The bones of the vertebrate skeleton are formed by two distinct processes. Most bones of the axial and appendicular skeleton develop by endochondral ossification, whereas the flat bones of the skull and parts of the clavicle and facial bones are formed by intramembranous ossification. Both processes are initiated by mesenchymal cells that condense, forming templates of the future skeletal elements.
During intramembranous ossification (reviewed in Eames and colleagues1 and Karsenty and colleagues2), signals from epithelia adjacent to the condensations induce the expression of Runx2, which promotes osteoblast differentiation and directly upregulates osteogenic genes like Osterix/Sp7 (Osx), Collagen I, and Osteocalcin/Bglap (Osc).3–8 With ongoing maturation, osteoblasts synthesize high amounts of extracellular matrix mainly composed of Collagen I and proteoglycans. This specialized matrix, the osteoid, serves as a scaffold for the deposition of hydroxyapatite crystals that convert the elastic protein matrix into calcified bone tissue. Osteoblasts that become embedded within the bone terminally differentiate into osteocytes, which play crucial roles during bone homeostasis.
Cells within endochondral condensations are committed to the chondrocyte lineage characterized by the expression of Sox9 and Collagen II (Col II). The outer cells of the condensations build the perichondrium and later differentiate into osteoblasts. The chondrocytes initially proliferate until they exit the cell cycle and become hypertrophic. Hypertrophic chondrocytes produce a mineralized matrix containing Collagen X (ColX) instead of Collagen II. In addition, early hypertrophic chondrocytes express Indian hedgehog (Ihh), which induces the expression of Runx2 in adjacent perichondrial cells. Like in intramembranous bones, Runx2-expressing perichondrial cells differentiate into osteoblasts, thereby converting the perichondrium into a mineralized periosteum, the bone collar, which will later give rise to the cortical bone.9, 10 In parallel to the onset of osteoblast differentiation, vascular endothelial growth factor (VEGF) expressed in hypertrophic cells initiates the invasion of blood vessels into the periosteum and subsequently the center of the hypertrophic region. With vascular invasion, osteoblasts and osteoclasts enter the cartilage anlagen and start to replace hypertrophic cartilage with trabecular bone and bone marrow.
In mice and human, Anoctamins form a family of 10 multitransmembrane proteins that show distinct but overlapping expression patterns in a variety of tissues and cell types during development11–15 (reviewed in Flores and colleagues16 and Galietta17). Recently, Anoctamin1/Tmem16A (Ano1) has been identified as a calcium-activated chloride channel acting in several tissues and cell lines.18–20 Studies on the electrophysiological properties of other family members revealed that Ano2, Ano6, Ano7, Ano8, and Ano9 can also confer chloride conductance in cell culture, but in vivo the channel function of these proteins has not been demonstrated yet.14, 19, 21–24 Interestingly, in addition to its potential function as a chloride channel, Ano6 has recently been identified as a regulator of phosphatidylserine (PS) scramblase activity.25
In the developing skeleton, Ano1, Ano3, Ano5, Ano6, and Ano9 are expressed on mRNA level, but a detailed analysis of the skeletal expression pattern in situ is only available for Ano5, which is expressed in articular and hypertrophic chondrocytes and in osteoblasts.12 Mice deficient for Ano1 suffer from tracheomalacia and die shortly after birth because of respiratory failure, but have not been reported to exhibit a skeletal phenotype.26, 27 Mouse models for the other anoctamins are lacking so far. Interestingly, mutations in the human ANO5/GDD1 gene cause gnathodiaphyseal dysplasia (GDD), an autosomal dominant inherited disease. GDD is characterized by increased bone fragility, fibro-osseous lesions of the jaw bones, and sclerosis and deformations of tubular bones supporting a role for Ano5 in skeletal development.28
Here we have analyzed the function of Ano6, the closest paralog of Ano5, during skeletal development. We have identified Ano6 in a subtractive hybridization screen as a potential target gene of Ihh signaling during endochondral ossification (unpublished data). We found that in the skeletal system, Ano6 is primarily expressed in osteoblasts, where it acts downstream of Runx2 and Osx. Deletion of Ano6 in mouse embryos results in reduced skeleton size and skeletal deformities. A molecular analysis of chondrocyte and osteoblast differentiation showed that these cell types develop normally in Ano6−/− mutants. However, Ano6-deficient osteoblasts display a reduced capacity to produce a calcified bone matrix in vivo and in vitro. Furthermore, we could demonstrate that phospholipid scrambling is disturbed in Ano6−/− osteoblasts, strongly supporting an Ano6-dependent, critical role of PS scrambling in the deposition of hydroxyapatite during bone mineralization.
