Abnormal calcification of the extracellular matrix of soft connective tissues is associated with a number of pathologies including vascular disease, chronic kidney disease, and multiple skeletal disorders.1–4 Ossification of paraspinal ligaments is often detected in middle-aged and elderly patients, presenting as one of two noninflammatory spondyloarthropathies: (1) diffuse idiopathic skeletal hyperostosis (DISH; formerly referred to as Forestier disease), which involves calcification of spinal ligaments and extra-axial structures including entheses4; and (2) ossification of the posterior longitudinal ligament (OPLL), which is common in the Japanese population.5 Although often misdiagnosed, DISH and OPLL are distinct from degenerative disc disease, osteoarthritis, and ankylosing spondylitis.6
DISH is typically diagnosed using radiographs of the thoracic spine or chest, which demonstrate: (1) flowing calcifications along the anterolateral aspect of at least four contiguous vertebral bodies; (2) preservation of intervertebral disc height (in contrast to degenerative disc disease); and (3) absence of bony ankylosis of facet joints and absence of sacroiliac erosion, sclerosis, or fusion (in contrast to ankylosing spondylitis).7 DISH often manifests as back pain associated with limited range of spinal motion, but can progress to the extent that lesions interfere with neighboring structures, including compression of the spinal cord and nerve roots.8, 9 Lesions in DISH can also cause dysphagia,10 and DISH is associated with increased susceptibility to spinal fractures11 and postsurgical heterotopic ossifications.12 Correlative studies have associated obesity, hypertension, diabetes mellitus, hyperinsulinemia, dyslipidemia, elevated growth hormone levels, elevated insulin-like growth factor-1, and hyperuricemia with DISH.13–15 It has been proposed that DISH lesions originate in fibrocartilaginous structures including entheses16; however, due to the lack of suitable animal models, the underlying pathogenesis remains obscure.
Several factors including genetic background have been postulated to be involved in the etiology of DISH; however, to date no single gene defect has been associated with the disease. In a subset of DISH patients of Asian descent, single-nucleotide polymorphisms in the COL6A1 and FGF2 genes have been shown to confer genetic susceptibility to DISH.17, 18 In contrast to DISH, OPLL has been studied extensively in the Japanese population, leading to identification of a number of associated genes, including NPSS,19, 20COL11A,21COL6A1,22BMP2,23TGFβ,24 and FGFR1.18
Recent advances point to a critical role for purine metabolism in the regulation of biomineralization in diseases associated with either insufficient or ectopic mineralization.25–30 For example, mutations in the gene encoding ectonucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1) have been associated with hypermineralization disorders.25, 26 Moreover, a recent study has linked ectopic arterial and joint calcifications with loss of ecto-5′-nucleotidase (5′NTD) function leading to decreased levels of extracellular adenosine.27 To further explore the role of purine metabolism in the regulation of biomineralization, we examined the phenotype of the mouse lacking the gene encoding the nucleoside transporter ENT1. ENT1 (equilibrative nucleoside transporter 1 or solute carrier family 29 member 1, encoded by the Slc29a1 locus) is the predominant nucleoside transporter expressed in mammalian cells.31 This sodium-independent, facilitative diffusion carrier is responsible for the movement of hydrophilic nucleosides, such as adenosine, across biological membranes. Loss of ENT1 activity would be expected to modify extracellular adenosine levels, thus altering overall purine metabolism and signaling through adenosine receptors. The present study reports the novel skeletal phenotype of the ENT1–/– mouse that resembles DISH in humans.
Subjects and Methods
ENT1–/– mice were generated through targeted deletion of exons 2 to 4 of the gene encoding ENT1 by a cre-loxP targeting strategy.32ENT1–/– mice were backcrossed with outbred C57BL/6 mice. The mouse colony was maintained through the breeding of heterozygous animals (ENT1+/–) to obtain wild-type (ENT1+/+) and knockout (ENT1–/–) littermates. Mice were housed in standard cages and maintained on a 12-hour light/dark cycle, with rodent chow and water available ad libitum. Genotyping was performed as described.33 Given the increased reported prevalence of DISH in males (25% of men versus 15% of women over 50 years of age),34 male mice were used for all experiments. Mice were euthanized at the following ages: 1 month (4–4.5 weeks), 2 months (8–11 weeks), 4 months (16–18 weeks), 6 months (26–30 weeks), and 12+ months (12–17 months). Experimental results were derived from groups of at least 3 wild-type and 3 ENT1–/– mice. All aspects of this study were conducted in accordance with the policies and guidelines set forth by the Canadian Council on Animal Care and were approved by the Animal Use Subcommittee of the University of Western Ontario, London, ON, Canada.
