We used polymerase chain reaction (PCR)-selective complementary DNA (cDNA) subtraction hybridization with an immortalized murine osteoclast (OCL) precursor cell line to identify genes that are highly expressed in OCLs compared with OCL precursors and which may be involved in the OCL differentiation process. ADAM8 was one of the 50 genes identified. ADAM (a disintegrin and metalloproteinase) peptides are membrane-bound proteins that can act as cell-to-cell and cell-to-matrix adhesion molecules, degrade the extracellular matrix, and play a role in tissue morphogenesis. Addition of antisense (AS) S-oligonucleotides for ADAM8 (1-10 nM) to mouse bone marrow cultures treated with 10−9 M 1,25-dihydroxyvitamin D3 [1,25-(OH)2D3] significantly inhibited OCL formation compared with treatment with the control S-oligonucleotide. Furthermore, conditioned media from 293 cells transiently transfected with a secretable form of the ADAM8 cDNA increased OCL formation in a dose-dependent manner. In addition, treatment of OCLs with soluble ADAM8 conditioned media significantly increased pit formation per dentin slice compared with control OCLs. Time course studies indicated that ADAM8 increased OCL formation only when it was present during days 4-7 of the 7-day culture period. Structural analysis, using truncated constructs of ADAM8, showed that the cysteine-rich/disintegrin domain was responsible for its OCL stimulatory activity. Western blot analysis confirmed that the soluble form of ADAM8 is present in normal marrow cultures. These data suggest that ADAM8 plays an important role in OCL formation and acts primarily at the later stages of OCL differentiation.
OSTEOCLASTS (OCLS) are derived from cells in the monocytic lineage.(1) During OCL differentiation, many genes are up-regulated and may affect OCL formation or activity. For example, cathepsin K, tartrate-resistant acid phosphatase (TRAP), c-Src, the calcitonin receptor, and metalloprotease 9 (MMP9) are up-regulated during OCL precursor differentiation and enhance or inhibit OCL differentiation/activity.(2, 3) Therefore, identification of genes that are up-regulated during OCL differentiation and affect OCL formation and activity is important to our understanding of the factors controlling osteoclastogenesis.
Recently, we have developed an OCL precursor cell line (B/T cells) from mice that were doubly transgenic for the Bcl-XL and Tag genes.(4) B/T cells are OCL precursors that are bipotent and can differentiate either to OCLs or macrophages. B/T cells were used as a source of highly purified OCL precursors and to obtain large numbers of OCLs, which formed when B/T cells were treated with receptor activator of nuclear factor κB ligand (RANKL) and macrophage colony-stimulating factor (M-CSF) to induce OCL formation. Polymerase chain reaction (PCR)-selective complementary DNA (cDNA) subtraction hybridization was then performed to identify genes that were up-regulated in OCLs compared with their precursors. ADAM8, a member of the ADAM gene family, was one of the genes overexpressed in OCLs. The ADAM (a disintegrin and metalloproteinase) gene family is composed of proteins that have transmembrane and cytoplasmic domains composed of MMP-like, disintegrin-like, cysteine-rich, and epidermal growth factor (EGF)-like (snake venom MMP-like) domains. ADAM family members have been implicated in cell fusion, cell-to-matrix adhesion,(5, 6) and can act as cell signal molecules, which are vital for normal cellular processes such as cell morphogenesis, wound healing, tumor cell invasion, and metastases.(7, 8) ADAM8 is expressed primarily in cells of the monocyte/macrophage lineage as a type 1a membrane-anchored protein.(9) We examined the effects of ADAM8 on OCL differentiation and formation using a soluble form of ADAM8 that was expressed in 293 cells because ADAM8 has been reported to be involved in myocyte fusion. We report that ADAM8 significantly stimulated OCL formation in mouse cultures and antisense (AS) to ADAM8 inhibited OCL formation in marrow cultures treated with 1,25-dihydroxyvitamin D3 [1,25(OH)2D3]. These results suggest that ADAM8 may have a physiological role in OCL formation.
