Isolation of a Human Homolog of Osteoclast Inhibitory Lectin ThatInhibits the Formation and Function of Osteoclasts

YUN SHAN HU,1 HONG ZHOU,1 DAMIAN MYERS,2 JULIAN MW QUINN,1 GERALD J ATKINS,3

CHI LY,1 CHRISTINE GANGE,1 VICKY KARTSOGIANNIS,1 JAN ELLIOTT,1 PANAGIOTA KOSTAKIS,4

ANDREW CW ZANNETTINO,4 BRETT CROMER,1 WILLIAM J MCKINSTRY,1 DAVID M FINDLAY,3

MATTHEW T GILLESPIE1 and KONG WAH NG5

ABSTRACT

Osteoclast inhibitory lectin (OCIL) is a newly recognized inhibitor of osteoclast formation. We identified a humanhomolog of OCIL and its gene, determined its regulation in human osteoblast cell lines, and established that it caninhibit murine and human osteoclast formation and resorption. OCIL shows promise as a new antiresorptive.

Introduction: Murine and rat osteoclast inhibitory lectins (mOCIL and rOCIL, respectively) are type II membraneC-type lectins expressed by osteoblasts and other extraskeletal tissues, with the extracellular domain of each,expressed as a recombinant protein, able to inhibit in vitro osteoclast formation.Materials and Methods: We isolated the human homolog of OCIL (hOCIL) from a human fetal cDNA library thatpredicts a 191 amino acid type II membrane protein, with the 112 amino acid C-type lectin region in the extracellulardomain having 53% identity with the C-type lectin sequences of rOCIL and mOCIL. The extracellular domain ofhOCIL was expressed as a soluble recombinant protein in E. coli, and its biological effects were determined.Results and Conclusions: The hOCIL gene is 25 kb in length, comprised of five exons, and is a member of asuperfamily of natural killer (NK) cell receptors encoded by the NK gene complex located on chromosome 12.Human OCIL mRNA expression is upregulated by interleukin (IL)-1� and prostaglandin E2 (PGE2) in atime-dependent manner in human osteogenic sarcoma MG63 cells, but not by dexamethasone or 1,25 dihy-droxyvitamin D3. Soluble recombinant hOCIL had biological effects comparable with recombinant mOCIL onhuman and murine osteoclastogenesis. In addition to its capacity to limit osteoclast formation, OCIL was alsoable to inhibit bone resorption by mature, giant-cell tumor-derived osteoclasts. Thus, a human homolog of OCILexists that is highly conserved with mOCIL in its primary amino acid sequence (C-lectin domain), genomicstructure, and activity to inhibit osteoclastogenesis.J Bone Miner Res 2004;19:89–99. Published online on December 15, 2003; doi: 10.1359/JBMR.0301215

Key words: osteoclasts, C-type lectin, natural killer cell receptors, natural killer gene complex, boneresorption

INTRODUCTION

OSTEOCLASTS ARE MULTINUCLEATED cells responsible forresorbing bone and are derived from mononuclear pre-

cursors of the monocyte-macrophage lineage. The differen-

tiation of monocyte-macrophage lineage cells toward that ofosteoclasts is considered to be the result of actions ofosteoblast-derived membrane-bound RANKL andmacrophage-colony-stimulating factor (M-CSF).(1,2) Thesefactors bind to their respective receptors (RANK and c-fms,respectively) that are expressed by monocyte/macrophagelineage cells, thereby stimulating some of these to formThe authors have no conflict of interest.

1Bone, Joint, and Cancer Unit, St. Vincent’s Institute of Medical Research, Fitzroy, Victoria, Australia.2Department of Physiology, The University of Melbourne, Parkville, Victoria, Australia.3Department of Orthopaedic Surgery and Trauma, University of Adelaide, Adelaide, South Australia, Australia.4Division of Haematology, Institute of Medical and Veterinary Science, Adelaide, South Australia, Australia.5Department of Endocrinology & Diabetes, St. Vincent’s Hospital, Fitzroy, Victoria, Australia.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 19, Number 1, 2004Published online on December 15, 2003; doi: 10.1359/JBMR.0301215© 2004 American Society for Bone and Mineral Research

89

osteoclasts. This process can be regulated by other factorsthat are expressed by osteoblasts. Principally, the secreteddecoy receptor of RANKL, osteoprotegerin (OPG), caninhibit osteoclast formation both in vitro and in vivo.(3) Therequirement for RANKL, RANK, and OPG in control ofosteoclast formation and activity have been exemplifiedthrough studies of genetically altered mice that are eithernull or overexpress a transgene. RANK�/� or RANKL�/�

mice display severe osteopetrosis caused by a complete lackof osteoclasts, and OPG�/� mice are osteoporotic.(4,5) Inaddition, an absence of lymph node development and re-duced numbers of mature T-cells and splenic B-cells werenoted in RANKL�/� mice, implicating RANKL as a regu-lator of lymph node organogenesis and of lymphopoiesis.(4)

Several other factors that are expressed by osteoblasts orlymphocytes have an impact on osteoclastogenesis to en-hance (interleukin [IL]-1, IL-6, and IL-8) or limit (IL-4,IL-10, IL-12, IL-13, and IL-18) osteoclast formation.(6) Wehave described a new family of molecules, osteoclast inhib-itory lectin (OCIL), that inhibit osteoclast formation.(7,8)

OCIL and its related proteins, OCILrP1 and OCILrP2, aretype II membrane-bound C-lectins expressed by osteoblastsand other extraskeletal tissues(7,8) and are encoded by sep-arate genes within the natural killer (NK) gene complex.(8)

Importantly, the extracellular domains of mOCIL, mO-CILrP1, or mOCILrP2 expressed as recombinant proteins inE. coli dose-dependently inhibited multinucleate osteoclastformation by adherent murine spleen cells treated withRANKL and M-CSF,(8) with their predominant actions toinhibit the proliferation of mononuclear osteoclast precur-sors. The tissue distribution of mOCIL/mOCILrP1/mOCILrP2 was similar to that of RANKL, strongly sug-gesting a functional interaction between these molecules inthe skeleton and in extraskeletal tissues. However, onlymOCIL was robustly regulated by calciotropic agents. mO-CILrP2 was constitutively expressed in osteoblasts, whereasmOCILrP1 was not expressed at all. These findings raise thepossibility that the genes of mOCIL, mOCILrP1, and mO-CILrP2 probably arose as a result of gene duplication eventsfrom a common ancestral gene.