Micro–computed tomography imaging
Micro–computed tomography (µCT) scanning, reconstruction, and analysis were performed based on reported protocols35 with the following modifications. Formalin-fixed or snap-frozen whole mice were imaged at Robarts Research Institute (London, ON, Canada) using a dedicated laboratory µCT scanner (eXplore speCZT; GE Healthcare, Waukesha, WI, USA). Data were acquired with an X-ray tube voltage of 90 kV and a current of 40 mA. In one continuous rotation, 900 views were obtained at an angular increment of 0.4 degrees and an exposure interval of 16 ms per view. The total scan time was 5 minutes per animal. A calibrating phantom, consisting of a vial of water, air, and a synthetic bone-mimicking epoxy (SB3; Gammex Inc. Middleton, WI, USA), was imaged together with the specimens. Images were acquired at isotropic voxel size of 100 µm and reconstructed into 3D images, using a modified cone-beam algorithm.36 The reconstructed data were expressed in Hounsfield units by calibrating the gray-level values against those of water and air. Mineralized tissue density within the volume of interest was expressed as hydroxyapatite (HA) equivalent density (mg HA cm−3), based on the calibration provided by the SB3 bone-mimicking material.
Images acquired for each animal were scored for severity of ectopic mineralization, based on the percentage of affected sites within the spine (sites defined as an intervertebral disc and/or associated paraspinal ligaments and entheses) in each anatomical region (cervical, thoracic, lumbar, and caudal). Values ranged from 0 to 4, with a score of 0 reflecting no detectable mineralized lesions, 1 indicating lesions involving 0% to 30% of sites within the anatomical region, 2 indicating lesions involving 30% to 60% of sites, 3 indicating lesions involving 60% to 90% of sites, and 4 indicating lesions involving 90% to 100% of sites.
Measurement of mineralized tissue density
Using data acquired by µCT, quantification of the density and volume of mineralized tissues was performed in a region of interest restricted to the spine (C1 to sacrum), rib cage, and sternum. For this purpose, a 3D volume of interest was defined by an operator within each volume image, using manually drawn contours lofted to create a volume that included only the anatomy of interest (MicroView 2.2; GE Healthcare Biosciences). Three measurements were obtained within this volume of interest: the volume of hypermineralized tissue, the volume of tissue with density equivalent to that of normal cortical bone, and the maximum density. Hypermineralized tissue was defined as material that exceeded the maximum density of cortical bone in the spine of wild-type mice within a specified age group. In the present study, we defined threshold values of 610, 630, and 710 mg HA cm−3 for maximum spinal bone density in animals of ages 2, 4, and 6 (or greater) months, respectively. Previous studies have reported a similar increase in murine cortical bone density during the first 24 weeks of postnatal development.37 Normal density tissue was defined as material that exceeded a minimum threshold (126 mg HA cm−3), but fell below the maximum thresholds defined above. Volume measurements are reported in cubic millimeters (mm3), representing the summation of all the volume elements that fell within the defined range of mineral density. Additionally, the maximum mineralized tissue density within the volume of interest was reported. Nonlinear least squares fits were obtained using GraphPad Prism; data for volume of hypermineralized lesions were fit with an exponential growth equation, and data for the volume of normal bone and maximum density values were fit with one-phase association equations.
Scanning electron microscopy and energy-dispersive X-ray spectroscopy
Scanning electron microscopy (SEM) imaging and energy-dispersive X-ray spectroscopy (EDX) microanalyses were performed based on previous reports38 using a LEO 1540XB FIB/SEM instrument (Carl Zeiss, Oberkochen, Germany) and X-ray analysis system (Oxford Instruments, Oxford, UK) at the Western Nanofabrication Facility (The University of Western Ontario). Prior to SEM imaging, dried samples were coated with 5 nm of osmium using a plasma coater (OPC-80T; Filgen Inc., Nagoya, Japan). EDX spectra were collected from bone and mineralized lesions identified by electron backscatter imaging at 20 keV beam energy. Elemental analysis of the EDX spectra was performed using INCA software (Oxford Instruments), including background correction and fitting of all peaks.
Formalin-fixed tissue samples were decalcified with Shandon TBD-2 Decalcifier (Thermo Scientific, Nepean, ON Canada) at a ratio of 10:1 (fluid:tissue) for 5 days with gentle rocking. Following standard histological processing, decalcified samples were embedded in paraffin and 5 µm-thick serial sections were cut. Tissues were sectioned, mounted on glass slides and baked for 48 hours at 45°C. Slides were stained with hematoxylin and eosin (H&E) and images were acquired using a Leica DM1000 microscope.
For visualization of cell nuclei, sections described above were dewaxed in xylene and rehydrated by successive immersion in descending concentrations of ethanol. Mounting was performed with VECTASHIELD Mounting Medium with 4′,6-diamidino-2-phenylindole (DAPI) (Burlingame, CA) and images were captured using a Leica DMI6000B microscope.
