Advanced Glycation Endproducts Stimulate Interleukin-6Production by Human Bone-Derived Cells*

MIKI TAKAGI,1 SOJI KASAYAMA,1 TAKEHISA YAMAMOTO,2 TAKASHI MOTOMURA,1

KUNIHIKO HASHIMOTO,1 HIROYASU YAMAMOTO,1 BUNZO SATO,3 SHINTARO OKADA,2

and TADAMITSU KISHIMOTO1

ABSTRACT

Advanced glycation endproducts (AGEs), which result from nonenzymatic reactions of glucose with tissue proteins,have been shown to accumulate on long-lived proteins in advanced aging and diabetes mellitus. Thus, AGEs havebeen implicated in some of the chronic complications associated with these disorders. In this study, we investigatedthe effects of the glucose-modified protein on the production of the potent bone resorption factors by cells derivedfrom explants of human bone. AGEs stimulated the release of interleukin-6 (IL-6) in the culture supernatants fromthe bone-derived cells and increased the levels of IL-6 mRNA in the cells. By contrast, the levels of IL-11 in theculture supernatants were not altered by AGEs, and the other bone resorption factors IL-1a and IL-1b wereundetectable (<1.0 pg/ml) either without or with the treatment of AGEs. Electrophoretic mobility-shift assaysrevealed that the transcription nuclear factor-kB, which is critical for the inducible expression of IL-6, wasactivated in the nuclear extracts frommouse osteoblasticMC3T3-E1 cells treated with AGEs. These results suggestthat AGEs are involved in bone remodeling modulation by stimulating IL-6 production in human bone-derivedcells. (J Bone Miner Res 1997;12:439–446)

INTRODUCTION

OSTEOPOROSIS is a group of skeletal disorders character-ized by a reduction in bone mass per unit of bone

volume. It may occur if the rate of bone resorption exceedsthat of bone formation, implying an uncoupling of thephases of bone remodeling.(1) The incidence of osteoporo-sis increases dramatically with age, becoming widely prev-alent in the elderly, in whom it has become a major publichealth problem.(2) However, the precise mechanism of thedisorder remains unresolved.Recent studies have demonstrated a role for cytokines in

the pathogenesis of osteoporosis. Interleukin-1 (IL-1),(3,4)

IL-6,(5,6) IL-11,(7) tumor necrosis factor-a (TNF-a),(8,9)

macrophage colony stimulating factor (M-CSF),(10) andgranulocyte macrophage colony stimulating factor(GM-CSF)(11) have been shown to stimulate bone resorp-tion. Such bone resorption cytokines are considered to playa pivotal role in the increased bone resorption in postmeno-pausal osteoporosis; IL-1, IL-6, TNF, M-CSF, and GM-CSF have been shown to be produced in greater abundancein the estrogen-deficient state,(12–17) resulting in the stim-ulation of osteoclastogenesis.Age-related osteoporosis affects the entire population of

aging men and women. The reason for this pathologicalstate is not well understood. Previous studies(18–20) havesuggested that humoral, nutritional, and metabolic changesin aging result in decreased bone formation as well asincreased bone resorption. Diabetes mellitus has also beenshown to be associated with osteopenia,(21–23) althoughconflicting findings have been reported, especially in pa-tients with non–insulin-dependent diabetes mellitus.(23–25)

*Part of this work was presented as an abstract in the 12thInternational Conference of Calcium Regulating Hormones, Mel-bourne, Australia, February 14–19, 1995.

1Department of Medicine III, Osaka University Medical School, Suita-City, Osaka, Japan.2Department of Pediatrics, Osaka University Medical School, Suita-city, Osaka, Japan.3The Third Department of Medicine, Nissei Hospital, Osaka-city, Osaka, Japan.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 12, Number 3, 1997Blackwell Science, Inc.q 1997 American Society for Bone and Mineral Research

439

Despite many investigations, there is no report regardingthe involvement of cytokines in the pathogenesis of age-and diabetes-related osteoporosis.Advanced glycation endproducts (AGEs) are the ulti-

mate products of nonenzymatic glycation of pro-teins.(26,27) The accumulation of AGEs on long-livedproteins have been shown to accelerate in advanced ag-ing and diabetes mellitus.(26,27) Therefore, AGEs arepostulated to be linked to the development of somecomplications in aging and diabetes. Several studies haveshown that AGE-modified proteins perturbed cellularfunctions, including cytokine and growth factor release inmacrophages,(28–30) increased permeability and proco-agulant activity in endothelial cells,(31) increased synthe-sis of extracellular matrix components in mesangialcells,(32) and induction of macrophage migration.(29) Thecellular interactions of AGEs have been shown to bemediated by the receptor for AGE (RAGE), a memberof the immunoglobulin superfamily.(33,34)

A recent study by Thomasek et al.(35) has demonstratedthat AGEs accumulated on collagen derived from corticalbones of diabetic or aged rats. This observation led us tohypothesize that AGEs, developed on collagenous matrix ofbone in diabetes and aging, could cause uncoupling of thephases of bone remodeling, resulting in osteoporosis. In thepresent investigation, we examined the effects of AGEs onthe production of the cytokines to stimulate bone resorp-tion by human bone-derived cells.

