Portal de Periódicos da CAPES

Expression Profile of Osteoblast Lineage at Defined Stages of Differentiation

2005; Elsevier BV; Volume: 280; Issue: 26 Linguagem: Inglês

10.1074/jbc.m413834200

1083-351X

Ivo Kalajzić, Ada Staal, Wen-Pin Yang, Yuli Wu, Susan E. Johnson, Jean H.M. Feyen, Winfried Krueger, Peter Maye, Fang Yu, Yifang Zhao, Lynn Kuo, Rishi Gupta, Luke E. K. Achenie, Hsin‐Wei Wang, Dong‐Guk Shin, David W. Rowe,

dental development and anomalies

The inherent heterogeneity of bone cells complicates the interpretation of microarray studies designed to identify genes highly associated with osteoblast differentiation. To overcome this problem, we have utilized Col1a1 promoter-green fluorescent protein transgenic mouse lines to isolate bone cells at distinct stages of osteoprogenitor maturation. Comparison of gene expression patterns from unsorted or isolated sorted bone cell populations at days 7 and 17 of calvarial cultures revealed an increased specificity regarding which genes are selectively expressed in a subset of bone cell types during differentiation. Furthermore, distinctly different patterns of gene expression associated with major signaling pathways (Igf1, Bmp, and Wnt) were observed at different levels of maturation. Some of our data differ from current models of osteoprogenitor cell differentiation and emphasize components of the pathways that were not revealed in studies based on a total cell population. Thus, applying methods to generate more homogeneous populations of cells at a defined level of cellular differentiation from a primary osteogenic culture is feasible and leads to a novel interpretation of the gene expression associated with increasing levels of osteoprogenitor maturation. The inherent heterogeneity of bone cells complicates the interpretation of microarray studies designed to identify genes highly associated with osteoblast differentiation. To overcome this problem, we have utilized Col1a1 promoter-green fluorescent protein transgenic mouse lines to isolate bone cells at distinct stages of osteoprogenitor maturation. Comparison of gene expression patterns from unsorted or isolated sorted bone cell populations at days 7 and 17 of calvarial cultures revealed an increased specificity regarding which genes are selectively expressed in a subset of bone cell types during differentiation. Furthermore, distinctly different patterns of gene expression associated with major signaling pathways (Igf1, Bmp, and Wnt) were observed at different levels of maturation. Some of our data differ from current models of osteoprogenitor cell differentiation and emphasize components of the pathways that were not revealed in studies based on a total cell population. Thus, applying methods to generate more homogeneous populations of cells at a defined level of cellular differentiation from a primary osteogenic culture is feasible and leads to a novel interpretation of the gene expression associated with increasing levels of osteoprogenitor maturation. Identification of genes that are associated with the onset of osteoblast differentiation or control the progression of osteoprogenitor cells has been largely based on culture models that involve osteogenic inducers (1Aubin J.E. Biochem. Cell Biol. 1998; 76: 899-910Crossref PubMed Scopus (382) Google Scholar, 2Malaval L. Liu F. Roche P. Aubin J.E. J. Cell. Biochem. 1999; 74: 616-627Crossref PubMed Scopus (249) Google Scholar). The acute response to the addition of bone morphogenetic protein (BMP) 1The abbreviations used are: BMP, bone morphogenetic protein; GFP, green fluorescent protein; Ibsp, integrin-binding sialoprotein; Bglap, bone Gla protein; Dmp1, dentin matrix protein; mCOB, mouse calvarial osteoblast; DMEM, Dulbecco's modified Eagle's medium; FCS, fetal calf serum; FDR, false discovery rate; FACS, fluorescence-activated cell sorter; IGF, insulin-like growth factor; SAM, significance analysis of microarrays; SPH, semiparametric hierarchical. 