High-performance liquid chromatography (HPLC) was used to analyze plasma for levels of adenosine and adenosine metabolites to determine the functional consequence of loss of ENT1. At 2 months of age, mice were anesthetized with pentobarbital. Blood was collected by cardiac puncture into a syringe containing NaCl (118 mM), KCl (5 mM), EDTA (13.2 mM), 5-iodotubercidin (10 µM) to inhibit adenosine kinase, erythro-9-(2-hydroxy-3-nonyl)adenine hydrochloride (EHNA, 100 µM) to inhibit adenosine deaminase, and dilazep (10 µM) to inhibit adenosine transport.39 Plasma was isolated by sedimentation at 3,000g for 10 minutes at 4°C. Plasma was applied to a 10-kDa cutoff ultrafiltration column and sedimented at 14,000g for 15 minutes at 4°C. Filtered plasma was analyzed on an Onyx monolithic C18 column as described,40 using a Hewlett Packard 1090 LC with UV detector. Adenosine was detected at 260 nm and adenosine metabolites at 250 nm.
To screen for systemic changes resulting from loss of ENT1, serum chemistry was performed using established panels of clinical chemistry parameters. At 2 months of age, mice were anesthetized with pentobarbital and blood was collected by cardiac puncture. Blood was allowed to coagulate at room temperature for 30 minutes. Samples were then sedimented at 3,000g for 10 minutes at 4°C and the serum supernatant was transferred to a fresh tube and frozen at –80°C. Chemical and biochemical analyses were performed by the Centre for Modeling Human Disease at the Toronto Centre for Phenogenomics (Toronto, Canada).
To assay plasma levels of inorganic pyrophosphate (PPi), mice were anesthetized with pentobarbital and blood was collected by cardiac puncture. Samples were transferred to microfuge tubes containing heparin (5 USP units/mL of blood) and plasma was isolated by sedimentation at 3,000g for 10 minutes at 4°C. Plasma was applied to 10-kDa cutoff ultrafiltration spin columns and sedimented at 14,000g for 20 minutes at 4°C. A fluorimetric PPi assay kit (ab112155; Abcam, Cambridge, MA, USA) was first validated using plasma samples from wild-type mice. Selected samples were supplemented with a saturating concentration of PPi (30 µM) in the presence or absence of inorganic pyrophosphatase (0.0012 units/µL of sample; Sigma-Aldrich, St. Louis, MO, USA) (Supplemental Fig. S1). Plasma samples were isolated from 2-month-old wild-type and ENT1–/– mice (n = 6 mice for each genotype) and analyzed according to the manufacturer's protocol with an incubation time of 20 minutes. Fluorescence was measured at excitation and emission wavelengths of 316 and 456 nm, respectively, using a SpectraMax M5 microplate reader (Molecular Devices, Sunnyvale, CA, USA) with Softmax Pro v5 software. Values were interpolated using linear regression. Background levels of plasma autofluorescence were negligible.
Quantitative real-time RT-PCR
At 6 months of age, littermate-paired wild-type and ENT1–/– mice (n = 8 for each genotype) were dissected to isolate intervertebral discs (IVDs). IVDs were segregated according to anatomical location as cervical-thoracic, lumbar, or caudal, and placed directly in TRIzol Reagent (Invitrogen, Life Technologies, Burlington, ON, Canada). The IVDs were homogenized using a Polytron benchtop homogenizer and total RNA was extracted according to the manufacturer's protocol. cDNA synthesis was performed using Superscript II Reverse Transcriptase (Invitrogen, Life Technologies). Gene expression patterns were determined by real-time PCR using the Bio-Rad CFX384 system. Gene transcript levels were determined by absolute quantification assays, using a six-point calibration curve made from known starting quantities of cDNA (0.1 µg to 3.2 × 10−5 µg) synthesized from a mix of wild-type heart, brain, kidney, muscle, intervertebral disc, and calvarial RNA. Values are expressed relative to the calibration curve. PCR primer sequences are provided in Supplemental Table 1.
Table 1. Mineralized Tissue Density in ENT1–/– and Wild-Type Mice
Max density (mg HA cm−3)
Max density (mg HA cm−3)
µCT-derived values for the volume of normal and hypermineralized tissue and the maximum density in ENT1–/– and wild-type mice. Data are means ± SD. Differences were evaluated by two-way analysis of variance (ANOVA) followed by a Bonferroni multiple comparison test.
ENT1 = equilibrative nucleoside transporter 1; HA = hydroxyapatite.
Significantly different from wild-type: *p < 0.05, **p < 0.001.