Materials and Methods
Generation of Ano6-deficient mouse lines
To generate the Ano6 targeting construct, a PAC clone (clone 379, PAC library RPCL21)29 containing exon 1 and flanking genomic sequences of the Ano6 gene, was identified by filter hybridization, Southern blot and PCR. A 3.5-kb 5' targeting fragment was amplified by PCR (XhoI-Ano6-5' arm-fw 5'-gaaaactcgagtggagctgttacttgg and Ano6-5' arm-XhoI-rv 5'-aaaactcgagcttctatgagctcacgagggtc) and cloned into the XhoI site of the targeting vector pWH9. For the 3' homology arm, two DNA fragments of 4.3 kb (SalI-Ano6-3' arm-fw 5'-aaattgtcgactgagctgtctgtaacttggga/Ano6-3' arm-BamHI-rv 5'-tatttggatccaaagcatcaagaagagagaatgag) and 3.6 kb (BamHI-Ano6-3' arm-fw 5'-gctttggatccagatgtagaactctcagttccttc/Ano6-3' arm-XbaI-rv 5'-tcacatctagaatacaaactgcaccaaaggtgg) were amplified by PCR and subcloned into pBSII SK (Agilent, La Jolla, CA, USA). The resulting 8-kb fragment was cloned via SalI and NotI into the targeting construct. The construct was linearized and electroporated into embryonic stem (ES) cells from SV129/J mice. Homologous recombination resulted in the replacement of exon 1 by the neomycin resistance cassette (Fig. 2A). Positive clones were identified by Southern blot hybridization of HindIII (5' probe) or EcoRV (3' probe) digested ES cell DNA using 32P-labelled probes Ano6-5' (Ano6-5'-probe-fw 5'-cctagaacatgcccttcagatac/Ano6-5'-probe-rv 5'-tccagtatggtgagccagtgag) and Ano6-3' (Ano6-3'-probe-fw 5'-aaatgcacgcaagattatgct/Ano6-3'-probe-rv 5'-actggacctggaaacctgaaag), which were generated by PCR. Two ES cell clones showing correct integration were used for injection into C57Bl6 blastocysts. Analyzed mice were backcrossed to C57Bl6 for at least six generations. Both mouse lines showed identical phenotypes. All animal studies were undertaken according to the institutional guidelines. Mice were genotyped by PCR using pimer pairs specific for the mutant (Ano6ko-fw 5'-ccttctatcgccttcttgac/Ano6ko-rv 5'-ctcctccttacttctaagacatc) and wild-type (Ano6wt-fw 5'-ttctctgtcctctttcggt/Ano6wt-rv 5'-gacttacactatactcgaca) allele. To generate Ano6gt/gt mice, the ES cell line RRF355 harboring a gene trap insertion cassette in intron 13 of the Ano6 gene was obtained from Bay Genomics (San Francisco, CA, USA, now hosted by Mutant Mouse Regional Resource Centers, MMRC) and injected into C57Bl6 blastocysts. Chimeric mice were backcrossed to C57Bl6, and mice were genotyped by PCR using primers Ano6-ex13gtfw (5'-ggtcacgctgtgtgcgagc) and Ano6gtvec70-R/RRF-rv (5'-cactccaacctccgcaaactc).
RT-PCR and qPCR
Total RNA was prepared from cells or tissues using Trizol (Life Technologies GmbH, Darmstadt, Germany) and cDNA generated using SuperscriptII (Life Technologies), according to the manufacturer's protocol. For semiquantitative and analytical PCR, the following oligonucleotides were used: Osc: 5'-ccgggagcagtgtgagctta and 5'-tagatgcgtttgtaggcggtc; Opn: 5'-ttcacagccacaaggacaag and 5'-ttacaacggtgtttgcatga; Ano6: ex8-rv 5'-agggaacgctgctttgtaga ex15-fw 5'-aactgaccacacagctgacg ex20-rv 5'-gggtttcggaagtcacgg. Quantitative RT-PCR was carried out using a Taqman 7500 (Applied Biosystems, Darmstadt, Germany) and SYBR Green reaction mix (Applied Biosystems). RNA from multiple tissues was isolated from ground shock-frozen tissue samples from 6-week-old C57BL/6 mice by the Trizol (Life Technologies) method.
In situ hybridization, histology and skeletal preparations
Formalin-fixed, paraffin-embedded 7-µm sections were stained by the von Kossa method as described in Kesper and colleagues.30 β-galactosidase was detected on sections or formalin-fixed whole-mount embryos as in Ratzka and colleagues.31 Alkaline or tartrate-resistent acid phosphatase (TRAP) were detected using the Leukocyte Alkaline Phosphatase or Acid Phosphatase, Leukocyte (TRAP) kits (Sigma, St. Louis, MO, USA). Skeletal preparations were prepared as described in Ratzka and colleagues.31 For in situ hybridizations, sections were hybridized with 33P-labelled riboprobes as described in Minina and colleagues.32 The following probes were used: Ihh,33Col10a1,34Osx,7Ibsp,35 and Osc.36 The Ano6 probe was amplified by PCR using oligonucleotides BamHI-Ano6ish-fw 5'-aaaggatcctcgcacgctgagcagcgcctg and Ano6ish-EcoRI-rv 5'-tatttggatccaaagcatcaagaagagagaatgag and cloned into pBSII SK.
