Gene Expression Profiling of Osteoclast Differentiation by Combined ...

4 downloads 98 Views 456KB Size Report
Gene Expression Profiling of Osteoclast Differentiation ... and cDNA Microarray Analysis ... ease, and cancer-induced osteolysis (Hofbauer et al., 2001).
DNA AND CELL BIOLOGY Volume 21, Number 8, 2002 © Mary Ann Liebert, Inc. Pp. 541–549

Gene Expression Profiling of Osteoclast Differentiation by Combined Suppression Subtractive Hybridization (SSH) and cDNA Microarray Analysis JAERANG RHO,1,*,† CURTIS R. ALTMANN,2,* NICHOLAS D. SOCCI,3 LUBOMIR MERKOV,4 NACKSUNG KIM,1 HONGSEOB SO,1 OKBOK LEE,1 MASAMICHI TAKAMI,1 ALI H. BRIVANLOU, 2 and YONGWON CHOI1

ABSTRACT Bone homeostasis is maintained by the balanced action of bone-forming osteoblasts and bone-resorbing osteoclasts. Multinucleated, mature osteoclasts develop from hematopoietic stem cells via the monocyte–macrophage lineage, which also give rise to macrophages and dendritic cells. Despite their distinct physiologic roles in bone and the immune system, these cell types share many molecular and biochemical features. To provide insights into how osteoclasts differentiate and function to control bone metabolism, we employed a systematic approach to profile patterns of osteoclast-specific gene expression by combining suppression subtractive hybridization (SSH) and cDNA microarray analysis. Here we examined how gene expression profiles of mature osteoclast differ from macrophage or dendritic cells, how gene expression profiles change during osteoclast differentiation, and how Mitf, a transcription factor critical for osteoclast maturation, affects the gene expression profile. This approach revealed a set of genes coordinately regulated for osteoclast function, some of which have previously been implicated in several bone diseases in humans. INTRODUCTION

O

(OCS ) ARE bone-resorbing multinucleated cells that, together with osteoblasts/stromal cells, remodel bone matrix, a process that is required for proper development and maintenance of vertebrate skeleton and mineral homeostasis (Suda et al., 1997, 1999). Hence, the abnormal activity, development, and function of OCs is implicated in a variety of human diseases including rheumatoid arthritis, periodontal disease, and cancer-induced osteolysis (Hofbauer et al., 2001). Osteoclasts are derived from hematopoietic stem cells of the monocyte–macrophage lineage. The same bone marrow precursor cells can also differentiate to macrophages (MFs) and dendritic cells (DCs) by utilizing distinct but overlapping differentiation factors (Suda et al., 1999; Teitelbaum, 2000). Osteoclast differentiation from bone marrow precursors is controlled by multiple factors produced by cells within a STEOCLASTS

specialized microenvironment of bone (Suda et al., 1999; Teitelbaum, 2000). In particular, macrophage colony stimulating factor (M-CSF) and a TNF family member TRANCE (TNFSF11) produced by osteoblasts were shown to be essential for osteoclast differentiation and maturation (Suda et al., 1999; Teitelbaum, 2000). Osteopetrotic (op/op) mutant mice result from the disruption of the M-CSF gene, are defective osteoclast development, and result in osteopetrosis (Suda et al., 1999; Teitelbaum, 2000; Weilbaecher et al., 2001). Similarly, mice deficient in TRANCE or its receptor, TRANCE-R/RANK, develop severe osteopetrosis due to a failure in osteoclastogenesis (Dougall et al., 1999; Kong et al., 1999; Kim et al., 2000). Moreover, it was shown that osteoblasts deficient in M-CSF or TRANCE fail to support osteoclast formation from bone marrow stem cells of wild-type mice, indicating that osteoblast-derived M-CSF and TRANCE are essential factors for proper osteoclast differentiation in vivo (Suda et al., 1999; Teitelbaum,

1 Abramson Family Cancer Research Institute, Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania. 2 Laboratory of Molecular Vertebrate Embryology, 3Laboratory of Theoretical Condensed Matter Physics, 4Laboratory of Infection Biology, Rockefeller University, New York, New York. *These authors contributed equally to this work. † Current address: Department of Microbiology, Chungnam National University, Daejeon, Korea.

541

542

RHO ET AL.

