Linked Deficiencies in Extracellular PPi and Osteopontin MediatePathologic Calcification Associated With Defective PC-1

and ANK Expression

KRISTEN JOHNSON,1 JAMES GODING,2 DEBORAH VAN ETTEN,1 ADNAN SALI,2 SHOU-IH HU,3

DAVID FARLEY,3 HOLLIS KRUG,4 LOVISA HESSLE,5 JOSE LUIS MILLAN,5 andROBERT TERKELTAUB1

ABSTRACT

Osteopontin and PPi both suppress hydroxyapatite deposition. Extracellular PPi deficiency causes spontane-ous hypercalcification, yet unchallenged osteopontin knockout mice have only subtle mineralization abnor-malities. We report that extracellular PPi deficiency promotes osteopontin deficiency and correction ofosteopontin deficiency prevents hypercalcification, suggesting synergistic inhibition of hydroxyapatite depo-sition.

Nucleotide pyrophosphatase phosphodiesterase (NPP) isozymes including PC-1 (NPP1) function partly togenerate PPI, a physiologic calcification inhibitor. PPi transport is modulated by the membrane channelprotein ANK. Spontaneous articular cartilage calcification, increased vertebral cortical bone formation, andperipheral joint and intervertebral ossific ankylosis are associated with both PC-1 deficiency and expressionof truncated ANK in ank/ank mice. To assess how PC-1, ANK, and PPi regulate both calcification and celldifferentiation, we studied cultured PC-1�/� and ank/ank mouse calvarial osteoblasts. PC-1�/� osteoblastsdemonstrated �50% depressed NPP activity and markedly lowered extracellular PPi associated with hyper-calcification. These abnormalities were rescued by transfection of PC-1 but not of the NPP isozyme B10/NPP3.PC-1�/� and ank/ank cultured osteoblasts demonstrated not only comparable extracellular PPi depression andhypercalcification but also marked reduction in expression of osteopontin (OPN), another direct calcificationinhibitor. Soluble PC-1 (which corrected extracellular PPi and OPN), and OPN itself (>15 pg/ml), correctedhypercalcification by PC-1�/� and ank/ank osteoblasts. Thus, linked regulatory effects on extracellular PPi

and OPN expression mediate the ability of PC-1 and ANK to regulate calcification. (J Bone Miner Res 2003;18:994–1004)

Key Words: ANK, hyperostosis, progressive ankylosis, ttw/ttw mouse

INTRODUCTION

TWO NUCLEOTIDE PYROPHOSPHATASE PHOSPHODIESTERASE

(NPP) family isozymes, PC-1 (or NPP1) and B10 (orNPP3), are strongly expressed by mature chondrocytes and

osteoblasts.(1–4) Each of these NPP family ecto-enzymeshydrolyzes the phosphodiester I bond in a variety of sub-strates with a purine and pyrimidine nucleoside monophos-phate backbone (EC 3.1.4.1).(5) However, the recent linkageof altered mineralization with defective PC-1 expression in“tiptoe walking” (or ttw/ttw) mice has suggested a uniquerole for PC-1 in regulating skeletal remodeling and calcifi-cation.(6) Specifically, ttw/ttw mice are homozygous for astop codon between the catalytic site and EF hand in PC-

Drs Goding and Sali receive research funding from Pfizer. DrsFarley and Hu are employed by Novartis Pharmaceuticals. Dr Ter-keltaub receives research funding from Novartis and also serves as aconsultant for Novartis. All other authors have no conflict of interest.

1Veterans Affairs Medical Center, University of California-San Diego, La Jolla, California, USA.2Monash University Medical School, Prahran, Australia.3Veterans Affairs Medical Center, University of Minnesota Medical School, Minneapolis, Minnesota, USA.4Novartis Pharmaceuticals, Summit, New Jersey, USA.5The Burnham Institute, La Jolla, California, USA.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 18, Number 6, 2003© 2003 American Society for Bone and Mineral Research

994

1.(6) The ttw/ttw mouse phenotype includes postnatal devel-opment of progressive ankylosing intervertebral and periph-eral joint hyperostosis, as well as spontaneous arterial andarticular cartilage calcification and increased vertebral cor-tical bone formation.(6–10)

Products released by NPP catalytic activity on nucleosidetriphosphates (EC 3.6.1.8) include PPi, a direct calcificationinhibitor.(5,11) PPi potently antagonizes the ability of Pi tocrystallize with calcium to form hydroxyapatite and PPi

suppresses hydroxyapatite crystal propagation.(11) Certainobservations suggest that PC-1 and PPi may further regulatecalcification by modulating function of cells in PC-1-expressing tissues.(11) Specifically, marked, systemic extra-cellular PPi lowering in the ttw/ttw mouse PC-1 deficiencystate is associated with organized periarticular and spineligament calcification with features of both membranousand enchondral ossification, rather than simply dystrophiccalcification.(6–10) In addition, �50% depression of plasmaand fibroblast NPP activity and extracellular PPi was re-cently linked to human PC-1 deficiency, but pathologicalcalcifications seemed limited to large arteries and periartic-ular tissues in the affected infant.(12)

PC-1 expression, NPP activity, PPi-degrading pyrophos-phates activities, and extracellular PPi are regulated byseveral growth factors, calciotropic hormones, andcytokines.(1–3,11,13–16) However, altered balance betweenPPi generation and degradation is not the sole mechanismdetermining extracellular PPi.

(11) Specifically, the multiple-pass membrane protein ANK promotes PPi channeling tothe cell exterior, thereby regulating both intracellular andextracellular PPi.

(17) “Gain of function” of wildtype ANKnormally stimulates increased extracellular PPi.

(17) How-ever, putative PPi channeling by ANK is abrogated bytruncation of the C-terminal ANK intracellular domain.(17)

Significantly, fibroblasts cultured from mice homozygousfor the truncation mutant ank demonstrated decreased ex-tracellular PPi.

(17) Moreover, the phenotypes of ank/ankmice and ttw/ttw mice are remarkably similar, includingshared and pathologically comparable enthesopathy and an-kylosing hyperostosis of the spine, as well as spontaneousarticular cartilage calcification and peripheral jointfusion.(6–10,17)

Here, we examined ex vivo calcification by PC-1�/�

mice, whose phenotype is nearly identical to ttw/ttwmice.(18) We tested the hypothesis that deficient extracellu-lar PPi generation by ecto-nucleoside triphosphate pyro-phosphohydrolase (NTPPPH) PC-1 activity mediates bothincreased calcification and altered osteoblast gene expres-sion. We studied primary calvarial osteoblasts from PC-1�/� mice and tested for shared findings in ank/ank mice.Prompted by recent mRNA differential display analysesrevealing altered expression of the mineralization regulatorOPN in ear cartilages from ttw/ttw mice,(19) we examinedosteoblast OPN expression. OPN mediates osteoclasticbone resorption and can promote skeletal remodeling.(20–24)

Like PPi, OPN also is a potent, direct inhibitor of bonemineral formation and hydroxyapatite crystal growth.(25–27)

Our results reveal linked regulatory effects on calcificationby PC-1 and ANK through regulation of extracellular PPi

and OPN.