16.4 ± 3.2** (n = 4)
527 ± 48*
1044 ± 77**
0.5 ± 0.4 (n = 3)
356 ± 35
744 ± 10
3.5 ± 1.1 (n = 3)
381 ± 55
921 ± 38*
0.6 ± 0.5 (n = 3)
351 ± 92
746 ± 54
1.9 ± 1.5 (n = 3)
341 ± 16
763 ± 110
0.5 ± 0.6 (n = 3)
318 ± 79
639 ± 65
0.4 ± 0.5 (n = 4)
256 ± 52
652 ± 84
0.5 ± 0.5 (n = 4)
259 ± 32
651 ± 58
Ectopic mineralization in ENT1–/– mice
Previous studies reported that ENT1–/– mice are phenotypically normal, with only a modest decrease (∼10%) in body weight compared with that of wild-type littermates.32 At young ages, open-field locomotor activity in ENT1–/– and wild-type mice does not differ. Moreover, there is no significant difference in spontaneous mortality rates up to 6 months of age. To date, there have been no reports of phenotypic changes in skeletal tissues related to mineralization32, 33, 41, 42; however, all reported studies were conducted on mice less than 4 months of age.
In the present study, we noted that the spines of 6-month-old ENT1–/– mice were extremely rigid. At 8 months, ENT1–/– mice demonstrated decreased hindlimb mobility that progressed to hindlimb paralysis by 12 to 17 months of age. Postmortem analysis of a 17-month-old ENT1–/– mouse revealed hard, chalky, white lesions protruding ventrally from the thoracic spine (Fig. 1A). µCT images revealed large radio-opaque lesions at distinct foci in the cervical and upper thoracic regions of ENT1–/– mice, localized to paraspinal and intervertebral tissues, and protruding in some cases into the spinal canal (Fig. 1B). Histological examination identified large accumulations of amorphous material (Fig. 1C) localized to the intervertebral discs (upper panel) and paraspinal tissues associated with the spinous processes (lower panel). The presence of calcified lesions within the spinal canal, impinging on the spinal cord, was consistent with the decreased mobility and paralysis observed in ENT1–/– mice.
Time course of phenotype development
µCT imaging was employed to assess the development and extent of ectopic mineralization in ENT1–/– mice over time (Fig. 2A). Images were scored for severity, based on the percentage of affected sites within the spine (intervertebral discs, and associated paraspinal ligaments and entheses) in each anatomical region (Fig. 2B). Ectopic mineralization of spinal tissues was not evident in wild-type animals at any time point (Fig. 2A), or in ENT1–/– mice at 1 month of age (not shown).
Ectopic mineralization in ENT1–/– mice proceeded temporally in a consistent anatomical pattern. At 2 months of age, maximum intensity projection (MIP) images revealed radio-opaque material in the paraspinal tissues of the cervical spine and the rib entheses of the upper thoracic spine (Fig. 2, top panels). At this age, there was no detectable involvement of intervertebral discs.
With advancing age, accumulation of ectopic mineral progressed caudally, with involvement of the lumbar spine at 6 months (Fig. 2, middle panels). In addition to paraspinal tissues and rib entheses, lesions were observed within the intervertebral spaces at 6 months in both the thoracic and lumbar spine. At this time point, lesions involved 60% to 90% of the cervical spine and 30% to 60% of the thoracic spine. Interestingly, lesions did not progress caudally in a uniform manner. Following involvement of the cervical and upper thoracic spine, lesion formation initially bypassed the mid-region of the thoracic spine and then appeared specifically in thoracic vertebrae T11–T12 (Fig. 2A, asterisk). At greater ages, lesions developed in the intervening mid-region of the thoracic spine.
By 12 months of age, ectopic mineralization was detected in >60% of the cervical, thoracic, and lumbar regions of the spine, as well as >30% of intervertebral discs of the caudal spine (Fig. 2, lower panel). In addition, there was notable decrease in kyphosis between 6 and 12 months of age in ENT1–/– mice compared to wild-type (Fig. 2A, lower panel). No evidence of ectopic mineralization was detected in the vasculature of ENT1–/– mice at any time point examined by µCT.
To examine the consequence of ENT1 haploinsufficiency, 6-month-old heterozygous ENT1+/– mice were examined by µCT (Supplementary Fig. S2). In contrast to ENT1–/– mice, heterozygous animals did not demonstrate ectopic mineral formation in paraspinal tissues, rib entheses, or intervertebral discs. Moreover, the loss of kyphosis observed in knockout animals was not apparent; ENT1 heterozygous animals were indistinguishable from wild-type controls.