2000). Although other factors regulating osteoclastogenesis are likely to be produced by osteoclasts in vivo, M-CSF and TRANCE pathways appear to be necessary and sufficient for osteoclast differentiation. Bone marrow precursors differentiate into mature osteoclasts in the absence of supporting osteoblasts when cocultured in M-CSF and TRANCE (Suda et al., 1999; Teitelbaum, 2000). Recent studies on TRANCE-induced signaling pathways have made an important advance in understanding osteoclast differentiation and function. The TRANCE-R/RANK-mediated signal is mainly mediated by recruiting signal adapters, TRAF proteins, which activate NF-kB transcription factors and MAPK kinase cascades (Wong et al., 1999b). In addition, TRANCE induces the activation of c-Src-dependent signal pathways via TRAF6, which plays a critical role in the regulation of osteoclast differentiation and function (Wong et al., 1999a). Both TRAF6 and Src knockout mice show osteopetrotic phenotypes due to defective osteoclast differentiation and function (Lowe et al., 1993; Lomaga et al., 1999; Naito et al., 1999). Despite the recent interest in the large-scale analysis of gene expression in various cell types, such studies in osteoclasts have been hampered, most likely because of the limited number of osteoclasts available for molecular and biochemical studies. Because cells of mature osteoclast properties can now be generated from bone marrow precursor cells in vitro, we sought here to apply a large-scale analysis of gene expression profile to further understand osteoclast differentiation and function. Here, we report mRNA profile analyses of genes during osteoclast differentiation and function using a combined analysis of cDNA microarrays and PCR-based suppression subtractive hybridization (SSH). By simultaneously examining gene expression profiles of mature osteoclasts derived from several independent in vitro differentiation protocols, we sought to obtain a highly selective set of genes expressed in osteoclasts. As validation of the approach, we identified a number of genes previously implicated in various bone-related diseases in humans.

MATERIALS AND METHODS Cell culture, poly A1 RNA preparation and Northern analysis OC, MFs, and DCs were prepared from bone marrow precursors as described (Kim et al., 2002). RAW264.7, a murine monocytic cell line, was also used for differentiation of OClike cells as described (Kim et al., 2002). For comparison of wild-type and Mitf-defective OCs, OCs were derived from splenocytes of wild-type and Mitfmi/mi mice (2–3 weeks of age) as described (Weilbaecher et al., 2001). For this, splenocytes were treated with 30 ng/ml of M-CSF, 1 mg/ml of mouse-soluble TRANCE (sTRANCE), and PGE2 (1026 M) for 6 days. On day 3, the culture was replaced with fresh media containing M-CSF, sTRANCE, and PGE2. OCs were visualized by TRAP staining as described (Kim et al., 2002). For the TRAP solution assay, the cultured OCs were disrupted with 1% Triton X-100 and incubated with p-nitrophenyl phosphate tablet solution dissolved in 50 mM sodium tartrate and 0.1 M sodium acetate buffer (pH 5.0) for 30 min at 37°C. After incubation, 1 N NaOH was added, and the color was measured at 405 nm by ELISA reader as de-

scribed (Li et al., 2000). For the Northern analysis, total RNA (10 mg) or poly A1 RNA (0.5 mg) was separated and transferred to nylon membranes as described (Kim et al., 2002). The membranes were hybridized with [32P] dCTP probe prepared using a Ready-to-go labeling kit and a ProbeQuant G-50 purification kit (Amersham/Pharmacia, Arlington Heights, IL).

Preparation of subtractive cDNA library and microarray A subtractive cDNA library was constructed by PCR-based SSH as described (Kim et al., 2002). In brief, 2 mg of poly A1 RNA from OC and MFs was used to make tester and driver cDNA, respectively. Subtractive PCRs were performed using the PCR-select cDNA subtraction kit according to the manufacturer’s protocol (Clontech, Palo Alto, CA). PCR products are subcloned into the pCR2.1 vector using TA cloning kit (Invitrogen, San Diego, CA). Plasmid DNAs from subtractive OC cDNA library were purified with a Qiagen 96-well preparation kit on a Qiagen 9600 Biorobot. A total of 2304 clones were purified (24 3 96 wells) and their inserts were amplified by a 96well PCR format using the following primers: upstream 59 CAC ACA GGA AAC AGC TAT GAC CAT GAT 39, downstream 59 TTG TAA TAC GAC TCA CTA TAG GGC GA 39. The amplified PCR products were precipitated, and resuspended in 50% DMSO and 0.53 SSC. The DNAs were printed on CMTGAPS aminosilane-coated slides (Corning) as previously described (Altmann et al., 2001).

Preparation of fluorescent probes and hybridization of microarrays Poly A1 RNA (0.5 mg) was labeled with Cy3 or Cy5 monoreactive dyes (Amersham/Pharmacia) using an Atlas glass fluorescent labeling kit (Clontech) following the manufacturer’s protocol with the following modifications. The labeled Cy3 or Cy5 probes were purified with a ProbeQuant G-50 purification kit (Amersham/Pharmacia). The purified probes were dried and resuspended in 20 ml of hybridization solution (25% formamide, 53 SSC, 0.1% SDS, 10 mg of ssDNA). The array was crosslinked with 200 mJ of UV-irradiation and incubated with prehybridization solution (25% formamide, 53 SSC, 0.1% SDS, and 10 mg/ml of BSA) at 42°C for 45 min in a Colpin jar. After the prehybridization, the array was rinsed once with distilled water and 100% ethanol. The array was dried and kept at room temperature until hybridization. The probes were denatured at 99°C for 2 min, cooled on ice, and centrifuged. The supernatant was applied onto the array and covered with a coverglass in a Corning hybridization chamber. Hybridization was performed at 42°C for 18 h. The slides were then washed once with 23 SSC–0.1% SDS at 42°C for 5 min, once with 0.13 SSC–0.1% SDS at room temperature for 10 min, and four times with 0.13 SSC at room temperature for 1 min. Finally, the slides were washed with distilled water, ethanol, and then dried.

