Research ArticleDevelopmental Biology

Mice lacking the intracellular cation channel TRIC-B have compromised collagen production and impaired bone mineralization

See allHide authors and affiliations

Science Signaling  17 May 2016:
Vol. 9, Issue 428, pp. ra49
DOI: 10.1126/scisignal.aad9055

The TRIC to building strong bones

During bone development, osteoblasts secrete a collagen-rich matrix that is necessary for bone mineralization. Defects in collagen deposition cause osteogenesis imperfecta (OI), a disease characterized by brittle bones. Zhao et al. found that mice lacking Tric-b, which encodes a trimeric intracellular cation channel that localizes to the endoplasmic reticulum (ER), had bone defects similar to those of OI patients. Although osteoblasts in Tric-b knockout mice synthesized collagen, it accumulated inside the cells instead of being secreted. The accumulation of intracellular collagen deposits was associated with morphological and biochemical markers of ER stress, including severe dilation, excess Ca2+ in the ER, and impaired Ca2+ release from the ER. These findings suggest that TRIC-B is necessary to maintain ER homeostasis, thus enabling osteoblasts to secrete the large amounts of collagen required to build strong bones.


The trimeric intracellular cation (TRIC) channels TRIC-A and TRIC-B localize predominantly to the endoplasmic reticulum (ER) and likely support Ca2+ release from intracellular stores by mediating cationic flux to maintain electrical neutrality. Deletion and point mutations in TRIC-B occur in families with autosomal recessive osteogenesis imperfecta. Tric-b knockout mice develop neonatal respiratory failure and exhibit poor bone ossification. We investigated the cellular defect causing the bone phenotype. Bone histology indicated collagen matrix deposition was reduced in Tric-b knockout mice. Osteoblasts, the bone-depositing cells, from Tric-b knockout mice exhibited reduced Ca2+ release from ER and increased ER Ca2+ content, which was associated with ER swelling. These cells also had impaired collagen release without a decrease in collagen-encoding transcripts, consistent with a defect in trafficking of collagen through ER. In contrast, osteoclasts, the bone-degrading cells, from Tric-b knockout mice were similar to those from wild-type mice. Thus, TRIC-B function is essential to support the production and release of large amounts of collagen by osteoblasts, which is necessary for bone mineralization.


Ca2+ release from the endoplasmic reticulum (ER) and sarcoplasmic reticulum (SR) regulates various important cellular functions (1). It is reasonable to expect that Ca2+ efflux from ER or SR mediated by inositol 1,4,5-trisphosphate receptors (IP3Rs) and ryanodine receptors (RyRs), unless balanced by counter-ion flux, would generate a negative potential in the ER or SR lumen, which would inhibit subsequent Ca2+ release from intracellular stores. Previous studies have proposed that counter-ion flux into the lumen neutralizes store membrane potential to facilitate physiological Ca2+ release (24). The channels and transporters mediating counter-ion movements have not been identified, although several ionic fluxes, such as K+, Cl, and H+ currents, have been detected in muscle SR (57). Two TRIC (trimeric intracellular cation) channels, TRIC-A and TRIC-B, are localized to ER, SR, and nuclear membranes (8, 9). TRIC proteins assemble into bullet-shaped homotrimers and likely form monovalent cation-selective channels (10). Gene knockout studies indicate that TRIC channels support Ca2+ release from intracellular stores. Double knockout mice lacking both TRIC subtypes die during early embryonic stages, and cardiomyocytes isolated from these embryos exhibit impaired RyR-mediated Ca2+ release (10). In Tric-a knockout mice, the vascular smooth muscle generates RyR-mediated transient localized Ca2+ signals called “Ca2+ sparks” at a reduced frequency, and the skeletal muscle exhibits irregular contractile responses that are likely due to compromised RyR-mediated Ca2+ release (1113). Moreover, Tric-b knockout mice die immediately after birth due to respiratory failure, and the alveolar epithelial cells from these mice exhibit insufficient production and secretion of surfactant lipids likely due to diminished IP3R-mediated Ca2+ release (14). TRIC channels, therefore, may contribute to the counter-ion movements required for physiological Ca2+ release in various cell types.

Osteogenesis imperfecta (OI) is a hereditary disease characterized by low bone mass, leading to increased bone fragility (15). OI patients present with diverse phenotypes, such as shortened and bowed limbs, brittle opalescent teeth, and persistently blue sclerae (16). Most OI cases result from defects in the structure, function, production, or trafficking of type I collagen (15), a main constituent of the bone extracellular matrix. Type I collagen both strengthens the bone and acts as the substrate for mineralization. OI-causing mutations also occur in several genes encoding proteins unrelated to collagen deposition, such as the osteoblast-specific transcription factor osterix and the osteoblast-specific transmembrane protein IFITM5 (16). Homozygous deletion and point mutations in the TRIC-B (also known as TMEM38B in humans) locus have been identified in unrelated OI pedigrees (1720). For example, one deletion mutation eliminates exons 1 and 2 from TRIC-B (19), and another point mutation disrupts the splicing of intron 3 (20). These mutations likely compromise TRIC-B function, but the pathophysiological mechanism in TRIC-B–associated OI is unknown. Here, we report that Tric-b knockout mice develop an OI-like bone phenotype. By analyzing bone tissue from the knockout animals, we found that TRIC-B is essential for active osteoblasts to produce the large amount of collagen that is necessary for bone mineralization.