Density and elemental composition of ectopic lesions
Mineralized tissue density of wild-type and ENT1–/– mice at different ages was quantified from µCT data. The region of interest for these analyses consisted of the cervical to lumbar spine, rib cage, and sternum. Hypermineralized tissue was detected in paraspinal and intervertebral tissues of the spine, as well as the sternocostal articulations (highlighted in red in Fig. 3A). Quantification revealed that the volume of hypermineralized tissue grew exponentially with age in the ENT1–/– mice, whereas there was little hypermineralized tissue detectable in wild-type mice at any age (Fig. 3B). We also quantified the volume of mineralized tissue with density equivalent to that of normal cortical bone (Fig. 3C). Whereas the wild-type animals appeared to reach a plateau of approximately 355 mm3 by day 200 (reflecting bone growth), ENT1–/– mice reached a plateau of 552 mm3 by day 400 (reflecting both bone growth and accumulation of ectopic mineral). Likewise, the maximum mineral density of the wild-type mice reached a plateau before an age of 200 days, whereas that of the ENT1–/– mice continued to increase past day 400 (Fig. 3D), reflecting hypermineralization of the ectopic lesions.
A summary and statistical analysis of these data is presented in Table 1. From 2 to 17 months of age, the volume of bone in wild-type mice increased from 259 to 356 mm3 with only about 0.5 mm3 appearing hypermineralized (regardless of age) according to the thresholds used. The mineralized tissue volume of ENT1–/– mice was similar to that of wild-type controls up to 6 months of age. In contrast, 12- to 17-month-old ENT1–/– mice had a significantly greater volume of mineralized tissue than the equivalent age of wild-type mice (527 versus 356 mm3). In addition, the ENT1–/– mice had a significantly greater volume of hypermineralized tissue than wild-type mice between 12 to 17 months old (16 versus 0.5 mm3), and the maximum mineral density was significantly greater in ENT1–/– mice than wild-type controls at both 6 and 12 months of age. Note that whereas the maximum mineral density observed in ENT1–/– mice (1044 mg HA cm−3) does not exceed the typical value for fully mineralized cortical bone (approximately 1050 mg HA cm−3), it is significantly greater than the mineral density of cortical bone observed in the spine of wild-type littermates (744 mg HA cm−3).
Samples from wild-type and ENT1–/– mice at 6 months of age were analyzed by EDX to determine the elemental content of both cortical bone and mineralized lesions. Mineralized lesions appeared by SEM as a disordered amorphous material (Supplemental Fig. S3A, B) and EDX revealed a high content of calcium and phosphorus (Supplemental Fig. S3C). The mean elemental content of bone and lesions is displayed as the percent atomic ratio in Table 2. The vertebral bone from ENT1–/– mice was not significantly different from that of wild-type mice in terms of elemental ratios. In contrast, ectopic mineralized lesions in ENT1–/– mice displayed higher levels of phosphorus and calcium relative to vertebral bone, consistent with the greater mineralized tissue density observed using µCT. The calcium/phosphorus ratio was similar in all samples.
Table 2. Energy-Dispersive X-Ray Spectroscopy
% Atomic ratio
Scanning electron microscopy was used to identify regions of vertebral cortical bone and ectopic mineralization (lesion) in samples isolated from 6-month-old WT and ENT1–/– mice. Energy-dispersive X-ray spectroscopy was then applied to these selected regions to obtain the % atomic ratios of the indicated elements. Data are mean ± SEM, n = 4.
We first assessed the intervertebral disc regions of wild-type and ENT1–/– mice between 12 and 17 months of age. Decalcified sections stained with H&E revealed the presence in ENT1–/– mice of large accumulations of amorphous, eosinophilic acellular material suggestive of niduses of mineralization in the cervical, thoracic, lumbar, and caudal regions (Fig. 4). Interestingly, in the intervertebral disc, lesions appeared to be localized within the annulus fibrosus, leading to lateral compression of the nucleus pulposus and extensive bulging of the annulus fibrosus out of the intervertebral space (Fig. 4, right panels). There was no evidence that lesions were associated with inflammation or increased vascularization. Because the sacral intervertebral discs undergo progressive fusion with the onset of skeletal maturity in mice, this region was not examined for ectopic mineralization.
Lesions were also detected in the sternocostal region of ENT1–/– mice between 12 and 17 months of age, by both µCT and histology (Figs. 2, 3A, and Supplemental Fig. S4A). In this region, extensive ectopic mineral was detected within the connective tissue of the sternocostal articulations, leading to deformation of the sternum. In contrast, examination of appendicular joints revealed no aberrant morphology or evidence of ectopic mineralization (Supplemental Fig. S4B–E) establishing that lesions are specifically associated with the axial skeleton.
Comparison of histological sections of spinal tissues from wild-type and ENT1–/– mice at 6 months of age revealed further insights into the development of lesions. Within the cervical spine of 6-month-old ENT1–/– mice, large lesions were associated with the paraspinal ligaments, but no distinct lesions were detected within the intervertebral discs in this region (Fig. 5A). However, the outer annulus fibrosus of the intervertebral discs did appear altered, with regions of metaplasia and disruption of normal tissue architecture (Fig. 5A, arrowhead). Within the thoracic spine of 6-month-old ENT1–/– mice, extensive lesions were localized within: (1) the annulus fibrosus; (2) the fibrocartilaginous tissue of the posterior paraspinal ligaments; and (3) the fibrocartilaginous tissue of the rib entheses (Fig. 5B). In the lumbar spine of 6-month-old ENT1–/– mice, small lesions were detected within the annulus fibrosus, leading to disruption of intervertebral disc structure and displacement of the nucleus pulposus (Fig. 5C). At this time point, no changes were detected within the caudal spine.
