Visualizing levels of osteoblast differentiation by ... - Wiley Online Library

' 2005 Wiley-Liss, Inc.

genesis 43:87–98 (2005)

ARTICLE

Visualizing Levels of Osteoblast Differentiation by a Two-Color Promoter-GFP Strategy: Type I CollagenGFPcyan and Osteocalcin-GFPtpz I. Bilic-Curcic,1 M. Kronenberg,1 X. Jiang,1 J. Bellizzi,1 M. Mina,2 I. Marijanovic,1 E.M. Gardiner,3 and D.W. Rowe1* 1

Department of Genetics and Developmental Biology, University of Connecticut Health Center, Farmington, Connecticut Department of Pediatric Dentistry, University of Connecticut Health Center, Farmington, Connecticut 3 Bone and Mineral Research Program and Molecular Modeling Facility, Garvan Institute of Medical Research, Sydney NSW, Australia 2

Received 15 December 2004; Accepted 28 June 2005

Summary: A 3.9 kb DNA fragment of human osteocalcin promoter and 3.6 kb DNA fragment of the rat collagen type1a1 promoter linked with visually distinguishable GFP isomers, topaz and cyan, were used for multiplex analysis of osteoblast lineage progression. Three patterns of dual transgene expression can be appreciated in primary bone cell cultures derived from the transgenic mice and by histology of their corresponding bones. Our data support the interpretation that strong pOBCol3.6GFPcyan alone is found in newly formed osteoblasts, while strong pOBCol3.6GFPcyan and hOCGFPtpz are present in osteoblasts actively making a new matrix. Osteoblasts expressing strong hOC-GFPtpz and weak pOBCol3.6GFPcyan are also present and may or may not be producing mineralized matrix. This multiplex approach reveals the heterogeneity within the mature osteoblast population that cannot be appreciated by current histological methods. It should be useful to identify and isolate populations of cells within an osteoblast lineage as they progress through stages of C 2005 Wiley-Liss, differentiation. genesis 43:87–98, 2005. V Inc.

Key words: GFP; osteocalcin promoter; osteoblast differentiation; type I collagen promoter; primary osteoblast culture; bone histology

INTRODUCTION Recognizing the cellular and molecular stages of increasing osteoblast differentiation within the osteoprogenitor lineage is an essential requirement for understanding diseases of bone, especially when caused by the inability of the lineage to produce sufficient numbers of functional osteoblasts (Aubin et al., 1995; Manolagas, 2000; Oreffo et al., 1998). Mutations of genes essential to the formation of a stable bone matrix or genes that participate in a molecular pathway essential to the forward maturation of osteoblast precursor cells can have a similar end result

of decreased bone mass (Byers et al., 1988; Erlebacher et al., 1995; Marie et al., 1991). Distinguishing between the two mechanisms is the first step in understanding the molecular and cellular basis for diseases of bone. Current methods for staging levels of osteoblast differentiation in murine models of bone disease include histochemistry, RNA phenotyping (Liu et al., 1994), or cell surface antigens (Aubin and Turksen, 1996; Chen et al., 1999; Gronthos et al., 1999; Turksen and Aubin, 1991; Van Vlasselaer et al., 1994; Zannettino et al., 2003). However, these approaches lack the ability to be easily transferred between primary cell culture and a corresponding intact bone of the affected mouse. The autofluorescent marker proteins derived from green fluorescent protein (GFP), when driven by promoters that are active in a restricted population of cells, have the potential to identify a unique subpopulation of cells within the osteoprogenitor lineage in both experimental settings. We have previously demonstrated the potential for visualizing cells within the osteprogenitor lineage using Col1a1-derived GFP constructs (Dacic et al., 2001; Kalajzic et al., 2002a). Using marrow stromal fibroblast Abbreviations: BSP, bone sialoprotein; CAT, chloramphenicol acetyl transferase; DMP1-GFP, name of transgenic mouse with dental matrix protein reporter construct; FAC, fluorescent activated cytometry; GFP, green fluorescent protein; mCOB, mouse calvarial osteoblast; MSC, marrow stromal cell; OC, osteocalcin; pOBCol3.6GFPcyan and hOCGFPtpz, names given to the two transgenes used in this study; pOBCol2.3GFP, name of type I collagen reporter that is restricted to mature osteoblasts; XO, xylenol orange. * Correspondence to: David W. Rowe, Department of Genetics and Developmental Biology, University of Connecticut Health Center, Farmington, CT 06030. E-mail: [email protected] Contract grant sponsor: Public Health Service (PHS), Contract grant number: R01 AR43457, Contract grant number: U01 DK63478. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/gene.20156