Impaired bone mineralization in Tric-b knockout mice

Tric-b knockout neonates die immediately after birth because of respiratory failure (14). Here, we analyzed osteogenesis in the knockout mice just before (embryonic day E18) and immediately after birth (postnatal day P0). In whole-body skeleton preparations, the P0 knockout mice were slightly smaller in body size than control mice (14) but macroscopically normal in overall skeletal features (Fig. 1A). After Alizarin red staining to detect calcium-rich deposits, the surface coloration of each bone was generally uniform in control mice. However, major bones from the knockout mice, including the skull, rib, and femoral bones, had patches of reduced staining, suggesting impaired mineralization. The skulls and femoral bones from both male and female knockout mice were physically fragile and easily fractured during handling. Therefore, Tric-b knockout mice congenitally develop systemic impairment of osteogenesis similar to OI.

Fig. 1 Impaired bone mineralization in Tric-b knockout mice.

(A) Skeletons from wild-type (WT) and Tric-b knockout (Tric-b KO) P0 mice. The preparations were stained with Alcian blue (cartilage) and Alizarin red (bone). Scale bars, 5 mm (left panels) and 0.5 mm (middle and right panels). (B) Kossa-stained mid–cross sections of femoral bones from P0 mice. Scale bar, 0.2 mm. Both the cross-sectional area and Kossa-positive area were determined from digitalized images, and the Kossa-positive fraction in the cross-sectional area (Kossa-stained ratio) was calculated (graphs). n values represent the numbers of mice examined and are shown in parentheses. (C) Hematoxylin and eosin–stained calvarial sections from the sincipital region of skulls from P0 mice. Scale bars, 250 μm (upper panels) and 10 μm (lower panels). The cortical layer thickness, bone matrix thickness, and eosin-stained intensity in bone matrix were measured in each section (graphs). n values represent the numbers of mice examined and are shown in parentheses. (D) Mid-longitudinal sections of femoral diaphyses from P0 WT and Tric-b KO mice stained with picrosirius red. Scale bar, 0.5 mm. The portion of the diaphysis stained with picrosirius red (picrosirius red–stained area), the ratio of the picrosirius red–stained area to the total area of the diaphysis section (picrosirius red–stained ratio), and relative staining intensity were determined in each section. In the bar graphs, the data represent means ± SEM, and the numbers of mice examined are shown in parentheses. Statistical differences between the genotypes are marked with asterisks (*P < 0.05 and **P < 0.01 in t test).

Kossa staining is a routine method for visualizing calcium deposits as metallic silver. In this assay, femoral bones from P0 knockout mice exhibited reduced mineralization both in the cortex and medulla regions (Fig. 1B). Micro-CT (computed tomography) further confirmed impaired ossification in the femoral bones from Tric-b knockout mice (fig. S1A). In hematoxylin and eosin–stained skull bones, the cortical layer and bone matrix were thin in the knockout mice (Fig. 1C). Eosin intensely stains collagen fibers in bone matrix, and eosin staining was uniform in wild-type bones. In contrast, we frequently detected eosin-negative regions in skull bones from the knockout mice. During embryonic development, bone growth depends on the availability of adequate circulating minerals (21). However, insufficient circulating minerals are unlikely to contribute to the OI-like phenotype because serum Ca2+ and phosphate contents were within the normal ranges in the E18 knockout mice (fig. S1B).

There are two histogenic mechanisms of bone formation: (i) intramembranous ossification, which involves bone formation directly within the primitive connective tissue, and (ii) endochondral ossification, which involves the use of cartilage as a precursor to bone (22). Osteoblasts and osteoclasts are important for both types of bone formation, whereas chondrocytes are only involved in endochondral ossification. The skull is a typical bone formed through intramembranous ossification; the femur is a typical bone formed through endochondral ossification. On the basis of the poor ossification observed in both types of bone in Tric-b knockouts, we hypothesized that the altered function of osteoblasts or osteoclasts or both, rather than chondrocytes, is responsible for the OI-like phenotype of Tric-b knockout mice.

Insufficient collagen matrix in bones from Tric-b knockout mice

Osteoblasts produce abundant collagen, which organizes the bone matrix (22). Picrosirius red staining for collagen fibers clearly indicated that the collagen matrix was abnormal in femoral bones from Tric-b knockout mice (Fig. 1D). Similar to the Kossa staining results, femurs from P0 knockout mice showed reduced collagen underneath the periosteum and a near absence of collagen in the central regions. These data, together with observations in skull bones, indicated that the collagen deficiency contributed to the impaired ossification of bones in the knockout mice. Reduced collagen synthesis by osteoblasts or increased bone resorption by osteoclasts, or both, could account for the OI-like phenotype observed in the knockout mice.

Bone mineralization is associated with matrix vesicles, which bud off from osteoblast cell membranes and attach to the bone matrix (23). Matrix vesicles are distinct from secretory vesicles containing collagen fibers in osteoblasts. Deposits of hydroxyapatite, the main mineral constituent of bone, form in the lumen of matrix vesicles from concentrated Ca2+ and phosphate. Matrix vesicles can be isolated from bones and actively form hydroxyapatite in vitro. Matrix vesicles prepared from the knockout mice appeared normal; we detected no obvious differences in the recovery yield, protein composition, morphology, or mineralizing activity between matrix vesicle preparations from P0 knockout and wild-type mice (fig. S2). Therefore, neither insufficient production nor altered performance of matrix vesicles seemed to contribute to the OI-like phenotype.

Osteoblasts and osteoclasts in Tric-b knockout mice

We analyzed femoral bones from Tric-b knockout mice using conventional histochemical methods (Fig. 2A). Osteoblasts and osteoclasts were visualized using endogenous alkaline phosphatase (ALP) activity and endogenous tartrate-resistant acid phosphatase (TRAP) activity, respectively. We visualized cartilage by staining with Safranin O, which labels the extracellular matrix secreted by chondrocytes. The diaphyseal region (the bone shaft) was narrowed in femurs from P0 knockout mice, but bones from knockout and wild-type mice exhibited similar regional ALP- and TRAP-staining densities at each end of the diaphysis. However, bones from knockout mice had few osteoblasts and osteoclasts in the middle of the diaphysis, whereas both cell types were observed at the mid-diaphysis in wild-type bones. Cartilage was present at similar densities in the epiphyses of bones from knockout and wild-type mice.

