CFTR mediates Cl^- transport in osteocytes to sustain cell viability and skeletal homeostasis

Abstract

Osteocytes are long-lived with underlying mechanisms largely unknown. Here, we report that osteocyte-specific knockout of cystic fibrosis transmembrane conductance regulator (CFTR) results in excessive osteocyte death, proinflammatory cytokine surge, osteoclast overactivation and bone formation impairment leading to bone loss in adult mice. Consistently in MLO‑Y4 osteocyte‑line, CFTR-knockout causes progressive cell death, which is reversed by CFTR overexpression or medium replenishment. A massive proinflammatory osteocyte secretome is evoked by CFTR-knockout, which deteriorates wild-type osteocytes, inhibits osteogenic differentiation, while robustly stimulates osteoclastogenic differentiation in vitro. Patch-clamp/Cl^--imaging verifies CFTR to mediate Cl^- transport in osteocytes, while Cl^--deprivation mimics CFTR-knockout to trigger transcriptomic/proteomic changes, cell stress and death. Additionally, osteocyte CFTR is downregulated in aged human bones; local delivery of CFTR via adenovirus or a CFTR modulator increases viable osteocytes and bone mass in aged mice. Together, the present study reveals a direct role of CFTR-mediated Cl^- transport in sustaining osteocyte viability and skeletal homeostasis.

Here, the authors show that the chloride channel CFTR keeps osteocytes alive and limits bone loss, revealing a new mechanism for maintaining long-lived bone cells and suggesting a potential strategy to protect bone health.

Introduction

Osteocytes are long-lived and the most abundant type of cells in the bone^1–3. Seemingly embedded in rigid mineralized bone matrix, osteocytes are in fact surrounded by a thin layer of liquid within the lacunar-canalicular space, which provides a route for osteocytes to communicate with other cells/systems⁴. Although electrolytes transport between osteocytes and lacunar-canalicular fluid remains largely unexplored, osteocyte-derived signaling molecules are known to effectively influence adjacent cells, bone tissues and even distant organ-systems^3,5,6. For instance, osteocytes secrete receptor activator of NFκB ligand (RANKL) and osteoprotegerin (OPG) to activate and inhibit osteoclastogenesis, respectively. Sclerostin derived from osteocytes reduces bone formation by inhibiting Wnt signaling pathway in osteoblasts^3,5. Osteocytes therefore play pivotal roles in bone remodeling and mineral homeostasis^7–9. Osteocyte senescence, necrosis and excessive death have been noted to cause bone loss^10–13. Antibodies against RANKL and sclerostin are being used/developed to treat osteoporosis¹⁴. However, the understanding of molecular mechanisms regulating osteocyte viability remains limited.

Cystic fibrosis transmembrane conductance regulator (CFTR), encoded by a single gene of the ATP-binding cassette transporter family, is a cAMP-activated anion channel known to be essential to epithelial fluid transport. Mutations of CFTR account for cystic fibrosis (CF), the most common genetic disease in Caucasians with airway epithelial dysfunction as the primary manifestation^15,16. Bone problems (e.g., osteopenia, osteoporosis) are present in CF^17,18, which is largely attributed to physical inactivity due to airway problems. Whether and how CFTR plays a direct role in osteocytes required for skeletal health remained unknown. In the present study, using a mouse model of Cre-LoxP-based osteocyte-specific knockout of CFTR, mice carrying ΔF508 (the most common loss-of-function mutation of CFTR), human bone samples and an osteocyte line with CRISPR/Cas9-based CFTR knockout in conjunction with technologies including patch-clamp, Cl^--imaging, bone functional analysis as well as RNA sequencing, mass-spectrometry and cytokine array, we revealed a direct role of CFTR in maintaining osteocyte viability for skeletal integrity and homeostasis. In addition, we tested the therapeutic potential of CFTR-containing adenovirus and CFTR-targeting pharmaceuticals¹⁹ in aged (>20-month-old) mice, which showed significant effect in sustaining osteocytes and benefiting aged bones.

Results

Osteocyte-specific knockout of CFTR induces bone loss in adult mice

To determine whether CFTR plays a direct role in osteocytes required for bone integrity, we built a transgenic mouse model where osteocytes were specifically removed of CFTR gene by crossing two transgenic mouse lines (Supplementary Fig. S1a), one inserted of an osteocyte-specific Dmp1-driven Cre recombinase gene (Dmp1-Cre) and the other of two LoxP sites flanking Cftr (Cftr^fl/fl, see methods). Western blotting and immunostaining (Supplementary Fig. S1b–e) for CFTR in the mouse bone tissues confirmed diminished expression of CFTR in osteocytes of the conditional (osteocyte-specific) knockout mice (cKO, Dmp1-Cre; Cftr^fl/fl) in comparison with the Cre-negative control (Ctrl, Cftr^fl/fl) mice. The cKO mice showed no significant difference in body weight from the Ctrl ones over a monitoring period from 4- to 24-week-old (Supplementary Fig. S2a). Bone lengths were also found similar between the two groups at 24-week-old (Supplementary Fig. S2b-c). We used micro-CT to examine the microarchitecture of different bones in these mice at 16- and 24-week-old. Although no statistical difference between the Ctrl and cKO (n = 5-10 mice each gender) was found in the femur at 16-week-old, cKO mice showed bone mass reduction from 16- to 24-week-old, compared to Ctrl mice (Supplementary Fig. S3). In particular, 24-week-old cKO mice of both genders exhibited significant reduction in volume, thickness and/or number of trabecular bones in lumbar vertebra, femoral and tibial bones, compared to age- and gender-matched Ctrl mice (n = 5–10 mice each gender, Fig. 1a–c, full data shown in Supplementary Figs. S4, S5 and S6). Cortical bones also showed some reduction in cKO mice but not as significant as trabecular bones at 24-week-old (Supplementary Figs. S5 and S6). Histological analysis confirmed substantial decreases in trabecular bone volume and thickness in cKO mice at 24-week-old (Fig. 1d). Additionally, compression tests suggested that vertebrae from cKO mice had much weaker strength compared to Ctrl mice (Fig. 1e). Three-point bending tests showed the maximal load resisted by the tibia or femur was also smaller in cKO mice compared to that in the Ctrl (Fig. 1f). These results demonstrate a phenotype of bone loss in adult mice triggered by CFTR knockout in osteocytes, confirming a direct role of CFTR in osteocytes required for skeletal health.

Fig. 1. Osteocyte-specific knockout of CFTR induces bone loss in adult mice.

a–c Micro-CT analysis of lumbar vertebra (a), femur (b) and tibia (c) bones in the conditional (osteocyte-specific) CFTR-knockout mice (cKO, Dmp1-Cre; Cftr^fl/fl) in comparison with the Cre-negative control (Ctrl, Cftr^fl/fl) mice at 24-week-old. Data from male mice are shown. Full data can be found in supplementary information. Upper: representative 2D or 3D construction images. L4 and L5: 4^th and 5^th lumbar vertebra. Lower: quantification of trabecular (Tb.) bones in bone volume (BV), tissue volume (TV), BV/TV, thickness (Tb.Th), number (Tb.N) and separation (Tb.Sp). *P < 0.05, **P < 0.01 by t-test. n = 5 or 6 mice. d Hematoxylin and eosin (H&E) stained femoral tissues from cKO or Ctrl mice at 24-week-old. e–f Compression test of isolated lumbar vertebra (e) and three-point bending test of tibiae or femora (f) from 24-week-old cKO and Ctrl mice. Left: representative curves of loading force-displacement. Right: summary of maximal compressing/bending forces. *P < 0.05, **P < 0.01, ***P < 0.001 by t-test. n = 4 to 10 mice. Scale bars = 500 (a–c) and 100 (d) μm. Source data are provided as a Source Data file.

