Cloning and characterization of osteoactivin, a novel cDNA expressed in osteoblasts

Journal of Cellular Biochemistry 84:12±26 (2002)

ARTICLES

Cloning and Characterization of Osteoactivin,A Novel cDNA Expressed in Osteoblasts

Fayez. F. Safadi,1* Jie Xu,1 Steven L. Smock,2 Mario C. Rico,1 Thomas A. Owen,2 and Steven N. Popoff1

1Department of Anatomy and Cell Biology, Temple University School of Medicine, Philadelphia,Pennsylvania 191402Department of Cardiovascular and Metabolic Diseases, P®zer Global Research and DevelopmentÐGrotonLaboratories, Groton, Connecticut 06340

Abstract Osteoblast development is a complex process involving the expression of speci®c growth factors andregulatory proteins that control cell proliferation, differentiation, and maturation. In this study, we used the rat mutation,osteopetrosis (op), to examine differences in skeletal gene expression between mutant op and normal littermates. TotalRNA isolated from long bone and calvaria was used as a template for mRNA differential display. One of many cDNAsthat were selectively expressed in either normal or mutant bone was cloned and sequenced and found to share somehomology to the human nmb and Pmel 17 genes. This novel cDNA was named osteoactivin. Osteoactivin has an openreading frame of 1716 bp that encodes a protein of 572 amino acids with a predicted molecular weight of 63.8 kD.Protein sequence analysis revealed the presence of a signal peptide and a cleavage site at position 23. The protein alsohas thirteen predicted N-linked glycosylation sites and a potential RGD integrin recognition site at position 556.Northern blot analysis con®rmed that osteoactivin was 3- to 4-fold overexpressed in op versus normal bone. RT-PCRanalysis showed that osteoactivin is most highly expressed in bone compared with any of the other non-osseous tissuesexamined. In situ hybridization analysis of osteoactivin in normal bone revealed that it is primarily expressed inosteoblasts actively engaged in bone matrix production and mineralization. In primary rat osteoblast cultures,osteoactivin showed a temporal pattern of expression being expressed at highest levels during the later stages of matrixmaturation and mineralization and correlated with the expression of alkaline phosphatase and osteocalcin. Our ®ndingsshow that osteoactivin expression in bone is osteoblast-speci®c and suggest that it may play an important role inosteoblast differentiation and matrix mineralization. Furthermore, osteoactivin overexpression in op mutant bone maybe secondary to the uncoupling of bone resorption and formation resulting in abnormalities in osteoblast geneexpression and function. J. Cell. Biochem. 84: 12±26, 2002. ß 2001 Wiley-Liss, Inc.

Key words: osteoactivin; osteoblast differentiation; osteopetrosis

Bone is a very dynamic tissue that undergoescontinuous modeling/remodeling. Because oste-oblastic bone formation and osteoclastic boneresorption are antagonistic processes that pro-ceed simultaneously, bone metabolism must betightly regulated at all times. During the ®rsttwo decades of life when the skeleton is growing,there must be a net increase in bone mass suchthat bone formation exceeds bone resorption.

Once the skeleton has reached maturity, theremust be a constant balance between formationand resorption to ensure that there is no netgain or loss of bone; this highly regulatedbalance is called coupling [Marks and Hermey,1996]. When the rate of bone resorption exceedsthat of bone formation there is a progressive lossof bone (osteopenia), and the resulting osteo-porosis is associated with a variety of conditionssuch as aging, post-menopausal estrogen de®-ciency, in¯ammation, and chronic steroid treat-ment. In fact, uncoupling of formation andresorption is a common feature of most meta-bolic bone diseases [Marks and Hermey, 1996].As bone loss progresses, the structural inte-grity of the skeleton is compromised resultingin an increased incidence of bone fractures.With an aging population, the exponential

ß 2001 Wiley-Liss, Inc.DOI 10.1002/jcb.1259

Grant sponsor: P®zer, Inc. (to S.N.P. and F.F.S.); Grantsponsor: Temple University School of Medicine (to BoneResearch Enterprise Program).

*Correspondence to: Fayez. F. Safadi, Ph.D., Department ofAnatomy and Cell Biology, Temple University School ofMedicine, 3400 N. Broad Street, Philadelphia, PA 19140.E-mail: [email protected]

Received 5 June 2001; Accepted 3 July 2001

increase in the incidence of osteoporotic frac-tures has become a major health care issueand a signi®cant cause of morbidity andmortality.

Osteoblasts are derived from mesenchymalstem cells and differentiate into pre-osteoblastsand then mature osteoblasts under the in¯u-ence of genetic, local and systemic factors[Trif®tt, 1996]. Our understanding of the rolethat these factors play in regulating osteoblastdifferentiation and function has dramaticallyincreased in recent years. A number of differenttechniques such as differential display [Ryooet al., 1997; Xu et al., 2000], subtractivehybridization [Petersen et al., 2000], and, mostrecently, gene array analysis [Beck et al., 2001]have led to the identi®cation of novel factorsthat regulate bone cell development and func-tion. Among these factors are OF45 [Petersenet al., 2000], connective tissue growth factor(CTGF) [Xu et al., 2000], the TGF-b superfamilymembers, lefty-1 [Seth et al., 2000], best-5[Grewal et al., 2000] and PROM-1 [Ryoo et al.,1997], murine osteoclast inhibitory lectin(mOCIL) [Zhou et al., 2001], and RANK-L[Horowitz et al., 2001; Teitelbaum, 2000].Some of these factors promote osteoblast dif-ferentiation and enhance osteogenesis whileothers are produced by osteoblasts and regu-late osteoclast development and/or function.Identi®cation of such factors has potentialtherapeutic application in cases where anincrease in bone formation or a decrease inresorption would have a bene®cial effect,either locally (as in fracture repair or localizedosteopenia) or systemically (as in generalizedosteopenia).

