- Email: [email protected]
Gene expression in human osteoblastic cells from normal and heterotopic ossification
Experimental and Molecular Pathology 76 (2004) 37 – 43 www.elsevier.com/locate/yexmp
Gene expression in human osteoblastic cells from normal and heterotopic ossification Christophe Chauveau, a Jean-Christophe Devedjian, a Marie-Claude Blary, a Christophe Delecourt, b Pierre Hardouin, a Joseph Jeanfils, a and Odile Broux a,* a
LR2B-Laboratoire de Recherche sur les Biomate´riaux et les Biotechnologies, Universite´ du Littoral Coˆte d’Opale, Boulogne-sur-mer et Berck-sur-mer, France b Etablissement He´lio-marin, Groupe Hopale, Berck-sur-mer, France Received 6 September 2003
Abstract Heterotopic ossification (HO), a possible complication of head injury, develops in sites where it is not normally present like at the vicinity of joints. It may cause pain, decrease motion and in severe cases complete joint ankylosis requiring surgical intervention. To our knowledge, no study has been made to analyze HO at the molecular level on human biopsies, whereas its etiology remains to be determined. We defined a procedure of cell fractionation from bone resections and developed quantitative RT-PCR to compare genetic expression patterns between human normal osteoblasts and heterotopic ossification forming cells. This quantitative study demonstrated a specific and strong overexpression of osteocalcin mRNA in HO-isolated cells associated with a significant upregulation of type 1 collagen and osteonectin mRNA while histological analysis showed only small cellular variations. Our results give a first molecular characterization of heterotopic ossification and we conclude that such overexpressions in HO-isolated cells could be associated with the high activity of this pathological bone. D 2003 Elsevier Inc. All rights reserved. Keywords: Heterotopic ossification; Head injury; Quantitative RT-PCR; Osteocalcin mRNA
Introduction Heterotopic ossification (HO) is characterized by the formation of bone in sites where it is not normally present. HO may be classified in a rare hereditary form or a most common acquired form. There are at least three distinct genetic hereditary forms in humans: fibrodysplasia ossificans progressiva (Cohen et al., 1993), progressive osseous heteroplasia (Kaplan et al., 1994) and Albright hereditary osteodystrophy (Brook and Valman, 1971). In acquired forms, HO frequently occurs after neurologic injury such as spinal trauma or head injury. It is commonly near a joint
* Corresponding author. LR2B-Laboratoire de Recherche sur les Biomate´riaux et les Biotechnologies, Universite´ du Littoral Coˆte d’Opale, Quai Masset, Bassin Napole´on, BP 120, 62327 Boulogne-sur-mer et Bercksur-mer, Cedex, France. Fax: +33-321994524. E-mail address: [email protected] (O. Broux). 0014-4800/$ - see front matter D 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.yexmp.2003.10.001
and may cause pain, decrease motion and, in severe cases, may require surgical resection (Ippolito et al., 1999; Pittenger, 1991; Sawyer et al., 1991; Shehab et al., 2002). The biological mechanisms of acquired HO formation remain to be determined. To understand the etiology of acquired heterotopic bone formation, studies focused mainly on two approaches: the research of a humoral factor as an inductive agent in neurogenic HO and the measure of osteoblast activity in HOisolated cells. For example, all the in vitro culture systems but one (Renfree et al., 1994) demonstrated a positive effect of sera, from patients with heterotopic ossification, on osteoblast proliferation (Bidner et al., 1990; Kurer et al., 1992). However, the osteoinductive factor, which could perhaps enhance osteogenesis, has not been identified. Other in vitro studies showed that human cells isolated from HO exhibit, in culture, increased osteoblastic activities, such as alkaline phosphatase activity and type 1 collagen synthesis, when compared with human normal osteoblasts (Kaysinger et al.,
38
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
1997; Sell et al., 1998). Nevertheless, to our knowledge, no study has been made to analyze neurogenic HO at the molecular level on human biopsies. In recent years, advances in human genetics led to the identification of many genes involved in the different steps of osteoblast differentiation. At the same time, a variety of powerful and sensitive methods to quantify mRNA expression has been developed. With these new tools at hands, we performed a study comparing genetic expression patterns of human HO-forming cells and of human normal osteoblasts. We first defined a procedure of cell fractionation from bone resections, allowing as assessed by histological analysis to obtain homogeneous osteoblastic populations that can be compared. Quantitative RT-PCR analysis was performed to determine mRNA expression levels of osteoblast markers. Some genes representative of the successive stages of osteoblastic differentiation have been chosen. Endoglin (END), identified by immunoprecipitation using the SH2 antibody, represents a marker which can discriminate undifferentiated human osteoblastic precursors and more mature bone cells (Barry et al., 1999; Haynesworth et al., 1992). Type 1 collagen (COL I), the major constituent of bone matrix, and alkaline phosphatase activity (ALP) are associated with matrix maturation and decrease when mineralization is in progress. Osteonectin (ON) is expressed at maximal levels during mineralization while osteocalcin (OC), a noncollagenous protein synthesized exclusively by osteoblasts, appears at a more final stage. Some of these genes were described as regulated by core-binding factor alpha 1 (Cbfa1) and we also quantified the expression level of this essential osteogenesis transcription factor (Ducy et al., 1997; Komori et al., 1997, Otto et al., 1997). The genetic expression patterns obtained by quantification of this panel of markers were relevant to the restricted cellular populations present in our preparations. In such homogeneous populations, this genetic analysis allowed us to compare the cellular activity of normal versus HOisolated cells and demonstrate a strong overexpression of OC in HO cells as well as a significant upregulation of type 1 collagen and osteonectin.
