ARTHRITIS & RHEUMATISMVol. 42, No. 9, September 1999, pp. 1986–1997© 1999, American College of Rheumatology

DIFFERENTIAL MECHANISMS OF INORGANIC PYROPHOSPHATEPRODUCTION BY PLASMA CELL MEMBRANE GLYCOPROTEIN-1

AND B10 IN CHONDROCYTES

KRISTEN JOHNSON, SUCHETA VAINGANKAR, YING CHEN, ALLISON MOFFA,MARY B. GOLDRING, KIMIHIKO SANO, PIAO JIN-HUA, ADNAN SALI,

JAMES GODING, and ROBERT TERKELTAUB

Objective. Increased nucleoside triphosphate py-rophosphohydrolase (NTPPPH) activity in chondro-cytes is associated with cartilage matrix inorganic pyro-phosphate (PPi) supersaturation in chondrocalcinosis.This study compared the roles of the transforminggrowth factor b (TGFb)–inducible plasma cell mem-brane glycoprotein-1 (PC-1) and the closely related B10NTPPPH activities in chondrocyte PPi metabolism.

Methods. NTPPPH expression was studied usingreverse transcriptase–polymerase chain reaction andWestern blotting. Transmembrane PC-1 (tmPC-1),water-soluble secretory PC-1 (secPC-1), and transmem-brane B10 were expressed by adenoviral gene transfer orplasmid transfection, and expression of PPi was as-sessed in cultured articular chondrocytes and immor-

talized NTPPPH-deficient costal chondrocytes (TC28cells).

Results. PC-1 and B10 messenger RNA weredemonstrated in articular cartilages in situ, in un-treated cultured normal articular chondrocytes, and inTC28 cells. Expression of tmPC-1 and secPC-1, but notB10, rendered the NTPPPH-deficient TC28 cells able toincrease expression of extracellular PPi, with or withoutaddition of TGFb (10 ng/ml) to the media. More plasmamembrane NTPPPH activity was detected in cells trans-fected with tmPC-1 than in cells transfected with B10.Furthermore, confocal microscopy with immunofluores-cent staining of articular chondrocytes confirmed pref-erential plasma membrane localization of PC-1, relativeto B10. Finally, both PC-1 and B10 increased the levelsof intracellular PPi, but PC-1 and B10 appeared to actprincipally in different intracellular compartments(Golgi and post-Golgi versus pre-Golgi, respectively).

Conclusion. PC-1 and B10 NTPPPH activitieswere not redundant in chondrocytes. Although in-creased PC-1 and B10 expression caused elevations inintracellular PPi, the major effects of PC-1 and B10were exerted in distinct subcellular compartments.Moreover, PC-1 (transmembrane and secreted), but notB10, increased the levels of extracellular PPi. Differen-tial expression of PC-1 and B10 could modulate carti-lage mineralization in degenerative joint diseases.

Articular cartilage chondrocytes have the uniqueability to constitutively elaborate relatively largeamounts of extracellular inorganic pyrophosphate (PPi),and cartilage is the major source of free PPi in joints(1–3). Moreover, in aging and osteoarthritic (OA) artic-ular cartilage, a dysregulated increase in PPi elaborationin chondrocytes and other changes in the chondrocytedifferentiation and matrix composition promote calcium

Ms Moffa’s work was supported by a Student InvestigatorAward from the Stein Research Institute of Aging at the University ofCalifornia, San Diego. Dr. Goldring’s work was supported by NIHgrant P01-AR-43578 and a Biomedical Sciences Research Award fromthe Arthritis Foundation. Dr. Sano’s work was supported by theMinistry of Education, Science, Sports and Culture, and the HyogoTotal Health Association, Japan. Dr. Goding’s work was supported bythe National Health and Medical Research Council of Australia. Dr.Terkeltaub’s work was supported by a Merit Review Award from theDepartment of Veterans Affairs Medical Service, NIH grant P01-AG-07996, and a Biomedical Sciences Research Award from the ArthritisFoundation.

Kristen Johnson, BA, Sucheta Vaingankar, PhD, Ying Chen,MD, Allison Moffa, BA, Robert Terkeltaub, MD: Department ofVeterans Affairs Medical Center, University of California, San Diego;Mary B. Goldring, PhD: Beth Israel Deaconess Medical Center andHarvard Medical School, Boston, Massachusetts; Kimihiko Sano, MD,PhD, Piao Jin-Hua, PhD: Kobe University, Kobe, Japan; Adnan Sali,BSc, James Goding, MD, PhD: Monash University Medical School,Prahran, Australia.

Address reprint requests to Robert Terkeltaub, MD, Depart-ment of Veterans Affairs Medical Center, 3350 La Jolla Village Drive,San Diego, CA 92161.

Submitted for publication February 2, 1999; accepted inrevised form May 5, 1999.

1986

pyrophosphate dihydrate (CPPD) crystal deposition inthe chondrocyte pericellular matrix (1).

PPi is generated by numerous biosynthetic pro-cesses and biochemical reactions (4). Excess PPi produc-tion in cartilage has been associated with increasedPPi-generating nucleoside triphosphate pyrophosphohy-drolase (NTPPPH) activity in idiopathic and OA-associated CPPD crystal deposition disease (1,2,5–8).Moreover, of all tissues tested, chondrocytes have thehighest specific activity of NTPPPH (9).

In idiopathic CPPD crystal deposition disease,NTPPPH overactivity is observed both in cartilage andin dermal fibroblasts grown in explant culture (10).Furthermore, intracellular PPi is elevated in chondro-cytes, dermal fibroblasts, and lymphoblasts from af-fected members of certain kindreds with CPPD crystaldeposition disease (1,11) and in dermal fibroblasts ofpatients with idiopathic/sporadic CPPD crystal deposi-tion disease (10). These observations have suggested thepresence of multisystemic primary disorders of bothintracellular and extracellular PPi metabolism and NT-PPPH activity in certain subjects with CPPD crystaldeposition disease (1,10,11).

NTPPPH activity produces free PPi by hydrolysisof the phosphodiester I bond in both purine and pyrim-idine nucleoside triphosphates (12–14). A homologousgroup of NTPPPH species is termed the plasma cellmembrane glycoprotein-1 (PC-1), or phosphodiesterasenucleotide pyrophosphatase (PDNP), family (12). Eachidentified NTPPPH is an ecto-enzyme, which is func-tional in the lumen of certain intracellular organelles(e.g., the endoplasmic reticulum [ER]) and outside thecell (12,15–17).

The most widely distributed NTPPPH is PC-1(PDNP1) (12,18). The alternatively spliced brain, intes-tinal, and melanocyte NTPPPH, autotaxin/PD1a, istermed PDNP2 (12,16), and the neural, enteric, andgenitourinary tract species, B10, is termed PDNP3(12,19). Each of the identified PDNP genes encodes acell-bound class II (intracellular N-terminus) transmem-brane glycoprotein of 120–130 kd that shares a highlyhomologous extracellular domain containing 2 somato-medin B–like regions and a highly conserved catalyticsite (Figure 1) (12,14). B10 is predicted to form ho-modimers, and homodimerization of PC-1 by disulfide-bonding is well established (12). Soluble monomericPC-1 with NTPPPH activity also is liberated by proteo-lysis of the parent molecule (20), and soluble PC-1polypeptides exist in synovial fluids (21,22), in the con-ditioned media of cultured chondrocytes and plasmacells (20,21), and in the serum (20).