In this study, we used an animal model ofosteopetrosis, the osteopetrosis (op) mutation inthe rat, to examine differential gene expressionin bone from normal and osteopetrotic animalsresulting from the severe skeletal phenotypeand abnormalities associated with skeletaldevelopment in op mutants [Marks and Popoff,1989]. The mammalian osteopetroses are aheterogeneous group of congenital or experi-mentally-induced mutations characterized by ageneralized increase in bone mass resultingfrom defective osteoclast-mediated bone re-sorption [Popoff and Schneider, 1996]. Thesemutations are pathogenetically heterogeneoussince the point at which osteoclast developmentor activation is intercepted differs for eachmutation [Popoff and Marks, 1995]. In some

osteopetrotic mutations, the primary defect isintrinsic to cells of the osteoclast lineage whilein others the defect is extrinsic to cells of theosteoclast lineage and involves cells within thebone microenvironment that regulate osteo-clast development and/or activity.

In the op mutation, the primary defectappears to be intrinsic to osteoclasts since boneresorption can be restored following the trans-plantation of bone marrow in which the donorcell population contains normal osteoclast pro-genitors [Popoff et al., 1994]. In op mutant rats,abnormalities involving osteoblast gene expres-sion [Shalhoub et al., 1991], mineral home-ostasis [Hermey et al., 1995], immune function[Yamamoto et al., 1994], and the endocrinesystem [Safadi et al., 1999] have also beendescribed. Using the technique of mRNA differ-ential display, we identi®ed a novel cDNA thatis highly upregulated in op compared to normalbone. Subsequent cloning and sequencing of thefull-length cDNA revealed a sequence withhomology to the previously reported humannmb and Pmel 17 cDNAs. In addition to beingthe ®rst report of this cDNA in a species otherthan human, these studies also show that themRNA is expressed by osteoblasts in bone. Thetemporal regulation of mRNA expression dur-ing osteoblast differentiation and analysis of thepredicted amino acid sequence are consistentwith a gene that has a functional role in bone.We propose the name osteoactivin for theprotein encoded by this cDNA.

MATERIALS AND METHODS

Reagents

All chemicals were of molecular biology gradeor higher and were purchased from Sigma (St.Louis, MO) unless otherwise stated. All cellculture media was purchased from Invitrogen(Life Technologies, Gaithersburg, MD).

Animals

An inbred colony of osteopetrotic (op) mutantrats, consisting of heterozygous breeders, ismaintained at Temple University School ofMedicine (Philadelphia, PA). Homozygousmutants (op) are distinguished from normallittermates (�/?) by radiographic analysisbetween 1 and 3 days after birth by the failureof the mutants to develop bone marrow cavities.Because the genotype of phenotypically normalrats cannot be distinguished, except by breeding

Osteoactivin Expression in Osteoblasts 13

experiments, the normal littermates used inthis study were of either heterozygous (�/op) orhomozygous (�/�) normal genotype. All ani-mals were maintained and used according to theprinciples in the NIH Guide for the Care andUse of Laboratory Animals [1985] and guide-lines established by the IACUC of TempleUniversity.

RNA Isolation

Total cellular RNA was isolated from calvariaand long bones (femurs and tibiae) harvestedfrom 2-week-old op mutant rats and theirnormal littermates. Prior to freezing, the endsof the long bones were removed at the growthplate and bone marrow was ¯ushed from theshafts of normal bones with saline at 48C.Flushing of the bone marrow was only possiblein normal rats because there were no marrowcavities in op mutants. Total RNA was preparedfrom pools of a minimum of six samples perphenotype and bone site (calvaria versus longbone). For osteoblast cultures, at the end of theculture period, medium was removed, the celllayer was rinsed, scraped into a 50 ml conicaltube, centrifuged and supernatant was dis-carded. Bone samples and cell pellets were ¯ashfrozen in liquid nitrogen and stored at ÿ 808C.Total RNA from long bones, calvaria or osteo-blasts was prepared as previously described[Thiede et al., 1994] with some modi®cations.Brie¯y, frozen samples were pulverized on dryice (for long bones and calvaria only), and thenhomogenized in 5 M guanidinium isothiocya-nate, 72 mM b-mercaptoethanol, and 0.5%sarkosyl. Homogenates were layered over aCsCl cushion (5.7 M CsCl and 30 mM NaAc),centrifuged at 100,000g overnight, and totalRNA recovered by precipitation of the resultingpellets. Soft tissue RNA was isolated usingTriZol (Invitrogen) according the manufac-turer's protocol. RNA concentration was quan-titated and its integrity was determined asdescribed previously [Xu et al., 2000].

Differential Display of mRNA

RNA samples isolated from long bone orcalvaria were treated with DNase I (RocheMolecular Biochemicals, Indianapolis, IN) toeliminate any potential contamination withgenomic DNA. The basic principle of mRNAdifferential display (DD) was ®rst described byLiang and Pardee [1992]. Brie¯y, 0.5 mg RNAfrom each sample (total of four independent

samples, mutant and normal/calvaria and longbone) was reverse transcribed using each of 12two-base-anchored oligo-dT primers provided inthe Hieroglyph mRNA pro®le kits (BeckmanCoulter Inc., Fullerton, CA) to subdivide themRNA population. First strand cDNAs wereampli®ed by the polymerase chain reaction(PCR) for 30 cycles using one of four upstreamarbitrary primers (also provided in the kit) andthe same anchoring primers used for ®rststrand synthesis as described previously [Xuet al., 2000]. This resulted in 48 possible primercombinations for each kit (total of ®ve kits) andeach PCR ampli®cation was run in duplicatefrom the same ®rst-strand cDNA template. Allampli®ed cDNAs were radiolabled with 33P-dATP ([a-33P]dATP, 2500 Ci/mmol, AmershamPharmacia Biotech, Piscataway, NJ). The radi-olabeled PCR products were electrophoresed on4.5% denaturing polyacrylamide gels and driedusing the Genomyx LR differential displayapparatus (Beckman Coulter). Following auto-radiography, bands were visually assessed andthose representing differentially expressedcDNAs (exclusively expressed or highly over-expressed in one phenotype and con®rmed induplicate PCR ampli®cations) were excisedfrom the gel. Each cDNA of interest was ream-pli®ed by PCR and used to probe a Northern blotto con®rm its differential expression.