MATERIALS AND METHODS

Reagents

All chemical reagents were from Sigma (St Louis, MO,USA), unless otherwise indicated. Bovine serum albumin(BSA) and recombinant murine myeloma cell line-expressed, His-tagged OPN of �95% purity, in 50 �gBSA/�g OPN (catalog 499260), were from Calbiochem (LaJolla, CA, USA).

PC-1 null and ank/ank mouse models and genotyping

Generation and initial characterization of PC-1 null micewere described previously, and herein we used mice on a129/Sv background.(18) To determine PC-1 genotypes,genomic DNA was isolated from tails or from culturedprimary osteoblasts and analyzed using a 3-primer polymer-ase chain reaction (PCR) protocol. Primers for the targetingregion in wildtype PC-1 genomic DNA sequence were5�-CCCTTTGTGGTACAAAGGACAG-3� and 5�-GCAT-GACCCATTATACACTTTGT-3�; the primer for the tar-geting vector PGK-PolyA was 5�-GGGTGAGAACA-GAGTACCTAC-3�. This PCR reaction generated distinctproducts (1.2-kb PC-1 null allele, 750-bp wildtype PC-1allele). Initial Southern blots were used to confirm repro-ducibility of PCR screening for each genotype.

The ank/ank mouse colony used was on a hybrid back-ground (derived originally from crossing a C3H and

FIG. 1. Morphological changes in PC-1 null mice. (A–D) X-rays of100-day-old female (A and C)wildtype mice or (B and D) PC-1�/�

mice. Exposures were under identical conditions (mammography ap-paratus, soft X-rays). C and D are magnified from A and B, respec-tively. In the PC-1�/� mice, there was heavy calcification adjacent tothe intervertebral disks, while the vertebral bodies, arches, and spines,and the shafts of long bones show reduced bone density. (E and F)Macroscopic view of the anterior aspect of the lumbar vertebrae of100-day-old female (E) wildtype mice and (F) PC-1�/� mice. Fineforceps and a metal brush were used to remove muscle tissue surround-ing the vertebrae. Note the heavy calcification in the region of theintervertebral disks and growth plates of the PC-1�/� mice.

995PPi DEFICIENCY AND OSSIFICATION

C57BL/6 hybrid male with BALB/c female).(28) ANK ge-notypes were analyzed by PCR.(17) Heterozygote breederswere used to generate distinct litters containing PC-1 nulland ank/ank mice and their respective heterozygotic andwildtype littermates.

Isolation, culture, and transfection of primaryosteoblasts and reverse transcriptase-PCR analyses

Primary cultures of osteoblasts were isolated from in-tramembranous bone (calvariae) of 0- to 3-day-old pupsthrough sequential collagenase digestion, as described.(4)

Calvariae of the same genotype were pooled and an en-riched cell population of osteoblastic phenotype seeded at�4 � 104 cells/cm2 in �-MEM (Gibco-BRL, Grand Island,NY, USA), containing 10% heat inactivated FCS, glutamine(2 mM), penicillin (50 U/ml), and streptomycin (0.5 mg/ml). For studies under mineralizing conditions, media weresupplemented with �-glycerophosphate (2.5 mM) andL-ascorbic acid-PO4 (25 �g/ml) every third day (mediumA). We transfected cDNA expression constructs (2 �gDNA) for wildtype human PC-1 and B10/NPP3 inpcDNA3.1(2,4) into primary cells using Lipofectamine Plus(Life Technologies, Grand Island, NY, USA) with transfec-tion efficiency �40%.(1) To further verify osteoblastic dif-ferentiation, mouse osteoblasts were confirmed to express

osteocalcin and type I collagen, but not aggrecan or type IIcollagen, by reverse transcriptase (RT)-PCR, as de-scribed.(2,29)

For RT-PCR, total RNA was isolated from osteoblastsusing 0.5 ml TriZOL (Life Technologies)/35-mm dish.RNA was extracted and reverse transcribed as described.(1)

Mouse OPN primers(30) were sense, 5�-CTCCCGGA-GAAAGTGACTGA-3�, which amplified a 831-bp product.Mouse matrix Gla protein (MGP) primers(31) were sense,5�-TGCGCTGGCCGTGGCAACCCT-3�; and antisense,5�-CCTCTCTGTTGATCTCGTAGGCA-3�, which ampli-fied a 181-bp product. Mouse cbfa1 primers(32) were sense,5�-GAGGCACAAGTTCTATCTGGA-3�; and antisense,5�-GGTGGTCCGCGATGATCTC-3�, which amplified a336-bp product. Mouse osteoprotogerin (OPG) primers(33)

were sense, 5�-CTGGACATCATTGAATGGAC-3�; andantisense 5�-AATTAGCAGGAGGCCAAATG-3�, whichamplified a 495-bp product. Primers for the housekeepinggene loading control, the ribosomal protein L30, were aspreviously described.(1)

Assessment of calcification

To quantify calcification by osteoblasts, we used a pre-viously described Alizarin red S binding assay using cellsthat reached confluency.(34) Osteoblasts (5000 cells/well)

FIG. 2. Depression of intracellular and extracellular PPi and NPP activity and augmented calcification associated with cultured osteoblasts ofPC-1–deficient mice. Primary calvarial osteoblasts from PC-1�/�, PC-1�/�, and PC-1�/� mice were plated in 35-mm dishes. At confluency(designated day 1), 2.5 mM �-glycerophosphate and 25 �g/ml of ascorbic acid-PO4 were added to complete �-MEM growth medium to stimulatemineralization (as described in the Materials and Methods section); the medium was changed every 3 days. (A–C) Conditioned media and celllysates were collected at each time point indicated for assays of PPi and of the specific activity of NPP (for which 1 U was defined as 1 �molof substrate hydrolyzed per hour), as described in the Materials and Methods section. (D) For studies of calcification, primary calvarial osteoblasts(5000 cells per well) from PC-1�/�, PC-1�/�, and PC-1�/� mice were plated in 96-well plates, and at confluency, 2.5 mM �-glycerophosphateand 25 �g/ml of ascorbic acid-PO4 were added to complete �-MEM growth medium and designated day 1 (as described in the Materials andMethods section). Media were changed every 3 days, and at each designated time point, the media was removed and calcification measured asmicromoles of bound Alizarin red S normalized per microgram cell DNA, as described in the Materials and Methods section. (A–C) Ten micefrom each genotype were studied in triplicate, and results shown are pooled from all experiments. (D) Results pooled from studies on eight miceof each genotype studied in replicates of eight. (A) p � 0.05 relative to PC-1�/� cells for extracellular PPi levels on days 4–17 in PC-1�/� cellsand days 1–17 in PC-1�/� cells. (B and C) *p � 0.05 relative to PC-1�/� cells. (D) *p � 0.05 relative to wildtype cells.

