Connexin hemichannels drive lactation-induced osteocyte acidification and perilacunar-canalicular remodeling

SUMMARY

The maternal skeleton experiences significant bone loss during lactation, followed by rapid restoration post weaning. Parathyroid-related protein (PTHrP)-induced acidification of the perilacunar matrix by osteocytes is crucial in this process, yet its mechanism remains unclear. Here, we identify Cx43 hemichannels (HCs) as key mediators of osteocyte acidification and perilacunar-canalicular remodeling (PLR). Utilizing transgenic mouse models expressing dominant-negative Cx43 mutants, we show that mice with impaired Cx43 HCs exhibit attenuated lactation-induced responses compared to wild-type and only gap junction-impaired groups, including lacunar enlargement, upregulation of PLR genes, and bone loss with compromised mechanical properties. Furthermore, inhibition of HCs by a Cx43 antibody blunts PTHrP-induced calcium influx and protein kinase A activation, followed by impaired osteocyte acidification. Additionally, impeded HCs suppress bone recovery during the post-lactation period. Our findings highlight the pivotal role of Cx43 HCs in orchestrating dynamic bone changes during lactation and recovery by regulating acidification and remodeling enzyme expression.

In brief

To meet the high calcium demand during lactation, one of the crucial aspects is the remodeling of the bone perilacunar matrix by osteocytes. Here, Hua et al. show the indispensable role of osteocytic Cx43 hemichannels in regulating intracellular acidification and activation of proteases necessary for perilacunar remodeling associated with lactation and post-lactation recovery.

Graphical Abstract

INTRODUCTION

Bone is an important endocrine organ for modulating mineral metabolism and homeostasis. The skeleton serves as a calcium reservoir during lactation. Bone mineral density (BMD) declines by 6%–10% over 6 months in nursing women, while rodents can lose up to 20%–30% of bone mass over 3 weeks of lactation.¹ This substantial mobilization of calcium into milk results from activated bone resorption and osteocyte perilacunar-canalicular remodeling (PLR). It has been proposed that osteocytes are able to remove their perilacunar matrix, a process known as “osteocytic osteolysis.”^2,3 This concept was recently demonstrated in lactating mice, which were under high calcium demand.⁴ Osteocytes, comprising 90%–95% of bone cells, function as orchestrators of bone-forming osteoblasts and bone-resorbing osteoclasts. They are terminally differentiated cells embedded in the bone mineral matrix and form an extensive lacuno-canalicular network through long dendritic processes.^5,6 As a result, osteocytes are exposed to a significantly larger surface area compared to the bone surface area, allowing for metabolite transport and communication.

Parathyroid hormone-related peptide (PTHrP) is a multifaceted protein that regulates the bone remodeling process, mediating both anabolic and catabolic effects.⁷ PTHrP is secreted into the circulation from the lactating mammary gland and induces the acidification and demineralization of osteocyte microenvironment through the parathyroid hormone 1 receptor (PTH1R) during lactation.⁸ PTHrP promotes the expression of several key genes, such as carbonic anhydrase 1/2 (Car1/2), proton pump vacuolar ATPase (V-ATPase), tartrate-resistant acid phosphatase (ACP5/TRAP), cathepsin K, and matrix metalloproteinases (MMPs).⁹ Mice lacking osteocytic PTH1R, cathepsin K, or MMP13 have been reported to be resistant to lactation-induced osteocyte PLR.^4,10,11 Importantly, the rapid bone loss associated with lactation is reversible after weaning, with osteocytes replacing bone matrix in their surrounding lacunar space.^12,13 However, how PTHrP acidifies extracellular matrix and activates PLR-related enzyme is largely unknown.

As the most predominantly expressed subtype of connexin (Cx) in osteocytes, Cx43 forms both gap junctions and hemichannels (HCs), which mediate the passage of small molecules (<1.2 kDa). Cx43 gap junctions enable cell-cell communication between adjacent osteocytes and osteocytes with other cells residing within the bone surface. Meanwhile, Cx43 HCs, which are unpaired gap junction channels, serve as a portal for communication between osteocytes and their extracellular lacuna-canalicular microenvironment.¹⁴ The open probability of HCs is low under normal physiological conditions. However, their opening can be stimulated by various factors, including low extracellular Ca²⁺ concentration, extracellular alkalinization, mechanical stimulation, and inflammatory cytokines.^15,16 Cx43 has been shown to play a crucial role in skeletal development, bone remodeling, bone cell differentiation and survival, and mechanotransduction.^17–19 Deletion of Cx43 from osteoblasts using the 2.3-kb Col1α1 promoter leads to low BMD and compromised bone strength.^20,21 Deletion of Cx43 in osteoblasts/osteocytes using the Bglap2 or 8-kb Dmp1 promoter results in increased osteocytic apoptosis, periosteal bone formation, and endocortical bone resorption.^22,23 Our previous work has demonstrated that impairment of Cx43 HCs in osteocytes critically impacts both bone formation and remodeling.^19,24

In this study, we demonstrated the unique role of Cx43 HCs in mediating the effect of PTHrP on PLR by regulating extracellular acidification of osteocytes. We used two transgenic mouse models that overexpress dominant-negative Cx43 mutants primarily in osteocytes: R76W, which inhibits gap junctions, and α130–136, which inhibits both gap junctions and HCs.²⁴ Additionally, we employed a Cx43 HC inhibitory antibody.^25,26 We investigated the regulation of PLR by Cx43 channels under normal, lactation, and post-lactation conditions. Our results unveil a previously unrecognized role of osteocytic Cx43 HCs in mediating calcium influx and cytosolic acidification, key steps leading to the activation of extracellular enzymes and perilacunar mineral removal and deposition. This highlights the crucial regulatory mechanism in PLR during lactation.

RESULTS

Impairment of Cx43 HC function disrupts the osteocyte perilacunar-canalicular network

Two transgenic mouse models driven by the 10-kb Dmp1 promoter were utilized to study the distinct roles of Cx43 gap junction channels and HCs in osteocytes. Femoral cortical bone sections from female wild-type (WT), R76W, and Δ130–136 mice were prepared. Picrosirius red staining under polarized light microscopy was used to evaluate collagen content and organization. In comparison to WT and R76W mice, Δ130–136 mice exhibited significant collagen disorganization in the femoral midshaft cortical bone (Figure 1A). Ploton silver staining revealed a severe deterioration of the osteocyte lacuno-canalicular network in Δ130–136 mice, with a 50% reduction in canalicular length and lacuno-canalicular area (Figures 1B and 1C). Immunohistochemistry staining also demonstrated reduced expression of MMP13 and cathepsin K in osteocytes of Δ130–136 mice (Figures S1A and S1B), indicating the involvement of Cx43 HCs in the regulation of gene expression of PLR markers. Considering the increased number of apoptotic osteocytes observed in Δ130–136 mice,²⁴ we investigated whether the suppressed osteocyte network resulted from the apoptosis of osteocytes. Using terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL)-labeled apoptotic cells and phalloidin staining of the actin cytoskeleton, we showed a reduction in filament density in Δ130–136 mice (Figures 1D and 1E). The decrease in dendrite numbers was mainly found in non-apoptotic osteocytes of Δ130–136 mice (Figures 1D and 1F). We also determined the perilacunar-canalicular network in male mice. Similar to female mice, we bserved deterioration of the osteocyte lacuno-canalicular network and collagen disorganization in male Δ130–136 mice compared to WT and R76W mice (Figures S1E and S1F). The level of Cx43 dominant-negative mutant and endogenous Cx43 was determined using western blotting (Figure S1G). There is an ~70-kDa Cx43-GFP band in transgenic mice, and the ratio of Cx43-GFP/Cx43 protein is averaged at 0.87. Additionally, we utilized the Cx43 flox/flox mice crossed with Dmp1 promoter (Cx43 cKO) and investigated the osteocyte lacuno-canalicular network using Ploton silver staining. Unlike Δ130–136 mice, there was no significant difference in the osteocyte network between Cx43 cKO mice and WT mice (Figures S1H and S1I). These results show the importance of Cx43 HCs in maintaining the osteocyte perilacunar-canalicular network under normal physiological conditions.

Figure 1. Impairment of Cx43 HC function disrupts the osteocyte network.