Cloning of Rat Osteoactivin cDNA

When the cDNA later determined to beosteoactivin was con®rmed to be differentiallyexpressed in Northern blot analysis, it wascloned into PCR-Script (Stratagene, LaJolla,CA), miniprep DNA was prepared, and plas-mids with the appropriately sized inserts weresequenced using standard dideoxy methodolo-gies. Gaps and ambiguities in the sequence werehandled by direct sequencing of requiredregions using speci®c primers. Approximately600 bp of sequence corresponding to the 30 end ofrat osteoactivin was obtained from the differ-ential display clone. This fragment was used asa probe to screen a lgt11 rat kidney cDNAlibrary (Clontech, Palo Alto, CA) by conven-tional means. A single clone was identi®ed afterthree rounds of screening. The insert in thisclone was ampli®ed by PCR and subcloned intopCR2.1-TOPO (Invitrogen). The completecDNA sequence of osteoactivin was obtainedfollowing transposon insertion (Primer Island;PE Applied Biosystems, Foster City, CA) and

14 Safadi et al.

sequence analysis revealed an open readingframe of 1,716 bp. The sequence of rat osteoac-tivin has been deposited in GenBank underaccession number AF184983.

Northern Blot Analysis

Twenty micrograms of total RNA from opmutant and normal bone, calvaria, normalosteoblast cultures or soft tissues were electro-phoresed on 1% formaldehyde-agarose gels andtransferred to nylon membranes (Scheicher &Schuell, Keene, NH). The cDNA for osteoactivinwas 32P-labeled ([a-32P]dCTP, 6,000 Ci/mmol)using RediprimeTMII kit (Amersham Pharma-cia Biotechnology). Blots were hybridized inChurch's buffer (1 mM EDTA, 1 mM Na2HPO4

pH 7.2, 1% BSA, 7% SDS) overnight at 658C.Blots were then autoradiographed, stripped andreprobed with an 18S rRNA probe used as acontrol to normalize for differences in loadingand transfer.

RT-PCR Analysis

Two micrograms total RNA isolated fromeither long bone, calvaria, osteoblast culturesor soft tissues was reversed transcribed tocDNA at 428C for 50 min in a volume of 20 mlcontaining the following components: 1�®rststrand buffer (6� 250 mM Tris, pH 8.3, 375 mMKCl and 15 mM MgCl2) 0.5 mM dNTPmix; 10 mM dithiothreitol (DTT); 0.5 mgoligo(dT)12±18 and 20 U Superscript II (RNaseH-free reverse transcriptase, Invitrogen). Thereactionwas then terminated at 708C for 15 min,and 1 U RNase H (Invitrogen) was added to thereaction mixture, followed by incubation at378C for 10 min to remove the RNA. Onemicroliter aliquots of the generated cDNA wasampli®ed in 50 ml of PCR reaction mixturecontaining 1 nM primer sets, 10 ml 10� buffer D(for osteoactivin) or buffer N (for 18S) (Invitro-gen) 10 nM dNTP mix, 1 ml DMSO, 1 ml Advant-age polymerase mix (Clonetech). The primersfor osteoactivin were (sense: 50-CCAGAAGAAT-GACCGGAACTCG-30 and anti-sense: 50-CAG-GCTTCCGTGGTAGTGG-30). These primerswere designed from the 50 end of the proteincoding region starting at a position 729 bp fromthe ATG start codon to position 1280. Primersfor 18S were (sense: 50-ACTTTCGATGG-TAGTCGCCGTGC-30 and anti-sense: 50-ATCT-GATCGTCTTCGAACCTCCGA-30). PCR wasperformed using Perkin-Elmer GeneAmp PCRSystem 9600. PCR parameters for osteoactivin

were: denaturation step at 948C for 3 min,followed by 25 cycles of 948C for 20 s, 628C for20 s and 688C for 20 s; with ®nal extension stepat 688C for 7 min. The expected osteoactivinPCR product was 552 bp. PCR parameters for18S were: denaturation step at 948C for 3 min,followed by 30 cycles of 948C for 30 s, 658C for30 s and 688C for 30 s; with ®nal extension stepat 728C for 7 min. The expected 18S PCRproduct was 700 bp. The PCR products wereanalyzed by 1% agarose gel electrophoresisstained with ethidium bromide. A 100 bp lad-der was used as a molecular weight marker(Invitrogen).

Synthesis of Osteoactivin RNA Riboprobe

Approximately 600 bp fragment of the 30 endof rat osteoactivin was cloned into the PCRScript plasmid vector ¯anked by T7 and T3promoters. Osteoactivin RNA sense and anti-sense riboprobes were synthesized in the pre-sence of digoxygenin (DIG)-modi®ed UTP usingan in vitro transcription kit (Roche) according tothe manufacturer's protocol.

In Situ Hybridization of Osteoactivin in Bone

Tibae or femurs from normal rats were har-vested, ®xed in 4% paraformaldehyde, decalci-®ed, embedded, and sectioned. Sections werebaked at 608C for 1 h, cleared in xylene,rehydrated in 100, 75, 50, 25% ethanol in PBT(PBS, 0.1% Triton X-100), and rinsed in PBTthree times for 10 min each. Sections weretreated with Proteinase K (10 mg/ml) in PBT andincubated at 378C for 15 min, then rinsed twicein PBT 5 min each. Sections were then re®xed in0.2% glutaraldehyde and 4% parafomaldehydein PBT at room temperature for 10 min and thenrinsed brie¯y in PBT. Sections were placed in ahumidi®ed chamber and prehybridized (50%formamide; 5�SSC; 2% blocking powder sup-plied in the kit (Roche Laboratories); 1 mg/mlyeast RNA; 5 mM EDTA and 50 mg/ml heparin)for 4 h at 608C. Sections were then hybridized(prehybridization solution and osteoactivin-riboprobe at 5 ng/ml) at 608C overnight. Thefollowing day, sections were subjected to differ-ent washes; 100% solution-1 (50% formamide,5�SSC, 0.1% Triton and 0.5% CHAPS); 75%solution-1 in 2�SSC; 50% solution-1 in 2�SSCand 25% solution-1 in 2�SSC. Sections wererewashed with 2�SSC and 0.1% CHAPS twicefor 30 min each at 658C; 0.2�SSC and 0.1%CHAPS twice for 30 min each at 658C and twice