MATERIALS AND METHODS

Preparation of glucose-modified protein

AGE-bovine serum albumin (AGE-BSA) was preparedas described previously.(36) BSA (essentially fatty acid free;Sigma, St. Louis, MO, U.S.A.) was incubated with 250 mMglucose-6-phosphate (G-6-P) at 378C for 8 weeks in thepresence of 1.5 mM phenylmethylsulfonyl fluoride and 0.5mM EDTA in phosphate-buffered saline (PBS). Unincor-porated G-6-P was removed by extensive dialysis againstPBS. BSA was incubated in parallel without G-6-P, asunmodified proteins. The glucose-modified BSA, but notthe unmodified BSA, has the characteristics of AGE-pro-teins in terms of specific absorption and fluorescencespectra.(27)

Human bone-derived cell culture

Bone samples were obtained from four patients whounderwent surgery. The samples of normal cancellous bonewere collected following hip replacement surgery or maxil-lodental surgery. Bone specimens were minced into smallfragments and incubated with 2 mg/ml collagenase and 0.1mg/ml DNAse I for 2 h at 378C with vigorous shaking. Bonefragments were then cultured in Dulbecco’s modified Ea-gle’s medium (DMEM) supplemented with 10% fetal calfserum (FCS; Bioserum, Victoria, Australia), 15 mMHEPES, 50 mg/ml ascorbic acid, 100 U/ml penicillin, and 30mg/ml kanamycin sulfate.(37) The bone fragments were re-moved after bone cells migrated. The cells were grown to

confluency and subcultured using 30 mg/ml Actinase E(Kaken, Tokyo, Japan) digestion. All experiments wereconducted by using third to eighth passages of the cells.These cells were able to produce bone gla protein whenthey were treated with 1,25-dihydroxyvitamin D3 (data notshown).

Measurement of cytokines

Human bone-derived cells were plated onto a 24-wellplate (2 3 104 cells/well) in DMEM supplemented with10% FCS and cultured to be grown to a subconfluent state.After the cultures were washed with PBS, the cells weretreated for 48 h with BSA or AGE-BSA in 0.3 ml DMEMsupplemented with 0.1% FCS and 50 mg/ml ascorbic acid.Conditioned media were collected and centrifuged at 3000rpm for 5 minutes to remove any particulate material.Mouse calvaria-derived osteoblastic cell lines, MC3T3-

E1,(37) was obtained from the RIKEN Cell Bank (Tsukuba,Japan). These cells were plated onto a 24-well plate (5 3104 cells/well) in a-MEM containing 10% FCS and culturedto subconfluency. The cells were treated for 48 h with BSAor AGE-BSA in 0.3 ml a-MEM supplemented with 0.1%FCS. Conditioned media were collected as described above.Human IL-6 concentrations in the culture supernatants

were determined by enzyme-linked immunosolvent assaysfrom Genzyme (Cambridge, MA, U.S.A.). Human IL-1a,IL-1b, and IL-11 concentrations and mouse IL-6 concen-trations were measured by specific enzyme-linked immuno-solvent assays from R & D Systems (Minneapolis, MN,U.S.A.).

Alkaline phosphatase assay

Subconfluent bone-derived cells were treated withBSA or AGE-BSA for the indicated periods. After theincubation, the cells were washed three times with phys-iological saline and lysed in 0.1% Triton X-100 withsonication. Alkaline phosphatase activity in the lysatewas determined spectrophotometrically using p-nitrophe-nol phosphate (PNP) (Sigma Chemical Co., St. Louis,MO, U.S.A.) as substrate, as described previously.(38)

The enzyme activity was expressed as nanomoles of PNPper 30 minutes per micrograms of DNA. DNA contentwas determined on aliquots of the cell lysates using afluorometric method.(39)

Northern blot analysis and reverse transcriptasepolymerase chain reaction

Total RNA was extracted from bone-derived cells bythe acid guanidinium thiocyanate-phenol-chloroform ex-traction method.(40) Heat-denatured RNA was electro-phoresed (10 mg/lane) in 1% agarose-2% formaldehydegels. The gels were transferred to nylon membranes (Hy-bond N1; Amersham, Buckinghamshire, U.K.), and themembranes were baked at 808C for 2 h. A 32P-labeledcDNA probe for human IL-6(41) was prepared by a ran-dom priming method.(42) Hybridization was carried outusing rapid hybridization buffer (Amersham) according

440 TAKAGI ET AL.

to the manufacture’s protocol. The same filters wererehybridized with the 32P-labeled cDNA probe for humanglyceraldehyde-3 phosphate dehydrogenase (G3PDH; Clon-tech, Palo Alto, CA, U.S.A.).In reverse transcriptase polymerase chain reaction

