Interactions between the Bone Matrix Proteins Osteopontin and Bone

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Vol. 268, No. 13, Issue of May 5 , pp.Printed 9901-9907,1993 in U.S.A.

Interactions between the Bone Matrix Proteins Osteopontin and Bone Sialoprotein and the Osteoclast Integrin CY& Potentiate Bone Resorption* (Received for publication, May 6, 1992)

F. Patrick Ross*§, Jean Chappel*, Jose I. AlvarezS, Diane Sanderll, William T. Butler 11, Mary C. Farach-Carson11, Keith A. Mintz**,Pamela Gehron Robey**, Steven L. TeitelbaumS, and David A. Chereshll From the $Department of Pathology and Laboratory Medicine, Jewish Hospital, Washington University Medical Center, St. Louis,Missouri 631 20,the llDepartment of Immunology, Scripps Research Institute, La Jolla, California 92037, the 1) Department of Biological Chemistry, Dental Branch, Universityof Texas Health Science Center, Houston, Texas 77225, and the **Bone Research Branch, National Institute of Dental Research, National Institutes of Health, Bethesda, Maryland 20892

tantly, an antibody toa,& blocks formation of characteristic bone resorptive pits by human osteoclasts (12). The specific matrix proteins to which this or other integrins bind when initiating resorption are unknown, but immunoelectron microscopic studies of whole bone have shown that osteoclast a& and the matrix protein osteopontin colocalize (13), suggesting that their interactionmay represent a mechanism by which the cell binds to bone. Regardless of the molecule to which it attaches, a& recognizes a common tripeptide sequence, RGD (Arg-Gly-Asp), which has been shown to be crucial for specific binding ( 5 , 9, 14) and which is present in the bone matrix proteins bone sialoprotein, osteopontin, thrombospondin, type 1 collagen, and fibronectin (15-17). The latter two proteins are poten(I1 integrins (8,18), tially significant, since osteoclasts express which are known to be involved in attachment to fibronectin and collagen (19), suggesting that such interactions may also play a role in the resorptive process. In this report we investigate the identity of bone matrix proteins critical to resorption and the osteoclast integrins with which they interact. We show that while both PI and (I3 integrins are present on chickenosteoclasts, the cells bind to intactbonematrix via a&, using osteopontinand bone Bone resorption is a multistep process mediated by the sialoprotein for attachment. Osteoclasts do not attach signifosteoclast, a multinucleated cell of the monocyte/macrophage icantly to themajor matrix protein collagen, suggesting that, if this cell while abundant, it does not initiate the resorptive process. lineage (1,Z). Recent studies have established that is todegrade matrix it mustdevelop a tightly sealed microen- Most importantly, a polyclonal antibody to osteopontin invironment at the osteoclast-matrix interface inwhich a steep hibits attachment of osteoclasts to both the isolated protein proton gradient is established and where matrix-degrading and whole bone, providing the first direct evidence for the enzymes are active(3). functional significance of such interactions. Generation of such a highly defined compartment is believed to be mediatedby interactions betweenspecific matrix MATERIALS ANDMETHODS proteins and integrins, which are heterodimeric membrane Attachment Proteins-Chicken fibronectin was obtained commerreceptors (4-6). Osteoclasts are known to express at least onecially (BiomedicaI Technologies, Stoughton, MA), while rat tail colsuch receptor, a& (7, 8) which recognizes a number of ad- lagen and rat albumin were purchased from Sigma. Purified type 1 hesive proteins including vitronectin, fibrinogen, thrombos- chicken collagen was a generous gift from Dr. R. Mayne, Department pondin,and von Willebrand’s factor (9-11). Mostimporof Cell Biology, University of Alabama, Birmingham, AL. Rat bone

We have investigated the mechanism by which osteoclasts adhere to and resorb bone. We show that these cells expressB1 and B3 integrins which are involvedin attachment to purified bone matrix proteins. Binding to osteopontin and bone sialoprotein is mediated by a,.&, while a B1 integrin is responsible for attachment to fibronectin. Both the rapid attachment by osteoclasts to intact bone particles and their subsequent resorption are blocked by a monoclonal antibody directed to the a& complex but not by an antibody against 81 integrins. Attachment of osteoclasts to bone is also inhibited withsoluble osteopontin, Arg-Gly-Asp-containing peptides derived from both osteopontin and bone sialoprotein, or a monospecific polyclonal antibody against osteopontin. We conclude that both osteoclast adherence to bone and subsequent resorption of its matrix are dependent on interactions between the bone matrix proteins osteopontin and/or bone sialoprotein and the integrin aVBs. Moreover, collagen, which constitutes 90% of its organic matrix, is minimally involved in binding of chicken osteoclasts to bone.