for 10 min each with TBT (0.05 M Tris (pH 7.5),0.1% Triton, 150 mM NaCl) at room tempera-ture. Sections were then blocked with 10%sheep serum, 2% BSA in TBT for 1 h at roomtemperature and incubated overnight at roomtemperature with anti-DIG antibody (1:500dilution in 10% sheep serum, 2% BSA in TBT).Following antibody treatment, sections werewashed as follows: three times for 30 min each in0.1% BSA in TBT at room temperature, twice30 min each in TBT, three times 10 min each inGenius buffer (Genius buffer; 100 mM NaCl,100 mM Tris HCl, 50 mM MgCl2 pH 9.5) and0.1% Tween 20. Signal was developed using thecolor developing solution (45ml NBT and 35ml X-phosphate in 10 ml of Genius buffer); sectionswere covered and placed in the dark at roomtemperature for 15±30 min or until the colorchange was observed. Color signal was ®xedwith 4% paraformaldehyde in PBT overnight at48C and sections were mounted in PBS-glyceroland examined using a Nikon-Eclipse E800 lightmicroscope.

Primary Osteoblast Cultures

Normal rat pups (1±4-day-old) were eutha-nized and a mid-line skin incision was made toexpose the calvaria. The periosteum wasremoved and frontal and parietal bones weredissected. These pieces of calvarial bone weretransferred to siliconized Ehrlenmeyer ¯askscontaining 10 ml of trypsin/collagenase P solu-tion (0.25% trypsin and 0.2% collagenase P inPBS) and placed in a 378C shaker for 5 min.Supernatant was discarded following the®rst digestion. Supernatants from the second(15 min at 378C) and third (25 min at 378C)digestions were pooled into MEM (Hank's) con-taining 10% FBS and penicillin (5,000 U/ml)and streptomycin (5 mg/ml). Cells were washedtwice in MEM (Earl's) supplemented with 10%FBS penicillin/streptomycin, ®ltered through a200 mm metal screen ®lter, followed by ®ltrationthrough an ether/ethanol treated ®lter. Cellswere counted and viability determined bytrypan blue exclusion. Cells were plated in100-mm dishes at a density of 5� 105 cells/dishin plating medium (MEM (Earl's) with 10% FBSand penicillin/streptomycin. On day 3 of theculture, cells were fed with plating mediumsupplemented with ascorbic acid (25 mg/ml). Onday 7 and every 3 days thereafter, cells were fedwith plating medium containing ascorbic acid(50 mg/ml) and b-glycerolphosphate (10 mM) to

induce osteoblast differentiation and promotemineralization. Osteoblast differentiation wasassessed by the formation of mineralized bonenodules assessed by von Kossa staining [Popoffet al., 2000].

RESULTS

Identi®cation of Osteoactivin

Total cellular RNA obtained from normal orop mutant calvaria or long bone was used asa template for differential display analysis.Many differences in gene expression were ob-served between mutant and normal bone with atypical autoradiogram shown in Figure 1. For

Fig. 1. Identi®cation of osteoactivin using differential displayanalysis. Total cellular RNA obtained from mutant (M) ornormal (N) calvaria (C) or long bone (L) was reversed transcribedusing a two-base anchored oligo-dT primer (Hieroglyph mRNApro®le kit). First strand cDNAs were used to generateradiolabeled PCR products using the same anchoring primerand an arbitrary upstream primer. Each PCR reaction wasperformed in duplicate and run on a denaturing acrylamide gel.The arrow indicates the band corresponding to osteoactivin andshows differential expression between mutant and normal bonesamples.

16 Safadi et al.

osteoactivin, the upper of the two intense bandsin mutant calvaria and long bone that is faintlyvisible in normal bone represents the band fromwhich the original cDNA fragment was isolated(Fig. 1). This band was cut from the gel, DNAwas extracted and used as a template for PCRreampli®cation. The resulting DNA was radi-olabeled and used as a probe for Northerncon®rmation of differential expression ofosteoactivin using RNA isolated from normaland mutant long bone and calvaria (Fig. 2).Densitometric analysis revealed that osteoacti-vin was highly (3- to 4-fold) overexpressed in opmutant compared with normal long bone(Fig. 2A) and calvaria (Fig. 2B).

Cloning and Characterization of Osteoactivin

The cDNA obtained from the differentialdisplay analysis was used as a probe to screena lgt11 rat kidney cDNA library in an effort toobtain a full-length clone of osteoactivin. Appro-ximately 106 clones were screened and onepositive was obtained. This clone was comple-tely sequenced and the sequence has beendeposited in GenBank under accession numberAF184983. The cloned rat osteoactivin has anopen reading frame of 1,716 bp with 114 bp of 50

untranslated sequence and 476 bp of 30 untrans-lated sequence prior to the poly-A tail (data notshown). The open reading frame encodes a

Fig. 2. Northern blot analysis of osteoactivin in bone.Representative Northern blots of osteoactivin expression inlong bone (A) and calvaria (B) from op mutant and normal rats.Twenty micrograms of total RNA isolated from mutant (M) ornormal (N) long bone (A) and calvaria (B) was electrophoresedin a 1% agarose-formaldehyde gel, blotted, and probed for

osteoactivin. The blot was then stripped and reprobed for 18SrRNA as a control for gel loading. Northern analysis wasrepeated three times using independent RNA samples. Densito-metric analysis showed the relative difference in osteoactivinexpression in mutant when compared to normal bone.


Fig. 3. Nucleotide sequence of rat osteoactivin open reading frame. The nucleotide sequence of ratosteoactivin is shown in comparison to mouse and human nmb. The open reading frame potentiallyencodes a protein of 572 amino acids with a predicted molecular weight of 63.8 kD.

protein of 572 amino acids with a predictedmolecular weight of 63.8 kD.