Histomorphometry and micro-computed tomography (µCT)
Quantitative analysis of femur and spine at postnatal stages was performed by µCT. Scanning and bone mass measurements were performed on a Scanco Medical (Brüttisellen, Switzerland) vivaCT40 with 10-µm voxel size. For histological analysis, dissected bones were fixed in phosphate-buffered 4% paraformaldehyde. After ethanol dehydration, samples were embedded in methylmethacrylate37 and sectioned at 5 µm with a hard tissue microtome (Leica, Wetzlar, Germany). Sections were stained with von Kossa/picrofuchsin or von Kossa/toluidin. Osteoid measurements were performed using the Osteomeasure (OsteoMetrics, Decatur, GA, USA) software. Histomorphometry on embryonic sections was carried out as described in Wuelling and colleagues.38
Primary calvarial osteoblast cultures
Frontal and parietal parts of the calvaria were dissected from embryonic day (E) 18.5 embryos and primary osteoblasts isolated by two 30-minute digestion steps in 0.1% Collagenase IV (Sigma). Cells from individual animals were cultured separately, seeded at a density of 104/cm2, and cultured in DMEM supplemented with 10% fetal bovine serum and penicillin/streptomycin. Upon confluency, cells were reseeded at a density of 104/cm2 and cultured in the presence of 10 mM glycerolphosphate and 5 mM ascorbate phosphate to induce osteogenic differentiation and mineralization (day 0). At culture days 4, 7, and 14, triplicate wells of each culture were fixed in methanol and washed in deionized water. Mineralization was visualized by alizarin red staining, which was subsequently quantified as described.39 Briefly, alizarin red S was extracted from the cell layer in 10% acetic acid, cell debris removed by centrifugation, and quantified at 405 nm in a microplate reader. Alkaline phosphatase activity of triplicate cultures was determined as described.40 Cells were lysed in ALP lysis buffer (100 mM glycine, 1% NP-40, 1 mM MgCl2, 1 mM ZnCl2, pH 9.6) and cell debris removed by centrifugation. One volume of ALP buffer (100 mM glycine, 5 mM p-nitrophenyl phosphate, 1 mM MgCl2, and 1 mM ZnCl2, pH 9.6) was added, lysates incubated for 30 minutes at 37°C, and p-nitrophenol release determined at 405 nm. Cells in S phase were determined by flow cytometry after 5-Bromo-2-deoxyuridine (BrdU) incorporation using the FITC BrdU Flow kit (BD Pharmingen, San Diego, CA, USA). Measurements were conducted on a FACScalibur flow cytometer and data analyzed with FlowJo 7.6 (Treestar, Ashland, OR, USA).
Flow cytometric detection of PS exposure
To induce PS translocation, primary osteoblasts were seeded at a density of 104 cells/cm2 and cultured for 24 hours. Cells were harvested by trypsinization, washed first in growth medium and then in Hank's Balanced Salt Solution (HBSS; Invitrogen) supplemented with 0.5 mM CaCl2 at 4°C. Subsequently, cells were treated with 10 µM ionomycin or DMSO for 1 or 5 minutes at 37°C. Ionomycin was removed by washing in ice-cold HBSS. Treated cells were then washed in AnxA5 staining buffer (10 mM HEPES, 150 mM NaCl, 2.5 mM CaCl2, pH 7.4) and stained with Dye490-labelled recombinant AnxA5 and propidium iodide (PI) as described.41 Flow cytometric analysis was conducted using a FACSCanto2 (Beckton Dickinson, Heidelberg, Germany) and data processed with FlowJo 7.6 (Treestar).
For assays conducted with primary calvarial osteoblasts, cultures from at least five animals per genotype that were recovered from three independent litters were analyzed. Cells from individual animals were maintained as individual cell lines. Mineralization, alkaline phosphatase, and metabolic assays were carried out in triplicate measurements for each assay, cell line, and time point. Statistical analyses were conducted by the Wilcoxon rank sum test using the RGui software version 2.9.1 (The R Foundation for Statistical Computing, Vienna, Austria), except histomorphometric and flow cytometric analyses that were tested by two-tailed, unpaired Student's t test. In all cases, p values less than 0.05 were considered significant.
Expression of Ano6 in skeletal tissues
To receive first insight into the role of Ano6 during skeletal development, we studied its expression in skeletal tissues during mouse embryogenesis. In situ hybridization (ISH) on sections of mouse embryos showed that Ano6 mRNA is most prominently expressed in ossifying tissues of endochondral and intramembranous bones. At E14.5 and E16.5, Ano6 expression is detected in differentiating osteoblasts of the perichondrium/periosteum of endochondral bones including vertebrae, ribs and long bones, and in intramembranous bones of the skull like calvaria and facial bones (Fig. 1A–E). At later stages, when chondrocytes in the center of endochondral bones are being replaced by bone, Ano6 is also expressed in the primary ossification centers (Fig. 1D, E).
For a detailed analysis of the Ano6 expression profile during osteoblast differentiation, we cultured primary mouse calvarial osteoblasts for 4 weeks in DMEM or DMEM supplemented with glycerolphosphate and ascorbate to induce differentiation. Semiquantitative RT-PCR analysis showed that Ano6 is expressed in freshly isolated osteoblasts as is the early osteoblast marker Osteopontin (Opn) (Fig. 1K). During 4 weeks in culture, when the cells differentiate into mineralizing, Osteocalcin (Osc)-expressing osteoblasts, Ano6 expression levels are not changed significantly, indicating a function of Ano6 throughout osteoblast differentiation and maturation.