Scanning and analysis of DNA microarrays Arrays were scanned with a GMS 418 Array Scanner (Affymetrix). Data were analyzed with Scanalyze (Stanford University) and uploaded for analysis in the TANGO program of the Rockefeller University array analysis system (http:// arrays.rockefeller.edu/xenopus).

GENE EXPRESSION PROFILE IN OSTEOCLASTS

543

FIG. 1. Identification of genes specifically expressed in mature osteoclasts. (A) The experimental scheme employed in this study. Mature osteoclasts (OC) are obtained from bone marrow stem cells of monocyte–macrophage lineage or from the monocytic cell line RAW264.7. Macrophages (MF) and dendritic cells (DC) are derived from bone marrow cells. Gene expression profiles are first compared in (bone marrow-derived OCs versus MFs) and also in (bone marrow-derived OCs versus DCs). Finally, genes identified in this comparative analysis were compared in RAW264.7-derived osteoclasts. Osteoclast genes expressed preferentially in mature OC in both experimental systems (designated by arrows “a” and “b”) are selected for further analysis and listed in Table 1. (B) Northern analysis of OC specific genes. Clones listed in Table 1 were used to prepare probes and hybridized to OC, MF, or DC RNA. Poly A1 RNAs (0.5 mg) were transferred to nylon membrane, and hybridized. GAPDH, glycerolaldehyde-3-phosphate dehydrogenase was used as controls. (C) Northern analysis of genes expressed in both OC and MF. (D) Northern analysis of genes expressed in both OC and DC.

544

RHO ET AL.

RESULTS Identification of genes highly expressed in mature osteoclasts To identify genes specifically expressed in osteoclasts more efficiently, we have generated a cDNA microarray from cDNA clones produced by suppression subtractive hybridization (SSH). A subtractive OC cDNA library of around 3000 clones was generated from bone marrow-derived mature osteoclasts (tester) and bone marrow-derived macrophages (driver) by

TABLE 1. GENES PREFERENTIALLY EXPRESSED

Group

IN

PCR-based SSH. A total of 2304 clones were isolated and printed in triplicate to generate an array containing 6912 spots. Each array was hybridized in duplicate using inverse dye-labeled probes to generate six data points for each clone, and data were uploaded for analysis in the TANGO system (Altmann et al., 2001). These arrays were used to identify genes differentially expressed in mature OC, compared to MF or DC, and the precursor cell line, RAW264.7. The overall experimental scheme for identifying osteoclast-specific genes used in this study is presented in Figure 1A.

BOTH BONE MARROW -DERIVED OCS

AND

RAW264.7-DERIVED OCS

GenBank accession number

Number of clones

OCa



DC

Clone identity

Selection criteria

NM_007388 AJ006033 NM_009801 Z27231 BC013341 AF218254 U13840 U13841

123 114 2 42 67 2 13 2

111 111 111 111 111 N.D.b 111 111

2 2 2 2 2 N.D. 1 1

6 2 2 2 1 N.D. 1 1

TRAP Cathepsin K Carbonic anhydrase 2 Collagenase type IV Arginase 1 V-ATPase subunit a3 V-ATPase subunit D-like V-ATPase subunit E

a, bc a, b a, b a, b a, b a a, b a, b

NM_011930 AF391159 AF274752 BAB29508

2 3 5 10

111 111 11 111

1 2 2 2

1 2 2 6

CIC-7 OSCAR ANK DC-STAMP

a, b a a, b a, b

M62470 AF270513

24 1

N.D. 6

N.D. 2

N.D. 11

Thrombospondin (THBS1) EMILIN-2

a, b a, b

NM_005719 NM_053961

1 1

111 11

6 2

2 2

a, b a, b

AF144695 AF143409 NP_035634

1 2 2

N.D. 111 111

N.D. 1 1

N.D. 11 11

AK017735

1

111

1

1

p21-ARC Endoplasmic reticulum protein ERp29 ERO1L Mediterranean fever Syntaxin binding protein 3 homolog DnaJ 10 like

AF048696 U88908

1 3

11 N.D.

1 N.D.

N.D. N.D.

Rac2 Inhibitor of apoptosis protein-1 (IAP-1)

a a, b

BC004807 AK018202 AK012515 AB020657

5 3 1 1

1 111 11 111

2 11 11 11

1 111 1 1

Putative protein Putative protein Putative protein Putative protein

a, b a, b a a, b

BC006966 NM_011909

1 2

11 N.D.

2 N.D.

1 N.D.

a, b a, b

NM_020615

1

N.D.