Comparison of the sequence of rat osteoacti-vin with public databases revealed that it ishighly homologous to the human (GenBankaccession X7653) and mouse (GenBank acces-sion AF322054 and AJ251685) nmb DNA andprotein sequences. A comparison of the nucleo-tide sequences of the open reading frames(Fig. 3) and predicted protein sequences(Fig. 4) of rat osteoactivin with human nmbreveals a 77% identity on the DNA level(excluding the coding sequences for the 14amino acid insertion in the predicted protein)and a 65% identity on the protein level. As canbe seen in Figure 4, the predicted proteinsequences of rat osteoactivin and mouse nmbhave a proline-serine rich insertion of 14 or 16amino acids, respectively, beginning at ratresidue 329 that is not present in the humannmb sequence.

Since no data were previously available as tothe function of osteoactivin or nmb, bioinfor-matic analysis of the osteoactivin proteinsequence was performed. We used the CBSSignal P V1.1 World Wide Web PredictionServer at the Center for Biological Sequence(http://www.cbs.dtu.dk) to analyze the se-quence of osteoactivin for the presence of asignal peptide. Signal P calculates the maximalC-(raw cleavage site score), S-(signal peptidescore) and Y-(combined cleavage site score)scores, and the mean S-score, between the N-terminal and the predicted cleavage site. SignalP analysis of osteoactivin revealed a mean Sscore of 0.907 indicating that it has a signalpeptide (Fig. 5, Table I). This signal peptideconsists of the ®rst 22 amino acids with apredicted cleavage site between residues 22and 23 (Fig. 4, arrow and Fig. 5). These datasuggest that osteoactivin is a secreted protein.Hydropathy analysis revealed the presence of

Fig. 3. (Continued)

Fig. 4. Comparison of the predicted amino acid sequences ofrat osteoactivin and mouse and human nmb. The predictedprotein sequences of rat osteoactivin and mouse and humannmb are compared. Amino acid residues that are identicalamong the three species are boxed. The predicted ratosteoactivin and mouse nmb proteins have a proline-serinerich insertion beginning at residue 329 of the rat protein that is

not present in the human nmb sequence. At the protein level,the sequences of rat osteoactivin and human nmb are 65%identical. The predicted rat osteoactivin protein contains aleader sequence with its cleavage site after residue 22 (arrow) aswell as 13 potential sites for N-linked glycosylation (*). [Color®gure can be viewed in the online issue, which is available atwww.interscience.wiley.com.]


several potential transmembrane spanningregions throughout the molecule as well as 13sites of potential N-linked glycosylation consis-tent with the data reported for human nmb[Weterman et al., 1995] (data not shown).Osteoactivin also contains four sites for proteinkinase C phosphorylation, 14 cysteine residues,and an RGD site located at position 556.

Expression of Osteoactivin inOther Tissues/Organs

To determine whether osteoactivin mRNA isexpressed in other tissues/organs and to com-pare its relative levels of expression withbone, RT-PCR and Northern blot analyseswere performed using total RNA isolatedfrom various soft tissues and bone as well asfrom primary osteoblast cultures. The results

of RT-PCR analysis demonstrated that osteoac-tivin mRNA is expressed at highest levels inlong bones, calvaria and cultured osteoblasts,with much lower levels in thymus, brain andskeletal muscle (Fig. 6). Osteoactivin mRNAwas not detectable in the other tissues exam-ined including liver, heart, spleen, kidney, skin,proximal small intestine, and bone marrow(Fig. 6). These results were con®rmed by North-ern blot analysis although the sensitivity waslower making it dif®cult to detect distinct bandsin the soft tissues with low levels of expression(data not shown).

In Situ Hybridization of Osteoactivin in Bone

Given the relatively high levels of osteoacti-vin expression in cultured rat osteoblasts, insitu hybridization was employed to localize the

TABLE I. Characterization of Predicted Osteoactivin Amino AcidSequence

Amino Acid Position Description

1±22 Signal peptide23 Cleavage site93, 134, 147, 200, 249, 275, 296, 300, 306, 312, 461, 469, 532 N-glycosylation site191, 280, 417, 455, 526 Protein kinase C phosphorylation site556 RGD site, cell attachment sequence5, 105, 120, 190, 323, 371, 418, 427, 435, 443, 452, 467, 508, 533 Cysteine residues166 Amidation site

Fig. 5. Analysis of osteoactivin predicted amino acid sequence. Signal peptide prediction plot using theneural network model indicates a potential leader sequence at the amino terminus of osteoactivin. Mean Sscore�0.907 which represents a signal peptide consisting of the ®rst 22 amino acids and a cleavage sitebetween position 22 and 23. [Color ®gure can be viewed in the online issue, which is available atwww.interscience.wiley.com.]

20 Safadi et al.

expression of osteoactivin mRNA in cells withinthe bone microenvironment. Sections of theproximal tibial metaphysis from 2-week-oldnormal rats were prepared and hybridized tosense and anti-sense osteoactivin riboprobes.Using the anti-sense riboprobe, the detectionsignal for osteoactivin mRNA was predomi-nantly located in rows of plump, cuboidalosteoblasts lining active bone forming surfaces(Fig. 7, arrows). Osteoactivin mRNA was alsodetected in some osteocytes, particularly thosemost recently encased in bone and close to thebone surface. In addition, a few cells scatteredamong the intertrabecular spaces of the meta-physis also exhibited signal for osteoactivinmRNA. These data demonstrate that osteoacti-vin is primarily expressed in osteoblasts.