996 JOHNSON ET AL.

were plated in 0.1 ml of medium A in a flat bottom 96-wellB-D Falcon 353075 culture plate (B-D Falcon, Bedford,MA, USA). At each time point, media were removed, and0.05 ml of diluted Alizarin red S (0.5% vol/vol Alizarin redS, pH 5.0) was added for 10 min.(34) Data were expressed asmicromoles of bound Alizarin red S per microgram DNA ineach well, determined chromogenically at OD570.(34)

Histology and immunohistochemistry

Mouse skeletal tissues were isolated by dissection andfixed in10% neutral buffered formalin for 2 days and thendecalcified in 4% hydrochloric acid, processed for histol-ogy, and embedded in paraffin. For immunohistologic anal-ysis of OPN, sections were deparaffinized, blocked with10% goat serum for 20 min, and incubated overnight at 4°Cwith rabbit polyclonal antibody to OPN (Chemicon, Te-mecula, CA, USA). Washed sections were incubated for 1 hat 22°C with biotinylated goat anti-rabbit IgG followed bya 1-h incubation with peroxidase-conjugated avidin. Perox-idase activity was detected using the Sigma Fast DABstaining kit, according to the manufacturer’s instructions.

Preparation of soluble PC-1

Stable expression of a soluble PC-1 cDNA construct inthe vector pSecTag2A (Invitrogen, Carlsbad, CA, USA)was established in 293 cells. The construct encoding solublePC-1 was generated by introducing a BamH1 site throughPCR at the junction of the transmembrane and extracellulardomains of PC-1, as described.(2) This site was used toligate PC-1 cDNA encoding the extracellular domain start-ing with the amino acid sequence PSCAKK into the BamHIsiteof pSEcTag2A, allowing PC-1 to be cleaved by signalpeptidase and secreted by transfected 293 cells. To isolatesoluble PC-1, 700 ml volume of conditioned cell mediacontaining soluble PC-1 was dialyzed overnight in 20 mMTris, pH 8.0, followed by additional dialysis for 2 h in freshbuffer. The dialyzed medium was fractionated on a

Q-Sepharose, Fast Flow column (Pharmacia, Piscataway,NJ, USA). The column was washed with 70 ml of 20 mMTris, pH 8.0, 50 mM NaCl, and soluble PC-1 eluted with 20mM Tris and 150 mM NaCl, pH 8.0. Soluble PC-1 wasconcentrated from fractions most enriched in NPP activity(and the preparation was verified to be PC-1–enriched bySDS-PAGE/Western blotting).

PPi, NPP, and alkaline phosphatase assays

PPi concentrations were measured radiometrically as de-scribed.(2) We determined specific activity of NPP colori-metrically using p-nitropheylthymidine monophosphate assubstrate for nucleotide phosphodiesterase I activity at al-kaline pH.(2) Specific activity of alkaline phosphatase (AP)was measured as described.(4) One unit of NPP (or AP) wasdefined as 1 �mol substrate hydrolyzed per hour (per mi-crogram protein/sample).

OPN ELISA

To measure OPN protein levels, we developed an ELISAbased on a previously published method,(35) using platescoated with a monoclonal antibody to native OPN (Chemi-con; using a 1:2500 dilution of OPN antibody in 100 �l/well0.1 M sodium bicarbonate, pH 9.0 at 4°C overnight). Wellswere blocked with 10 mM Tris, 150 mM NaCl, and 0.05%Tween-20, pH 8.0, for 1 h at 22°C. Samples, diluted in 1%BSA, 10 mM Tris, and 150 mM NaCl, pH 8.0, were addedto the wells and incubated for 1 h at 37°C (with dilutions of1:2–1:8 quantified). Recombinant murine OPN was used asstandard. Washed wells were incubated sequentially for 1 hat 37°C with rabbit anti-OPN (1:1000; Chemicon), biotin-ylated goat anti-rabbit IgG (1:1000), and streptavidin con-jugated with AP (1:500 dilution), and color was developedusing p-nitrophenyl phosphate and read at 405 nM.

FIG. 3. NPP family isozyme-specific rescue ofmatrix hypercalcification by PC-1 in mouse os-teoblasts lacking PC-1 expression: Primary cal-varial osteoblasts from PC-1�/� and control un-transfected PC-1�/� mice were isolated, and 5 �105 cells were plated in 60-mm dishes. ThePC-1�/� cells were transfected with 2 �g ofempty plasmid DNA or plasmid DNA expres-sion constructs for PC-1 or the NPP isozymeB10/NPP3, and aliquots of 5000 cells were stud-ied in 96-well plates as described in the Materi-als and Methods section. At confluency, 2.5 mM�-glycerophosphate and 25 �g/ml of ascorbicacid-PO4 were added to complete �-MEMgrowth medium, which was changed every 3days. (A) At each time point indicated, calcifi-cation was measured as micromoles of boundAlizarin red S normalized per microgram cellu-lar DNA, as described above. The results for (B)PPi and (C) NPP on day 10. Five mice from eachgenotype were studied in replicates of eight.Results shown are pooled from all experiments.*p � 0.05 relative to vector-transfected PC-1�/�

osteoblasts in A–C.

997PPi DEFICIENCY AND OSSIFICATION

Statistics

Where indicated, error bars represent SD. Statistical anal-yses were performed using Student’s t-test (paired two-sample testing for means).

RESULTS

Skeletal pathology in PC-1�/� mice

PC-1�/� mice were phenotypically indistinguishablefrom PC-1�/� mice. In PC-1�/� mice, we observed featuresessentially identical to the previously described phenotypeof ttw/ttw mice.(6–10) These included the development ofhyperostosis, starting at approximately 3 weeks of age, in aprogressive process that culminated in ossific intervertebralfusion (Figs. 1B, 1D, and 1F) and peripheral joint ankylosis,as well as Achilles tendon calcification (data not shown).

PC-1 and PPi in osteoblast calcification

We studied cultured primary calvarial osteoblasts to iden-tify basic mechanisms involved in abnormal mineralizationin PC-1�/� mice. PC-1 is one of several NPP familyisozymes expressed in the skeleton, yet PC-1 plays a dis-proportionately large role in supporting chondrocyte extra-cellular PPi relative to other NPP isozymes.(34) We observedthat extracellular PPi, intracellular PPi, and cell-associatedNPP specific activity (measured as alkaline nucleotide phos-phodiesterase I) all were relatively lower by approximately50% at early time points in culture in PC-1�/� osteoblastsrelative to PC-1�/� littermate control cells (Figs. 2A–2C).NPP activity and intracellular and extracellular PPi levelsrose in association with time in culture in osteoblasts of allgenotypes (Figs. 2A–2C). However, absolute differencesbetween PPi levels associated with PC-1�/� and PC-1�/�

osteoblasts remained similar over the same time in culture(Figs. 2A and 2B). In the PC-1�/� cells, levels of PPi andNPP were intermediate between those of wildtype and PC-1�/� cells (Figs. 2A–2C). The extent and rapidity of calci-fication also were markedly increased in PC-1�/� primarycalvarial osteoblasts relative to controls (Fig. 2D). In thesestudies, done on cells that had reached confluency, cellnumbers at confluency were comparable between genotypesand relative changes were the same whether the calcificationwas measured per well or per cell DNA (data not shown).