Postnatally, we used quantitative real-time PCR (qPCR) to detect Ano6 in tissues from 6-week-old mice and cultured cells. This revealed highest Ano6 expression in the lung, skin, and long bones, whereas it is weakly expressed in calvaria and cortical bone, which mainly consists of osteocytes, and nearly absent from brain, eye, and muscle tissue (Fig. 1L). In cultured cells of the skeleton, primary undifferentiated and differentiated osteoclasts as well as the osteocyte-like cell line MLO-Y4 show markedly lower Ano6 expression than osteoblasts, indicating that, similar to embryonic development, Ano6 primarily functions in bone-producing cells.
As Ano6 has been identified as a potential target of Ihh signaling, we analyzed the expression of both genes on parallel sections of E16.5 forelimbs. We found that the distal border of Ano6 expression in the perichondrium/periosteum correlates with the border of Ihh-expressing cells in early hypertrophic chondrocytes (Fig. 1E, F). No Ano6 expression could be detected in regions of the perichondrium adjacent to proliferating chondrocytes. A similar expression in early differentiating osteoblasts of the perichondrium/periosteum has been described for Runx2 and Osx, which are induced by Ihh signaling.9 To receive insight into the epistatic relationship of Ihh and Ano6, we investigated Ano6 expression in E16.5 Ihh−/− mutants but could not detect Ano6 expression at this stage (Fig. 1G, H). To investigate if this is because Ihh mutants do not form osteoblasts, we analyzed Ano6 expression in forelimbs of E16.5 Runx2−/− and Osx−/− mutants. We could not detect any Ano6 expression in either mutant. We can thus conclude that Ano6 expression is initiated downstream of the differentiation-inducing factors Ihh, Runx2, and Osx (Fig. 1G, I, J). If these directly regulate Ano6 expression on a transcriptional level or if other factors are required has to be investigated in future studies.
Generation of Ano6-deficient mouse lines
The murine Ano6 coding sequence comprises 20 exons spanning 183.9 kb on mouse chromosome 15 (Fig. 2A). By automated sequence analysis (http://www.ensembl.org release 61), Ano6 has been predicted to encode two alternative transcripts. The full-length transcript (Ano6-201, ENSMUST00000071874) is composed of 20 exons encoding a protein of 911 amino acid residues with eight predicted transmembrane domains (TMHMM topology prediction).42 RT-PCR with oligonucleotides spanning exons 1 to 20 followed by sequence analysis confirmed the expression and predicted exon composition of the full-length Ano6-201 transcript (data not shown). RT-PCR amplifying overlapping cDNA regions of 2 to 5 exons did not reveal evidence for major alternative splicing (data not shown). In contrast to these in vitro data, in silico sequence analysis predicts a second alternative transcript (Ano6-202, ENSMUST00000109239), which starts in a putative exon 12a of 30 nucleotides and includes the C-terminal 484 amino acid residues of Ano6 (Fig. 2A).
Considering the predicted complexity of the gene, we followed two approaches to generate mice lacking functional Ano6 protein. First, mouse ES cells carrying a gene trap insertion cassette in intron 13 were used to generate Ano6gt/gt mice (Fig. 2A). In these mice, the gene trap insertion should disrupt both predicted Ano6 transcripts and delete the six C-terminal transmembrane domains. β-galactosidase (β-gal) staining of E14.5 and E16.5 Ano6gt/+ embryos and forelimb sections showed that the transgene is expressed in the mineralizing areas of skeletal tissues as calvaria, ribs, radius, and ulna, mimicking the expression of the endogenous Ano6 gene (Fig. 2B–E and Fig. 1).
In a parallel inactivation approach, Ano6−/− mice were generated by targeted deletion of exon 1, which contains the transcriptional and translational start of Ano6-201 (Fig. 2A). RT-PCR with primer pairs covering exons 3 to 8 and 15 to 20, respectively, confirmed that Ano6-201 is not expressed in primary osteoblasts of Ano6−/− mice (Fig. 2F). Furthermore, lack of exon 15 to 20 expression indicates that the predicted short Ano6-202 transcript is not expressed in osteoblasts of Ano6−/− mice, at least not at detectable levels (Fig. 2F). To support this result, we investigated Ano6-202 expression in wild-type mice. Using oligonucleotides spanning exons 12a to 20, we could not detect the short Ano6-202 transcript in primary wild-type osteoblasts by RT-PCR, indicating that the short form of the protein has no major function in this cell type (data not shown).