Femoral midshaft cortical bone sections from female WT, R76W, and Δ130–136 mice were prepared.

(A) Collagen organization was assessed using picrosirius red staining under polarized light. Scale bar, 50 μm.

(B) Representative images of the osteocyte lacuno-canalicular network of WT, R76W, and Δ130–136 mice revealed by Ploton silver staining. Scale bar, 10 μm; n = 5–6 mice/group.

(C) Quantitative analysis of canalicular length (upper panel) and lacuno-canalicular area normalized to bone area (lower panel).

(D) Confocal images of phalloidin (green) and TUNEL (red) double-labeled sections showing the osteocyte network and apoptotic osteocytes in WT, R76W, and Δ130–136 mice. Scale bar, 5 μm.

(E) Quantifications of the percentage of the acellular bone matrix occupied by phalloidin-positive filaments.

(F) Dendrite number was quantified in both TUNEL-positive and TUNEL-negative osteocytes of WT, R76W, and Δ130–136 mice. n = 3–4 mice/group. Data are presented as mean ± SEM. ***p < 0.001; ****p < 0.0001. Statistical analysis was performed using one-way ANOVA with Tukey’s post hoc test.

Cx43 HCs mediate lacunar enlargement and PLR gene expression during lactation

PLR is necessary during lactation to meet the high calcium demand for milk production by resorbing the perilacunar matrix. To investigate the impact of osteocytic Cx43 HCs on the skeletal responses induced by lactation, 3-month-old female WT, R76W, and Δ130–136 mice were allowed to become pregnant and lactated for 12 days (Figure 2A). Backscatter scanning electron microscopy (SEM) was utilized to visualize osteocyte lacunar in the tibial diaphysis of WT, R76W, and Δ130–136 virgin or lactating mice. The experiment was conducted under a low-calcium diet to maximize perilacunar osteolysis. When compared to WT virgin mice, WT lactating mice exhibited enhanced osteocyte perilacunar resorption (Figure 2B), as evidenced by a significant increase in osteocyte lacunar area and lacunar perimeter (Figures 2C and 2D, left panels). The R76W group showed similar changes to the WT group in lacunar size (Figures 2B–2D). In contrast, these significant alterations were not observed in Δ130–136 lactating mice compared to corresponding virgin control (Figures 2B–2D). When comparing the lactation/virgin ratio among the groups, the lactation-induced lacunar enlargement was attenuated in the Δ130–136 group, with the lactation/virgin ratio close to 1 for lacunar area and perimeter quantification (Figures 2C and 2D, right panels). Notably, at the basal level, Δ130–136 mice exhibited a larger lacunar area compared to WT and R76W virgin mice (Table S1). Acid-etched SEM and Ploton silver staining were used to image osteocyte lacunae, further validating the measurement obtained by backscatter SEM (Figure 2E). Consistently, increased osteocyte lacunar size was observed in WT and R76W mice during lactation but not in Δ130–136 mice.

Figure 2. Attenuation of osteocyte lacunar changes and perilacunar-canalicular remodeling gene expression in Δ130–136 mice during lactation.

(A) Experimental diagram depicting the timeline for lactation induction and sample collection.

(B) Representative backscatter scanning electron microscopy (SEM) images of the tibial diaphysis from virgin or lactating WT, R76W, and Δ130–136 mice. Scale bar, 50 μm.

(C and D) Quantification of (C) lacunar area and (D) lacunar perimeter. n = 5–7 mice/group.

(E) Lactation-induced osteocyte lacunar changes demonstrated by acid-etch SEM (upper panels) and Ploton silver staining (lower panels). Scale bar, 10 μm.

(F–I) Relative gene expression of (F) ACP5 (TRAP), (G) MMP13, (H) cathepsin K, and (I) RANKL/OPG was determined by RT-qPCR in bone marrow-flushed osteocyte-enriched femoral and tibial diaphysis samples from WT, R76W, and Δ130–136 virgin and lactating mice. n = 5–9 mice/group. The right panels represent the corresponding lactation/virgin ratio for each genotype. Data are presented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Statistical analysis was performed using t test for virgin and lactation within each genotype and one-way ANOVA with Tukey’s post hoc test for the lactation/virgin ratio.

It has been reported that osteocytes express several osteoclast-specific genes to resorb their perilacunar matrix during lactation.^4,9 In this study, we examined the levels of several key regulatory genes involved in PLR. Osteocyte-enriched femoral and tibial diaphysis samples were prepared from WT, R76W, and Δ130–136 virgin or lactating female mice. Lactation induced a 3- to 5-fold increase in the expression of genes required for osteocyte perilacunar bone resorption, including ACP5/TRAP, MMP13, and cathepsin K in WT and R76W groups (Figures 2F–2H; Table S2). However, Δ130–136 mice did not show a response to lactation-induced PLR gene expression, with no significant differences between Δ130–136 virgin and lactating groups (Figures 2F–2H, left panels). Consistently, the lactation/virgin ratio for the expression changes of ACP5/TRAP, MMP13, and cathepsin K in Δ130–136 mice was significantly lower than in WT and R76W mice (Figures 2F–2H, right panels). Additionally, the levels of bone remodeling markers osteoprotegerin (OPG) and receptor activator of nuclear factor kappa-B ligand (RANKL) were measured in response to lactation. WT lactating mice exhibited a reduction in OPG expression and an increase in RANKL expression compared to virgin control (Figures S2A and S2B). OPG acts as a secreted decoy receptor for RANKL, protecting bone from excessive resorption by binding to RANKL. Thus, the RANKL/OPG ratio is indicative of a potentially catabolic environment for bone remodeling.²⁷ Lactating WT and R76W mice showed a significant elevation of the RANKL/OPG ratio compared to corresponding virgin controls, whereas the RANKL/OPG change was not evident in Δ130–136 mice during lactation (Figure 2I; Table S2). Considering the functional differences of R76W and Δ130–136 mutants on gap junctions and gap junctions/HCs, respectively, these results demonstrate that Cx43 HCs act as a crucial mediator for PLR during lactation.

Attenuation of lactation-induced bone loss and mechanical property changes with impaired Cx43 HCs

To assess the impact of Cx43 HCs on lactation-induced bone loss and mechanical property changes, we first measured whole-body BMD by dual-energy X-ray absorptiometry (DEXA) in WT and transgenic mice during lactation. The results showed a 15% reduction in whole-body BMD in lactating WT mice under normal diet conditions and a 30% reduction under low-calcium diet conditions (Figures S3A and S3B). Both transgenic mouse groups also exhibited significant reductions in BMD during lactation (Figures S3A and S3B, left panels). However, the ratio of BMD change in lactating Δ130–136 mice was significantly higher than WT and R76W mice (Figures S3A and S3B, right panels). Furthermore, we assessed lactation-induced changes in bone microarchitecture using three-dimensional micro-computed tomography (μCT) measurements under normal diet condition. Representative μCT images of femoral metaphyseal trabecular bone are shown in Figure 3A. Compared to corresponding virgin controls, lactating WT and R76W mice exhibited similar decreases in bone volume fractions (BV/TV) (Figure 3B). There was a reduction in trabecular number (Tb.N) and an increase in trabecular separation (Tb.Sp) in lactating WT and R76W lactating mice (Figures 3C and 3E). In contrast, lactating Δ130–136 mice did not show significant differences in BV/TV, Tb.N, and Tb.Sp compared to their virgin controls (Figures 3B, 3C, and 3E, left panels). Analysis of the lactation/virgin ratio revealed significant changes in BV/TV, Tb.N, and Tb.Sp in lactating Δ130–136 mice compared to WT and R76W groups (Figures 3B, 3C, and 3E, right panels). Lactation-induced decreases in trabecular thickness (Tb.Th) were observed in WT and both transgenic mice during lactation (Figure 3D). Additionally, we analyzed lactation-induced bone loss in the fifth lumbar vertebrae (L5) using μCT analysis. Representative μCT images are shown in Figure S3C. In line with our previous study,²⁸ Δ130–136 mice exhibited higher trabecular BV/TV and Tb.Th than WT and R76W groups (Table S3). Lactation resulted in decreased BV/TV and Tb.Th in WT and both transgenic mouse groups, with a significantly higher lactation/virgin ratio in Δ130–136 mice compared to WT and R76W groups (Figures S3D and S3F; Table S3). While lactating WT and R76W mice exhibited decreased Tb.N and increased Tb.Sp, these alterations were not observed in lactating Δ130–136 mice (Figures S3E and S3G). Taken together, these findings suggest that impairment of Cx43 HCs attenuates the effects of lactation-induced trabecular bone loss in the femur and vertebrae.