We describe the isolation, cloning, and functions of thehuman homolog of OCIL (hOCIL) and show, in addition toits ability to inhibit osteoclast formation, that it also inhibitsthe osteoclast activity independent of osteoclast formation.

MATERIALS AND METHODS

Materials

Recombinant murine RANKL was obtained from Pepro-tech Inc. (Canton, MA, USA). Eagle’s minimum essentialmedium (MEM; Gibco Invitrogen, Mount Waverley, Vic-toria, Australia) and penicillin/streptomycin solution werepurchased from Sigma Chemical Co. (St Louis, MO, USA).Nonessential amino acids and fetal bovine serum (FBS)were purchased from CSL Biosciences (Parkville, Austra-lia). Ficoll-Paque was purchased from Pharmacia Biotech(Uppsala, Sweden). Tissue culture flasks, dishes, and plateswere purchased from Greiner Labortechnik (Frickenhasen,Germany). Recombinant stem cell factor (rhSCF), rhGM-CSF, and rhIL-3 were purchased from Chemicon (Te-

mecula, CA, USA). Recombinant human M-CSF was gen-erously provided by Genetics Institute (Cambridge, MA,USA). Human osteosarcoma cell line, MG63, cells wereroutinely grown in �-MEM containing 10% FBS. Incuba-tion was carried out at 37°C in a humidified atmosphereequilibrated with 5% CO2 in air. All other reagents were ofanalytic grade obtained from standard suppliers.

Isolation of hOCIL cDNA

A [32P] �-dCTP–labeled rOCIL (GenBank Accession no.AF321552) fragment was used to probe a human fetalcDNA library under low stringency hybridization condi-tions at 55°C in a hybridization buffer containing 4� SSPE,5� Denhardt’s solution, and 0.5% SDS for 24 h, and thenwashed with low stringency at 2� SSC with 0.1% SDS at40°C for 15 minutes and 1� SSC/0.1% SDS at 40°C for 15minutes. A 960-bp cDNA segment was obtained with 64%identity with mOCIL over a length of 346 bp (GenBankAccession no. AF321553). A search of the expressed se-quence tag (EST) database showed that this clone has a99.5% sequence identity over 209 bp with an EST clone ofunknown function obtained from human pregnant uterus(GenBank Accession no. AA029932). This EST clone wasobtained and further sequenced. To obtain the full-lengthcDNA of hOCIL, 5�- and 3�-rapid amplification of cDNAend (RACE) strategies were used. The specific antisenseprimer used for 5�-RACE was OCILh1 5�-CTCTGC-TCAGCCCAATCCAGTGATCAG-3� complementary tosequences within the hOCIL clone. First-strand cDNA syn-thesis was performed on either human placental total RNA(Clontech) or total RNA isolated from MG63 cells, a humanosteosarcoma cell line, using a SMART RACE cDNA Am-plification Kit (Clontech). The cDNA was further amplifiedby polymerase chain reaction (PCR) using OCILh1 andUNP primer as the 5� anchored primer using a touchdownPCR protocol with denaturation at 94°C for 1 minute, 5cycles at 94°C for 30 s, 72°C for 1 minute, and 5 cycles of94°C for 30 s, 70°C for 30 s, and 72°C for 1 minute,followed by 50 cycles of 94°C for 30 s, 65°C for 30 s, and72°C for 1 minute. A 3�-RACE strategy was also used toobtain the 3� ends of the cDNA using the SMART RACEcDNA Amplification Kit (Clontech). The sense specificprimer used was OCILh3�-1 5�-GCTGATCTTGCTCAG-GTTGAAAGCTTCC-3� complementary to sequenceswithin hOCIL. First-strand cDNA was synthesized fromtotal RNA isolated from MG63 cells. The cDNA was fur-ther amplified by PCR using OCILh3�-1 and UNP primer asthe 3� anchored primer. The PCR conditions used a touch-down PCR protocol with denaturation at 94°C for 1 minute,and 5 cycles at 94°C for 30 s, 72°C for 1 minute, and 5cycles of 94°C 30 s, 70°C for 30 s, and 72°C for 1 minute,followed by 30 cycles of 94°C for 30 s, 65°C for 30s, and72°C for 1 minute.

Expression and regulation of hOCIL mRNA

Total RNA was extracted from MG63 cells and normalhuman bone cells. First-strand cDNA was synthesized, andreverse transcriptase (RT)-PCR was carried out usingOCIL-h2 (5�-AAAGCTACCTTAATTTGGCG-3�) andOCIL-h3 (5�-TGGGCCTTTATATCTCAACAG-3�) as a

90 HU ET AL.

sense and antisense primer, respectively. The PCR was runat 94°C for 5 minutes, and 30 cycles of 94°C for 30 s, 60°Cfor 30 s, and 72°C for 1 minute, followed by a finalextension step of 72°C for 10 minutes. Southern blot anal-ysis was carried out using an internal antisense strand oli-gonucleotide probe complementary to hOCIL (OCIL-h4,5�-AGCTTTCTGGGCATGCAGC-3�). To ensure equalstarting quantities of RNA in each sample, the reversetranscribed material was also amplified using oligonucleo-tide primers specific for glyceraldehyde-3-phosphate dehy-drogenase (GAPDH) as described.(7)

Isolation of the hOCIL gene

To obtain the hOCIL gene, the 680-bp cDNA insert ofclone AA029932 was isolated as a probe and screened byGenome Systems, Inc. against the genomic BAC HumanRelease II Hybridization library, and one positive BAC wasidentified. After EcoRI restriction endonuclease digestion ofBAC DNA, Southern blot analysis was performed with the680-bp cDNA insert of clone AA029932. Two EcoRI frag-ments (4 and 10 kb) that hybridized with this clone wereidentified; the 4-kb fragment was cloned into the EcoRI siteof pBS (Stratagene), the 10-kb fragment was further di-gested with PstI and cloned into PstI- or PstI/EcoRI-digested pBS, and the nucleotide sequence was determinedfor each construct. In concert with this approach, we madeuse of the known intron-exon boundaries for rOCIL, mO-CIL, mOCILrP1, and mOCILrP2, which are conserved.(8)

Oligonucleotides were designed to the hOCIL cDNA se-quence near the predicted splice site junctions and wereused in cycle sequencing of the BAC clone DNA, its sub-clones in pBS, and MG63 genomic DNA. This approachfacilitated the determination of 10 kb of genomic nucleotidesequence (GenBank Accession no. AY144606).