Materials and methods Chemicals were purchased from Sigma (L’Isle d’Abeau, France) unless otherwise stated. Specimens All bone specimens were obtained after surgical operations justified by medical reasons and would have been otherwise discarded. Samples of heterotopic bone were removed per-operatively from seven head-injured patients who had developed heterotopic ossification around the hip. Three patients were male and four were female. The range
of age was 19 –60 years and the average age was 37 years. Normal bone pieces used as control were obtained peroperatively from the proximal femoral shaft of five male and two female patients who had received prosthesis of the hip joint. The age range was 32 –56 years and the mean age was 48 years. Bone fragments were minced using scalpels and scissors and washed with phosphate-buffered saline (PBS). Short enzymatic collagenase type IV digestion (0.25% in PBS) was performed for 15 min at 37jC. Treated bone fragments were then frozen in liquid nitrogen and stored at 70jC. Additionally, some of the treated bone fragments were prepared for histological analyses. Histological preparation All specimens had been prepared according to the nondecalcified method. They were fixed in neutral-buffered formalin and dehydrated in progressive alcohol before embedding in methylmetacrylate. Sections of 8-Am thickness were stained by May Gru¨nwald-Giemsa coloration to be examined under light microscopy. The number of osteocytes measured automatically with an image analyzer was expressed for 1 mm2. RNA extraction Bone fragments (0.5 –1 g) were crushed in 5 –10 ml of extraction reagent (Eurobio, Les Ulis, France). RNA extraction was then performed following recommendations of the manufacturer. Total RNA was quantified by spectrophotometer at 260-nm wavelength and the integrity of RNA was controlled by the 28S/18S rRNA ratio after agarose gel electrophoresis. Contaminating DNA was removed from RNA samples in a 30-min digestion at 37jC with DNase I (Roche Diagnostics, Meylan, France). Reverse transcription Each RNA sample (1 Ag) was used for reverse transcription performed under standard conditions with Superscript II reverse transcriptase (Life technologies, Cergy Pontoise, France) and random hexamer primers (Amersham Pharmacia Biotech, Saclay, France) in a 20 Al final volume. The reaction was carried out at 42jC for 30 min and stopped with incubation at 99jC for 5 min. The RT reactions were then diluted to 100 Al in water. Stock cDNA template (1 Al) was used in subsequent PCR reactions. Quantitative PCR experiments Quantitative PCR was performed using a LightCycler system (Roche Diagnostics) according to the manufacturer ‘s instructions. Reactions were performed in 10-Al volume
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
with 0.5-AM primers, 4 mM MgCl2 and 1 Al of LightCycler-FastStartDNA Master SYBR Green I mix (Roche Diagnostics). Protocol consisted of a hot start step (8 min at 95jC) followed by 40 cycles including a 10-s denaturation step (95jC), a 10-s annealing step and an elongation step at 72jC varying from 15 to 40 s. Efficiencies of PCR were optimized according to Roche Diagnostic’s recommendations on a standard sample expressing all studied genes. To confirm amplification specificity, PCR products were subjected to a melting curve analysis and a subsequent gel electrophoresis. Quantification data represented the mean of two experiments. The sequences for the primers used for each of the genes analyzed were as follows. h actin upstream primer 5V-ATCTGGCACCACACCTTCTA-3V and downstream primer 5V-AGCTCGTAGCTCTTCTCCAG-3V. Alkaline phosphatase upstream primer 5V-CCCAAAGGCTTCTTCTTG-3Vand downstream primer 5V-CTGGTAGTTGTTGTGAGC-3V. Cbfa1 upstream primer 5V-GCTGTTATGAAAAACCAAGT-3V and downstream primer 5V-GGGAGGATTTGTGAAGAC-3V. Type 1 collagen upstream primer 5V-GGACACAATGGATTGCAAGG-3V and downstream primer 5V-TAACCACTGCTCCACTCTGG-3V. Endoglin upstream primer 5V-GTGATTGCCATCTTTGCCCT-3V and downstream primer 5V-CATTGTCCAAAGCCCACTTC-3V. Glyceraldehyde-3-phosphate deshydrogenase (GAPDH) upstream primer 5V-GTTCCAATATGATTCCACCC-3V and downstream primer 5V-AGGGATGATGTTCTGGAGAG-3V. Osteocalcin upstream primer 5V-ATGAGAGCCCTCACACTCCTC-3V and downstream primer 5V-GCCGTAGAAGCGCCGATAGGC-3V. Osteonectin upstream primer 5V-GATGAGGACAACAACCTTCTGAC-3V and downstream primer 5V-TTAGATCACAAGATCCTTGTCGAT-3V. Relative quantification analyses were performed by RelQuant 1.01 Software (Roche Diagnostics). Statistical analysis Statistical significance of differences between groups was determined by the Mann – Whitney U-test.
Results Concerning the patient groups, the mean age of the controls is a decade older. We checked the possibility that the variations of the metabolic markers studied could be due to an age effect alone between the two groups. As shown in Fig. 1, we failed to find any correlation between osteocalcin mRNA expression level and patients age. We also failed to correlate other patient factors with mRNA variations (e.g., gender, heterotopic bone dimensions or time from injury to biopsy). Radionucleide bone scans were performed in each HO patient before surgical resection and showed that the HO and normotopic bone shared similar anabolisms. It ensured
39
Fig. 1. Variations of osteocalcin mRNA expression level according to patient’s age, normalized against h actin. Control (v), HO cells (x).
that heterotopic ossifications were in the later stage of their development (data not shown). Histological evaluation was performed on all the HO samples and on four normal bone samples. Appearance of the two types of bone was similar consisting of a trabecular bone but the number of osteocytes/mm2 tissue was increased from 356 F 35 in the normal specimens to 542 F 94 in the heterotopic ossifications ( P < 0.01). A 15-min collagenase digestion allowed eliminating almost all nonosteoblastic cells, essentially bone marrow cells (Fig. 2). Treated bone fragments exhibited mainly osteocytes in the mineralized matrix with some remaining bordering osteoblasts associated with osteoid. For each gene presented, preliminary experiments were performed to normalize the study. The quality of RNA preparation (i.e., absence of DNA contamination) was controlled by PCR done on non-reverse-transcribed samples. Amplification specificity of PCR products was confirmed by the observation of a specific melting curve and a single band on gel electrophoresis. Even after 40 PCR cycles, no primer dimer was observed. For all markers,
40
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
Fig. 2. Histological analysis. Bone sections of 8 Am stained by May Gru¨nwald-Giemsa observed under light microscopy. 1: osteoblasts; 2: osteocytes; 3: adipocytes. (A and C) Normal bone pieces obtained from the proximal femoral shaft. (B and D) Heterotopic ossification surrounding the hip. Observation of osteocytes, osteoblasts and bone marrow cells before collagenase digestion (A and B). After 15 min of collagenase digestion, osteocytes represent 90 – 95% of cellular population (C and D).
