than Prx1-Cre heterozygotes.) As Ca_V1.2 proteins were depleted in the mutant mouse limbs upon tamoxifen injections (SI Appendix, Fig. S7), the phenotypes were likely to be attributable to defects in Ca²⁺ signaling via Ca_V1.2. Most prominently, Cacnalc mutant mice displayed shortened limbs (Fig. 4). For example, the mutant hindlimbs were 76.9% of the length of controls (Fig. 4C). The mutant hindlimbs were also missing digits. Forelimbs showed similar kinds of defects; however, in general, the phenotypes were relatively more severe in hindlimbs than forelimbs (Fig. 4 C and D), correlating with a relatively higher mRNA expression level of CreER in hindlimb buds of E10.5 embryos than in forelimbs (SI Appendix, Fig. S8). Mice displayed similar kinds of phenotypes, albeit to a lesser extent, when tamoxifen was administered at E10.5 and E11.5 rather than at E9.5 and E10.5 (SI Appendix, Fig. S9). These data demonstrated a role for Ca_V1.2 in limb development.

Fig. 4.

Ca_V1.2 is required for development of skeletal elements in mouse limbs. (A and B) Alcian blue/Alizarin red skeletal preparations of E18.5 Cacna1c^fl/fl (control, A), and Prx1-CreER^+/+; Cacna1c^fl/fl mouse embryos (mutant, B). Tamoxifen injections were performed at E9.5 and E10.5. Forelimbs (FL) and hindlimbs (HL) are separately displayed. (Scale bars: 5 mm in A.) (C and D) Quantification for relative length of the limbs (C) and the number of digits (D) of controls (n = 20 limbs) and mutants (n = 26). Error bars represent SEM. *P < 0.05, **P < 0.01.

Lack of Ca_V1.2 Activity Affects Cell Death but Not Proliferation in the Early Limb Bud.

Given that the mutant limbs were shorter than the wild-type, we scrutinized whether cell proliferation and programmed cell death were affected in the differentiating limb mesenchyme of the mutants, making use of a mitotic marker, phosphorylated histone H3 (pH3), and an apoptotic marker, cleaved caspase3 (Cas3). Quantification of pH3⁺ cells in the forelimb and hindlimb buds at E11.5 showed no significant differences between control and mutant limbs (Fig. 5 A–D and I). In contrast, Cas3 analyses revealed that excess cell death occurred in the limb buds of the mutant (Fig. 5 E–H and J). While the observed cell death might contribute to the overall decrease in the size of the mutant limbs, it is notable that the apoptotic cells are distributed across the entire limb bud and, if anything, are present to a lesser extent in the core of the limb bud where chondrogenesis takes place. Thus, decreased survival would not be a probable explanation for the observed defects in the skeletal elements.

Fig. 5.

Increased cell death in Ca_V1.2 mutant limb buds. (A–H) Sagittal sections of fore- or hindlimb buds from E11.5 Control (A, C, E, and G) and mutant mouse embryos (B, D, F, and H) were stained with antibodies for phospho-histone H3 (pH3; A–D) or cleaved-caspase3 (Cas3; E–H). (Scale bars: 200 μm in A and E.) (I and J) Quantification of the number of pH3⁺ or Cas3⁺ cells in the limb buds (n = 12 each for pH3-counting, n = 18 each for Cas3-counting). Error bars represent SEM. **P < 0.01.

Ca_V1.2 Controls Chondrocyte Differentiation by Regulating Sox9 Expression.