* This work

wassupported by National Institutes of Health Grants CA47526 and CA50286 (to D. A. C.), AR39273 (to W. T. B.), and DE05413 and Shriners Hospital Grant 15960 (to S. L. T.). The costs of publication of this article were defrayed in partby the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18U.S.C. Section 1734 solely to indicate this fact. I To whom all correspondence should be addressed Dept. of Pathology and Laboratory Medicine, Jewish Hospital, Washington University Medical Center, 216 South Kingshighway, St. Louis, MO 63110. Tel.: 314-454-8463:Fax: 314-454-5505,

matrix protein fractions were separated by minor modifications of a published method (20), and thefractions were characterized by SDSPAGE’ and staining with either Coomassie Blue or Stains-all (21). The procedure yielded samples highly enriched for osteocalcin, solubilized bone collagen, osteonectin, bone sialoprotein, and a mixture of proteoglycans. Rat osteopontin and a protein shown to be the rat homologue of human a2HS-glycoprotein were purified by published methods (22, 23). Pure bone sialoprotein from the osteosarcoma cell line ROS 17/2.8 was purified as described and donated by Dr. Ron The abbreviations used are: PAGE, polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline.

9901

9902

Interactions Osteoclast Integrin-mediated Matrix

Midura, National Institutes of Health (24). Synthetic peptides, containing the RGD sequences in human osteopontin and bone sialoprotein and purified to homogeneity by high performance liquid chromatography, were provided by Dr. Frank Robey, National Institutes of Health. The sequence of these peptides was as follows: human osteopontin, TVDTYDGRGDSVVYGLRSKS; human bone sialoprotein, YESENGEDRGDNYRAYED. Preparation and Use of Osteoclasts-Chicken osteoclasts were prepared as described (25). Briefly, laying hens were maintained on a calcium-free diet (Purina, St. Louis) for 3-4 weeks, at which time bone marrow was harvested. The mononuclear cells were isolated by Ficoll-hypaque gradient centrifugation, and the 24-h adherent cells were cultured at high density in the presence of cytosine arabinoside. Under these conditions the cells fused within 5-7 days and were characterized as osteoclasts by a varietyof functional, morphological, and immunohistochemical criteria. To assess cell attachment, osteoclasts were generated in 150-mm plates and when multinucleated (>95%) they were lifted by incubation at 37 "C for 60-90 min in 2 mM EDTA in PBS. The cells were then washed free of EDTA with PBS, resuspended at theappropriate density, and used within 1 h. For otherstudies osteoclasts were generated in 48-well tissue culture dishes which also acted as containers for the subsequent bone attachment and resorption assays. MetabolicLabeling and Immunoprecipitation of Osteoclast Integrins-Cells were either surface-labeled with lZ5Ior labeled metabolically with [35S]cysteineand methionine as described previously (14). Cells were lysed with a buffer containing 2% Renex detergent and then subjected to immunoprecipitation and analysis by SDS-PAGE according to methods already detailed (14). Gels were run under both reducing and non-reducing conditions and radiolabeled proteins were visualized by autoradiography as described previously (9). Cell Attachment Assays-Proteins used in the attachment assays were dissolved in PBS atbetween 0.6 and 200 pg/ml, and thesolutions were used to coat 96-well microtiter plates (Falcon, 1008).A 2-pl drop was placed in the center of each well, and the entireplate was incubated at room temperature in 100% humidity for 3 h. The wells were washed with PBS and used immediately. Each well was filled with 50 pl of prewarmed PBS, followed by the addition of a further 50 plof PBS containing 2500 freshly lifted osteoclasts. The plates were incubated at 37 "C for 30 min, following whichthey were rinsed three times with room temperature PBS. Thecells and proteins were fixed with 2% formaldehyde for 30 min and rinsed again with PBS. Cells were stained with toluidine blue (26), and all adherent multinucleated cells were counted. In studiesdesigned to test the inhibition of attachment, theosteoclasts were preincubated for 30 min at 37 "C with one of a number of anti-integrin antibodies or peptides derived from the sequence of bone matrix proteins and treated thereafter as described above. To investigate whether antibodies to specific matrix proteinscould inhibit attachment, the entire surface of the well in 96-cells plates was coated with rat osteopontin, and after 1 h excess protein was washed off. The wells were dried overnight, and the following day diluted preimmune serum or monospecific anti-serum was added as a drop in the center of each well. Three h later the serum was rinsed off and 5000 freshly isolated osteoclasts were added to each well. Thirty min later non-adherent cells were washed off, and adherent cells were fixed and stained with toluidine blue. Diluted goat antirabbit IgG conjugated to horseradish peroxidase was added, followed by 3-amino-9-ethylcarbazole to develop color and hence identify the area of primary antibody binding. The number of cells attached to the area defined by the color was quantitated by counting. Bone Attachment-Attachment of bone particles to adherent cells was determined according to methods developed in our laboratory (27). Briefly, radiolabeled rat bone was obtained by a published method (3), in which rats were injected with [3H]proline, and after 10 days long bones were harvested and scraped clean, split and rinsed free of marrow, dried, and ground in a mill. Finally, the material was sieved to a consistent size(25-63 pm), sterilized with either 70% ethanol or by irradiation with ultraviolet light, rinsed, and stored in sterile PBS.2 Aliquots (100 pgof bone, 3-5 X lo3 counts/minute in 50 p1 of PBS) were added to cells in 48-well plates, and after 30 min the wells were rinsed gently and the amount of adherent bone was determined by solubilization in 1 M HCl, followed by hydrolysis at In earlier studies we had determinedthat sterilization of the bone particles by either method gave equivalent results in the resorption assay, indicating that treatment with ethanol did not denature the proteins.