Figure 3. Attenuation of bone loss in femoral metaphyseal trabecular bone of Δ130–136 mice during lactation.

μCT was used to assess the distal femur trabecular bone of virgin or lactating WT, R76W, and Δ130–136 mice.

(A) Representative 3D models of the metaphyseal trabecular bone for all groups. Scale bar, 500 μm.

(B–E) Bone (B) volume fraction (BV/TV), (C) trabecular number (Tb.N), (D) trabecular thickness (Tb.Th), and (E) trabecular separation (Tb.Sp) are shown. n = 5–6 mice/group. The right panels represent the corresponding lactation/virgin ratio for each genotype. Data are presented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Statistical analysis was performed using t test for virgin and lactation within each genotype and one-way ANOVA with Tukey’s post hoc test for lactation/virgin ratio.

Similar attenuation of lactation-induced cortical bone loss was also observed in Δ130–136 mice. μCT analysis was conducted at the femoral midshaft cortical bone (Figure 4A). The results showed that lactating WT and R76W mice exhibited significant reductions in bone area (B.Ar) and bone area fraction (B.Ar/T.Ar), accompanied by an enlarged bone marrow area fraction (M.Ar/T.Ar) compared to corresponding control mice. However, these lactation-induced differences were diminished in Δ130–136 mice (Figures 4B–4D and 4E). Lactation did not have an effect on bone total area (T.Ar) in any of the genotypes (Figure 4C). Notably, B.Ar and T.Ar were significantly larger in Δ130–136 virgin mice compared to WT and R76W groups (Table S4), consistent with findings from our previous studies.^24,28 The cortical thickness (Ct.Th) was significantly reduced in lactating WT, R76W, as well as Δ130–136 mice (Figure 4F). Torsional strength, determined by polar moment of inertia (pMOI), decreased in lactating WT and R76W mice but not in Δ130–136 mice (Figure 4G). The lactation/virgin ratio further confirmed the attenuated response in Δ130–136 mice, showing reduced changes in B.Ar, B.Ar/T.Ar, M.Ar/T.Ar, and Ct.Th compared to WT and R76W groups (Figures 4B–4D and 4F, right panels). Biomechanical properties were evaluated through three-point bending flexural testing of the femurs. Similar to our previous observations,²⁴ Δ130–136 mice exhibited a lower elastic modulus and higher ultimate force compared to WT and R76W mice at the basal level (Table S4). Significant reductions in elastic modulus and ultimate stress induced were observed in lactating WT and R76W mice, while lactating Δ130–136 mice had preserved biomechanical properties, reflected by a higher lactation/virgin ratio compared to WT and R76W groups (Figures 4H and 4I). However, ultimate force and stiffness were decreased in all genotypes in response to lactation (Figures S4A and S4B). These findings provide evidence that Cx43 HCs play a crucial role in regulating cortical bone structure and mechanical properties during lactation.

Figure 4. Mitigation of lactation-induced cortical bone loss and mechanical property changes in Δ130–136 mice.

μCT was used to assess the femoral midshaft cortical bone (55% site) of virgin or lactating WT, R76W, and Δ130–136 mice.

(A) Representative 3D models of the midshaft cortical bone for all groups. Scale bar, 500 μm.

(B–G) Bone area (B) (B.Ar), (C) total area (T.Ar), (D) bone area fraction (B.Ar/T.Ar), (E) bone marrow area (M.Ar/T.Ar), (F) cortical thickness (Ct.Th), and (G) polar moment of inertia (pMOI) are shown. The three-point bending assay was performed for the femur of virgin or lactating WT, R76W, and Δ130–136 mice.

(H and I) Elastic modulus (H) and ultimate stress (I). n = 5–6 mice/group. The right panels represent the corresponding lactation/virgin ratio for each genotype. Data are presented as mean ± SEM. *p < 0.05; **p < 0.01; ****p < 0.0001. Statistical analysis was performed using t test for virgin and lactation within each genotype and one-way ANOVA with Tukey’s post hoc test for lactation/virgin ratio.

Cx43 HCs mediate osteolysis and endosteal bone remodeling during lactation

The expression of TRAP, a marker commonly used to assess osteolysis activities, was evaluated in osteocytes and osteoclasts. The results showed that lactation induced a significant increase in TRAP-positive osteocytes (indicated by black arrows in Figure 5A) in the midshaft femoral cortical bone of WT and R76W mice under low-calcium diet conditions. The increase in TRAP-positive osteocytes was 10.84- ± 1.42-fold in WT group and 8.63- ± 1.36-fold in the R76W group (Figures 5A and 5B). Similar levels of induction were observed in lactating WT and R76W mice under normal diet conditions, 9.35- ± 1.09-fold in WT group and 7.24- ± 0.79-fold in R76W group (Figures S5A and S5B). However, the osteolysis response to lactation was significantly mitigated in Δ130–136 mice as indicated by reduced TRAP expression in osteocytes (Figures 5B and S5B). This result is in good agreement with the result of blunted osteocyte lacunar enlargement, as described above. Furthermore, the study assessed endosteal osteoclastogenesis during lactation and found that it was also attenuated in Δ130–136 mice under both low-calcium and normal diet conditions. The Δ130–136 mice showed less osteoclast surface to bone surface (Oc.S/BS) changes compared to WT and R76W groups (Figures 5C and S5C). Dynamic histomorphometric analysis using calcein and alizarin red double labeling assay was performed to assess endosteal bone formation. The results showed that lactation caused significant reductions of mineral apposition rate (MAR), bone formation rate (BFR/BS), and mineralizing surface (MS/BS) in WT and R76W mice under normal diet condition compared to virgin controls. However, these bone formation changes were impaired in the Δ130–136 group (Figures 5D–5G). Taken together, these results suggest that impaired osteocytic Cx43 HCs in Δ130–136 mice attenuate perilacunar resorption and endosteal bone remodeling responses during lactation.

Figure 5. Reduction of osteolysis and osteoclastogenesis, and altered bone formation in Δ130–136 mice during lactation.

(A) Representative images of the femoral midshaft endosteal surface and osteocytes stained for TRAP (black arrows) in virgin or lactating WT, R76W, and Δ130–136 mice. Scale bar, 50 μm.

(B and C) Histomorphometric quantification of TRAP-positive osteocytes per bone perimeter (B) and osteoclast surface per bone perimeter (C). n = 5–6 mice/group. Dynamic histomorphometric analysis was performed on the tibial midshaft cortical endosteal surface in virgin or lactating WT, R76W, and Δ130–136 mice.

(D) Representative images of calcein (green) alizarin (red) double labeling. Scale bar, 200 μm.

(E–G) Mineral apposition rate (E) (MAR), (F) mineralizing surface/bone surface (MS/BS), and (G) bone formation rate (BFR/BS) were assessed. n = 6 mice/group. The right panels represent the corresponding lactation/virgin ratio for each genotype. Data are presented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Statistical analysis was performed using t test for virgin and lactation within each genotype and one-way ANOVA with Tukey’s post hoc test for lactation/virgin ratio.

Impaired osteolytic HCs have minimal effect on normal serum/milk calcium levels with high serum 1,25-dihydroxyvitamine D3

Given the markedly attenuated PLR in Δ130–136 lactating mice, we asked whether maternal serum calcium level and milk calcium supply to offspring were affected. Male and female pups from WT, R76W, and Δ130–136 lactating mice were weighed separately at day 12 after delivery. No differences in average pup weight per litter were observed among WT and the two transgenic mice groups under both normal and low-calcium diet conditions (Figure S6A). Milk calcium and serum calcium levels of Δ130–136 mice were also maintained at normal levels compared to WT and R76W groups (Figures S6B and S6C). During lactation, serum 1,25-dihydroxyvitamine D3 (1,25 (OH)₂D₃), which is the most active hormonal form of vitamin D, modulates calcium absorption in the intestine and bone calcium mobilization.²⁹ Compared to virgin controls, there was an increase in 1,25 (OH)₂D₃ level in lactating WT and transgenic mice groups (Figure S6D). Noticeably, Δ130–136 mice showed a significantly amplified increase compared to WT and R76W mice groups during lactation (Figure S6D). These results suggest that appropriate milk calcium supply to the offspring is maintained despite the attenuated PLR in Δ130–136 lactating mice.