PCR efficiency ranged from 1.85 to 2.00. These efficiencies allowed comparison between genes. Two housekeeping genes, h actin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), were used to standardize real-time RT-PCR results. Crossing-point values obtained with both of these housekeeping genes were homogeneous within all the samples (data not shown), indicating they were expressed at constant levels. Fig. 3 shows the expression levels of all studied genes for normal and HO-isolated cells normalized with h actin and GAPDH. Whatever the housekeeping gene used, the obtained patterns were comparable. First, similar levels of endoglin were expressed in HO- and control-isolated cells. ALP mRNA levels were slightly higher and with a weak
significance in HO-isolated cells using GAPDH as housekeeping gene (2, P < 0.05), while no difference was detected using the other housekeeping gene. Real-time RTPCR results demonstrated an overexpression of the other studied genes in HO-isolated cells. A considerable upregulation for osteocalcin was observed in HO specimens (65, P <0.005 when normalized with h actin; 42, P < 0.005 when normalized with GAPDH). Significant upregulation was also observed for type 1 collagen (12, P < 0.005 when normalized with h actin; 7, P < 0.005 when normalized with GAPDH) and osteonectin (10, P < 0.001 when normalized with h actin; 6, P < 0.005 when normalized with GAPDH) even if it did not reach levels similar to osteocalcin. A small overexpression of cbfa1 was observed
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
Fig. 3. Quantitative RT-PCR analysis. mRNA expression levels were normalized against two different housekeeping genes (a: h actin; and b: GAPDH) in control (o) and HO cells (4). *P V 0.05, **P V 0.005, ***P V 0.001.
when normalized with h actin (4, P < 0.05), but it was not significant after normalization with GAPDH.
Discussion In this paper we describe the development and validation of real-time RT-PCR to compare gene expression levels in
41
normal and heterotopic human bone biopsies. Among the markers of osteoblastic differentiation, we investigated the expression of alkaline phosphatase, type 1 collagen, osteonectin, osteocalcin, endoglin and core binding factor alpha 1. Being a very powerful and sensitive method, quantitative RT-PCR appears to be an interesting way to compare gene expression directly in resected tissue without culturing cells. But its use requires particular care in sample preparation as well as corrections for experimental variations by normalization to a housekeeping gene. We first defined a procedure of cell fractionation from bone resections. As assessed by histological analysis, 15-min collagenase digestion of bone samples eliminated almost all nonosteoblastic cells, like bone marrow cells. Treated bone samples exhibited mainly osteocytes in the mineralized matrix with some remaining bordering osteoblasts. Longer collagenase treatments allowed obtaining osteocytes with a higher degree of purification, but the RNA integrity was not preserved (data not shown). Following RNA extraction, we tested different housekeeping genes because the use of a particular housekeeping gene may sometimes be discussed (Giulietti et al., 2001). We determined that h actin and GAPDH genes gave the most constant and reproducible expression levels and were suitable endogenous controls for our quantification assays. Quantification data normalized with both of these housekeeping genes were comparable, showing that significant differences regarding gene expression levels can be determined between human cells from normal and heterotopic bone. Expression of the genes encoding alkaline phosphatase, endoglin and cbfa1 was not significantly different in HO and control specimens, whereas a strong upregulation for osteocalcin was observed in HO specimens as well as a significant upregulation for type 1 collagen and osteonectin. Such increased expressions are likely to be specific of HO because no correlation was found between mRNA levels and any of the nonpathologic factors that may differ (Fig. 1 as example of age impact). Most of the previous studies performed to understand heterotopic ossification etiology were done in vitro (Bidner et al., 1990; Kaysinger et al., 1997; Kurer et al., 1992; Renfree et al., 1994, Sell et al., 1998). In contrast to our results, Sell et al. (1998) showed that cultured cells from heterotopic ossification exhibited a lower level of osteocalcin expression compared with human osteoblast-like cells, whereas Kaysinger et al. (1997) showed an increased alkaline phosphatase activity in osteogenic cells obtained from HO but with a great variation among cultured cells from different patients. The differences between in vivo and in vitro studies may easily be explained by phenotype alteration during culture steps. Recently, development of HO following total hip arthroplasty was associated with an increase in serum osteocalcin (Wilkinson et al., 2003). Our genetic study suggests that human normal and heterotopic bone cells display variability in cell type distribution, maturity or activity. Histological analyses are relevant to a restricted cellular population committed to the
42
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
osteoblastic lineage with a slight increase in osteocyte density for HO samples (1.5-fold). Variations of measured mRNA levels for housekeeping genes, alkaline phosphatase, cbfa1 and endoglin are of the same range. In contrast, variability in gene expression levels detected for osteocalcin (about 50-fold), as well as osteonectin and type 1 collagen, could not be the reflection of the slight difference in osteocyte density. In bone cells, osteonectin and type 1 collagen are expressed during earlier stages of differentiation than osteocalcin, which is found in clearly differentiated osteoblasts and osteocytes as well as in the mineralized bone matrix (Stein et al., 1989; Zhou et al., 1994). The coexisting overexpression of mRNA encoding these three proteins does not support the assumption of a difference in differentiation stage between HO and normal human bone cells. Another more likely explanation would come from the phenomenon of enhanced osteogenesis in heterotopic ossification. HO has been described as being highly active, with a high bone turnover and a probably rapid bone formation (Bord et al., 1999; Puzas et al., 1987). This requires a rapid matrix production and type 1 collagen, osteonectin and osteocalcin, all three interact to synthesize the bone matrix. Thus, the elevated osteocalcin mRNA expression level in HO cells may be associated with the high bone turnover observed in the previous stages of this pathological ossification. Furthermore, this may be consistent with the variations observed between patients (variations which were not correlated with patient gender or age or days from trauma to excision of HO) and may reflect differences in the heterotopic bone activity. In conclusion, the method described here gives a first molecular characterization of heterotopic ossification. We have demonstrated that significant differences exist in genetic expression patterns between human cells from normal and heterotopic bone. Overexpression of osteocalcin, osteonectin and type 1 collagen mRNA levels could be associated with the high activity of this pathological bone. Further studies will be performed with known osteocalcin regulators to define more accurately the origin of hyperactivity in HO cells.