110 "C overnight and counting in a 0 spectrometer after addition of scintillant. In some experiments 500 pg of bone particles were preincubated for 30 min with non-immune rabbit serum or the same monospecific antibody to osteopontin described above, each diluted 150. Equal aliquots were taken for either quadruplicate determination of osteoclast attachment asdescribed above or for direct analysis of radiolabel content. Results areexpressed as theratio of the percentage of counts bound after incubation with pre-immune or immune serum. Bone Resorption-The same procedure as described abovefor attachment was used for the resorption assay. However the samples were left undisturbed for 24 h, at which time supernatant (700 pl) was removedand counted as before. In specific experiments involving both bone attachment and resorption, the adherent cells were pretreated with antibodies or peptides for 30 min, the reagents were removed by rinsing, and the bone particles were then added. For the attachment assays, the cells were cultured in a-minimal essential medium without serum, while for bone resorption the conditions of culture were 1%fetal calf serum in a-minimal essential medium. MiscellaneousReagents-General reagents were purchased from Sigma while chicken and fetal calf sera for cell culture were from Hazelton Biologics, Lenexa, KS. 3H-labeled proline was provided by Amersham Corp.,while lZ5I-NaI and [35S]methionine/cysteine, as Translabel, were from ICN, Irvine, CA. The monoclonal antibodies LM 609, which binds to a functional domain of a,P3 (28) and CSAT, which binds to chick Dl integrins (29, 30), were purified from ascites or spent culture medium by Protein A-Sepharose chromatography. The proteins were concentrated, dialyzed into PBS, sterile filtered, and kept at 4 "C. Irrelevant, isotype-matched antibodies (IgG1 for LM 609, IgGzb for CSAT) were obtained from The Binding Site, La Jolla, CA. A rabbit antiserum to rat osteopontin which is known to contain small amounts of cross-reactivity to BAG, an acidic glycoprotein found in bone (31), was purified of this contamination by passage over an osteopontin affinity column. On subsequent analysis this antiserum was monospecific for rat osteopontin (data not shown). Statistical Methods-The results for cell binding are from representative experiments and are presented as the mean and standard deviation. Statistical significance was determined by the used of the paired Student's t test.

RESULTS

Chicken OsteoclastsAdhere Selectively to Bone Matrix Proteins-To determine the nature of the adhesive ligands in bone, osteoclasts were allowed to attach to microtiter wells coated with different matrix proteins. The fractions tested represented the major proteins in bone matrix and included all thosewith known chemotactic and/or attachmentactivity. As shown in Fig. 1,there was specific binding to osteopontin, bone sialoprotein, and fibronectin, but not to any other protein fraction, whether partially purified or in pure form. In particular, cells did not adhere significantly to collagen, either in the pure form or as isolated peptides obtained from the fractionation of total matrix. The minimal binding seen could be increased only slightly by using a medium containing M$+ in place of Ca2+ (data not shown). Replacement of rat tail collagen with that purified from chicken gizzard led to only a slight increase in the minimal binding. Incontrast, when osteoblasts isolated from fetal avian calvariae were used in this assay, there was very significant binding to all three collagen fractions, and this binding couldbe inhibited by CSAT, an antibody to theP1 integrin ~ u b u n i tThe . ~ cells were also non-adherent to wells coated with the control protein bovine serum albumin. Chicken OsteoclastsExpress a& and Multiple PI IntegrinsTo examine the repertoire of integrins expressed on chicken osteoclasts cells were surface labeled with lactoperoxidase/ lZ5I,the lysate was immunoprecipitated with the monoclonal antibodies CSAT (anti-P1) and LM 609 (anti-add, and the resultant fractions were subjected to autoradiographic analyR. Dunlay, S. L. Teitelbaum, K. Sago, and F. P. Ross, unpublished results.

9903

Integrin-mediated MatrixOsteoclast Interactions

FIG. 1. Attachment of chicken osteoclasts to bone matrix proteins. Osteoclastswere allowed to attach for 30 min to various protein fractions in coated microtiter plates. Non-adherent cells were washed off and the attached multinucleated cells were counted after staining with toluidine blue. Significant binding was seen only in the case of rat osteopontin (rOPN),rat bone sialoprotein (rBSP), and chickenfibronectin (cFN). Fractions to which cells did not adhere included those enriched for rat osteocalcin (roc), rat osteonectin (rON),mixed rat bone matrix proteoglycans (rPGs), soluble rat bone collagen ( S O L - s ) as well as pure albumin ( r A L B ) , rat serumglycoprotein a 2 - H S (aZ-HS), rat tail collagen ( C O L I ) , or buffer alone ( P B S ) .

I60 -

n

w

120 -

I?