To better understand the skeletal abnormalities in the Ca_V1.2-deficient limbs, we examined the pattern of chondrogenesis during cartilage differentiation by staining with the marker Col2al at E13.5 (Fig. 6 A and B). Not only are less chondrogenic cells observed, the spacing of the skeletal elements (for example, between the digits of the mutant limbs) was also misregulated. To see if the earliest steps of chondrocyte differentiation were disturbed in the Ca_V1.2-deficient mice in vivo, as seen in the Nifedipine-treated micromass cultures, we examined Sox9 expression in the mutant mice at both E11.5 and E13.5 (Fig. 6 A–D). Indeed, immunostaining for Sox9 protein demonstrated reduction of the proteins in the hindlimb buds of mutant embryos at both stages (Fig. 6 A–D). Consistent with this, qPCR analyses revealed significant decrease of Sox9 as well as the late-stage chondrogenic marker Aggrecan1 (Agc1) (Fig. 6E). This effect on chondrogenic differentiation appears to be specific, as no differences (examined by qPCR) were seen between control and Ca_V1.2 mutant limbs in the expression of other key genes known to regulate limb pattern, including genes acting through the Wnt pathway (Axin2), the Fgf pathway (Fgf8, Fgf10, Dusp6), the hedgehog pathway (Shh, Ptch1), genes involved in regulating key domains of cell death (Msx2), or genes establishing a proximodistal pattern (Hoxd13, Meis2) (SI Appendix, Fig. S10). The 1 exception to this was when we evaluated proximodistal markers by in situ hybridization. Proximodistal patterning in limbs was examined using Meis1 and Hoxa13 as proximal and distal markers, respectively (SI Appendix, Fig. S11). Hoxa13 was distally localized in an apparently normal expression domain at both E11.5 and E13.5 in mutant embryos (SI Appendix, Fig. S11 B and D). Meis1 was similarly detected at the correct location in the proximal third of the limb bud at E11.5. However, unlike Hoxa13, which continues to be expressed around the differentiating skeletal elements at E13.5, Meis1 is normally down-regulated by E13.5 as chondrogenic differentiation proceeds. However, in the mutant limbs, Meis1 expression is aberrantly sustained at E13.5 (SI Appendix, Fig. S11 A and C). Maintenance of Meis1 expression is likely secondary to the defective differentiation seen in the mutant limbs.

Fig. 6.

Requirement of Ca_V1.2 for chondrocyte differentiation in mouse limbs. (A and B) Sagittal sections of E13.5 control (A) and mutant hindlimbs (B) were stained with antibodies for Sox9 (red) and Col2 (cyan). Cartilage elements were marked by anti-Col2 antibodies as indicated by arrowheads. (C and D) Sagittal views of hindlimbs of E11.5 and E13.5 mouse embryos. Chondrocytes were visualized by immunostaining against Sox9. (E) Sox9 and Agc1 expression levels were measured by qPCR and normalized to β-actin expression (n = 7 for each group). (F and G) Immunostaining of phosphorylated Smad1/5 (pSmad) in E11.5 and E13.5 hindlimbs from control (F) and mutant embryos (G). Error bars represent SEM. **P < 0.01. (Scale bars: 200 μm in A–C and F.)

These data indicate a specific role of Ca_V1.2 in regulating the earliest steps of chondrogenic differentiation. The Bone morphogenetic protein (Bmp) pathway is also known to play a critical role in the initiation of chondrogenesis in the developing limb (33–35). To ask whether the reduction of Sox9 is attributable to failure of Bmp activation, we performed staining of phospho-Smad (pSmad), a readout of Bmp signaling. pSmad staining was seen in the peripheral regions of E11.5 in equivalent domains in the mutant and control limb buds (Fig. 6 F and G). Similarly, at E13.5, signals were observed interdigitally in both the controls and mutants (Fig. 6 F and G). These results suggest that Ca_V1.2 regulates chondrogenesis in a Bmp-independent manner.

NFATc1 Mediates Ca²⁺/Ca_V1.2 Signaling during Limb Mesenchyme Chondrogenesis.

Since the impact of Ca_V1.2 on limb bud chondrogenesis appears to be independent of Bmp signaling, we considered the possibility that its effect could be cell-autonomous. One important set of intracellular effectors of Ca²⁺ signaling is the NFAT (nuclear factor of activated T-cells) family of transcription factors. There are 5 members of the NFAT family (NFATc1 to -5), and subcellular localization of NFATs, except NFATc5, is known to be regulated by Ca²⁺ signaling. In the presence of Ca²⁺, calmodulin, a well-established calcium sensor, initiates a cascade leading to dephosphorylation of the NFAT proteins, thereby activating them and resulting in their nuclear translocation. We observed that 1 of these members, NFATc1, is broadly expressed in the early mouse limb bud and is very specifically expressed in chondrogenic regions, such as in the developing digits (Fig. 7A). To verify that NFATc1 expression coincides with mesenchymal chondrogenesis, we looked at expression of NFATc1 proteins in cultured limb mesenchyme harvested from either control or Ca_V1.2 mutant embryos. Nuclear localized NFATc1 (the active form of NFATc1) and Sox9 protein colocalize in the chondrogenic micromass cells in control cultures. Both proteins were seen to a lesser degree in the mutant cells (Fig. 7 B–D). Moreover, the fluorescent intensity of NFATc1 and Sox9 signals were positively correlated with one another (Fig. 7E).