Temporal Expression of Osteoactivin inPrimary Osteoblast Cultures

Primary cultures of osteoblasts derived fromcalvarial digests of newborn normal rat calvariawere established and used to examine theexpression of osteoactivin mRNA during osteo-blast differentiation. In these cultures, thereare three distinct stages of osteoblast develop-ment including an initial period of cell prolif-eration (from initiation to day 7), followed by a

period of matrix formation and maturation (day7±14), and ending with a ®nal phase of matrixmineralization (day 14±21) [Owen et al., 1990].For each experiment, cultures were establishedat the same time and terminated at various timepoints for RNA isolation. Northern blot analysisclearly shows a temporal pattern of expressionfor osteoactivin, which, although detectable atlow levels in proliferating osteoblasts, progres-sively increases as the cells differentiate beinghighest during the period of matrix mineraliza-tion (Fig. 8). The pattern of osteoactivin expres-sion in these cultures is very similar to that ofalkaline phosphatase, a gene highly expressedduring the periods of maturation and matrixmineralization. Peak expression of osteoactivinalso correlates with the peak expression, atday 21, of osteocalcin, a marker for osteoblastterminal differentiation.

DISCUSSION

In the present study, we used mRNA differ-ential display (DD) to compare the expression ofgenes in bone from osteopetrotic (op) andnormal rats. This technique was initiallydescribed by Liang and Pardee [1992]; Lianget al. [1993] as a way to bring the remarkable

Fig. 6. PCR analysis of osteoactivin mRNA in multiple tissues.Total RNA was isolated from soft tissues, bone and osteoblastcultures, as described in the methods. Two micrograms of totalRNA was reversed transcribed to generate ®rst strand cDNA.The generated cDNAs from different tissues and osteoblastcultures were subjected to PCR using speci®c primers forosteoactivin and 18S as internal control. PCR products wereseparated on 1% agarose gel and stained with ethedium

bromide. PCR product for osteoactivin was 552 bp and for18S was 700 bp. cDNA for osteoactivin was used as an internalpositive control. PCR was repeated three times using indepen-dent cDNA samples with similar results. Although low levels ofosteoactivin expression were detected in thymus, brain, andskeletal muscle, maximum levels were detected in long bones,calvaria and primary osteoblast cultures.


power and sensitivity of the polymerase chainreaction (PCR) to bear on questions of differ-ences in gene expression. The advantages of DDinclude its lack of bias and high sensitivity. Theabsence of bias means that no prior assumptionsneed to be made as to the biological mechanismsand the method is capable of detecting novelgene products. The PCR-based sensitivity of DDalso permits the detection of minor changes ingene expression. The main disadvantage of themethod is that it is prone to producing falsepositives and this requires careful attention toexperimental conditions and the use of appro-priate controls. Most importantly, any differ-ences observed by DD must be con®rmed byNorthern blot analysis.

Mason et al. [1997] utilized this approach tostudy bone cell gene expression and found thatthe glutamate signaling pathway is involved inthe response of bone to mechanical loading. Weused this technique to clone and characterizeconnective tissue growth factor (CTGF) that

was highly overexpressed in op mutant com-pared to normal bone [Xu et al., 2000]. CTGF is asecreted, extracellular matrix-associated pro-tein that was previously shown to be producedby various cell types including ®broblasts,endothelial cells, and chondrocytes [Lau andLam, 1999]. CTGF has been shown to regulatediverse cellular functions including cell prolif-eration, attachment, migration, differentiation,survival and matrix production [Moussad andBrigstock, 2000]. Our studies demonstratedthat CTGF is also produced by osteoblasts andappears to play an important role in regulatingosteoblast differentiation and function (unpub-lished observations).

In this study, we identi®ed a novel cDNA,named osteoactivin that was overexpressed inop mutant compared to normal bone. Con®rma-tion by Northern blot analysis demonstratedthat osteoactivin expression was increased 3- to4-fold in op versus normal long bones andcalvaria. Osteoactivin cDNA was cloned and

Fig. 7. In situ hybridization (ISH) of osteoactivin in bone.Sections of proximal tibial metaphyses of 2-week-old normalrats were processed for ISH as described in the methods. A:Sections incubated with the osteoactivin sense riboprobe wereused as a control. Note the absence of any reaction product;image was captured using Nomarski DIC microscopy to allow

visualization of some detail of the unstained tissue. B±D: Low,medium and high power photomicrographs of sections hybrid-ized with the osteoactivin anti-sense riboprobe. Note thereaction product indicative of osteoactivin mRNA expressionspeci®cally associated with osteoblasts (arrows) lining the bonytrabeculae (b), m: marrow spaces.

22 Safadi et al.

sequenced; it was found to have 77% identity onthe DNA level and 65% identity of the proteinlevel to human nmb. Human nmb was isolatedby subtractive hybridization of MV1 and MV3melanoma cell lines that exhibit low and highmetastatic potential in nude mice, respectively;nmb was found to be overexpressed in MV1when compared to MV3 cells [Weterman et al.,

1995], although the function of nmb isunknown. Osteoactivin also shares sequencehomology with Pmel 17, a melanocyte-speci®cgene that may function as a catalyst in melaninbiosynthesis [Kwon et al., 1991]. Osteoactivinhas an open reading frame (ORF) of 1,716 bpwith a short 50 untranslated region and encod-ing a protein of 572 amino acids with a predicted

Fig. 8. Temporal expression of osteoactivin mRNA in primaryrat osteoblast cultures. A: Twenty micrograms of total RNAisolated from osteoblasts cultured for 3, 7, 14, 17, and 21 days,was electrophoresed, blotted, and probed for osteoactivin. Blotwas stripped and reprobed for alkaline phosphatase (AP),osteocalcin (OC) and, ®nally, 18S rRNA as a loading control.

B: The results of three independent experiments were quanti-tated by scanning densitometry and plotted relative to themaximal expression of each transcript corrected using the 18SrRNA control. [Color ®gure can be viewed in the online issue,which is available at www.interscience.wiley.com.]


molecular weight of 63.8 kD. Its sequenceresembles that of Pmel 17 which has an ORFthat encodes a protein of 645 amino acids and amolecular weight of 68.6 kD. The ®rst 22 aminoacids of osteoactivin and the ®rst 23 amino acidsof Pmel 17 represent potential signal peptides,characteristic of secreted proteins. These simi-larities suggest that osteoactivin may belong tothe same gene family as Pmel 17.