80 -

_1

w 0 st

40

-

PBS

rOPN rBSP

cFN rCOL-S rOC

rON COL-1 rALB rPGs

COATING AGENT (20pg/ml)

sis. Material precipitatedby CSAT showed anumber of bands particles were added to adherent osteoclasts, either with or ranging from 110 to 170 kDa (Fig. 2A, lanes2 and 4 ) , indicat- without pretreatment of the cells with specific antibodies or cells. In both fragments of matrix proteins. Therewas rapid and significant ing thatseveral p1integrins are present on these binding of bonetothe cells in the absence of anyprior these and the later studies precipitation with non-immune incubation. Of the total of 100 pg of bone added, some 25sera yielded no specific bands. Analysis of the material precipitated by the monoclonal 35% adhered within 30 min. As shown in Fig. 5 attachment antibody LM 609, directed t o avp3,showed two bands which, wasblockedby LM 609, but not by CSAT, while control on the basis of their size on both reduced and non-reduced antibodies were without effect. Both intact osteopontin and SDS-PAGE (Fig. 2 A , lanes 1 and 3 ) , were identical to that synthetic peptides containing the RGD sequence in human bonesialoproteinandosteopontin also inhibited binding. previously reported for thisintegrin (28). T o confirmthe presence of a,P3 cells were metabolically labeled with [35S] These results indicate that the integrina,p3 is directly implimethionine and cysteine and the lysate was passed over an cated in bone attachment, while p1integrins are not. Bone Resorption Zs Dependent on a&-Mediated InteracGRGDSPK-Sepharose affinity column, which is known to tions between Osteoclasts and Bone-In order to determine bind human a,p3 (28). Elution with the peptide GRGESP resulted in about 5% of the applied radiolabel being removed whether the cell attachment observed is required for osteoa solution of clast-dependent bone resorption, the labeled particles were from the column (not shown). In contrast, when cellsfor 24 h. A s shown in Fig. 6 GRGDSP was applied about 85% of the applied counts were coincubatedwiththe eluted, which, when immunoprecipitatedby LM 609, gave rise osteoclasts rapidly solubilized the bone, reflecting resorption. to bands of the same size as those seen from surface-labeled In typical experiments 10-15%of the radioisotope was discells (Fig. ZB), establishing the presenceof the chicken inte- solved within the first 24 h, a process which was inhibited of a& from an affinity about 30% by the presence of LM 609. The addition of small grin a&. These results on the elution volumes of LM 609 to the cultures at 6 hourlyintervals column comprising the matrix GRGDSP are similar to earlier produced significantlymore inhibition. Once again the RGDreports (9). containing peptides were active while both CSAT and the Osteoclast Attachment to Individual Matrix Proteins Is Mediated via Eitherp1or p3 Integrins-To identify the natureof control antibodies were without effect. the interactions between the integrins identified on the cell DISCUSSION surface and individualbone matrix proteins, osteoclastswere We havedeveloped a model in which large numbers of avian preincubated with monoclonal antibodies LM 609 (anti-a,p3) osteoclasts can be generated in vitro (25). These cells are orCSAT(anti-P1)priortoadditiontotheprotein-coated wells. As shown in Fig. 3, attachment to both bone sialopro- indistinguishable biochemically, morphologically andfunctionally from freshly isolated chicken osteoclasts. We have tein and osteopontin was blocked by LM 609, with CSAT a number of questions relathaving no effect. In contrast, CSAT blocked attachment to now used this system to address ing to osteoclast-matrix interactions and in particular therole fibronectin while LM 609 was inactive.Inbothinstances blocking was dose-dependent (data not shown) and the iso- of integrins in resorption. In order toprove that our osteoclasts both synthesize and type-matched irrelevant antibodieswere without activity. Antibodies to a Specific Matrix Protein inhibit Osteoclast express the integrin avp3on their surface, we subjected cells Binding-To determine the specificity of osteoclast-matrix to metabolic and surfacelabeling, followed by immunoprecipinteractions, we used a monospecific polyclonal antibody to itation with thespecific monoclonal antibody LM 609, allowosteopontin. This antibody totally eliminated attachment of ing us to characterize the protein biochemically. Since the osteoclasts to osteopontin-coated plates (Fig. 4A) and also material also bound to an affinity column containing an RGD markedly inhibited(50%) binding of osteoclasts towhole bone sequence, a result which is characteristic of avP3(9), we could conclude that chicken osteoclasts indeed synthesize and ex(Fig. 4 B ) . Non-immune serum had noeffect. Bone Attachment to Adherent Osteoclasts Is Mediated by press on their plasma membranea protein which is structurCY& but Notp1 Integrins and CanB e Blocked by Specific Bone ally and functionally similar to its mammalian counterpart. Matrix Proteins-To assess therole of integrins in the inter- These findings are consistent with studies in several transaction between osteoclasts and intactbone, radiolabeled bone formed cell lines (32, 33) and human osteoclasts (8).

Integrin-mediated Matrix

9904

I30 -

IO5

-

@

I) - I60

- 95

yi

1

2

FIG.2. Identification of several

integrinsand a& in chicken osteoclasts. Cells were surface labeled with "'I, and the lysate was immunoprecipitated with the monoclonal antibodies CSAT (anti &) and LM 609 (anti a&). The precipitated material was analyzed under both non-reducing and reducing conditions. Lanes 1 and 2 in panel A and lane 2 in panel B represent the results of the non-reducing gels, with lanes 3 and 4 in panel A and lane I in panel B those of the reducing gels. Several integrins were present (panel A, lanes 2 and 4 ) , with the expected sizes for the PI subunit (115 kDa non-reduced and 125 kDa reduced) and with several a subunits in the range 130-160 both reduced and non-reduced. These integrins were not characterized further. Bands corresponding to a& were also present, with sizes of 9.5 and 110 kDa and 160 and 130 kDa for the non-reduced and reduced a and subunits, respectively (panel A, lanes I and 3 ) . Cells were also labeled metabolically with [%] methionine/cysteine, lysed, and the lysate purified by passage over a column containing bound GRGDSPK. The fraction eluting with a solution of GRGDSP was immunoprecipitated with LM 609 and analyzed by SDS-PAGE (panel B ) . The material exhibited bands at 9.5 and 160 kDa (lane 2 ) , which upon reduction migrated at 110 and 130 kDa (lane I ) , characteristic of av&