1The abbreviations used are: BMP, bone morphogenetic protein; GFP, green fluorescent protein; Ibsp, integrin-binding sialoprotein; Bglap, bone Gla protein; Dmp1, dentin matrix protein; mCOB, mouse calvarial osteoblast; DMEM, Dulbecco's modified Eagle's medium; FCS, fetal calf serum; FDR, false discovery rate; FACS, fluorescence-activated cell sorter; IGF, insulin-like growth factor; SAM, significance analysis of microarrays; SPH, semiparametric hierarchical. has been used in a number of immortalized cell lines, and the lineage progression to osteogenic nodule formation has been studied in primary murine, rat, and human cultures when grown in the presence of ascorbic acid, β-glycerophosphate, and dexamethasone. From these studies, a generalized pattern of diminished cell proliferation followed by an increase in the expression of genes associated with bone matrix production has been observed. Progression of differentiation has also been associated with the suppression of genes that are required for differentiation into other cell lineages, including myocyte and adipocyte lineages (3Balint E. Lapointe D. Drissi H. van der Meijden C. Young D.W. van Wijnen A.J. Stein J.L. Stein G.S. Lian J.B. J. Cell. Biochem. 2003; 89: 401-426Crossref PubMed Scopus (146) Google Scholar, 4Harris S.E. Guo D. Harris M.A. Krishnaswamy A. Lichtler A. Front. Biosci. 2003; 8: S1249-S1265Crossref PubMed Scopus (64) Google Scholar, 5Korchynskyi O. Dechering K.J. Sijbers A.M. Olijve W. ten Dijke P. J. Bone Miner. Res. 2003; 18: 1177-1185Crossref PubMed Scopus (51) Google Scholar, 6Peng Y. Kang Q. Cheng H. Li X. Sun M.H. Jiang W. Luu H.H. Park J.Y. Haydon R.C. He T.C. J. Cell. Biochem. 2003; 90: 1149-1165Crossref PubMed Scopus (164) Google Scholar).Several studies highlight the complexity of using mesenchymal stem cells to identify genes associated with bone cell lineage progression and commitment (3Balint E. Lapointe D. Drissi H. van der Meijden C. Young D.W. van Wijnen A.J. Stein J.L. Stein G.S. Lian J.B. J. Cell. Biochem. 2003; 89: 401-426Crossref PubMed Scopus (146) Google Scholar, 4Harris S.E. Guo D. Harris M.A. Krishnaswamy A. Lichtler A. Front. Biosci. 2003; 8: S1249-S1265Crossref PubMed Scopus (64) Google Scholar, 5Korchynskyi O. Dechering K.J. Sijbers A.M. Olijve W. ten Dijke P. J. Bone Miner. Res. 2003; 18: 1177-1185Crossref PubMed Scopus (51) Google Scholar, 6Peng Y. Kang Q. Cheng H. Li X. Sun M.H. Jiang W. Luu H.H. Park J.Y. Haydon R.C. He T.C. J. Cell. Biochem. 2003; 90: 1149-1165Crossref PubMed Scopus (164) Google Scholar, 7Roman-Roman S. Garcia T. Jackson A. Theilhaber J. Rawadi G. Connolly T. Spinella-Jaegle S. Kawai S. Courtois B. Bushnell S. Auberval M. Call K. Baron R. Bone. 2003; 32: 474-482Crossref PubMed Scopus (59) Google Scholar). The major reason for this has to do with the limited number of cells within the culture that become mature osteoblasts, thereby representing only a small proportion of the total population. Therefore, when changes in gene expression are observed, it is never certain whether these changes occur in fully differentiated osteoblasts or other cell populations (7Roman-Roman S. Garcia T. Jackson A. Theilhaber J. Rawadi G. Connolly T. Spinella-Jaegle S. Kawai S. Courtois B. Bushnell S. Auberval M. Call K. Baron R. Bone. 2003; 32: 474-482Crossref PubMed Scopus (59) Google Scholar).The heterogeneity of a primary bone cell culture can be appreciated when different promoter-GFP reporter constructs are used as visual surrogates for osteoblast differentiation. We have previously