MATERIALS AND METHODS
RANKL was generously provided by Immunex Corp. (Seattle, WA, USA), and M-CSF was purchased from R & D (Minneapolis, MN, USA). Cytotoxic antisera against mouse fibroblasts and juvenile rabbit complement were obtained from Dr. T. Yoneda (University of Texas Health Science Center, San Antonio, TX, USA). Mac I antibody and anti-rat immunoglobulin G (IgG) were purchased from Dako Corp. (Carpinteria, CA, USA). Restriction enzymes, Taq polymerase, and tissue culture media were obtained from Gibco/BRL (Grand Island, NY, USA), and all other chemicals were from Sigma Corp. (St. Louis, MO, USA).
Preparation of OCL precursors and OCLs
OCL precursors and OCLs were prepared from B/T cells as previously described.(4) Briefly, B/T cells were maintained in α-minimal essential medium (α-MEM) with 10% fetal calf serum (FCS) and cocultured with the MC3T3 osteoblast cell line. To remove the feeder cells, B/T cell cultures were treated with a cytotoxic antiserum against mouse fibroblasts at a dilution of 1:1000, followed by addition of complement at the dilution of 1:300 for 1 h at 37°C. The cells were washed extensively with phosphate-buffered saline (PBS) and then were trypsinized and collected by centrifugation. Mac I panning was then performed to obtain enriched B/T cells that were more than 99% pure, as previously described.(10)
Purified B/T cells were cultured with 100 ng/ml of RANKL and 20 ng/ml of M-CSF in α-MEM/20% FCS. Every 3 days, the media were changed, and at day 7, OCLs were scored. The OCLs were purified by trypsinization for 10 minutes to remove mononuclear cells, and the purified OCL RNA was extracted with RNAzol B (Tel-Test, Inc.; Friendswood, TX, USA) and total RNA was isolated according to the manufacturer's protocol.
PCR-selective subtraction hybridization
PCR-selective subtraction hybridization was performed as described in the manufacturer's protocol using a kit from Clontech Co. (Palo Alto, CA, USA). Briefly, messenger RNAs (mRNAs) from purified B/T cells and purified OCLs formed by B/T cells were converted into cDNA using reverse transcriptase (RT). Both cDNAs were digested with RsaI to obtain shorter, blunt-end cDNAs. cDNAs from OCLs were used as “tester” cDNA, and cDNAs from B/T cells were used as “driver” cDNA. The OCL cDNAs were divided into two pools and each pool was ligated with a different double-stranded cDNA adaptor. Excess B/T cell cDNA was added to each of the two OCL cDNA pools at a 1:4 (OCL:B/T cDNA) ratio. The two hybridization reactions were heat-denatured and then allowed to anneal for 48 h at 68°C. The two hybridization reactions were then mixed together to allow the single-stranded differentially expressed OCL cDNA remaining in the pools to hybridize and form double-stranded sequences that contained both types of adaptors. Enriched OCL cDNA fragments that were overexpressed in OCLs compared with B/T cells were then amplified by PCR using both adaptors as primers and32P-deoxycytosine triphosphate (dCTP) under high-stringency conditions. The PCR reaction conditions were varied to enhance detection of overexpressed OCL cDNAs by varying either the annealing temperature from 62°C to 66°C, the Mg+2 ion concentration from 1.0 to 2.5 mM, the dimethylsulfoxide (DMSO) concentration from 0% to 2.5%, or the number of PCR cycles from 20 to 25. Control PCR reactions were performed using only OCL cDNA or B/T cDNA as the template. Fifty PCR products were identified as being highly overexpressed by these techniques and were reamplified by secondary PCR. After cloning the PCR products in the TA vector (Invitrogen; Carlsbad, CA, USA), the DNA sequences were determined and compared with those in the National Center for Biotechnical Information (NCBI) database using the Blast program.