Fig. 2 Histological defects in Tric-b knockout bones.

(A) Histological analysis of osteoblasts (ALP staining), osteoclasts (TRAP staining), and cartilage (Safranin O staining) in longitudinal sections of femurs from WT and Tric-b KO P0 mice. Scale bar, 0.5 mm. (B) Bone and diaphyseal sizes in E16.5 and P0 femoral bones. Mid-longitudinal sections prepared from femurs were subjected to Kossa staining, and the bone size and diaphyseal length (double-headed arrows) were measured in each section. Scale bar, 0.5 mm. (C) Quantitative real-time polymerase chain reaction (qRT-PCR) analysis in P0 femoral bones. Marker genes examined are categorized according to cell type–specific expression. In the graph charts, the data represent means ± SEM, and the numbers of mice examined are shown in parentheses. Statistical differences between the genotypes are marked with asterisks (*P < 0.05 and **P < 0.01, t test). Ct, cycle threshold.

Bones from day E16.5 embryos contain tiny diaphyses and are mainly composed of cartilage. At this stage, Tric-b knockout femurs maintained regular histology, and we observed no difference in diaphyseal size between the bones of knockout and wild-type mice by Kossa staining (Fig. 2B), indicating that Tric-b ablation did not grossly affect chondrocyte-mediated cartilage formation. After this stage, matrix-degrading proteinases released from osteoclasts and hypertrophic chondrocytes destroy this cartilage, and osteoblasts replace the cartilage with bone and expand the diaphysis (22). Kossa staining confirmed that the diaphyseal regions were significantly narrower in the femurs of P0 knockout mice (Fig. 2B). Therefore, Tric-b ablation impaired diaphyseal expansion during late embryonic development.

To assess whether bone cell differentiation was altered in Tric-b knockouts, we performed transcript analysis for cell type–specific marker genes in femurs from P0 mice (Fig. 2C). For example, Bglap (encoding osteocalcin) and On (osteonectin) are osteoblast marker genes, and Ctr (calcitonin receptor) and Ctsk (cathepsin K) are osteoclast marker genes, whereas Rn18s (18S ribosomal RNA) and Gapdh are expressed in all cell types. The abundance of transcripts that serve as markers of osteoblasts or osteoclasts was decreased in bones from knockout mice, suggesting that there were fewer osteoblasts and osteoclasts in the knockouts or that these cells were less differentiated or had reduced expression of the cell type–specific markers. The altered expression profile may reflect an increased ratio of cartilage-derived cells to osteoblasts and osteoclasts in the bones of the P0 knockout mice.

ER dilation in osteoblasts from Tric-b knockouts

The connective tissue that covers the outer surface of bones, the periosteum, maintains osteoblast progenitors, and immature osteoblasts liberated from the periosteal inner layer migrate into the diaphysis for maturation (24). Mature ALP-positive osteoblasts are highly active in bone matrix production and are characterized by stacked rough ER (rER) layers with electron-dense lumens that are filled with procollagen. As bones mature, osteoblasts stop producing collagen and differentiate into osteocytes, which are surrounded by mineralized bone matrix (24, 25). We performed electron microscopic (EM) analysis of sections of femurs from P0 Tric-b knockout mice, which revealed osteoblasts with dilated rER (Fig. 3A, upper panels). In the knockout mice, a substantial fraction of active (collagen-producing) osteoblasts exhibited dilated rER (Fig. 3B), whereas such dilated rER was not detected in immature osteoblasts near the periosteum or in osteocytes (Fig. 3C). In wild-type mice, active osteoblasts also exhibited dilated rER elements, but did so at a low frequency, and the dilation phenotype was relatively mild, as judged by measuring the rER luminal areas (Fig. 3B). Therefore, osteoblasts from Tric-b knockout mice exhibited severe rER dilation during the time when collagen production was greatest.

Fig. 3 ER dilation in Tric-b knockout osteoblasts.

(A) Electron micrographs of active osteoblasts (upper panels) containing dilated rER elements in femurs from P0 WT and Tric-b KO mice, Golgi cisternae (middle panels, white arrowheads) and electron-dense secretory vesicles (black arrows) in active osteoblasts, and collagen-based bone matrix and mineralizing matrix vesicles (lower panels, arrowheads) surrounding active osteoblasts. Os, osteoblast; Ca, calcium phosphate deposit. Scale bars, 500 nm. (B) Quantification of dilated rER elements and secretory vesicles in active osteoblasts. Dilated rER elements were defined as ribosome-studded organelles with expanded lumina (>0.05 μm2 in cross-sectional area) of the organelle structures studded with ribosomes. Upper panel: Frequency of osteoblasts containing dilated rER. We observed 403 osteoblasts from Tric-b KO mice and 513 osteoblasts from WT mice to determine the frequency of cells with dilated ER and performed statistical comparison (n = 4 mice of each genotype). Middle panel: Severity of rER dilation. In each dilated rER-containing osteoblast, we determined an averaged luminal size for statistical comparison (n = 22 cells from four KO mice and n = 15 cells from four WT mice). Lower panel: Frequency of secretory vesicles. In osteoblast images containing Golgi stacks, we counted secretory vesicles surrounding the stacks for statistical comparison (n = 34 cells from four KO mice and n = 17 cells from four WT mice). Statistical differences between the genotypes are marked with asterisks (**P < 0.01, t test). (C) Immature osteoblasts, osteocytes, and multinuclear osteoclasts in P0 femoral bones. Scale bar, 1 μm.