Blocking Cx43 HCs blunts PTHrP-induced calcium influx, cytosolic acidification, and expression of PLR genes in osteocytes

During lactation, PTHrP is secreted from the mammary gland and functions on PTH1R in osteocytes to initiate acidification and perilacunar resorption. Cx43 HCs mediate the influx and efflux of ions, including ions and small molecules, across plasma membrane. To investigate if Cx43 HCs are involved in the PTHrP-induced intracellular acidification process, we treated the MLO-Y4 osteocytes with 100 nM PTHrP and the Cx43(E2) antibody, which specifically inhibits Cx43 HCs.²⁵ The activities of Cx43 HCs were determined using the ethidium bromide (EtBr) dye uptake assay. PTHrP induced Cx43 HC opening, which was blocked by the Cx43(E2) antibody (Figure 6A). We next assessed PTHrP receptor downstream signaling, protein kinase A (PKA) activation.^30,31 PTHrP mediated PKA activation in MLO-Y4 cells, and this activation was inhibited by the Cx43(E2) antibody (Figure 6B). We next determined the expression of PTH1R after PTHrP treatment. MLO-Y4 cells showed a trend of time-dependent increase in PTH1R from 24 to 72 h (Figure S6E). We further examined the PTH1R mRNA level under lactation condition in vivo. As shown in Figure S6F, there was a trend of increase in PTH1R mRNA level (p = 0.09) in bone from WT lactating mice. PTHrP-induced calcium influx was determined using Cal-520, a free calcium indicator. Representative calcium images and real-time quantification of calcium signals indicated that the Cx43 blocking antibody dramatically impaired intracellular calcium responses after PTHrP treatment (Figures 6C and 6D). Quantifications further showed a significant decrease in the amplitude, occurrence, and response time of calcium peak in the Cx43(E2) group (Figure 6E). Moreover, intracellular pH was measured using the pH-sensitive dye 5-(and-6)-carboxy SNARF-1. PTHrP treatment for 48 h significantly decreased cytosolic pH value, while blocking Cx43 HCs with the Cx43(E2) antibody reversed the acidification to untreated level (Figure 6F). Accordingly, genes responsible for acidification, including Car2 and ATP6V1G1, were increased 2-fold in response to PTHrP treatment, and this increase was blunted by the Cx43(E2) antibody (Figure 6G). The expression of osteocyte perilacunar resorption genes, such as cathepsin K, MMP2, and MMP13, was also increased in MLO-Y4 cells treated with PTHrP and was inhibited using the Cx43 HC blocking antibody (Figure 6H). These data support the key roles of Cx43 HCs in mediating PTHrP-induced osteocyte acidification and the PLR process.

Figure 6. Blocking Cx43 HCs by Cx43(E2) antibody blunts PTHrP-induced acidification and expression of perilacunar/canalicular remodeling genes in MLO-Y4 osteocytes.

(A) Representative images of HC dye uptake performed with ethidium bromide (EtBr) (red fluorescence) in treated MLO-Y4 cells (left panels). Scale bar, 50 μm. The intensity of EtBr fluorescence was quantified (right panel). n = 4 biologically independent experiments/group.

(B) PKA activity was determined in treated MLO-Y4 cells. n = 3 biologically independent experiments/group.

(C) Representative Ca²⁺ images of control and Cx43(E2)-treated MLO-Y4 cells at 100, 300, and 600 s after PTHrP application. Live cell images were taken at 2-s intervals for a total of 15 min. Scale bar, 10 μm.

(D) Real-time quantification of calcium signals in the control group (upper panel) and the Cx43(E2)-treated (lower panel) group. Each curve represents one cell. n = 6/group from three independent experiments.

(E) Summary graphs of ΔF/F₀ (left panel), Ca²⁺ signal response numbers (middle panel), and response time after PTHrP application (right panel) in the control and Cx43(E2)-treated MLO-Y4 cells.

(F) Representative images of intracellular pH determined by SNARF-1 AM dye (left panels). Scale bar, 20 μm. After treatment and dye loading, MLO-Y4 cells were excited at 514 nm, and the emission fluorescence was measured at both 580 and 640 nm. The intracellular pH value of treated MLO-Y4 cells was calculated using a calibration curve derived from MLO-Y4 cells with pH fixed to known values (right panel). n = 7/group from three independent experiments.

(G) Relative gene expression of Car2 and ATP6V1G1 was determined by RT-qPCR in MLO-Y4 cells after treatment. n = 3 biologically independent experiments/group.

(H) Relative gene expression of cathepsin K, MMP2, and MMP13 was determined by RT-qPCR in treated MLO-Y4 cells. n = 3 biologically independent experiments/group. Data are presented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.. Statistical analysis was performed using one-way ANOVA with Tukey’s post hoc test.

Impairment of Cx43 HCs impedes bone structure, mechanical properties, bone formation, and bone remodeling gene expression changes during post-lactation recovery

In addition to perilacunar matrix resorption during lactation, osteocytes also function to deposit perilacunar matrix during post-lactation recovery in the maternal skeleton. On day 12 of lactation, pups were removed from female WT, R76W, and Δ130–136 mice to induce skeletal recovery (Figure 7A). The raw data of post-lactation recovery are summarized in Tables S2–S5. Femoral and vertebral bone structure was determined by μCT analysis. The ratio of post-lactation to lactation level was calculated to reflect the bone accrual in each genotype. Impairment of Cx43 HCs in Δ130–136 mice impeded the increase in femoral trabecular BV/TV and Tb.N and the decrease in Tb.Sp during post weaning, although no significant difference in Tb.Th was observed (Figures 7B–7F). Similarly, Δ130–136 mice showed attenuated BV/TV, Tb.N, and Tb.Th increase in L5 vertebrae compared to WT and R76W mice during post-lactation recovery (Figures S7A–S7D). Femoral cortical μCT revealed a blunted Ct.Th recovery and a close to significant reduction of B.Ar/T.Ar in Δ130–136 mice (Figures 7G and 7H), with no significant differences in M.Ar and pMOI among the three groups (Figures S7E and S7F). Bone mechanical properties were assessed by three-point bending tests. Impaired recovery of elastic modulus and stiffness were observed in post-weaning Δ130–136 mice (Figures 7I and S7H), while there were no significant differences in ultimate stress and ultimate force among WT and two transgenic mice groups (Figures 7J and S7G). We found that bone resorption and osteolysis (indicated by TRAP⁺ osteocytes) levels were significantly higher in Δ130–136 mice compared to WT and R76W mice during post-lactation recovery (Figures 7K–7M). We also showed that expression levels of PLR genes, including ACP5, cathepsin K, and MMP13, remained high in Δ130–136 mice after post-lactation recovery (Figures 7O–7P). Moreover, gene expression analysis showed a higher RANKL/OPG in Δ130–136 mice during post-lactation recovery, indicating a more catabolic bone remodeling environment (Figure 7Q). Dynamic histomorphometric analysis demonstrated inhibition of the increase in bone formation parameters during the post-lactation period, including MAR, MS/BS, and BFR/BS in Δ130–136 mice compared to WT and R76W groups (Figures 7R–7T), whereas calcium levels are comparable among these mice (Figure S7I). Altogether, these data suggest that Cx43 HCs also play crucial roles in regulating post-weaning bone recovery.

Figure 7. Impairment of Cx43 HC function impedes bone structure, mechanical properties, bone formation, osteolysis, and perilacunar-canalicular remodeling gene expression changes during post-lactation recovery.

(A) Experimental diagram depicting the timeline for post-lactation recovery and sample collection. μCT was used to assess femoral trabecular and cortical bone of WT, R76W, and Δ130–136 mice after 7 days of post-lactation recovery.