Fig. 7.

NFATc1 is reduced in the Ca_V1.2 cKO cells and is sufficient to up-regulate chondrocyte differentiation in vivo and in vitro. (A) mRNA expression patterns of NFATc1 in E10.5, E11.5, E12.5, and E13.5 mouse limbs. Arrows mark NFATc1 expression at both fore- and hindlimb buds. (B and C) Immunostaining for NFATc1 and Sox9 in cultured mouse limb cells harvested from E11.5 control (B) and mutant limb buds (C). (D) Quantification of the number of NFATc1⁺ or Sox9⁺ cells in the cultured cells (n = 3,436 cells for control, n = 1,585 cells for mutant group). (E) Measurement of fluorescent intensity for NFATc1 and Sox9 signals (n = 270 cells from control group). (F and G) Plasmids carrying ZsGreen1 (control, F) or NFATc1^nuc-IRES-ZsGreen1 (G) were electroporated into chicken forelimbs. Misexpression of NFATc1^nuc resulted in inducing a Sox9⁺ ectopic aggregate (an arrow in G; n = 5 of 15), which was not seen in the controls (n = 0 of 15). (H and I) mCherry or NFATc1^nuc was overexpressed in micromass-cultured cells obtained from E11.5 control and mutant embryos by using lentiviruses (see SI Appendix, Fig. S10 for experimental procedures). (I) Quantification of Alcian blue staining (n = 8 for each group). Error bars represent SEM. **P < 0.01. (Scale bars: 500 μm in A, 100 μm in B, 200 μm in F, and 2 mm in H.)

To test whether NFATc1 is capable of activating the chondrogenic program in vivo, we introduced a plasmid that bicistronically expresses NFATc1^nuc (a constitutively activated, Ca²⁺-independent form of the protein) (36) and ZsGreen1, along with a transposase-expressing vector, into chicken limb progenitors by electroporation (Fig. 7 F and G). The NFATc1^nuc-electroporated cells form Sox9⁺ aggregates ectopically (Fig. 7G), showing that NFATc1 does indeed have prochondrogenic activity.

We reasoned if Ca_V1.2 regulates chondrogenesis through modulation of the cartilage-promoting activity of NFATc1, then introduction of an activated form of NFATc1 should be able to restore chondrogenesis in the context of the Ca_V1.2 mutant. Harvested limb mesenchyme was placed into micromass cultures and infected with a lentivirus carrying either mCherry (control) or NFATc1^nuc (SI Appendix, Fig. S12). When NFATc1^nuc was overexpressed, the reduced chondrocyte formation was partially restored, as assessed by Alcian blue intensity (Fig. 7 H and I).

Discussion

We have identified the L-type voltage-gated channel Ca_V1.2 as a key player in chondrogenesis, and more broadly implicated the regulation of membrane potential as a crucial parameter in controlling cell differentiation decisions during limb development. We started by interrogating the V_mem dynamics during limb development in both chick and mouse embryos. DiBAC staining particularly highlighted differentiating chondrocytes as being strongly depolarized. Indeed, depolarization appears to be crucial for chondrocyte differentiation as the pharmacological inhibition of ENaC, a channel known to modulate membrane depolarization in other contexts (9, 37), blocked cartilage formation in micromass cultures, and Ca²⁺ transients activated by depolarization are themselves required for cartilage formation. It is, however, unclear whether ENaC itself is required for limb cartilage formation in vivo.