Recombinant hOCIL and mOCIL protein constructs

To express the glutathione S-transferase (GST) fusionprotein for hOCIL, a PCR fragment specifying the extracel-lular domain of hOCIL (amino acid residues 71–191) wasamplified from the cDNA clone by PCR using a senseprimer 5�-GCGGGATCCCTTCAAGCTGCATGCCC-3�with a BamHI site and an antisense primer 5�-CCTGGA-ATTCGCTTTGCTGTAACATCTAGAC-3� with an EcoRIsite. The PCR product was digested with EcoRI and BamHIand cloned into pGEX-2T (Pharmacia Biotech). To expressthe GST-mOCIL protein as a positive control, a DNAfragment encoding the extracellular domain (amino acidresidues 76–207) of mOCIL was also constructed intopGEX-2T plasmid. The DNA fragment was amplified frommOCIL cDNA clone by PCR using OCILm67 5�-CCC-GGGGACCTATGCTGCTTGCCCGC-3� as a sense primerwith a SmaI site and OCILm68 5�-GAATTCCAGGCT-AAAAAGCGTCTCTTGG-3� as an antisense primer withan EcoRI site. The PCR product was digested with SmaI andEcoRI and cloned into pGEX-2T. The open reading frameand GST fusion for both constructs were confirmed bysequencing.

Expression and purification of mOCIL and hOCIL

Competent strain BL21 E. coli cells were transformedwith the constructs and the corresponding fusion proteins in-duced with isopropyl-1-thio-�-D-galactopyranoside (IPTG).The GST-hOCIL fusion protein was isolated from the solublebacterial fraction using glutathione sepharose 4B affinity chro-matography as outlined in the manufacturer’s instructions(Pharmacia Biotech). The eluant fractions were subjected toSDS-PAGE and transferred to polyvinylidene diflouride mem-branes (Roche Molecular Biochemicals). Western blot analy-ses were performed with a rabbit anti-GST serum (New En-gland Biolabs Inc.) and a BM chemiluminescence blottingsubstrate (POD) detection system (Roche Molecular Bio-chemicals). Fractions containing the GST-OCIL fusion proteinwere pooled and concentrated using an Amicon ultrafiltrationYM10 membrane (Millipore, Bedford, MA, USA). The pro-tein concentration was ascertained in a BCA protein assay(Pierce) and by SDS-PAGE stained with Coomasie BrilliantBlue against bovine serum albumin (BSA) standard. MBP-OCIL was prepared as previously described.(7)

Isolation of normal human bone cells

Normal human bone cells (NHBCs) were cultured, aspreviously described,(9) from trabecular bone fragments ob-tained either from patients (58–80 years old) during routinejoint replacement surgery in the Department of OrthopaedicSurgery and Trauma at the Royal Adelaide Hospital, oralternatively, from bone fragments isolated from bone mar-row (BM) aspirates obtained from iliac crest biopsy ofnormal volunteers (�35 years old) from the Division ofHematology, Institute of Medical and Veterinary Science(IMVS), Adelaide, South Australia (with Institutional Hu-man Ethics Committee approval). NHBCs were cultured in75-cm2 tissue-culture flasks in �-MEM supplemented with10% FBS, 2 mM L-glutamine, and 100 �M L-ascorbate-2-phosphate (Novachem, Melbourne, Australia). Single cellsuspensions were obtained from confluent primary NHBCcultures by enzymatic digestion, consisting of collagenase(3 mg/ml, collagenase type I; Worthington BiochemicalCo., Freehold, NJ, USA) and dispase (4 mg/ml, NeutralProtease grade II; Boehringer Mannheim) for 30 minutes at37°C, followed by trypsin digestion for 5 minutes at 37°C.Cell suspensions were washed with growth medium beforebeing passed through a Falcon cell strainer (Becton Dick-inson Labware, Franklin Lakes, NJ, USA) to obtain singlecell suspensions.

Giant cell tumors

Giant cell tumor (GCT) specimens were collected andcryopreserved as previously described.(10) Freshly thawedGCT cells were plated onto 16-mm2 slices of sperm whaledentine in 6-mm-diameter wells at 2 � 105 cells/well in�-MEM medium containing 10% FBS. Recombinant hu-man GST-OCIL was titrated from 2 to 500 ng/ml, and GSTfrom 1.25 to 325 ng/ml; this concentration of GST wasequivalent in a molar ratio to GST-OCIL. The cells werecultured for 3 days before being processed for scanningelectron microscopy. For this, the bone slices were treatedwith 1 M ammonium hydroxide for 1 h and ultrasonicated

91CHARACTERIZATION OF HUMAN OCIL

for 30 minutes to remove the adherent cells. After this, thebone slices were washed three times in deionized water,dehydrated by washing in 100% ethanol, and air-dried be-fore being mounted on stubs and carbon-gold coated forvisualization using a Philips XL-20 scanning electron mi-croscope. Contiguous resorption sites were counted fromrandom, low magnification (100�) fields. Data are ex-pressed as the mean � SE of at least four fields.

Murine osteoclast formation assay

Mouse bone marrow cells were prepared from adultC57BL/6J mice. Cells were cultured for 7 days in culturemedium containing 10% FBS and 25 ng/ml M-CSF and 100ng/ml RANKL in the absence or presence of recombinantGST (control) or recombinant GST-hOCIL and MBP-mOCIL fusion proteins, respectively, at various concentra-tions (as indicated). Treatments were added on day 0 andwhen the medium was changed on day 3. After 7 days, cellswere fixed and stained for TRACP as described.(11)

Human osteoclastogenesis and resorption assays

Human osteoclastogenesis assays were performed usingCFU-M/CFU-GM cells derived from human umbilical cordblood as described.(12) Human umbilical cord blood wasobtained from healthy donors under a protocol approved bythe Barwon Health Research and Ethics Committee. Amononuclear cell fraction containing monocytes and lym-phocytes (CBMCs) was isolated by Ficoll-Paque densitygradient centrifugation. CBMCs (1 � 106 cells/culture)were cultured in semi-solid medium based on Iscove’sMDM containing 1% methylcellulose, 30% FBS, 1% BSA,10�4 M 2-mercaptoethanol, 2 mM L-glutamine, 50 ng/mlrecombinant SCF (rhSCF), 10 ng/ml rhGM-CSF, and 10ng/ml rhIL-3. The cells were plated in a volume of 1.0 ml in35-mm culture dishes and incubated at 37°C in a humidifiedatmosphere of 5% CO2-air for 10 days. Pooled colonieswere harvested into PBS for osteoclast formation assays.