After the DNA sequences were obtained, sense (SS) and AS primers for the PCR products were designed and cycle-dependent RT-PCR was performed to confirm that the genes of interest were up-regulated in OCLs. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) primers 5′-ACC ACA GTC CAT GCC ATC AC-3′ (SS) and 5′-TCC ACC ACC CTG TTG CTG TA-3′ (AS) were used as a control. The expression levels of other ADAM gene family members in OCLs also were determined. The specific primer sequences for mouse ADAM gene family members used in the cycle-dependent RT-PCR reactions were as follows: 5′-CAT AGT GAA ACC AAA GAG GCC-3′ (SS) and 5′-ATA GGA GCA GTG GTA TCT CC-3′ (AS) for ADAM8 (350 base pairs [bp]); 5′-TAA CAT CAC AGA TGT CCC GG-3′ (SS) and 5′-CGA TCA GCA TTA GAC AGC AC-3′ (AS) for ADAM1 (394 bp); 5′-GGT GTG TAC TCA GTT ATT GGC-3′ (SS) and 5′-ACA GCT GAG AGA GTA ATC CC-3′ (AS) for ADAM9 (MDC9) (517 bp); 5′-CCT GTT GTT ATT GGT CCT CG-3′ (SS) and 5′-TGC TTC TCT AGT GTT TGG GG-3′ (AS) for ADAM15 (MDC15) (419 bp); and 5′-CCC TGT CTC TGT TTC ATC AC-3′ (SS) and 5′-AAG GGG TCC TTC TCA AAT CC-3′ (AS) for ADAM17 (243 bp). The PCR conditions were 94°C for 30 s, 58°C for 30 s, and 72°C for 1 minute for 14-22 cycles for GAPDH and 18-26 cycles for ADAM1, ADAM8, ADAM9, ADAM15, and ADAM17. The specificity of the PCR products for the various ADAM family members was confirmed by DNA sequence analysis after cloning in TA cloning vector (Invitrogen). PCR reactions containing32P-dCTP were sampled every two cycles and analyzed by 5% polyacrylamide gel electrophoresis (PAGE) to assess the linearity of PCR reaction. After drying, the gel was autoradiographed on X-ray film followed by densitometric analysis.
Generation of a full-length mouse ADAM8 cDNA and construction of soluble and truncated forms of ADAM8
The full-length ADAM8 open reading frame cDNA was generated by PCR, using the specific primers 5′-CAG GCC AGT GAT GAA GAG AG-3′ (SS) and 5′-TAG CTG GTC ACC TCT TCT GG-3′ (AS). After cloning the PCR product in the TA vector, DNA sequence analysis was performed to confirm that it was mouse ADAM8. A soluble form of ADAM8 using the AS primer 5′-CTA CGC TGC TTG TTC ATC TGA-3′, which did not contain the transmembrane and cytoplasmic domains of ADAM and could be secreted into conditioned media, also was generated by PCR. The soluble form of the ADAM8 gene was subcloned into the pcDNA3 mammalian expression vector (Invitrogen), transfected into 293 cells, and expressed, as reported previously.(11)
To map the domains of ADAM8 responsible for OCL formation, truncated forms of the ADAM8 cDNA were generated, which contained the structurally distinct domains of ADAM8 (Fig. 1). To generate truncated forms of ADAM8, the following SS and AS specific primers for ADAM8 were used for PCR: SS1 (−63 from ATG) 5′-CAG GCC AGT GAT GAA GAG AG-3′; SS2 (+1165) 5′-CCG CTCGAG CTG ACC AAT GTT CCA GAT GTC-3′; AS1 (+62) 5′-CCG CTCGAG CAA AGG AGG TCC AGG GG-3′; AS2 (+1515) 5′-TTA CCC ATC AAA GCA GTA GCC C-3′; AS3 (+1624) 5′-TTA GAT CCT GCC AGA GTA CAT CC-3′; and AS4 (+1968) 5′-CTA CGC TGC TTG TTC ATC TGA-3′. For the generation of the Tru 4 deletion construct of ADAM8, which is secreted by using the signal peptide of ADAM8, the two PCR products of SS1/AS1 and SS2/AS3 were digested with XhoI (underlined in primer sequences), ligated together, and cloned. These subclones were confirmed by sequence analysis and recloned in the pcDNA3 mammalian expression vector. The capacity of these deletion constructs to stimulate OCL formation was then tested in murine bone marrow cultures, as described.