In contrast to the depolarization observed in differentiating chondrocytes, undifferentiated limb mesenchyme is relatively hyperpolarized, excluding DiBAC staining, and taking up the low-voltage indicator CC2-DMPE. In zebrafish fins, gain-of-function mutations in kcnk5b, a gene that encodes a 2-pore domain K⁺ channel, lead to membrane hyperpolarization and significantly promote cell proliferation of fin mesenchyme (8). This raises the possibility that the hyperpolarization we observe in the early chick and mouse limb-bud mesenchyme could reflect a role in modulating the proliferation within the early limb bud.

Although Ca_V1.2 was previously known to be expressed in limb chondrocytes, its role in promoting chondrogenesis had not been elucidated. A missense mutation of Cacna1c (G406R), which is constitutively activated, and hence inappropriately increases intracellular Ca²⁺ concentration, causes limb malformation in Timothy syndrome patients (23), and limb-specific overexpression of the G406R mutant in mice resulted in increase of bone mass in the limbs (38). These animals did not, however, display aberrant chondrogenesis, per se, indicating that Ca_V1.2 activity is not sufficient on its own to drive these processes. A requirement of VGCCs in digit formation was implied in a second study, in which Nifedipine was administered to pregnant rabbits and the fetuses showed reduction of digits and abnormal structure of phalanges (39). It was, however, concluded that these phalangeal phenotypes were secondarily induced by decreased blood flow caused by excessive hypotension, rather than direct inhibitory actions of the blocker on VGCCs (40, 41). In contrast, our analysis of limb-specific Ca_V1.2 cKO mice, supported by in vitro culture studies, strongly indicate that Ca_V1.2 plays an essential role in regulating chondrogenesis in the limb buds.

Ca_V1.2 is important, not only for chondrocyte differentiation, but also for protecting limb mesenchyme from inappropriate apoptosis. Bmp signaling has many roles in vertebrate limb development, including initiating chondrogenic differentiation and inducing programmed cell death to sculpt the autopod (35, 42). Even though expression levels of Sox9 were decreased in the mutant limb buds, pSmad proteins were seen at similar levels, compared to the controls, suggesting that Bmp signals were activated normally in the mutants. These results suggest that Ca_V1.2 regulates Sox9 expression independent of Bmp activity.

It has previously been shown that increases in intracellular Ca²⁺ concentration are involved in promoting chondrocyte proliferation (38, 43). However, this would appear to be independent of Ca_V1.2, as cKO limbs display no significant change in number of mitotic cells.

We show that Ca_V1.2 signaling is important for early phases of limb chondrocyte differentiation. However, Ca_V1.2 is not absolutely required for the initiation of chondrogenesis, as Sox9 is detected in mutant E11.5 hindlimb buds. Moreover, the mutant mice form skeletal elements, albeit greatly reduced in size. Ca_V1.2 is nonetheless essential for proper differentiation at the earliest stages of skeletogenesis. Ca_V1.2 does not appear to be as crucial at later stages, however, as delayed tamoxifen injection causes only modest defects in cartilage formation in mouse limbs. Similarly, late Nifedipine treatment in micromass cultures does not affect Ca²⁺ transients in the cultured cells (except for myogenic cells), and only modestly affects chondrogenesis. Other Ca²⁺ channels may be responsible for Ca²⁺ influx at later stages of chondrocyte differentiation. One candidate channel is Ca_V3.2 because it is also expressed in limb chondrocytes and micromass-cultured limb mesenchyme from the Ca_V3.2^−/− mice exhibited attenuated cartilage formation (22). Another ion channel that potentially regulates late chondrogenesis is TRPM7 (transient receptor potential cation channel subfamily M member 7), because TRPM7-mediated Ca²⁺ fluctuations in growth plate chondrocytes were found to be critical for chondrocyte maturation (44).

We further found that Ca²⁺ influx via Ca_V1.2 appeared to activate nuclear translocalization of NFATc1, which in turn up-regulated expression of Sox9, a master regulator for chondrocyte differentiation in the limbs. In addition to chondrogenic differentiation, Sox9 is known to regulate proliferation and survival of differentiating limb mesenchyme (45); thus, NFATc1 and related family members, such as NFATc2 that could act synergistically with NFATc1, may be critical for multiple aspects of limb development. Consistent with our results, Lin et al. (22) previously demonstrated that Ca²⁺ entry via Ca_V3.2 regulated Sox9 expression in tracheal cartilage, which was mediated by an NFATc1 binding site within the Sox9 promoter. Also, in the pancreas, an NFATc1/c-Jun complex up-regulates Sox9 expression, which in turn initiates pancreatic adenocarcinoma by binding to the promoter regions of Sox9, albeit in a different context from chondrocytes (46). This evidence suggests that the mechanistic regulation of Sox9 expression by NFAT may be widely conserved in biological events, such as development and cancer.