In active osteoblasts, well-developed Golgi bodies and many secretory vesicles containing collagen fibers are present in the vicinity of the rER stacks. However, in the osteoblasts with dilated rER from the knockout mice, the Golgi apparatus was small (Fig. 3A, middle panels), and the frequency with which we observed secretory vesicles was significantly reduced (Fig. 3B). A dense collagen matrix accompanied by mineralizing matrix vesicles always surrounded osteoblasts in bones from wild-type mice, whereas only sparse collagen fibers were present around osteoblasts in bones from the knockout mice (Fig. 3A, lower panels). Because procollagen is transported from the rER to the Golgi by coat protein complex II (COPII)–coated vesicles, we hypothesized that the dilated rER, smaller Golgi stacks, and reduced secretory vesicles in osteoblasts of Tric-b knockouts may reflect a defect in intracellular collagen sorting and trafficking processes. In contrast to the observation with osteoblasts, osteoclasts from Tric-b knockout mice exhibited no ultrastructural abnormalities (Fig. 3C).

Impaired collagen production in cultured Tric-b knockout osteoblasts

To assess collagen production in osteoblasts, we isolated osteoblast progenitors from neonatal skull bones and expanded the cells in culture. We then stimulated their maturation by adding bone morphogenetic protein-2 (BMP-2). Osteoblast progenitors prepared from Tric-b knockout and wild-type mice exhibited similar growth profiles under culture conditions (fig. S3A, left panel). After 10 days of BMP-2–induced maturation, we detected dense mineralized deposits and collagen matrix, as demonstrated by Alizarin red staining and collagen immunostaining, respectively, in the culture dishes on which wild-type osteoblasts were grown. The dishes on which the knockout osteoblasts were grown had less mineral and collagen matrix deposits (Fig. 4, A and B). To assess the amount of matrix components within the cells, we solubilized intracellular proteins in a deoxychorate-containing solution and removed extracellular mineralized matrix by centrifugation. Tric-b knockout osteoblasts had increased amounts of intracellular collagen fibers, represented by two immunoreactive bands for α1 and α2 chains (Fig. 4C). However, the increase in intracellular procollagen was not the result of increased gene expression or transcript stabilization because we detected similar amounts of transcripts encoding type I collagen (Col1a1 and Col1a2) in the knockout and wild-type cells (fig. S3B).

Fig. 4 Impaired collagen production in Tric-b knockout osteoblasts.

Osteoblast cultures were subjected to the analyses after maturation in differentiation medium for 10 days. (A) Alizarin red staining to show mineralization in osteoblasts cultured from WT and Tric-b KO mice. Staining densities were photometrically measured for quantification. (B) Immunostaining for matrix collagen. Type I collagen immunoreactivity was photometrically quantified. (C) Immunodetection of intracellular procollagen. Intracellular fractions were prepared from cultured osteoclasts, and equal volumes of the preparations were subjected to Western blotting. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was examined as an internal control, and blots were digitalized for quantification. (D) Immunocytochemical observations using an antibody that recognizes type I collagen (green). Left panel: Collagen-positive meshwork structures were observed in most of the osteoblasts from WT mice. Middle panel: Enlarged deposits of collagen (>5 μm in diameter) in minor populations of the osteoblasts from WT mice. Right panel: Enlarged deposits of collagen in osteoblasts from KO mice. Scale bar, 5 μm. Cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI), and >350 cells from each of four preparations were examined for statistical analysis of the frequency and size of the collagen deposits. (E) Osteoblasts double-labeled with antibodies recognizing collagen, KDEL, and GM130. Scale bars, 20 μm. In the bar graphs, the data present means ± SEM, and the numbers of osteoblast preparations derived from different mice are shown in parentheses. Significant differences between the genotypes are denoted by asterisks (*P < 0.05 and **P < 0.01, t test).

In 99% of the cultured wild-type osteoblasts, collagen formed a cytoplasmic meshwork (Fig. 4D). In rare cases (less than 1%), we detected dense collagen deposits in wild-type osteoblasts (Fig. 4D). However, in Tric-b knockout cells, these dense deposits occurred at a significantly higher frequency (3.3%) and were significantly increased in size (Fig. 4D). In double-staining experiments with antibodies that recognize organelle-specific markers (Fig. 4E), the collagen signal was distinct from that of the ER marker KDEL and frequently colocalized with the Golgi marker GM130 in wild-type osteoblasts. In contrast, much of the collagen signal in cells from Tric-b knockout mice colocalized with KDEL, but none colocalized with GM130, suggesting that the intracellular collagen deposits corresponded to the dilated rER elements (Fig. 3). These cytological defects are consistent with the reduction in collagen matrix deposition exhibited by bones from the knockout mice. Therefore, we propose that the accumulation of collagen and severe dilation of rER elements underlie the impaired production of collagen matrix in the Tric-b knockout osteoblasts.

Compromised Ca2+ handling in cultured Tric-b knockout osteoblasts

Because TRIC-B is a cation channel implicated in the maintenance of the internal Ca2+ stores and thus Ca2+ signaling, we investigated Ca2+ handling in BMP-2–induced Tric-b knockout osteoblasts by imaging cellular Ca2+ signals with the Ca2+-sensitive dye Fura-2. Activation of G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptors, such as the endothelin and purinergic P2Y receptors, stimulates phosphatidylinositol turnover to evoke IP3-induced Ca2+ release from the ER in osteoblasts (26, 27). Although resting Ca2+ concentrations were similar between osteoblasts from wild-type and Tric-b knockout mice in normal and Ca2+-free bathing solutions, Ca2+ transients triggered by endothelin-1 or adenosine 5′-triphosphate (ATP) were significantly reduced in the cells from knockout mice (Fig. 5, A and B). The impaired Ca2+ signal was unlikely to be due to defective activation of the signaling cascade upstream of IP3R because the abundance of transcripts encoding major components of the IP3 signaling cascade, for example, P2y1 and P2y2 (encoding the purinergic receptor subtypes), Plcb1 and Plcb3 (phospholipase C subtypes), and Ip3r1-3 (IP3R subtypes), was similar in cells from knockout and wild-type mice (fig. S3B).