CFTR is functionally expressed to mediate Cl^- transport in osteocytes

To understand the role of CFTR in osteocytes, we next examined MLO-Y4, a commonly used mouse osteocyte line. mRNA expression of CFTR was detected in MLO-Y4, the sequence of which was confirmed to align with mouse CFTR gene (Supplementary Fig. S7a). Western blot detected a band of about 160 kDa, suggesting protein expression of CFTR in MLO-Y4 (Supplementary Fig. S7b). We next performed patch-clamp in MLO-Y4 to detect possible CFTR channel activity. In whole-cell mode and at the holding voltage of −60 mV, the addition of Forskolin (10 µM, an activator of adelyne cyclase) into the bath for 5–10 min induced an inward current in MLO-Y4 cells, which could be blocked quickly by subsequent addition of a selective inhibitor of CFTR, Inh172 (10 µM) (Fig. 2a). A series of voltage stimuli ranging from −100 to +80 mV were applied onto the cells, which elicited time- and voltage-independent whole-cell currents that were enhanced by Forskolin and blocked by Inh172 (Fig. 2b), indicating CFTR channel activities. Additionally, we used a Cl^- sensitive fluorescence dye (MQAE) to monitor intracellular Cl^- level ([Cl^-]i) in MLO-Y4 cells, which showed that the inhibition of CFTR by Inh172 (1-10 µM) induced an elevation of [Cl^-]i (as indicated by MQAE intensity reduction, Fig. 2c, d), suggesting CFTR in mediating Cl^- efflux in MLO-Y4 cells. To confirm these observations, CRISPR-Cas9 based technology was used to create an MLO-Y4 line with CFTR gene knocked out (CFTR-KO, Supplementary Fig. S8). Sequencing results showed that the exon 4 of CFTR gene was destroyed in the CFTR-KO line (Fig. 2e). Western blotting confirmed that the CFTR band was largely diminished in the CFTR-KO line (Fig. 2f). In addition, the Forskolin-activated and Inh172-sensitive whole-cell currents observed in wild-type cells were no longer detected in the KO line by patch-clamp (Fig. 2g); nor was the Inh172-sensitive [Cl^-]i change shown in the KO line (Fig. 2h). These results therefore revealed CFTR function to mediate Cl^- transport in osteocytes.

Fig. 2. CFTR is functionally expressed to mediate Cl^- transport in osteocytes.

a Representative patch-clamp recording of whole-cell currents (holding voltage at −60 mV) in an MLO-Y4 cell when Forskolin (an activator of adenyl cyclase, 10 µM) and subsequently a selective CFTR inhibitor (Inh172, 10 µM) were added into the bath. Double slashes are recording breaks. b Representative whole-cell currents (left) elicited from −100 to +80 mV in an MLO-Y4 cell before (Control) and after the addition of Forskolin or Inh172. Right: Corresponding current-voltage curves. n = 3 cells. c Representative images (left) of MLO-Y4 cells loaded with MQAE, a fluorescent dye sensitive to Cl^- concentration, and time-course measurement (right) of MQAE intensity in the cells before and after Inh172 was added into the bath. d Calibration between change in intracellular Cl^- concentration (Δ[Cl^-]_i) and change in MAQE intensity (left), and quantification of Δ[Cl^-]_i induced by Inh172 (1–10 µM) or DMSO as vehicle control (right). *P < 0.05, ***P < 0.001 by One-way ANOVA. n (number of cells) is shown in each group. e Sequencing for exon 4 of mouse cftr in MLO-Y4 cells with CFTR knockout by CRISPR-Cas9 (CFTR-KO), in comparison with wild-type (WT) mouse cftr (NM_021050). f Western blot for CFTR in WT or CFTR-KO MLO-Y4 cells. Actin was used as the loading control. g Patch-clamp experiments (as described in a and b) in CFTR-KO cells. n = 3 cells. h Representative MQAE images taken before (-) and after Inh172 addition with time-course changes in MQAE intensity and quantification of Inh172-induced Δ[Cl^-]_i in WT or CFTR-KO MLO-Y4 cells. *P < 0.05 by t-test. n = 20-64 cells. Pseudo-colors from purple to red of the images (in c and h) indicate MQAE intensity from low to high. Scale bars = 50 µm. Data are mean ± s.e.m. Source data are provided as a Source Data file.

Knockout of CFTR deteriorates osteocyte viability

Apart from the observed defective Cl^- transport, we quickly detected that the CFTR-KO MLO-Y4 line grew much slower than the wild-type line (Fig. 3a). At 48 h after seeded, the CFTR-KO line exhibited, as compared to the wild-type line, increased number of cells positively stained for beta-galactosidase (β-gal, a senescence-associated enzyme, Fig. 3b), more cells of multinucleated morphology (0.7 ± 0.1% in wild-type versus 5.0 ± 0.9% in CFTR-KO, Fig. 3c), increases in mRNA level of genes associated with cellular senescence including Serpinb2, Tbx3, Igfbp3, Ets2, Plau, Nox4, Map2k6, and Cdkn2b (Fig. 3d), as well as a higher level of intracellular reactive oxygen species (ROS, Fig. 3e). Osteocyte key genes Dmp1 and Phex were significantly downregulated in the CFTR-KO line, although sclerostin (Sost) and fibroblast growth factor 23 (Fgf23) were not affected at mRNA level (Fig. 3f). On day 7-8 after seeded, while the wild-type cells continued to grow, the number of viable CFTR-KO cells substantially dropped (Fig. 3a), suggesting cell death in the KO line as the culture was prolonged. TUNEL (Fig. 3g) and Annexin V/Propidium iodide (Fig. 3h) assays revealed that although minimal apoptosis/necrosis was detected in both wild-type and CFTR-KO lines on day 3, the number of apoptotic and necrotic cells drastically increased in the CFTR-KO line as the culture progressed to day 7.

Fig. 3. Knockout of CFTR deteriorates osteocyte viability.

a Cell viability test in WT or CFTR-KO MLO-Y4 cells in 8 days of culture. Data are normalized to those on day 1. n = 6 wells of cells. b–c β-galactosidase (β-gal) staining (b, blue) and fluorescence staining (c) for actin (green) and nucleus (blue) with quantifications. n = 4 wells of cells. d–e Quantitative PCR (qPCR) for genes associated with cellular senescence (d) and reactive oxygen species (ROS) assay (e) in WT CFTR-KO MLO-Y4 cells 24 h after seeding. n = 3-6 reactions. qPCR was repeated for 2–3 times. f qPCR for genes associated with osteocyte functions in WT or CFTR-KO MLO-Y4 cells 24 h after seeding. n = 6 reactions from two batches of experiments. g–h Representative fluorescence images showing TUNEL labeling (g, green) and flow cytometry detection of annexin V and propidium iodide (PI) labeling (h) for apoptotic/necrotic cells on day 3 and day 7 after seeding WT or CFTR-KO MLO-Y4 cells. n = 3 assays. Data are mean ± s.e.m., *P < 0.05, **P < 0.01, ***P < 0.001 by two-way ANOVA (a and h) or t-test (b–f). Scale bars = 100 µm. Source data are provided as a Source Data file.

To confirm the loss of CFTR as the determinant for these severe defects observed in the CFTR-KO line, we tried to rescue the CFTR-KO line by introducing back the CFTR gene. Overexpressing CFTR by transfecting the CFTR-KO cells with plasmids containing human CFTR gene (pCFTR, Fig. 4a) recovered CFTR expression (Fig. 4b) and Inh172-sensitive Cl^- efflux activities (Fig. 4c), and, importantly, reversed the viability of the cells by 29.1 ± 2.0 % on day 8 post transfection (Fig. 4d), and reduced ROS level by 32.1 ± 6.5 % (Fig. 4e), as compared to the controls transfected with vector plasmids (pVector). In addition, TUNEL-positive cells were observed to be largely reduced after the CFTR overexpression (Fig. 4f). Together, these results identified CFTR as an essential factor for osteocyte viability and sustainability.

Fig. 4. Overexpression of CFTR restores osteocyte viability.

a–f PCR detection of mRNA expression of human CFTR gene (hCFTR, a), western blot for CFTR (b), MQAE detection of 172inh-sensitive Cl^- activity (c), MMT assay (d), ROS test (e) and TUNEL labeling (f) in CFTR-KO MLO-Y4 cells after transfected with plasmids containing human CFTR (pCFTR) or vector plasmids (pVector). Data are mean ± s.e.m., *P < 0.05, **P < 0.01, ***P < 0.001 by t-test (c, e) or two-way ANOVA (d). n = 34–39 cells (c), 6 wells of cells (d) and 10 wells of cells (e). Scale bars = 20 µm. Source data are provided as a Source Data file.