Cells from semi-solid medium were settled onto dentineslices (4 � 104 cells/well) in 96-well tissue culture platescultured in 200 �l MEM containing 10% FBS, nonessentialamino acids, penicillin/streptomycin solution (penicillin 50U/ml; streptomycin 50 �g/ml), 2 mM L-glutamine, M-CSF(25 ng/ml), and RANKL (125 ng/ml). In all experiments,there were eight replicates in each group, and at least threeindependent experiments were performed for each protocol.One-half of the medium changes were made twice weeklyby adding double concentrations of the cytokine additives ortreatments as described in the Results section. At 14 days,cultured cells were fixed in 1% formalin and stained forTRACP activity. The formation of multinucleated OCs wasassessed by transmission light microscopy and quantifiedusing microcomputer image analysis software (MCID-Imaging Research Inc., Ontario, Canada). Subsequent tothis, cells were removed from dentine slices by brief agita-tion in chloroform:methanol (2:1). Xylene-free black inkwas applied to the resorbed surface of each slice, andresidual ink was removed by wiping against absorbent pa-per. Resorption was assessed by transmission light micros-

copy, and the percentage area resorbed was quantified usingMCID software.

Results were analyzed by one-way ANOVA with Fisher’smultiple comparison using Minitab software (Minitab Inc.).

RESULTS

Predicted hOCIL protein structure

The full-length hOCIL cDNA was 845 bp in length,encoding a 191 amino acid C-lectin type II membrane-bound protein. The protein is predicted to have an intracel-lular domain of 30 amino acids, transmembrane domain of29 amino acids, and an extracellular domain of 132 aminoacids; no alternatively spliced variants were identified thatmay encode secreted forms of hOCIL. Within the extracel-lular domain is a C-lectin domain of 112 amino acids(amino acids 75–186; Fig. 1). The deduced protein sequenceshowed 42% identity overall to the deduced protein se-quences of mOCIL(7) (53% identity in the C-lectin domain),and two proteins closely related to mOCIL in structure andfunction, designated as mOCILrP1 and mOCILrP2.(8) Fig-ure 1 compares the protein structure of hOCIL with those ofmOCIL, mOCILrP1, mOCILrp2, and human CD69. CD69belongs to a family of C-type lectins with a 36% homologynoted between the C-type lectin domains of human CD69and mOCIL. The differences are principally within theintracellular domains at the N termini, and there is 80%identity between the mouse and human OCIL proteins in theC-lectin domain. The extracellular domain of the hOCIL hasfive cysteine residues and two predicted N-linked glycosyl-ation sites at residues 95 and 147, and no potential post-translational cleavage sites were present in the primaryamino acid sequence that would predict production of ashed form of hOCIL.

Regulation of hOCIL mRNA expression

hOCIL mRNA is constitutively expressed in human os-teoblastic cells and the MG63 human osteosarcoma cell line(Fig. 2A). In MG63 cells, hOCIL mRNA expression wasincreased by 7-fold at 24 h when treated with 1 ng/ml IL-1�(Fig. 2B), and 4-fold at 8 h by 10�7 M PGE2 (Fig. 2C). Incontrast, dexamethasone, 1,25-dihydroxyvitamin D3 (Fig.2D), and parathyroid hormone (PTH; data not shown) werewithout effect; similarly, dexamethasone or 1,25-dihydroxyvitamin D3 did not affect OCIL expression innormal human osteoblast populations (data not shown).Furthermore, other known inhibitors of osteoclast formationincluding IL-4, IL-12, IL-18, interferon �, and GM-CSF didnot regulate OCIL mRNA production in primary osteoblastsor osteosarcoma cell lines (data not shown). IL-1� andPGE2 also dose-dependently increased cyclic adenosinemonophosphate (cAMP) levels, whereas there was nocAMP response to PTH(1-34) treatment in MG63 cells(data not shown), indicating that there is no PTH receptorexpression in these cells.

Gene structure of hOCIL

We determined 10 kb of genomic sequence (GenBankAccession no. AY144606), corresponding to the cDNAsequence from 190 to 845 bp of hOCIL (covering exons 3,

92 HU ET AL.

4, and 5; Fig. 3). Our genomic sequence had 100% identityto a sequence within a human genomic clone of 178,607 bp(GenBank Accession no. AC007068). The 5� flanking re-gion, promoter region, and the first 189 bp of hOCIL cDNAsequences are represented in the 74,801-bp sequence depos-ited in GenBank (Accession no. AC010186). The hOCILgene is approximately 25 kb in length and contains fiveexons, and the positions of each of the exon/intron bound-aries are conserved with those for mOCIL (Fig. 3). Theintron-exon splice junctions of hOCIL gene were confirmedby direct cycle sequencing of the genomic clone DNA usingprimers against exon sequences and are shown in Table 1.

Inhibition of osteoclast formation by hOCIL

The primary amino acid sequence of hOCIL does notpredict that this protein would be shed from the transmem-

brane, and genomic structure does not indicate alternativemRNA species that would result in secreted forms of hO-CIL, suggesting that OCIL requires cell-to-cell contact forits activity. Therefore, a recombinant protein for the extra-cellular domain of hOCIL fused to GST was prepared todetermine if hOCIL has a similar biological action to mO-CIL in inhibiting murine and human osteoclastogenesis.Murine marrow cells, when cultured in the presence ofRANKL (100 ng/ml) and M-CSF (25 ng/ml) for 7 days,form multinucleate TRACP� osteoclasts (Fig. 4A). Recom-binant mOCIL dose-dependently inhibited osteoclast forma-tion over a comparable dose as previously described.(7,8)

Like mOCIL, GST-hOCIL dose-dependently inhibited mu-rine osteoclast formation, with 80% inhibition observed atthe highest dose of 500 ng/ml (Fig. 4B). This is comparablewith the complete inhibition of murine osteoclast formation

FIG. 1. Structural alignment ofdeduced amino acid sequencesof hOCIL, mOCIL, mOCILrP1,mOCILrP2, and hCD69. The de-duced amino acid sequences areshown, segmented into cytoplasmic,transmembrane (TM; above align-ment for hOCIL and below align-ment for mOCIL, mOCILrP1,mOCILrP2, and CD69), and extra-cellular domains. C-lectin domainsare boxed. ƒ conserved cysteines;�, cysteine residues that are notconserved; [, potential N-linked gly-cosylation sites.