Acknowledgments The authors thank Dr B. Bouxin for work expertise and assistance. The authors also thank Drs K. Anselme, N. Benabid, F. Danze, D. Darriet, P. Rigaux and B. Sutter, members of HO working group, for helpful discussions. This work was supported in part by grants from the SFR: Socie´te´ Francaise de Rhumatologie.
References Barry, F.P., Boynton, R.E., Haynesworth, S., Murphy, J.M., Zaia, J., 1999. The monoclonal antibody SH-2, raised against human mesenchymal
stem cells, recognizes an epitope on endoglin (CD105). Biochem. Biophys. Res. Commun. 265, 134 – 139. Bidner, S.M., Rubins, I.M., Desjardins, J.V., Zukor, D.J., Goltzman, D., 1990. Evidence for a humoral mechanism for enhanced osteogenesis after head injury. J. Bone Jt. Surg., Am. 72 (8), 1144 – 1149. Bord, S., Horner, A., Beeton, C.A., Hembry, R.M., Compston, J.E., 1999. Tissue inhibitor of matrix metalloproteinase-1 (TIMP-1) distribution in normal and pathological human bone. Bone 24 (3), 229 – 235. Brook, C.G., Valman, H.B., 1971. Osteoma cutis and Albright’s hereditary osteodystrophy. Br. J. Dermatol. 85, 471 – 475. Cohen, R.B., Hahn, G.V., Tabas, J.A., Peeper, J., Levitz, C.L., Sando, A., Sando, N., Zasloff, M., Kaplan, F.S., 1993. The natural history of heterotopic ossification in patients who have fibrodysplasia ossificans progressiva. A study of forty-four patients. J. Bone Jt. Surg. 75A, 215 – 219. Ducy, P., Zhang, R., Geoffroy, V., Ridall, A., Karsenty, G., 1997. Osf2/ Cbfa1: a transcriptional activator of osteoblast differentiation. Cell 89, 747 – 754. Giulietti, A., Overbergh, L., Valckx, D., Decallonne, B., Bouillon, R., Mathieu, C., 2001. An overview of real-time quantitative PCR: applications to quantify cytokine gene expression. Methods 25, 386 – 401. Haynesworth, S.E., Baber, M.A., Caplan, A.L., 1992. Cell surface antigens on human marrow-derived mesenchymal cells are detected by monoclonal antibodies. Bone 13, 69 – 80. Ippolito, E., Formisano, R., Caterini, R., Farsetti, P., Penta, F., 1999. Operative treatment of heterotopic hip ossification in patients with coma after brain injury. Clin. Orthop. Relat. Res. 365, 130 – 138. Kaplan, F.S., Craver, R., MacEwen, G.D., Gannon, F.H., Finkel, G., Hahn, G., Tabas, J., Gardner, R.J., Zasloff, M.A., 1994. Progressive osseous heteroplasia: a distinct developmental disorder of heterotopic ossification. Two new case reports and follow-up of three previously reported cases. J. Bone Jt. Surg. 76A, 425 – 436. Kaysinger, K.K., Ramp, W.K., Lang, G.J., Gruber, H.E., 1997. Comparison of human osteoblasts and osteogenic cells from heterotopic bone. Clin. Orthop. Relat. Res. 342, 181 – 191. Komori, T., Yagi, H., Nomura, S., Yamaguchi, A., Sasaki, K., Deguchi, K., Shimizu, Y., Bronson, R.T., Gao, Y.H., Inada, M., Sato, M., Okamoto, R., Kitamura, Y., Yoshiki, S., Kishimoto, T., 1997. Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell 79, 755 – 764. Kurer, M.H.J., Khoker, M.A., Dandona, P., 1992. Human osteoblast stimulation by sera from paraplegic patients with heterotopic ossification. Paraplegia 30, 165 – 168. Otto, F., Thornell, A.P., Crompton, T., Denzel, A., Gilmour, K.C., Rosewell, I.R., Stamp, G.W., Beddington, R.S., Mundlos, S., Olsen, B.R., Selby, P.B., Owen, M.J., 1997. Cbfa1, a candidate gene for cleidocranial dysplasia syndrome, is essential for osteoblast differentiation and bone development. Cell 89, 765 – 771. Pittenger, D.E., 1991. Heterotopic ossification. Orthop. Rev. 20, 33 – 39. Puzas, J.E., Brand, J.S., Evarts, C.M., 1987. The stimulus for bone formation. In: Brand, J.S. (Ed.), The hip. Mosby CV, St Louis, pp. 25 – 38. Renfree, K.J., Banovac, K., Hornicek, F.J., Lebwohl, N.H., Villanueva, P.A., Nedd, K.J., 1994. Evaluation of serum osteoblast mitogenic activity in spinal cord and head injury patients with acute heterotopic ossification. Spine 19 (7), 740 – 746. Sawyer, J.R., Myers, M.A., Rosier, R.N., Puzas, J.E., 1991. Heterotopic ossification: clinical and cellular aspects. Calcif. Tissue Int. 49, 208 – 215. Sell, S., Gaissmaier, C., Fritz, J., Herr, G., Esenwein, S., Kusswetter, W., Volkmann, R., Wittkowski, K.M., Rodemann, H.P., 1998. Different behavior of human osteoblast-like cells isolated from normal and heterotopic bone in vitro. Calcif. Tissue Int. 62, 51 – 59. Shehab, D., Elgazzar, A.H., Collier, D., 2002. Heterotopic ossification. J. Nucl. Med. 43 (3), 346 – 353.
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43 Stein, G.S., Lian, J.B., Gerstenfeld, L.G., Shalhoub, V., Aronow, M., Owen, T., Markose, E., 1989. The onset of progression of osteoblast differentiation is functionally related to cellular proliferation. Connect. Tissue Res. 20, 3 – 13. Wilkinson, J.M., Stockley, I., Hamer, A.J., Barrington, N.A., Eastell, R., 2003. Biochemical markers of bone turnover and development of het-
43
erotopic ossification after total hip arthroplasty. J. Orthop. Res. 21, 529 – 534. Zhou, H., Choong, P., Mc Carthy, R., Chou, S.T., Martin, T.J., Ng, K.W., 1994. In situ hybridization to show sequential expression of osteoblast gene markers during bone formation in vivo. J. Bone Miner. Res. 9, 1489 – 1499.