Knockout of CFTR provokes osteocyte proinflammatory secretome

We then asked how CFTR knockout can affect MLO-Y4 cell viability. Interestingly, despite drastic cell death observed in prolonged cultures, the CFTR-KO line survived and could be sub-cultured for over 30 passages. Daily replenishment of the culture medium substantially restored the viability of CFTR-KO line by 86 ± 12% in 6 days (Fig. 5a), compared to non-replenished cells, suggesting that CFTR-KO cells might secrete substances harmful to themselves. Indeed, treating the wild-type MLO-Y4 cells with a conditioned medium collected 72 h after incubation with the CFTR-KO cells drastically decreased the number of viable wild-type cells (Fig. 5b). We then examined the conditioned medium with a cytokine array, which showed that the levels of a number of proinflammatory cytokines/chemokines including KC, IL-6, IP-10, MIP-2, G-CSF, RANTES, MIP-1α, GM-CSF, I-309, TNF-α, IL-3 and Eotaxin were significantly increased in the medium from the CFTR-KO line as compared to that from the wild-type line (Fig. 5c and Supplementary Fig. S9). Additionally, CFTR-KO cells showed high production of RANKL versus OPG (Fig. 5d), and abundant activation of NFκB (Fig. 5e), in comparison with wild-type cells. We then applied the conditioned medium to the induction of in vitro osteogenic and osteoclastogenic differentiation (see methods) from primary cultures of mouse bone marrow cells (see method). Results showed that the conditioned medium from the CFTR-KO line significantly retarded the induced osteogenic differentiation from bone marrow cells (Fig. 5f), as compared to the control medium from the wild-type cells. Whereas, the CFTR-KO conditioned medium robustly stimulated osteoclastogenic differentiation as indicated by giant TRAP (tartrate-resistant acid phosphate)-positive multinuclear cells (Fig. 5g), while very few TRAP-positive cells were present in cultures treated with the medium from the wild-type line (Fig. 5g). Similar results were achieved using MC3T3-E1 (an osteoblast line) and RAW 264.7 (a pre-osteoclast line) cells (Supplementary Fig. S10). These results suggested that the loss of CFTR induced a proinflammatory osteocyte secretome, which is detrimental to osteocyte survival and osteogenic differentiation, while over-activating osteoclasts.

Fig. 5. Knockout of CFTR in osteocytes induces a proinflammatory secretome.

a Cell viability test in WT or CFTR-KO MLO-Y4 cells with and without daily replenishing the culture medium. n = 3 wells of cells. b WT MLO-Y4 cells were cultured in a conditioned medium (collected after 72 h incubation with WT or CFTR-KO cells, CM_WT or CM_CFTR-KO, see methods) and tested for viability for 6 days. n = 3 wells of cells. c Cytokine array with quantification (right) of significantly changed (P < 0.05, t-test) cytokines in the conditioned medium. Experiments were repeated twice. Full data can be found in supplementary information. d ELISA measurements of RANKL and OPG levels in the conditioned medium. n = 6 wells of cells. Data are normalized to the average level in CM_WT. e Western blots for pNFκB and NFκB in WT or CFTR-KO MLO-Y4 cells. GAPDH was used as a loading control. n = 3. f Alizarin red S (ARS) staining with quantification (right) in mouse bone marrow cell cultures 2 weeks after incubation with the conditioned medium (CM_WT or CM_CFTR-KO) in the presence of osteogenic reagents (see method). n = 6 wells of cells. g Representative photographs of tartrate-resistant acid phosphate (TRAP) staining indicating osteoclastogenic differentiation in mouse bone marrow cell cultures 7 days after incubation with CM_WT or CM_CFTR-KO (see method). Osteoclasts are defined by TRAP-positive staining and multinuclear morphology. Quantification of osteoclast numbers is shown (right). n = 6 wells of cells. Data are mean ± s.e.m., *P < 0.05, **P < 0.01, ***P < 0.001 by two-way ANOVA (a–c) or t-test (d–g). Scale bars = 100 µm (f, g). Source data are provided as a Source Data file.

Cl^- disturbance mimics CFTR knockout to cause osteocyte stress and death

To further understand the mechanism underlying CFTR’s role in maintaining osteocyte viability, we performed a transcriptome analysis by RNA sequencing, which showed that at 48 h after seeded, the CFTR-KO cells, as compared to the wild-type, exhibited differential expression of a large number of genes (q < 0.05), 1849 of which changed by over 2 folds. KEGG pathway enrichment analysis of the differentially expressed genes suggested alterations in signaling pathways related to cell cycle, cellular senescence, apoptosis and necroptosis as well as inflammatory pathways and osteoclast differentiation (Supplementary Fig. S11, Fig. 6a, b). Proteomic analysis of the cells confirmed similar changes at protein expression level suggesting high cellular stress and inflammation caused by CFTR knockout (Fig. 6c).

Fig. 6. Cl^- disturbance mimics CFTR knockout to cause osteocyte stress and death.

a KEGG pathway enrichment analysis of differentially expressed genes (DEGs) between WT and CFTR-KO MLO-Y4 cells (seeded for 48 h) as detected by RNA sequencing. n = 3 wells of cells. b Heatmap showing the DEGs between WT or CFTR-KO cells belonging to pathways of cellular senescence and osteoclast differentiation. Row Z-scores are used. n = 3 wells of cells. c, d Mass spectrometry detection of proteins differentially expressed in WT MLO-Y4 cells after 2 h incubation with a Cl^--free solution in comparison with cells incubated in a normal Margo’s solution (Ctrl, see methods), or in CFTR-KO cells in comparison with WT cells. KEGG pathway enrichment (c) and heatmap (d) show similar profiles of altered pathways and proteins caused by Cl^- deprivation and CFTR-KO. n = 3 wells of cells. e, f ROS assay (e) and cell viability test (f) in WT MLO-Y4 cells treated with a Ctrl, low Cl^- or Cl^--free solution for up to 36 h. n = 3–9 wells of cells. Data are mean ± s.e.m., ***P < 0.001 by two-way ANOVA. Source data are provided as a Source Data file.

Given the observed Cl^- channel activities of CFTR in MLO-Y4 cells (Fig. 2), we wondered if Cl^- disturbance could underlie the effect of CFTR knockout. Indeed, treating the wild-type MLO-Y4 cells with a Cl^- free solution (see methods) to mimic Cl^- transport disturbance, induced, in 2 h, alterations in the expression levels of proteins associated with cell growth/death and inflammatory pathways (Fig. 6c, d and Supplementary Fig. S12), in comparison with the cells treated in normal Cl^- solution as the control (Ctrl). Interestingly, such a Cl^--disturbance-induced proteomic profile alteration showed high similarity with the effect of CFTR knockout (Fig. 6c, d and Supplementary Fig. S12). We then incubated wild-type MLO-Y4 cells with such a Cl^--free solution for up to 36 h, which resulted in significantly increased ROS level (Fig. 6e) and accelerated cell death compared to the Ctrl (Fig. 6f), mimicking the effect of CFTR knockout. Together, these results suggested that disturbed Cl^- transport across osteocyte plasma membrane as a result of CFTR deficiency triggered cellular stress and inflammation leading to osteocyte death.

Defective CFTR triggers excessive osteocyte death and bone resorption in vivo

We then asked whether the CFTR knockout-provoked osteocyte stress, inflammation and death observed in vitro could be the reason for the phenotype of progressive bone loss in the cKO mice in vivo. Indeed, staining for TRAP in the mouse tibial and femoral tissues showed strikingly stronger labeling in the cKO bone tissues (both male and female, at 24-week-old) as compared to that of the Ctrl (Fig. 7a), indicating overwhelmed activation of osteoclasts in the cKO mice. In line with this, higher NFκB activation was detected in femoral tissues from cKO mice, compared to that of the Ctrl (Supplementary Fig. S13). In addition, the cKO mice exhibited higher serum levels of CTX-1 (C-terminal telopeptide of type I collagen) versus PINP (N-terminal propeptide of type I procollagen), compared to Ctrl mice (Fig. 7b), indicating again high bone resorption activities. We also performed a cytokine array test of the serum samples from the mice, which showed that most of the detectable cytokines were elevated in cKO mice. Significantly increased (cKO versus Ctrl) were KC (keratinocyte-derived cytokine, or CXCL-1, by 11.5 folds), BLC (B lymphocyte chemoattractant, or CXCL-13, by 6.8 folds), SDF-1 (stromal cell-derived factor 1, or CXCL-12, by 2.45 folds), C5/C5a (complement component 5/5a, by 1.5 folds) (Fig. 7c) Moreover, counting osteocytes and empty lacunae as well as TUNEL assay (see methods) showed significant reduction of osteocyte numbers and increased osteocyte death in cKO mice, compared to age- and gender-matched Ctrl mice (Fig. 7d, e). Measurement of bone formation rate from 16- to 18-week-old confirmed retarded mineral apposition rate and bone formation in cKO mice compared to the Ctrl (Fig. 7f). Furthermore, another mouse model carrying the loss-of-function mutation (ΔF508) of CFTR showed similar osteocyte death and shift to bone resorption in bone tissues (Supplementary Fig. S14). These in vivo results are therefore in line with in vitro observations suggesting essential role of osteocyte CFTR in maintaining osteocyte viability and the equilibrium of bone absorption/formation.