(RT-PCR) analysis, total RNA (1 mg) from human bone-derived cells was used as a template for cDNA synthesis ina 30-ml volume containing the following reagents: 0.5 mMdNTP (Pharmacia, Piscataway, NJ, U.S.A.); 1.25 U/ml oligo(dT); 100 mg/ml BSA; 4 U/ml RNAse inhibitor (Promega,Madison, WI, U.S.A.); 20 U/ml Moloney murine leukemiavirus reverse transcriptase (GIBCO BRL, Gaithersburg,MD, U.S.A.); 10 mM dithiothreitol; 3 mM MgCl2; 75 mMKCl; and 50 mM Tris-HCl (pH 8.3). The reaction wasincubated at 378C for 60 minutes. The PCR was carried outat a concentration of 13 PCR buffer (10 mM Tris-HCl, pH8.3, 50 mM KCl, 1.5 mM MgCl2), 0.2 mM dNTP, 0.8 mMeach of 59 and 39 primers, and 0.025 U/ml Taq polymerase(Perkin-Elmer Cetus, Norwalk, CT, U.S.A.). Analysis of thetranscripts for RAGE was performed using the followingamplification profile: a denaturation step at 968C for 1minute, annealing at 558C for 45 s, and extension at 728C for2 minutes for 30 cycles. The primers used were those de-scribed previously.(36) The PCR-amplified products wereseparated on an agarose gel and hybridized with 32P-labeledoligonucleotide probe (59-AACCGTAACCCTGACCTGTGAAGTC-39) homologous to the middle region of theexpected RAGE amplification product. The PCR productswere sequenced to confirm their identity by the dideoxychain termination method using an Applied Biosystems373A automated DNA sequencer (Applied Biosystems,Foster City, CA, U.S.A.).

Nuclear extraction and electrophoreticmobility-shift assay

MC3T3-E1 cells were treated with BSA or AGE-BSAin the serum-free medium, and the nuclear extracts wereprepared by the methods of Schreiber et al.(43) Electro-phoretic mobility-shift assay (EMSA) was performed us-ing a gel shift assay kit from Stratagene (La Jolla, CA,U.S.A.). The sequence of the double-stranded oligonu-cleotides used to detect the DNA binding activitiesof nuclear factor-kB (NF-kB) was 59-GATCGAGGGGACTTTCCCTAGC 39.(44) Five micrograms of nuclearproteins were incubated with 500 pg of 32P-labeled oli-gonucleotides for 30 minutes at room temperature, andthe samples were loaded onto 5% nondenaturing acryl-amide gels, according to the manufacture’s protocol. Insome experiments, nuclear extracts were incubated witheither unlabeled oligonucleotides or rabbit polyclonalIgG against p50 or p65 (Santa Cruz Biotechnology, SantaCruz, CA, U.S.A.) before the incubation with 32P-labeledoligonucleotides. After electrophoresis, gels were ex-posed to X-ray films (X-Omat, Kodak, Rochester, NY,U.S.A.).

Statistics

All values were shown as means 6 SE. When the signif-icant difference was discussed, unpaired Student’s t-test orWelch’s t-test was employed.

RESULTS

Effects of AGEs on the production of cytokines byhuman bone-derived cells

Human bone-derived cells were cultured with BSA orAGE-BSA (1000 mg/ml) in the presence of 0.1% FCS for48 h, and the concentrations of IL-6, IL-a, IL-1b, and IL-11in the culture supernatants were determined. Table 1 showsthat the bone-derived cells released significant amounts ofIL-6 and IL-11 in this experimental condition. The treat-ment with AGE-BSA increased the IL-6 concentrations to2.7-fold above that with BSA. In contrast, AGE-BSA in-duced no significant change in the IL-11 concentrations inthe culture supernatants. Neither IL-1a nor IL-1b concen-trations in the culture supernatants were at detectable lev-els (,1.0 pg/ml) even when the bone-derived cells weretreated with or without AGE-BSA. As shown in Fig. 1,AGE-BSA increased the IL-6 concentrations in the culturesupernatants of the bone-derived cells in a dose-dependentmanner. In this experiment, the concentrations of BSA(including AGE-BSA and unmodified BSA) were kept at1000 mg/ml by adjustment with unmodified BSA. AGE-BSA, at concentrations of 500 and 1000 mg/ml, induced asignificant increase in the IL-6 concentrations.To examine whether AGE-BSA increases the levels of

IL-6 mRNA in the human bone-derived cells, Northern blotanalysis was employed (Fig. 2). The treatment with AGE-BSA for 24 h increased the IL-6 mRNA levels which wasdependent on the amounts of AGE-BSA added. By con-trast, expression of G3PDH mRNA appeared not to beaffected by AGE-BSA.Next, the effects of AGE-BSA on IL-6 production were

examined in the bone-derived cells from four patients. Asshown in Table 2, the production of IL-6 was detected inall the bone-derived cells, although its concentrationswere variable. The treatment with AGE-BSA consis-

TABLE 1. EFFECTS OF AGES ON IL-6, IL-1a, IL-1b, andIL-11 Release by Human Bone-Derived Cells

TreatmentIL-6(pg/ml)

IL-1a(pg/ml)

IL-1b(pg/ml)

IL-11(pg/ml)

BSA 4830 6 383 ,1.0 ,1.0 160 6 10.0AGE-BSA 12900 6 1300* ,1.0 ,1.0 133 6 3.3 (NS)

Human bone-derived cells were treated for 48 h with 1000 mg/mlBSA or AGE-BSA in DMEM supplemented with 0.1% FCS and 50mg/ml ascorbic acid. IL-6, IL-1a, IL-1b, and IL-11 concentrationswere determined by specific enzyme-linked immunosorbent assays.Values represent means 6 SE in triplicate assay. NS, no significantdifference; * p , 0.001, significant difference when compared withBSA.