Gene expression in human osteoblastic cells from normal and heterotopic ossification Christophe Chauveau, a Jean-Christophe Devedjian, a Marie-Claude Blary, a Christophe Delecourt, b Pierre Hardouin, a Joseph Jeanfils, a and Odile Broux a,* a
LR2B-Laboratoire de Recherche sur les Biomate´riaux et les Biotechnologies, Universite´ du Littoral Coˆte d’Opale, Boulogne-sur-mer et Berck-sur-mer, France b Etablissement He´lio-marin, Groupe Hopale, Berck-sur-mer, France Received 6 September 2003
Abstract Heterotopic ossification (HO), a possible complication of head injury, develops in sites where it is not normally present like at the vicinity of joints. It may cause pain, decrease motion and in severe cases complete joint ankylosis requiring surgical intervention. To our knowledge, no study has been made to analyze HO at the molecular level on human biopsies, whereas its etiology remains to be determined. We defined a procedure of cell fractionation from bone resections and developed quantitative RT-PCR to compare genetic expression patterns between human normal osteoblasts and heterotopic ossification forming cells. This quantitative study demonstrated a specific and strong overexpression of osteocalcin mRNA in HO-isolated cells associated with a significant upregulation of type 1 collagen and osteonectin mRNA while histological analysis showed only small cellular variations. Our results give a first molecular characterization of heterotopic ossification and we conclude that such overexpressions in HO-isolated cells could be associated with the high activity of this pathological bone. D 2003 Elsevier Inc. All rights reserved. Keywords: Heterotopic ossification; Head injury; Quantitative RT-PCR; Osteocalcin mRNA
Introduction Heterotopic ossification (HO) is characterized by the formation of bone in sites where it is not normally present. HO may be classified in a rare hereditary form or a most common acquired form. There are at least three distinct genetic hereditary forms in humans: fibrodysplasia ossificans progressiva (Cohen et al., 1993), progressive osseous heteroplasia (Kaplan et al., 1994) and Albright hereditary osteodystrophy (Brook and Valman, 1971). In acquired forms, HO frequently occurs after neurologic injury such as spinal trauma or head injury. It is commonly near a joint
* Corresponding author. LR2B-Laboratoire de Recherche sur les Biomate´riaux et les Biotechnologies, Universite´ du Littoral Coˆte d’Opale, Quai Masset, Bassin Napole´on, BP 120, 62327 Boulogne-sur-mer et Bercksur-mer, Cedex, France. Fax: +33-321994524. E-mail address: [email protected] (O. Broux). 0014-4800/$ - see front matter D 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.yexmp.2003.10.001
and may cause pain, decrease motion and, in severe cases, may require surgical resection (Ippolito et al., 1999; Pittenger, 1991; Sawyer et al., 1991; Shehab et al., 2002). The biological mechanisms of acquired HO formation remain to be determined. To understand the etiology of acquired heterotopic bone formation, studies focused mainly on two approaches: the research of a humoral factor as an inductive agent in neurogenic HO and the measure of osteoblast activity in HOisolated cells. For example, all the in vitro culture systems but one (Renfree et al., 1994) demonstrated a positive effect of sera, from patients with heterotopic ossification, on osteoblast proliferation (Bidner et al., 1990; Kurer et al., 1992). However, the osteoinductive factor, which could perhaps enhance osteogenesis, has not been identified. Other in vitro studies showed that human cells isolated from HO exhibit, in culture, increased osteoblastic activities, such as alkaline phosphatase activity and type 1 collagen synthesis, when compared with human normal osteoblasts (Kaysinger et al.,
38
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
1997; Sell et al., 1998). Nevertheless, to our knowledge, no study has been made to analyze neurogenic HO at the molecular level on human biopsies. In recent years, advances in human genetics led to the identification of many genes involved in the different steps of osteoblast differentiation. At the same time, a variety of powerful and sensitive methods to quantify mRNA expression has been developed. With these new tools at hands, we performed a study comparing genetic expression patterns of human HO-forming cells and of human normal osteoblasts. We first defined a procedure of cell fractionation from bone resections, allowing as assessed by histological analysis to obtain homogeneous osteoblastic populations that can be compared. Quantitative RT-PCR analysis was performed to determine mRNA expression levels of osteoblast markers. Some genes representative of the successive stages of osteoblastic differentiation have been chosen. Endoglin (END), identified by immunoprecipitation using the SH2 antibody, represents a marker which can discriminate undifferentiated human osteoblastic precursors and more mature bone cells (Barry et al., 1999; Haynesworth et al., 1992). Type 1 collagen (COL I), the major constituent of bone matrix, and alkaline phosphatase activity (ALP) are associated with matrix maturation and decrease when mineralization is in progress. Osteonectin (ON) is expressed at maximal levels during mineralization while osteocalcin (OC), a noncollagenous protein synthesized exclusively by osteoblasts, appears at a more final stage. Some of these genes were described as regulated by core-binding factor alpha 1 (Cbfa1) and we also quantified the expression level of this essential osteogenesis transcription factor (Ducy et al., 1997; Komori et al., 1997, Otto et al., 1997). The genetic expression patterns obtained by quantification of this panel of markers were relevant to the restricted cellular populations present in our preparations. In such homogeneous populations, this genetic analysis allowed us to compare the cellular activity of normal versus HOisolated cells and demonstrate a strong overexpression of OC in HO cells as well as a significant upregulation of type 1 collagen and osteonectin.