Fig. 7. Osteocyte-specific knockout of CFTR provokes inflammation and osteocyte death in vivo.

a Representative photographs of TRAP with quantification of osteoclast numbers in femoral and tibial tissues from 24-week-old cKO or Ctrl mice. n = 7 or 8 mice. b ELISA measurements of CTX-I, and P1NP in the serum from 24-week-old cKO or Ctrl mice. n = 5 mice. c Cytokine array test of serum samples from cKO or Ctrl mice at 24-week-old. Samples were from 4 mice of each group (cKO or Ctrl). d Representative H&E images and quantifications of empty lacuna and osteocyte number in the tibia or femur cortical bones from 24-week-old cKO or Ctrl mice. n = 9 mice. Arrows indicate empty lacunae. e TUNEL labeling (red) with quantification in tibial tissues from 24-week-old cKO or Ctrl mice. n = 4 or 5 mice. f Representative fluorescence images of calcein green/xylenol orange labeling in cortical and trabecular bones of the femur/tibia from cKO and Ctrl mice (n = 6). The cortical (Ct.) or trabecular (Tb.) mineral apposition rate (MAR) was calculated as the distance (yellow mark) between midlines of the two labels divided by the interval days. Bone formation rate (BFR) was calculated as MAR × (mineralizing surface/bone surface). Data are mean ± s.e.m., *P < 0.05, **P < 0.01, ***P < 0.001 by two-way ANOVA (a, d) or t-test (b, c, e, f). Scale bars = 100 µm. Source data are provided as a Source Data file.

Enhancing CFTR sustains osteocytes and alleviates osteoporosis in aged mice

We next explored possible implication of osteocyte CFTR in treating bone diseases. We collected human bone samples from orthopedic surgeries (subject data shown in Supplementary Table 2). CFTR expression in osteocytes, particularly its plasma membrane expression, was detected by immunostaining in the human bone tissues, which was, however, nearly absent in osteoporotic bones from elderly subjects (Fig. 8a). We then used aged (>20-month-old) mice as an osteoporotic model. Our micro-CT analysis of mouse femoral tissues confirmed low bone mass, especially of trabecular bones, in both male and female mice of >20-month-old. In these mice, we did local injection of adenovirus carrying human CFTR gene (Adv_hCFTR) into one femur of each aged mouse with the other femur of the mouse injected with empty vector (Adv_Vector) as the control (see methods). In 5 weeks, the Adv_hCFTR-injected femoral tissues showed the mRNA expression of the delivered human CFTR gene (Fig. 8b) as well as significant upregulation of osteocyte genes, compared to Adv_Vector-injected ones (Fig. 8c). Staining for CFTR showed fewer CFTR-positive osteocytes in the aged mouse bones (Fig. 8d), which was substantially reversed by the Adv_hCFTR treatment (Fig. 8d). Importantly, two doses of such Adv-hCFTR injection significantly increased, in 10 weeks, bone volume (measured by micro-CT) of trabecular bones in the femora, compared to Adv_Vector-injected controls (Fig. 8e). Beneficial effects were also observed in cortical bones of middle and distal portions of the injected femur (Supplementary Fig. S15). The number of osteocytes in cortical bones was increased while ratio of empty lacuna was reduced in Adv_hCFTR-treated bones compared to Adv_Vector-injected (Fig. 8f), suggesting the Adv_hCFTR treatment in promoting osteocyte viability. Additionally, we used VX809, a pharmaceutical drug known to enhance CFTR¹⁹. After confirming in MLO-Y4 cells that VX809 treatment enhanced CFTR protein expression and Cl^- channel function (Supplementary Fig. S16a–b), we performed local injections of VX809 (every other day for 4 weeks) near L4 lumbar vertebrae of aged mice, which, with no visible side-effect on body weight and general health (Supplementary Fig. S16c), significantly increased BV, TV and BV/TV and trabecular thickness of the vertebrae (Fig. 8g, Supplementary Fig. S16d), as compared to the vehicle injected controls. Together, these results suggested the therapeutic potential of enhancing CFTR for osteocyte sustainability to benefit aged bones.

Fig. 8. Enhancing CFTR benefits osteoporotic bones.

a Representative confocal images of immunofluorescence staining for CFTR (red) in human bone tissue samples obtained from subjects during orthopedic surgeries, with quantifications of osteocyte CFTR expression. Nuclei were labeled in blue with DAPI. Analyses included CFTR expression versus age across all subjects (n = 13), and a group comparison between subjects aged 34-56 years (n = 7) and 75–84 years (n = 6). b, c qPCR detection of mRNA expression of human (hCFTR) and mouse CFTR (mCFTR) (b) and bone cell genes (c) in mouse femoral tissues 5 weeks after treated with adenovirus carrying human CFTR gene (Adv_hCFTR) or control adenovirus (Adv-Vector) (see method). n = 3 or 4 mice. d Immunohistochemistry staining for CFTR (brown) in the femoral tissues from mice aged >20-month-old treated with Adv-Vector or Adv_hCFTR. e Micro-CT measurement of trabecular bones (Tb.) in mouse femora 10 weeks after the two injections of Adv_hCFTR or Adv-Vector in male and female mice aged >20-month-old. Adv_hCFTR and Adv-Vector were injected into the right and left femora, respectively, in each mouse. n = 7 mice (4 females and 3 males). f Quantification of empty lacunae and osteocytes in the femur tissues 10 weeks after treatment with Adv_hCFTR or Adv-Vector. Arrows indicate empty lacunae. n = 3 mice. g Micro-CT measurement of L4 lumbar vertebra of 28-month-old female mice after 4 weeks treatment with VX809 (10 nmole per injection site, every 2 days) or DMSO. n = 8 mice. Data are mean ± s.e.m., *P < 0.05, **P < 0.01, ***P < 0.001 by Pearson test (a) or t-test (a, c, e–g). Scale bars = 5 (a), 50 (d), 25 (f), and 500 µm (e, g). Source data are provided as a Source Data file.

Discussion

In summary, the present study has demonstrated, for the first time, that CFTR is functionally expressed in osteocytes to mediate Cl^- transport. Cl^- disturbance due to CFTR deficiency causes osteocyte stress evoking a proinflammatory osteocyte secretome, which deteriorates osteocytes, inhibits osteoblast differentiation, while triggers overwhelmed activation of osteoclasts, leading to disruption of skeletal homeostasis and consequent bone loss; enhancing CFTR corrects Cl^- transport and improves osteocyte viability to benefit bone health (Fig. 9).

Fig. 9. Schematic model for the role of osteocyte CFTR in skeletal homeostasis.

CFTR is functionally expressed in osteocytes to mediate Cl^- transport between osteocytes and lacuna-canalicular fluid. Loss of CFTR in osteocytes leads to Cl^- disturbance, evoking cellular stress, and massive secretion of proinflammatory factors, which deteriorates osteocyte viability, inhibits osteoblast differentiation, while triggers excessive activation of osteoclasts, resulting in disruption of skeletal homeostasis and bone loss. Enhancing CFTR boosts osteocyte genes, reserves osteocyte viability and thus improves bone health.

Osteocytes sustain a long lifespan in the body, for which the present study has revealed a previously undefined molecular mechanism that requires CFTR. Clearly, MLO-Y4 cells need CFTR to survive longer in cultures. We screened more than 50 clones of cells after the CRISPR-Cas9 treatment to knock out CFTR, only one of which was found successful. Such a low rate suggests that most of the cells knocked out of CFTR might have died early. In the clone that survived, the cellular stress level is obviously high as evident by results of multiple assessments including functional ones (e.g., MTT) as well as gene expression analysis (e.g., qPCR and RNAseq). Importantly, delivery back of CFTR gene alleviated the stress to a significant extent and improved the survival of the knockout line, determining the role of CFTR itself in MLO-Y4 sustainability. Consistently, osteocytes were found affected in vivo by CFTR knockout, as evident by the increased number of empty lacunae and level of osteocyte death in the osteocyte-specific knockout (cKO) and/or ΔF508 mice. Moreover, both in vitro and in vivo data supported that deteriorated osteocyte viability by CFTR knockout can affect surrounding cells and disturb skeletal homeostasis, which cumulatively lead to bone loss. Of note, different from the whole-body mutant ΔF508 mice^18,20, our cKO mice did not show severe bone problems during the growing/developmental period (i.e., before 16-week-old). Since CFTR is known to be expressed by osteoblasts and important to osteoblast differentiation²¹, massive bone formation driven by osteoblasts can override the effect on osteocytes in the cKO mice before 16-week-old. Additionally, we demonstrate the bone problems in the cKO to be caused by a proinflammatory osteocyte secretome (discussed below). The active circulation system in vivo may timely remove what osteocytes secrete, which may alleviate its detrimental effect in vivo. Skeletal growth in mice typically reaches a plateau from 16-week-old onward, which sustains till about 18-month-old when age-related decline begins. Due to escalated cellular stress/inflammation and early death of osteocytes, the cKO mice in the present study exhibited a decline in bone mass between 16- to 24-week-old, suggesting skeletal premature aging induced by CFTR deficiency in osteocytes. In this sense, the present study has provided a new etiology of bone problems that originates from osteocyte dysfunction. Our analysis of limited human bone samples from orthopedic surgeries showed some association of clinical osteoporosis with deficiency in osteocyte CFTR, although it needs verification in the future using a larger number of human subjects with healthy ones included. Given that over 2000 mutations of CFTR have been identified in humans²² and most of them are mild not causing CF, it may be worth investigating whether CFTR mutations could underlie idiopathic bone diseases in the future.