Skeletal phenotype of Ano6 mutant mice
To get an overview of the skeletal phenotype, we stained wild-type and Ano6-deficient embryos with Alcian blue for cartilage proteoglycans and Alizarin red S for mineralized bone. Ano6 mutant embryos from both lines were recovered at Mendelian frequency. Postnatally, the number of recovered Ano6−/− mutants was reduced by approximately 60% (Fig. 2G), and we observed that most Ano6−/− pups died shortly after birth from asphyxiation. In wild-type embryos, mineralization of the skeletal elements is usually well detectable at E15.5. At this stage, Ano6 heterozygous mice of either line displayed no overt phenotype (data not shown), whereas Ano6−/− and Ano6gt/gt embryos are smaller than wild-type littermates and devoid of mineralized bone (Fig. 3A–D). Consequently, Ano6 mutant embryos develop skeletal malformations like shortened and curved limbs and ribs, a narrow rib cage, absence of the olecranon processes, and delayed mineralization of intramembranous skull bones (Fig. 3). At E16.5, the mutants show distinct mineralization in all skeletal elements, but the ossified regions are markedly smaller than in wild-type embryos (Fig. 3E, F). During later stages of development, mineralization progresses, but most skeletal malformations persist and newborns exhibit wider sutures of the skull bones (Fig. 3G and data not shown). In general, Ano6gt/gt embryos display a similar but slightly milder phenotype than Ano6−/− mutants. We thus focused our subsequent analyses on the Ano6−/− strain.
Delayed osteoblast mineralization in Ano6−/− embryos
To further characterize the skeletal defect in Ano6−/− mutants, we analyzed the expression of markers for distinct subpopulations of chondrocytes and osteoblasts on paraffin sections of embryonic forelimbs. Ihh and Col10 are markers for hypertrophic chondrocytes.43, 44 In E14.5 and E16.5 Ano6−/− embryos, the expression domains of both genes are comparable in size and position to that of wild-type embryos, indicating that chondrocyte differentiation is not obviously affected in mutant embryos (Fig. 4A, B and data not shown). To monitor distinct stages of osteoblast differentiation, we investigated the expression of Osx, integrin-binding sialoprotein (Ibsp), and Osteocalcin (Osc). In addition, we analyzed alkaline phosphatase activity (Alp) in Ano6-deficient embryos. At E14.5 and E16.5, Osx is prominently expressed in terminal hypertrophic chondrocytes and in differentiating osteoblasts of the perichondrium/periosteum flanking the hypertrophic region in wild-type embryos (Fig. 4C and data not shown). Ibsp is expressed in early osteoblasts of the periosteum, whereas Osc expression is detected at later maturation stages (Fig. 4D, E). In Ano6−/− embryos, no changes in the expression of either gene could be detected at E14.5 or E16.5 (Fig. 4C–E and data not shown). Similar results were obtained for Alp activity, which is found in comparable patterns in wild-type and mutant embryos (Fig. 4F). Together these data strongly indicate that the osteoblast differentiation program is not altered by loss of Ano6 expression. Tartrate-resistent acid phosphatase (TRAP) staining for osteoclasts on sections of E16.5 forelimbs did reveal no differences in osteoclast differentiation (data not shown).
To investigate the process of osteoblast mineralization, we analyzed sections of embryonic forelimbs by von Kossa staining. Mineralization in wild-type embryos is first detected at E14.5 in a restricted area of the perichondrium/periosteum flanking the central region of hypertrophic cells (Fig. 4G). At E16.5 mineralization was detected in the perichondrium/periosteum flanking hypertrophic chondrocytes and in the newly formed diaphyseal trabeculae (Fig. 4H). In Ano6 mutants, no mineralization could be detected in radius and ulna at E14.5, whereas at E16.5 both the perichondrium and diaphyseal trabeculae are clearly mineralized (Fig. 4G, H). However, compared with wild-type embryos, trabecular von Kossa staining appeared dispersed at E16.5 (Fig. 4G, J) and mineralization of the perichondrium/periosteum was restricted to a smaller region. To further characterize the defect in mineralization, we compared the mineralized region of the periosteum demarcated by von Kossa staining to the region of cells expressing Ibsp, which precedes mineralization.45 On parallel sections of the radius, we measured the distance from the joint to the onset of Ibsp expression and von Kossa-stained mineral deposition, respectively. We found that Ibsp expression is induced at a similar distance from the joint in wild-type and Ano6−/− mutant embryos (Fig. 4I). Compared with the onset of Ibsp expression, the border of mineralized matrix, as detected by von Kossa staining, is located at a slightly larger distance from the joint in wild-type embryos (Fig. 4J). In Ano6−/− mutants, this distance is increased by 71%, displaying an extended domain of Ibsp-expressing osteoblasts that have not yet deposited mineralized matrix (p = 0.00009, n = 10 limbs per genotype; Fig. 4I–K). These data demonstrate that although the differentiation program of osteoblasts seems not to be affected by loss of Ano6, osteoblasts display a reduced capacity to form calcified bone in vivo.