93CHARACTERIZATION OF HUMAN OCIL

obtained with 500 ng/ml MBP-mOCIL (Fig. 4A). GSTprotein alone had no effect on osteoclast formation (Fig.4B). In the human system, osteoclastogenesis assays wereperformed using CFU-M/CFU-GM cells derived from hu-man umbilical cord blood. A mononuclear cell fractioncontaining monocytes and lymphocytes was allowed togrow and differentiate into multinucleate osteoclasts afteran initial selection process by Ficoll-Paque density gradientcentrifugation and culture in semi-solid medium containingSCF, IL-3, and GM-CSF. The cells were treated with humanM-CSF (25 ng/ml) and RANKL (125 ng/ml) for 14 days toallow formation of multinucleate TRACP� osteoclasts. Incontrast with the actions of OCIL on murine osteoclastformation, neither mOCIL nor hOCIL could completelyablate human osteoclast formation. GST-hOCIL dose-dependently inhibited human osteoclast formation, with50% inhibition observed at the highest dose of 500 ng/ml

(Fig. 5A). GST-mOCIL at 32 and 500 ng/ml was equipotentto GST-hOCIL in its ability to inhibit human osteoclasto-genesis. Concomitant with the ability of hOCIL to inhibitosteoclast formation, it also inhibited bone resorption, witha significant effect noticed at 125 ng/ml (Fig. 5B). Micro-graphs of osteoclasts produced from human cord blooddemonstrated both multinuclear and mononuclear cell for-mation (Fig. 6A), with hOCIL affecting the differentiationof both cell types (Fig. 6B). The reduction in resorptionnoted when human CFU-GM cells were cultured on dentinein the presence of hOCIL is illustrated in Figs. 6C and 6D.

Inhibition of bone resorption by hOCIL of humanGCT-derived osteoclasts

hOCIL significantly inhibited the formation of resorptionpits by mature osteoclasts present in disaggregated GCT

FIG. 2. hOCIL mRNA expres-sion. Semiquantitative RT-PCRwas carried out to determine theexpression of hOCIL mRNA lev-els. (A) hOCIL mRNA was con-stitutively expressed in four pa-tient samples of NHBCs andMG63 cells. MG63 cells weretreated with (B) IL-1� (1 ng/ml),(C) PGE2 (10�7 M), (D) 1,25-dihydroxyvitamin D3 (10�8 M),or dexamethasone (10�7 M) in atime course, with samples takenat 0, 2, 4, 6, 8, and 24 h. Quanti-fied at the right are hOCILmRNA levels normalized againstGAPDH mRNA and relative tothe control level.

94 HU ET AL.

specimens in a dose-dependent manner in doses rangingfrom 2 to 500 ng/ml. A significant effect was observed evenwith the lowest dose (2 ng/ml). GST alone (control) in dosesranging from 1.25 to 325 ng/ml had no effect on boneresorption (Fig. 7).

DISCUSSION

hOCIL was isolated from a human fetal cDNA libraryusing rOCIL as a probe under low stringency hybridizationconditions. hOCIL is highly conserved with mOCIL, mO-CILrP1, and mOCILrP2 in its primary amino acid sequence.The hOCIL cDNA predicts a 191 amino acid type II mem-brane protein, with the 112 amino acid C-type lectin regionin the extracellular domain having 53% identity with theC-type lectin sequences of rOCIL and mOCIL. Like themOCIL gene, the hOCIL gene contains five exons, and theintron/exon boundaries are conserved with those for mO-CIL. However, unlike mOCIL, hOCIL is not controlled byan inverted TATA promoter.(7,8) The 5� flanking regionadjacent to exon 1 in hOCIL (identified by 5�-RACE strat-egies) contains a GAATCA sequence. This sequence hasbeen reported as a promoter in c-fes gene.(13)

Although constitutively expressed by human osteoblasticand MG63 osteosarcoma cells, hOCIL mRNA expression inMG63 cells was responsive to IL-1� and PGE2, as wasmOCIL RNA in mouse osteoblasts but not mOCILrP1/rP2.However, in contrast to mOCIL, hOCIL mRNA expressionwas not upregulated by IL-11, 1,25(OH)2 vitamin D3, reti-noic acid, and PTH.(7) This suggests some similarity inregulatory control between mOCIL and hOCIL expression,but also marked differences. These differences may beattributed to the number of closely related, and apparentlyredundant, OCIL genes that exist in rodents compared withhumans.

The cDNA sequence of hOCIL is identical to that of anew lectin-like transcript (LLT1) reportedly expressed onhuman NK cells.(14) LLT1 is a member of a superfamily ofNK cell receptors with lectin-like domains encoded by theNK gene complex located on chromosome 12 in the humanand chromosome 6 in the mouse.(15–17) LLT1 shares 59%similarity with the C-lectin domain of the activation-induced C-type lectin (AICL) and 56% identity to CD69,respectively.(18–20) Similarities are also present to other NKcell receptors belonging to the family of C-type lectins, such

FIG. 3. Genomic structure of the hOCIL gene compared with the mOCIL gene. Schematic representations of the genomic organization of hOCILand mOCIL genes are illustrated, and the positions of the intron-exon splice junctions are presented in relation to predicted structural features. Theintronic sequences are indicated by the horizontal line, and exons are boxed; exons specifying the coding regions are shaded gray, whereas theregion specifying the transmembrane domain is in black, and exons encoding the untranslated regions are unfilled. The numbers below intronsrepresent the intron size, and the cDNA size in base pairs is indicated at the end of each gene. The arrows represents the putative promoters (TATAfor mOCIL and GAATCA for hOCIL).