presented evidence that the GFP driven by a 3.6-kb Col1a1 promoter fragment (pOBCol3.6GFP, referred to as 3.6GFP) becomes visible when alkaline phosphatase-positive cell colonies are detected 5–7 days before the colonies develop into mineralizing multilayered nodules. In contrast, a 2.3-kb Col1a1 promoter fragment driving GFP (pOBCol2.3GFP, referred to as 2.3GFP) remains inactive until the nodules begin to mineralize (8Kalajzic I. Kalajzic Z. Kaliterna M. Gronowicz G. Clark S.H. Lichtler A.C. Rowe D. J. Bone Miner. Res. 2002; 17: 15-25Crossref PubMed Scopus (314) Google Scholar). Based on these expression patterns we have associated the early expression of the 3.6GFP with cells reaching the preosteoblast level of differentiation, which continues during early osteoblast development, whereas the 2.3GFP expression represents a cell that has acquired the ability to form a mineralized matrix. Because the 2.3GFP-positive cells represent less than 15% of the total cell population of a well mineralized osteoblast culture, it is likely that a microarray study based on a heterogeneous cell mixture would not accurately represent the expression of a fully differentiated osteoblast (8Kalajzic I. Kalajzic Z. Kaliterna M. Gronowicz G. Clark S.H. Lichtler A.C. Rowe D. J. Bone Miner. Res. 2002; 17: 15-25Crossref PubMed Scopus (314) Google Scholar).This study was designed to contrast the pattern of gene expression of a mixed cell population with one in which two populations from the same sample are isolated based on their stage of differentiation. Mouse calvarial cultures at days 7 and 17 underwent microarray analysis either as a total cell population or as cells separated based on GFP expression. This study demonstrates the importance of using defined and relatively homogeneous cell populations to more completely understand the gene expression profile of cells at a particular stage of development.MATERIALS AND METHODSCell Culture—Transgenic mice used in this study have been generated and characterized previously (8Kalajzic I. Kalajzic Z. Kaliterna M. Gronowicz G. Clark S.H. Lichtler A.C. Rowe D. J. Bone Miner. Res. 2002; 17: 15-25Crossref PubMed Scopus (314) Google Scholar). Mouse calvarial osteoblast (mCOB) cells were isolated from 6–9-day-old transgenic mice by sequential 0.05% trypsin/collagenase digestions as described previously (8Kalajzic I. Kalajzic Z. Kaliterna M. Gronowicz G. Clark S.H. Lichtler A.C. Rowe D. J. Bone Miner. Res. 2002; 17: 15-25Crossref PubMed Scopus (314) Google Scholar). Cells were plated at a density of 1.5 × 104 cells/cm2, and at 1 week of culture, differentiation was induced with 50 μg/ml ascorbic acid and 4 mm β-glycerophosphate.Preparation of Cells for Sorting—7-Day-old cultures were digested using 0.25% trypsin, 1 mm EDTA for 5–10 min. Trypsin was inactivated by addition of DMEM, 10% FCS followed by centrifugation at +4 °C. Cells were resuspended in phosphate-buffered saline and filtered using a 70-μm strainer. Following a second centrifugation cells were resuspended in phosphate-buffered saline, 2% FCS and filtered through a 45-μm filter. Cells grown for 17 days were digested using a 0.2% collagenase, 0.2% hyalouronidase, 2.5% trypsin. Enzymes were neutralized using DMEM, 10% FCS, and cells were centrifuged and prepared as described above. Cell sorting was done on a FACS Vantage (BD Biosciences) with excitation at 488 nm and 530/30 emission filter. Cells were separated using a 100-μm nozzle and collected into DMEM, 30% FCS media. Prior, during, and following sorting the cell suspensions were kept cold to minimize changes in gene expression.RNA Extraction and Northern Blot Analysis—RNA was extracted using a TRIzol LS reagent (Invitrogen) according to the manufacturer's instructions, and Northern blots were prepared using procedures described previously (8Kalajzic I. Kalajzic Z. Kaliterna M. Gronowicz G. Clark S.H. Lichtler A.C. Rowe D. J. Bone Miner. Res. 2002; 17: 15-25Crossref PubMed Scopus (314) Google Scholar). Membranes were probed with the following fragments: 0.7 kb of GFP, 0.7 kb of EcoRI Igf1 (9Bell G.I. Stempien M.M. Fong N.M. Rall L.B. Nucleic Acids Res. 1986; 14: 7873-7882Crossref PubMed Scopus (172) Google Scholar), 900 bp of PstI rat Col1a1 (pα1R2), (10Genovese C. Rowe D. Kream B. Biochemistry. 