Articular chondrocytes, similar to certain othertissues (17), express more than 1 NTPPPH species,including distinct but incompletely characterized intra-cellular, plasma membrane, and water-soluble secretedNTPPPH activities (2,17). However, use of a monoclo-nal antibody specific for native PC-1 has shown thatPC-1 accounts for more than half (although not all) ofthe cell-associated NTPPPH activity in human articularchondrocytes (23). In this context, PC-1 expression, totalNTPPPH activity, and PPi elaboration all are induced bytransforming growth factor b (TGFb) (and inhibited byinterleukin-1 [IL-1]) in articular chondrocytes (23). Sim-ilarly, TGFb (and IL-1) regulate the amount of PC-1released in membrane-limited apoptotic bodies and ma-trix vesicles (24,25).

PC-1 increases NTPPPH activity and intracell-ular PPi expression without the addition of exogenousATP in fibroblasts (11). In the present study, we ob-served that PC-1 and B10 are both transcribed in normalhuman articular chondrocytes in vivo and in vitro. Thus,to shed light on how PPi is generated both inside andoutside chondrocytes, we directly examined and com-pared the functions of PC-1 and B10.

Figure 1. Models for structures of transmembrane (wild-type) plasmacell membrane glycoprotein-1 (tmPC-1), secretory PC-1 (secPC-1),and B10. The ecto-enzymes PC-1 and B10, although likely products ofantecedent gene duplication (see ref. 12), lack significant homology intheir cytoplasmic tails. They do share highly homologous intraluminal/extracellular domains with conserved catalytic sites and divalentcation-binding EF hand domains, although only B10 has an RGD celladhesion sequence in the ecto-enzyme domain. PC-1 is consistentlyexpressed as a covalent (disulfide-bonded) homodimer. B10 also ispredicted to dimerize, but it is not established if B10 consistently formscovalent disulfide-bonded homodimers. SecPC-1 was engineered to besecreted via replacement of the cytosolic and transmembrane domainsof PC-1 with the influenza hemagglutinin signal peptidase cleavagedomain.

PC-1 AND B10 IN CHONDROCYTE PPi METABOLISM 1987

MATERIALS AND METHODS

Materials. Unless otherwise indicated, chemical re-agents were from Sigma (St. Louis, MO). Recombinant humanTGFb1 was from R&D Systems (Minneapolis, MN).

Cartilage samples and cell culture. The sources of OAand normal human knee articular cartilage samples, andmethods for isolation and culture of knee articular chondro-cytes have been previously described (26). Chondrocytes weregrown in monolayer culture in Dulbecco’s modified Eagle’smedium (DMEM) high glucose with 10% fetal calf serum(FCS), 1% glutamine, 100 units/ml penicillin, and 50 mg/mlstreptomycin (26). Only first-passage cells were studied. Thecell line TC28 (or T/C-28), established after immortalization ofprimary juvenile costal chondrocytes using a retrovirus ex-pressing SV40-TAg and the neomycin-resistance selectionmarker (neoR) (27), was cultured in DMEM low glucose/Ham’s F12 (1:1) supplemented as for articular chondrocytes,and continuously passaged after reaching ;90% confluence ata 1:10 split ratio. Cell proliferation was measured by incorpo-ration of fluorogenic H33258 (Calbiochem, La Jolla, CA)according to the manufacturer’s instructions, and cell viabilitywas assessed by trypan blue exclusion.

Reverse transcriptase–polymerase chain reaction (RT-PCR), Western blotting, and immunodetection of B10 andPC-1. We used Trizol (Life Technologies, Grand Island, NY)to collect total RNA from cells and from whole cartilages notsubjected to further culture (in situ messenger RNA [mRNA]expression), and RNA was reverse transcribed as previouslydescribed (25,26). Following denaturation (94°C for 5 min-utes), complementary DNA (cDNA) was amplified in 1 PCRround (40 cycles at 94°C for 30 seconds, 55°C for 30 seconds,72°C for 1 minute, and a final extension for 5 minutes at 72°C).G3PDH was used to confirm equal RNA loading (26). Weconfirmed expression of aggrecan (28), type II collagen (29),and PC-1 (23) using previously referenced primer sets andprotocols.

B10 was detected using the sense primer 59-GGAGACACATCGCCTCTG-39 and the antisense primer59-AAGTCAAGCCCAGTGAGAAGTTC-39, which ampli-fied a 595-basepair B10 fragment. Autotaxin (12,16) wasdetected using the sense primer 59-CACATCGAATTAAG-AGAGCAGAAGG-39 and the antisense primer 59-TTGCTG-CCTTTCTTCATGTATGA-39, which amplified a 460-bp au-totaxin fragment. We verified the cDNA products to be eitherhuman B10 or autotaxin, by generation of the correctly sizedfragments following endonuclease digestion using specific in-ternal restriction sites, as well as by hybridization to theirrespective full-length human cDNA in Southern blotting assays.

Western blotting for immunodetection of PC-1 andB10 was performed as previously described, using the rabbitanti–PC-1 antibody R1769, which is specific for the 150–aminoacid C-terminal domain of PC-1 (25), and a rabbit antiserum toB10 (amino acids 193–616), which is encoded after insertion(Xmn I–Pst I) of a B10 cDNA fragment into pMAL-c2(Pharmacia, Piscataway, NJ) to produce a maltose-bindingprotein–B10 fusion protein in DH5 Escherichia coli. The fusionprotein, extracted using 0.1% Triton X-100, was purified byamylose resin chromatography.

For immunofluorescent staining/confocal microscopy,articular chondrocytes were seeded on 18-mm2 coverslips

coated with poly-D-lysine at a density of 3 3 105 cells/coverslip.At 48 hours, cells were rinsed with phosphate buffered saline(PBS) and fixed in 4% paraformaldehyde in PBS for 30minutes at 22°C, and washed 3 times. Where indicated, cellswere permeabilized with 0.1% Triton X-100 in blocking buffer(PBS with 2% goat serum and 0.02% thiomersal) for 10minutes, and cells were again treated with blocking buffer for45 minutes. Cells were stained with monoclonal antibodies thatwere specific for native PC-1 (3E8) (23) or RB13-6 (30)(provided by Dr. H. Deissler, Essen, Germany) for 18 hours at4°C in blocking buffer, washed 3 times in PBS, and incubatedwith Alexa 488–conjugated goat anti-mouse IgG (MolecularProbes, Eugene, OR) at 1:400 in blocking buffer for 1 hour at22°C. Coverslips were mounted with Slowfade media (Molec-ular Probes) and cells were studied using a Zeiss Axiovert100M laser scanning microscope (absorbance at 495 nm,emission at 519 nm).

Biochemical assays, and collection of membrane frac-tions. PPi expression was determined radiometrically andequalized for cell DNA, as previously described (26). Todetermine intracellular PPi, washed cells were heated at 65°Cfor 45 minutes, washed again, and lysed in 1% Triton X-100,1.6 mM MgCl2, and 0.2M Tris, pH 8.1 (lysis buffer). Wedetermined the amount of cell protein, as well as the specificactivity of NTPPPH and alkaline phosphatase as previouslydescribed (25,26).