TABLE 1. NUCLEOTIDE SEQUENCES OF THE INTRON-EXON SPLICE JUNCTIONS OF THE HUMAN OCIL GENE

Exon

Exonsize(bp) 5� Splice junctions

Intronsize(bp) 3� Splice junctions

Aminoacid(s)

interrupted

1 79 AAC CCA G gtaagaagctgggctctacagagtaagtgt 11,127 ttacaccatctgaattgcctgcaaacccag GT TGT CTG G21

2 111 TTA AGC G gtaagtgaactgtaatctttattcatctag 5,281 ttattttatttatttttattttattttcag CA ATA AGA A58

3 185 CAG GAA CTG gtaagaaaatagttctggccagaatcaaag 4,741 aatgaatatcatctctgtgttttctgaaag AAT TTC CTG L119/N120

4 104 ACA AGA CA gtaagttctaaaaatctggcagtaatattt 1,828 ggccaactgattctgcttcttctcttgcag G TTT CCT Q154

5 366 NA NA NA NA

The splice donor and acceptor splice sequences of the human OCIL gene are presented. The exonic nucleotide sequences are represented in capital lettersand the intronic sequences in lowercase letters. Exons are numbered from the 5� transcriptional start site for the hOCIL gene. Exon and intron sizes areindicated in base pairs and were determined by RT-PCR and gene sequences. The amino acids interrupted are indicated.

NA, not applicable.

95CHARACTERIZATION OF HUMAN OCIL

as NKG2-D (53%), CD94 (51%), and Ly-49D (41%).(21–24)

Similarly, a number of C-type lectin-related (Clr) geneshave been identified in the NK gene complex on murinechromosome 6, and within this complex are the genes forNK1.1, NKR-P1 family, and CD69.(25,26) mOCIL is identi-cal to Clr-b, the splice variant of mOCILrP2 (mOCILrP2b)is identical to Clr-g, and mOCILrP1 matches with the partialamino acid sequence provided for Clr-d.(8,25) The mOCIL/Clr family of genes map within a 700-kb region of murinechromosome 6 immediately juxtaposed to the CD69 locus

telomeric to A1c.(8,25) In the case of LLT1, it is located inthe human NK gene complex within 100 kb of the CD69gene. Unlike other type II receptors in the NK gene complexthat are restricted to NK cells and a subset of T-cells, CD69and AICL are expressed in most cells of hematopoieticorigin.(27–30) Studies in mice deficient in CD69 throughtargeted gene deletion suggested that CD69 plays a role inB-cell development, with otherwise normal hematopoieticcell development as well as normal T-cell subpopulationsand functions(31); no bone defects have been ascribed tomice deficient for CD69 expression.

mOCIL belongs to a family of closely related genes thatprobably arose through the process of gene duplications.(8)

The presence of multiple transcripts of human LLT1 inNorthern blot analysis(14) thus raises the possibility that asimilar situation could occur with hOCIL, and human geneshomologous to mOCILrP1 and mOCILrP2 might yet bediscovered.

The extracellular domain of hOCIL was expressed as asoluble recombinant protein in E. coli, and its biologicaleffects were compared with recombinant mOCIL on murineand human osteoclastogenesis. Human osteoclastogenesiswas performed with human cord blood–derived CFU-GM

FIG. 4. Effect of recombinant mOCIL and hOCIL on murine oste-oclast formation. Murine marrow cell cultures were treated with solubleRANKL (50 ng/ml) and M-CSF (25 ng/ml) for 7 days. (A) Control andMBP-treated (500 ng/ml) cultures are compared with cultures treatedwith 5, 50, and 500 ng/ml MBP-mOCIL. (B) Control and GST-treated(500 ng/ml) cultures are compared with increasing concentrations (8,32, 125, 500 ng/ml) of GST-hOCIL. TRACP� multinucleate oste-oclasts (MNCs) were counted at the end of the incubation period. Barsrepresent mean � SE; n � 4 for each treatment. *p � 0.05, **p � 0.01,***p � 0.001 vs. vector control alone.

FIG. 5. Effect of recombinant hOCIL and mOCIL on osteoclastgeneration and dentine resorption from human CFU-GM cells. Oste-oclasts were generated from CFU-M/CFU-GM–derived cells culturedon dentine slices in the presence of M-CSF (25 ng/ml) and RANKL(125 ng/ml) for 14 days. OCIL (human or murine) was added at theconcentrations indicated, and (A) inhibition of osteoclast generationand (B) resorption on sperm whale dentine slices were quantitated asdescribed in the Materials and Methods section. Results represent themean � SE of three separate experiments with eight replicates in eachexperiment; *p � 0.05 relative to control.

96 HU ET AL.

cells cultured on dentine slices in the presence of M-CSFand RANKL. Treatment with hOCIL resulted in a 50%inhibition of human osteoclast formation at the highestconcentration used (500 ng/ml), and mOCIL was equipotent

to hOCIL in its ability to inhibit human osteoclastogenesis.Consistent with this reduction in osteoclast numbers, bothhOCIL and mOCIL significantly reduced resorption pitformation in the same cultures. Similarly, in murine oste-oclast formation assays using mouse marrow cells, hOCILdose-dependently inhibited osteoclast formation; however,only 80% inhibition was observed at the highest dose of 500ng/ml compared with complete inhibition obtained with 500ng/ml mOCIL. Thus, there seems to be minor differences incross-species efficacy in the inhibition of osteoclast forma-tion and function by hOCIL with mOCIL.

Significantly, in this study, we showed the ability ofhOCIL to inhibit bone resorption by mature osteoclastsderived from GCTs of bone. GCTs, also known as oste-oclastomas, are primary neoplasms of the skeleton. They arelocally destructive, causing extensive osteolysis. GCTs con-tain within the tumor mass variable numbers of large,multinucleated cells that are phenotypically and function-ally osteoclasts capable of extensive bone resorption invitro. The stromal cells of GCTs are believed to be thetumor cells, and they induce osteoclastic bone resorption byrecruiting osteoclast precursors, promoting their differenti-ation into functional osteoclasts,(32) and they do so, at leastin part, in a RANKL-dependent manner.(10) The inhibitionof bone resorption in vitro by mature osteoclasts by hOCILshows an action that is distinct from its ability to inhibitosteoclast formation from mononuclear precursors; this wasshown in the murine and human osteoclastogenesis assays.

While the precise mechanism of OCIL action remains tobe determined, its ability to inhibit osteoclast formation andresorption was akin to the functions of OPG.(3) However,

FIG. 6. Inhibition of human os-teoclastogenesis and resorptionby hOCIL. Human cord blood–derived CFU-M/CFU-GM cellswere cultured on dentine slices in�-MEM (10% vol/vol FBS) withM-CSF (25 ng/ml) and RANKL(125 ng/ml) in the (A) absence or(B) presence GST-hOCIL (125ng/ml; �100). Cells were re-moved, and resorption pits werestained for transmission micros-copy; (C) control and (D) GST-hOCIL at 125 ng/ml.