Transfection of PC-1 into PC-1�/� osteoblasts restoredintracellular and extracellular NPP activity levels and PPi,as well as the hypercalcification to approximately baselinelevels for wildtype cells (Fig. 3). Significantly, the restor-ative effects of PC-1 transfection on NPP activity andextracellular PPi were sustained at up to 10 days aftertransfection (Fig. 3). To assess if the correction of hyper-calcification was selective for PC-1, we compared the ef-fects of transfection of the NPP family isozyme B10/NPP3.Transfected B10/NPP3 significantly elevated NPP activityin the PC-1�/� osteoblasts, but the induced NPP activityremained predominantly intracellular (Fig. 3). B10/NPP3also significantly elevated intracellular but not extracellularPPi in the PC-1�/� osteoblasts (Fig. 3). In contrast to PC-1,B10/NPP3 induced a small decrease in hypercalcificationrelative to control cells (Fig. 3).

PC-1�/� cells also demonstrated a marked decrease inexpression of OPN mRNA relative to wildtype cells (Fig.4A). OPN expression was not appreciably distinct betweenthe cells from PC-1�/� and PC-1�/� mice (data not shown).The selectivity of the altered OPN expression was tested byevaluating expression in osteoblasts of the mineralization-regulating genes OPG, MGP, and cbfa1.(31–33) In contrast toOPN, only modest differences in expression of OPG, MGP,and cbfa1 mRNA were observed in PC-1�/� osteoblastsrelative to wildtype cells (Fig. 4A).

Next, we assessed if depressed OPN expression wasattributable to extracellular PPi deficiency in PC-1�/� os-teoblasts. Predicated on a difference in extracellular PPi

concentrations of approximately 500 pM between PC-1�/�

cells and PC-1�/� cells at multiple time points (Fig. 2), weoptimized pulsing of the extracellular medium with sodiumPPi to intermittently restore extracellular PPi in PC-1�/�

cells to the approximate level of PC-1�/� cells. Specifically,we determined that adding 0.05 pmol of sodium PPi per 0.1ml volume at days 3, 5, and 7 to the cultured PC-1�/� cellselevated extracellular PPi to within 5% of the levels of

FIG. 4. Diminished OPN expression by primary calvarial osteoblastsfrom PC-1�/� mice. We cultured PC-1�/� and PC-1�/� osteoblasts forthe number of days indicated under mineralizing conditions in com-plete �-MEM growth medium supplemented with 2.5 mM�-glycerophosphate and 25 �g/ml of ascorbic acid-PO4 every 3 days,as described above. (A) We analyzed the mRNA expression of selectedosteoblast genes, including OPN, OPG, MGP, and cbfa1 by RT-PCR incell culture, as described in the Materials and Methods section. Assayof the ribosomal “housekeeping gene” L30 served as the control forRNA loading. (B) To assess for potential reversal of the above-described changes in OPN expression by extracellular PPi supplemen-tation, PC-1�/� and PC-1�/� osteoblasts were cultured for the numberof days indicated in complete �-MEM growth medium supplementedwith 2.5 mM �-glycerophosphate and 25 �g/ml of ascorbic acid-PO4

every 3 days. Analyses of OPN and L30 mRNA expression by RT-PCRwere performed. Where indicated (as PC-1�/� � PPi), the PC-1�/�

cells were pulsed on days 0, 2, and 5 with exogenous sodium PPi (1pmol added to 2.0 ml medium to elevate extracellular PPi by approx-imately 500 pM, as described above). Four mice from each genotypewere studied. Results shown are representative of all mice studied.

998 JOHNSON ET AL.

PC-1�/� cells for a 24-h period (data not shown). Accord-ingly, we intermittently pulsed aliquots of mineralizing PC-1�/� cells at early time points in culture with 0.05 pmol ofsodium PPi per 0.1 ml (Fig. 4B). The decreased OPNexpression in PC-1�/� osteoblasts at day 1 was reversed bythe pulsing with sodium PPi (Fig. 4B). This effect of PPI

was not demonstrable at later time points in culture.

Reversal of hypercalcification in both PC-1�/� andank/ank osteoblasts by soluble PC-1 and OPN

Similar to PC-1�/� cells, the ank/ank osteoblasts hadmarked depression of extracellular PPi relative to wildtypecells (Fig. 5). However, intracellular PPi was significantlyelevated in the ank/ank osteoblasts (Fig. 5), similar to thecase in ank/ank fibroblasts.(17) Calcification was signifi-cantly accelerated and more extensive in cultured ank/ankosteoblasts relative to both wildtype ANK/ANK and het-erozygotic ANK/ank littermate cells (Fig. 5).

We focused on OPN mRNA expression at early timepoints in culture in ank/ank osteoblasts (Fig. 6). AlthoughOPN mRNA levels fluctuated widely between days 1–6 inthe ank/ank osteoblasts, there was depression of OPNmRNA expression at days 1 and 6 relative to ANK/ANKcells (Fig. 6A). By comparison, there was little change inOPG, MGP, and cbfa1 mRNA expression in ank/ank rela-tive to the wildtype osteoblasts (Fig. 6A), as was the casefor PC-1�/� cells above. As for the PC-1�/� cells above, thedecrease in OPN expression in ank/ank osteoblasts, wasreversible by PPi supplementation at day 1 in culture (Fig.6B), although this effect of PPI was not clearly demonstra-ble at later time points in culture.

Next, we examined for functional linkage between defec-tive PC-1 and ANK expression and OPN expression in

FIG. 6. PPi -reversible decrease in OPN expression by primarycalvarial osteoblasts from ank/ank mice. Using the same experimentalmethods as described for Fig. 3, we cultured osteoblasts of the indi-cated genotypes. (A) Analyses of mRNA for OPN, OPG, MGP, cbfa1,and L30 by RT-PCR, as described above. (B) Assessment for potentialreversal of the decreased OPN expression in ank/ank cells by extracel-lular PPi supplementation. As described above, osteoblasts of theindicated genotypes were studied. Where indicated (as ank/ank � PPi),the ank/ank cells had been pulsed on days 0, 2, and 5 with exogenoussodium PPi (1 pmol added to 2.0 ml medium to elevate extracellular PPi

by approximately 500 pM). Four mice from each genotype werestudied. Results shown are representative of all mice studied.