(B) Representative 3D models of the metaphyseal trabecular bone (upper panels) and midshaft cortical bone (lower panels) for all groups. Scale bar, 500 μm.

(C–H) Bone (C) volume fraction (BV/TV), (D) trabecular number (Tb.N), (E) trabecular thickness (Tb.Th), (F) trabecular separation (Tb.Sp), (G) bone area fraction (B.Ar/T.Ar), and (H) cortical thickness (Ct.Th) are shown. n = 5–6 mice/group. The three-point bending assay was performed for femur of WT, R76W, and Δ130–136 mice.

(I and J) Elastic modulus (I) and ultimate stress (J).

(K) Representative images of the TRAP-stained femoral midshaft endosteal surface (red arrows) and osteocytes (black arrows) in WT, R76W, and Δ130–136 mice during post-lactation recovery. Scale bar, 50 μm.

(L and M) Histomorphometric quantification of (L) TRAP-positive osteocytes per bone perimeter and (M) osteoclast surface per bone perimeter. n = 6 mice/group.

(N–Q) Relative gene expression of (N) ACP5 (TRAP), (O) MMP13, (P) cathepsin K, and (Q) RANKL/OPG was determined by RT-qPCR in bone marrow-flushed osteocyte-enriched femoral and tibial diaphysis of WT, R76W, and Δ130–136 mice after 7 days of post-lactation recovery. Dynamic histomorphometric analysis was performed on the tibial midshaft cortical endosteal surface. n = 5–7 mice/group.

(R–T) Mineral apposition rate (R) (MAR), (S) mineralizing surface/bone surface (MS/BS), and (T) bone formation rate (BFR/BS) were assessed. n = 6 mice/group. Data are presented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Post-lactation/lactation ratio for each genotype was determined. Statistical analysis was performed using one-way ANOVA with Tukey’s post hoc test.

DISCUSSION

In this study, we identified the important role of osteocytic Cx43 HCs in PLR and PTHrP-mediated cytosolic acidification using a lactation model in WT and transgenic mice with osteocytic Cx43 dominant-negative mutants. Compared to WT and R76W groups, Δ130–136 mice attenuated osteocyte PLR under normal and lactation conditions, resulting in blunted lactation-induced cortical and trabecular bone responses, as well as impaired post-weaning bone recovery. Given the impaired gap junction channels and HCs in Δ130–136 mice, but only impaired gap junction channels in R76W mice, our results suggest that osteocytic Cx43 HCs play predominant roles in mediating the resorption and replacement of their parilacunar matrix. The underlying cellular mechanism involves the regulation of PTHrP-induced osteocyte HC opening, PKA activation, calcium influx, and intracellular acidification process.

Osteocyte PLR plays a fundamental role in bone homeostasis and is responsible for maintaining the lacuno-canalicular network and bone quality.^32,33 The PLR process is highly conserved among species and is triggered in the context of metabolic stress.^34–36 In the current study, we used a mouse lactation model, which is ideal for investigating the dynamic regulation of the PLR process within a short period of time. During lactation, osteocytes express osteoclast-specific markers, including cathepsin K, TRAP, and MMPs, to mobilize calcium from the skeleton to breast milk. After weaning, the skeleton undergoes substantial recovery, and osteocytes express osteoblast-specific genes to restore mineralization.^12,13 Our previous studies have demonstrated the importance of Cx43 HCs in modulating bone homeostasis.²⁴ Δ130–136 mice exhibited increased cortical B.Ar, T.Ar, and pMOI compared to WT and R76W mice, along with lower elastic modulus and higher ultimate force. Our previous study revealed enhanced osteoclastic endocortical bone resorption and periosteal bone formation in Δ130–136 mice,²⁴ resulting in a net increase in the diameter of cortical bone. These cortical bone changes are associated with higher whole-bone strength as evaluated by three-point bending tests. However, when calibrated by the cross-section, Δ130–136 mice showed significantly decreased material properties. In addition, we observed increased apoptotic osteocytes and empty lacunae in Δ130–136 mice,^24,28 further compromising bone quality. The increased trabecular bone volume may be attributed in part to the higher trabecular bone formation rate in Δ130–136 mice, while no difference in trabecular bone resorption was observed.²⁴ In this study, we implicate osteocytic Cx43 HCs as a pivotal regulator of PLR. Impairment of Cx43 HCs in Δ130–136 mice resulted in a severe deterioration of osteocyte perilacunar-canalicular network, along with dysregulation of PLR genes. Interestingly, Δ130–136 mice showed increased lacunar size, which may be associated with elevated TRAP (ACP5) expression in the osteocytes. Lactation-induced PLR was mitigated in Δ130–136 mice, as was lactation-induced bone loss. To assess the changes in cortical and trabecular bone structure, we used μCT analysis of the femur and vertebrae. Mice with impaired Cx43 HCs exhibited reduced femoral thickness (Ct.Th) or trabecular thickness (Tb.Th) in response to lactation, while other parameters showed no significant responsiveness. These findings are consistent with previous studies using lactation mouse models with deficiencies in osteocytic PLR regulators, such as cathepsin K and TGFβ receptor II.^10,37 Furthermore, it has been reported that lactation-induced bone resorption and post-lactation-induced bone formation are more pronounced in trabecular-rich compartments compared to the appendicular skeleton.^12,38 Consistently, our results showed a 38.4% decrease and a 58.7% increase in trabecular BV/TV during lactation and 7 days post weaning, respectively, while there was a 21.3% decrease and a 1.7% increase of femoral cortical B.Ar/T.Ar.

Using the CD1 mouse strain, which has an average litter size of 8–13 pups, Qing et al. has demonstrated that the osteocyte lacunar area is enlarged during lactation.⁴ However, a later study using a transgenic mouse line with a low litter size of 4.5 average pups showed an insufficient calcium demand during lactation, resulting in a non-significant increase in osteocyte lacunar area.⁹ Due to the limited litter size of our R76W and Δ130–136 transgenic mice, we employed both a low-calcium diet and a normal diet in our study. The low-calcium diet was initiated on the first day after delivery, and an enlargement of lacunae area by 24% was observed in the lactating WT group, which is comparable to the previous report.⁴ However, lactation-induced changes in bone structure, mechanical properties, bone formation, and gene expression were presented under normal diet, which mimics a more physiological condition. When comparing the reduction of BMD, the low-calcium diet resulted in a more profound decrease (30%) than the normal diet (15%) in lactating WT mice. In contrast, high and low dietary calcium intakes do not alter the loss of BMD in nursing women,^39,40 indicating potential differences in lactational bone physiology between human and rodents. Interestingly, the milk calcium level was similar between lactating mice under a normal diet or a low-calcium diet and also among lactating WT, R76W, and Δ130–136 mice. This suggests compensation by hormonal changes to maintain mineral homeostasis and ensure the unaltered skeletal development of the offspring.

The molecular mechanism by which osteocytes control PLR remains largely elusive, and the role of Cx43 HCs in response to PTHrP-induced osteocyte PLR has not been investigated. Our work here demonstrates that impairment of Cx43 HCs attenuates lactation-induced PLR by inhibiting PTHrP-mediated calcium influx, PKA activation, and the cytosolic acidification process. Cytosolic acidification exports protons to the extracellular space, acidifying the extracellular environment, which, in turn, leads to the activation of extracellular enzymes needed for PLR.^41,42 Specifically, PTHrP exerts its functions through PTH1R, which induces rapid calcium uptake and promotes cAMP/PKA signaling.^43,44 PKA phosphorylation, followed by PTH1R activation, promotes V-ATPase-mediated endosomal acidification. Cathepsin K requires V-ATPase to keep an acidic environment for its participation in extracellular matrix degradation.^30,45 Although expression of MMPs is not directly dependent on pH, they are activated by cathepsin-dependent cleavage at low pH.^46,47 A previous study has revealed that the inhibition of Na⁺/H⁺ exchange blocks PTH/PTHrP-induced cytosolic acidification in osteoblasts.^31,48 The opening of Cx43 HCs allows the transmembrane passage of ions such as Ca²⁺, Na⁺, and K⁺, with relatively low selectivity.^16,49 In the MLO-Y4 osteocyte cell line, our results showed that Cx43(E2) antibody inhibited the opening of Cx43 HCs after PTHrP treatment and blunted PTHrP-induced cytosolic acidification and PLR gene expression. The underlying mechanism is unlikely to involve direct proton transport through Cx43 HCs since protons are generated inside the osteocytes, and intracellular pH is lower than extracellular during PLR.⁵⁰ However, we showed that the PTHrP-induced intracellular Ca²⁺ signals were inhibited by the Cx43 HC blocking antibody, suggesting that PTHrP-activated HCs facilitate Ca²⁺ influx, given that the extracellular Ca²⁺ levels are significantly higher than intracellular levels. Previous studies have shown that changes in cytosolic Ca²⁺ are closely associated with intracellular acidification in various types of cells.^51–53 Ca²⁺ plays a role in a variety of activities for V-ATPase,^54–56 and a Ca²⁺-dependent exocytosis machinery for cathepsin K release has been described in active osteoclasts.^57,58 The degradation activities of MMPs also require Ca²⁺.^59,60 Additionally, it has been reported that Ca²⁺ can regulate Car2 activity.^61,62 Future studies are needed to further investigate the mechanisms and enhance our understanding of how Cx43 HCs regulate PTHR1 function, which could potentially lead to improved treatments for disorders resulting from defects in PTH/PTHrP signaling.