1984; 23: 6210-6216Crossref PubMed Scopus (545) Google Scholar), 440 bp of PstI/EcoRI mouse Bglap1 (p923), (11Celeste A.J. Rosen V. Buecker J.L. Kriz R. Wang E.A. Wozney J.M. EMBO J. 1986; 5: 1885-1890Crossref PubMed Scopus (351) Google Scholar), 0.7 kb of Dmp1 (12MacDougall M. Gu T.T. Luan X. Simmons D. Chen J. J. Bone Miner. Res. 1998; 13: 422-431Crossref PubMed Scopus (134) Google Scholar), and 1000 bp of EcoRI mouse Ibsp (13Young M.F. Ibaraki K. Kerr J.M. Lyu M.S. Kozak C.A. Mamm. Genome. 1994; 5: 108-111Crossref PubMed Scopus (94) Google Scholar). Probes were radiolabeled, and hybridization was performed using the protocols described previously (8Kalajzic I. Kalajzic Z. Kaliterna M. Gronowicz G. Clark S.H. Lichtler A.C. Rowe D. J. Bone Miner. Res. 2002; 17: 15-25Crossref PubMed Scopus (314) Google Scholar, 14Sambrook J. Fritsch E.F. Maniatis T. Molecular Cloning: A Laboratory Manual. 1. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY1989: 4.9.1-4.9.9Google Scholar).Array Hybridization and Data Analysis—The cRNA preparation and array hybridization were performed according to the Affymetrix protocol (Affymetrix, Santa Clara, CA). Briefly, cRNA was prepared from 10 μg of total RNA. The RNA was denatured at 70 °C with T7-tagged oligo(dT) primers and then reverse transcribed with Superscript II (Invitrogen) at 42 °C for 1 h. Second-strand cDNA was synthesized by adding DNA polymerase I, Escherichia coli DNA ligase, and RNase H, and incubation was carried out for 2 h at 16 °C. After phenol/chloroform extraction, the synthesized cDNA was used for in vitro transcription using the BioArray high yield RNA transcript labeling kit (Enzo Life Sciences, New York, NY). Labeled cRNA was purified with RNeasy columns (Qiagen) and fragmented before hybridization (10 μg/chip).Affymetrix murine U74A, -B, and -C v2 oligo array cartridges were prehybridized at 45 °C for 10 min and hybridized for 16 h at 45 °C with 60 rpm rotation. After hybridization, the chips were washed and stained in a fluidics station using the antibody amplification protocol from Affymetrix and scanned using a Hewlett-Packard GeneArray scanner. The data were analyzed using GeneChip software MAS4 (Affymetrix). An intensity value and presence/absence call was derived from the hybridization signal for each gene to represent its expression level. Data were normalized with the slope of hybridization intensity of a control sample before comparison. The annotations of Affymetrix chip probe sets were derived from NetAffx (Affymetrix).We applied two statistical methods to select differentially expressed genes. 1) Significance analysis of microarrays (SAM) orders the genes by using the modified t statistic and declares a gene to be up-regulated (down-regulated) if the observed modified t statistic is above (below) the global cutoff point (15Tusher V.G. Tibshirani R. Chu G. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 5116-5121Crossref PubMed Scopus (9705) Google Scholar). This procedure allows estimation of the median of false discovery rates. In this study we selected the cutoff points to control this median to be at most 5%. 2) The semiparametric hierarchical (SPH) mixture model was developed by Newton et al. (16Newton M.A. Noueiry A. Sarkar D. Ahlquist P. Biostatistics. 