FIG. 5. Altered PPi levels associated with hy-percalcification in ank/ank primary calvarial os-teoblasts. (A) Primary calvarial osteoblasts fromANK/ANK, ANK/ank, and ank/ank mice, platedin 35-mm dishes, treated with medium supple-mented with 2.5 mM �-glycerophosphate and 25�g/ml ascorbic acid-PO4, were assessed for in-tracellular and extracellular PPi concentrations,and the specific activity of NPP at the timepoints indicated, as described above for Fig. 3.Ten mice from each genotype were studied intriplicate. Results shown are pooled from allexperiments. *p � 0.05 relative to ANK/ANK.(B) Confluent primary calvarial osteoblasts fromANK/ANK, ANK/ank, and ank/ank mice weretreated with 2.5 mM �-glycerophosphate and 25�g/ml ascorbic acid-PO4 in complete �-MEMgrowth medium (as described in the Materialsand Methods section), and matrix calcificationwas measured as bound Alizarin red S, as above.Results pooled from studies on eight mice ofeach genotype studied in replicates of eight.*p � 0.05 relative to wildtype cells.

999PPi DEFICIENCY AND OSSIFICATION

hypercalcification. First, we analyzed OPN expression qual-itatively by immunohistochemistry of PC-1�/� mousespines. OPN staining was only weakly detectable in thedisorganized trabecular bone in these samples, in contrast torobust OPN staining in trabecular bone in wildtype mice(Fig. 7, panels at 25�). We incidentally noted thinning ofthe nucleus pulposus in PC-1�/� mouse spines (Fig. 7,panels at 25�). OPN expression was detectable in thenucleus pulposus, as well as in the intervertebral ligaments(Fig. 7, panels at 6.25� and 13�).

Quantitative assessment of OPN expression in unfraction-ated newborn calvarial tissue by OPN ELISA demonstrateddepression (�50%) in OPN levels in PC-1�/� relative toPC-1�/� mice (Fig. 8A). Comparable in situ depression ofOPN was observed in calvarial extracts of ank/ank relative

to ANK/ANK mice (Fig. 8A). Correspondingly, OPN levelswere significantly depressed in vitro (by �15 pg/ml) inconditioned media of both cultured PC-1�/� and ank/ankcalvarial osteoblasts relative to their wildtype controls (Fig.8B). Next, we supplemented the culture media of PC-1�/�

and ank/ank osteoblasts with commercial recombinant mu-rine OPN purified from murine myelomatous cells andstabilized in BSA. OPN supplementation (at 1.5, 15, and150 pg/ml) did not significantly alter extracellular PPi (datanot shown). However, OPN supplementation, at �15 pg/ml,prevented the hypercalcification by both PC-1�/� and ank/ank osteoblasts (Fig. 9). We verified that addition of equiv-alent amounts of BSA to those in the OPN preparation (upto 7.5 ng/ml) had no effect under these conditions (data notshown).

FIG. 7. OPN expression in the spines of PC-1�/� mice. Deparaffinized sections of lumbarvertebrae from 4-month-old PC-1�/� and PC-1�/� mice were analyzed for immunohistochem-ical staining of OPN using rabbit polyclonalanti-OPN (1:100), biotinylated goat anti-rabbitIgG, and peroxidase-conjugated avidin, as de-scribed in the Materials and Methods section.For the negative staining control, an equal dilu-tion of rabbit serum was used in place of theprimary antibody. Note particularly the stainingfor OPN associated with trabecular bone of wild-type mice (arrows in upper left 25� panel) incontrast with the disorganized trabecular bone inthe PC-1�/� mouse associated with weak OPNstaining. T, trabecular bone; E, endochondralossification center of the vertebral body; N, nu-cleus pulposus.

1000 JOHNSON ET AL.

Last, we prepared soluble PC-1 and optimized restora-tion of depressed NPP specific activity of PC-1�/� pri-mary calvarial osteoblasts by titrated addition of solublePC-1. Specifically, addition of 4.92 U NPP/ml raisedextracellular NPP and extracellular PPi of PC-1�/� cellsto within 5% of their respective values in PC-1�/� cul-tures at each time point studied (data not shown). Wedemonstrated that addition of soluble PC-1 in this mannerlargely corrected the depression of extracellular PPi, thedepression of secreted OPN levels, and the hypercalcifi-cation in not only PC-1�/� but also ank/ank osteoblasts(Fig. 10).

DISCUSSION

In this study, we tested the functional significance inregulating calcification of PPi generation by PC-1 and PPi

transport by ANK. We directly demonstrated that deficientNPP-catalyzed PPi generation in PC-1�/� mice contributesto pathological calcification. Specifically, osteoblasts fromPC-1�/� mice demonstrated �50% depression of NPP-specific activity and intracellular and extracellular PPI atday 1 in culture. Support of extracellular PPi levels was aNPP family isozyme-specific function of PC-1 with uniqueimplications for regulation of calcification. Specifically,

FIG. 8. Depressed OPN levels in situ in calvariae and in vitro incultured osteoblasts from both PC-1�/� and ank/ank mice: (A) Cal-variae were isolated from newborn PC-1�/�, PC-1�/�, ANK/ANK, andank/ank mice and homogenized before the sequential digestion in 0.5%Triton X-100 in PBS. Diluted aliquots of the tissue extracts (containing100 ng–1 �g protein) were added to plates precoated with monoclonalanti-OPN antibody for 1 h, and the OPN ELISA was performed asdescribed in the Materials and Methods section. n � 8 calvariae of eachgenotype, studied in replicates of three at three dilutions (1:2, 1:4, and1:8), *p � 0.01. (B) OPN protein levels, in the conditioned media ofcultured PC-1�/� and ank/ank calvarial osteoblasts relative to theirwildtype littermate controls, were measured by ELISA. n � 8 of eachgenotype, *p � 0.01.

FIG. 9. Reconstitution with OPN prevents hypercalcification by bothPC-1�/� and ank/ank osteoblasts: We tested if correction of OPNlevels (back to the OPN levels of wildtype controls, which were �15pg/ml higher in vitro) would alter hypercalcification by cultured PC-1�/� and ank/ank osteoblasts. To do so, we supplemented the culturemedia of PC-1�/� and ank/ank osteoblasts with recombinant murineOPN (at 1.5, 15, or 150 pg/ml where indicated). We carried the cellsunder mineralizing conditions for up to 10 days in culture and thenmeasured extent of calcification by Alizarin red S binding. (n � 6,*p � 0.05 for the correction of matrix hypercalcification by OPN addedat 15 and 150 pg/ml).

1001PPi DEFICIENCY AND OSSIFICATION

“gain of function” of NPP activity in PC-1�/� osteoblastsinduced by transfection of B10/NPP3 restored intracellularPPi levels to normal. However, B10/NPP3, unlike PC-1,failed to restore depressed extracellular PPI to normal. Con-currently, the capacity of B10/NPP3 to reduce hypercalci-fication by PC-1�/� cells was modest relative to PC-1.