In summary, this study unveils a previously unrecognized role of osteocytic Cx43 HCs in mediating PLR during lactation and post weaning. Lack of Cx43 HC function in vivo abolishes lactation-induced osteocyte lacunar enlargement and attenuates bone loss by regulating endocortical bone formation/resorption. It also impairs the post-lactation recovery of bone structure and strength. Furthermore, blocking Cx43 HC function in vitro blunts osteocyte acidification after PTHrP treatment. These findings highlight the crucial role of Cx43 HCs in regulating PTHrP/PTHR1 signaling and the resorption and deposition of perilacunar mineral during the reproductive cycle.

Limitations of the study

There are three limitations in the current study. First, the Cx43 transgenic mouse models used are driven by the 10-kb Dmp1 promoter. Since 10-kb Dmp1 expression is not restricted to osteocytes, we cannot rule out the possibility of the effects from other organs. Secondly, we did not investigate the effects of Cx43 deletion in osteocytes on lactation-induced perilacunar remodeling. Although the Cx43 cKO and Δ130–136 mouse models share several similar features, such as increased osteocyte apoptosis, enlarged bone marrow cavity, and compromised mechanical properties, they represent different models. Deletion of Cx43 in osteocytes disrupts the function of both HCs and gap junctions, with potential compensatory effects affecting the expression of other proteins. Our transgenic models with dominant-negative Cx43 mutations provide unique insights into dissecting the specific functions of Cx43 HCs and gap junctions. However, we acknowledge the possibility of other altered responses of osteocytes in Cx43 cKO mice during lactation-induced perilacunar remodeling, which warrants further investigation. Thirdly, our study mainly focuses on lactation-induced lacunar size changes, while the changes of canalicular dimensions under lactation in our transgenic models would be a gap to be addressed in future studies.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Jean X. Jiang (jiangj@uthscsa.edu).

Materials availability

All unique/stable reagents and mice generated in this study will be made available on request to the lead contact as indicated above, but we may require a completed materials transfer agreement if there is potential for commercial application.

Data and code availability

• All data reported in this paper will be shared by the lead contact by request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Animal models

Two transgenic mouse models, R76W and Δ130–136, overexpressing dominant-negative Cx43 mutants driven by 10 kb Dmp1 promoter, were generated for this study.²⁴ Female mice (WT, and two transgenic) of the C57BL/6J strain, aged 3-month and weighing approximately 20–23 g, were used. The mice were allowed to become pregnant and lactated for a period of 12 days. After delivery, the litter size was adjusted to 6–7 pups to maintain similar breast milk requirements. On the 12th day of lactation, the pups were removed for forced weaning, which induces skeletal recovery. Mice were euthanized after 12 days of lactation or 7 days post-lactation. In addition, a low calcium diet (Teklad Diet TD.95027, Envigo, Houston, TX, USA) was also utilized to enhance calcium demand. For the low calcium diet condition, the litter size was adjusted to 4–6 pups. Age- and weight-matched virgin mice were used as controls. The mice were housed in a temperature-controlled room with a light/dark cycle of 12 h at the University of Texas Health Science Center at San Antonio (UTHSCSA) Institutional Lab Animal Research facility. All animal protocols were conducted in accordance with the National Institutes of Health guidelines for care and use of laboratory animals and were authorized by the UTHSCSA Institutional Animal Care and Use Committee (IACUC).

Bone mineral density (BMD) and micro-computed tomography (μCT) measurement

The BMD values of the whole body were measured for mice after 12 days of lactation and their corresponding virgin controls. Mice were anesthetized and placed on a specimen tray. A Lunar PIXImus dual-energy X-ray absorptiometry (DEXA) densitometer from GE Medical Systems (Piscataway, NJ, USA) was used for the measurements, following a previously described procedure.²⁸

To evaluate bone structural properties, a Brüker SkyScan 1172 scanner from Brüker microCT (Kontich, Belgium) was used. The scanner was operated under the following parameters: 59 kV voltage, 167 μA beam intensity, 0.5 mm aluminum filter, 750 ms exposure time, 0.4-degree rotation step, 4-frame averaging, 1024 × 1024 pixel matrix, and a 9.98 μm isotropic voxel dimension. The acquired images underwent background noise removal by eliminating disconnected objects smaller than 4 pixels in size. Two volumes of interest (VOI) were selected for analysis, one in the metaphysis and the other in the midshaft regions of the femur. In the distal femoral metaphysis, the trabecular bone VOI was positioned 50 slices proximal to the growth plate and extended 150 slices in the proximal direction. An irregular contour was manually drawn along the perimeter of the cortical bone a few pixels away from the endosteal boundary. These contours were created every 10 slices and interpolated to obtain a 3D VOI. A grayscale value threshold of 90 was set for a set of 8-bit slices. The analysis of cortical bone structure was performed over 50 slices centered at the 55% position (from proximal to distal) in the femur diaphysis. The cortex was selected using automated, density-driven contouring with a threshold of 120, for a set of 8-bit slices. For the lumbar vertebral body 5 (L5) VOI, the centrum of the vertebral body, from endplate to endplate and inside the endosteal margin, was considered. The structural morphometric properties of the cortical and trabecular regions were analyzed using CT Analyser software (CTAn 1.18.8.0, Bruker Skyscan).

Mechanical testing of femur bone

After μCT scanning, three-point bending tests were performed on the femur. The remaining soft tissues were carefully removed, and the femur was thawed to room temperature prior to testing. The tests were carried out using a micromechanical testing system (Mach-1 V500CST, Biomomentum, Laval, Canada). During the three-point bending test, the femur was placed on supports with a span distance of 8 mm. The loading pin was positioned at the femur midshaft of the femur. The test was conducted in displacement control mode, with a constant rate of 0.05 mm/s. Data were collected at a sampling rate of 200 Hz for all measurements. To calculate mechanical properties, accurate cross-sectional areas and moments of inertia for each individual sample were determined from the μCT images. These parameters were used in the calculation of the mechanical properties of the femur.

Dynamic bone histomorphometry

To assess bone formation, intraperitoneal (IP) injection of calcein and alizarin red were administered to the mice. Seven days before euthanization, the mice received IP injection of calcein (C0875, Sigma-Aldrich, St. Louis, MO, USA) at a dosage of 20 mg/kg of body weight. This was followed by an injection of alizarin red (A5533, Sigma-Aldrich, St. Louis, MO, USA) injection at a dosage of 30 mg/kg of body weight five days later. After euthanization, the tibias were dissected, fixed in 70% ethanol, and embedded in methylmethacrylate (MMA) for plastic sectioning. A precision wafering saw (PICO 155, PACE Technologies, Tucson, AZ, USA) was used to create 80 μm thick cross-sectional surfaces, which were then polished using sandpaper with a grit size of P1200 on the PHOENIX 4000 system (Buehler, Lake Bluff, IL, USA), as previously described.⁶³ Two-color fluorescent images were obtained using a fluorescence microscope (Keyence BZ-X710, Osaka, Japan). Quantification of the following parameters was performed at the midshaft of the tibia using NIH ImageJ software (NIH, USA): total perimeter (BS); single label perimeter (sLS); double label perimeter (dLS), and double-label area (dL.Ar). From these measurements, the following values were calculated: mineralizing surface [MS/BS = (sLS/2+ dLS)/BS], mineral apposition rate [MAR = (dL.Ar/dLS)/5], and bone formation rate (BFR/BS = MAR × MS/BS). These parameters provide information on bone formation rates and dynamics.