Fig. 5 Compromised Ca2+ handling in Tric-b knockout osteoblasts.

Cultured osteoblasts were subjected to Fura-2 ratiometric imaging after maturation in differentiation medium for 10 days. (A) Endothelin-1 (ET-1)–evoked transients and ionomycin (IM)–induced responses in osteoblasts from WT and Tric-b KO mice. (B) ATP-evoked transients (ATP) and IM-induced responses. (C) Thapsigargin (TG)–induced transients and SOCE responses. (D) IM-induced transients and SOCE responses. Representative recording traces from cell preparations are shown, and the average Ca2+ responses in the preparations derived from different mice were statistically analyzed. In the bar graphs, the data represent means ± SEM, and the numbers of cell preparations derived from different mice are shown in parentheses. Statistical differences between the genotypes are marked with asterisks (*P < 0.05 and **P < 0.01, t test).

To assess whether store-operated calcium entry (SOCE) was altered in osteoblasts from Tric-b knockouts, we induced store Ca2+ depletion with the ER-localized Ca2+ ATPase (SERCA) inhibitor thapsigargin or the Ca2+ ionophore ionomycin in Ca2+-free solution and then added extracellular Ca2+ to enable SOCE (Fig. 5, C and D). We also assessed store Ca2+ contents as the amount of cytosolic Ca2+ that accumulated in cells in response to thapsigargin- and ionomycin-induced store depletion in the absence of extracellular Ca2+. Although SOCE responses were similar between osteoblasts from wild-type and Tric-b knockout mice, Ca2+ responses induced by both thapsigargin or ionomycin were increased in osteoblasts from the knockouts (Fig. 5, A to D). Thus, given that resting Ca2+ concentrations were not altered, cells from the knockout mice seemed to store more Ca2+ in the ER than did cells from wild-type mice, but did not have impaired SOCE. On the basis of these observations, we propose that Tric-b ablation inhibited IP3R-mediated Ca2+ release, which would not only impair Ca2+ signaling but also lead to Ca2+ store overload in osteoblasts.

Link between Ca2+ release, ER morphology, and collagen production in osteoblasts

We administered recombinant retroviruses carrying Tric-a or Tric-b cDNA (complementary DNA) (fig. S4A) to osteoblast precursors cultured from Tric-b knockouts 7 days after the addition of BMP-2 (3 days before full maturation). As expected, treatment with retroviruses carrying Tric-b rescued the defects observed in cells from the knockout mice: store Ca2+ handling became nearly normal (fig. S4B), and both ER morphology and collagen production were significantly improved (fig. S4, C and D). In contrast, compromised Ca2+ handling, ER dilation, and impaired collagen production continued in the cells infected with empty viruses or those carrying Tric-a. Therefore, TRIC-A channels cannot substitute for TRIC-B to support IP3R-mediated Ca2+ release from intracellular stores in osteoblasts. This observation is consistent with our previous studies suggesting that TRIC-A and TRIC-B channels preferentially facilitate RyR- and IP3R-mediated Ca2+ release, respectively (9).

We tested whether pharmacological inhibition of IP3Rs phenocopied the defects observed in osteoblasts from Tric-b knockout mice. We used the IP3R blockers 2-aminoethoxydiphenylborate (2-APB) and xestospongin C (XeC), although both compounds can affect multiple Ca2+-handling proteins, including SERCA and cell surface Ca2+ channels (28, 29). We applied each blocker individually to cultured wild-type osteoblasts for 36 hours before the end of the 10-day maturation period and then subjected the cells to analysis. In Fura-2 imaging, osteoblasts exposed to 30 μM 2-APB or 3 μM XeC had reduced endothelin-induced Ca2+ responses and increased store Ca2+ contents (Fig. 6A). By immunocytochemical analysis, we observed an increase in the frequency of ER dilation (Fig. 6B) and significant reductions in collagen production (Fig. 6C) in the cells exposed to the IP3R blockers. Therefore, ER morphology and collagen synthesis are linked to handling of Ca2+ stores. On the basis of the observations using the blockers and retroviruses, we concluded that Tric-b deficiency inhibits IP3-induced Ca2+ release, thus increasing store Ca2+ content, which results in impaired collagen trafficking, collagen accumulation in the ER, and ER dilation in osteoblasts.

Fig. 6 Impaired osteoblastic functions by IP3R blockers.

Cultured WT osteoblasts were treated with the IP3R blockers 2-APB and XeC for 36 hours before full maturation and then subjected to analysis. (A) Weakened IP3R-mediated Ca2+ release in osteoblasts treated with 2-ABP, XeC, or vehicle [dimethyl sulfoxide (DMSO)]. In the blocker-treated groups, >97 cells in culture preparations derived from three different mice were analyzed, and ET-1–evoked transients and IM-induced responses were averaged. (B) Trends toward ER dilation in osteoblasts treated with 2-ABP or XeC. ER elements were visualized using an antibody that recognizes KDEL, and >795 cells from three different mice were analyzed to detect enlarged ER vesicles (>5 μm in diameter, arrowheads) for each treatment group. Scale bars, 20 μm. (C) Immunostaining for matrix collagen. Staining intensities were quantified photometrically. In the bar graphs, the average data from osteoblast cultures derived from different mice were statistically analyzed; the data represent means ± SEM, and the numbers of mice examined are shown in parentheses. Statistical differences between the genotypes are marked with asterisks [*P < 0.05 and **P < 0.01, analysis of variance (ANOVA) and Dunnett’s test in comparison with control DMSO treatments].