GLUCOSE-MODIFIED PROTEINS AND IL-6 PRODUCTION 441

tently increased the IL-6 concentrations about 2- to3-fold in the culture supernatants from each of the cells.

Effects of AGEs on alkaline phosphatase activity andDNA contents of human bone-derived cells

In the next experiments, we examined the effects ofAGEs on DNA contents and alkaline phosphatase activity

in human bone-derived cells. The treatment withAGE-BSA for 48 h had no apparent effect on DNA con-tents per well in each of the bone-derived cells from fourpatients, compared with the BSA treatment. In addition,the AGE-BSA treatment also exerted no significant effecton alkaline phosphatase activity in those cells. Exposure ofthe bone-derived cells to AGE-BSA for longer periods of 8days, also had no effects on alkaline phosphatase activityand DNA contents (data not shown).

Expression of RAGE mRNA in humanbone-derived cells

Since various biological effects of AGEs have beenshown to be mediated by RAGE,(33,34) we examinedwhether human bone-derived cells express mRNA forRAGE. RT-PCR reaction using specific primers for humanRAGE(36) was followed by hybridization with 32P-labeledoligonucleotide probe homologous to the middle region ofthe expected amplified product. The result demonstratedappearance of the amplified fragment of 480 bp. There wasno amplified fragment by direct PCR of total RNA withoutRT (Fig. 3). Next, the 480 bp fragment was sequenced bythe dideoxy chain termination method. It revealed that thenucleotide sequence of the RT-PCR products was com-pletely concordant with the reported sequence of humanRAGE.(34) Thus, the PCR products obtained were found tobe specific for human RAGE mRNA.

Effects of AGEs on activation of NF-kB in mouseMC3T3-E1 cells

Recent observations(45,46) have demonstrated that AGEsinduced activation of the transcription nuclear factorNF-kB in endothelial cells. The NF-kB binding site is iden-tified within the promoter region of the IL-6 gene, which isconsidered to be important for the transcriptional activa-

FIG. 1. Dose response of the effects of AGE-BSA on IL-6concentrations in the culture supernatants from humanbone-derived cells. The cells were cultured in DMEM–0.1% FCS containing 50 mg/ml ascorbic acid with variousconcentrations of AGE-BSA for 48 h. In this experiment,the concentrations of BSA (including AGE-BSA and un-modified BSA) were always set at 1000 mg/ml. After theincubation, IL-6 concentrations in the culture supernatantswere determined. Bars represent means 6 SE in triplicateassay. *p , 0.001, significant difference from cultures nottreated with AGE-BSA.

FIG. 2. Northern blot analyses of IL-6 mRNAs in humanbone-derived cells. The cells were treated for 24 h withvarious concentrations of AGE-BSA. Total RNA (10 mg)was electrophoresed, transferred to nylon membranes, andhybridized to specific probes for human IL-6 and G3PDH.Arrows indicate the positions of the mRNAs for IL-6 andG3PDH. The data are typical of four experiments.

TABLE 2. EFFECTS OF AGES ON IL-6 RELEASE BY BONE-DERIVED CELLS FROM FOUR PATIENTS

Patient

IL-6

BSA(pg/ml)

AGE-BSA(pg/ml)

A 1530 6 163 3240 6 154*B 1433 6 49 3680 6 334*C 620 6 17 1090 6 16*D 267 6 8.7 656 6 12*

Human bone-derived cells were obtained from surgical samplesof four patients. The cells used in these experiments were rangedbetween third and eighth passages. Cells, 23 104, were plated onto24-well plates in DMEM–10% FCS and were cultured to be grownto subconfluency. The subconfluent cells were treated for 48 h with500 mg/ml BSA or AGE-BSA in DMEM supplemented with 0.1%FCS and 50 mg/ml ascorbic acid. Values indicate means 6 SE intriplicate assay. NS, no significant difference; * p , 0.001, signifi-cant difference when compared with BSA.

442 TAKAGI ET AL.

tion of the gene by IL-1, TNF, or lipopolysaccharide.(47)

Thus, we examined whether AGEs also activate NF-kB inbone-derived cells. For its purpose, mouse osteoblasticMC3T3-E1 cells were treated with BSA or AGE-BSA. Theexperiments revealed that the IL-6 concentrations in the

culture supernatants of these cells were 50 6 1.5 pg/ml,which were significantly ( p , 0.001) increased to 87 6 2.7pg/ml by the treatment with AGE-BSA. Therefore, EMSAwas performed using the nuclear extracts from MC3T3-E1cells and the oligonucleotide probe specific for NF-kB bind-ing. The results showed enhanced intensity of gel-retardedbands when the cells were treated for 4 h with AGE-BSA,compared with BSA (Fig. 4, lanes 1 and 2). The gel-re-tarded bands were specific for NF-kB, since it disappearedin the presence of excess unlabeled oligonucleotides (Fig. 4,lane 3) but not unrelated oligonucleotides (Fig. 4, lane 4).Addition of anti-p50 antibody or anti-p65 antibody partiallysupershifted the gel-retarded bands, in which the supershiftby anti-p50 antibody was minor compared with that inducedby anti-p65 antibody (Fig. 4, lanes 5 and 6). Addition of theboth antibodies supershifted the retarded bands completely(Fig. 4, lane 7).