2004; 5: 155-176Crossref PubMed Scopus (399) Google Scholar) to capture the complicated variations observed in microarray data. According to this method, we first estimated the shape parameter in the γ distribution by the moment method. Then we fit the mixture model by a nonparametric expectation and maximization algorithm. Posterior probabilities for the three hypotheses (un-regulated, up-regulated, down-regulated) were computed numerically for each gene. We ranked the genes according to the posterior probabilities for a hypothesis. To determine the cutoff point, a SPH mixture model controlled the false discovery rate (FDR) from the posterior probabilities. The expected FDR was computed by averaging mistake probabilities from the top to subsequent ones. The cutoff gene was selected when the expected FDR reached at most 5%.RESULTSIsolation of the GFP-positive (GFPpos) and GFP-negative (GFPneg) Cells from Primary Calvarial CultureCultures were established from GFP transgenic mice, and the onset of GFP expression during the osteoblast differentiation was obtained by fluorescent imaging. The 3.6GFP signal activates between day 4 and day 7 of culture (Fig. 1A) in forming colonies and associates with the onset of alkaline phosphatase staining (data not shown). The majority of these colonies will develop into bone nodules. Cells lacking the GFP signal are more prevalent at the periphery and between the developing colonies. At this time, FACS analysis shows that ∼30–40% of cells express the 3.6GFP transgene (Fig. 1C). Following induction of osteogenic differentiation (day 7), the culture develops multilayered nodules that will mineralize. By day 15–17 of culture, Ibsp and Bglap1, molecular markers of mature osteoblast differentiation, are expressed. At this time, about 10% of cells within the culture express the 2.3GFP transgene (Fig. 1, B and E), and these cells are restricted to the mineralizing nodules (Fig. 1B, arrows).GFPpos and GFPneg cells were separated by flow cytometry. FACS reanalysis was carried out on isolated cell populations demonstrating that the sorting process highly enriched GFPpos and GFPneg populations. For 3.6GFP, the percentage of positive cells increased from 30 to >97%, whereas 2.3GFP increased from 10 to >85% (Fig. 1, C–F).Prior to microarray analysis, Northern blot analysis was performed to verify the effectiveness of the sorting process and the quality of the RNA. Fig. 2 shows the presence of a strong GFP mRNA band in the 3.6GFPpos and 2.3GFPpos populations and its absence in the GFPneg population. The Northern blot demonstrated that expression of the type I collagen was similar in the 3.6GFPpos and 3.6GFPneg cells, but it was increased in the 2.3GFPpos and 2.3GFPneg population as well as in the day 17 unsorted population. Interestingly, the expression of Igf1, which showed no changes among sorted or unsorted cell types at day 7 of calvaria culture, was substantially reduced in the 2.3GFPpos population at day 17 of calvaria culture. Additionally, intermediate and late markers of bone cell differentiation including Ibsp, Bglap1, and Dmp1 were strongly expressed in the unsorted culture on day 17 and greatly enriched in the 2.3GFPpos-sorted population.Fig. 2Northern blot analysis of bone-associated gene expression. Bone-related genes were analyzed by Northern blot to compare with the microarray data. Purity of sorted cells was detected by hybridization with GFP cDNA. Bone markers (Col1a1, Ibsp, Dmp1, and Bglap1) as well as Igf1 showed a high correlation with microarray results.