Properties of cultured Tric-b knockout osteoclasts

Promyeloid precursors derived from hematopoietic stem cells differentiate into osteoclasts, macrophages, or dendritic cells in response to different signaling factors. Osteoclastogenesis depends on receptor activator of nuclear factor κB ligand (RANKL), which is present on the cell membrane of osteoblasts (25). To prepare cultured osteoclasts, we prepared fetal liver cells and propagated macrophage-osteoclast precursors in the presence of macrophage colony-stimulating factor (M-CSF). We then stimulated differentiation of these cells into osteoclasts by applying a soluble form of RANKL to the culture medium. Osteoclasts differentiated from fetal liver cells from Tric-b knockout or wild-type mice exhibited similar growth profiles in culture (fig. S3A, right panel). During RANKL-induced differentiation, osteoclasts generate Ca2+ oscillations and undergo full maturation by activating the Ca2+-dependent calcineurin–nuclear factor of activated T cell (NFAT) pathway (30). We detected Tric-b expression in osteoclasts differentiated from wild-type fetal liver cells by qRT-PCR but did not detect Tric-b in the cells from the knockout animals (fig. S3B). Osteoclasts differentiated from either genotype exhibited similar Ca2+ oscillations and ionomycin-induced responses by Fura-2 imaging (fig. S5). The presence of regular Ca2+ oscillations suggested that calcineurin-NFAT signaling is activated normally in the cells from the knockout mice. TRAP-positive osteoclasts differentiated from wild-type or Tric-b knockout fetal liver cells in a manner that depended on the dose of RANKL, and these cells were morphologically normal (Fig. 7A). Moreover, mature osteoclasts differentiated from knockout or wild-type liver cells exhibited similar activities in a resorption assay (Fig. 7B). Therefore, osteoclast differentiation, cytomorphogenesis, or bone resorption was independent of TRIC-B.

Fig. 7 Regular features in Tric-b knockout osteoclasts.

(A) TRAP staining of cultured osteoclasts. Cultured cells derived from fetal livers were stimulated in differentiation medium with different RANKL doses (0 to 50 ng/ml) for 3 days, and the resulting mature osteoclasts were subjected to TRAP staining. The numbers of TRAP-positive multinuclear cells (MNCs) were counted in cell preparations from several different mice for quantification. (B) Resorption activities of cultured osteoclasts. The fetal liver cells were seeded on Osteo Assay plates in differentiation medium for 3 days to monitor resorption activities. The plates were photographed after washing out the osteoclasts, and representative resorption pits are shown. TRAP-positive MNCs and pit resorption area were photomicroscopically determined and averaged in each cell preparation for quantification. Scale bars, 300 μm (image panels). In the bar graphs, the data represent means ± SEM, and the numbers of mice used for cell preparations are shown in parentheses. No significant differences are detected between the genotypes in t tests.


In bones from Tric-b knockout mice, impaired ossification is associated with insufficient collagen matrix. Tric-b ablation impairs collagen deposition by inhibiting the transport of collagen from the rER to the Golgi in osteoblasts. Tric-b knockout osteoblasts also have impaired IP3R-mediated release of Ca2+ from the ER, resulting in store Ca2+ overloading. Both impaired Ca2+ release and store overload have also been observed in alveolar epithelial cells from Tric-b knockout mice (14). Moreover, the defects in Ca2+ handling, ER morphology, and collagen production were rescued by the expression of Tric-b, but not Tric-a, indicating that TRIC-B channels have a subtype-specific function in osteoblasts. Biochemical assays indicated that bone matrix vesicles were produced normally by Tric-b knockout osteoblasts, indicating that defective or insufficient matrix vesicles did not cause the impaired ossification. Furthermore, it is also unlikely that excess resorption by hyperactive osteoclasts plays a role in the OI-like phenotype because we observed no obvious abnormalities in Tric-b knockout osteoclasts in vivo or in culture. Mutations in various genes have been reported to cause OI in different pedigrees, and divergent types of OI are classified by overlapping but distinct sets of phenotypes. However, all forms of OI are associated with altered bone matrix, mainly caused by insufficient or defective deposition of collagen (15, 16). Our data indicate that the form of OI caused by mutations in TMEM38B likely affects collagen deposition by osteoblasts.

ER Ca2+ overloading may directly inhibit collagen release by Tric-b knockout osteoblasts by affecting the activities of ER-resident chaperones and processing enzymes, many of which function in a Ca2+-dependent manner (31, 32). If these become aberrantly activated or inactivated under Ca2+-overloaded conditions, this could interfere with posttranslational processing within or trafficking of collagen from the ER, leading to the accumulation of immature procollagen fibers in the rER. Alternatively, insufficient Ca2+ release may directly inhibit vesicular trafficking because the Ca2+-binding protein ALG-2 may promote COPII assembly for ER-Golgi transport in a Ca2+-dependent manner (33, 34). In Tric-b knockout osteoblasts, impaired Ca2+ release could, therefore, inhibit the assembly of COPII-coated vesicles, leading to ER retention of collagen. The accumulated collagen may exacerbate ER stress as well as ER dilation.