It had been shown that freshly isolated human monocytes while fail to express the integrina,.&, but elaborate the dimer in culture (34). Therefore, we were concerned that all of our studies implicating this molecule as a critical mediatorof bone resorption may be artifactual. However, our previous studies demonstrating that the monoclonal antibodyLM 609 (which

Osteoclast Interactions specifically recognizes a&) reacted with both generated and freshly isolated osteoclasts (25) suggest that this is notlikely to be the case. The ability of the cells to bind bone-derived proteins was examined usinga n assay inwhich onlythe centerof microtiter plates was coated with individual proteins. This simple and rapid method permits the of usesmall amountsof rare protein fractions and each well acts as its own control, since under the described conditions osteoclastswere totally non-adherent to the surrounding uncoated plastic. Among the manyknown proteins in bone (16), osteoclasts bound significantly only to fibronectin, osteopontin, and bone sialoprotein. Interestingly, the latter two proteins were not derived from chicken bone, underscoring thehomology of the RGD domains in osteopontin and bone sialoprotein in the chicken and a number of mammalian species (35-39). Collagen comprises 90% of the organic matrix of bone and therefore would be a logical candidate to play an important role in osteoclast recognition, particularly as it contains both an RGD motif known to serve as a cell-binding site (40) and an RGD-independent binding domain(41). Moreover, human osteoclasts express the a2 integrinsubunit (18) and hence must be presumedtohave a collagenreceptor. Thus, the failure of chicken osteoclasts to bind to either rat tail or bone collagen fragments, even in the presence of magnesium ions which have been shown to support attachment via the integrin a& (42), is a critical finding of this study. This observation suggeststhat,incontrasttotheirhumancounterparts, chicken cells do not have thiscollagen receptor (the unavailability of anantibodyprecludedusfromperformingthis experiment) or that the appropriate sequence in collagen is unavailable for binding. The significance of this finding is that whilecollagen isthe overwhelmingly abundantbone protein (16), it appears toplay, a t best, a minimal role in the attachment of osteoclasts to matrix. Monoclonal antibodies to individual integrins decreased cell attachment to osteopontin, bone sialoprotein, and fibronectin in a highly specific manner. Inhibition of cell binding by LM 609, directed to a& (28), is consistent with earlier reports on the ability of this receptor to mediate attachment to osteo(43). On the other hand the pontin and bone sialoprotein monoclonal antibody CSAT,which binds to PI integrins (29), interfered with attachment only to fibronectin. While we have not identified which integrin is involved in such binding, our results suggest that it is likely to be the high affinity fibronectin receptor aaP1.The fact that the antibody toa& only inhibited partially both binding and resorption (Figs. 5 and 6) suggests that other integrin- and/or non-integrin-mediated attachment mechanisms may be active. In support of these ideas it has been demonstrated in a number of systems that the processes of initial attachment and subsequent spreading are independent (44-47). Matrix-integrininteractionsareprobablyconformationdependent. Thus, it hasbeen shown that soluble and surfaceattached formsof fibronectin have different affinitiesfor the same receptor (48). Furthermore, studies on the interactions between a& and synthetic peptides have also demonstrated marked conformation dependence(43). Since osteopontin and bone sialoprotein had been purified in the presenceof strong chaotropicagents (20,22), partialdenaturationmayhave resulted, and the proteinsmay not have been in their native configuration. This latter reason and proven conformation dependence of fibronectin binding to its receptor raised the possibility that the results obtained with proteins coated on plastic were not representative of the in vivo situation. T o address this issue, intact bone particles were incubated with

9905

Integrin-mediated Matrix Osteoclast Interactions FIG.3. Antibodies to both 4 3 and chicken f l l integrins block attachment of osteoclasts to bone matrix proteins. Osteoclasts were preincubated for 30 min with no antibody, purified LM 609, CSAT, and irrelevant antibodies all at 200 pg/ml, prior to plating in microtiter plates coated with fibronectin, rat bone sialoprotein, and rat osteopontin. After 30 min, the number of adherent cells was determined by staining and counting. CSAT inhibited binding to fibronectin, while LM 609 reduced attachment tobone sialoprotein and osteopontin. Coincubation of both antibodies revealed no synergism and the use of irrelevant, isotype-matched antibodies (IgGI and IgGz) had no effect (neither result shown). * p < 0.01.

A.

200

r

”

Fibronectin Osteopontin Bone Sialoprotein

r B.