Our findings suggested thresholds for both PC-1 andextracellular PPi deficiency to heighten calcification by os-teoblasts to lie between the 25% and 50% depression ofPC-1 NPP activity and extracellular PPi relative to wildtypecells seen at early time points in culture. Specifically, os-teoblasts from phenotypically normal PC-1�/� mice dem-onstrated extracellular PPi levels (and NPP levels) interme-diate between PC-1�/� and PC-1�/� mice. However,cultured osteoblasts from PC-1�/� and PC-1�/� mice didnot show significant quantitative differences in calcification.Parallel findings for extracellular PPi and calcification wereseen comparing ank/ANK cells from phenotypically normalmice with ank/ank mouse-derived cells.

Cultured osteoblasts of PC-1�/� and ank/ank mice dem-onstrated decreased expression of OPN relative to wildtypecells. Changes in the OPN expression were at least partiallyselective, because the mineral resorption regulator OPG,(33)

the osteoblastic transcription factor cbfa1,(32) and the phys-iologic mineralization inhibitor MGP(31) demonstrated onlymodest differences in expression in cultured PC-1�/� andank/ank osteoblasts relative to wildtype cells. Importantly,the depressed OPN expression preceded the marked rise incalcification seen in both PC-1�/� and ank/ank osteoblastsin culture and therefore was not likely to be simply second-ary to heightened calcification.

Taken together, our results suggested that the remarkablephenotypic similarities between ank/ank and PC-1�/� miceat least partly reflect the common depression of extracellularPPi and OPN. First, hypercalcification and depression ofextracellular PPI and OPN expression were strikingly com-parable in cultured primary calvarial osteoblasts from ank/ank mice and PC-1�/� mice. Second, OPN protein levelswere diminished in PC-1�/� and ank/ank calvariae in situ.Third, soluble extracellular PC-1, which was added to pro-vide a stable source of continuing PPi generation and whichwas validated to effectively restore of PPi by over 0–14days in culture, jointly corrected extracellular PPi levels andOPN levels in both PC-1�/� and ank/ank osteoblasts. Fur-thermore, such restoration by soluble PC-1 of extracellularPPi and OPN levels corrected hypercalcification in osteo-blasts from both PC-1�/� and ank/ank mice.

In this study, a single pulse of exogenous PPI reversed thedepression in OPN mRNA expression by both PC-1�/� andank/ank osteoblasts at day 1 in culture. However, the ca-pacity of a pulse of exogenous PPi to reverse the decreasedOPN expression was attenuated with further time in cultureunder pro-mineralizing conditions. It is possible that metab-olism and disposition of pulsed exogenous PPi by pyrophos-phatases and other factors may alter in association withchanges in differentiation of osteoblasts with time in cul-ture. Thus, identification of the microenvironment(s) inwhich PC-1 and ANK both act to regulate PPi levels may bepertinent to further understanding of the regulation of OPNexpression and other functionally significant events in min-

eralizing cells. In this regard, intracellular PPi(17) was ele-

vated in ank/ank mouse osteoblasts but was depressed inPC-1�/� cells. We speculate that such differences in intra-cellular PPi may contribute to subtle distinctions betweenPC-1�/� and ank/ank mice. These include a longer time toinitial development of peripheral and spinal joint fusion inank/ank relative to PC-1�/� mice.(6,7,28,36)

Focal upregulation of OPN expression, already docu-mented in arterial calcification and most forms of patholog-ical “soft tissue” calcification described to date, is believedto play a physiologic restraining role in calcification.(20–27,37)

Post-translationally phosphorylated OPN,(20,27,38) like PPi,binds to hydroxyapatite crystals and inhibits hydroxyapatitecrystal propagation, and exogenous OPN potently inhibitscalcification by cultured cells.(20,21,26,27) Thus, the down-regulation of OPN demonstrated in PC-1�/� and ank/ankosteoblasts in vitro in this study was unexpected.

When challenged by ovariectomy or administration ofparathyroid hormone (PTH), OPN-deficient tissues clearlyshow impaired bone resorption.(21,22) However, unchal-lenged OPN knockout mice show only subtle changes inbone mineralization, such as greater hydroxyapatite crystalsize and crystallinity detectable by sensitive evaluation.(25)

Indeed, the grossly normal OPN null mouse phenotypestands in contrast to the phenotype of extracellularPPi-depleted PC-1�/� and ank/ank mice.(6–10,17,28,36) Thus,the capacity of restoration of OPN levels to normal usingexceedingly small doses (�15pg/ml) of OPN to prevent themarked hypercalcification by cultured PC-1�/� and ank/ankosteoblasts was unexpected in this study. We speculate thatthe direct inhibitory effect of OPN on hydroxyapatite dep-osition becomes substantially more significant at sites whereactively mineralizing cells are challenged by extracellulardepletion of PPi. In essence, the inhibitory effects of PPi andOPN on hydroxyapatite crystal deposition may be synergis-tic. Decreased OPN expression seems to be a permissivefactor for hypercalcification associated with extracellularPPi deficiency, and remarkably, extracellular PPi deficiencypromoted OPN deficiency. It will be of interest to determineif extracellular PPi affects additional activities of OPN, suchas mobilization of macrophages,(39) stimulation of activa-tion of several metalloproteinases,(40) modulation ofosteoclast-mediated bone resorption,(21–23) and promotionof connective tissue turnover in certain inflammatorystates.(41)

The basic cell signaling mechanisms by which decreasedPC-1 expression and extracellular PPi levels promoted bothincreased calcification and decreased OPN expression willbe of interest to define. There is no documentation of plasmamembrane receptor recognition of PPi or of active uptake ofextracellular PPi in mammalian cells.(11) However, extra-cellular PPi is a precursor of inorganic phosphate (Pi),including tissue nonspecific AP (TNAP)-catalyzed PPi hy-drolysis.(11) TNAP is an essential promoter of bone miner-alization that also acts as a PC-1 antagonist.(4,16) AlthoughPi stimulates calcification,(11,42) Pi uptake and signalingtriggered by Pi also promote OPN expression.(43) Given themutual correction of mineralization abnormalities by cross-ing PC-1�/� and TNAP�/� mice,(16) we speculate thatPC-1, ANK, and TNAP, through their effects on extracel-

1002 JOHNSON ET AL.

lular PPi, critically interface in the function of cells involvedin calcification.

In conclusion, our observations support the potential roleof therapeutic modulation of extracellular PPi for control ofsome forms of dysregulated calcification. The PPi analoguephosphocitrate(44) was previously ascertained to markedlysuppress calcification in vivo in ank/ank mice.(36) The dis-covery that either soluble PC-1 or OPN rescue hypercalci-fication of both PC-1�/� and ank/ank osteoblasts in vitrosuggests the potential therapeutic utility of these moleculesto regulate calcification.

ACKNOWLEDGMENTS

We gratefully acknowledge the technical assistance ofAndrew Wigg in preparation of the soluble PC-1. This studywas supported by grants from the Department of VeteransAffairs, National Institutes of Health (P01AGO7996,AR47908, AR40770, CA42595, DE12889), Monash Uni-versity, Novartis, and the National Health and MedicalResearch Council of Australia.