Mouse bone marrow assays
Mouse bone marrow assays were performed as described previously to assess the OCL stimulatory activity of ADAM8.(12) Briefly, freshly isolated mouse bone marrow cells (106/well) were cultured for 8 days in α-MEM/15% FCS in the presence of conditioned media from 293 cells transfected with the soluble ADAM8 cDNA or empty vector. At the end of the culture period, the cells were fixed and stained for TRAP using a TRAP staining kit (Sigma Corp.) to identify OCL-like multinucleated cells (MNCs). MNCs were scored with an inverted microscope. In selected experiments, conditioned media from 293 cells transfected with the soluble form of ADAM8 cDNA were added to marrow cultures for the first 2 days, days 2-4, days 4-6, or for the entire culture period. The cultures were continued for a total of 7 days, and the number of TRAP(+) MNCs formed was determined. In selected experiments, murine bone marrow cells were cultured on slices of sperm whale dentin in 48-well plates. After 8 days of culture, the dentin slices were stained and the number of TRAP(+) MNCs was scored. The cells on the dentin slices were removed gently by rubbing the slices, and the number of bone resorption pits and area resorbed were measured by image analysis techniques as previously described.(13)
Effects of AS and SS oligonucleotides to ADAM8 on OCL formation
To determine if native ADAM8 was involved in OCL formation, we designed AS and SS S-oligonucleotides (5′-CCA AGC ATG ATG GGA TCT GGA GCA C-3′ and 5′-GTG CTC CAG ATC CCA TCA TGC TTG G-3′) that included the ATG and ribosome binding site of the ADAM8 gene. The AS and SS oligonucleotides were added to mouse bone marrow culture stimulated with 10−9 M 1,25(OH)2D3. Every 3 days, half of the media was replaced with fresh media containing the S-oligonucleotide and 10−9 M 1,25(OH)2D3. At the end of the culture period, the cells were stained for TRAP activity, and the number of TRAP(+) MNCs and TRAP(+) mononuclear cells formed were determined.
Western blot analysis for ADAM8 in cell lysates and conditioned media from mouse bone marrow cultures
A polyclonal antibody to recombinant ADAM8 was raised in rabbits as previously described(11) by immunizing them with two synthetic polypeptides that comprised the cysteine/disintegrin domain of ADAM8. Structural analysis had determined these regions to be highly hydrophilic and antigenic regions. These polypeptides (468CKVKPAGEV and 490CDGRKPT) were purified by high-performance liquid chromatography (HPLC), conjugated with keyhole limpet hemocyanin (KLH), and injected into rabbits. The polyclonal antibody was used for detection of ADAM8 in cell lysates and conditioned media from 293 cells transfected with the ADAM8 cDNA or mouse bone marrow cultures stimulated with 1,25(OH)2D3 (10−9 M) for 7 days as follows. Briefly, the cell lysates and conditioned media were concentrated on an Amicon 10 column, analyzed on PAGE gels as described previously,(11) and then incubated with the polyclonal ADAM8 antibody or β-actin (a generous gift of Dr. Chen, University of Texas Health Science Center, San Antonio, TX, USA) for Western blot analysis. In specific experiments, mouse marrow cells were treated with ADAM8 AS S-oligonucleotide or control SS S-oligonucleotide, and the conditioned media and cell lysates were subjected to Western blot analysis as described previously.
All cultures were performed in quadruplicate, and the mean ± SEM for the number of OCLs formed was determined. The means of individual treatment groups in independent experiments were compared using Student's t-test, and the results were considered significantly different for p < 0.05.
PCR-selective subtraction hybridization
Using PCR-selective subtraction hybridization, approximately 200 bands were detected that were overexpressed in OCLs compared with B/T precursor cells. We selected 50 bands that were the most highly overexpressed in OCLs for further sequence analysis. Overexpression of these bands was confirmed in a second PCR reaction (data not shown). DNA sequences were compared with those in the NCBI database; and three bands that were independent regions of the mouse complement component C3, one band that was macrophage inflammatory protein 1-α, and one band that encoded mitogen-activated protein (MAP) kinase kinase were identified. All of these genes have OCL stimulatory activity in vitro.(14–16) In addition, ADAM8, which induces cell-to-cell fusion in myocytes, and one expressed sequence tag (EST) DNA sequence were detected. ADAM8 was selected for further study because of its known function in cell fusion and its unknown role in OCL formation.
Expression of ADAM8 and the other ADAM gene family members in OCL precursors and OCLs
Cycle-dependent RT-PCR was performed to confirm that expression of ADAM8 (350 bp) was increased in OCLs. As shown in Fig. 2, ADAM8 expression was increased about 4-fold in OCLs compared with the GAPDH (451 bp) control. In contrast, the expression levels of other ADAM family members, ADAM1 (394 bp),(17) ADAM9 (517 bp),(18) ADAM15 (419 bp),(19) and ADAM17 (243 bp),(20) were not increased in OCLs.