N.D.

N.D.

Glutathione reductase 1 Ubiquitin specific protease 18 (USP18) ATP synthase (ATP5C1)

Northern analysis

Osteoclast function

Transmembrane protein

Extracellular matrix Protein secretion and cytoskeleton

a, b a, b a, b a

Signal transduction

Putative protein

Other

a

a These are the data from Northern analysis on mature osteoclast (OC), macrophase (MØ), and dendritic cell (DC) that are derived from bone marrow precursors. b N.D.: not determined. 111: high, 11: moderate, 1: low, and 2: no expression. c Clones selected are marked by selection criteria in Figure 1A.

FIG. 2. Differentiation of RAW264.7 cells in culture and identification of differentially expressed genes. TRAP-negative (TRAP2 ) cells were cultured for 4 days in the presence of sTRANCE to induce the formation of osteoclast-like cells. (A) The development of osteoclast-like cells was confirmed by the presence of multinucleated TRAP-positive (TRAP1 ) cell staining. (B) Cells shown in A were lysed and used to measure TRAP activity. The activity of TRAP increased rapidly, beginning on day 2. (C) The number of regulated spots and the fold change identified increases during osteoclast differentiation.

FIG. 3. Cluster analysis of cDNA microarray and Northern analysis of temporal changes in gene expression during osteoclast differentiation. (A) Clusters showing the different types of gene expression profiles obtained. For each group, the kinetic points (0, 0.5, 1, 2, 3, and 4 day) are represented. Expression levels greater than the mean are shaded in gray. USP18; ubiquitin specific protease 18, UCE12; ubiquitin conjugating enzyme 12, LDH-A; lactate dehydrogenase A, PARL; presenilins associated rhomoid-like protein. Unidentified clones in GenBank database are marked as clone name. (B) The clusters enhanced by TRANCE activation are graphically shown. The scale indicates SDs above or below the mean. (C) Northern analysis of the selected genes in TRANCe enhanced clusters. GAPDH, glycerolaldehyde-3-phosphate dehydrogenase, was used as controls.

546 When cDNA probes from OC and MF were used, a total of 2009 spots were identified as differentially expressed at least threefold with two times the standard deviation (SD) of the background, and represented 854 clones (average 2.4 spots per clone). Comparing the mRNA expression profile between mature OC and DC identified 2632 spots with threefold differences in their expression levels and represented a total of 811 clones (average 2.9 spots per clone). As depicted in Figure 1A, these clones were further compared to the data from the mRNA profile analysis of OC differentiation, in which the gene expression pattern of mature OC-like cells is compared to that in the OC precursor cell line, RAW264.7. By comparing gene expression profiles of three independent experimental designs, we sought to increase the accuracy for the selection of osteoclastspecific clones. The clones in the criteria set in Figure 1A were sequenced, and genes showing threefold higher expression in osteoclasts were listed in Table 1. Most of the clones selected by the experimental scheme represented OC-specific or OC-enriched genes (Fig. 1B, Table 1). These results suggest that the comparison of gene expression profiles in osteoclasts with their closely related cell types, and with osteoclasts derived by independent in vitro culture methods, was quite efficient in identifying genes important for osteoclasts. Some of the clones showed predominant expression in both OC and MF or in both OC and DC (Fig. 1C and D), as predicted from the scheme shown in Figure 1A. Therefore, a similar experimental scheme of combinatorial analysis of various cell types as shown here may enhance the efficiency of rapid identification of differentially expressed genes. The most abundant clones identified were cathepsin K, tartrate-resistant acid phosphatase, arginase 1, collagenase type IV, thrombospondin, and vacuolar H1 -ATPase subunits, all of which are previously known to play critical roles in osteoclast function (Teitelbaum, 2000; Alper, 2002). Among genes that were not previously implicated in osteoclast development and function were several transmembrane proteins identified to be predominantly expressed in osteoclasts. OSCAR is a member of the Ig-like protein in the leukocyte receptor complex, which has been recently shown to regulate OC differentiation and function (Kim et al., 2002). ClC-7 and ANK were previously identified to play critical roles in bone homeostasis (Kornak et al., 2001; Nurnberg et al., 2001). Loss of function of a chloride channel protein ClC-7 results in osteopetrosis, and mutations in this gene are associated with human infantile malignant osteopetrosis and retinal degeneration (Kornak et al., 2001). ANK is another multipass transmembrane protein implicated in arthritis. Mutations of ANK lead to progressive ankylosis in the mouse (Ho et al., 2000). Although the function of ANK in osteoclast per se was not examined, our data suggest its potential involvement in osteoclast function. A number of genes involved in cellular trafficking, protein folding, and cytoskeleton were highly expressed in osteoclasts, and may be related to the bone resorption process acquired in mature osteoclasts, which involves the secretion of various degradative enzymes and matrix binding proteins, and also the high motility of mature osteoclasts during bone remodeling. For example, the actin-related protein 2/3 complex subunit 3 (p21-ARC), a component of the Arp2/3 complex, is important for assembly of actin filaments in eukaryotes and may regulate the formation of actin rings in OC, which is critical for bone