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Gene Expression by Microarray AnalysisSeven-day-old cultures were harvested to obtain 3.6GFPpos and GFPneg (at the preosteoblastic stage), and 17-day-old cultures were used to isolate 2.3GFP GFPpos and GFPneg cells at the mature osteoblast stage. The entire process of cell culture, FACS sorting, RNA extraction, probe labeling, and microarray hybridization was performed three times, and the results were analyzed as three replicates. The 74A chip was used to compare the expression profile of the FACS-separated cells with the parental total cell population taken at day 7 and day 17 of culture.3.6GFPpos Versus GFPneg CellsThe two populations of cells were isolated from a 7-day-old culture for RNA extraction. No genes showed higher expression in 3.6GFPpos when compared with GFPneg cells. However, 69 genes showed significant decrease in expression by one or both of the statistical tests, SAM and the SPH mixture model (genes listed in Tables I and II). When genes with lower expression were grouped based on similar decrease in expression in unsorted day 7 to day 17 cell populations, 11 genes were identified, all of which are associated with smooth muscle cells or pericytes (Table I). This type of result has been observed in other microarray studies that use total cell populations (7Roman-Roman S. Garcia T. Jackson A. Theilhaber J. Rawadi G. Connolly T. Spinella-Jaegle S. Kawai S. Courtois B. Bushnell S. Auberval M. Call K. Baron R. Bone. 2003; 32: 474-482Crossref PubMed Scopus (59) Google Scholar) and provides supportive evidence that the sorting process separates bone progenitor cells from other cell types that have lost these progenitor cell markers. The remainder of genes with lower expression in the 3.6GFPpos cells showed an increase in expression as the total cell population moved from day 7 to day 17. Many of these genes encode proteins that modify or interact with the extracellular space or are involved in cytokine signaling. A number of genes are associated with dendritic cells or cells within the macrophage/osteoclast lineage (Table II). Of particular interest are Tyrobp, which is a RANKL adaptor molecule that displays the osteopetrotic phenotype when disrupted (17Kaifu T. Nakahara J. Inui M. Mishima K. Momiyama T. Kaji M. Sugahara A. Koito H. Ujike-Asai A. Nakamura A. Kanazawa K. Tan-Takeuchi K. Iwasaki K. Yokoyama W.M. Kudo A. Fujiwara M. Asou H. Takai T. J. Clin. Investig. 2003; 111: 323-332Crossref PubMed Scopus (282) Google Scholar), and the Csf1 receptor. This latter group of genes may reflect a population of monocyte/osteoclast progenitor cells that continues to increase in number and level of differentiation explaining their increased expression in day 17 culture. However, their fall in expression in the 3.6GFPpos cells suggests that these cells do not directly contribute to the osteoblast lineage.Table IGenes associated with smooth muscle/pericyte level of differentiationNameSymbolDay 7 unsortedDay 17 unsorted3.6GFPneg3.6GFPposActin, α 1, skeletal muscleActa11,7793091,573420Actin, α, cardiacActc11,116301,10814ClusterinClu1,6049861,569505MyoglobinMb26310324376Matrix metalloproteinase 3, stromelysin-1Mmp35,5631,0125,3811,088Myosin, heavy polypeptide 3, skeletal muscle, embryonicMyh31832412712MLC1F/MLC3F gene for myosin alkali light chainMy111,8682412,068344Myosin light chain, alkali, cardiac atriaMyla1,3602341,406311Nik-related kinaseNrk39712641961Troponin C, cardiac/slow skeletalTncc81155929136Troponin T2, cardiacTnnt2281160294104 Open table in a new tab Table IIGenes associated with dendritic or macrophage/osteoclast lineagesNameSymbolDay 7 unsortedDay 17 unsorted3.6GFPneg3.6GFPposA disintegrin and metalloprotease domain 8Adam81,1205,1201,10546Apolipoprotein EApoe2,49166,3311,698389Complement component 1, q subcomponent, a polypeptideClqa3,05325,2012,801347Complement component 