Materials and Methods

Experimental Animals.

White leghorn eggs were obtained from Charles River. Chicken embryos were staged according to the Hamburger and Hamilton stages (HH) (47). Mouse colonies were maintained in the vivarium at the New Research Building of Harvard Medical School. Cacna1c^tm3hfm/J mice (also known as Ca_V1.2-floxed) were purchased from the Jackson Laboratory (Stock no. 024714). Prx1-CreER-IRES-GFP (Prx1-CreER) mice were provided by Shunichi Murakami, Case Western Reserve University, Cleveland, OH (48). Ai9 (Gt[ROSA]26Sor^{tm9[CAG-tdTomato]Hze}) and Col2a1-CreER (49) mouse strains were obtained from the Jackson Laboratory (Stock no. 007909 and 006774, respectively). Ai9 mice were crossed with Col2a1-CreER mice to obtain Col2a1-tdTomato reporter embryos. All animal experiments were performed under the guidelines of the Harvard Medical School Institutional Animal Care and Use Committee.

Micromass Culture and Alcian Blue Staining.

Limb buds from HH20 chicken or E11.5 mouse embryos were used to generate micromass cultures. Briefly, after the surface ectoderm was removed in 0.25% Trypsin-EDTA solution (Sigma-Aldrich), limb mesenchyme was dissociated gently by pipetting and pelleted by centrifugation. The cells were redissociated, and 2 × 10⁵ cells per 20 μL of 10% FBS/DMEM were dropped into each well of a 24-well plate. After being attached, the cells were cultured in 10% FBS/DMEM for 5 d. Alcian blue staining was performed as previously described (50). To extract Alcian blue dye, the stained cultures was incubated with 250 μL of 6 M guanidine hydrochloride for 24 h at 37 °C, and the OD of extracted Alcian blue was measured at 620 nm. Benzamil, Nifedipine, and 4-Bromo A23187 were obtained from Sigma-Aldrich.

Imaging with DiBAC₄(3) and CC2-DMPE.

DiBAC₄(3) and CC2-DMPE staining was performed as previously described with slight modifications (24). DiBAC was obtained from Thermo Fisher Scientific, and was prepared in a loading buffer (160 mM NaCl, 4.5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM Glucose, 10 mM Hepes, 1% FBS, pH 7.4). Primary cultured limb mesenchyme was soaked at a concentration of 0.5 μg/mL for 30 min at 37.5 °C, and slice-cultured limbs were incubated with 1 μg/mL for 1 h with gently rocking at 37.5 °C, to allow for sufficient dye permeation before imaging. Sliced limb buds were embedded with 1% low melting agarose (Invitrogen) in a 35-mm glass-bottom dish (MatTek Corporation) after the incubation for the confocal imaging (Carl Zwiss LSM710). To induce membrane depolarization, sliced chicken limb buds were soaked in the loading buffer containing 40 mM K-gluconate for 10 min prior to the embedding. CC2-DMPE (Invitrogen) was used at a concentration of 5 μM that was mixed in DiBAC solution.

Skeletal Preparation.

Skeletal preparations were carried out as described previously (51). The E18.5 embryos were skinned and eviscerated, and fixed in 95% ethanol overnight and then in acetone overnight at room temperature. Subsequently, embryos were stained in Alcian blue 8GX (Sigma-Aldrich) solution for 24 h at room temperature, and incubated in 1% KOH for several hours. Then, embryos were stained with Alizarin red S (Sigma-Aldrich) overnight, and permeabilized through a graded series of 1.5% KOH, 1% KOH, 1% KOH/20% glycerol, 0.6% KOH/40% glycerol, 0.4% KOH/60% glycerol, and 80% glycerol for imaging.