Scanning electron microscopy (SEM)

To image osteocyte lacunae of cortical bone at the mid-diaphysis, the backscatter SEM was employed. Tibias were dissected and fixed using freshly prepared 4% paraformaldehyde in PBS (pH 7.4). After fixation, samples were transferred to 70% ethanol for preservation and subsequently dehydrated in ascending concentrations of ethanol. The samples were then embedded in MMA resin. The MMA-embedded blocks were sectioned in an anteroposterior direction using a water-cooled diamond-impregnated circular saw (Buehler, Lake Bluff, IL, USA). To obtain smooth surfaces, the sample blocks were polished using alumina alpha micropolish II solutions with decreasing particle sizes of 1 mm, 0.3 mm, and 0.05 mm, applied with a soft cloth rotating wheel. For backscattered SEM imaging, the polished samples were coated with gold and palladium. For the acid-etched SEM, the samples were subjected to a 2 to 10-s treatment with 37% phosphoric acid, followed by a 5-min rinse with 5.25% sodium hypochlorite. Following the acid etching, the samples were also coated with gold and palladium to enable secondary electron image analysis. The analysis was conducted using an AJEOL JSM-6300 scanning electron microscope (JEOL Limited, Tokyo, Japan), as previously reported.⁶⁴ The quantification of osteocyte lacunar area and perimeter was performed by measuring over 100 individual osteocyte lacunae measurements in each sample using ImageJ software (NIH, USA). This allowed for the evaluation and measurement of the osteocyte lacunae characteristics.

Histology and immunohistochemistry

Femurs were collected and fixed in 4% PFA for 2 days and decalcified using 10% EDTA (pH 7.5) for 3 weeks. The decalcified femurs were then embedded in paraffin, and 5-μm-thick sections were obtained. TRAP staining was performed to determine osteoclast activity.²⁸ Osteoclast surface (Oc.S) and bone surface (BS) on the femoral midshaft along the endosteal surface were quantified using ImageJ software (NIH, USA). Picro-sirius red staining was performed to assess collagen fiber orientation. Briefly, paraffin-embedded sections were stained in a saturated aqueous solution of picric acid and 0.1% Direct Red 80. The sections were imaged under polarized light microscopy. For visualization of the osteocyte lacuno-canalicular network, paraffin embedded sections were subjected to Ploton silver staining.⁶⁵ The sections were incubated in a solution of 50% silver nitrate (S181–25, Thermo Fisher Scientific, Waltham, MA, USA), 1% formic acid, and 2% gelatin solution. Slides were then washed in 5% sodium thiosulfate (S-8503, Sigma-Aldrich, St. Louis, MO, USA). Canalicular length, lacuno-canalicular area, and bone area on the femoral midshaft were quantified using ImageJ software (NIH, USA). For immunohistochemistry, after antigen retrieval, paraffin sections were probed with anti-MMP13 (ab39012, 1:200, Abcam, Cambridge, MA, USA) or anti-Cathepsin K (ab19027, 1:200, Abcam, Cambridge, MA, USA) overnight at 4°C. The sections were probed with a biotin-labeled secondary antibody and ABC Reagent. Staining was visualized with DAB Chromogen (SK-4100, Vector Laboratories, Burlingame, CA, USA) and hematoxylin was used as a counterstain. Images were captured using an optical microscope (BZ-X710, KEYENCE).

For the phalloidin/TUNEL double labeling assay, femurs were embedded in OCT compound, and 12-μm-thick frozen sections were prepared. The In Situ Cell Death Detection Kit, TMR red (Roche, Pleasanton, CA, USA), was used to detect apoptotic osteocytes, followed by staining with Alexa Fluor 488-conjugated phalloidin (1:100) for confocal imaging (Zeiss LSM 810, Jena, Germany). Quantifications of the filament density and dendrite number were performed with ImageJ software (NIH, USA).

RNA extraction and RT-qPCR

Total RNA was isolated from bone tissues and MLO-Y4 cells using TRI Reagent (Molecular Research Center, TR118, Cincinnati, OH, USA) following the manufacturer’s instructions. The long bones (femurs and tibias) were isolated, removing soft tissues, and the bone shafts were prepared by removing the proximal and distal ends. The bone marrow was completely flushed out with RNase-free PBS. The diaphyses were subjected to three sequential digestions, as previously described.^66,67 Briefly, the diaphyses were incubated with α-MEM containing 0.2% type 1 collagenase (Sigma, St. Louis, MO, USA) for 30 min at 37°C. This was followed by incubation with 5 mM EDTA/0.1% BSA, pH 7.4 for 30 min after washing with PBS three times. Subsequently, the diaphyses were digested with 0.2% type 1 collagenase additionally and then rinsed with PBS. The osteocyte-enriched bones were pulverized using a frozen mortar and pestle in liquid nitrogen.

cDNA was synthesized from 1 μg of total RNA using the high-capacity RNA-to-cDNA kit (43–889-50, Applied Biosystems, Bedford, MA, USA). Real-time PCR was performed using an ABI 7900 PCR device and SYBR Green (1725124, Bio-Rad Laboratories, Hercules, CA, USA) with a two-step protocol consisting of an initial denaturation at 94°C for 15 s and 60°C for 60 s. The ΔΔCT method was used for qPCR data analysis, and GAPDH was used as the housekeeping gene control. The primer sequences used in this study are listed in Table S6. Experiments were conducted in triplicates.

Western blotting

Bone marrow-flushed femurs and tibias were pulverized in liquid nitrogen and lysed in lysis buffer (5 mM Tris, 5 mM EDTA/EGTA, pH 8.0). The membrane protein was prepared as previously reported.^28,63 Briefly, lysates were centrifuged at 45,000 × g for 45 min and the pellet was resuspended in lysis buffer with 1% sodium dodecyl sulfate (SDS). The membrane protein extract was collected, and protein concentration was determined using a BCA assay (#23225, Thermo Scientific, Rockford, IL, USA). Proteins were separated by SDS-PAGE and probed with antibodies against affinity-purified Cx43 antibody (1:300 dilution),⁶⁸ and visualized by a Licor Odyssey Infrared Imager (Lincoln, NE, USA). The band intensity was quantified by the NIH ImageJ software (NIH, USA).

Serum and milk analysis

Blood was collected from virgin and lactating mice after a 3-h fasting, and serum was obtained by centrifugation. Serum 1,25 (OH)₂VD₃ level was measured using the ELISA kit (MBS2602146, MyBioSource, San Diego, CA, USA). Serum calcium concentration was determined using the QuantiChrom Calcium Assay kit (DICA-500, BioAssay Systems, Hayward, CA, USA). Milk was collected from lactating mice after stimulation with 0.1 mL (2 IU) of oxytocin (O3251, Sigma, St. Louis, MO, USA), as previously described.⁶⁹ The milk was diluted 1:100 for calcium concentration measurement, and protein concentration was determined using the Micro BCA protein kit (Thermo Fisher Scientific, Waltham, MA, USA) for normalization.

Cell culture

MLO-Y4 cells, kindly provided by Dr. Lynda Bonewald (Indiana University), were cultured on plates coated with 0.15 mg/mL type I collagen. The cells were grown in α-MEM medium supplemented with 2.5% fetal bovine serum, 2.5% bovine calf serum, and 1% penicillin/streptomycin at 37°C and 5% CO₂. For the experiments, cells were pre-incubated with Cx43(E2) antibody for 30 min, and then treated with 100 nM PTHrP for 30 min. PKA activity was determined using the PKA Colorimetric Activity Kit (EIAPKA, Invitrogen, Frederick, USA). For gene expression analysis, MLO-Y4 cells were pre-incubated with Cx43(E2) antibody (2 μg/mL) for 30 min, followed by incubation with 100 nM PTHrP for 48 h before RNA collection.