I

IOoo 4

E

800

t

T

T

NIL

IgG,

-

8

I

3 II

I 600 0

2 t

w

I-

s

z

B

8

a

-

04

400

02

200 -

?t

”

n NanImmune

Immune

NonImmune

Immune

SERUM

FIG. 4. A monospecific antibody to rat osteopontin inhibits binding of osteoclasts to both osteopontin and whole bone. Panel A , microtiter wells werecoated with osteopontin, excess protein was washed off, and the plate dried overnight. A drop of preimmune or anti-osteopontin goat antiserum, diluted 1:50 in PBS, was added for 30 min, after which freshly lifted osteoclasts were added, allowed to bind for 30 min, and then rinsed off prior to fixing and staining. The antibody-containing area (marked with a horseradish peroxidaselabeled second antibody to goat IgG) was identified, and the number of bound cells was counted. The results represent the mean of two separate experiments. The specific antibody totally inhibits attachment of osteoclasts to osteopontin. Panel B, intact bone particles were preincubated with either PBS alone or the same sera described in panel A . After 30 min quadruplicate aliquots were added to adherent osteoclasts, while a fifth sample was processed directly to determine isotope content. Non-adherent bone was rinsed off and that which adhered was determined by dissolution and hydrolysis, as described previously. The results are expressed as the ratio of bone bound in the presence of immune and preimmune serum. (Bone incubated with non-immune serum bound about 70% as well to osteoclasts as did that incubated with PBS alone (not shown)). The results shown are the mean of four separate experiments with different preparations of osteoclasts. The anti-osteopontin antibody decreased attachment by 50% ( p < 0.05).

lgGz

CSAT

LM 609 hOPN(1) hBSP(I) rOPN

BLOCKING AGENT (200 pgirnl)

FIG.5. Binding of intact bone to osteoclasts is inhibited by LM 609, osteopontin, and bone sialoprotein fragments, but not CSAT. Adherent osteoclasts in 24-well plates were treated for 30 min with either PBS alone ( N I L ) or PBS which contained LM 609, CSAT, intact rat osteopontin ( O P N ) , RGD-containing fragments of human osteopontin (OPN-f) and bone sialoprotein (BSPf) and irrelevant antibodies (IgG1 and IgG2b), all at 200 pg/ml, followed by the addition of a suspension of [3H]proline-labeled rat bone particles (100 rg, 3-5 X lo3 counts/minute). Thirty min later non-adherent material was rinsed off, and the attached bone was determined as described under “Materials and Methods.” The rapid binding of bone to cells was inhibited by blocking avoB,but not PI, function. The soluble bone matrix protein and fragments also inhibited attachment. *p < 0.01.

and the integrin a,P3 are involved in attachment. Further confirmation of this was obtained in experiments in which polyclonal antibodies to rat osteopontin totally inhibited cell attachment to surfaces coated with the protein. Most importantly, this antibodyalso decreased attachment of osteoclasts to the same intact bone particles used in the bone attachment assays. Whilewe cannot prove that the targetfor the integrin in these studies is the peptidesequence RGD (e.g. by the use osteoclasts. In this instance attachment was mediated through of domain-specific antibodies), the ability of peptides containa,p3 but not (3,-containing integrins, since the antibody LM ing thissequence t o mimic this result supports this hypothesis 609 but not CSAT inhibited theprocess. Further support for strongly. These studies complement an earlier report in which such specificity was obtained in experiments where RGD- the peptide GRDGSP inhibited the formation of characteriscontaining peptides derived from osteopontin and bone sial- tic pits by osteoclasts (49). In this work the inactive peptide oprotein also interfered with binding. These critical results GRGESP was much less efficacious, as expected. revealed that interactions between these two matrix proteins Taken as a whole these results provides the first direct,

9906

Integrin-mediated Matrix

FIG. 6. Bone resorption is blocked by pretreatment of osteoclasts with osteopontin, bone sialoprotein, and the antibody LM 609, but not CSAT. The experiment described in Fig. 4 was repeated, except that incubation was extended to 24 h. At this time the supernatant was assayed for the presence of [3H]hydroxyproline/proline. In the absence of any additions, about 10-15% of the isotope was released (not shown), and thisamount was decreased by treatment of the cells with LM 609 and both intactrat osteopontin and RGD-containing fragments of the proteins osteopontin and bone sialoprotein, but notby CSAT. Repeated additions of LM 609 (6 hourly, 50 pg/ml) increased the degree of inhibition. *p < 0.05 ; **p < 0.01.

Tf N

z

0

w m

8 2

U

W

z

$2

Osteoclast Interactions

I

T

T

i ** T

4 8

functional data for the importance of the interactions between the integrin and osteopontin. Conversely, they suggest that &-mediated attachment of cells to fibronectin, while occurring in uitro, is not demonstrable and hence may have limited significance in uiuo. As an extension of these studies the bone particles were left in contact with the osteoclasts for 24 h to monitor resorption in the presence of various antibodies to integrins. Paralleling the bone adhesion experiments, both selected RGD-containing peptides and LM 609, but neither CSAT nor the isotypematched controls, inhibited this process. When LM 609 was added at 6 hourly intervals therewas an increase in inhibition, suggesting that thecells metabolize the antibody. A recent report on the inhibition of osteoclast function in both mammalian and avian osteoclasts by the snake venom echistatin (50) has partially corroborated our results. These authors showed that this RGD-containing protein inhibited both bone resorption by and response of the cells to calcitonin in a manner suggesting that an av integrin was implicated. However, they failed to identify clearly the integrin involved in these events, a question of growing importance since it is now known that, in addition to its classical interaction with the P3 subunit, av can also associate with either PI, f i b , P6, or Pa (51-57). We have now shown by both immmunoprecipitation and blocking of adhesion with an antibody which recognizes specifically a& that this integrin is present and involved directly in these events. A t this time we cannot exclude the presence and function of other integrins involving the av chain and studies are inprogress to address this issue. In summary, we have demonstrated that chicken osteoclasts express the integrin a& which mediate attachment of the cells to specific bone matrix proteins. Binding to fibronectin is mediated by an unidentified integrin, possibly a&, and can be demonstrated only when the protein is coated on plastic. In contrast a,P3 modulates attachment to osteopontin and bone sialoprotein both on plastic-coated plates and in,for the former molecule, in intact bone. While the predominant event in the dissolution of the organic matrix is the proteolysis by an acidic collagenase of the major matrix component collagen (3), this same protein (collagen) appears to play no role in the attachment of avian osteoclasts to bone or in subsequent matrix resorption, despite the presence of an RGD sequence. Thus, these studies have demonstratedseparate roles for different matrix proteins found in bone. Most significantly,