1, q subcomponent, β polypeptideC1qβ99712,05482323Complement component 1, q subcomponent, c polypeptideC1qc2,37924,2341,947480Complement component 3a receptor 1C3ar18703,738828185CD53 antigenCd539396,920844296CD68 antigenCd682,23514,5902,101329Carboxypeptidase ECpe4,9166,7714,2891,950Colony-stimulating factor 1 receptorCsf1r1,1899,4691,047350Cytotoxic T lymphocyte-associated protein 2 αCt1a2a278898321121Cytotoxic T lymphocyte-associated protein 2 βCtla2b30397533050Cathepsin CCtsc1,3162,8971,029404Cathepsin SCtss3,13240,0362,286211Adipose fatty acid-binding protein, AP2Fabp4285417369124Glycoprotein 49 AGp49a1,4338,4341,34998Glycoprotein 49 BGp49b7143,63561342Histocompatibility 2, T region locus 18H2-T181,5865,9551,930883Lysosomal-associated protein transmembrane 5Laptm51,89613,2601,938756Lymphocyte cytosolic protein 2Lcp228189434182LIM domain only 2Lmo21,0384,0301,159479Lymphocyte antigen 57Ly572211,68914849Lymphoblastomic leukemiaLyl16043,258527139LysozymeLyzs7,58845,3027,414523P lysozyme structuralLzp-s9,08278,7418,163420Matrix metalloproteinase 12Mmp121,48614,665722122Macrophage C-type lectinMpc12712,72935632Macrophage-expressed gene 1Mpeg11,94217,0531,897426Mannose receptor, C type 1Mrc17798,68755922Macrophage scavenger receptor 1Msr16651,263544174Properdin factor, complementPfc3661,390397143Paired Ig-like receptor A3Pira31,4096,6491,286498Plastin 2, LPls21,4556,5281,243395Small inducible cytokine A12Scya12680967797116Small inducible cytokine A9Scya99854,1611,080231Sialyltransferase 8 (α-2, 8-sialytransferase) DSiat8d17788620474Transferrin receptorTrfr6081,405735359TYRO protein tyrosine kinase-binding proteinTyrobp1,4199,9871,140116 Open table in a new tab 2.3GFPpos Versus 2.3GFPneg CellsA similar comparison of 2.3GFPpos and 2.3GFPneg cells in day 17 cultures was made with total unsorted cultures undergoing differentiation (day 17 versus day 7), yielding a much larger number of differentially expressed genes. Four major patterns of expression were observed.Group A: Genes with Higher Expression in the Unsorted Differentiating Cultures (Day 17 Versus Day 7) That Are Also Highly Expressed in Sorted Mature Osteoblasts (2.3GFPpos Versus 2.3GFPneg)—The expected higher expression of osteoblast-specific genes, Bglap1 and Ibsp, observed in unsorted (day 17 versus day 7) culture was reflected by significantly higher levels in the sorted 2.3GFPpos versus 2.3GFPneg population (supplemental Table 1). Similarly, a number of known osteoblast differentiation-regulated genes such as transcription factor distal-less homeobox 3 (Dlx3), dentin matrix protein 1 (Dmp1), dickkopf homolog 1 (Dkk1), parathyroid hormone receptor 1 (Pthr1), prostaglandin I2 synthase (Ptgis), and phosphate-regulating neutral endopeptidases (Phex) were also found in this category. Interestingly, a few genes with currently no appreciated role in bone biology, such as hepatic lipase (Lipc), bone morphogenetic protein 8a (Bmp8a), leukemia inhibitory factor receptor (Lifr), cysteine dioxygenase 1 (Cdo1), the membrane-bound type 13 collagen (Col13α1), and potassium channel, subfamily K (Kcnk1), were highly expressed in mature osteoblasts (2.3GFPpos).Group B: Genes with Unchanged or Lower Expression in the Unsorted (Day 17 Versus Day 7) Population but with Higher Level of Expression in Mature Osteoblasts—In this category are genes with higher expression in the 2.3GFPpos compared with the 2.3GFPneg population, the change in which is not shown by the total cell population analysis (Table III). A role for many of these genes in this osteoblast function is not obvious, and confirmation of their importance will require gene deletion studies. For example, mice with a total deficiency of the Cpe gene are obese, although no analysis of the bone phenotype