At 2 months of age, when ectopic mineral was first detected in ENT1–/– mice by µCT, histological examination also demonstrated lesions within the paraspinal ligaments of the cervical spine (Fig. 6A). Interestingly, these lesions were associated with large regions of metaplasia, with a disruption of normal tissue architecture and increased cellularity (Fig. 6A, arrowheads). In the upper thoracic spine, the mineralized lesions detected by µCT in ENT1–/– mice were localized to the fibrocartilaginous structures of the paraspinal ligaments and rib entheses, within larger regions demonstrating evidence of hyperplasia and desmoplasia (Fig. 6B, arrowheads). At this time point, no lesions were detectable in intervertebral discs within any region of the spine. Early lesions evident at 2 months of age were surrounded by a clearly demarcated transition zone (TZ; Fig. 6C). The periphery of this zone consisted of typical fibrocartilaginous tissue and, as expected, cells with basophilic nuclei that stained positively with DAPI (Fig. 6C, lower panels). However, the more central region of the transition zone was characterized by metaplasia, with cells displaying eosinophilic nuclei that stained positively with DAPI, revealing an intact nuclear structure. The TZ surrounded the presumptive mineralized lesion, which contained large accumulations of amorphous matrix and necrotic cell debris. There was no evidence of apoptotic nuclei within the lesion or TZ.
Differences in blood chemistry and gene expression associated with the ENT1–/– phenotype
Blood plasma or serum from 2-month-old wild-type and ENT1–/– mice was analyzed for components related to mineralization as well as adenosine and adenosine metabolites (Table 3). HPLC analysis of plasma showed 2.8-fold greater adenosine levels in ENT1–/– than wild-type mice. In contrast, there were no significant differences in levels of the adenosine metabolites xanthine and uric acid, arguing against the possibility that ectopic lesions involved deposition of monosodium urate crystals. Importantly, serum levels of alkaline phosphatase, calcium, phosphate, and magnesium were not significantly different in ENT1–/– and wild-type mice. On the other hand, quantification of plasma PPi levels revealed 3.4 ± 1.0-fold greater levels in ENT1–/– mice than in wild-type controls (n = 6 two-month-old mice for each genotype, p < 0.05).
Table 3. Blood Chemistry
Wild-type (n = 4)
ENT1–/– (n = 5)
Data are means ± SEM. Serum or plasma was obtained from wild-type and ENT1–/– mice at 2 months of age.
ALP = total alkaline phosphatase; ENT1 = equilibrative nucleoside transporter 1; Pi = inorganic phosphate.
Significantly different from wild-type (p < 0.05, Student's t test).
Given this alteration in PPi levels, we interrogated the expression of genes associated with pyrophosphate metabolism in intervertebral disc tissues isolated from ENT1–/– and wild-type littermate control mice at 6 months of age. Real-time PCR analysis demonstrated a significant decrease in the expression of ectonucleotide pyrophosphatase/phosphodiesterase 1 (Enpp1; Fig. 7A), Ank (a putative PPi transporter) (Fig. 7B), and tissue-nonspecific alkaline phosphatase (Alpl; Fig. 7C) in ENT1–/– mice. In contrast, no significant difference was detected in the expression of unrelated genes including the adenosine A3 receptor (Adora3; Fig. 7D) in ENT1–/– and wild-type mice, indicating specificity.
To determine if the specific anatomical pattern of ectopic mineralization was influenced by endogenous levels of ENT1, expression was assessed in 6-month-old wild-type mice. No differences were detected in ENT1 transcript levels in intervertebral disc tissues isolated from cervical-thoracic, lumbar, or caudal regions of the spinal column (Fig. 7E).
This is the first report of a role for ENT1 in regulating biomineralization. We discovered that ENT1–/– mice develop ectopic mineralization with distribution restricted to the fibrous connective tissues of the spine and sternum. In the spine, pathological mineralization begins in the paraspinal fibrocartilaginous tissues and progresses to involve the annulus fibrosus of intervertebral discs. Aberrant mineralization is first observed in ENT1–/– mice between 6 and 8 weeks of age in the paraspinal connective tissues of the cervical vertebrae. With advancing age, lesions increase in severity and progress to other regions of the spine, with ectopic mineralization eventually involving the thoracic, lumbar, and caudal spine, as well as the sternocostal articulations.