Skeletal phenotype in adult Ano6-deficient mice
To investigate whether the skeletal defects observed in Ano6−/− embryos persist postnatally in surviving mutants, we analyzed bone parameters of femora and vertebrae of 3-week-old and 2-month-old mice. No significant differences in bone volume/tissue volume (BV/TV) or trabecular number (Tb.N), spacing (Tb.Sp), and thickness (Tb.Th) between mutant and wild-type mice were found in µCT investigations, neither in juvenile or adult mice (Fig. 5A, Supplemental Figs. S1A and S2A). Furthermore, histomorphometric analysis of Pichrofuchsin/von Kossa-stained femora sections of 3-week-old mice did not reveal differences in cortical thickness (Ct.Th, Supplemental Fig. S3C). Growth plate width (Gr.Pl.Wi) was also unchanged between wild-type and mutants, demonstrating that, similar as in the embryo, Ano6 deficiency does not affect postnatal growth plate architecture (Supplemental Fig. S3). In contrast, Picrofuchsin/von Kossa staining of trabecular bone from vertebrae and femora indicated expanded layers of nonmineralized osteoid surrounding the trabeculae of the secondary spongiosa (Fig. 5C, D and Supplemental Figs. S1B–E and S2B, D). This finding was further supported by a histomorphometric analysis of osteoid parameters that revealed an increase in osteoid volume/bone volume (OV/BV) of approximately 100% in 2-month-old Ano6−/−-deficient mice (Fig. 5B and Supplemental Fig. S1H). In addition, osteoid surface/bone surface (OS/BS) and osteoid thickness (O.Th) were increased by approximately 30% in mutants of this stage (Fig. 5B and Supplemental Fig. S1H). In 3-week-old Ano6 mutants, osteoid parameters were also elevated, but the data were not yet statistically significant (Supplemental Fig. S2D), indicating that the phenotype manifests over time. To exclude a defect in calcium homeostasis or matrix production, we analyzed blood serum levels of calcium, phosphate, and cleaved procollagen I N-terminal propeptide (PINP) as a marker for collagen synthesis. We could not detect significant alterations in these parameters in adult Ano6−/− mice (Supplemental Fig. S1I and data not shown). We can thus conclude that although bone parameters of Ano6−/− mice are normal, the mutants show an increased amount of unmineralized osteoid, which is not the result of altered calcium homeostasis or matrix production.
Primary calvarial osteoblast cultures
Because mineralization is dependent on the functional integrity of osteoblasts, we next investigated the role of Ano6 during mineral deposition in primary osteoblasts isolated from frontal and parietal skull bones of E18.5 embryos. First, we determined the proliferation rate in the cultures using fluorescence-activated cell sorting (FACS) after Bromo deoxyuracil (BrdU)-labelling. A total of 3.47% (± 2.2%, n = 8) of control and 3.24% (± 1.2%, n = 7, p = 0.8) of Ano6−/− calvarial cells were BrdU+, indicating that there were no significant differences in osteoblast proliferation (Fig. 6A). In addition, the rate of apoptosis was determined by incubating cells with a luminescent substrate for Caspase-3/7. In both wild-type and Ano6−/− cultures, apoptosis was below the detection limit unless triggered by addition of nocodazole, which was used as positive control (data not shown). We then cultured the osteoblasts for up to 14 days under osteogenic conditions and monitored differentiation and mineralization by alkaline phosphatase measurements and Alizarin red S staining at days 0, 4, 7, and 14. As expected from the in vivo data, alkaline phosphatase levels did not differ significantly between wild-type and mutant cultures (p > 0.1; nwt = 7 and nko = 5 for each time point; Fig. 6C and Supplemental Fig. S4A–C). Alizarin red staining showed that after 4 days in culture, wild-type and Ano6-deficient osteoblasts start to form mineralized nodules, producing a comparable amount of calcified matrix (Fig. 6B). Although mineralization increased in both cultures, mineral deposition was markedly reduced in Ano6−/− osteoblasts after 7 and 14 days (Fig. 6B). Acidic extraction and subsequent colorimetric determination of the dye revealed that mineralization of Ano6−/− cultures was reduced to approximately 72% of wild-type cells at day 7 and 56% at day 14 (Fig. 6D and Supplemental Fig. S4D–F). Interestingly, although differences in mineralization were already clearly apparent at culture days 7 and 14, we could not detect statistically significant alterations in the expression levels of the negative regulators of mineralization Ank, Enpp1, Mepe, and Mgp or the positive regulator Dmp1 between wild-type and Ano6-deficient cultures (Supplemental Fig. S5). These results sustain the in vivo findings that Ano6−/− osteoblasts differentiate normally but are impaired in mineral deposition. They furthermore support a cell autonomous role of Ano6 in osteoblasts. In summary, Ano6-deficient osteoblasts proliferate and differentiate normally at early stages but display a cell autonomous, reduced capacity to produce calcified matrix.
Regulation of PS translocation in osteoblasts by Ano6
Mineralization of osteoblasts and chondrocytes is thought to be promoted by the formation of matrix vesicles, which in these cell types are highly enriched in the phospholipid PS.46–50 Furthermore, at least in hypertrophic chondrocytes, the release of mineralization-competent matrix vesicles coincides with the translocation of PS from the inner to the outer leaflet of the plasma membrane.47 This process, denoted as phospholipid scrambling, is regulated by members of the phospholipid scramblase protein family. Interestingly, for the pro B cell line Ba/F3, it has recently been shown that Ano6 regulates inside-out scrambling of PS in response to an increase in intracellular calcium levels.25
Based on these data, the reduced mineralization observed in Ano6−/− mice might be attributed to a reduced scramblase activity in osteoblasts. To test if increased intracellular calcium levels would induce PS scrambling in osteoblasts, we treated primary osteoblasts with the calcium ionophore ionomycin51, 52 and analyzed PS exposure by flow cytometry using fluorescence-labeled Annexin 5 (AnxA5-Dye490), which binds to PS on the cell surface, and PI that is incorporated into the DNA of dead cells. Treatment with DMSO or 1 µM ionomycin had no effect on wild-type or mutant cells (Fig. 7 and data not shown). Exposure of wild-type osteoblasts to 10 µM ionomycin for 1 minute resulted in 23.6% (± 8.6%, n = 5) AnxA5+/Pi– cells representing viable cells that expose PS to the outside. Prolonged treatment for 5 minutes resulted in similar numbers of viable PS-exposing cells (18.2 ± 11.2%, n = 10), but we observed a higher toxicity of the elongated treatment (Fig. 7).