DISCUSSION

AGEs have been shown to exert biological activities onvarious kinds of cells, such as macrophages, endothelialcells, mesangial cells, renal cell carcinoma cells, and fibro-blasts.(28–33,36,48) In the present study, we demonstrated forthe first time that AGEs enhanced the production of thebone resorption factor IL-6 in normal human bone-derivedcells. AGEs also increased the levels of IL-6 mRNA inthese cells. By contrast, AGEs did not influence the pro-duction of another bone resorption cytokine, IL-11, in thebone-derived cells. The concentrations of IL-1a and IL-1bin the culture supernatants were undetectable (,1.0 pg/ml),even when the cells were incubated either with or withoutAGE-BSA. TNF-a mRNA was not detected in the bone-derived cells by RT-PCR analysis (our unpublished obser-vation), which is consistent with the previous report.(49)

Thus, among various cytokines examined that are knownto stimulate bone resorption, both IL-6 and IL-11 areproduced in significant amounts by human bone-derivedcells, but only IL-6 production is up-regulated by AGEs.In this relation, Girasol et al.(7) have shown that IL-11 isessential for osteoclastogenesis in general, whereas IL-6is important for osteoclastogenesis in an estrogen-defi-cient state.It remains obscure how the signals of AGEs are conveyed

into the cells. Several studies(33,34,45,46) have revealed thatRAGE has a central role in mediating the interactions ofAGEs with cellular surfaces. In the present investigation,we showed for the first time that RAGE mRNA is ex-pressed in human bone-derived cells. Therefore, it is pos-sible that the binding of AGEs with the cellular surfaceRAGE leads to the expression of IL-6 genes in these cells.Recently, it has been shown that AGEs induce oxidantstress and thereby activate the transcription factor NF-kB invascular endothelium.(45,46) In this relation, we demon-strated that AGEs enhanced activation of NF-kB in mouseosteoblastic cells. EMSA using antibodies against p50(NFKB1) or p65 (Rel A) revealed that at least these twoproteins were involved in the NF-kB activation in the cells.It is also possible that other NF-kB/Rel family proteins,

FIG. 3. Identification of mRNA for receptor for AGEs(RAGE) in human bone-derived cells by RT-PCR. TotalRNA was isolated from the bone-derived cells. RNA (1 mg)was reverse-transcribed into cDNA. PCR reactions wereperformed using specific primers for human RAGE. ThePCR products were run on 1.5% agarose gel and hybridizedwith 32P-labeled oligonucleotide probe homologous to themiddle region of the expected 480 bp RAGE amplifiedproduct. RT, reaction with reverse transcriptase. The pre-dicted length of the PCR product (480 bp) is shown.

FIG. 4. Effects of AGE-BSA on DNA binding activity ofNF-kB. Nuclear proteins were extracted from MC3T3-E1cells treated with 1000 mg/ml of BSA (lane 1) or AGE-BSA(lanes 2–7) for 4 h. EMSA using the oligonucleotide probespecific for NF-kB binding was performed on nuclear ex-tracts preincubated with no reagents (lanes 1 and 2), 50-foldexcess of unlabeled NF-kB probe (lane 3), 50-fold excess ofunrelated oligonucleotides (lane 4), anti-p50 antibody (lane5), anti-p65 antibody (lane 6), or both anti-p50 and anti-p65antibodies (lane 7).

GLUCOSE-MODIFIED PROTEINS AND IL-6 PRODUCTION 443

such as p52 (NFKB2), c-Rel, and Rel B, participate. TheNF-kB binding site in the promoter region of the IL-6 genehas been shown to be important for the transcriptionalregulation of the IL-6 gene,(47) and in vivo targeting of p50subunit resulted in reduced expression of the IL-6 gene.(50)

Thus, AGEs may stimulate IL-6 gene transcription in os-teoblastic cells by activating NF-kB. In contrast, the IL-11gene does not contain an NF-kB binding site in its promoterregion.(51) This seems to be consistent with our results thatAGEs did not alter the concentrations of IL-11 in theculture supernatants from the bone-derived cells.In age- and diabetes-related osteoporosis, decreased

bone formation(18,24) as well as increased bone resorp-tion(20,52) have been observed. In the present study, wefailed to show that AGEs had direct influence on DNAcontents and alkaline phosphatase activity of the humanbone-derived cells. In this respect, AGEs appeared not toalter the proliferation and differentiation of the bone-de-rived cells. It is also possible that IL-6, whose release isstimulated by AGEs, has some effects on these cells in vivo,since soluble IL-6 receptor present in sera is able to coop-erate with IL-6 in activating a gp130-mediated pathway.(53)