Analysis of the predicted amino acid sequencefor osteoactivin revealed 13 potential N-linkedglycosylation sites and it is known that glyco-sylation of a protein can have a signi®cantimpact on its biological activity. Other featuresthat are also likely to regulate its biologicalactivity include the four sites for protein kinaseC phosphorylation and 14 cysteine residues. AnRGD site, located at position 556, represents apotential site for cell attachment and may beimportant for osteoactivin binding and itsmechanism of action. Since there are no pre-vious studies on osteoactivin and the function ofhuman nmb is unknown, there is a great deal ofwork that needs to be done to determine thefunctions of osteoactivin and its mechanism(s)of action.

Tissue distribution of osteoactivin expressionmeasured by RT-PCR analysis showed thatosteoactivin is expressed at highest levels inbone and primary osteoblast cultures withlower levels of expression in brain, thymus,and skeletal muscle. Its high level of expressionin developing bone suggests that it may play arole as a local regulator of bone cell metabolism.In situ hybridization was performed to localizeosteoactivin mRNA expression in cells of thebone microenvironment. These experimentsrevealed that osteoactivin mRNA is primarilylocalized in plump, cuboidal osteoblasts liningbone surfaces.

These data led us to examine osteoactivinexpression in primary osteoblast culturesderived from newborn rat calvaria. This culturesystem is routinely employed to examine theeffects of local and systemic factors (e.g.,hormones, cytokines, growth factors) on osteo-blast development and function. Primary osteo-blast cultures undergo three distinct stagesbeginning with cell proliferation (day 0±7)associated with the expression of histone H4,TGF-b and ®bronectin, followed by a period ofmatrix maturation (day 7±14) associated withthe expression of alkaline phosphatase andmatrix Gla protein, and ending with a stage of

matrix mineralization (day 14±21) associatedwith the expression of genes characteristic ofthe terminally differentiated osteoblast includ-ing osteocalcin and osteopontin [Stein et al.,1996]. In this study, we examined the temporalpattern of osteoactivin expression and showedthat it increased progressively as the cellsdifferentiated with maximum expression dur-ing the ®nal stage of matrix mineralization.Osteoactivin expression was also correlatedwith the expression of the osteoblast-relatedgenes alkaline phosphatase and osteocalcin inthese cultures. The data suggest that osteoacti-vin may play an important role in osteoblastdifferentiation and matrix mineralization.Clearly, additional studies are necessary todetermine whether it is secreted in cultureand to examine its functional role. Experimentsto block its expression using an anti-senseoligonucleotide and/or a neutralizing antibodyare planned and expected to shed some light onthe functional properties of this novel protein.

The overexpression of osteoactivin in opmutant compared to normal bone is an inter-esting ®nding that also requires further inves-tigation. The rat mutation, osteopetrosis (op)mutation is among a group of spontaneous,autosomal recessive mutations in rodents thatexhibit the characteristic feature of generalizedskeletal sclerosis [Seifert et al., 1993]. Char-acterization of op bone has revealed thatosteoclasts are present but exhibit atypicalmorphological features, suggesting that thedefect involves cell function instead of develop-ment [Marks and Popoff, 1989]. Furthermore,the ability of normal bone marrow transplanta-tion to restore osteoclastic bone resorption andcure the skeletal manifestations provides directevidence that the defect is inherent to cells of theosteoclast lineage [Popoff et al., 1994]. Despitethe fact that the op rat mutation is caused by aprimary defect in osteoclast-mediated boneresorption, abnormalities involving osteoblastgene expression have been reported. Theexpression of some osteoblast-related genesare markedly increased in op versus normalbone [Shalhoub et al., 1991]. Therefore, it isperhaps not surprising that osteoactivin,another osteoblast-related gene, is also upregu-lated in op bone. These abnormalities are likelyto be secondary to the absence of normalosteoclast function causing an uncoupling ofbone resorption from bone formation. Theresulting dysregulation of osteoblast gene

24 Safadi et al.

expression and function would be consistentwith our ®ndings, although the factor(s) thatdrive this overexpression are unknown. Futurestudies will focus on systemic and local factorsknown to be upregulated in the op mutation,such as 1,25(OH)2D3, PTH, TGF-b, c-fos andothers, and are expected to shed light ontranscriptional regulation of osteoactivin aswell as provide an explanation for its upregula-tion in op bone.

ACKNOWLEDGMENTS

The authors would like to thank members ofthe DNA sequencing facility at P®zer GlobalResearch and Development for performing thesequencing for this project. The authors wouldalso like to thank Dr. Victoriya Zakhaleva forher technical assistance and Cherie Glanzmannfor her assistance in preparing the manuscript.This study was supported by a research grantfrom P®zer, Inc. (to S.N.P. and F.F.S.) andby funds awarded to the Bone Research Enter-prise Program by Temple University School ofMedicine.

REFERENCES

Beck GR Jr, Zerler B, Moran E. 2001. Gene array analysisof osteoblast differentiation. Cell Growth Differ 12:61±83.

Grewal TS, Genever PG, Brabbs AC, Birch M, Skerry TM.2000. Best5: a novel interferon-inducible gene expressedduring bone formation. FASEB J 14(3):523±531.

Guide for the care and use of laboratory animals. 1985. U.S.Dept. of Health and Human Services, National Institutesof Health Publ. No. 86-23.

Hermey DC, Ireland RA, Zerwekh JE, Popoff SN. 1995.Regulation of mineral homeostasis in osteopetrotic (op)rats. Am J Physiol 268:E312±E317.

Horowitz MC, Xi Y, Wilson K, Kacena MA. 2001. Control ofosteoclastogenesis and bone resorption by members ofthe TNF family of receptors and ligands. CytokineGrowth Factor Rev 12(1):9±18.

Kwon BS, Chintamaneni C, Kozak CA, Copeland NG,Gilbert DJ, Jenkins N, Barton D, Francke U, KobayashiY, Kim KK. 1991. A melanocyte-speci®c gene, Pmel 17,maps near the silver coat color locus on mouse chromo-some 10 and is in a syntenic region on human chromo-some 12. Prac Natl Acad Sci 88:9228±9232.

Lau LF, Lam SCT. 1999. The CCN family of angiogenicregulators: the integrin connection. Exper Cell Res 248:44±57.