Effects of soluble ADAM8 conditioned media on TRAP(+) MNC formation in mouse bone marrow cultures
The full-length mouse ADAM8 cDNA (3.0 kilobase pair [kbp]) was generated by PCR, its sequence was confirmed by sequence analysis, and the PCR product was then used for the construction of a soluble form of ADAM8 (sADAM8). sADAM8 did not contain the transmembrane and cytoplasmic domains and was cloned into the mammalian expression vector pcDNA3. After transient transfection into 293 cells, conditioned media were harvested and tested for their capacity to induce TRAP(+) MNC formation in mouse bone marrow cultures in the absence of any added osteoclastogenic stimulator. As shown in Fig. 3A, sADAM8 (1-10%, vol/vol) increased OCL-like TRAP(+) MNC formation in a dose-dependent pattern compared with control conditioned media from 293 cells transfected with the empty vector.
Western blot analysis of conditioned media from 293 cells transfected with the full-length ADAM8 cDNA (wADAM8), which contained a transmembrane domain, showed that the 70-kDa soluble form was present in the conditioned media, which was identical to the ADAM8 band from conditioned media from mouse bone marrow cultures stimulated with 1,25(OH)2D3. Conditioned media from 293 cells transfected with a soluble form of ADAM8 cDNA (sADAM8) also showed the same 70-kDa band (Fig. 3B). The membrane-bound form of ADAM8 was not detected on Western blots of cell lysates (data not shown). ADAM8 was not detected in mouse bone marrow (mBM) conditioned media not treated with 1,25(OH)2D3 mBM. These data suggest that either very small amounts of the membrane form of ADAM8 are normally present or our antibody does not detect the membrane-bound form of ADAM8.
Effects of sADAM8 conditioned media on bone resorption on dentin slices by OCLs formed in murine bone marrow cultures
To determine the bone-resorbing capacity of OCLs formed in cultures treated with sADAM8, we placed dentin slices in marrow cultures treated with various concentrations of sADAM8 conditioned media. As shown in Fig. 4, treatment of marrow cultures with sADAM8 conditioned media significantly increased the number of bone resorption pits and the resorption area in a dose-dependent manner (p < 0.05) compared with control media from 293 cells transfected with the empty vector.
Time course studies
To determine when ADAM8 was acting during OCL formation, ADAM8 conditioned media were added on days 0-2, 2-4, and 4-6, or for the entire 7 days of culture, and the TRAP(+) MNC formation was determined. TRAP(+) MNC formation was only increased significantly compared with the vector-only control media when ADAM8 was present for the later stages of the culture (days 4-6) or for the entire culture period. ADAM8 did not increase MNC formation if it were only present during the early stage of the cultures (Fig. 5).
Effects of AS S-oligonucleotide of ADAM8 on TRAP(+) MNC formation, mononuclear cell numbers, and ADAM8 protein expression levels in mouse bone marrow cultures
To determine if native ADAM8 was involved in OCL formation, we tested the effects of AS S-oligonucleotide on MNC formation in murine cultures treated with 1,25(OH)2D3 (10−9 M). When mouse bone marrow cultures were treated with ADAM8 AS S-oligonucleotide at a concentration of 1-10 nM, TRAP(+) MNC formation was reduced by about 40% (Fig. 6A) compared with the control SS S-oligonucleotide (Fig. 6B). At a concentration of 100 nM, AS and SS oligonucleotides were toxic to mouse bone marrow cultures. Western blot analysis of this conditioned medium indicated that the AS S-oligonucleotide to ADAM8 decreased ADAM8's expression but not β-actin in murine marrow cultures and that a soluble form of ADAM8 is normally present in mouse bone marrow cultures treated with 1,25(OH)2D3 (Fig. 6C). The identity of this band as ADAM8 was confirmed by competition analysis using 5 μg/ml of the synthetic ADAM8 peptide as competitor (data not shown).
In murine bone marrow cultures stimulated with 1,25(OH)2D3 (10−9 M) and treated with 10 nM ADAM8 SS S-oligonucleotide for 7 days, the number of TRAP(+) mononuclear cells was similar to the levels present in control cultures (Fig. 6D). In contrast, treatment of marrow cultures with AS S-oligonucleotide significantly increased TRAP(+) mononuclear cell numbers compared with cultures treated with SS oligonucleotide or control cultures (Fig. 6D).