Materials and methods Chemicals were purchased from Sigma (L’Isle d’Abeau, France) unless otherwise stated. Specimens All bone specimens were obtained after surgical operations justified by medical reasons and would have been otherwise discarded. Samples of heterotopic bone were removed per-operatively from seven head-injured patients who had developed heterotopic ossification around the hip. Three patients were male and four were female. The range
of age was 19 –60 years and the average age was 37 years. Normal bone pieces used as control were obtained peroperatively from the proximal femoral shaft of five male and two female patients who had received prosthesis of the hip joint. The age range was 32 –56 years and the mean age was 48 years. Bone fragments were minced using scalpels and scissors and washed with phosphate-buffered saline (PBS). Short enzymatic collagenase type IV digestion (0.25% in PBS) was performed for 15 min at 37jC. Treated bone fragments were then frozen in liquid nitrogen and stored at 70jC. Additionally, some of the treated bone fragments were prepared for histological analyses. Histological preparation All specimens had been prepared according to the nondecalcified method. They were fixed in neutral-buffered formalin and dehydrated in progressive alcohol before embedding in methylmetacrylate. Sections of 8-Am thickness were stained by May Gru¨nwald-Giemsa coloration to be examined under light microscopy. The number of osteocytes measured automatically with an image analyzer was expressed for 1 mm2. RNA extraction Bone fragments (0.5 –1 g) were crushed in 5 –10 ml of extraction reagent (Eurobio, Les Ulis, France). RNA extraction was then performed following recommendations of the manufacturer. Total RNA was quantified by spectrophotometer at 260-nm wavelength and the integrity of RNA was controlled by the 28S/18S rRNA ratio after agarose gel electrophoresis. Contaminating DNA was removed from RNA samples in a 30-min digestion at 37jC with DNase I (Roche Diagnostics, Meylan, France). Reverse transcription Each RNA sample (1 Ag) was used for reverse transcription performed under standard conditions with Superscript II reverse transcriptase (Life technologies, Cergy Pontoise, France) and random hexamer primers (Amersham Pharmacia Biotech, Saclay, France) in a 20 Al final volume. The reaction was carried out at 42jC for 30 min and stopped with incubation at 99jC for 5 min. The RT reactions were then diluted to 100 Al in water. Stock cDNA template (1 Al) was used in subsequent PCR reactions. Quantitative PCR experiments Quantitative PCR was performed using a LightCycler system (Roche Diagnostics) according to the manufacturer ‘s instructions. Reactions were performed in 10-Al volume
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
with 0.5-AM primers, 4 mM MgCl2 and 1 Al of LightCycler-FastStartDNA Master SYBR Green I mix (Roche Diagnostics). Protocol consisted of a hot start step (8 min at 95jC) followed by 40 cycles including a 10-s denaturation step (95jC), a 10-s annealing step and an elongation step at 72jC varying from 15 to 40 s. Efficiencies of PCR were optimized according to Roche Diagnostic’s recommendations on a standard sample expressing all studied genes. To confirm amplification specificity, PCR products were subjected to a melting curve analysis and a subsequent gel electrophoresis. Quantification data represented the mean of two experiments. The sequences for the primers used for each of the genes analyzed were as follows. h actin upstream primer 5V-ATCTGGCACCACACCTTCTA-3V and downstream primer 5V-AGCTCGTAGCTCTTCTCCAG-3V. Alkaline phosphatase upstream primer 5V-CCCAAAGGCTTCTTCTTG-3Vand downstream primer 5V-CTGGTAGTTGTTGTGAGC-3V. Cbfa1 upstream primer 5V-GCTGTTATGAAAAACCAAGT-3V and downstream primer 5V-GGGAGGATTTGTGAAGAC-3V. Type 1 collagen upstream primer 5V-GGACACAATGGATTGCAAGG-3V and downstream primer 5V-TAACCACTGCTCCACTCTGG-3V. Endoglin upstream primer 5V-GTGATTGCCATCTTTGCCCT-3V and downstream primer 5V-CATTGTCCAAAGCCCACTTC-3V. Glyceraldehyde-3-phosphate deshydrogenase (GAPDH) upstream primer 5V-GTTCCAATATGATTCCACCC-3V and downstream primer 5V-AGGGATGATGTTCTGGAGAG-3V. Osteocalcin upstream primer 5V-ATGAGAGCCCTCACACTCCTC-3V and downstream primer 5V-GCCGTAGAAGCGCCGATAGGC-3V. Osteonectin upstream primer 5V-GATGAGGACAACAACCTTCTGAC-3V and downstream primer 5V-TTAGATCACAAGATCCTTGTCGAT-3V. Relative quantification analyses were performed by RelQuant 1.01 Software (Roche Diagnostics). Statistical analysis Statistical significance of differences between groups was determined by the Mann – Whitney U-test.
Results Concerning the patient groups, the mean age of the controls is a decade older. We checked the possibility that the variations of the metabolic markers studied could be due to an age effect alone between the two groups. As shown in Fig. 1, we failed to find any correlation between osteocalcin mRNA expression level and patients age. We also failed to correlate other patient factors with mRNA variations (e.g., gender, heterotopic bone dimensions or time from injury to biopsy). Radionucleide bone scans were performed in each HO patient before surgical resection and showed that the HO and normotopic bone shared similar anabolisms. It ensured
39
Fig. 1. Variations of osteocalcin mRNA expression level according to patient’s age, normalized against h actin. Control (v), HO cells (x).
that heterotopic ossifications were in the later stage of their development (data not shown). Histological evaluation was performed on all the HO samples and on four normal bone samples. Appearance of the two types of bone was similar consisting of a trabecular bone but the number of osteocytes/mm2 tissue was increased from 356 F 35 in the normal specimens to 542 F 94 in the heterotopic ossifications ( P < 0.01). A 15-min collagenase digestion allowed eliminating almost all nonosteoblastic cells, essentially bone marrow cells (Fig. 2). Treated bone fragments exhibited mainly osteocytes in the mineralized matrix with some remaining bordering osteoblasts associated with osteoid. For each gene presented, preliminary experiments were performed to normalize the study. The quality of RNA preparation (i.e., absence of DNA contamination) was controlled by PCR done on non-reverse-transcribed samples. Amplification specificity of PCR products was confirmed by the observation of a specific melting curve and a single band on gel electrophoresis. Even after 40 PCR cycles, no primer dimer was observed. For all markers,
40
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
Fig. 2. Histological analysis. Bone sections of 8 Am stained by May Gru¨nwald-Giemsa observed under light microscopy. 1: osteoblasts; 2: osteocytes; 3: adipocytes. (A and C) Normal bone pieces obtained from the proximal femoral shaft. (B and D) Heterotopic ossification surrounding the hip. Observation of osteocytes, osteoblasts and bone marrow cells before collagenase digestion (A and B). After 15 min of collagenase digestion, osteocytes represent 90 – 95% of cellular population (C and D).