ER dilation is an ultrastructural characteristic indicative of severe Ca2+ overloading and ER stress (3537). Tric-b ablation disrupted either collagen maturation or intracellular trafficking, or both, in osteoblasts. Excess collagen retention may promote ER stress and induce the unfolded protein response (UPR) in the knockout osteoblasts, which is evident by the dilated rER elements. Consistent with ER stress and induction of the UPR, expression of the UPR-related gene Bip was increased in cultured Tric-b knockout osteoblasts (fig. S3), although only a small population of the cultured cells had enlarged intracellular collagen deposits. We also detected increased expression of the UPR-related gene Derl3, which encodes a putative component of the ER-associated degradation system, in bones from Tric-b knockout mice (Fig. 2). During bone development, active osteoblasts normally exhibit a moderate UPR, which stimulates the expression of genes encoding ER chaperones to facilitate collagen folding and processing, and this UPR is essential for efficient collagen production (38). However, the increased frequency and severity of ER dilation in osteoblasts from Tric-b knockout mice suggested that Tric-b ablation aggravated the ER stress to pathological or dysfunctional levels. In addition, severe ER stress can suppress specialized cellular functions to preserve basic housekeeping functions. For example, intensive UPR induction reduces cellular translation efficiency through the phosphorylation of the initiation factor eIF2 (37). Therefore, increased ER stress may disrupt migration and functional maturation of osteoblasts in the knockouts. These defects would indirectly affect osteoblast-dependent osteoclastogenesis. As indicated by the narrow diaphyses and retention of osteoblasts underneath the priosteum, overall ossification processes were delayed during the E16.5-P0 development in Tric-b knockout mice. Therefore, aggravated ER stress in osteoblasts could account for the delay of diaphyseal expansion in bones from the perinatal knockout mice.

In conclusion, our data establish Tric-b knockout mice as a model for studying the bone defects of OI. These mice could provide a unique experimental system for further study of the pathophysiological details of OI, exploring effective treatments for the bone-related symptoms of OI, and developing new drugs for bone-related diseases.


Anatomical analyses

All experiments in this study were conducted with the approval of the Animal Research Committee according to the regulations on animal experimentation at Kyoto University. Tric-b knockout mice were generated and genotyped as described previously (10). For skeleton preparations, mice were fixed in 70% ethanol and then eviscerated. After skin and fat were removed, the specimens were treated with 0.01% Alcian blue (Nacalai Tesque) and 0.001% Alizarin red (Wako Pure Chemical). For histological analysis, mouse bones were fixed in 10% formalin and embedded in paraffin to prepare the tissue sections. Mineralized portions of bone were visualized using the Von Kossa Method for Calcium Kit (Polysciences Inc.), and ALP-, TRAP-, and Safranin O–stained samples were generated using the TRACP & ALP Assay Kit (Takara). Collagen matrix was stained using the Picrosirius Red Stain Kit (Polysciences Inc.). For hematoxylin and eosin staining, the sections were stained with Mayer’s hematoxylin solution (Muto Pure Chemicals) and 0.25% eosin (Wako Pure Chemical). Microscopic images were quantitatively analyzed using ImageJ software (U.S. National Institutes of Health). EM analysis was performed using a JEM-200CX (JEOL) transmission electron microscope as described previously (14).

Micro-CT analysis

Femoral bones were fixed in 70% ethanol and subjected to micro-CT scanning using a Scan Xmate-L090 (Comscantechno Ltd.) in the Kureha Special Laboratory (Tokyo, Japan). Three-dimensional (3D) structural images were reconstructed using the TRI/3D-BON software (RATOC Systems Ltd.) based on regional bone mineral density, and then cross-sectional images of the mid-diaphyseal region were produced from the 3D reconstructions.

Antibodies and immunochemistry

Target proteins of the primary antibodies used in this study were annexin V (GeneTex, GTX113384), type I collagen (Novus, NB600-408), KDEL (MBL, M181-3), GM130 (BD Biosciences, 610823), and GAPDH (Sigma-Aldrich, G9545). For Western blotting, the primary antibodies were used at the following concentrations: 1:2000 (annexin V), 1:500 (type I collagen), and 1:1000 (GAPDH). Secondary antibodies used were goat anti-rabbit IgG (immunoglobulin G) with HRP (horseradish peroxidase) (Agilent Technologies, 1:2000), goat anti-rabbit IgG with Alexa Fluor 488 (Thermo Fisher Scientific, 1:2000), and goat anti-mouse IgG with Alexa Fluor 555 (Thermo Fisher Scientific, 1:2000). Western blot and immunocytochemical analyses were performed as described previously (14).

Matrix vesicles

Bone matrix vesicles were prepared and analyzed as described previously (39). Briefly, upper and lower limb bones were aseptically collected, minced into small pieces, and then collagenase-digested (200 U/ml, 2 hours). The tissue suspension was passed through a nylon filter and then centrifuged to remove tissue debris. The resulting supernatant was subjected to ultracentrifugation to collect matrix vesicles. The ability of the recovered vesicles to generate mineral deposits was determined as described previously (40). The matrix vesicles were suspended in a medium containing ascorbic acid (50 μg/ml) and 2 mM β-glycerophosphate and incubated for 5 days. The amorphous precipitates that formed were fixed with a solution containing 2% paraformaldehyde, 2.5% glutaraldehyde, and 0.1 M sodium cacodylate (pH 7.4), and dehydrated. The fixed specimens were mounted on aluminum holders with adhesive carbon tape and examined using an energy dispersive x-ray spectrometer (EDAX Genesis, Ametec) equipped with a field emission scanning EM (JSM-7400F, JEOL). The percentage of Ca and P in the precipitates was calculated using commercial software (AUTO-ZAF).

qRT-PCR analysis

qRT-PCR analysis was performed as described previously (12). Briefly, total RNAs were extracted from tissues and cultured cells using ISOGEN (Nippongene) and reverse-transcribed using the ReverTra ACE qPCR RT kit (Toyobo). Resulting cDNAs were examined using a real-time PCR system (Thermal Cycler, TP800, Takara). The cycle threshold was determined from the amplification curve as an index for relative mRNA content in each reaction. The qRT-PCR primer sets used in this study are listed in table S1.