ENT1–/– mice were first used to investigate the role of adenosine transport in the central nervous system pathways regulating alcohol consumption.32 In that study, male mice were examined at approximately 10 weeks of age and appeared normal in their anatomy, physiology, mortality rates, and consumption of water. However, ENT1–/– mice consumed twice as much alcohol compared to wild-type controls. This behavior was associated with a decrease in endogenous adenosine tone, which was not due to loss of A1 receptors or decreases in A1 receptor affinity. ENT1–/– mice have also been shown to exhibit reduced anxiety-like behaviors.43 Levels of endogenous extracellular adenosine were not reported in these studies.
More recently, ENT1-null mice have been used to investigate the role of this transporter in the cardiovascular system. Microvascular endothelial cells isolated from ENT1–/– mice have enhanced expression of the A2A adenosine receptor and adenosine deaminase.33 Cardiomyocytes and microvascular endothelial cells from ENT1–/– mice are relatively resistant to ischemic insult and ENT1–/– mice show decreased heart damage in response to ischemia and hypoxia.44 Furthermore, loss of ENT1 protects against ischemic acute kidney injury through control of postischemic renal perfusion.45 Overall, these studies point to a cardioprotective role of ENT1. However, it is unlikely that changes in the nervous and cardiovascular systems give rise to ectopic mineralization in ENT1–/– mice, especially of the inner annulus fibrosus, which is considered an avascular and aneural tissue.46, 47
In the present study, we found that plasma adenosine concentrations are significantly elevated in the ENT1–/– mice compared to wild-type littermates, as reported previously.41 The greater plasma concentration of adenosine in ENT1–/– mice likely reflects extracellular accumulation of adenosine due to lack of uptake by cells that normally express ENT1, one of the primary uptake pathways for adenosine.31, 48 The role of elevated extracellular adenosine in the formation of abnormal mineral deposits in fibrocartilaginous tissues remains to be explored, but may involve local changes in adenosine receptor signaling.
In this regard, recent advances point to a critical role for purine metabolism and signaling in the regulation of biomineralization and in diseases associated with either insufficient or ectopic mineralization.25–30 Once released by the cell, extracellular adenosine triphosphate (ATP) is sequentially metabolized by cell-surface enzymes, leading first to the production of PPi and adenosine monophosphate (AMP), which in turn are converted to inorganic phosphate and adenosine. Ectonucleotide pyrophosphatase/phosphodiesterase 1 (Enpp1) is responsible for the first step in this process, and the release of PPi has been shown to inhibit ectopic calcification in soft tissues.49 Interestingly, we show that ENT1-null mice demonstrate significant downregulation of Enpp1 expression in intervertebral disc tissues relative to wild-type littermate controls. These findings are in keeping with the reported association between mutations in the gene encoding Enpp1 and hypermineralization disorders such as idiopathic infantile calcification25 and OPLL.26 Furthermore, Enpp1–/– mice display soft tissue mineralization in the Achilles tendon, paraspinal ligaments, and intervertebral discs, as well as hyperostosis of peripheral joints and calcification of articular cartilage.50 However, elements of the appendicular skeleton are not affected in ENT1–/– mice; changes are limited to fibrocartilaginous tissues of the axial skeleton.
We also observed a significant decrease in Ank expression in the intervertebral discs of ENT1–/– mice. Ank is a transmembrane protein that is thought to mediate PPi transport. In mice, loss-of-function mutations of Ank lead to arthritis, ectopic crystal formation, and generalized joint fusion,51 whereas, in humans, dominant mutations are associated with craniometaphyseal dysplasia52, 53 and familial chondrocalcinosis.54, 55 Thus, like Enpp1, decreased Ank expression in the spinal tissues of ENT1–/– mice would be expected to further suppress extracellular PPi levels, permitting the formation of ectopic mineral deposits.
Unexpectedly, expression of Alpl was also found to be decreased in the intervertebral discs of ENT1-null mice. Alpl encodes tissue-nonspecific alkaline phosphatase, which is responsible for the hydrolysis PPi,50 and Alpl expression has been shown previously to be regulated by extracellular adenosine.27 In humans, disruption of ALPL causes hypophosphatasia, characterized by skeletal hypomineralization56 and presumably a result of excessive accumulation of PPi. Thus, decreased Alpl expression in the spinal tissues of ENT1–/– mice would be expected to increase extracellular PPi levels, counteracting the decrease in PPi arising from changes in Enpp1 and Ank expression. Moreover, it is conceivable that a generalized decrease in Alpl expression could account for the increase in plasma PPi that we observed in ENT1–/– mice.