PCR efficiency ranged from 1.85 to 2.00. These efficiencies allowed comparison between genes. Two housekeeping genes, h actin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), were used to standardize real-time RT-PCR results. Crossing-point values obtained with both of these housekeeping genes were homogeneous within all the samples (data not shown), indicating they were expressed at constant levels. Fig. 3 shows the expression levels of all studied genes for normal and HO-isolated cells normalized with h actin and GAPDH. Whatever the housekeeping gene used, the obtained patterns were comparable. First, similar levels of endoglin were expressed in HO- and control-isolated cells. ALP mRNA levels were slightly higher and with a weak
significance in HO-isolated cells using GAPDH as housekeeping gene (2, P < 0.05), while no difference was detected using the other housekeeping gene. Real-time RTPCR results demonstrated an overexpression of the other studied genes in HO-isolated cells. A considerable upregulation for osteocalcin was observed in HO specimens (65, P <0.005 when normalized with h actin; 42, P < 0.005 when normalized with GAPDH). Significant upregulation was also observed for type 1 collagen (12, P < 0.005 when normalized with h actin; 7, P < 0.005 when normalized with GAPDH) and osteonectin (10, P < 0.001 when normalized with h actin; 6, P < 0.005 when normalized with GAPDH) even if it did not reach levels similar to osteocalcin. A small overexpression of cbfa1 was observed
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
Fig. 3. Quantitative RT-PCR analysis. mRNA expression levels were normalized against two different housekeeping genes (a: h actin; and b: GAPDH) in control (o) and HO cells (4). *P V 0.05, **P V 0.005, ***P V 0.001.
when normalized with h actin (4, P < 0.05), but it was not significant after normalization with GAPDH.
Discussion In this paper we describe the development and validation of real-time RT-PCR to compare gene expression levels in
41
normal and heterotopic human bone biopsies. Among the markers of osteoblastic differentiation, we investigated the expression of alkaline phosphatase, type 1 collagen, osteonectin, osteocalcin, endoglin and core binding factor alpha 1. Being a very powerful and sensitive method, quantitative RT-PCR appears to be an interesting way to compare gene expression directly in resected tissue without culturing cells. But its use requires particular care in sample preparation as well as corrections for experimental variations by normalization to a housekeeping gene. We first defined a procedure of cell fractionation from bone resections. As assessed by histological analysis, 15-min collagenase digestion of bone samples eliminated almost all nonosteoblastic cells, like bone marrow cells. Treated bone samples exhibited mainly osteocytes in the mineralized matrix with some remaining bordering osteoblasts. Longer collagenase treatments allowed obtaining osteocytes with a higher degree of purification, but the RNA integrity was not preserved (data not shown). Following RNA extraction, we tested different housekeeping genes because the use of a particular housekeeping gene may sometimes be discussed (Giulietti et al., 2001). We determined that h actin and GAPDH genes gave the most constant and reproducible expression levels and were suitable endogenous controls for our quantification assays. Quantification data normalized with both of these housekeeping genes were comparable, showing that significant differences regarding gene expression levels can be determined between human cells from normal and heterotopic bone. Expression of the genes encoding alkaline phosphatase, endoglin and cbfa1 was not significantly different in HO and control specimens, whereas a strong upregulation for osteocalcin was observed in HO specimens as well as a significant upregulation for type 1 collagen and osteonectin. Such increased expressions are likely to be specific of HO because no correlation was found between mRNA levels and any of the nonpathologic factors that may differ (Fig. 1 as example of age impact). Most of the previous studies performed to understand heterotopic ossification etiology were done in vitro (Bidner et al., 1990; Kaysinger et al., 1997; Kurer et al., 1992; Renfree et al., 1994, Sell et al., 1998). In contrast to our results, Sell et al. (1998) showed that cultured cells from heterotopic ossification exhibited a lower level of osteocalcin expression compared with human osteoblast-like cells, whereas Kaysinger et al. (1997) showed an increased alkaline phosphatase activity in osteogenic cells obtained from HO but with a great variation among cultured cells from different patients. The differences between in vivo and in vitro studies may easily be explained by phenotype alteration during culture steps. Recently, development of HO following total hip arthroplasty was associated with an increase in serum osteocalcin (Wilkinson et al., 2003). Our genetic study suggests that human normal and heterotopic bone cells display variability in cell type distribution, maturity or activity. Histological analyses are relevant to a restricted cellular population committed to the
42
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43
osteoblastic lineage with a slight increase in osteocyte density for HO samples (1.5-fold). Variations of measured mRNA levels for housekeeping genes, alkaline phosphatase, cbfa1 and endoglin are of the same range. In contrast, variability in gene expression levels detected for osteocalcin (about 50-fold), as well as osteonectin and type 1 collagen, could not be the reflection of the slight difference in osteocyte density. In bone cells, osteonectin and type 1 collagen are expressed during earlier stages of differentiation than osteocalcin, which is found in clearly differentiated osteoblasts and osteocytes as well as in the mineralized bone matrix (Stein et al., 1989; Zhou et al., 1994). The coexisting overexpression of mRNA encoding these three proteins does not support the assumption of a difference in differentiation stage between HO and normal human bone cells. Another more likely explanation would come from the phenomenon of enhanced osteogenesis in heterotopic ossification. HO has been described as being highly active, with a high bone turnover and a probably rapid bone formation (Bord et al., 1999; Puzas et al., 1987). This requires a rapid matrix production and type 1 collagen, osteonectin and osteocalcin, all three interact to synthesize the bone matrix. Thus, the elevated osteocalcin mRNA expression level in HO cells may be associated with the high bone turnover observed in the previous stages of this pathological ossification. Furthermore, this may be consistent with the variations observed between patients (variations which were not correlated with patient gender or age or days from trauma to excision of HO) and may reflect differences in the heterotopic bone activity. In conclusion, the method described here gives a first molecular characterization of heterotopic ossification. We have demonstrated that significant differences exist in genetic expression patterns between human cells from normal and heterotopic bone. Overexpression of osteocalcin, osteonectin and type 1 collagen mRNA levels could be associated with the high activity of this pathological bone. Further studies will be performed with known osteocalcin regulators to define more accurately the origin of hyperactivity in HO cells.