To collect crude plasma membrane fractions (31),washed cell pellets were covered in hypotonic lysis buffer (5mM MgCl2, 1 mM EDTA, 0.5 mM dithiothreitol [DTT], 2 mMphenylmethylsulfonyl fluoride, 10 mg/ml leupeptin, 25 mMHEPES [pH 7.5], without detergent) at 4°C for 2 minutes. Theswollen cells were then scraped off the dish with a rubberpoliceman, lysed by vigorous vortexing for 2 minutes, and thencentrifuged at 1,000g at 4°C for 15 minutes to pellet anddiscard nuclei and unlysed cells, followed by centrifugation at100,000g for 30 minutes at 4°C (31). We collected the resultingsupernatant, which contained the cytosol, as well as otherproteins released from organelles by the described cell lysisprocedure, and termed this the “intracellular” fraction. Thepellet from the initial 100,000g spin was resuspended inhypotonic lysis buffer supplemented with 0.1% Triton X-100,mixed at 4°C for 2 hours, and centrifuged at 100,000g for 20minutes. We thereby isolated the Triton-soluble, crude “plas-ma membrane” fraction and verified enrichment (;5-fold) ofalkaline phosphatase in this “plasma membrane” fraction ineach preparation. NTPPPH activity was not detected in anyTriton-insoluble pellets from the 100,000g spins. Thus, suchpellets were not further studied.

To assay ATP, we modified the standard luciferaseassay (32). Briefly, we washed and pelleted 2.5 3 106 cells in 10mM PBS and snap-froze the cell pellets in a dry ice/ethanolbath, then added 0.25 ml of 0.2M HCl followed by 0.3 ml 2 mMglycine (pH 9.0). Samples were heated at 100°C for 5 minutes,then placed on ice. The standard curve used 0–500 pmoles ofpurified ATP (disodium salt). The cell extract (or purifiedATP), in 0.1 ml, was injected into 0.1 ml of the ATP assaymixture (containing MgSO4, DTT, EDTA, luciferase, andluciferin all diluted 1:25; from the Sigma BioluminescenceAssay Kit). Following a delay of 5 seconds and integration timeof 15 seconds, samples were read on a TD-20/20 Luminometer(Turner Designs, Sunnyvale, CA).

1988 JOHNSON ET AL

Complementary DNA constructs and plasmid trans-fection. We used cDNA constructs to encode full-lengthhuman transmembrane PC-1 (tmPC-1) (in pcDNA3) andfull-length human B10 (in pBKCMV) (models for PC-1 andB10 structures are schematized in Figure 1). We also used afull-length enzyme-deficient mutant of mouse PC-1 (mutPC1),which is 81% identical to human PC-1 (in pSVT7, containingan alanine-for-threonine substitution in the active site [13,14]).In addition, we generated a water-soluble, secretory form ofhuman PC-1 (secPC-1). To do so, we used PCR to introduce aBam HI site at the junction of the transmembrane andextracellular domains of PC-1, which did not alter PC-1expression (20). We used this site to ligate the PC-1 extracel-lular domain to the signal peptide sequence of influenzahemagglutinin in pSHT, downstream of the SV40 early pro-moter (33). Thus, secPC-1 lacked the transmembrane andcytosolic domains, but was able to be targeted to the ER,cleaved by signal peptidase, and secreted (33) (Figure 1).

For plasmid transfections of TC28 cells, we usedLipofectamine Plus (Life Technologies) as previously de-scribed (25), achieving a transfection efficiency of $50% ofTC28 cells, as estimated using b-galactosidase in pcDNA3.Briefly, we diluted 2 mg of plasmid DNA in serum-freemedium and complexed it with the Lipofectamine Plus reagentfor 15 minutes at 22°C, after which Lipofectamine (6 ml),diluted in serum-free media, was added to this mixture andincubated at 22°C for 15 minutes. Then, 0.8 ml of media (with1% FCS, 1% glutamine) was added to each well and theDNA/Lipofectamine Plus mixture was added to the cells andincubated for 3 hours at 37°C, followed by replacement withfresh complete medium and incubation for 72 hours.

Adenoviral gene transfer. We subcloned cDNA encod-ing wild-type full-length human PC-1 (the Bam HI–Xba Ifragment from pcDNA3.PC-1) and cDNA encoding the mouseenzyme-deficient PC-1 mutant (released by Hind III digestionfrom pSVT7) into the Bam HI– and Xba I–digested or HindIII–digested pACCMVpLpA (which contains the genomicsequence of the replication-defective E1 mutant of adenovirus5 and the SV40 poly A tail) (25). To generate adenovirus thatexpressed secPC-1, we released a cassette containing thecytomegalovirus promoter, influenza signal peptide domain,and 59 truncated PC-1 from pcDNA3 containing secPC-1, byNru I and Xba I digestion. We inserted the released cDNA insense orientation in the promoterless adenoviral 5 E1 mutantshuttle vector ATV (from Dr. Tom Kipps, University ofCalifornia, San Diego) digested with Eco RV and Xba I.Resulting plasmids (5 mg) were each cotransfected, usingSuperfect (Qiagen, Valencia, CA), with 5 mg of adenoviralplasmid pJM17 into 293 cells, which provide adenoviral E1 intransfection. After plaque purification, individual recombinantplaques were amplified, and isolated by CsCl banding.

Statistical analysis. Where indicated, values are themean 6 SD. Statistical analysis was performed using Student’st-test (paired 2-sample testing for means).

RESULTS

PC-1 and B10 expression by normal articularchondrocytes. We detected unequivocal in situ PC-1 andB10 transcription, but only rare autotaxin transcription,

by articular chondrocytes that were isolated from normaland OA knees (Figure 2). Furthermore, cultured artic-ular chondrocytes from normal knees transcribed themRNA of PC-1 and B10, but autotaxin mRNA expres-sion was consistently below the limit of detection (Fig-ure 2).

We also used clonal immortalized human costalchondrocytes (TC28 cells) to assess chondrocyte func-tion (25,27,34–36), and confirmed their expression oftype II collagen and aggrecan (results not shown). PC-1and B10, as well as autotaxin, mRNA expression weredetectable in TC28 cells (Figure 2). However, basalNTPPPH activity was relatively sparse in TC28 cells(Figure 3), which facilitated direct evaluation of theroles of PC-1 and B10 in PPi metabolism in TC28 cells.

PPi elaboration by chondrocytes in vitro. We firstexamined the effects of PC-1 and B10 NTPPPH activi-ties on the ability of TGFb to increase extracellular PPi(37–39). TGFb moderately, but significantly, increasedcell-associated NTPPPH activity in articular chondro-cytes (23,37), but not in TC28 cells (Figure 3). Further-more, TGFb increased both intracellular PPi and extra-

Figure 2. Qualitative reverse transcriptase–polymerase chain reaction(RT-PCR) screening for nucleoside triphosphate pyrophosphohydro-lase (NTPPPH) expression. We screened for expression of mRNA of3 members of the phosphodiesterase nucleotide pyrophosphatase/NTPPPH family (plasma cell membrane glycoprotein-1 [PC-1], B10,and autotaxin). RNA was prepared and reverse transcribed andstudied by RT-PCR as described in Materials and Methods. Weassessed expression in situ by isolating RNA immediately at the time ofisolation of knee articular cartilages during total knee replacementfrom 3 donors with end-stage osteoarthritis (lane 1), and at the time ofautopsy from 3 donors with normal knees (lane 2). We also assessedcultured articular chondrocytes from the normal knees of 2 differentdonors (lane 3) and cultured TC28 cells (immortalized human costalchondrocytes) (lane 4), treated as described in Materials and Methods,and mRNA was assessed after 24 hours in culture. Results (not shown)were similar in 3–5 additional human donors of articular cartilages andchondrocytes.

PC-1 AND B10 IN CHONDROCYTE PPi METABOLISM 1989

cellular PPi (23,37) in articular chondrocytes, but not inTC28 cells (Figure 3).