Adenosine and ATP can regulate cell behavior and gene expression through cell surface receptors—the adenosine receptor family57 and the P2 family of nucleotide receptors,58, 59 respectively. Adenosine increases intracellular cyclic AMP (cAMP) levels via its A2A and A2B receptor subtypes. cAMP has been shown to induce abnormal calcification of vascular smooth muscle cells via a mechanism involving reduction in extracellular PPi accumulation.60, 61 In addition, activation of the P2X7 subtype of ATP receptors promotes bone formation and mineralization.59, 62
Taken together, our data suggest that disruption of adenosine signaling and PPi metabolism in ENT1–/– mice is associated with disease onset and progression. Although there are no reported associations between mutations in the gene encoding ENT1 in humans and DISH, the high incidence of this disease in the human population argues against an underlying single gene defect. However, disruption of adenosine or PPi metabolism could result from alterations in the function of one or more regulatory or metabolic proteins—either through genetic defects or as a consequence of cell aging.
In ENT1–/– mice, the only location outside of the spine that exhibited ectopic mineralization was the sternocostal region. This distinct anatomical pattern of mineral deposition in ENT1–/– mice suggests that some common element in spinal and sternal fibrocartilaginous tissues makes them susceptible to ectopic mineralization in the absence of ENT1. Relatively scant vascular perfusion of the affected fibrocartilaginous tissues could lessen the clearance of extracellular adenosine, giving rise to sustained adenosine signaling, which may in turn lead to deregulation of PPi metabolism.
The spatial and temporal pattern of ectopic mineralization observed in the ENT1–/– mice, along with the absence of inflammation in affected tissues, resemble characteristics of DISH in humans. DISH affects about 20% of the male population over the age of 55 years, with a slightly lower prevalence in women.34 Similarly, in ENT1–/– mice, lesions developed gradually over time, with functional impairments noticeable by 8 months of age. Similar to DISH in humans, which is often first diagnosed in the cervical or thoracic spine,63 lesions in ENT1–/– mice began in the cervicothoracic spine and spread caudally. Moreover, sternal involvement is also noted in DISH patients.64 On the other hand, extra-axial calcifications are present in some DISH patients, but were not detected by µCT or histology in ENT1–/– mice. Interestingly, the reproducible pattern of mineral deposition observed in ENT1–/– mice was not related to differences in endogenous ENT1 expression levels in the affected tissues. The factors contributing to the timing of mineral deposition at these sites remains an intriguing question for ongoing investigation. The radiographic appearance of DISH in humans bears striking resemblance to that of ectopic mineralization in ENT1–/– mice. DISH is characterized by tortuous paravertebral calcifications generally anterior to the vertebral bodies.65 On gross examination the appearance is likened to that of candle wax dripping down the spine. This appearance is remarkably similar to that observed in the present study by µCT analysis of ENT1–/– mice.
In summary, this is the first report of a role for the primary membrane transporter for adenosine, ENT1, in regulating the calcification of soft tissues. Disruption of purine homeostasis by removal of ENT1 leads to the ectopic mineralization of paraspinal ligaments and intervertebral discs in mice, resembling lesions seen in the human condition, DISH. Pathogenesis appears to be associated with both local and systemic changes in PPi homeostasis. The ENT1–/– mouse may prove useful as a model for investigating the mechanisms underlying ectopic mineralization associated with DISH and for the evaluation of therapies for the prevention and reversal of DISH and associated pathologies.
All authors state that they have no conflicts of interest.
These studies were funded by the Canadian Institutes of Health Research (CIHR) (MOP-115068) and the Canadian Arthritis Network (CAN) (10-DAP-11). DBJB was supported by a CIHR Doctoral Research Award. DWH holds the Dr. Sandy Kirkley Chair in Musculoskeletal Research at The University of Western Ontario. CAS is supported by a Network Scholar Award from The Arthritis Society and CAN. MD is a Career Investigator of the Heart and Stroke Foundation of Ontario. We thank: Tom Chrones, Holly Dupuis, and Linda Jackson for assistance with histology; Joseph Umoh for µCT scanning; Drs. Tom Daley and Ian Welch for guidance on histopathology; Drs. David Freeman and Murray Cutler for help with HPLC analyses; and Dr. Todd Simpson (Western Nanofabrication Laboratory) for assistance with SEM and EDX. Requests for ENT1–/– mice should be addressed to Dr. Doo-Sup Choi, Department of Molecular Pharmacology and Experimental Therapeutics, Mayo Clinic College of Medicine, Rochester, MN 55905, USA.
Authors' roles: Study design: SJD, JRH, and CAS. Data collection: SW, DBJB, DQ, HI, and DWH. Data analysis: SW, DBBJ, HI, DWH, MD, JRH, and CAS. Data interpretation: SW, DBJB, DSC, DWH, MD, SJD, JRH, and CAS. Drafting manuscript: SW, DBJB, DWH, and MD. Revising manuscript content: DSC, SJD, JRH, MD, and CAS. Approving final version of manuscript: all authors. JRH and CAS take responsibility for the integrity of the data analysis.