Live cell calcium imaging

MLO-Y4 cells were cultured at a total of 5× 10⁴ cells on a collagen-coated glass bottom 35 mm culture dish for 48 h before live cell Ca²⁺ imaging. Cells were loaded with a free Ca²⁺ indicator, Cal-520 (21130, AAT Bioquest Inc., Pleasanton, CA, USA), at a working concentration of 5 μM at 37°C for 2 h. Before imaging, Cx43(E2) antibody (2 μg/mL) was pre-incubated with MLO-Y4 cells for 30 min. The medium was then rinsed and replaced with Hank’s and HEPES buffer. Images were captured with a Zeiss LSM-810 confocal laser scanning microscope at 488 nm, using a 40× water objective. The entire imaging process lasted for 900 s, with a 2-s interval between each frame (512 × 512 pixels), and PTHrP was loaded at 180 s.

Intracellular pH measurement

Intracellular pH was measured using the pH-sensitive fluorescent dye 5-(and-6)-carboxy SNARF-1, AM (C1272, Thermo Fisher Scientific, Waltham, MA, USA). MLO-Y4 cells were pre-incubated with Cx43(E2) antibody (2 μg/mL) for 30 min, followed by incubation with 100 nM PTHrP for 48 h. After treatment, cells were washed in HBSS and loaded with 10 μM dye for 30 min at 37°C. Cellular fluorescence was excited at 514 nm, and the emission fluorescence was measured at both 580 nm and 640 nm using confocal imaging (Zeiss LSM 810, Jena, Germany). The pH value was calculated by determining the ratio of the fluorescence intensity at 580 nm to that at 640 nm. This ratio was then compared to a calibration curve derived from cells with known intracellular pH.⁵⁰ The calibration curve was established using cells with their intracellular pH fixed at known values ranging from 6.0 to 7.5 in a calibration solution containing 115 mM K⁺ and 10 μM nigericin.

Dye uptake assay

Cells were incubated with a mixture of 0.1 mM ethidium bromide (EtBr, MW 394 Da) and 1 mg/mL FITC-dextran (MW 10 kDa) for 5 min. EtBr was used as a tracer to detect HC activity, while FITC-dextran, which is too large to pass through HCs but is taken up by dying cells, served as a negative control. MLO-Y4 cells were treated with 100 nM PTHrP for 24 h and then incubated with Cx43(E2) antibody (2 μg/mL) for 30 min. After that, cells were rinsed five times with PBS and fixed with 2% paraformaldehyde for 10 min. At least six microphotographs of fluorescence fields were captured under a 20X fluorescent microscope (Keyence, BZ-X710, Osaka, Japan). For each image, the average intensity of EtBr fluorescence was measured and quantified from at least 30 random cells using ImageJ software (NIH, Bethesda, USA). Experiments were repeated 3 times, and the collected data were illustrated as the pixel mean in arbitrary units.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis

Statistical analysis was performed using GraphPad Prism8 statistics software (GraphPad, San Diego, USA). All data are presented as mean ± SEM. T test, One-way ANOVA and two-way ANOVA with Tukey test and multiple comparisons was used for statistical analysis. Asterisks indicate the degree of significant differences compared with the controls (*, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001).

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit polyclonal anti-MMP13	Abcam	Cat# ab39012; RRID: AB_776416
Rabbit polyclonal anti-Cathepsin K	Abcam	Cat# ab19027; RRID: AB_2261274
Rabbit polyclonal anti-Cx43(E2)	Siller-Jackson et al.25	N/A
Rabbit polyclonal anti-Cx43(CT)	Cherian et al.68	N/A
Chemicals, peptides, and recombinant proteins
Calcein	Sigma-Aldrich	Cat# C0875; CAS: 154071-48-4
Alizarin red	Sigma-Aldrich	Cat# A5533; CAS: 130-22-3
Alexa Fluor™ 488 Phalloidin	Thermo Fisher Scientific	Cat# A12379
MEM α	Thermo Fisher Scientific	Cat# 11900073
TRIzol Reagent	Molecular Research Center	Cat# TR118
Oxytocin	Sigma-Aldrich	Cat# O3251; CAS: 50-56-6
Picric acid	Sigma-Aldrich	Cat# 197378; CAS: 88-89-1
Direct red 80	Sigma-Aldrich	Cat# 365548; CAS: 2610-10-8
Silver nitrate	Thermo Fisher Scientific	Cat# S181-25; CAS: 7761-88-8
Sodium thiosulfate	Sigma-Aldrich	Cat# S-8503; CAS: 10102-17-7
Formic acid	Thermo Fisher Scientific	Cat# 423755000
Gelatin	Sigma-Aldrich	Cat# G9382
Formaldehyde	Thermo Fisher Scientific	Cat# 28908
Methyl methacrylate	Thermo Fisher Scientific	Cat# 127140010; CAS: 80-62-6
Ethylenediaminetetraacetic acid	Thermo Fisher Scientific	Cat# A107130B; CAS: 60-00-4
Type 1 collagenase	Worthington Biochemical Corporation	Cat# LS004196
Bovine serum albumin	Fisher Scientific	BP1600-100
Type I collagen	Corning	Cat# 354236
0.25% Trypsin-EDTA	Corning	Cat# 25053CI
Penicillin-Streptomycin	Corning	Cat# 30001CI
Fetal bovine serum	Corning	Cat# 35-015-CV
Bovine calf serum	Cytiva HyClone	Cat# SH30073.03H
Ethidium bromide	Amresco	Cat# 0492
FITC-dextran	Sigma-Aldrich	Cat# FD10S; CAS: 60842-46-8
PTH-related protein (1–34)	Bachem	Cat# H-6630.0500
Cal-520	AAT Bioquest Inc	Cat# 21130
Nigericin	Sigma-Aldrich	Cat# N7143; CAS: 28643-80-3
HEPES	Sigma-Aldrich	Cat# H4034; CAS: 7365-45-9
5-(and-6)-carboxy SNARF-1, AM	Thermo Fisher Scientific	Cat# C1272
Critical commercial assays
DAB Chromogen Kit	Vector Laboratories	Cat# SK-4100
In Situ Cell Death Detection Kit	Roche	Cat# 12156792910
1,25 (OH)2VD3 ELISA kit	MyBioSource	Cat# MBS2602146
QuantiChrom Calcium Assay kit	BioAssay Systems	Cat# DICA-500
PKA Colorimetric Activity Kit	Thermo Fisher Scientific	Cat# EIAPKA
High-capacity RNA-to-cDNA kit	Applied Biosystems	Cat# 43-889-50
Micro BCA protein kit	Thermo Fisher Scientific	Cat# 23225
Experimental models: Cell lines
Mouse: MLO-Y4 cells	Dr. Lynda Bonewald	RRID: CVCL_M098
Experimental models: Organisms/strains
Mouse: R76W	Xu et al.24	N/A
Mouse: Δ130-136	Xu et al.24	N/A
Oligonucleotides
Primers for qPCR	Table S6	N/A
Software and algorithms
ImageJ	National Institutes of Health	https://imagej.nih.gov/ij/
GraphPad Prism	GraphPad, Inc	https://www.graphpad.com/scientific-software/prism/
CTAn	Brüker	https://www.bruker.com/en/products-and-solutions/preclinical-imaging/micro-ct/3d-suite-software.html
Other
Keyence imaging system	Keyence	BZ-X710
Confocal imaging system	Zeiss	LSM 810
Dual-energy X-ray absorptiometry densitometer	GE Medical Systems	Lunar PIXImus
Brüker microCT scanner	Brüker	SkyScan 1172
Micromechanical testing system	Biomomentum	Mach-1 V500CST
Wafering saw	PACE Technologies	PICO 155

Highlights.

• Cx43 hemichannels are key mediators for lactation-induced perilacunar-canalicular remodeling

• PTHrP activates osteocytic Cx43 hemichannels, leading to calcium influx and PKA activation

• Blocking Cx43 hemichannels inhibits PTHrP-mediated osteocyte acidification

• Impairment of Cx43 hemichannels suppresses bone accrual during post-lactation recovery