Alternatively, there is a possibility that AGEs exert theirinhibitory effect on bone formation by indirect mechanisms.Recently, Fong et al. demonstrated that formation of AGEson bone matrix inhibits the matrix-induced bone differen-tiation, probably involving alterations of binding sites forextractable proteins with bone inductive properties such asbone morphogenetic protein-2.(54)

It has been suggested that multifactorial changes arerelated to osteoporosis in advanced aging and diabetesmellitus.(18–25) AGEs accumulate on long-lived extracellu-lar matrix proteins in aging and diabetes.(26,27) It has beendemonstrated that the levels of collagen-linked fluores-cence in cortical bones were increased in aged and diabeticrats, as a result of the formation of AGEs.(35) In the presentinvestigations, we have not directly demonstrated thatAGEs induce bone resorption. However, the current studyimplies that AGEs formed on bone matrix proteins wouldinduce the osteoblasts to produce IL-6, which in turn maystimulate osteoclastogenesis and thereby bone resorption,leading to osteoporosis unless bone formation is efficientlycoupled. Thus, it introduces a potential mechanism for age-and diabetes-related osteoporosis, in which AGEs are cru-cial in triggering an uncoupling event in bone remodelingsystems.

ACKNOWLEDGMENTS

We thank Dr. Yuko Takagaki (Kanagawa Dental Col-lege, Yokosuka, Japan) and Dr. Kazuo Hiroshima (OsakaNational Hospital, Osaka, Japan) for providing us withnormal human bone samples. We also thank Keiko Tsujiifor her excellent secretarial assistance in preparing themanuscript. This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education,Science, and Culture of Japan and by the Enami MemorialFoundation for Diabetes Research.

REFERENCES

1. Riggs BL, Melton LJ III 1986 Involutional osteoporosis. NewEngl J Med 314:1676–1686.

2. Tohme JF, Silverberg SJ, Lindsay R 1990 Osteoporosis. In:Becker KL (ed.) Principles and Practice of Endocrinology andMetabolism. Lippincott/East Washington Square, Philadel-phia, PA, U.S.A., pp 491–504.

3. Gowen M, Wood DD, Ihrie EJ, McGuire MKB, Russell RGG1983 An interleukin-1-like factor stimulates bone resorption invitro. Nature 306:378–380.

4. Boyce BF, Aufdemorte TB, Garrett IR, Yates AJP, Mundy GR1989 Effects of interleukin-1 on bone turnover in normal mice.Endocrinology 125:1142–1150.

5. Ishimi Y, Miyaura C, Jin CH, Akatsu T, Abe E, Nakamura Y,Yamaguchi A, Yoshiki S, Matsuda T, Hirano T, Kishimoto T,Suda T 1990 IL-6 is produced by osteoblasts and induces boneresorption. J Immunol 145:3297–3303.

6. Black K, Garrett IR, Mundy GR 1991 Chinese hamster ovariancells transfected with the murine interleukin-6 gene causehypercalcemia as well as cachexia, leukocytosis and throm-bocytosis in tumor-bearing nude mice. Endocrinology128:2657–2659.

7. Girasole G, Passeri G, Jilka RL, Manolagas SC 1994 Interleu-kin-11: A new cytokine critical for osteoclast development.J Clin Invest 93:1516–1524.

8. Bertolini DR, Nedwin GE, Bringman TS, Smith DD, MundyGR 1986 Stimulation of bone resorption and inhibition of boneformation in-vitro by human tumor necrosis factors. Nature319:516–518.

9. Johnson RA, Boyce BF, Mundy GR, Roodman GD 1989Tumors producing human tumor necrosis factor induced hy-percalcemia and osteoclastic bone resorption in nude mice.Endocrinology 124:1424–1427.

10. Tanaka S, Takahashi N, Udagawa N, Tamura T, Akatsu N,Stanley ER, Kurokawa T, Suda T 1993 Macrophage colony-stimulating factor is indispensable for both proliferation anddifferentiation of osteoclast progenitors. J Clin Invest91:257–263.

11. Schneider GB, Relfson M 1989 Pluripotent hemopoietic stemcells give rise to osteoclasts in vitro: Effect of rGM-CSF. BoneMiner 5:129–138.

12. Pacifici R 1996 Review: Estrogen, cytokines, and pathogenesisof postmenopausal osteoporosis. J Bone Miner Res11:1043–1051.

13. Pacifici R, Rifas L, Titelbaum S, Slatopolsky E, McCracken R,Bergfeld M, Lee W, Avioli LV, Peck WA 1987 Spontaneousrelease of interleukin 1 from human blood monocytes reflectsbone formation in idiopathic osteoporosis. Proc Natl Acad SciUSA 84:4616–4620.

14. Jilka RL, Hangoc G, Girasole G, Passeri G, Williams DC,Abrams JS, Boyce B, Broxmeyer H, Manolagas SC 1992 In-creased osteoclast development after estrogen loss: mediationby interleukin-6. Science 257:88–91.

15. Pacifici R, Brown C, Rifas L, Avioli LV 1990 TNF-a andGM-CSF secretion from human blood monocytes: Effect ofmenopause and estrogen replacement. J Bone Miner Res 5:145(abstract).