REFERENCES

1. Johnson K, Moffa A, Chen Y, Pritzker K, Goding J, Terkeltaub R1999 Matrix vesicle plasma cell membrane glycoprotein-1 regu-lates mineralization by murine osteoblastic MC3T3 cells. J BoneMiner Res 14:883–892.

2. Johnson K, Vaingankar S, Chen Y, Moffa A, Goldring MB, SanoK, Jin-Hua P, Sali A, Goding J, Terkeltaub R 1999 Differentialmechanisms of PPi production by plasma cell membraneglycoprotein-1 (PC-1) and B10 in chondrocytes. Arthritis Rheum42:1986–1997.

3. Solan JL, Deftos LJ, Goding JW, Terkeltaub R 1996 Expression ofthe nucleoside triphosphate pyrophosphohydrolase PC-1 is in-duced by basic fibroblast growth factor (bFGF) and modulated byactivation of the protein kinase A and C pathways in osteoblast-like osteosarcoma cells. J Bone Miner Res 11:183–192.

4. Johnson KA, Hessle L, Vaingankar S, Wennberg C, Mauro S,Narisawa S, Goding JW, Sano K, Millan JL, Terkeltaub R 2000Osteoblast tissue-nonspecific alkaline phosphatase antagonizes andregulates PC-1. Am J Physiol Cell Physiol 279:R1365—R1377.

5. Goding J, Terkeltaub R, Maurice M, Deterre P, Sali A, Belli S1998 Ectophosphodiesterase pyrophosphatase of lymphocytes andnonlymphoid cells: Structure and function of the PC-1 family.Immunol Rev 161:11–26.

6. Okawa A, Nakamura I, Goto S, Moriya H, Nakamura Y, IkegawaS 1998 Mutation in Npps in a mouse model of ossification of theposterior longitudinal ligament of the spine. Nat Genet 19:271–273.

7. Sakamoto M, Hosoda Y, Kojimahara K, Yamazaki T, YoshimuraY 1994 Arthritis and ankylosis in twy mice with hereditary mul-tiple osteochondral lesions: With special reference to calciumdeposition. Pathol Int 44:420–427.

8. Okawa A, Goto S, Moriya H 1999 Calcitonin simultaneouslyregulates both periosteal hyperostosis and trabecular osteopenia inthe spinal hyperostotic mouse (twy/twy) in vivo. Calcif Tissue Int64:239–247.

9. Baba H, Furusawa N, Fukuda M, Maezawa Y, Imura S, KawaharaN, Nakahashi K, Tomita K 1997 Potential role of streptozotocin inenhancing ossification of the posterior longitudinal ligament of the

FIG. 10. Corrective effects of soluble PC-1 onextracellular PPi, OPN levels, and hypercalcifi-cation by cultured PC-1�/� and ank/ank osteo-blasts. Primary calvarial osteoblasts from PC-1�/�, PC-1�/�, ANK/ANK, and ank/ank micewere isolated and cultured under mineralizingconditions. Where indicated, we added aliquotsof purified soluble PC-1 to achieve 4.92 U ofNPP activity to the cells on days 1, 3, 5, 7, 9, 11,and 14. At each time point indicated, the me-dium was removed and the wells analyzed forcalcium precipitation by the Alizarin red S as-say. The conditioned media were analyzed forextracellular PPi levels and OPN levels. Resultsare from replicates of eight samples derivedfrom osteoblasts pooled from 12 mice of eachgenotype. *p � 0.05. (A) Addition of solublePC-1 in this manner corrected the depression ofextracellular PPi in both PC-1 null and ank/ankosteoblasts. (B) Addition of soluble PC-1 in thismanner significantly corrected the depression ofsecreted OPN levels in both PC-1 null and ank/ank osteoblasts by day 14. (C) Addition of sol-uble PC-1 in this manner corrected the matrixhypercalcification by both PC-1�/� osteoblastsand ank/ank osteoblasts.

1003PPi DEFICIENCY AND OSSIFICATION

cervical spine in the hereditary spinal hyperostotic mouse (twy/twy). Eur J Histochem 41:191–202.

10. Furusawa N, Baba H, Imura S, Fukuda M 1996 Characteristics andmechanism of the ossification of posterior longitudinal ligament inthe tip-toe walking Yoshimura (twy) mouse. Eur J Histochem40:199–210.

11. Terkeltaub R 2001 Inorganic pyrophosphate (PPi) generation anddisposition in pathophysiology. Am J Physiol Cell Physiol 281:C1–C11.

12. Rutsch F, Vaingankar S, Johnson K, Goldfine I, Maddux B,Schauerte P, Kalhoff H, Boisvert WA, Superti-Furga A, TerkeltaubR 2001 PC-1 nucleoside triphosphate pyrophosphohydrolase defi-ciency in idiopathic infantile arterial calcification. Am J Pathol158:543–554.

13. Oyajobi BO, Caswell AM, Russell RG 1994 Transforming growthfactor beta increases ecto-nucleoside triphosphate pyrophosphataseactivity of human bone-derived cells. J Bone Miner Res 9:99–109.

14. Oyajobi BO, Russell RG, Caswell AM 1994 Modulation of ecto-nucleoside triphosphate pyrophosphatase activity of humanosteoblast-like bone cells by 1 alpha, 25-dihydroxyvitamin D3,24R, 25-dihydroxyvitamin D3, parathyroid hormone, and dexa-methasone. J Bone Miner Res 9:1259–1266.

15. Lotz M, Rosen F, McCabe G, Quach J, Blanco F, Dudler J, SolanJ, Goding J, Seegmiller JE, Terkeltaub R 1995 Interleukin 1 betasuppresses transforming growth factor-induced inorganic pyro-phosphate (PPi) production and expression of the PPi -generatingenzyme PC-1 in human chondrocytes. Proc Natl Acad Sci USA92:10364–10368.

16. Hessle L, Johnson KA, Anderson HC, Narisawa S, Sali A, GodingJW, Terkeltaub R, Millan JL 2002 Tissue nonspecific alkalinephosphatase and plasma cell membrane glycoprotein-1 are centralantagonistic regulators of bone mineralization. Proc Natl Acad SciUSA 99:9445–9449.

17. Ho AM, Johnson MD, Kingsley DM 2000 Role of the mouse ankgene in control of tissue calcification and arthritis. Science 289:265–270.

18. Sali A, Favaloro JM, Terkeltaub R, Goding JW 1999 Germlinedeletion of the nucleoside triphosphosphate (NTPPPH) plasma cellmembrane glycoprotein (PC-1) produces abnormal calcification ofperiarticular tissues. In: Vanduffel L, Lemmens R (eds.) Ecto-ATPases and Related Ectonucleotidases. Shaker Publishing BV,Maastricht, The Netherlands, pp. 267–282.