has been made. Other genes include an ion transfer gene (chloride channel 3) and cationic amino acid transporter, genes involved in cell metabolic activity (stearoyl-coenzyme A desaturase 1, cytochrome-c oxidase subunit VIa, asparaginase synthetase), a cytoskeletal gene (ankyrin 3 epithelial), and two transcription regulation factors (Fyn proto-oncogene and Kruppel-like factor 4). The increased expression of farnesyl pyrophosphate synthetase and fatty acid-binding protein 3 is noteworthy because of the beneficial effect of statins on bone mass (18Mundy G. Garrett R. Harris S. Chan J. Chen D. Rossini G. Boyce B. Zhao M. Gutierrez G. Science. 1999; 286: 1946-1949Crossref PubMed Scopus (1581) Google Scholar, 19Garrett I.R. Gutierrez G. Mundy G.R. Curr. Pharm. Des. 2001; 7: 715-736Crossref PubMed Scopus (157) Google Scholar).Table IIIGenes with unchanged or lower expression in the unsorted population but with higher level of expression in mature osteoblastsNameSymbolDay 7 unsortedDay 17 unsorted2.3GFPneg2.3GFPposAnkyrin 3, epithelialAnk3522374294800Asparagine synthetaseAsns1,761576292765CD24a antigenCd24a7037939193,056Chloride channel 3Clcn3495545273849Cytochrome-c oxidase, subunit VI aCox6a21,4513932732,097Carboxypeptidase ECpe5,3826,8563,17016,015Fatty acid-binding protein 3Fabp36916856572,984Farnesyl pyrophosphate synthaseFpps13,3495,8745,14911,170Fyn proto-oncogeneFyn1,9682,6921,5005,073Kruppel-like factor 4 (gut)Klf44,3143,8922,0275,868Stearoyl-coenzyme A desaturase 1Scd16,8818,4644,15813,170Solute carrier family 7 (cationic amino acid transporter, y+ system), member 3Slc7a31,213224191548 Open table in a new tab Group C: Genes That Have Lower Expression in Unsorted (Day 17 Versus Day 7) Cultures as Well as in Sorted Mature Osteoblasts—Lower gene expression with increasing cell differentiation likely reflects the loss of progenitor potential and the restriction of the synthetic capacity of the genes characteristic of the differentiated cell type (supplemental Table 2). Thus, observing many genes that decrease in expression during osteoblastic differentiation is not unexpected in either the unsorted or 2.3GFPpos-sorted cell population. Msx2, a transcriptional factor with importance in early osteoblastogenesis is a gene with a known reduction in expression during differentiation. Other genes that are suppressed but have roles in the osteogenic lineage that are less recognized include fibulin 2, necdin, cysteine-rich protein 2, and inhibin β-B.Group D: Genes with Unchanged or Higher Expression Levels in Unsorted Differentiating Cultures (Day 17 Versus Day 7) That Show Significantly Lower Expression in Sorted Mature Osteoblasts—The heterogeneity of osteoblastic culture is evident in genes that show either no change or higher expression in unsorted cultures from day 7 to day 17, when in fact their expression is lower in the isolated 2.3GFPpos cell population (Tables IV and V). Examples include extracellular matrix genes more associated with non-osseous tissues such as procollagen IVa1 (endothelium), matrix Gla protein (smooth muscle cells and chondrocyte lineage cells), fibulin 1 (elastic tissue), thrombospondin and tissue inhibitor of metalloproteinase 2 (wound repair), intracellular adhesion molecule 1 (ICAM1) (endothelial), Meox2 (muscle), Mest (kidney), and osteoglycin (cartilage). This suggests that the 2.3GFPpos cell population is distinguishing itself from a heterogeneous and less differentiated cell population (2.3GFPneg). However, other growth factors sometimes associated with osteoblast differentiation also show lower expression. Included in this category are fibroblast growth factor 7 and two Igf-binding proteins known to affect bone mass. More surprisingly, a group of genes that shows h