16. Pacifici R, Vannice JL, Rifas L, Kimble RB 1993 Monocyticsecretion of interleukin-1 receptor antagonist in normal andosteoporotic women: Effect of menopause and estrogen/pro-gesterone therapy. J Clin Endocrinol Metab 77:1135–1141.

17. Pacifici R, Brown C, Puscheck E, Friedrich E, Slatopolsky E,Maggio D, McCracken R, Avioli LV 1991 Effect of surgicalmenopause and estrogen replacement on cytokine release from

444 TAKAGI ET AL.

human blood mononuclear cells. Proc Natl Acad Sci USA88:5134–5138.

18. Lips P, Courpron P, Meunier P 1978 Mean wall thickness oftrabecular bone packets in human iliac crest: Changes with age.Calcif Tissue Res 26:13–17.

19. Gallagher JC, Riggs BL, Eisman J, Hamstra A, Aranaud SB,DeLuca HF 1979 Intestinal calcium absorption and serumvitamin D metabolites in normal subjects and osteoporoticpatients: Effect of age and dietary calcium. J Clin Invest64:729–736.

20. Delmas PD, Stenner D, Wahner HW, Mann KG, Riggs BL1983 Increase in serum bone g-carboxyglutamic acid proteinwith aging in women. J Clin Invest 71:1316–1321.

21. Levin ME, Boisseau VC, Avioli LV 1976 Effects of diabetesmellitus on bone mass in juvenile and adult-onset diabetes.New Engl J Med 294:241–245.

22. Auwerx J, Dequeker J, Bouillon R, Geusens P, Nijs J 1988Mineral metabolism and axial skeleton in diabetes mellitus.Diabetes 37:8–12.

23. Ishida H, Senio Y, Matsukura S, Ikeda M, Yawata M,Yamashita G, Ishizuka S, Imura H 1985 Diabetic osteopeniaand circulating levels of vitamin D metabolites in type 2 (non-insulin-dependent) diabetes. Metabolism 43:797–801.

24. Barrett-Connor E, Holbrook TL 1992 Sex differences in osteo-porosis in older adults with non-insulin-dependent diabetesmellitus. JAMA 268:3333–3337

25. van Dale PLA, Stolk RP, Bueger H, Algra D, Grobbee DE,Hofman A, Birkenhager JC, Pols AP 1995 Bone density innon-insulin-dependent diabetes mellitus: The Rotterdamstudy. Ann Intern Med 122:409–414.

26. Brownlee M, Cerami A, Vlassara H 1988 Advanced glycosyl-ation end products in tissue and the biochemical basis ofdiabetic complications. New Engl J Med 318:1315–1320.

27. Monnier VM, Kohn RR, Cerami A 1984 Accelerated age-related browning of human collagen in diabetes mellitus. ProcNatl Acad Sci USA 81:583–587.

28. Vlassara H, Blownlee M, Manogue KR, Dionarello CA, Pasa-gian A 1988 Cachectin/TNF and IL-1 induced by glucose-modified proteins: role in normal tissue remodeling. Science240:1546–1548.

29. Kirstein M, Brett J, Radoff S, Ogawa S, Stern D, Vlassara H1990 Advanced protein glycosylation induces transendothelialhuman monocyte chemotaxis and secretion of PDGF: Role invascular disease of diabetes and aging. Proc Natl Acad Sci USA87:9010–9014.

30. Kirstein M, Aston C, Hintz R, Vlassara H 1992 Receptor-specific induction of insulin-like growth factor I in humanmonocytes by advanced glycosylation end product-modifiedproteins. J Clin Invest 90:439–446.

31. Esposito C, Gerlach H, Brett J, Stern D, Vlassara H 1989Endothelial receptor-mediated binding of glucose-modified al-bumin is associated with increased monolayer permeability andmodulation of cell surface coagulant properties. J Exp Med170:1387–1407.

32. Doi T, Vlassara H, Kirstein M, Yamada Y, Striker GE, StrikerLJ 1992 Receptor-specific increase in extracellular matrix pro-duction in mouse mesangial cells by advanced glycosylationendproducts is mediated via platelet-derived growth factor.Proc Natl Acad Sci USA 89:2873–2877.

33. Schmidt AM, Yan SD, Brett J, Mora R, Nowygrod R, Stern D1993 Regulation of human mononuclear phagocytic migrationby cell surface-binding proteins for advanced glycosylation endproducts. J Clin Invest 92:2155–2168.

34. Neeper M, Schmidt AM, Brett J, Yan SD, Wang F, Pan Y-CE,Elliston K, Stern D, Shaw A 1992 Cloning and expression of a

cell surface receptor for advanced glycosylation end productsof proteins. J Biol Chem 267:14998–15004.

35. Tomasek JJ, Meyers SW, Basinger JB, Green DT, Shew RL1994 Diabetic and age-related enhancement of collagen-linkedfluorescence in cortical bones of rats. Life Sci 55:855–861.

36. Miki S, Kasayama S, Miki Y, Nakamura Y, Yamamoto M, SatoB, Kishimoto T 1993 Expression of receptors for advancedglycosylation end products of renal cell carcinoma cells in vitro.Biochem Biophys Res Commun 196:984–989.