RHO ET AL. resorption (Teitelbaum, 2000; Volkmann et al., 2001). Syntaxin 3, a member of the SNARE family of proteins, resides on the plasma membrane and is known to be involved in vesicle transport and exocytosis (Hepp and Langley, 2001). Rac-2 is a GTP binding protein of the Rho family, which mediates signaling pathways, regulating a variety of cellular processes, many of which are associated with dynamic cytoskeletal reorganization (Price and Collard, 2001; Ridley, 2001). Interestingly, the Mediterranean fever gene associates with microtubules responsible and colocalizes with actin filaments, and is highly expressed in OCs (Centola et al., 2000). Mutations in this gene are responsible Familial Mediterranean fever (FMF); whether there is any defect in OC function in FMF remains to be determined. Among the genes identified were many with no identified domains or homologies. Most of these genes had been identified by EST sequencing strategies, particularly the efforts from the RIKEN group in Japan, which relied on a novel cap trapper approach to isolated full-length genes (Carninci and Hayashizaki, 1999; Carninci et al., 2000). Four putative clones (BC004807, AK018202, AK012515, and AB020657) were identified but are not predicted to have known domains as determined by SMART (a Simple Modular Architecture Research Tool) (Schultz et al., 1998; Ponting et al., 1999).

Temporal analysis of gene expression during osteoclast differentiation We examined temporal gene expression changes during OC differentiation in an attempt to distinguish genes expressed early (presumably involved in fate determination and differentiation) from those expressed late (potentially involved in osteoclast function). We examined the temporal gene expression pattern of RAW264.7 cells treated with sTRANCE over a period of 4 days, at which point RAW264.7 cells become mature OCs (Fig. 2) (Kim et al., 2002). The differentiation of osteoclasts was confirmed histologically (Fig. 2A). TRAP positive cells were first identified on day 2, and the numbers increased subsequently (Fig. 2B). Beginning at 12 h after TRANCE stimulation, regulated (both up and down) target spots were identified, and both the number and level of differentially expressed target increased over the 4-day period. At the end of 4 days, over 200 (,3% spots) spots were identified as being regulated at least 10-fold, and over 650 spots were identified as regulated at least threefold (,9–10%) (Fig. 2C). To differentiate the gene expression changes and distinguish different classes of expression we performed a nonlinear normalization to each individual array experiment and clustered using a k-means-based algorithm. Initially, sequence information was not included in the analysis as a control so that we could identify meaningful clusters by the inclusion of duplicate clones. To further identify appropriate clustering, clusters were analyzed to identify local versus global minima. These results suggested that the majority of the genes fell into eight general classes of two categories: upregulated and downregulated data (data not shown). This unbiased approach correctly included the previously characterized osteoclast genes into two main clusters. As a final step, we combined the duplicates, which cosegregated as a single value to simplify the output and eliminate redundancy. Subsequent sequencing further confirmed the

GENE EXPRESSION PROFILE IN OSTEOCLASTS validity of the approach by identifying additional TRAP clones. The results of the two clusters, which included all of the previously characterized genes, are presented in Figure 3A and B. To confirm the temporal expression determined by cDNA microarray, Northern analysis was performed (Fig. 3C). Although many of the genes identified in the bone marrow-derived mature OCs and RAW264.7-derived OCs were identical, some regulated genes were not shared due to differences in the selection criteria. The clusters containing downregulated genes included the apoptosis inhibitor MDM2 as well as clecsf9, dchil. These genes may be involved in the initial specification of the OC cell rather than the function or terminal differentiation of these cells. The involvement of these genes in osteoclast development and differentiation are being examined further.

Microarray analysis of Mitf mi / mi mutant mice To identify gene expression changes associated with transcription factor Mitf loss of function, cDNA microarray analysis was performed on OCs derived from wild-type and Mitfmi/mi mice. Mitf function is required for proper craniofacial development, and Mitfmi/mi mice have Microphthalmia. In addition

547 to the facial defects, terminal osteoclast differentiation is affected in Mitfmi/mi mice (Luchin et al., 2000). Although TRAP positive OCs can be generated from Mitf-deficient precursor cells, Mitfmi/mi OCs are significantly smaller and contain less nuclei when compared to wild-type mature OCs (Fig. 4A and B). In addition, Mitfmi/mi OCs show an irregular ruffle border, suggesting defective bone resorption (Motyckova et al., 2001). Array analysis identified a number of genes whose expression is significantly decreased in Mitfmi/mi mice (Table 2, Fig. 4C). Consistent with previous reports, TRAP and cathepsin K were identified by microarray analysis to be less expressed in Mitfmi/mi OCs (Luchin et al., 2000; Motyckova et al., 2001). In addition, we showed here that the expression level of bone matrix degradation enzyme collagenase IV is also reduced in Mitfmi/mi OCs, which may account for their defects in bone resorption. Moreover, Mitfmi/mi OCs have impaired expression of several proteins required for the interaction of mature osteoclasts with extracellular bone matrix, which is important for proper function of osteoclasts (Suda et al., 1999; Teitelbaum, 2000). Thus, it appears that the transcription factor Mitf regulates the expression of multiple genes required for proper function of mature osteoclasts. Expression of these genes may be directly regulated by the transcription factor Mitf via the E-box