Under these conditions, TGFb induced a prolif-erative response (i.e., 50% more TC28 cells relative tocontrol at 24 hours) as well as doubling of intracellularATP (P , 0.001 [n 5 8]) (results not shown). Thus, theinability of TC28 cells to increase extracellular PPi inresponse to TGFb was not due to global unresponsive-ness to TGFb. Low NTPPPH activity was associatedwith low extracellular PPi (and intracellular PPi) inTC28 cells relative to articular chondrocytes (Figure 3).Therefore, we compared how increased expression ofPC-1, by itself, regulated PPi metabolism.

Direct induction of PC-1 expression. Efficienttransfection of human articular chondrocytes with plas-mid DNA is not generally feasible (40). Thus, to directlyup-regulate PC-1 expression by these cells, we generatedrecombinant replication-defective adenovirus that ex-pressed wild-type tmPC-1 (Figure 1), enzyme-deficienttransmembrane PC-1 (mutPC-1), and secPC-1 (Figure1), as described in Materials and Methods.

Extracellular PPi increased significantly in responseto adenoviral gene transfer of both tmPC-1 and secPC-1 inarticular chondrocytes and TC28 cells (Figure 4A). In thisregard, adenoviral gene transfer of tmPC-1 in humanarticular chondrocytes doubled cell-associated NTPPPHactivity at 72 hours (Figure 4B). Significantly, adenoviralgene transfer of secPC-1 more than doubled extracellularNTPPPH at 72 hours, but little associated change occurred

Figure 4. Effects of adenoviral expression of tmPC-1 and secPC-1 onextracellular PPi and NTPPPH activity in articular chondrocytes andTC28 cells. First-passage normal articular chondrocytes and TC28 cellswere infected with 5 3 107 plaque-forming units of adenoviral humantmPC-1, secPC-1, or the naked pJM17 adenovirus for 6 hours in 2%fetal calf serum–containing medium. Medium was then changed to themedia as described in Materials and Methods and cells were culturedfor 72 hours. Intracellular and extracellular PPi (A), NTPPPH (B), andalkaline phosphatase (data not shown) were measured at 72 hours.Using adenoviral b-galactosidase and X-gal staining, we confirmedthat at least 80% of cells were infected under these conditions.Alkaline phosphatase was not altered, relative to controls, by anyadenoviral infection, and an additional control, adenoviral mutPC-1,did not affect PPi or NTPPPH (data not shown). Bars show themean 6 SD of 8 experiments performed in triplicate. p 5 P , 0.01; #5 P , 0.001, versus controls. See Figures 1–3 for definitions.

Figure 3. Elaboration of inorganic pyrophosphate (PPi) and NTPPPHactivity in articular chondrocytes and TC28 cells. We isolated humanarticular chondrocytes from the knees of normal donors and studied thecells in first passage in monolayer culture (2 3 105 cells per well, asdescribed in Materials and Methods). The same number of TC28 cellswere studied. Where indicated, transforming growth factor b (TGFb; 10ng/ml) was added throughout 72 hours of culture. Intracellular andextracellular PPi (A), NTPPPH (B), and alkaline phosphatase (data notshown) were measured. There were no significant changes in alkalinephosphatase activity under these conditions (data not shown). Bars showthe mean and SD of 7 experiments performed in triplicate using 7different articular chondrocyte donors. The mean values for PPi levelsand NTPPPH activity in unstimulated TC28 cells and articular chondro-cytes were comparable with the control values obtained in subsequentexperiments. p 5 P , 0.01 versus controls. See Figure 2 for otherdefinitions.

1990 JOHNSON ET AL

in cell-associated NTPPPH activity at 72 hours in articularchondrocytes. NTPPPH activity similarly increased inTC28 cells in response to adenoviral gene transfer, butincreases in NTPPPH were proportionately greater in therelatively NTPPPH-poor TC28 cells (Figure 4). The addi-tional negative control, adenoviral mutPC-1, had no signif-icant effect on NTPPPH or PPi (results not shown).

Differential effects of PC-1 and B10 on PPigeneration. To assess whether the ability of PC-1 toincrease extracellular PPi was selective, we comparedthe effects of PC-1 and B10. We were not able totechnically achieve successful recombination of adeno-viral constructs based on pJM17 that could incorporatethe particularly large (4-kb) B10 cDNA. Thus, we nextused plasmid DNA transfection to express B10 and PC-1in TC28 cells.

B10 and tmPC-1 transfection increased both cell-associated NTPPPH activity and the intracellular con-

centration of PPi (Figure 5). Furthermore, doubling ofcell-associated NTPPPH activity in association with in-creased expression of tmPC-1 rendered TC28 cells ableto further increase extracellular PPi in response toTGFb (Figure 5). However, B10, unlike PC-1, failed toincrease extracellular PPi in unstimulated and TGFb-stimulated TC28 cells (Figure 5). Moreover, in cellstransfected with B10, TGFb did not increase intracell-ular PPi beyond the increase induced by B10 alone,unlike that seen with tmPC-1 (Figure 5).

The differential ability of PC-1 and B10 to in-crease extracellular PPi suggested the possibility ofdifferential localization at the extracellular surface ofchondrocytes. To test this possibility, we first isolatedthe chondrocyte subcellular fraction enriched in plasmamembranes, and we also studied a crude “intracellular”fraction containing cytosol and other proteins liberatedby the cell lysis procedure, as described in detail inMaterials and Methods. Using Western blotting, moreconstitutive PC-1 than B10 expression was detectable inthe crude plasma membrane fractions of untransfectedarticular chondrocytes (Figure 6). We next confirmedthis finding by demonstrating sparse immunofluorescentstaining of B10 relative to PC-1 on the surface ofnonpermeabilized articular chondrocytes studied byconfocal microscopy (Figures 7b and d). Permeabiliza-tion of articular chondrocytes enabled detection of B10(in a compartmentalized perinuclear distribution) (Fig-

Figure 5. Effects of transfection of PC-1 compared with B10 on theability of TGFb to influence extracellular PPi and NTPPPH in TC28cells. We transfected TC28 cells with plasmid expression constructs fortmPC-1 and B10, or empty plasmid DNA, as described in Materialsand Methods, and added TGFb (10 ng/ml) where indicated, followedby measurement of PPi (A) and NTPPPH (B) after 72 hours. Barsshow the mean and SD from 6 experiments. p 5 P , 0.01 versuscontrols. See Figures 1–3 for definitions.

Figure 6. Preferential localization in the plasma membrane of plasmacell membrane glycoprotein-1 (PC-1) relative to B10 in chondrocytes.We separated crude intracellular and plasma membrane fractions ofcultured normal articular chondrocytes and of TC28 cells, as describedin detail in Materials and Methods, at 72 hours after transfection oftransmembrane PC-1 or B10. We performed qualitative analysis forthe presence of immunoreactive B10 (A) and PC-1 (B) on 0.1 mg ofprotein from each of these fractions, which was separated by sodiumdodecyl sulfate–polyacrylamide gel electrophoresis under reducingconditions followed by Western blotting, using antibodies that werepredetermined to be specific and to discriminate between PC-1 andB10. Arrows indicate the detection of PC-1 and B10 polypeptides of;125–130 kd.

PC-1 AND B10 IN CHONDROCYTE PPi METABOLISM 1991

ure 7a). Permeabilization has the potential fordetergent-induced release of proteins from the plasmamembrane, and in this context, overall PC-1 stainingdiminished somewhat after permeabilization (Figure 7cversus Figure 7d).