37. Sudo H, Kodama H-A, Amagai Y, Yamamoto S, Kasai S 1983In vitro differentiation and calcification in a new clonal osteo-genic cell line derived from new born mouse calvaria. J CellBiol 96:191–198.

38. Yamamoto T, Ecarot B, Glorieux FH 1991 In vivo osteogenicactivity of isolated human bone cells. J Bone Miner Res6:45–51.

39. Labarca C, Paigen K 1980 A simple, rapid and sensitive DNAassay procedure. Anal Biochem 102:344–352.

40. Chomczynski P, Sacchi N 1987 Single-step method of RNAisolation by acid guanidium thiocyanate-phenol-chloroform ex-traction. Anal Biochem 162:156–159.

41. Hirano T, Yasukawa K, Harada H, Taga T, Watanabe Y,Matsuda T, Kashiwamura S, Nakajima K, Koyama K,Iwamatsu K, Tsunasawa S, Sakiyama F, Matsui H, Takahara T,Taniguchi T, Kishimoto T 1986 Complementary DNA for anovel human interleukin (BSF-2) that induce B lymphocytes toproduce immunoglobulins. Nature 324:73–76.

42. Feinberg AP, Vogelstein B 1983 A technique for radiolabelingDNA restriction endonuclease fragments to high specific ac-tivity. Anal Biochem 132:6–13.

43. Schreiber E, Matthias P, Muller MM, Schaffner W 1989 Rapiddetection of octamer binding proteins with ‘mini-extracts,’ pre-pared from a small number of cells. Nucleic Acids Res 17:6419.

44. Shirakawa F, Chedid M, Suttles J, Pollok BA, Mizel SB 1989Interleukin 1 and cyclic AMP induce k immunoglobulin light-chain expression via the activation of an NF-kB-like DNA-binding protein. Mol Cell Biol 9:959–964.

45. Wautier J-L, Wautier M-P, Schmidt A-M, Anderson GM, HoriO, Zoukourian C, Capron L, Chappey O, Yan S-D, Brett J,Guillausseu P-J, Stern D 1994 Advanced glycation end prod-ucts (AGEs) on the surface of diabetic erythrocytes bind to thevessel wall via a specific receptor inducing oxidant stress in thevasculature: a link between surface-associated AGEs and dia-betic complications. Proc Natl Acad Sci USA 91:7742–7746.

46. Schmidt AM, Hori O, Chen JX, Li JF, Crandall J, Zhang J,Cao R, Yan SD, Brett J, Stern D 1995 Advanced glycationendproducts interacting with their endothelial receptor induceexpression of vascular cell adhesion molecule-1 (VCAM-1) incultured human endothelial cells and in mice: A potentialmechanism for the accelerated vasculopathy of diabetes. J ClinInvest 96:1395–1403.

47. Shimizu H, Mitomo K, Watanabe T, Okamoto S, YamamotoK-I 1990 Involvement of a NF-kB like transcription factor inthe activation of the interleukin-6 gene by inflammatory lym-phokines. Mol Cell Biol 10:561–568.

48. Vlassara H, Bucala R, Striker L 1994 Pathogenic effects ofadvanced glycosylation: biochemical, biologic, and clinical im-plications for diabetes and aging. Lab Invest 70:138–151.

49. Brich MA, Ginty AF, Walsh CA, Fraser WD, Gallagher JA,Bilbe C 1993 PCR detection of cytokines in normal human andPagetic osteoblast-like cells. J Bone Miner Res 8:1155–1162.

50. Sha WC, Liou H-C, Tuomanen EI, Baltimore D 1995 Targeteddisruption of the p50 subunit of NF-kB leads to multifocaldefects in immune responses. Cell 80:321–330.

51. McKinley D, Wu Q, Yang-Feng T, Yang Y-C 1992 Genomic

GLUCOSE-MODIFIED PROTEINS AND IL-6 PRODUCTION 445

Sequence and chromosomal location of interleukin-11 gene(IL-11). Genomic 13:814–819.

52. Mathiassen B, Nielsen S, Johansen JS, Hartwell D, Ditzel J,Rødblo P, Christiansen C 1990 Long-term bone loss in insulin-dependent diabetic patients with microvascular complications.J Diabet Complications 4:145–149.

53. Tamura T, Udagawa N, Takahashi N, Miyaura C, Tanaka S,Yamada Y, Koishihara Y, Ohsugi Y, Kumaki K, Taga T,Kishimoto T, Suda T 1993 Soluble interleukin-6 receptor trig-gers osteoclast formation by interleukin 6. Proc Natl Acad SciUSA 90:11924–11928.

54. Fong Y, Edelstein D, Wang EA, Brownlee M 1993 Inhibition

of matrix-induced bone differentiation by advanced glycationend-products in rats. Diabetologia 36:802–807.

Address reprint requests to:Soji Kasayama, M.D.

Department of Medicine IIIOsaka University Medical School

2-2 Yamada-okaSuita-City, Osaka 565, Japan

Received in original form August 12, 1996; in revised form October7, 1996; accepted October 24, 1996.

446 TAKAGI ET AL.