FIG. 4. Gene expression profile in Mitfmi/mi mice. (A) Mitfmi/mi mice-derived OCs showed mono- or small-size multinucleated osteoclast morphology. Splenocytes from 2–3-week-old wild-type and Mitfmi/mi mice were cultured with sTRANCE and M-CSF where indicated. Arrows indicate multinucleated osteoclasts. (B) Total number of TRAP-positive multinucleated osteoclasts (MNC) are shown. (C) Northern analysis of genes whose expression is downregulated in Mitfmi/mi OCs. Poly A1 RNAs (0.5 mg) were transferred to nylon membrane, and hybridized. Mitf and GAPDH, glycerolaldehyde-3-phosphate dehydrogenase, were used as controls.

548

RHO ET AL. TABLE 2.

Group

MITFmi/mi REGULATED GENES

Gene

GenBank no.

Description

TRAP Cathepsin K Collagenase IV Arginase 1 Osteopontin Thrombospondin Tumor protein D52 Calveolin-1 alpha Lysozyme ERp29

NM_007388 AJ006033 Z27231 BC013341 X51834 M62470 U44426 AB029929 M21050 NM_053961

Bone resorption Bone resorption Bone resorption Bone resorption Bone matrix protein Bone matrix protein Breast carcinoma-associated D52 Metastasis regulated gene Lysosomal vesicle protein Endoplasmic reticulum protein

Lymphocyte antigen 6 M

AF118557

Lymphocyte marker

Downregulated genes

Upregulated gene

in their promoters, or indirectly through the loss or decreased expression of secondary signaling molecules (Luchin et al., 2000; Motyckova et al., 2001; Weilbaecher et al., 2001). It has been reported that TRAP and cathepsin K gene expression is regulated by Mitf directly via E-box elements in their promoter regions (Motyckova et al., 2001; Mansky et al., 2002). Preliminary examination of the promoter regions of collagenase IV, thrombospondin, and osteopontin (1–3 kb upstream of the transcription start site, GenBank accession numbers AF403768, M62449, and D14816, respectively) led to the identification of E-box consensus Mitf binding sites, which may function for their transcriptional activation in mature OCs.

DISCUSSION We have presented a global analysis of genes involved in the differentiation of OC cells under the influence of TRANCE from both in vivo-derived precursors as well as a tissue culture model using RAW cells. By employing combinatorial analysis of various related cell types, we have identified genes whose expression is highly enriched in mature osteoclasts such as TRAP, cathepsin K, arginase, collagenase type IV, and V-ATPase subunits. Therefore, combinatorial schemes similar to that presented in this study are likely to enhance the efficiency of rapid identification of differentially expressed genes using cDNA microarray. In our study, three genes that were previously associated with bone-related diseases in humans were isolated to be highly expressed in osteoclasts. Loss of function of a chloride channel protein ClC-7 is associated with human infantile malignant osteopetrosis and retinal degeneration (Kornak et al., 2001). ANK, mutation of which lead to progressive ankylosis in the mouse, is a multitransmembrane protein (Ho et al., 2000; Nurnberg et al., 2001). Mediterranean fever gene is responsible Familial Mediterranean fever (FMF) and colocalizes with actin filaments, being highly expressed in OCs (Centola et al., 2000). Although the role of ClC-7 is a regulator of bone resorption by osteoclasts has been established, the importance of ANK and Mediterranean fever gene in osteoclast remains to be determined. Because our data contained a highly selective set of

genes for osteoclasts, it suggests potential roles of ANK and Mediterranean fever genes regulating bone resorption. In summary, we generated a cDNA microarray for osteoclast differentiation using a PCR-based SSH approach. By comparing cells obtained by multiple experimental schemes simultaneously, a highly selective set of genes specifically expressed in mature osteoclasts are identified. Many of the known genes previously implicated in osteoclast function are identified in this study. Our results are likely to aid the dissection of the function of osteoclasts and the cause of various bone-related diseases in humans.

ACKNOWLEDGMENTS We thank A. Santana for technical assistance. This work was supported in part by grants from HHMI, NIH, University of Pennsylvania (to Y. Choi), HD 32105 and EY12370 (to A. Brivanlou). L. Merkov is a Predoctoral Fellow of the Howard Hughes Medical Institute. We would like to thank Dr. M.C. Ostrowski (Ohio State University) for Mitfmi/mi mice.