Membrane PC-1 and B10 were both sparse inuntransfected TC28 cells, but we readily detected anincrease in PC-1, but not B10, in the crude plasmamembrane fractions of transfected TC28 cells by West-ern blotting (Figure 6). We also observed more cellsurface PC-1 staining than B10 staining after transfec-tionoftheirrespectivecDNA,usingimmunofluorescence/confocal microscopy (results not shown). Concurrently,

NTPPPH activity in crude plasma membrane fractionsmarkedly increased in TC28 cells transfected withtmPC-1, but not in those transfected with B10 (Figure8). In TC28 cells, plasma membrane NTPPPH did notincrease in response to TGFb (10 ng/ml) alone (Figure8). However, in articular chondrocytes, plasma mem-brane NTPPPH activity increased markedly in responseto both adenoviral gene transfer of tmPC-1 and stimu-lation with 10 ng/ml TGFb (Figure 8).

Thus, an increase in NTPPPH activity at the cellsurface in the plasma membrane appeared essential forthe action of tmPC-1 (and TGFb) to increase the levelsof extracellular PPi. To further distinguish where PC-1and B10 acted to generate PPi, we carried out time-

Figure 7. Confocal microscopy and immunofluorescent staining ofuntransfected articular chondrocytes for B10 and plasma cell mem-brane glycoprotein-1 (PC-1). Articular chondrocytes were seeded on18-mm2 coverslips coated with poly-D-lysine at a density of 3 3 105

cells/coverslip. At 48 hours, washed cells were fixed in 4% paraformal-dehyde in phosphate buffered saline (PBS) for 30 minutes at 22°C,and, where indicated, cells were permeabilized with 0.1% Triton X-100in blocking buffer (PBS with 2% goat serum and 0.02% thiomersal) for10 minutes (a, c, and e), as described in Materials and Methods. Cellswere stained for native B10 (a and b) or native PC-1 (c and d) (usingmonoclonal antibodies RB13-6 or 3E8, respectively), for 18 hours at4°C, followed by incubation with Alexa 488–conjugated goat anti-mouse IgG. Control cells incubated without primary antibody areindicated (e and f). Stained cells were studied using a Zeiss Axiovert100M laser scanning microscope (absorbance at 495 nm, emission at519 nm). The composite image of the fluorescence and differentialinterference contrast (Nomarski) images is shown for each condition.Arrows in a–d indicate selected areas of positive immunofluorescentstaining.

Figure 8. Plasma membrane NTPPPH activity before and after forcedexpression of PC-1 and B10. TC28 cells (A) were treated with TGFb(10 ng/ml) or transfected with empty plasmid, tmPC-1, or B10. Thecrude plasma membrane and intracellular fractions were isolated 72hours later, as described in detail in Materials and Methods. Articularchondrocytes (B) were treated with TGFb or adenovirally infectedwith PC-1 or empty control virus. The specific activity of NTPPPH perequivalent amounts of protein was measured for each fraction (relativeto untransfected controls). Bars show the mean and SD of 5 separateexperiments. There was no significant change in alkaline phosphatase–specific activity (which was 5-fold enriched in plasma membranerelative to intracellular fractions) under any of these conditions (datanot shown). Mean control values for plasma membrane and intracell-ular NTPPPH activity were 0.262 and 0.175 units/mg protein, respec-tively, in TC28 cells, and 0.743 and 0.762 units/mg protein, respectively,in articular chondrocytes. p 5 P , 0.001 versus untransfected controls.See Figures 1–3 for definitions.

1992 JOHNSON ET AL

course studies and treated transfected TC28 cells withbrefeldin A (BFA) (41). By collapsing the Golgi complexinto the ER, BFA disrupts protein translocation to theplasma membrane (41–43). Secreted proteins that beara transient signal peptide (33), as well as most mem-

brane proteins, use a BFA-sensitive Golgi–ER exocy-totic pathway (41,42).

In dose-response studies, we selected conditionsto treat TC28 cells with BFA (5 mg/ml BFA added at 48hours after transfection) in which .90% of cells re-

Figure 9. Effects of the Golgi-disrupting agent Brefeldin A (BFA) on NTPPPH movement to the plasma membrane (A) and on PPi concentrations(B) in transfected TC28 cells. TC28 cells were treated with BFA (5 mg/ml BFA added to culture 48 hours after transfection, and thereafter left inthe medium of cultured cells). Under these conditions, we verified that . 90% of cells remained viable at 72 hours after transfection, and there wasno inhibition of expression of NTPPPH in response to transfection of tmPC-1, secPC-1, or B10 (results not shown). Plasma membrane(detergent-soluble fractions) and intracellular fractions were collected at the times indicated, and NTPPPH and PPi were measured in triplicate.Results are representative of 4 separate experiments. Results for inhibition by BFA of intracellular PPi and of movement of NTPPPH into theplasma membrane fraction and of BFA-induced accumulation of intracellular NTPPPH activity all were significant (P , 0.01) in tmPC-1–transfectedcells. See Figures 1–3 for other definitions.

PC-1 AND B10 IN CHONDROCYTE PPi METABOLISM 1993

mained viable, and no change in cellular ATP levels orexpression of transfected NTPPPH cDNA occurred at72 hours after transfection (results not shown). BFAtreatment did not decrease constitutive NTPPPH activ-ity (Figure 9A) or alkaline phosphatase in the plasmamembrane, and did not appear to enhance proteolyticdegradation of either B10 or PC-1 in TC28 cells (resultsnot shown). BFA treatment also did not hinder theincrease in extracellular PPi generation in response tosecPC-1, which induced marked release of extracellularNTPPPH activity prior to BFA (Figure 9).

The ability of transfected tmPC-1 to elevateextracellular PPi was delayed relative to secPC-1 (Figure9B), consistent with gradual transport of tmPC-1 to thecell surface. BFA inhibited movement to the membrane,and possible recycling to the membrane, of most NTP-PPH after transfection with tmPC-1 (Figure 9A). BFAincreased intracellular retention of a fraction of NTP-PPH in cells transfected with PC-1 (Figure 9A). How-ever, BFA did not inhibit the ability of tmPC-1/NTPPPHthat had reached the plasma membrane to increaseextracellular PPi (Figure 9).

BFA treatment did not significantly block theincreases in intracellular NTPPPH activity and intracell-ular PPi caused by transfection with B10 (Figure 9). Incontrast, the BFA treatment attenuated the ability ofnewly expressed PC-1 to increase intracellular PPi (Fig-ure 9B). Thus, the intracellular compartments in whichPC-1 and B10 generated PPi appeared distinct.

DISCUSSION

This study identified that NTPPPH species tran-scribed by human articular chondrocytes included notonly PC-1 (21,23), but also B10, and to a lesser degree,autotaxin. The results also suggested a paradigm inwhich PC-1 and B10 generated PPi largely in distinctcellular compartments (Figure 10). TmPC-1 and B10both increased intracellular PPi in NTPPPH-poor TC28cells. However, more tmPC-1 than B10 moved to the cellsurface, and tmPC-1 and secPC-1, but not B10, in-creased extracellular PPi. The ability of PC-1 to increaseextracellular PPi was not limited by intracellular PPigeneration. Furthermore, only a modest increase inplasma membrane–associated NTPPPH appeared suffi-cient for PC-1 to augment extracellular PPi.