Liang P, Pardee AB. 1992. Differential display of eukar-yotic messenger RNA by means of the polymerase chainreaction. Science 257:967±971.

Liang P, Averboukh L, Pardee AB. 1993. Distribution andcloning of eukaryotic mRNAs by means of differentialdisplay: re®nements and optimization. Nucleic Acids Res21:3269±3275.

Marks SC Jr, Hermey DC. 1996. The structure anddevelopment of bone. In: Bilezikian JP, Raisz LG, RodanGA, editors. Principles of bone biology. San Diego:Academic Press. p 69±85.

Marks SC Jr, Popoff SN. 1989. Osteoclast biology in theosteopetrotic (op) rat. Am J Anat 186:325±334.

Mason DJ, Suva LJ, Genever PG, Patton AJ, Steuckle S,Hillam RA, Skerry TM. 1997. Mechanically regulatedexpression of a neural glutamate transporter in bone: arole for excitatory amino acids as osteotropic agents?Bone 20:199±205.

Moussad EEA, Brigstock DR. 2000. Connective tissuegrowth factor: what's in a name? Molec Genet Metab71:276±292.

Owen TA, Aronow M, Shalhoub V, Barone LM, Wilming L,Tassinari MS, Kennedy MB, Pockwinse S, Lian JB, SteinGS. 1990. Progressive development of the rat osteoblastphenotype in vitro: reciprocal relationships in expressionof genes associated with osteoblast proliferation anddifferentiation during formation of the bone extracellularmatrix. J Cell Physiol 143:420±430.

Petersen DN, Tkalcevic GT, Mansolf AL, Rivera-GonzalezR, Brown TA. 2000. Identi®cation of osteoblast/osteocyte45 (OF45), a bone-speci®c cDNA encoding an RGD-containing protein that is highly expressed in osteoblastsand osteocytes. J Biol Chem 275(46):36172±36180.

Popoff SN, Marks SC Jr. 1995. The heterogeneity of theosteopetroses re¯ects the diversity of cellular in¯uencesduring skeletal development. Bone 17:437±445.

Popoff SN, Schneider GB. 1996. Animal models of osteope-trosis: the impact of recent molecular developments onnovel strategies for therapeutic intervention. Mol MedToday 2:349±358.

Popoff SN, Osier LK, Zerwekh JE, Marks SC Jr. 1994.Interdependence of skeletal sclerosis and elevated circu-lating levels of 1.25-dihydroxyvitamin D in osteopetrotic(op and tl) rats. Bone 15:515±522.

Popoff SN, Xu J, Smock S, Mendis M, Zakhaleva V, OwenTA, Safadi FF. 2000. Regulation of connective tissuegrowth factor in primary rat osteoblast cultures by1,25(OH)2D3. In Vitamin D Endocrine System: Struc-tural, Biological, Genetic and Clinical Aspects. Haw-thorne, NY: Walter de Gruyer. p 192±132.

Ryoo HM, van Wijnen AJ, Stein JL, Lian JB, Stein GS.1997. Detection of a proliferation speci®c gene duringdevelopment of the osteoblast phenotype by mRNAdifferential display. J Cell Biochem 64(1):106±116.

Safadi FF, Seifert MF, Popoff SN. 1999. Skeletal resistanceto 1,25-dihydroxyvitamin D3 in osteopetrotic rats. Endo-crine 11:309±319.

Seifert MF, Popoff SN, Jackson ME, MacKay CA, CielinskiM, Marks SC Jr. 1993. Experimental studies ofosteopetrosis in laboratory animals. Clin Orthopaed294:23±33.

Seth A, Lee BK, Qi S, Vary CP. 2000. Coordinate expressionof novel genes during osteoblast differentiation. J BoneMiner Res 15(9):1683±1696.

Shalhoub V, Jackson ME, Lian JB, Stein GS, Marks SC Jr.1991. Gene expression during skeletal development inthree osteopetrotic rat mutations. J Biol Chem 266:9847±9856.

Stein GS, Lian JB, Stein JL, van Wijnen AJ, Frennkel B,Montecino M. 1996. Mechanisms Regulating osteoblastproliferation and differentiation. In: Bilezikian JP,


Raisz LG, Rodan GA, editors. Principles of bone biology.San Diego: Academic Press. p. 69±85.

Teitelbaum SL. 2000. Bone resorption by osteoclasts.Science 289:1504±1508.

Thiede MA, Smock SL, Petersen DN, Grasser WA,Thompson DD, Nishimoto SK. 1994. Presence of mes-senger ribonucleic acid encoding osteocalcin, a marker ofbone turnover, in bone marrow megakaryocytes andperipheral blood platelets. Endocrinology 135:929±937.

Trif®tt JT. 1996. The stem cell of the osteoblast. In: Gold L,Cleveland D, editors. Oral and maxillofacial surgeryclinics of North America. Philadelphia: WB Saunders.p 39±50.

Weterman MA, Ajubi N, van Dinter IM, Degen WG, vanMuijen GN, Ruitter DJ, Bloemers HP. 1995. nmb, a novel

gene, is expressed in low-metastatic human melanomacell lines and xenografts. Int J Cancer 60(1):73±81.

Xu J, Smock SL, Safadi FF, Rosenzweig AB, Odgren PR,Marks SC Jr, Owen TA, Popoff SN. 2000. Cloning thefull-length cDNA for rat connective tissue growth factor:implications for skeletal development. J Cell Biochem77:103±115.

Yamamoto N, Lindsay DD, Naraparaju VR, Ireland RA,Popoff SN. 1994. A defect in the in¯ammation-primedmacrophage-activation cascade in osteopetrotic rats. JImmunol 152(10):5100±5107.

Zhou H, Kartsogiannis V, Hu YS, Elliott J, Quinn JM,McKinstry WJ, Gillespie MT, Ng KW. 2001. A novelosteoblast-derived c-type lectin that inhibits osteoclastformation. J Biol Chem 276(18):14916±14923.

26 Safadi et al.

Documents

Cloning and characterization of osteoactivin, a novel cDNA expressed in osteoblasts