Cultured osteoblasts

Primary osteoblasts were isolated, cultured, and analyzed as described previously (41). Briefly, osteoblast precursors were liberated from neonatal calvaria by treatment with collagenase (1 mg/ml) (Wako Pure Chemical) and dispase (2 mg/ml) (Godo Shusei), and tissue debris were removed by filtration through a 160-μm mesh (Tokyo Screen). The isolated cells were grown in a proliferation medium containing 10% fetal calf serum (FCS). After reaching semiconfluency, osteoblasts were replated in a differentiation medium containing BMP-2 (100 ng/ml) (R&B Systems) and 10% FCS for 5 to 10 days to stimulate cellular maturation. The cells were washed out, and the calcium deposits in the culture dishes were stained by 2% Alizarin red for 30 min, and unbound pigment was then washed out with phosphate-buffered saline. Mineralization was quantified by monitoring the absorbance of Alizarin red released in 10% acetic acid. To detect collagen matrix, culture dishes were reacted with an antibody that recognizes type I collagen (Novus, NB600-408) and a secondary antibody conjugated to HRP (P0448, Dako). Immunoreactivity was visualized using commercial kits that included the peroxidase substrate diaminobenzidine (SK-4100, Vector Laboratories Inc.) for microscopic observation or the substrate tetramethyl benzidine (TMB, Moss Inc.) for photometric quantification. To analyze intracellular procollagen contents, cell fractionation was performed as described previously (42). Briefly, cultured osteoblasts were homogenized in a deoxycholate-containing buffer, and then both extracellular matrix and debris were removed by centrifugation. Resulting supernatants corresponding to intracellular fractions were subjected to immunoblot analysis.

Cultured osteoclasts

Cultured osteoclasts were differentiated from fetal liver cells as described previously (43). Briefly, liver cells prepared from E14.5 embryos were cultured for 2 days in a proliferation medium containing M-CSF (10 ng/ml) (PeproTech) and 10% FCS. Osteoclastic maturation was then stimulated in a differentiation medium containing RANKL (5 to 50 ng/ml) (PeproTech), M-CSF (10 ng/ml), and 10% FCS for 2 to 3 days. TRAP-positive cells were stained as described previously (44). For the resorption assay, fetal liver cells were seeded on Osteo Assay Surface 96-well plates (Corning Life Sciences) at a density of 5 × 105 cells/ml and cultured in the proliferation medium for 3 days. After maturation in the differentiation medium for 3 days, adherent osteoclasts were removed using sodium hypochlorite (Coopclean), and the resorption area was determined using a BZ-X710 microscope and BZ-X Analyzer software (Keyence).

Ca2+ imaging

Fura-2 imaging was performed as described previously (14). Briefly, osteoblasts and osteoclasts cultured on glass-bottom dishes were incubated with 10 μM Fura-2AM (Dojindo) for ratiometric imaging. Excitation light of 340 and 380 nm was delivered, and emission light at greater than 510 nm was detected by a cooled CCD (charge-coupled device) camera mounted on a microscope equipped with a polychrometer (Meta Fluor Imaging System, Universal Imaging). The bathing solutions used were physiological salt solutions: 150 mM NaCl, 4 mM KCl, 1 mM MgCl2, 2 mM CaCl2, 5.6 mM glucose, and 5 mM Hepes (pH 7.4) for osteoblasts, and 115 mM NaCl, 5.4 mM KCl, 1 mM MgCl2, 2 mM CaCl2, 10 mM glucose, and 20 mM Hepes (pH 7.4) for osteoclasts.

Retroviral cDNA expression

Mouse Tric-a and Tric-b cDNAs (10) were cloned into the Eco RI/Xho I sites of the retroviral vector pMCs-IRES-GFP (Cell Biolabs Inc.) and introduced into Plat-E packaging cells (Cell Biolabs Inc.) to prepare virus-containing culture supernatants according to the instruction manual. The viral supernatants were concentrated using Retro-X Concentrator (Clontech), supplemented with polybrene (8 μg/ml) (Sigma-Aldrich), and then used for infection of cultured Tric-b knockout osteoblasts 3 days before harvest. Infected cells were identified by GFP (green fluorescent protein) fluorescence.


Fig. S1. Micro-CT analysis and blood mineral measurements.

Fig. S2. Analysis of matrix vesicles.

Fig. S3. Growth and gene expression in cultured osteoblasts and osteoclasts.

Fig. S4. Retroviral rescue trials in Tric-b knockout osteoblasts.

Fig. S5. Ca2+ imaging in Tric-b knockout osteoclasts.

Table S1. Primer sets for qRT-PCR analysis in this study.


Acknowledgments: We thank J. Ma and R. Sitsapesan for valuable suggestions to this work. Funding: This work was supported in part by the MEXT (Ministry of Education, Culture, Sports, Science and Technology)/JSPS (Japan Society for the Promotion of Science) (KAKENHI 15H04676, 26670028, and 15H05652, Core-to-Core Program and Platform for Drug discovery, Informatics and Structural Life Science), Takeda Science Foundation, Kobayashi International Scholarship Foundation, Nakatomi Foundation, Salt Science Research Foundation, Vehicle Racing Commemorative Foundation, the Keihanshin Consortium for Fostering the Next Generation of Global Leaders in Research (K-CONNEX), established by Human Resource Development Program for Science and Technology, MEXT, and Japan Foundation for Applied Enzymology. Author contributions: C.Z., A.I., and N.Q. are equally contributing first authors. K.Y., A.S., and F.A. were responsible for histological analysis. S. Komazaki and C.S. were responsible for EM analysis. C.Z., A.I., N.Q., T.I., D.Y., S. Kakizawa, and M.N. conducted biochemical and cell physiological analysis. C.Z., A.I., and D.Y. conducted cultured osteoblast analysis. N.Q., N.N., and M.A. conducted cultured osteoclast analysis. H.T. oversaw this project. A.I. and H.T. drafted the manuscript, and all authors reviewed the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: There are no material transfer agreements or restrictions. All data and materials are freely available.
View Abstract

Navigate This Article