NIL

IgG1 lgG2

CSAT LM609 hOPN(1) hBSP(1)

BLOCKING AGENT (200pg/ml)

OPN LM 609 (4x50 Pg)

we have shown directly that interference in the interactions between the integrin a,P3 and a protein in the context of whole bone led to decreased ability of osteoclasts to bind to and resorb this tissue. This result suggests an approach for the design of drugs which will beefficacious in the treatment of a wide range of clinically important bone diseases. Acknowledgments-We thank Dr. Clayton Buck for providing the CSAT hybridoma, Dr. Richard Mayne for the gift of chicken collagen, Dr. Ron Midura for the rat bone sialoprotein, and Jan Brunn for assistance in the purification of osteopontin and the osteopontin antibody. We acknowledge Jane Wodicker for expert secretarial assistance. REFERENCES 1. Vaes, G. (1988) Clin. Orthop. Rel. Res. 231, 239-271 2. Mundy, G. R. & Roodman, G. D. in Bone and Mineral Research15 (Peck, W. A,, ed) pp. 209-279, Elsevier, Amsterdam

3. Blair, H. C., Kahn, A. J., Crouch, E. C. Jeffrey, J. J. & Teitelbaum, S. L. (1986) J. Cell Biol. 1 0 2 , 1164-1172 4. Hynes, R. 0.(1987) Cell 48,549-554 5. Ruoslahti, E. & Pierschbacher, M. D.(1987) Science 238,491-497 6. Hemler, M. E. (1990) Annu. Reu. Immunol. 8,365-400 7. Davies, J., Warwick, J., Totty, N., Philp, R., Helfrich, M. & Horton, M. (1989) J. Cell Brol. 109,1817-1826 8. Zambonin-Zallone, A., Teti, A., Grano, M., Rubinacci, A., Abbadini, M., Gamboli, M. & Marchisio, P. C. (1989) I+p. Cell Res. 182, 645-652 9. Cheresh, D. A. (1987) Proc. Natl. Acad. Scr. U. S. A. 8 4 , 6471-6475 10. Charo, I. F., Fitzgerald, L. A., Steiner, B., Rall, S. C., Bekeart, L. S. & Philips, D. R. (1986) Proc. Natl. Acad. Sei. U. S. A. 83, 8351-8355 11. Lawler, J., Weinstein, R. & Hynes, R. 0.(1988) J. Cell Biol. 107,23512361 12. Chambers, T. J., Fuller, K., Darby, J. A. Pringle, J. A. S. & Horton, M. (1986) Bone Miner. 1,127-135 13. Reinholt, F. P., Hultenby, K., Oldberg, A. & Heinegard, D. (1990) Proc. Natl. Acad. SCL.U. S. A. 87,4473-4475 14. Cheresh, D. A., Berliner, S. A., Vincente, W. & Ruggeri, S. M. (1989) Cell 58.945-953 . ~ . ", .~~ 15. Lawler, J. W. & Hynes, R. 0.(1986) J. Cell Biol. 103,1635-1648 16. Heinegard, D. & Oldberg, A. (1989) FASEB J.3, 2042-2051 17. Gehron Robev. P.. Young. M. F.. Fisher, Larrv W. & McClain, T. D. ( 1989) J. Cell Biol '108,719c127 18. Horton, M. A. & Davies, J. (1989) J . Bone Min. Res. 4,803-808 19. Ruoslahti, E. (1991) J. Clin. Znuest. 8 7 , l - 5 20. Fisher. L. W.. Hawkins. G. R.. Tuross. N. & Termine. J. D.(1987) J. Biol. Chem. 262,9702-9708 21. Termine, J. D.,Belcourt, A. B., Christner, P. J., Conn, K. M. & Nylen, M. U.(1980) J. Biol. Chem. 256,9760-9768 22. Prince, C. W., Oosawa, T., Butler, B. T.,Tomana, M., Bhown, A. S., Bhown. M. & Schrohenloher, R. E. (1987) J. Biol. Chern. 262, 29002907 23. Mizuno M., Farach-Carson, M. C., Pinero, G. J., Fujisawa, R., Brunn, J. C., d y e r , J. M., Bousfield, G. R., Mark, M. P. & Butler, W. T. (1991) Bone Miner. 13, 1-21 24. Midura, R. J., McQuillan, D. J., Benham, K. J., Fisher, L. W. & Hascall, V. C. (1990) J. Biol. Chem. 266,5285-5291 25. Alvarez, J., Teitelbaum, S. L., Blair, H. C., Greenfield, E. M., Athanasou, N. A. & Ross, F. P. (1991) Endocrinology 128,2324-2335 26. Goldman, R. & Bar-Shavit, Z. (1979) Natl. Cancer Zmt. 63,1009-1010 27. Bar-Shavit, Z., Teitelbaum, S. L. & Kahn, A. J. (1983) J. Clin. Invest. 72, 516-525 ~