FIG. 7. Effect of hOCIL on bone resorption by mature osteoclasts.Cell suspensions derived from disaggregated human GCTs of bone thatcontained numerous mature osteoclasts were plated onto slices ofsperm whale dentine. Recombinant human GST-OCIL and GST (con-trol) were added at doses ranging from 2 to 500 ng/ml. Cells werecultured for 3 days before being processed for scanning electronmicroscopy. Resorption lacunae were counted from random, low-magnification (�100) fields. Data are expressed as the average of fourfields � SE. This experiment was performed twice with similar results;**p � 0.01, ***p � 0.001 vs. GST alone.

97CHARACTERIZATION OF HUMAN OCIL

unlike in OPG, which mediates its effects by binding to andinactivating RANKL biological actions, no protein interac-tion between OCIL and RANKL was shown (data notshown). In contrast, and similar with other C-type lectins,OCIL can bind to high molecular sulfated sugars includingfucoidan, �-carrageenan, and dextran sulfate (C. Gange,unpublished data, 2001). This finding suggests that carbo-hydrate recognition may be important for OCIL engagementto a hitherto unrecognized receptor present on the osteoclastand its progenitor cells. Receptors for other C-type lectinshave not been identified, and biological functions for manyof these, including CD69, the closest family member toOCIL, still remain to be determined. The reported tissuedistribution of human LLT1, a mRNA transcript that ap-pears identical to that of hOCIL, established expression inhuman peripheral blood leukocytes, lymph nodes, thymus,and spleen, but not in fetal liver and bone marrow(14); thelocalization studies were limited to these tissues. In previousstudies with mOCIL, we established a wide tissue distribu-tion of mOCIL by in situ hybridization and immunohisto-chemistry using an antibody raised against a peptide in theextracellular domain of mOCIL. mOCIL mRNA and pro-tein were expressed in osteoblasts, chondrocytes, skeletalmuscle, brain, lung, heart, kidney, small intestine, liver,skin, and spleen.(7) The tissue distribution of mOCIL andmOCILrP2 was identical to that previously described forRANKL.(8,33) One of the more intriguing aspects aboutfactors regulating osteoclastogenesis has been the demon-stration that OCIL, RANKL, and its decoy receptor, OPG,are widely distributed in tissues other than bone. OCIL,along with OPG, may thus act as a tonic inhibitor ofosteoclast formation in normal physiology in extraskeletaltissues, with OCIL acting in a contact-dependent processcaused by its transmembrane topology, whereas OPG mayhave distant actions from the cell expressing it because of itbeing a secreted inhibitor of osteoclast formation and activ-ity. Furthermore, each molecule probably has a multifunc-tional role, with the possibility of different effects depend-ing on tissue-specific microenvironments. Such functionsfor OCIL may be revealed by studies in genetically alteredmice (knockout or transgenic) to determine physiologicalroles or in appropriate animal models such as cancer-induced bone loss or arthritis for pathophysiological func-tions. Obviously, the potential of hOCIL as a therapy needsto be explored, as well as determining its actions on theskeleton and potentially on cells of the immune system,where it is abundantly expressed.

ACKNOWLEDGEMENTS

This work was supported by Program Grant 003211 (toHZ, MTG, and KWN) from the National Health and Med-ical Research Council, Australia.

REFERENCES

1. Lacey DL, Timms E, Tan HL, Kelley MJ, Dunstan CR, Burgess T,Elliott R, Colombero A, Elliott G, Scully S, Hsu H, Sullivan J,Hawkins N, Davy E, Capparelli C, Eli A, Qian YX, Kaufman S,Sarosi I, Shalhoub V, Senaldi G, Guo J, Delaney J, Boyle WJ 1998Osteoprotegerin ligand is a cytokine that regulates osteoclast dif-ferentiation and activation. Cell 93:165–176.

2. Yasuda H, Shima N, Nakagawa N, Yamaguchi K, Kinosaki M,Mochizuki S, Tomoyasu A, Yano K, Goto M, Murakami A, TsudaE, Morinaga T, Higashio K, Udagawa N, Takahashi N, Suda T1998 Osteoclast differentiation factor is a ligand forosteoprotegerin/osteoclastogenesis-inhibitory factor and is identi-cal to TRANCE/RANKL. Proc Natl Acad Sci USA 95:3597–3602.

3. Simonet WS, Lacey DL, Dunstan CR, Kelley M, Chang MS, LuthyR, Nguyen HQ, Wooden S, Bennett L, Boone T, Shimamoto G,DeRose M, Elliott R, Colombero A, Tan HL, Trail G, Sullivan J,Davy E, Bucay N, Renshaw-Gegg L, Hughes TM, Hill D, PattisonW, Campbell P, Sander S, Van G, Tarpley J, Derby P, Boyle WJ1997 Osteoprotegerin: A novel secreted protein involved in theregulation of bone density. Cell 89:309–319.

4. Kong YY, Yoshida H, Sarosi I, Tan HL, Timms E, Capparelli C,Morony S, Oliveira-dos-Santos AJ, Van G, Itie A, Khoo W,Wakeham A, Dunstan CR, Lacey DL, Mak TW, Boyle WJ, Pen-ninger JM 1999 OPGL is a key regulator of osteoclastogenesis,lymphocyte development and lymph-node organogenesis. Nature397:315–323.

5. Anderson DM, Maraskovsky E, Billingsley WL, Dougall WC,Tometsko ME, Roux ER, Teepe MC, DuBose RF, Cosman D,Galibert L 1997 A homologue of the TNF receptor and its ligandenhance T-cell growth and dendritic-cell function. Nature 390:175–179.

6. Martin TJ, Romas E, Gillespie MT 1998 Interleukins in the controlof osteoclast differentiation. Crit Rev Eukaryot Gene Expr 8:107–123.

7. Zhou H, Kartsogiannis V, Hu YS, Elliott J, Quinn JM, McKinstryWJ, Gillespie MT, Ng KW 2001 A novel osteoblast-derived c-typelectin that inhibits osteoclast formation. J Biol Chem 276:14916–14923.

8. Zhou H, Kartsogiannis V, Quinn JM, Ly C, Gange C, Elliott J, NgKW, Gillespie MT 2002 Osteoclast inhibitory lectin, a family ofnew osteoclast inhibitors. J Biol Chem 277:48808–48815.

9. Gronthos S, Zannettino AC, Graves SE, Ohta S, Hay SJ, SimmonsPJ 1999 Differential cell surface expression of the STRO-1 andalkaline phosphatase antigens on discrete developmental stages inprimary cultures of human bone cells. J Bone Miner Res 14:47–56.