In understanding how CFTR deficiency deteriorates osteocyte viability, we provided evidence to suggest the involvement of proinflammatory osteocyte secretome. Both in vitro and in vivo data of the present study showed osteocytes removed of CFTR to be capable of abundantly releasing cytokines/chemokines (e.g., KC, BLC, MIP-2, G-CSF, MIP-1α, GM-CSF). Such a provoked proinflammatory secretome is suggested to be the reason/upstream for other degenerative changes of the osteocytes (e.g., Phex, Dmp1 downregulation), given that (1) the culture medium replenishment restored cell viability by nearly 90%, and (2) the proinflammatory medium caused healthy osteocytes to die. This is, in fact, consistent with the increasingly recognized role of chronic inflammation in driving aging and other diseases²³. Since osteocytes account for over 95% of all bone cells and a substantial portion of all cells in the body²⁴, the inflammation caused by osteocyte dysfunction could be significant to a severe extent. Indeed, the present study detected cytokines surge in the serum of cKO mice, suggesting that it is not only deteriorating the bone but also possibly affecting other parts of the body through the circulation system. The global impact of osteocyte dysfunction caused by CFTR deficiency may be worth further investigation.

For recent decades, CFTR has been demonstrated to play a regulatory role beyond its channel function, that is to interact with multiple cellular or intercellular signaling pathways^20,25,26. The present study, however, suggests disturbance in Cl^- transport induced by CFTR-deficiency to be a key reason for the cellular stress we observed in osteocytes, since eliminating extracellular Cl^- triggered quickly proteomic changes, surges in ROS level and cell death, similar to the effect of CFTR knockout. Cl^- is known to be important to multiple cellular functions such as fluid transport, membrane potential shift and volume changes etc^27–31. Intracellular level of Cl^- is typically low and believed to be actively transported out for the cell to accommodate other negatively charged constituents (nucleic acids, metabolites) while maintaining cytosolic electroneutrality³². Cl^- disturbance is documented to associate with infectious airway diseases³³. The present study interestingly demonstrates CFTR to be crucial to osteocyte secretion of Cl^-. Very little is known about the lacunar-canalicular fluid given its low accessibility, although its Cl^- concentration was reported to be about 130 mM³⁴. The present study shows that in a bath solution containing about 140 mM Cl^-, CFTR mediates Cl^- efflux from MLO-Y4 cells, as indicated by the observed Inh172-induced rise in intracellular Cl^- in MLO-Y4 cells. It thus suggests CFTR channel in MLO-Y4 cells to be open, at least to some degree, and electrochemical gradient of Cl^- across MLO-Y4 cell membrane to be outward under the experimental condition. However, we learned from nervous system that Cl^- channels can be either excitatory or inhibitory depending on the electrochemical gradient of Cl^- across the neuron³⁵. CFTR should be able to contribute to either Cl^- efflux or influx in osteocytes as well, given a certain context that sets Cl^- electrochemical gradient of osteocytes in vivo. It should be noted that Cl^- disturbance or CFTR-knockout triggers massive transcriptomic/proteomic changes in genes and pathways, which suggests the underlying mechanism not through a single molecular target or pathway. Ultimate elucidation of the mechanisms for Cl^- homeostasis across osteocytes will require evaluation of intracellular electrical potentials, pH, osmotic pressure, negatively and positively charged proteins, lipids, nucleic acids and other small molecules, as well as extracellular matrixes and electrolytes, which awaits further investigation. Interestingly, similar to epithelial CFTR, our data demonstrate CFTR in MLO-Y4 cells to be activated by cAMP rise (i.e., by Forskolin) too, which suggests the capacity of osteocyte CFTR to respond to stimuli from neurotransmitters/hormones/paracrine factors that can evoke cAMP signaling pathway, and thus to transduce the extracellular signals into osteocyte behaviors under different physiological/pathological conditions. It will be interesting to study in the future such roles of CFTR and its possible cooperation with other ion channels/transporters^36,37 in achieving Cl^- and other electrolytes homeostasis surrounding osteocytes.

Intriguingly, local injection of CFTR expressing virus or CFTR enhancers exerted a beneficial effect on aged bones in wild-type mice. The implication of the present findings is therefore beyond CFTR mutation-associated bone problems, potentially reaching a non-CF general population with bone problems. Currently, treatment methods for osteoporosis are primarily antiresorptive drugs, while bone formation promoting strategies are lacking. Although antibodies against osteocyte-derived factors (RANKL and sclerostin) are used, less attention has been paid to strengthening osteocyte survival and integrity to counteract skeletal degeneration, despite osteocytes’ abundance in the bone. The present data suggest that the overexpression of CFTR may mitigate bone loss through strengthening osteocytes since significantly increased viable osteocyte numbers and boosted osteocyte marker genes (i.e., DMP1, PHEX, MEPE, and OPG) were observed after the treatment, while osteoblastic marker genes are less affected. It should also be noted that bone fracture is a frequent problem encountered by people with osteoporosis. The presently observed beneficial effect of locally injected CFTR-enhancing drugs/virus on aged bones in mice might suggest it as a possible local treatment method to assist in the healing of osteoporotic fracture. In the present study, we only did a proof-of-concept study testing the effect of a CFTR modulator (Lumacaftor/VX809) using a single dosage. VX809 alone is reported to be limited of its effectiveness or specificity^38,39, while newly developed combined CFTR modulators/potentiators treatment has emerged as highly effective to alleviate CF symptoms^40,41. More systematic and translational investigations can be done in the future to validate the effect of CFTR-targeting drugs or methods in treating related bone diseases.

Methods

Animals

The Cftr^fl/fl transgenic mouse line was constructed in Cyagen Biosciences Inc. (Guangzhou, China) by inserting two loxP sequences (5’-ATA ACTTCGTATAGCATACATTATACGAAGTTAT-3’) to flank the exon 5 to 6 of mouse CFTR gene (ENSMUST00000045706) by CRISPR-Cas9 in C57BL/6 mice. The size of intron 4 for 5’-loxP site insertion is 6305 bp, and the size of intron 6 for 3’-loxP site insertion is 856 bp. The size of effective knockout region is about 2097 bp. Stable heterozygous mice (Cftr^fl/wt) were obtained by crossing the homozygous mice with C57BL/6 wild-type mice. The Cftr^fl/wt mice were then crossed with the Dmp1-Cre mouse line (B6N.FVB-Tg (Dmp1-cre)1Jqfe/BwdJ) purchased from the Jackson laboratory (USA). The CFTR mutant (ΔF508) transgenic mice (Cftr^tm1Kth) were purchased from Laboratory Animal Service Center of the Chinese University of Hong Kong. Genotyping of the mice was done using the Phire Tissue Direct PCR Master Mix Kit (Thermo Scientific, F170S) with primers for ΔF508 (5’-TTCAAGCCCAAGCTTTCGCGAG-3’, 5’-CTCCCTTCTTCTAGTCACAACCG-3’ and 5’-CATCTTGATAGAGCCACGGTGC-3’), Cftr-flox (5’-AGATGACTGTCTACCCTACTCCTT-3’, 3’-CACATTAGACATGAAAATGGAGGC-5’) and Dmp1-Cre (5’-TTGCCTTTCTCTCCACAGGT-3’, 3’-CATGTCCATCAGGTTCTTGC-5’). All mice were maintained at a 12-h light-dark cycle and a constant temperature (25 °C) at Centralized Animal Facility of Hong Kong Polytechnic University. All animal experiments were conducted in accordance with ethical approvals at Hong Kong Polytechnic University (17-18/15-BME-R-HMRF, 17-18/48-BME-R-GRF, 20-21/259-BME-R-HMRF and 24-25/1408-BME-R-AoE) as well as ARRIVE guidelines.