19. Koshizuka Y, Ikegawa S, Sano M, Nakamura K, Nakamura Y2001 Isolation of novel mouse genes associated with ectopic os-sification by differential display method using ttw, a mouse modelfor ectopic ossification. Cytogenet Cell Genet 94:163–168.

20. Sodek J, Ganss B, McKee M 2000 Osteopontin. Crit Rev Oral BiolMed 1:279–303.

21. Ihara H, Denhardt DT, Furuya K, Yamashita T, Muguruma Y,Tsuji K, Hruska KA, Higashio K, Enomoto S, Nifuji A, RittlingSR, Noda M 2001 Parathyroid hormone-induced bone resorptiondoes not occur in the absence of osteopontin. J Biol Chem 276:13065–13071.

22. Yoshitake H, Rittling SR, Denhardt DT, Noda M 1999Osteopontin-deficient mice are resistant to ovariectomy-inducedbone resorption. Proc Natl Acad Sci USA 96:8156–8160.

23. Yamate T, Mocharla H, Taguchi Y, Igietseme JU, Manolagas SC,Abe E 1997 Osteopontin expression by osteoclast and osteoblastprogenitors in the murine bone marrow: Demonstration of itsrequirement for osteoclastogenesis and its increase after ovariec-tomy. Endocrinology 138:3047–3055.

24. Mckee M, Nanci A 1996 Osteopontin at mineralized tissue inter-faces in bone, teeth, and osseointegrated implants: Ultrastructuraldistribution and implications for mineralized tissue formation,turnover, and repair. Microsc Res Tech 33:141–164.

25. Boskey A, Spevak L, Paschalis E, Doty S, McKee M 2002 Os-teopontin deficiency increases mineral content and mineral crys-tallinity in mouse bone. Calcif Tissue Int 71:145–154.

26. Wada T, McKee MD, Steitz S, Giachelli CM 1999 Calcification ofvascular smooth muscle cell cultures: Inhibition by osteopontin.Circ Res 84:166–178.

27. Jono S, Peinado C, Giachelli CM 2000 Phosphorylation of os-teopontin is required for inhibition of vascular smooth muscle cellcalcification. J Biol Chem 275:20197–20203.

28. Krug HE, Taurog JD 2000 HLA-B27 has no effect on the pheno-typic expression of progressive ankylosis in ank/ank mice. J Rheu-matol 27:1257–1259.

29. Hahnel AC, Rappolee DA, Millan JL, Manes T, Ziomek CA,Theodosiou NG, Werb Z, Pedersen RA, Schultz GA 1990 Twoalkaline phosphatase genes are expressed during early develop-ment in the mouse embryo. Development 110:555–564.

30. Craig AM, JH Smith, Denhardt DT 1989 Osteopontin, atransformation-associated cell adhesion phosphoprotein, is inducedby 12-O-tetradecanoylphorbol 13-acetate in mouse epidermis.J Biol Chem 264:9682–9689.

31. Bostrom K, Tsao D, Shen S, Wang Y, Demer LL 2001 Matrix GLAprotein modulates differentiation induced by bone morphogeneticprotein-2 in C3H10T1/2 cells. J Biol Chem 276:14044–14052.

32. Ducy P, Zhang R, Geoffroy V, Ridall A, Karsenty G 1997 Osf2/Cbfa1: A transcriptional activator of osteoblast differentiation. Cell89:747–754.

33. Takai H, Kanematsu M, Yano K, Tsuda E, Higashio K, Ikeda K,Watanabe K, Yamada Y 1998 Transforming growth factor-� stim-ulates the production of osteoprotegrin/osteoclasto-genesis inhibi-tory factor by bone marrow stromal cells. J Biol Chem 273:27091–27096.

34. Johnson K, Hashimoto S, Lotz M, Pritzker K, Goding J, TerkeltaubR 2001 Up-regulated expression of the phosphodiesterase nucleo-tide pyrophosphatase family member PC-1 is a marker and patho-genic factor for knee meniscal cartilage matrix calcification. Ar-thritis Rheum 44:1071–1081.

35. Bautista DS, Saad Z, Chambers AF, Tonkin KS, O’Malley FP,Singhal H, Tokmakejian S, Bramwell V, Harris JF 1996 Quanti-fication of osteopontin in human plasma with an ELISA: Basallevels in pre and postmenopausal women. Clin Biochem 29:231–239.

36. Krug HE, Mahowald ML, Halverson PB, Sallis JD, Cheung HS1993 Phosphocitrate prevents disease progression in murine pro-gressive ankylosis. Arthritis Rheum 36:1603–1611.

37. Nitta K, Ishizuka T, Horita S, Hayashi T, Ajiro A, Uchida K,Honda K, Oba T, Kawashima A, Yumura W, Kabaya T, Akiba T,Nihei H 2001 Soluble osteopontin and vascular calcification inhemodialysis patients. Nephron 89:455–458.

38. Zhu X, Luo C, Ferrier JM, Sodek J 1997 Evidence of ectokinase-mediated phosphorylation of osteopontin and bone sialoprotein byosteoblasts during bone formation in vitro. Biochem J 323:637–643.

39. McKee MD, Nanci A 1996 Secretion of Osteopontin by macro-phages and its accumulation at tissue surfaces during wound heal-ing in mineralized tissues: A potential requirement for macrophageadhesion and phagocytosis. Anat Rec 245:394–409.

40. Philip S, Bulbule A, Kundu GC 2001 Osteopontin stimulatestumor growth and activation of promatrix metalloproteinase-2through nuclear factor-kappa B-mediated induction of membranetype 1 matrix metalloproteinase in murine melanoma cells. J BiolChem 276:44926–44935.

41. Yumoto K, Ishijima M, Rittling SR, Tsuji K, Tsuchiya Y, Kon S,Nifuji A, Uede T, Denhardt DT, Noda M 2002 Osteopontin defi-ciency protects joints against destruction in anti-type II collagenantibody-induced arthritis in mice. Proc Natl Acad Sci USA 99:4556–4561.

42. Jono S, McKee M, Murry C, Shioi A, Nishizawa Y, Mori K, MoriiH, Giachelli C 2000 Phosphate regulation of vascular smoothmuscle cell calcification. Circ Res 87:E10–E17.

43. Beck GR Jr, Zerler B, Moran E 2000 Phosphate is a specific signalfor induction of osteopontin gene expression. Proc Natl Acad SciUSA 97:8352–8357.

44. Cheung HS 2001 Phosphocitrate as a potential therapeutic strategyfor crystal deposition disease. Curr Rheumatol Rep 3:24–28.

Address reprint requests to:R Terkeltaub, MD

VA Medical Center3350 La Jolla Village DriveSan Diego, CA 92161, USA

E-mail: rterkeltaub@ucsd.edu

Received in original form August 27, 2002; in revised form No-vember 11, 2002; accepted December 10, 2002.

1004 JOHNSON ET AL.