REFERENCES ALPER, S.L. (2002). Genetic diseases of acid-base transporters. Annu. Rev. Physiol. 64, 899–923. ALTMANN, C.R., BELL, E., SCZYRBA, A., PUN, J., BEKIRANOV, S., GAASTERLAND, T., and BRIVANLOU, A.H. (2001). Microarray-based analysis of early development in Xenopus laevis. Dev. Biol. 236, 64–75. CARNINCI, P., and HAYASHIZAKI, Y. (1999). High-efficiency fulllength cDNA cloning, Methods Enzymol. 303, 19–44. CARNINCI, P., SHIBATA, Y., HAYATSU, N., SUGAHARA, Y., SHIBATA, K., ITOH, M., KONNO, H., OKAZAKI, Y., MURAMATSU, M., and HAYASHIZAKI, Y. (2000). Normalization and subtraction of cap-trapper-selected cDNAs to prepare full-length cDNA libraries for rapid discovery of new genes. Genome Res. 10, 1617–1630. CENTOLA, M., WOOD, G., FRUCHT, D.M., GALON, J., ARINGER, M., FARRELL, C., KINGMA, D.W., HORWITZ, M.E., MANSFIELD, E., HOLLAND, S.M., et al. (2000). The gene for familial

GENE EXPRESSION PROFILE IN OSTEOCLASTS Mediterranean fever, MEFV, is expressed in early leukocyte development and is regulated in response to inflammatory mediators. Blood 95, 3223–3231. DOUGALL, W.C., GLACCUM, M., CHARRIER, K., ROHRBACH, K., BRASEL, K., DE SMEDT, T., DARO, E., SMITH, J., TOMETSKO, M.E., MALISZEWSKI, C.R., et al. (1999). RANK is essential for osteoclast and lymph node development. Genes Dev. 13, 2412–2424. HEPP, R., and LANGLEY, K. (2001). SNAREs during development. Cell Tissue Res. 305, 247–253. HO, A.M., JOHNSON, M.D., and KINGSLEY, D.M. (2000). Role of the mouse ank gene in control of tissue calcification and arthritis. Science 289, 265–270. HOFBAUER, L.C., NEBAUER, A., and HEUFELDER, A.E. (2001). Receptor activator of nuclear factor-kappaB ligand and osteoprotegerin: Potential implications for the pathogenesis and treatment of malignant bone diseases. Cancer 932, 460–470. KIM, N., ODGREN, P.R., KIM, D.K., MARKS, S.C., JR., and CHOI, Y. (2000). Diverse roles of the tumor necrosis factor family member TRANCE in skeletal physiology revealed by TRANCE deficiency and partial rescue by a lymphocyte-expressed TRANCE transgene. Proc. Natl. Acad. Sci. USA 97, 10905–10910. KIM, N., TAKAMI, M., RHO, J., JOSIEN, R., and CHOI, Y. (2002). A novel member of the leukocyte receptor complex regulates osteoclast differentiation. J. Exp. Med. 195, 201–209. KONG, Y.Y., YOSHIDA, H., SAROSI, I., TAN, H.L., TIMMS, E., CAPPARELLI, C., MORONY, S., OLIVEIRA-DOS-SANTOS, A.J., VAN G., ITIE, A., et al. (1999). OPGL is a key regulator of osteoclastogenesis, lymphocyte development and lymph-node organogenesis. Nature 397, 315–323. KORNAK, U., KASPER, D., BOSL, M.R., KAISER, E., SCHWEIZER, M., SCHULZ, A., FRIEDRICH, W., DELLING, G., and JENTSCH, T.J. (2001). Loss of the ClC-7 chloride channel leads to osteopetrosis in mice and man. Cell 104, 205–215. LI, J., SAROSI, I., YAN, X.Q., MORONY, S., CAPPARELLI, C., TAN, H.L., MCCABE, S., ELLIOTT, R., SCULLY, S., VAN, G., et al. (2000). RANK is the intrinsic hematopoietic cell surface receptor that controls osteoclastogenesis and regulation of bone mass and calcium metabolism. Proc. Natl. Acad. Sci. USA 97, 1566–1571. LOMAGA, M.A., YEH, W.C., SAROSI, I., DUNCAN, G.S., FURLONGER, C., HO, A., MORONY, S., CAPPARELLI, C., VAN, G., KAUFMAN, S., et al. (1999). TRAF6 deficiency results in osteopetrosis and defective interleukin-1, CD40, and LPS signaling. Genes Dev. 13, 1015–1024. LOWE, C., YONEDA, T., BOYCE, B.F., CHEN, H., MUNDY, G.R., and SORIANO, P. (1993). Osteopetrosis in Src-deficient mice is due to an autonomous defect of osteoclasts. Proc. Natl. Acad. Sci. USA 90, 4485–4489. LUCHIN, A., PURDOM, G., MURPHY, K., CLARK, M.Y., ANGEL, N., CASSADY, A.I., HUME, D.A., and OSTROWSKI, M.C. (2000). The microphthalmia transcription factor regulates expression of the tartrate-resistant acid phosphatase gene during terminal differentiation of osteoclasts. J. Bone Miner. Res. 15, 451–460. MANSKY, K.C., SULZBACHER, S., PURDOM, G., NELSEN, L., HUME, D.A., REHLI, M., and OSTROWSKI, M.C. (2002). The microphthalmia transcription factor and the related helix-loop-helix zipper factors TFE-3 and TFE-C collaborate to activate the tartrate-resistant acid phosphatase promoter. J. Leukoc. Biol. 171, 304–310. MOTYCKOVA, G., WEILBAECHER, K.N., HORSTMANN, M., RIEMAN, D.J., FISHER, D.Z., and FISHER, D.E. (2001). Linking osteopetrosis and pycnodysostosis: Regulation of cathepsin K expression by the microphthalmia transcription factor family. Proc. Natl. Acad. Sci. USA 98, 5798–5803.