Tissue fixation and histology

Bone tissues were fixed with 4% paraformaldehyde (PFA) for 48 h, decalcified in 10% EDTA for 8 days, dehydrated and embedded in paraffin overnight before cut into 7 µm tissue sections with a microtome (Leica RM2235). After de-paraffinized in xylene and rehydrated through series of ethanol, hematoxylin (Sigma, HHS32) and eosin Y (Sigma, 230251) were used to stain the tissues. To quantify the numbers of osteocytes and empty lacunae in the mouse tibiae and femora, three to four pictures of each sample were randomly taken using 20x objective of a microscope (Eclipse 80i, Nikon; Tokyo, Japan). The pictures were carefully and blindly examined to identify osteocytes and empty lacunae based on the morphology and manually counted. ImageJ software was used to measure cortical bone area in each image.

Immunostaining

De-paraffinized tissue sections were quenched in 3% H₂O₂ in methanol for 10 min, followed by antigen retrieval in an alkaline buffer (10 mM Tris Base, 1 mM EDTA, 0.05% Tween, pH 9) at 65 °C for 20 min. Cultured cells were fixed with 4% PFA for 20–30 min and permeabilized with 0.01% Triton X-100 and 1% SDS in PBS for 20 min. After blocking in 1% bovine serum albumin (BSA) in PBS, primary antibodies against CFTR (1:50, Alomone Lab, ACL006) were used for incubating the tissue sections or cells. Horseradish peroxidase (HRP)-or fluorescence-conjugated secondary antibodies were used. HRP signals were detected using DAB Substrate Kit (Abcam, ab64238). DAPI (Vector Lab, H-1800) or hematoxylin was used to label nuclei. Phalloidin-iFluor 488 (1:500, Abcam, ab176753) was used to label actin.

Micro-CT

Bone specimens (lumbar vertebra, femur and tibia) collected from mice were fixed with 4% PFA for 48 h before transferred into 70% ethanol and scanned by a micro-CT machine (SkyScan 1276, Bruker, Kontich, Belgium)⁴² with a resolution of 6.5 to 13 µm per voxel (specified in Supplementary Table 1). To scan whole-body skeletons, in vivo micro-CT was conducted. Mice were anesthetized and secured on a 30-mm cassette before placed into the machine and scanned with a resolution of 20 μm per voxel. 3D reconstruction was done using NRecon software (Version 2.0.0.5, Bruker). To analyse the data, the region of interest (ROI) was selected from 2D images with threshold and longitudinal span specified in Supplementary Table 1. For quantification, Data Viewer software (Version 1.5.6.2, Bruker) and CTAn software (Version 1.20.8.0 + , Bruker) were used. Morphometric parameters including total tissue volume (TV), bone volume (BV), bone mineral density (BMD), cortical bone thickness (Ct.th), trabeculae number (Tb.N), trabecular thickness (Tb.Th) and trabecular separation (Tb.Sp) were quantified. In one set of experiments, lumbar vertebra collected from aged mice were wrapped in wet gauze and fitted in sample tubes before scanned by another micro-CT machine (µCT40, Scanco Medical, Switzerland) with a resolution of 15 µm per voxel. Regions between intervertebral discs were selected as ROIs with a threshold >180, and a low-pass gaussian filter (Sigma = 0.8, Support = 1.0) was used for 3D reconstruction.

Bone mechanical strength test

A Hounsfield Test Equipment (H25KS UK) was used to perform three-point bending tests on mouse tibiae and femora, as well as uniaxial compression tests on vertebrae. After removal of soft tissues, tibia and femur specimens were mounted, with the anterior-posterior axis in line with the loading force direction, on two supporting bars spaced 10 mm apart in the machine. The 4^th or 5^th lumbar (L4 or L5) vertebra were carefully dissected out with soft tissues removed and transversal processes trimmed off for the specimens to properly fit and be mounted vertically in the machine. A 250 N loading force from the top was applied at a velocity of 5 mm/min to the midshaft of the femur/tibia or to the superior surface of vertebrae body until failure. Load-displacement curves were generated using the built-in QMAT software (Redhill, Surrey, UK) to read the maximal force resisted by the bones.

Cell culture

MLO-Y4 (Karafast, EKC002) and MC3T3-E1 subclone 4 (ATCC, CRL-2593) cells were cultured in Alpha Minimum Essential Medium (αMEM, Gibco, 11900024) supplied with fetal bovine serum (FBS, 5% for MLO-Y4 and 10% for MC3T3-E1, Gibco, 10270106), penicillin (100 U/mL) and streptomycin (100 µg/mL, Gibco, 15140122) at 37 °C in 5% CO₂ incubator. For MLO-Y4, all the culture-wares were coated with 0.15 mg/mL Collagen I (Gibco, A1048301) in 0.02 M acetate acid at room temperature for 1 h and washed by 1x PBS before used. RAW 264.7 cells (ATCC, TIB-71) were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco, 12100046) with supplement of 10% FBS, 100 units/mL of penicillin and 100 µg/mL of streptomycin in 5% CO₂ incubator at 37 °C.

CRISPR-Cas9-based CFTR knockout in MLO-Y4 cells

Three sets of sgRNA oligos, sgRNA1 (Forward: 5’- ACA CCG ATT TTT GGC CTT CAT CGC ATG-3’, Reverse: 5’-AAA ACA TGC GAT GAA GGC CAA AAA TCG-3’), sgRNA2 (Forward: 5’-ACA CCG CCT TGG TTT ACT GAT AAT CCG-3’, Reverse: 5’-AAA ACG GAT TAT CAG TAA ACC AAG GCG-3’), and sgRNA3 (Forward: 5’-ACA CCG TAC CAT ATC TGT ACG GCA GTG-3’, Reverse: 5’-AAA ACA CTG CCG TAC AGA TAT GGT ACG-3’) were designed through an online platform (https://portals.broadinstitute.org) to target mouse CFTR gene at exons 4, 6 and 8, respectively. sgRNA1, sgRNA2 and sgRNA3 were purchased (from Thermo Scientific) and individually ligated using T4 DNA Ligase (Invitrogen, 15224041) into MLM3636 plasmids to construct recombinant plasmids, MLM_sgCFTR_1, MLM_sgCFTR_2 and MLM_sgCFTR_3, respectively. These plasmids were transformed into competent E.Coli for amplification and afterwards sequenced (by BGI, Hong Kong) for verification. To transfect MLO-Y4 cells, cells were seeded at a density of 1.0 × 10⁴ cells/cm². 24 h after the seeding, three combinations of two different recombinant plasmids (0.6 µg each) were transfected together with 6.25 μg CRISPR-Cas9 protein (Invitrogen, A36496) and 12.5 μL Lipofectamine (Invitrogen, CMAX00015) in 125 μL OptiMEM medium (Gibco, 3198508) into cells in each well of a 6-well-plate. The three plasmid combinations were: MLM_sgCFTR_1 + MLM_sgCFTR_2, MLM_sgCFTR_2 + MLM_sgCFTR_3, and MLM_sgCFTR_1 + MLM_sgCFTR_3. 48 h after the transfection, cells were resuspended and seeded into 96-well plates at a density of 0.5 cells per well for single colony expansion. 53 single colonies from three 96-well plates were found survived and all of them were individually expanded, from which genomic DNAs were extracted using a purification kit (Thermo Scientific, K0721), amplified by PCR at exons 4, 6 and 8, and subsequently sequenced individually. Primers used for PCR and sequencing were: exon 4 (Forward: 5’- TTA ATC ACT GCC TTC TCC TG-3’, Reverse: 5’- CAC TGA AGA CAT GTC AGG TAG-3’), exon 6 (Forward: 5’- GGA CTT GCC TTG GCA CAT TT-3’, Reverse: 5’- CTG TAC TTC ACC ATC ATC TTC-3’), and exon 8 (Forward: 5’- GGT GGA GCT GAA AAT GAC CCG-3’, Reverse: 5’- CTA CAA GGT GAG CAT TCC AG-3’). Only one out of the 53 colonies, which was transfected with the combination of MLM_sgCFTR_1 + MLM_sgCFTR_3, was found to be misaligned after comparing the sequencing results with mouse CFTR gene using BLAST, and the misalignments were at exon 4 but not at exon 8. Off-target analysis through online tools (https://zlab.bio/guide-design-resources) showed predicted top 10 off-target sites of sgRNA1 and sgRNA3 only at non-coding areas in the mouse genome. This single colony of MLO-Y4 cells was then used as the CFTR knockout line in other experiments.