The differences in the predominant subcellularlocalizations and functions of PC-1 and B10 in chondro-cytes may reflect the distinct PC-1 and B10 cytosolic tails(12) (Figure 1). In this context, PC-1 and B10 areconcomitantly expressed in rat hepatocytes, but differ-

ences in the PC-1 and B10 cytosolic tails account for theselective localization of PC-1 to the basolateral mem-brane and of B10 to the apical membrane (44).

Our data suggested a model in which B10 movedinto the ER and/or other pre-Golgi compartments.However, chondrocytes might selectively glycosylate B10into forms that preferentially cluster in intracellularorganelles rather than translocate to the plasma mem-brane (45). Studies using BFA supported a proposedmodel (11,46) by which extracellular PPi elaboration bychondrocytes is critically regulated by Golgi-mediatedtranslocation to the membrane or by secretion of PC-1or by another secreted NTPPPH activity (Figure 10).Within chondrocytes, PC-1 might principally generatePPi in post-ER compartments, including the Golgi. Inthe cartilage matrix in idiopathic chondrocalcinosis (47),supersaturation with PPi may partly reflect elevatedintraarticular ATP, which may provide ample substratefor membrane and secreted NTPPPH (2,46–48).

Figure 10. Hypothesized model for differential mechanisms of PPiproduction by PC-1 and B10 in chondrocytes. The model, based onbackground knowledge and the results of this study, illustrates thedifferences in how transmembrane PC-1, transmembrane B10, andsoluble, secreted NTPPPH activities affect intracellular and extracel-lular PPi generation in chondrocytes. B10 is proposed to moveprincipally to pre-Golgi compartments, likely including the endoplas-mic reticulum (ER), where B10 could then generate intracellular PPiin the lumen. In comparison with B10, PC-1 is proposed to preferen-tially move to the plasma membrane through Golgi-mediated translo-cation, and PC-1 is proposed to principally generate PPi withinpost-ER intracellular compartments including the Golgi, as well as inthe cell exterior. See Figures 1–3 for other definitions.

1994 JOHNSON ET AL

Our results suggest that dysregulated PC-1 ex-pression could contribute to the elevations of bothintracellular and extracellular PPi seen in chondrocytesand dermal fibroblasts in certain patients with chondro-calcinosis (1,10,11). Although PPi does not freely diffuseacross membranes, it is co-secreted with collagen bychondrocytes (49). Thus, leakage and/or active extracel-lular transport of intracellular PPi and the NTPPPHsubstrate ATP from intracellular stores have been pro-posed (46,50). However, results with B10 transfection,and BFA treatment of TC28 cells transfected withtmPC-1 and secPC-1, suggest that intracellular PPiproduction is not needed for NTPPPH to raise extracel-lular PPi (Figure 10).

In this study, induction of PC-1, but not B10,expression enabled NTPPPH-deficient TC28 cells toincrease extracellular PPi in response to TGFb. TC28cells were responsive to TGFb, as evidenced by bothincreased proliferation and elevation of intracellularATP. However, TGFb did not induce increased NTP-PPH activity or translocation to the plasma membranein TC28 cells. Furthermore, immunoreactive PC-1 wasnot detectable in the plasma membrane of TC28 cellswithout a forced, direct increase in PC-1 expression.

The capacity of TGFb to increase the extracellu-lar PPi concentration of articular chondrocytes is uniquein comparison with most chondrocyte growth factors(37–39). In articular chondrocytes, TGFb stimulatesanaerobic glycolysis, an ATP-generating process with arelatively high basal rate in these cells, even underaerobic conditions (51). Our results suggest that to beable to induce an increase in extracellular PPi in chon-drocytes, TGFb likely requires a threshold level ofmembrane PC-1 or secreted NTPPPH activity (Figure9). Furthermore, the capacity of TGFb to raise extra-cellular PPi in articular chondrocytes relies on synergis-tic abilities to induce PC-1 expression (23) and toenhance movement of PC-1 to the plasma membraneand extracellular space (23) (Figure 9).

TGFb did not augment the ability of B10 trans-fection to raise intracellular PPi. Thus, it will be ofinterest to test if TGFb selectively increases the avail-ability of ATP to PC-1/NTPPPH in the extracellularspace or in the Golgi lumen and Golgi-derived vesicles,where ATP is known to be actively pumped in (52)(Figure 10). It should be noted that extracellular ATPlevels in cultured chondrocytes are below the conven-tional limits of detection, and use of an alternative methodused to test hypothetical TGFb-induced extracellular ATPrelease, using 14C-adenine labeling, previously yieldednegative results in porcine articular chondrocytes (37).

The results of this study indicate that individualarticular chondrocyte NTPPPH activities are not redun-dant. Thus, a dysregulated increase in PC-1 expressionappears more likely than an increase in B10 expressionto promote pathologic supersaturation of the cartilagematrix with PPi and thereby favor CPPD crystal forma-tion in the pericellular matrix in chondrocalcinosis (1).

Our results are also consistent with the conceptthat in normal articular cartilage chondrocytes, B10expression may not compensate for deficiencies in PC-1expression. In this regard, a critical physiologic role ofabundant, NTPPPH-mediated PPi elaboration by chon-drocytes in articular cartilage is to produce an excess offree PPi to Pi, which inhibits basic calcium phosphate(hydroxyapatite) crystal deposition (25,53). Moreover,deficient expression of enzymatically active PC-1 in vivois associated with pathologic mineralization of articularcartilage with hydroxyapatite in tiptoe-walking mice (54).

We recently demonstrated that the ability ofPC-1 to restrain matrix mineralization with hydroxy-apatite is exerted, in part, by the amounts of both PC-1and PPi directly associated with matrix vesicles (25).Therefore, we speculate that a shift in predominantNTPPPH expression from PC-1 to B10 has the potentialto decrease extracellular PPi elaboration and promotehydroxyapatite deposition by articular chondrocytes indegenerative joint disease (55,56). In this context, IL-1,which has been implicated in the pathogenesis of OA,inhibits PC-1 expression by chondrocytes (23). Furtherstudies to compare the regulation of PC-1 and B10expression in healthy and diseased articular chondro-cytes should be of interest.

In conclusion, PC-1 and B10 are both expressedby articular chondrocytes, but they generate PPi princi-pally in different cellular compartments. PC-1 has agreater capacity than B10 to move to the chondrocyteplasma membrane, and transmembrane and secretedPC-1 have a greater capacity than transmembrane B10to increase the concentration of extracellular PPi inchondrocytes. Our findings show how differential regu-lation of the expression of NTPPPH species in chondro-cytes could modulate pathologic mineralization of artic-ular cartilage in degenerative arthropathies.

ACKNOWLEDGMENTS

J. Quach and Dr. M. Lotz (The Scripps ResearchInstitute, La Jolla, CA) characterized and provided the artic-ular cartilages. Dr. H. Deissler generously provided a mono-clonal antibody (RB13-6) specific for native human B10. Dr.M. Stracke (National Cancer Institute, Frederick, MD) suggestedthe specific autotaxin RT-PCR primers used in this study.

PC-1 AND B10 IN CHONDROCYTE PPi METABOLISM 1995

REFERENCES

1. Ryan LM, McCarty DJ. Calcium pyrophosphate crystal depositiondisease, pseudogout, and articular chondrocalcinosis. In: Koop-man W, editor. Arthritis and allied conditions: a textbook ofrheumatology, 13th ed. Baltimore: Williams and Wilkins; 1997. p.2103–26.