When Ano6−/− osteoblasts were treated with 10 µM ionomycin, no increase in the number of viable PS-exposing AnxA5+/Pi– cells could be observed, neither after 1 minute or after 5 minutes (2.8 ± 1.5%, n = 4 for short and 2.9 ± 2.0%, n = 9 for long treatments), whereas the number of dead Ano6−/− cells was similarly elevated as in wild-type cells (Fig. 7). This demonstrates that in contrast to wild-type osteoblasts, Ano6−/− osteoblasts do not induce PS scrambling in response to raised intracellular calcium levels.
Because a prolonged ionomycin treatment of osteoblasts led to a significant increase in the number of dead cells, it is possible that ionomycin triggers the apoptotic pathway, which is also characterized by PS exposure.53, 54 However, several lines of evidence indicate that PS exposure during apoptosis is not mediated by PS-scramblases in a calcium-dependent manner, but rather results from oxidation of PS by cytochrome c during apoptotic breakdown of intracellular compartments.55–58 Subsequently, oxidized PS cannot be maintained on the cytoplasmic side of the plasma membrane, possibly because of its enhanced diffusion within the membrane.59 To test whether Ano6 specifically regulates calcium-dependent PS scrambling or if it also plays a role during oxidation-induced apoptotic PS exposure, we induced apoptosis in primary osteoblasts by treatment with the protein kinase inhibitor staurosporine.60 At all time points analyzed, approximately 20% of staurosporine-induced wild-type or Ano6-deficient cells were AnxA5+/Pi–, representing early apoptotic cells that expose PS to the outside (Supplemental Fig. S4). In addition, we observed a continuous increase in the number of dead cells over time, indicating that staurosporine induces apoptosis in cells from both genotypes to similar extents (Supplemental Fig. S6). Together, our data demonstrate that Ano6 does not act during apoptotic PS exposure but is specifically required to induce calcium-dependent PS scrambling in osteoblasts.
Ano6−/− mice display a mineralization phenotype
Ano6 has previously been shown to be expressed in various nonskeletal tissues, such as the respiratory system, the gastrointestinal tract, ovary, kidney, and skin (Fig. 1).11, 13, 15 In the current study, we have characterized the role of Ano6 during differentiation and mineralization of the mouse skeleton. Analysis of wild-type and Ano6−/− mice revealed that Ano6 is expressed at early osteoblast differentiation stages. Its expression is maintained in the differentiating and mineralizing osteoblasts, but the low Ano6 levels we detected in the osteocyte-like cell line MLO-Y4 and in cortical bone, in which terminally differentiated osteocytes are the major cell type, indicate that the gene is downregulated in terminally differentiated osteocytes.
Both Ano6-deficient mouse lines we generated exhibited growth defects and deformities of long bones, ribs, and skull. Detailed analysis of mutant embryos revealed that during embryonic development, mineralization is severely delayed in endochondral and intramembranous bones. Postnatally, the skeletal phenotype is comparably mild and bone parameters are not significantly altered. However, bone trabeculae are still surrounded by a widened area of unmineralized osteoid. Together, these results demonstrate a critical function of Ano6 during osteoblast mineralization.
Although some homozygous mutants were recovered postnatally, about 60% died at birth. Thus, surviving mutants are likely to represent individuals with a weakened phenotype. We observed that at least some homozygous mutants have difficulties in breathing and suffocate shortly after birth. This is likely a consequence of the narrow rib cage of Ano6−/− pups. Alternatively, because Ano6 is highly expressed in the respiratory system,26 we cannot exclude defects in tracheal or lung development as has been observed in Ano1 mutants, which develop a cystic fibrosis-like tracheomalacia and die within 3 weeks after birth.26
Interestingly, inactivating mutations in Ano6 have recently been identified in patients with Scott syndrome, a rare, genetic bleeding disorder caused by a defect in PS scrambling to the outer leaflet of the plasma membrane in platelets.25 At present, none of the patients has been reported to display skeletal abnormalities (reviewed in Zwaal and colleagues61). However, according to our findings, they might represent only the mild end of a variable phenotype. It is also possible that Ano6 function differs slightly in humans and rodents. The increased bleeding phenotype might be more prominent in humans, whereas in mice the skeleton is more severely affected. We have not investigated platelet function and bleeding times in Ano6−/− animals in detail, but did not observe evidence for excessive bleeding, e.g., in females after giving birth, one of the major problems in Scott syndrome patients.
How does Ano6 regulate skeletal mineralization?