Patch clamp

MLO-Y4 cells were cultured on 35 mm culture dishes coated with rat tail type 1 collagen for 48 h with the seeding density of 2000 cells/cm² before the patch-clamp recording. 2–5mΩ glass pipettes were pulled by a micropipette puller (P1000, Sutter Instrument Co., USA). Electrical signals were acquired by MultiClamp 700B and Digidata 1550B (Molecular Devices, USA). For the CFTR current measurement, cells were bathed in Margo ringers’ solution (in mM): NaCl 130, KCl 5, MgCl₂ 1, CaCl₂ 2.5, HEPES 20 and glucose 5.8. (pH=7.4), with a pipette solution (CsCl 135, MgCl₂ 2, Mg-ATP 2, HEPES 10, EGTA 10, glucose 5.8). After the formation of Giga seal, the capacitance of cell was measured. A series of voltage stimulations of 1000 ms duration stepped from −100 mV to +80 mV with 20 mV interval were applied to elicit whole-cell currents.

Intracellular Cl^- imaging

MLO-Y4 cells were cultured on glass coverslips coated with Collagen I. Before the measurement, cells were washed with Margo Ringer’s solution (in mM): NaCl 130, KCl 5, MgCl₂ 1, CaCl₂ 2.5, HEPES 20 and glucose 5.8. (pH = 7.4) and then incubated with a Cl^- sensitive fluorescent dye, N-(ethoxycarbonylmethyl)−6-methoxyquinolinium bromide (MQAE, 10 mM, MedChemExpress, HY-D0090), in the bath solution at 37 °C for 30 min. Afterwards, cells were washed by the bath solution to remove residual dyes. MAQE-loaded cells on the coverslip were transferred to a mini chamber, which was then mounted to a fluorescence microscope (Eclipse Ti, Nikon). MQAE fluorescence in the cells was excited at 340 nm, and emission signals were recorded at 460 nm. The fluorescence intensity of MQAE inversely correlates with Cl^- concentration. Calibration of intracellular Cl^- concentration with MQAE intensity was done by stimulating the cells with a series of KCl (20-80 mM) in the presence of vancomycin (5 μM) and potassium thiocyanate (KSCN, 105 mM).

Cell viability assay

Cells were seeded into 96-well plates at a density of 2000 cells/well. From the second day after seeding till day 8, cells were incubated with 0.5 mg/mL MTT (3-(4,5-Dimethylthiazol-2-yl)−2,5-Diphenyltetrazolium Bromide, Invitrogen, M6494) in the culture medium at 37 °C for 2 h before the medium was removed, DMSO (100 µL per well) was added and the absorbance was measured at 595 nm by a microplate reader (Labexim Products LEDETECT 96). In a group of experiments, cells were seeded in a 96-well plate at an initial density of 1000 cells/well and cultured for 6 days with the medium daily replenished. In another group, cells were seeded at 5000 cells/well and grew for 4 days before the conditioned medium from CFTR-KO or WT cells (see below) was added.

β-galactosidase assay

Cells were seeded at a density of 10,000 cells/cm² 24 h before the assay. Senescence β-Galactosidase Staining Kit (Cell Signaling Technology, 9860) was used to stain fixed cells.

Quantitative PCR

Bone samples (e.g., femur, vertebra) were collected, carefully removed of soft tissues, flushed with PBS to remove bone marrows and quickly minced in liquid nitrogen before collecting into TRIzol (Invitrogen, 15596018) for RNA extraction. RNAs were also extracted from cultured cells using TRIzol. Quantity and purity of extracted RNAs were determined by Nanodrop spectrophotometer (Thermo, ND-ONE-W). 1 μg total RNA was reverse-transcribed to cDNA using the High-Capacity cDNA Reverse Transcription Kit (Thermo, 4368814). 1 µL of total cDNA of each sample was amplified in a 10 µL reaction system containing SYBR Green Premix Ex Taq Mix (Takara, RR420A) and primers (Supplementary Table 3) using real-time PCR system (Bio-Rad, CFX96 or Applied Biosystems QuantStudio 7). The expression of the target genes was normalized to that of Gapdh. Relative gene expression fold was calculated using the 2^−∆∆Cq formula.

Reactive oxygen species assay

Cells were seeded at a density of 2000 cells/well in a 96-well plate 24 h before the assay. Cellular ROS Assay Kit (Abcam, ab113851) was used.

TUNEL staining

Tissues were fixed and sectioned as above mentioned. Cells were fixed with 4% PFA and permeabilized with proteinase K in PBS for 15 min at room temperature before a TUNEL assay kit (ThermoFisher Scientific, Cat#C10617 or Cat#C0618) was used to label the 3ʹ-end of fragmented DNA of the apoptotic cells according to the manufacturer’s protocols.

Annexin V and propidium iodide assay

Cells (1 × 10⁶ cells/mL) were resuspended and stained using Annexin V-FITC/PI kit (ThermoFisher Scientific, Cat# BMS500FI-300) according to the manufacturer’s instruction. A flow cytometer (BD Accuri C6, USA) was used. Fluorescent signals from 50,000 cells per sample were acquired for analysis. Gating was applied using forward scatter area (FSC-A) versus side scatter area (SSC-A) was used to rule out cell debris, and subsequently FSC-A versus forward scatter height (FSC-H) to keep singlets. Unstained cells were used to determine the negative boundary and eliminate background signals. FlowJo™ v10.8.1 software was used for the analysis.

Western blot

Cells or tissues were lysed in ice-cold RIPA buffer (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1% NP-40, 0.5% DOC, and 0.1% SDS) with a protease inhibitor cocktail (cOmplete™, Roche, Cat#11836170001), incubated for 30 min on ice before centrifuged at 13,000 rpm at 4 °C for another 30 min and the supernatant collected as extracted proteins. To facilitate bone tissue protein extraction, a commercial kit (Minute™ Total Protein Extraction Kit, Invent Biotechnology, SA-02-BT) was used. Briefly, freshly collected bone tissues from the mice were cut into small pieces before minced and homogenized in ice-cold RIPA using the plastic beads from the kit. After incubating for 30 min on ice in RIPA, the homogenate was filtered and centrifuged at 13,000 rpm at 4 °C for 30 min. Extracted proteins were denatured and equal amounts of protein were resolved by SDS-polyacrylamide gel electrophoresis and electroblotted onto an equilibrated nitrocellulose (NC) or polyvinylidene fluoride (PVDF) membrane. After blocking in Tris-buffered saline containing 5% non-fat milk, the membranes were immunoblotted with primary antibody of target proteins overnight at 4 °C. Antibodies against CFTR (1:1000 to 1:500, Alomone Lab, ACL006), CFTR (1:1000, Millipore, MAB3482), NFκB (1:1000, CST, 8242), Phospho-NFκB P65 (1:1000, CST, 3033), β-tubulin (1:1000, Thermo, PA5-16863), GAPDH (1:1000, Abcam, ab9484), Actin (1:1000, Millipore, MAB1501R) were used. After three washes in TBS containing 0.1% Tween 20, membranes were further incubated with HRP-conjugated antibodies and visualized by the enhanced chemiluminescence assay (GE Healthcare, UK) or SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo Scientific™, 34095) according to the manufacturer’s instructions. Densitometry of Western blots was performed by ImageJ software (National Institute of Health, USA).

Overexpression of CFTR in vitro

The pCDNA3.1 plasmid expressing human CFTR (hCFTR) as described previously²⁵ or empty pCDNA3.1 plasmid was used to transfect CFTR-KO MLO-Y4 cells. DNA (3–7.5 µg per well for six-well plates) was transfected into the cells with Lipofectamine 2000 Transfection Reagent (Invitrogen; 11668027) and Optimal-MEM Reduced Serum Medium (Thermo Fisher Scientific, 31985070).

Conditioned medium collection

MLO-Y4 cells with or without CFTR knockout (WT or CFTR-KO) were seeded at a density of 10,000 cells/cm² in αMEM (with 5% FBS) and incubated for 72 h before the medium was collected and removed of floating cells/particles by centrifugation. Such conditioned medium (CM) was used for incubation of cells or other analysis.

Cytokine array

The collected conditioned medium or mouse serum was used in Proteome Profiler Mouse Cytokine Array Kit (R&D system, ARY006) according to the manufacturer’s instructions. The pixel density of each dot was quantified using ImageJ software (National Institute of Health, USA). In some experiments, mixed serum samples from 4 mice of each group were used.

ELISA

ELISA assays for RANKL (R&D system, MTR00), osteoprotegerin (R&D system, MOP00), BALP (Cubasio, CSB-E11914m), PⅠNP (Cubasio, CSB-E12775m) and CTX-Ⅰ (Cubasio, CSB-E12782m) were performed according to the manufacturer’s instructions.