2. Ryan LM, McCarty DJ. Understanding inorganic pyrophosphatemetabolism: toward prevention of calcium pyrophosphate dihy-drate crystal deposition. Ann Rheum Dis 1995;54:939–41.

3. Rosenthal AK, McCarty BA, Cheung HS, Ryan LM. A compari-son of the effect of transforming growth factor b1 on pyrophos-phate elaboration from various articular tissues. Arthritis Rheum1993;36:539–42.

4. Rachow J, Ryan L. Inorganic pyrophosphate metabolism in arthri-tis. Rheum Dis Clin North Am 1988;14:289–302.

5. Muniz O, Pelletier J-P, Martel-Pelletier J, Morales S, Howell D.NTP pyrophosphohydrolase in human chondrocalcinotic and os-teoarthritic cartilage. I. Some biochemical characteristics. ArthritisRheum 1984;27:186–92.

6. Pattrick M, Hamilton E, Hornby J, Doherty M. Synovial fluidpyrophosphate and nucleoside triphosphate pyrophosphatase:comparison between normal and disease and between inflamedand non-inflamed joints. Ann Rheum Dis 1991;50:214–8.

7. Hamilton E, Pattrick M, Doherty M. Inorganic pyrophosphate,nucleoside triphosphate pyrophosphatase, and cartilage fragmentsin normal human synovial fluid. Br J Rheumatol 1991;30:260–4.

8. Tenenbaum J, Muniz O, Schumacher HR, Good AE, Howell DS.Comparison of phosphohydrolase activities from articular carti-lage in calcium pyrophosphate deposition disease and primaryosteoarthritis. Arthritis Rheum 1981;24:492–500.

9. Cardenal A, Masuda I, Haas AL, McCarty DJ. Specificity of aporcine 127-kd nucleotide pyrophosphohydrolase for articulartissues. Arthritis Rheum 1996;39:245–51.

10. Ryan LM, Wortmann RL, Karas B, Lynch MP, McCarty DJ.Pyrophosphohydrolase activity and inorganic PPi content of cul-tured human skin fibroblasts: elevated levels in some patients withcalcium pyrophosphate dihydrate crystal deposition disease. J ClinInvest 1986;77:1689–93.

11. Terkeltaub R, Rosenbach M, Fong F, Goding J. Causal linkbetween nucleotide pyrophosphohydrolase overactivity and in-creased intracellular inorganic pyrophosphate generation demon-strated by transfection of cultured fibroblasts and osteoblasts withplasma cell membrane glycoprotein-1: relevance to calcium pyro-phosphate dihydrate deposition disease. Arthritis Rheum 1994;37:934–41.

12. Goding J, Terkeltaub R, Maurice M, Deterre P, Sali A, Belli S.Ectophosphodiesterase/pyrophosphatase of lymphocytes and non-lymphoid cells: structure and function of the PC-1 family. Immu-nol Rev 1998;161:11–26.

13. Belli SI, Goding JW. Biochemical characterization of human PC-1,an enzyme possessing alkaline phosphodiesterase I and nucleotidepyrophosphatase activities. Eur J Biochem 1994;226:433–43.

14. Belli SI, Mercuri FA, Sali A, Goding JW. Autophosphorylation ofPC-1 (alkaline phosphodiesterase I/nucleotide pyrophosphatase)and analysis of the active site. Eur J Biochem 1995;228:669–76.

15. Hickman S, Wong-Yip YP, Rebbe N, Greco JM. Formation oflipid-linked oligosaccharides by MOPC 315 plasmacytoma cells:decreased synthesis by a nonsecretory variant. J Biol Chem1985;260:6096–106.

16. Stracke ML, Clair T, Liotta LA. Autotaxin, tumor motility-stimulating exophosphodiesterase. Adv Enzyme Regul 1997;37:135–44.

17. Masuda I, Hamada J-I, Haas A, Ryan L, McCarty D. A uniqueectonucleotide pyrophosphohydrolase associated with porcinechondrocyte-derived vesicles. J Clin Invest 1995;95:699–704.

18. Harahap AR, Goding JW. Distribution of the murine plasma cellantigen PC-1 in nonlymphoid tissues. J Immunol 1988;41:2317–20.

19. Jin-Hua P, Goding JW, Nakamura H, Sano K. Molecular cloningand chromosomal localization of PD-Ibeta (PDNP3), a newmember of the human phosphodiesterase I genes. Genomics1997;45:412–5.

20. Belli SI, van Driel IR, Goding JW. Identification and character-ization of a soluble form of the plasma cell membrane glycoproteinPC-1. Eur J Biochem 1993;217:421–8.

21. Huang R, Rosenbach M, Vaughn R, Provvedini D, Rebbe N,Hickman S, et al. Expression of the murine plasma cell nucleotidepyrophosphohydrolase PC-1 is shared by human liver, bone, andcartilage cells: regulation of PC-1 expression in osteosarcoma cellsby transforming growth factor-b. J Clin Invest 1994;94:560–7.

22. Masuda I, Cardenal A, Ono W, Hamada J, Haas AL, McCartyDJ. Nucleotide pyrophosphohydrolase in human synovial fluid.J Rheumatol 1997;24:1588–94.

23. Lotz M, Rosen F, McCabe G, Quach J, Blanco F, Dudler J, et al.IL-1b suppresses TGFb-induced inorganic pyrophosphate (PPi)production and expression of the PPi-generating enzyme PC-1 inhuman chondrocytes. Proc Natl Acad Sci U S A 1995;92:10364–8.

24. Hashimoto S, Ochs RL, Rosen F, Quach J, McCabe G, Solan J, etal. Chondrocyte-derived apoptotic bodies and calcification ofarticular cartilage. Proc Natl Acad Sci U S A 1998;95:3094–9.

25. Johnson K, Moffa A, Pritzker K, Chen Y, Goding J, Terkeltaub R.Matrix vesicle plasma cell membrane glycoprotein-1 (PC-1) regu-lates mineralization by murine osteoblastic MC3T3 cells. J BoneMiner Res 1999;14:883–92.

26. Terkeltaub R, Lotz M, Johnson K, Deng D, Hashimoto S,Goldring MB, et al. Parathyroid hormone–related protein isabundant in osteoarthritic cartilage, and the parathyroidhormone–related protein 1-173 isoform is selectively induced bytransforming growth factor b in articular chondrocytes and sup-presses generation of extracellular inorganic pyrophosphate. Ar-thritis Rheum 1998;41:2152–64.

27. Goldring M, Birkhead J, Suen L-F, Yamin R, Mizuno S, GlowackiJ, et al. Interleukin-1b-modulated gene expression in immortalizedhuman chondrocytes. J Clin Invest 1994;94:2307–16.

28. Lum ZP, Hakala BE, Mort JS, Recklies AD. Modulation of thecatabolic effects of interleukin-1 beta on human articular chon-drocytes by transforming growth factor-beta. J Cell Physiol 1996;166:351–9.

29. Holderbaum D, Malemud CJ, Moskowitz RW, Haqqi TM. Humancartilage from late stage familial osteoarthritis transcribes type IIcollagen mRNA encoding a cysteine in position 519. BiochemBiophys Res Comm 1993;192:1169–74.

30. Blass-Kampmann S, Kindler-Rohrborn A, Deissler H, D’Urso D,Rajewsky MF. In vitro differentiation of neural progenitor cellsfrom prenatal rat brain: common cell surface glycoprotein on threeglial cell subsets. J Neurosci Res 1997;48:95–111.