10. Atkins GJ, Bouralexis S, Haynes DR, Graves SE, Geary SM,Evdokiou A, Zannettino AC, Hay S, Findlay DM 2001 Osteopro-tegerin inhibits osteoclast formation and bone resorbing activity ingiant cell tumors of bone. Bone 28:370–377.

11. Quinn JM, Morfis M, Lam MH, Elliott J, Kartsogiannis V, Wil-liams ED, Gillespie MT, Martin TJ, Sexton PM 1999 Calcitoninreceptor antibodies in the identification of osteoclasts. Bone 25:1–8.

12. Hodge JM, Kirkland MA, Aitken CJ, Myers DE, Nicholson GC2002 Osteoclastic potential of human cord blood mononuclearcells and derived CFU-GM: Effects of M-CSF, GM-CSF and IL-3.J Bone Miner Res 12:S455.

13. Heydemann A, Boehmler JH, Simon MC 1997 Expression of twomyeloid cell-specific genes requires the novel transcription factor,c-fes expression factor. J Biol Chem 272:29527–29537.

14. Boles KS, Barten R, Kumaresan PR, Trowsdale J, Mathew PA1999 Cloning of a new lectin-like receptor expressed on humanNK cells. Immunogenetics 50:1–7.

15. Yabe T, McSherry C, Bach FH, Fisch P, Schall RP, Sondel PM,Houchins JP 1993 A multigene family on human chromosome 12encodes natural killer-cell lectins. Immunogenetics 37:455–460.

16. Brown MG, Scalzo AA, Matsumoto K, Yokoyama WM 1997 Thenatural killer gene complex: A genetic basis for understandingnatural killer cell function and innate immunity. Immunol Rev155:53–65.

17. Yokoyama WM, Seaman WE 1993 The Ly-49 and NKR-P1 genefamilies encoding lectin-like receptors on natural killer cells: TheNK gene complex. Annu Rev Immunol 11:613–635.

18. Ziegler SF, Ramsdell F, Hjerrild KA, Armitage RJ, Grabstein KH,Hennen KB, Farrah T, Fanslow WC, Shevach EM, Alderson MR1993 Molecular characterization of the early activation antigenCD69: A type II membrane glycoprotein related to a family ofnatural killer cell activation antigens. Eur J Immunol 23:1643–1648.

19. Lopez-Cabrera M, Santis AG, Fernandez-Ruiz E, Blacher R, EschF, Sanchez-Mateos P, Sanchez-Madrid F 1993 Molecular cloning,expression, and chromosomal localization of the human earliestlymphocyte activation antigen AIM/CD69, a new member of the

98 HU ET AL.

C-type animal lectin superfamily of signal-transmitting receptors. JExp Med 178:537–547.

20. Hamann J, Fiebig H, Strauss M 1993 Expression cloning of theearly activation antigen CD69, a type II integral membrane proteinwith a C-type lectin domain. J Immunol 150:4920–4927.

21. Weis WI, Kahn R, Fourme R, Drickamer K, Hendrickson WA1991 Structure of the calcium-dependent lectin domain from a ratmannose-binding protein determined by MAD phasing. Science254:1608–1615.

22. Wong S, Freeman JD, Kelleher C, Mager D, Takei F 1991 Ly-49multigene family. New members of a superfamily of type IImembrane proteins with lectin-like domains. J Immunol 147:1417–1423.

23. Weis WI, Taylor ME, Drickamer K 1998 The C-type lectin super-family in the immune system. Immunol Rev 163:19–34.

24. Houchins JP, Yabe T, McSherry C, Bach FH 1991 DNA sequenceanalysis of NKG2, a family of related cDNA clones encoding typeII integral membrane proteins on human natural killer cells. J ExpMed 173:1017–1020.

25. Plougastel B, Dubbelde C, Yokoyama WM 2001 Cloning of Clr, anew family of lectin-like genes localized between mouse Nkrp1aand Cd69. Immunogenetics 53:209–214.

26. Plougastel B, Matsumoto K, Dubbelde C, Yokoyama WM 2001Analysis of a 1-Mb BAC contig overlapping the mouse Nkrp1cluster of genes: Cloning of three new Nkrp1 members, Nkrp1d,Nkrp1e, and Nkrp1f. Immunogenetics 53:592–598.

27. Wagtmann N, Rojo S, Eichler E, Mohrenweiser H, Long EO 1997A new human gene complex encoding the killer cell inhibitoryreceptors and related monocyte/macrophage receptors. Curr Biol7:615–618.

28. Testi R, D’Ambrosio D, De Maria R, Santoni A 1994 The CD69receptor: A multipurpose cell-surface trigger for hematopoieticcells. Immunol Today 15:479–483.

29. Lanier LL 1998 NK cell receptors. Annu Rev Immunol 16:359–393.

30. Hamann J, Montgomery KT, Lau S, Kucherlapati R, van Lier RA1997 AICL: A new activation-induced antigen encoded by thehuman NK gene complex. Immunogenetics 45:295–300.

31. Lauzurica P, Sancho D, Torres M, Albella B, Marazuela M,Merino T, Bueren JA, Martinez AC, Sanchez-Madrid F 2000Phenotypic and functional characteristics of hematopoietic celllineages in CD69-deficient mice. Blood 95:2312–2320.

32. James IE, Dodds RA, Lee-Rykaczewski E, Eichman CF, ConnorJR, Hart TK, Maleeff BE, Lackman RD, Gowen M 1996 Purifi-cation and characterization of fully functional human osteoclastprecursors. J Bone Miner Res 11:1608–1618.

33. Kartsogiannis V, Zhou H, Horwood NJ, Thomas RJ, Hards DK,Quinn JM, Niforas P, Ng KW, Martin TJ, Gillespie MT 1999Localization of RANKL (receptor activator of NF kappa B ligand)mRNA and protein in skeletal and extraskeletal tissues. Bone25:525–534.

Address reprint requests to:Kong Wah Ng, MD, PhD

Department of Endocrinology and DiabetesSt. Vincent’s Hospital

Victoria ParadeFitzroy 3065, Victoria, Australia

E-mail: k.ng@medicine.unimelb.edu.au

Received in original form May 30, 2003; in revised form July 8,2003; accepted August 6, 2003.

99CHARACTERIZATION OF HUMAN OCIL