549 NAITO, A., AZUMA, S., TANAKA, S., MIYAZAKI, T., TAKAKI, S., TAKATSU, K., NAKAO, K., NAKAMURA, K., KATSUKI, M., YAMAMOTO, T., and INOUE, J. (1999). Severe osteopetrosis, defective interleukin-1 signalling and lymph node organogenesis in TRAF6-deficient mice. Genes Cells 4, 353–362. NURNBERG, P., THIELE, H., CHANDLER, D., HOHNE, W., CUNNINGHAM, M.L., RITTER, H., LESCHIK, G., UHLMANN, K., MISCHUNG, C., HARROP, K., et al. (2001). Heterozygous mutations in ANKH, the human ortholog of the mouse progressive ankylosis gene, result in craniometaphyseal dysplasia. Nat. Genet. 28, 37–41. PONTING, C.P., SCHULTZ, J., MILPETZ, F., and BORK, P. (1999). SMART: Identification and annotation of domains from signalling and extracellular protein sequences. Nucleic Acids Res. 27, 229–232. PRICE, L.S., and COLLARD, J.G. (2001). Regulation of the cytoskeleton by Rho-family GTPases: Implications for tumour cell invasion. Semin. Cancer Biol. 11, 167–173. RIDLEY, A.J. (2001). Rho proteins: Linking signaling with membrane trafficking. Traffic 2, 303–310. SCHULTZ, J., MILPETZ, F., BORK, P., and PONTING, C.P. (1998). SMART, a simple modular architecture research tool: Identification of signaling domains. Proc. Natl. Acad. Sci. USA 95, 5857–5864. SUDA, T., NAKAMURA, I., JIMI, E., and TAKAHASHI, N. (1997). Regulation of osteoclast function. J. Bone Miner. Res. 12, 869–879. SUDA, T., TAKAHASHI, N., UDAGAWA, N., JIMI, E., GILLESPIE, M.T., and MARTIN, T.J. (1999). Modulation of osteoclast differentiation and function by the new members of the tumor necrosis factor receptor and ligand families. Endocr. Rev. 20, 345–357. TEITELBAUM, S.L. (2000). Bone resorption by osteoclasts. Science 289, 1504–1508. VOLKMANN, N., AMANN, K.J., STOILOVA-MCPHIE, S., EGILE, C., WINTER, D.C., HAZELWOOD, L., HEUSER, J.E., LI, R., POLLARD, T.D., and HANEIN, D. (2001). Structure of Arp2/3 complex in its activated state and in actin filament branch junctions. Science 293, 2456–2459. WEILBAECHER, K.N., MOTYCKOVA, G., HUBER, W.E., TAKEMOTO, C.M., HEMESATH, T.J., XU, Y., HERSHEY, C.L., DOWLAND, N.R., WELLS, A.G., and FISHER, D.E. (2001). Linkage of M-CSF signaling to Mitf, TFE3, and the osteoclast defect in Mitf(mi/mi) mice. Mol. Cell 8, 749–758. WONG, B.R., BESSER, D., KIM, N., ARRON, J.R., VOLOGODSKAIA, M., HANAFUSA, H., and CHOI, Y. (1999a). TRANCE, a TNF family member, activates Akt/PKB through a signaling complex involving TRAF6 and c-Src. Mol. Cell 4, 1041–1049. WONG, B.R., JOSIEN, R., and CHOI, Y. (1999b). TRANCE is a TNF family member that regulates dendritic cell and osteoclast function. J. Leukoc. Biol. 65, 715–724.

Address reprint requests to: Yongwon Choi, Ph.D. Abramson Family Cancer Research Institute Department of Pathology and Laboratory Medicine University of Pennsylvania School of Medicine Room 308, BRB II/III 421 Curie Boulevard Philadelphia, PA 19104-6160 E-mail: [email protected] Received for publication March 18, 2002; accepted May 10, 2002.