'

Integrin-mediated Matrix 28. Cheresh, D. A. & Spiro, R. C. (1987) J. Bid. Chem. 262,17703-17711 29. Knudsen, K. A., Horwitz, A. F. & Buck, C. A. (1985) Exp. Cell Res. 157, 218-226 30. Hynes, R. O., Marcantonio, E. E., Stepp, M. A,, Urry, L. A. & Yee, G. H. (1989) J. Cell Biol. 109.409419 31. Gorski, J. P. & Shimizu, K. (1988) J. Biol. Chem. 263,15938-15945 32. Cheresh, D. A. & Harper, J. R. (1987) J. Bid. Chem. 262, 1434-1437 33. Oldberg, A,, Franzen, A. & Heinegard, D. (1986) Proc. Natl. Acad. Sci. U. S. A. 83,8819-8823 34. Krissansen, G. W., Elliott, M. J., Lucas, C. M. Stomski, F. C., Berndt, M. C., Cheresh, D. A,, Lopez, A. F. & Burns, b. F. (1990) J. Bid. Chem. 265,823-830 35. Kiefer, M.C., Bauer, D.M. & Barr, P. J. (1989) Nucleic Acids Res. 17, mnfi 36. Fiiclr: L. W., McBride, 0. W., Termine, J. D. & Young, M. F. (1990) J. Biol. Chem. 265,2347-2351 37. Gotoh, Y., Gerstenfeld, L.C. & Glimcher, M. J. (1990) Eur. J . Biochem. 187.49-58 38. Zhang,'QI, Domenicucci, C., Goldberg, H. A,, Wrana, J. L. & Sodek, J. (1990) J. Bid. Chem. 265, 7583-7589 39. Cralg, A. M., Smith, J. H. & Denhardt, D. T. (1989) J . Biol. Chem. 264, 9682-9689 40. Dedhar, S., Ruoslahti, E. & Pierschbacher, M. (1987) J . Cell Biol. 104,

-""

SR6-hW ""y

41. Staatz, W. D., Fok, K. F., Zutter, M. M., Adams, S. P., Rodriguez, B. A, & Santoro, S. A. (1991) J. Bkol. Chem. 266,7363-7367 42. Staatz, W. D., Rajpara, S. M., Wayner, E. A,, Carter, W. G. and Santoro,

Osteoclast Interactions

9907

S. A. (1989) J Cell Bid. 108, 1917-1924 43. Oldberg, A,, Franzen, A,, Heinegard, D., Pierschbacher, M. & Ruoslahti, E. (1988) J. Bid. Chem. 263. 19433-19436 44. Sauk, J.' J., Van Kampen, C: L., Norris, K., Moehring, J., Foster, R. A. & Somerman, M. J. (1990) Exp. Cell Res. 188, 105-110 45. Oka. J. A. & Weieel. P. H. (1986) J. Cell Biol. 103.1055-1060 46. Runyan, R. B., Versalovic, J. & Shur, B. D. (1988) 2. Cell Biol. 107,18631871 47. Beacham, D. A. & Jacobson, B. S. (1990) Exp. Cell Res. 189,69-80 48. Garcia-Pardo, A,, Ferreira, 0. C., Valinsky, J. E., & Bianco, C. (1989) Exp. Cell Res. 181,420-431 49. Horton, M. A., Taylor, M. L., Arnett, T. R. & Helfrich, M. H. (1991) Ezp. Cell Res. 195.~R6R-375 ~~ . ., 50. Sato, M., Sardana, M. K., Grasser, W. A., Garsky, V. M., Murray, J. M. & Gould, R. J. (1990) J. Cell Bid. 111, 1713-1723 51. Vogel, B. E., Tarone, G., Giancotti, F. G., Gailit, J. & Ruoslahti, E. (1990) J. Biol. Chem. 265 5934-5937 52. Bodary, S. C. & McLean, J. W. (1990) J. Biol. Chem. 265, 5938-5941 53. Dedhar. S. & Grav. V. (1990) J. Cell Biol. 110. 2185-2193 54. Cheresh, D. A., Smith,' J. W . , Cooper, H. M.'& Quaranta, V. (1989) Cell 57,59-69 55. Smith, J. W., Vestal. D. J.. Irwin. S. V., Burke. T. A. & Cheresh. D. A. (1990) J . Biol. Chem. 265, 11008-11013 56. Sheppard, D., Rozzo, C., Starr, L., Quaranta, V., Erle, D. J. & Pytela, R. (1990) J. Biol..Chem. 265,11502-11507 57. Moyle, M.,Napxer,M.A. & McLean, J. W. (1991) J. Biol. Chem. 266, 19650-19658 ~

~

""