Bone marrow cell isolation and culture

Femora were collected from 6-week-old C57BL/6 mice before cut off the two ends for the bone marrow to be flushed out by a sterile PBS-filled syringe into a centrifuge tube. After centrifugation at 1200 rpm for 5 min, the cell pellets were collected as mixed bone marrow cells, which were then resuspended in α-MEM supplemented with 10% FBS, 100 units/mL penicillin, and 100 µg/mL streptomycin, and cultured in a 5% CO₂ incubator at 37 °C. Non-adherent cells were removed after 72 h. The 1^st passage of bone marrow cells was seeded for induction of osteoclast differentiation, and the 2^nd passage was used for osteogenic differentiation.

In vitro osteogenic differentiation

Osteogenic reagents, (+)-sodium L-ascorbate (50 µg/mL), β-glycerophosphate (10 mM) and dexamethasone (1 nM), were added into the collected MLO-Y4-conditioned medium before the medium was used to incubate bone marrow cells or MC3T3-E1 cells for up to 2 weeks to induce possible osteogenic differentiation. Cells were fixed with 70% ethanol for 5 min at room temperature, washed with ddH₂O followed by incubation with 2% (w/v) Alizarin Red S (ARS, pH 4.2, Sigma, A5533) at room temperature for 10 min before washed with ddH₂O. After the staining, pictures were taken by the scanner (EPSON, perfection V700) at a resolution of 300dpi and an inverted microscope (ECLIPSE Ti2-A, Nikon; Tokyo, Japan). Quantification was done by ImageJ software as previously reported⁴³.

In vitro osteoclastogenic differentiation

The MLO-Y4-conditioned medium was mixed with DMEM (with 10% FBS) or αMEM (with 10% FBS) in a 1:1 (v:v) ratio before being used to incubate bone marrow cells or RAW 264.7 cells (seeded at a density of 2000 cells/cm²) for 5-6 days, to induce osteoclastogenesis.

Tartrate-resistant acid phosphatase (TRAP) staining

TRAP staining kits (Solarbio, Beijing, China, and Sigma, 387A-1KT, USA) were used according to the manufacturer’s instructions. For quantification, three to four images per sample were taken in distal femur and proximal tibia bone by 20x objective of a microscope (Eclipse 80i, Nikon; Tokyo, Japan). The pictures were carefully and blindly examined to identify TRAP-positive multinuclear cells as osteoclasts and manually counted. The length of bone surface was measured for normalization.

Bone formation rate measurement

Mice at 16-week-old were intraperitoneally injected with Calcein green (10 mg per kg body weight, Sigma-Aldrich, USA), 9 days afterwards with xylenol orange (90 mg per kg body weight, Sigma-Aldrich, USA) and sacrificed another 3 days later. Bone tissues were collected, fixed with 4% PFA for 48 h and treated in 5% KOH for another 96 h as previously reported⁴⁴, before sliced into 10 µm sections using a cryostat (Leica, CM1950). Fluorescence images of these bone sections were taken. Mineral apposition rate (MAR) was measured as the distance between mid-lines of calcein and xylenol labeling divided by the interval days (i.e., 9 days). Bone surface (BS), double labeled surface (dLS) and single labeled surface (sLS) were measured. Mineralizing surface (MS) was calculated as dLS + (sLS/2) and bone formation rate (BFR) was calculated as MAR × (MS/BS). NIS-Elements software (Nikon) was used for the measurement.

RNA sequencing and data analysis

RNAiso Plus (Takara, 9108) was used to extract RNAs from wild-type and CFTR-KO MLO-Y4 cells 48 h after seeding according to the manufacturer’s instructions, which were then sent to BGI Corporation for RNA quality, library complexity, and alignment check before bulk RNA sequencing was done. FPKM (fragments per kilobase of transcript sequence per millions of base pairs sequenced) was used to estimate transcription level of genes’ expression and identify differentially expressed genes (DEGs). KEGG pathway enrichment analysis was done through the BGI online platform Dr. Tom (https://biosys.bgi.com/#/report/login) by mapping all DEGs to each entry in KEGG database (http://www.genome.jp/kegg/), calculating the number of genes per entry, and then applying a hypergeometric test to find the KEGG functions/pathways that are significantly enriched of the DEGs compared to all background genes of the species. Q value (corrected P value) was obtained by correcting the P value through multiple tests. Q value $\le$ 0.05 was used as the threshold.

Cl^- free incubation

MLO-Y4 cells were incubated in the Margo ringers’ solution as control or a modified Cl^- free solution (in mM): Na-gluconate 130, K-gluconate 5, Ca-gluconate 2.5, MgSO₄ 1, HEPES 20 and glucose 10 (pH=7.4) at 37 °C. In some experiments, Cl^- free and normal Margo ringers’ solutions were mixed by 1:1 in volume to make a low Cl^- solution. Cells were treated for 2 h before mass spectrometry experiments, for 1 to 24 h in ROS assay or for 12 to 36 h before viability test.

Mass spectrometry

Whole-cell proteins were extracted and purified by EasyPep mini-MS sample prep kit (Thermo scientific, A40006). A vacuum centrifuge was used to dry the proteins, which were then resuspended in 0.1% formic acid. Samples were analyzed by a liquid chromatography-tandem mass spectrometry (LC-MS/MS). Proteome Discoverer software (Version 2.1) was used for raw data to be aligned with UniprotKB protein database for Mus musculus (UP000000589) with 55,466 entries. The label-free quantitation (LFQ) algorithm was applied to calculate protein abundances as the sum of the peptide group abundances associated with that protein. Significantly differential (p < 0.05) genes/proteins were used for KEGG (Kyoto Encyclopedia of Genes and Genomes) enrichment analysis through an online platform KOBAS.

Human bone samples

Human bone specimens (Supplementary Table 2) otherwise disposable were collected during orthopedic surgeries from patients at the First Affiliated Hospital, Shantou University Medical College and at the Department of Orthopedics, Shenzhen Nanshan People’s Hospital with related ethical approval (Ref. No: SUMC2019-060, No.KY-2025-071830). All patients provided written informed consent for their tissue to be used in the present study. All patients are Chinese. Ischemic/necrotic/collapse sites were avoided and trabecular bones in proximal femur neck or subchondral bone without visible lesions were used. These specimens were fixed by 4% PFA at room temperature after collection.

Adenovirus-based overexpression of CFTR in vivo

Adenovirus containing human CFTR sequence (Adv-hCFTR, pAdeno-MCMV-CFTR-T2A-3Flag-T2A-mCherry) and null sequence (Adv-Vector, pAdeno-MCMV-3Flag-T2A-mCherry) as control were purchased from OBiO Technology (Shanghai) Corp., Ltd. 10⁸ PFU Adv-CFTR or Adv-Vector in 50 µL saline was locally injected into the periosteum surface of mid-shaft femur in 20 to 28-month-old male or female wild-type C57BL/6 mice. In some mice, two doses of the virus were injected on day 1 and day 8, respectively. Adv-hCFTR and Adv-Vector were injected into right and left femora, respectively, in each mouse for paired comparisons.

Local injection of VX809

28-month-old female wild-type C57BL/6 mice were randomly divided into groups. In each mouse, 50 µL saline with VX809 (Selleckchem, S1565, 200 μM) or DMSO (0.2% v/v) as the vehicle control was locally injected at the bone surface near L4 lumbar vertebra every two days for 4 weeks.

Statistical analysis

Sex- and age-matched animals were divided into groups randomly. The sample size was based on preliminary data. Data are presented as means ± s.e.m. Two-tailed unpaired or paired Student’s t-test was used for comparison between two groups. One-way ANOVA with Bonferroni’s post hoc test was used for multiple-group comparisons. Two-way ANOVA with Bonferroni’s post hoc test was used when there were two categorical independent variables. Pearson test was used for correlation analysis. P < 0.05 was considered statistically significant. All experiments were repeated independently at least twice with similar results, except RNA sequencing and mass spectrometry, which were conducted once with 3 biological replicates. No data were excluded. GraphPad Prism 10.0 software was used for statistical analysis.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conception: Y.C.R.; Experiments and/or data analysis: P.H., W.D., M.C., J.C., X.Z., Z.C., J.G., H.C., R.X., X.C., J.X. and Y.C.R.; Intelligence input and technical advisory: Ling Q., H.C.C., X.J., W.Y.L. and J.X.; Clinical data and consultant: Lei Q., J.X. and J.H.; Schematic drawing: Y.C.R.; Manuscript writing: Y.C.R.

Peer review

Peer review information

Nature Communications thanks Fréderic Velard, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The RNA sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive under accession code PRJNA1434652. The Mass Spectrometry raw data generated in this study have been deposited in the MetaboLights database under the accession code MTBLS14165. Source data are provided. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-72349-0.