Expression analysis with chondrocyte and osteoblast markers revealed that in the embryonic skeleton, Ano6 deficiency did not impair the differentiation program of chondrocytes, osteoblasts, or osteoclasts. Instead, a close comparison of van Kossa-stained mineral deposits and Ibsp expression in the periosteum revealed an enlarged region of Ibsp osteoblasts that have not yet deposited hydroxyapatite. Similarly, primary osteoblast cultures show a diminished capacity to calcify the extracellular matrix, indicating that Ano6 plays a specific cell autonomous role in this process.
Earlier studies revealed that, like other members of the anoctamin family, Ano6 might act as a relatively weak calcium-activated chloride channel in the plasma membrane.14, 21 In addition, more recent experiments identified Ano6 as a regulator of PS scrambling in platelets.25 Using the calcium ionophore ionomycin to induce PS translocation in osteoblasts, we show here that primary osteoblasts of Ano6−/− mutants are deficient in calcium-dependent PS scrambling, providing first evidence that Ano6 is required for scramblase activity in vivo. Interestingly, PS translocation in Ano6−/− cells is not affected during apoptosis, demonstrating that Ano6 is not required for PS translocation per se. Furthermore, Ano6 deficiency does not lead to an altered apoptosis rate in osteoblasts, a result in accordance with a recent study, in which Ano6 overexpression resulted in constitutive PS exposure in lymphoma cells but did not induce apoptosis in these cells.62
In the skeletal system, PS plays an essential role in the deposition of hydroxyapatite in mineralizing cells. In vitro experiments have shown that nucleation of hydroxyapatite cores and the rate of mineralization are supported by high levels of PS, which initiates hydroxyapatite crystallization by forming a PS-Ca2+-Pi nucleation core.49, 50, 63, 64 In addition, according to the current view, mineralization of chondrocytes and osteoblasts is initiated in extracellular, lipid-bilayer enclosed matrix vesicles that are rich in PS65 and are released through budding from the plasma membrane.47, 66 In these matrix vesicles, calcium and phosphate ions are actively concentrated to generate an environment for the regulated nucleation of hydroxyapatite crystals. At least for vesicle-producing hypertrophic chondrocytes, it has been demonstrated that they express phospholipid scramblase 1 and expose PS to the cell surface.47 Similarly, mineralization-competent matrix vesicles released from vascular smooth muscle cells in an arteriosclerosis culture model contain increased levels of externalized PS,67 supporting the importance of PS exposure in tissue mineralization.
How Ano6 regulates osteoblast mineralization is difficult to predict, but one could speculate that Ano6-dependent PS translocation facilitates matrix vesicle release from the cell membrane, a prerequisite for initiating the mineralization core. Annexins, which bind to PS and collagen fibers in vitro, have been proposed to link matrix vesicles to the collagen network, and this might be a different step affected by impaired PS translocation.68–73 The dispersed matrix calcification observed in mutant embryos might support such a function (Fig. 4). Interestingly, it has also been postulated that PS-annexin-collagen interactions regulate calcium influx into matrix vesicles and cells in vitro,74–76 thereby regulating the concentration of Ca2+ ions. Mice double-deficient for annexins A5 and A6 did not show any mineralization defects and displayed a normal distribution of matrix vesicles in vivo, showing that bone mineralization is facilitated rather independently of these annexins.77 However, it is possible that other annexins expressed in bone, e.g., AnxA1 and AnxA2, functionally compensate for loss of AnxA5 and AnxA6. If Ano6-dependent PS scrambling regulates mineralization at the level of matrix vesicle release, targeting of matrix vesicles to the collagen network or by directly influencing hydroxyapatite nucleation inside the matrix vesicles will be a challenging question for future studies.
Another important question to be addressed is the role of other anoctamin proteins in regulating scramblase activity and mineral deposition. Although initially disturbed, later stages of bone formation display relatively mild mineralization defects, indicating that other members of the anoctamin family rescue the initial delay. This is supported by the finding that GDD patients carrying mutations in ANO5 also show a bone mineralization defect. Furthermore, besides Ano5 and Ano6, at least Ano1, Ano3, and Ano9 are expressed in osteoblasts, although no bone phenotype has been observed in Ano1−/− mice.26 A detailed analysis of the relative expression and function of the different anoctamins during osteoblast differentiation will give insight into this question. Similarly, most anoctamins seem to mediate chloride channel activity, at least in vitro, but only Ano6 has been linked to PS translocation. It will be important to decipher if this is a specific function of a subset of anoctamins and how the proposed chloride channel function may be linked to the regulation of scramblase activity.
All authors state that they have no conflicts of interest.
We thank B de Crombrugghe for kindly providing Osx mutants and D Hoffmann for help with statistical analyses. This work was funded by grants from the BMBF (01GM0317) and the DFG (Vo620/10) to AV. We thank Sabine Stumpp for help with bone sections and histomorphometry, and Claire Schlack for expression analyses.
Authors' roles: HE, MC, UK, BB, MW, and AV designed the study. HE, MC, UK, MM, YK, HMM, and SG performed and analyzed the experiments. HE, UK, and AV wrote the manuscript. MC, MM, YK, BB, and MW revised the manuscript and contributed important intellectual content. HE, MC, and UK are responsible for the integrity of the data. All authors approve of this final version of the manuscript.