Determination of domain(s) of ADAM8 responsible for stimulation of osteoclast formation
As shown in Fig. 7, ADAM8 constructs containing only the MMP domain did not induce OCL stimulation. In contrast, ADAM8 constructs containing the cysteine-rich/integrin domain indicated OCL stimulatory activity, and a truncated form of ADAM8 (Tru4) that contained only the cysteine-rich/disintegrin domain retained OCL stimulatory activity.
PCR-selective subtraction hybridization is a powerful method to identify genes that are overexpressed in one cell population compared with another cell population. This method is similar to traditional differential display RT-PCR but it is more accurate and results in less false positives because of the high stringency PCR conditions used and because it detects positively selected bands. Using this technique with OCLs and OCL precursors, we detected three known genes that can affect OCL formation, the mouse complement C3 gene, which was reported by Suda et al.(14) to increase OCL formation; macrophage inflammatory protein 1-α;(15) and MAP kinase kinase, which is involved in signal transduction.(16) Furthermore, we detected ADAM8 and one novel EST clone that was not previously reported to affect OCL formation and currently is being studied.
ADAM8 is involved in myocyte fusion and expressed in cells of the myelomonocytic lineage. The ADAM family of peptides is a recently described gene family that contains MMP-like, disintegrin-like, cysteine-rich, and EGF-like transmembrane and cytoplasmic domains. These type 1a membrane-bound proteins are involved in cell-to-cell fusion and cell-to-matrix adhesion, have proteolytic activity for degrading the extracellular matrix, and play a role in tissue morphogenesis,(6) possibly including osteoclastogenesis.
The mechanism of action responsible for the effects of ADAM8 on OCL formation has not been defined clearly, but our data suggest that it acts at the later stages of OCL formation when cell fusion rather than proliferation occurs. The finding that the number of TRAP(+) mononuclear cells was increased in marrow cultures treated with AS S-oligonucleotide of ADAM8 is consistent with ADAM8 affecting cell fusion. In support of these observations are reports that several adhesion molecules and proteases play a similar role in cell morphogenesis. For example, cadherins, immunoglobulins, selectins, integrins, and syndecans can induce cell-to-cell fusion and cell-to-matrix adhesion events.(5) Alternatively, Lum et al.(21) reported that ADAM family members can have a tumor necrosis factor α (TNF-α) converting enzyme (TACE)-like activity. This enzyme can cleave TNF-α-related proteins from the cell membrane and then converts to a soluble protein. RANKL, a potent osteoclastogenic factor, is a member of the TNF-α-related protein, which has been reported to be released in small amounts by TACE.(21) This soluble form of RANKL increased the survival of dendritic cells and stimulated OCL formation. However, ADAM8 does not have TACE activity (unpublished results).
ADAM8 also may enhance OCL formation because of MMP activity or be involved in signal transduction through its cytoplasmic domain. However, this seems unlikely. Alternatively, ADAM8 could induce release of osteoclastogenic factors by marrow stromal cells or OCLs, analogous to the release of interleukin-6 (IL-6) by osteoblasts treated with ADAM9.(22) However, we could not detect up-regulation of IL-6, RANKL, or IL-1 in osteoblasts or marrow stromal cells treated with ADAM8 (data not shown).
Our studies with deletion mutants for ADAM8 suggest that the cysteine-rich/integrin domain is involved in ADAM8's effects on OCL formation, but the mechanisms responsible for these effects are unknown. Possibly, the effects of ADAM8 are mediated through interactions between integrin receptors on OCL precursors and the integrin domain of ADAM8 to induce fusion.
In summary, ADAM8 expression is increased in OCLs compared with OCL precursors and stimulates OCL formation in murine bone marrow cultures. ADAM8 naturally occurs as both a membrane and a soluble form, although the soluble form appears to be the predominant form in bone marrow cultures. Blocking ADAM8 expression in marrow cultures inhibits OCL formation at the differentiation/fusion stage, rather than the proliferative stage of the culture, suggesting an important role for ADAM8 in the later stages of osteoclastogenesis.