Acknowledgments The authors thank Dr B. Bouxin for work expertise and assistance. The authors also thank Drs K. Anselme, N. Benabid, F. Danze, D. Darriet, P. Rigaux and B. Sutter, members of HO working group, for helpful discussions. This work was supported in part by grants from the SFR: Socie´te´ Francaise de Rhumatologie.
References Barry, F.P., Boynton, R.E., Haynesworth, S., Murphy, J.M., Zaia, J., 1999. The monoclonal antibody SH-2, raised against human mesenchymal
stem cells, recognizes an epitope on endoglin (CD105). Biochem. Biophys. Res. Commun. 265, 134 – 139. Bidner, S.M., Rubins, I.M., Desjardins, J.V., Zukor, D.J., Goltzman, D., 1990. Evidence for a humoral mechanism for enhanced osteogenesis after head injury. J. Bone Jt. Surg., Am. 72 (8), 1144 – 1149. Bord, S., Horner, A., Beeton, C.A., Hembry, R.M., Compston, J.E., 1999. Tissue inhibitor of matrix metalloproteinase-1 (TIMP-1) distribution in normal and pathological human bone. Bone 24 (3), 229 – 235. Brook, C.G., Valman, H.B., 1971. Osteoma cutis and Albright’s hereditary osteodystrophy. Br. J. Dermatol. 85, 471 – 475. Cohen, R.B., Hahn, G.V., Tabas, J.A., Peeper, J., Levitz, C.L., Sando, A., Sando, N., Zasloff, M., Kaplan, F.S., 1993. The natural history of heterotopic ossification in patients who have fibrodysplasia ossificans progressiva. A study of forty-four patients. J. Bone Jt. Surg. 75A, 215 – 219. Ducy, P., Zhang, R., Geoffroy, V., Ridall, A., Karsenty, G., 1997. Osf2/ Cbfa1: a transcriptional activator of osteoblast differentiation. Cell 89, 747 – 754. Giulietti, A., Overbergh, L., Valckx, D., Decallonne, B., Bouillon, R., Mathieu, C., 2001. An overview of real-time quantitative PCR: applications to quantify cytokine gene expression. Methods 25, 386 – 401. Haynesworth, S.E., Baber, M.A., Caplan, A.L., 1992. Cell surface antigens on human marrow-derived mesenchymal cells are detected by monoclonal antibodies. Bone 13, 69 – 80. Ippolito, E., Formisano, R., Caterini, R., Farsetti, P., Penta, F., 1999. Operative treatment of heterotopic hip ossification in patients with coma after brain injury. Clin. Orthop. Relat. Res. 365, 130 – 138. Kaplan, F.S., Craver, R., MacEwen, G.D., Gannon, F.H., Finkel, G., Hahn, G., Tabas, J., Gardner, R.J., Zasloff, M.A., 1994. Progressive osseous heteroplasia: a distinct developmental disorder of heterotopic ossification. Two new case reports and follow-up of three previously reported cases. J. Bone Jt. Surg. 76A, 425 – 436. Kaysinger, K.K., Ramp, W.K., Lang, G.J., Gruber, H.E., 1997. Comparison of human osteoblasts and osteogenic cells from heterotopic bone. Clin. Orthop. Relat. Res. 342, 181 – 191. Komori, T., Yagi, H., Nomura, S., Yamaguchi, A., Sasaki, K., Deguchi, K., Shimizu, Y., Bronson, R.T., Gao, Y.H., Inada, M., Sato, M., Okamoto, R., Kitamura, Y., Yoshiki, S., Kishimoto, T., 1997. Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell 79, 755 – 764. Kurer, M.H.J., Khoker, M.A., Dandona, P., 1992. Human osteoblast stimulation by sera from paraplegic patients with heterotopic ossification. Paraplegia 30, 165 – 168. Otto, F., Thornell, A.P., Crompton, T., Denzel, A., Gilmour, K.C., Rosewell, I.R., Stamp, G.W., Beddington, R.S., Mundlos, S., Olsen, B.R., Selby, P.B., Owen, M.J., 1997. Cbfa1, a candidate gene for cleidocranial dysplasia syndrome, is essential for osteoblast differentiation and bone development. Cell 89, 765 – 771. Pittenger, D.E., 1991. Heterotopic ossification. Orthop. Rev. 20, 33 – 39. Puzas, J.E., Brand, J.S., Evarts, C.M., 1987. The stimulus for bone formation. In: Brand, J.S. (Ed.), The hip. Mosby CV, St Louis, pp. 25 – 38. Renfree, K.J., Banovac, K., Hornicek, F.J., Lebwohl, N.H., Villanueva, P.A., Nedd, K.J., 1994. Evaluation of serum osteoblast mitogenic activity in spinal cord and head injury patients with acute heterotopic ossification. Spine 19 (7), 740 – 746. Sawyer, J.R., Myers, M.A., Rosier, R.N., Puzas, J.E., 1991. Heterotopic ossification: clinical and cellular aspects. Calcif. Tissue Int. 49, 208 – 215. Sell, S., Gaissmaier, C., Fritz, J., Herr, G., Esenwein, S., Kusswetter, W., Volkmann, R., Wittkowski, K.M., Rodemann, H.P., 1998. Different behavior of human osteoblast-like cells isolated from normal and heterotopic bone in vitro. Calcif. Tissue Int. 62, 51 – 59. Shehab, D., Elgazzar, A.H., Collier, D., 2002. Heterotopic ossification. J. Nucl. Med. 43 (3), 346 – 353.
C. Chauveau et al. / Experimental and Molecular Pathology 76 (2004) 37–43 Stein, G.S., Lian, J.B., Gerstenfeld, L.G., Shalhoub, V., Aronow, M., Owen, T., Markose, E., 1989. The onset of progression of osteoblast differentiation is functionally related to cellular proliferation. Connect. Tissue Res. 20, 3 – 13. Wilkinson, J.M., Stockley, I., Hamer, A.J., Barrington, N.A., Eastell, R., 2003. Biochemical markers of bone turnover and development of het-
43
erotopic ossification after total hip arthroplasty. J. Orthop. Res. 21, 529 – 534. Zhou, H., Choong, P., Mc Carthy, R., Chou, S.T., Martin, T.J., Ng, K.W., 1994. In situ hybridization to show sequential expression of osteoblast gene markers during bone formation in vivo. J. Bone Miner. Res. 9, 1489 – 1499.