31. Whitfield JF, Isaacs RJ, Jouishomme H, MacLean S, Chakra-varthy BR, Morley P, et al. C-terminal fragment of parathyroidhormone-related protein, PTHrP-(107-111), stimulatesmembrane-associated protein kinase C activity and modulates theproliferation of human and murine skin keratinocytes. J CellPhysiol 1996;166:1–11.

32. Lin S, Cohen HP. Measurement of adenosine triphosphate con-tent of crayfish stretch receptor cell preparations. Anal Biochem1968;24:531–40.

33. Madison E, Bird P. Vector, pSHT, for the expression and secretionof protein domains in mammalian cells. Gene 1992;121:179–80.

34. Loeser RF, Varnum BC, Carlson CS, Goldring MB, Liu ET,Sadiev S, et al. Human chondrocyte expression of growth-arrest–specific gene 6 and the tyrosine kinase receptor axl: potential rolein autocrine signaling in cartilage. Arthritis Rheum 1997;40:1455–65.

35. Moulton PJ, Hiran TS, Goldring MB, Hancock JT. Detection ofprotein and mRNA of various components of the NADPH oxidase

1996 JOHNSON ET AL

complex in an immortalized human chondrocyte line. Br J Rheu-matol 1997;36:522–9.

36. Cawston TE, Curry VA, Summers CA, Clark IM, Riley GP, LifePF, et al. The role of oncostatin M in animal and humanconnective tissue collagen turnover and its localization within therheumatoid joint. Arthritis Rheum 1998;41:1760–71.

37. Rosenthal AK, Cheung HS, Ryan LM. Transforming growthfactor b1 stimulates inorganic pyrophosphate elaboration by por-cine cartilage. Arthritis Rheum 1991;34:904–11.

38. Olmez U, Ryan L, Kurup I, Rosenthal A. Insulin-like growthfactor-1 suppresses pyrophosphate elaboration by transforminggrowth factor b1-stimulated chondrocytes and cartilage. Osteoar-thritis Cartilage 1994;2:149–54.

39. Rosen F, McCabe G, Quach J, Solan J, Terkeltaub R, SeegmillerJE, et al. Differential effects of aging on human chondrocyteresponses to transforming growth factor b: increased pyrophos-phate production and decreased cell proliferation. ArthritisRheum 1997;40:1275–81.

40. Terkeltaub RA. The immortalized chondrocyte: art without sacri-fice. J Clin Invest 1994;94:2173.

41. Klausner RD, Donaldson JG, Lippincott-Schwartz J. Brefeldin A:insights into the control of membrane traffic and organelle struc-ture. J Cell Biol 1992;16:1071–80.

42. Salamero J, Sztul ES, Howell KE. Exocytic transport vesiclesgenerated in vitro from the trans-Golgi network carry secretoryand plasma membrane proteins. Proc Natl Acad Sci U S A1990;87:7717–21.

43. Ripley C, Fant J, Bienkowski R. Brefeldin A inhibits degradationas well as production and secretion of collagen in human lungfibroblasts. J Biol Chem 1993;268:3677–82.

44. Scott LJ, Delautier D, Meerson NR, Trugnan G, Goding JW,Maurice M. Biochemical and molecular identification of distinctforms of alkaline phosphodiesterase I expressed on the apical andbasolateral plasma membrane surfaces of rat hepatocytes. Hepa-tology 1997;25:995–1002.

45. Fukushi M, Amizuka N, Hoshi K, Ozawa H, Kumagai H, OmuraS, et al. Intracellular retention and degradation of tissue-nonspecific alkaline phosphatase with a Gly3173Asp substitutionassociated with lethal hypophosphatasia. Biochem Biophys ResCommun 1998;246:613–8.

46. Masuda I, Ryan LM, McCarty DJ. Inorganic pyrophosphatemetabolism. In: Smyth SJ, Holers VM, editors. Gout, hyperurice-mia, and other crystal-associated arthropathies. New York: MarcelDekker; 1999. p. 359–67.

47. Ryan LM, Rachow JW, McCarty DJ. Synovial fluid ATP: apotential substrate for the production of inorganic pyrophosphate.J Rheumatol 1991;18:716–20.

48. Park W, Masuda I, Cardenal-Escarcena A, Palmer DL, McCartyDJ. Inorganic pyrophosphate generation from adenosine triphos-phate by cell-free human synovial fluid. J Rheumatol 1996;23:665–71.

49. Ryan L, Kurup I, Cheung H. Stimulation of cartilage inorganicpyrophosphate elaboration by ascorbate. Matrix 1991;11:276–81.

50. Rosenthal A, Ryan L. Probenecid inhibits transforming growthfactor-b1 induced pyrophosphate elaboration by chondrocytes.J Rheumatol 1994;21:896–900.

51. Stefanovic-Racic M, Stadler J, Georgescu HI, Evans CH. Nitricoxide and energy production in articular chondrocytes. J CellPhysiol 1994;159:274–80.

52. Capasso JM, Keenan TW, Abeijon C, Hirschberg CB. Mechanismof phosphorylation in the lumen of the Golgi apparatus: translo-cation of adenosine 59-triphosphate into Golgi vesicles from ratliver and mammary gland. J Biol Chem 1989;264:5233–40.

53. Meyer JL. Can biological calcification occur in the presence ofpyrophosphate? Arch Biochem Biophys 1984;231:1–8.

54. Okawa A, Nakamura I, Goto S, Moriya H, Nakamura Y, IkegawaS. Mutation in Npps in a mouse model of ossification of theposterior longitudinal ligament of the spine. Nat Genet 1998;19:271–3.

55. Halverson PB, McCarty D. Basic calcium phosphate (apatite,octacalcium phosphate, tricalcium phosphate) crystal depositiondiseases; calcinosis. In: Koopman W, editor. Arthritis and alliedconditions: a textbook of rheumatology. 13th ed. Baltimore:Williams and Wilkins; 1997. p. 2127–46.

56. Derfus B, Kranendonk S, Camacho N, Mandel N, Kushnaryov V,Lynch K, et al. Human osteoarthritic cartilage matrix vesiclesgenerate both calcium pyrophosphate dihydrate and apatite invitro. Calcif Tissue Int 1998;63:258–62.

ErrataIn the article by McGonagle et al published in the June 1999 issue of Arthritis & Rheumatism (pp1080–1086), it was stated that HLA–B27 transgenic rats showed radiographic evidence of enthesitis. Infact, the radiographic features of disease were not reported in the article cited (reference 54 of theMcGonagle article) and have yet to be investigated. However, some of the histologic changes in axialand peripheral joints are strongly suggestive of an enthesitis-associated pathology.

In the article by Wilson et al published in the July 1999 issue of Arthritis & Rheumatism (pp 1309–1311),the second footnote of Table 1 was incorrect. The footnote should have read, “Pregnancy morbiditycriteria were mainly developed by Branch and Silver (3).”

In the article by Del Rincon and Escalante published in the July 1999 issue of Arthritis & Rheumatism(pp 1329–1338), the X-axis label for the far right data column in Figure 1B was incorrect. The labelshould have read “DRB1*08 POSITIVE/EPITOPE POSITIVE.” Also, in the first full paragraph on page1336, the value of the confidence interval was listed incorrectly. The sentence should have read,“Excluding these 2 alleles, 48 of 253 patient alleles (19%) had D rather than Q or R at position 70,compared with 56 of 93 control alleles (60%) (OR 0.35, 95% CI 0.21–0.60).

We regret the errors.

PC-1 AND B10 IN CHONDROCYTE PPi METABOLISM 1997