Tamoxifen and 4-Hydroxy Tamoxifen (4-OHT) Treatment.

Tamoxifen was dissolved in corn oil (Sigma-Aldrich), and 1 mg/d of tamoxifen was given to E9.5 to E11.5 pregnant dams by intraperitoneal injections; 5 μM of 4-OHT (Calbiochem) was used for in vitro micromass culture experiments.

In Ovo Electroporation and Expression Vectors.

For the in ovo electroporation, eggs were incubated for about 54 h at 38 °C. DNA solution was prepared at 4 μg/μL, and injected into the coelomic cavity of HH14 embryos. Three electric pulses of 50 V, 2 ms, were given, followed by 7 pulses of 5 V, 10 ms, with 10-ms interval between pulses (Super Electroporator NEPA21-type II, NEPA GENE). pT2A-CAGGS-IRES2-ZsGreen1 was described previously (52, 53). For pT2A-CAGGS-NFATc1^nuc-IRES2-ZsGreen1, coding sequences (CDS) of NFATc1^nuc were amplified from pCR4-huNFATc1^nuc (Addgene plasmid # 23988) (36) by PCR using Q5 DNA polymerase (New England Biolabs), and subcloned into pT2A-CAGGS-IRES2-ZsGreen1 by Gibson Assembly Cloning Kit (New England Biolabs). The plasmid was sequence verified prior to the electroporation. pCAGGS-T2TP was shared by Yoshiko Takahashi, Kyoto University, Kyoto, Japan. For preparation of GCaMP-labeled cells, RCAS-GCaMP6s-2A-mCherry plasmid (30) was electroporated into HH14 embryos.

Time-Lapse Imaging.

For in vitro time-lapse imaging, GCaMP6s-2A-mCherry–electroporated limb mesenchymal cells were micromass-cultured on a 35-mm glass-bottom dish. The cells were cultured in 10% FBS/DMEM in an incubation chamber attached to the LSM10 confocal microscope, and images were taken every minute. For artificial depolarization, 40 mM potassium-gluconate was used, and 250 μM of Nifedipine was added to medium to inhibit Ca²⁺ transients.

Immunostaining.

For immunological staining on histological sections, the following antibodies were used as described previously (54): Mouse anti-Collagen 2A1 (1:100; DSHB, II-6B3), rabbit anti-pH3 (1:1,000; Millipore, 06-570), rabbit anticleaved caspase-3 (1:300; Cell Signaling, #9661), rabbit anti-Ca_v1.2 (1:200; Alomone Laboratories, ACC-003), rabbit anti-Sox9 (1:500; Millipore, AB5535), rabbit anti-pSmad1/5 (1:300, Cell Signaling, #9516), and mouse anti-NFATc1 (1:25; Santa Cruz, sc-7294). For staining of Col2A1 and Ca_V1.2, an antigen retrieval using Target Retrieval Solution (DAKO, S2369) was performed in advance of blocking. To detect pSmad, TSA Plus Cy3 system (PerkinElmer) was used as previously described (54).

qPCR.

RNA was extracted using the RNeasy Mini Kit (Qiagen) and reverse-transcribed by SuperScript III First-Strand Synthesis System (Thermo Fisher). PCR was performed by using Brilliant III Ultra-Fast SYBR Green QPCR kit (Agilent) and CFX Touch Real-Time PCR Detection System (Bio-Rad). Relative expression levels were calculated by the ΔΔCq method. Genotypes for Prx1-Cre^ER-GFP allele was determined according to the Cq values obtained by qPCR using the SYBR Green QPCR kit. Sequences (5′-3′) of primers for qPCR are as follows:

• cßactin-F, TCTGACTGACCGCGTTACTC;

• cßactin-R, CCATCACACCCTGATGTCTG;

• cSox9-F, AGGAAGCTGGCTGACCAGTA;

• cSox9-R, CGTTCTTCACCGACTTCCTC (55);

• cCol2a1-F, GGACAGCGTGGTGAGAGAGGCTT;

• cCol2a1-R, TCCCTGCCTGGGAGACCGTCA (56);

• cIhh-F, CATCCACTGCTCCGTCAAGT;

• cIhh-R, ATGACGTGGAAGGCAGTGAG;

• mßactin-F, GCTGTATTCCCCTCCATCGTG;

• mßactin-R, CACGGTTGGCCTTAGGGTTCAG;

• mSox9-F, TAAGTTCCCCGTGTGCATCC;

• mSox9-R, TTGCCCAGAGTCTTGCTGAG;

• mAgc1-F, GGTCACTGTTACCGCCACTT;

• mAgc1-R, CCCCTTCGATAGTCCTGTCA;

• mDusp6-F, CGGAAATGGCGATCTGCAAG;

• mDusp6-R, GACGACTCGTACAGCTCCTG;

• mMsx2-F, CAAGTGAAGGGGGAGGTGTA;

• mMsx2-R, CAGGGACCTGACATGGAGTT;

• mAxin2-F, GAGGAGATCGAGGCAGAAGC;

• mAxin2-R, CACCTCTGCTGCCACAAAAC;

• mFgf8-F, AGAGAGACAGTGTCCTGCCT;

• mFgf8-R, TCTTCTGCCATGGCGTTGAT;

• mFgf10-F, AGCGGGACCAAGAATGAAGACTGT;

• mFgf10-R, CCTGCCATTGTGCTGCCAGTTAAA (57);

• mMeis2-F, GTTCCGTCATTTCGTTCTCC;

• mMeis2-R, CTTTCTCGCTCGCTCTCACT;

• mShh-F, CCGAACGATTTAAGGAACTCACCC;

• mShh-R, TGGTTCATCACAGAGATGGCCAAG (58);

• mPtch1-F, GACCGGCCTTGCCTCAACCC;

• mPtch1-R, CAGGGCGTGAGCGCTGACAA;

• mHoxd13-F, TGTCCACTTTTGGATCCGGG;

• mHoxd13-R, TTCTTCCTTCCCCGTCGGTA;

• Cre-F, GCATTACCGGTCGATGCAAC;

• Cre-R, ACATCTTCAGGTTCTGCGGG.

Probes and In Situ Hybridization.

Mouse cDNA fragments for NFATc1 and Meis1 were isolated by PCR using primers (5′–3′):

• NFATc1-F, ATAACGCGTGCTACAGCTGTTCATTGGGA;

• NFATc1-R, ATAGCTAGCCTTGCACAGGTCCCGGTCAG;

• Meis1-F, ATAACGCGTATGGCGCAAAGGTACGACGA;

• Meis1-R, ATAGCTAGCGATCCTCCTTCTCTATCATC.

The PCR fragments were digested by MluI and NheI (New England Biolabs), and subcloned into pBS-D. Plasmids for generating antisense probes for mHoxa13 were provided by Richard Maas, Brigham and Women’s Hospital, Boston, MA. Whole-mount in situ hybridizations were performed as described in ref. 54. The color reaction was carried out with BM Purple (Sigma-Aldrich).

Lentivirus Production and Infection.

FUW-M2rtTA (59), psPAX2, pMD2.G, and pLV-tetO-mCherry (60) plasmids were obtained from Addgene (#20342, #12260, #12259, and #70273). For FUW-tetO-NFATc1^nuc, PCR products of NFATc1^nuc CDS were introduced into FUW-tetO-MCS (Addgene plasmid #84008) (61) by using a Gibson Assembly Kit (New England Biolabs). VSV-G-coated lentiviral production was performed in 293T cells through cotransfection of lentiviral vectors with psPAX2 and pMD2.G by Polyethylenimine. Virus containing supernatant was collected 48 h after transfection, filtered through 0.45-μm SFCA syringe filters (Corning), and frozen for further use. Infection was performed on 24-well dishes for 24 h in the presence of 8 μg/mL polybrene (Sigma-Aldrich). Next, 2 μg/mL of doxycycline (Sigma-Aldrich) was used to activate M2rtTA.

Statistical Analysis.

P values were obtained using a 2-tailed, unpaired Student’s t test or 1-way ANOVA (GraphPad Prism 7, GraphPad Software).

Data Availability.

All data generated or analyzed during this study are in the main text and SI Appendix, or available from the corresponding author upon request.