88

BILIC-CURCIC ET AL.

and neonatal calvarial cell cultures, it is possible to view progression of osteoblast differentiation in real time and correlate it with transgene expression in intact bone (Wang et al., 2005). Two overlapping isoforms of GFP marker gene (GFPemd and GFPtpz) were used to generate transgenic mice expressing visual markers that activate at different stages of the osteoblastic lineage. The onset of low-intensity pOBCol3.6GFPtpz expression can be associated with a level of preosteoblast differentiation based on its temporal correlation with alkaline phosphatase staining and type I collagen production and its location in the cluster of cells that will ultimately form the bone nodule. The GFP signal becomes much stronger as the nodule acquires markers of early osteoblast differentiation (prior to the onset of pOBCol2.3GFP transgene, see below) and remains active through the later stages of nodule maturation. In bone, it is found in the periosteal layer, endosteal osteoblasts, and a few osteocytes in cortical areas adjacent to the endosteal surface, but is generally not expressed in deeply embedded osteocytes (Jiang et al., 2005). In contrast, pOBCol2.3GFPemd activates when BSP and osteocalcin are expressed in cultured osteoblasts and fluorescence is restricted to cells within the mineralizing nodules. In bone its expression is limited to surface osteoblasts and to most of the osteocytes throughout the bone matrix. These patterns of activity suggest that the Col3.6 construct preferentially identifies cells early in the lineage commitment, while the Col2.3 promoter drives expression of GFP in mature osteoblasts and osteocytes. However, the similarity of the promoter constructs and the overlapping spectral properties of the two GFPs preclude a clear distinction between the different levels of osteoblast differentiation. Because osteocalcin expression is associated with full osteoblast differentiation, our first objective was the development of a GFP reporter construct that identified a more restrictive population of osteoblastic cells. The initial osteocalcin (OC) promoter construct utilized the 1.7-kb murine fragment that has been widely used to express a variety of proteins in transgenic animals (Kalajzic et al., 2002b). As a promoter to drive GFP, it lacked sufficient strength to mark the spectrum of cells anticipated by immunohistochemical or in situ studies. Subsequently, a larger fragment derived from the human gene containing 3.8 kb of upstream sequence and 3.5 kb of downstream flanking sequence was shown to drive a CAT (chloramphenicol acetyl transferase) transgene that was detectable by immunohistochemistry on the surface of the trabecular bone, the endosteal surface of the cortical bone, and in mature chondrocytes (Kesterson et al., 1993; Sims et al., 1997). This fragment has been used in a variety of gene regulatory studies in which it was shown to be responsive to vitamin D (Kerner et al., 1989; Morrison et al., 1989), while the murine fragment lacked this property (Clemens et al., 1997; Sims et al., 1997). This construct appears to be sufficiently active to mark osteoblastic cells and its expression pattern is detailed below.

The other objective of this study is to show the feasibility of generating transgenic mice containing two promoter constructs driving distinguishable GFP isomers that are active at different stages of osteoblast differentiation as a tool to identify cells at different levels of development both in cell culture and intact mouse bone. Mice with a single transgene were produced and characterized expressing either a Col3.6 promoter driving cGFP, identified here as pOBCol3.6GFPcyan, or a human osteocalcin CAT reporter transgene (GOSCAS) that was reengineered to express GFPtpz and identified here as hOCGFPtpz. Crossing pOBCol3.6GFPcyan with hOC-GFPtpz transgenic mice allowed us to compare temporal and spatial expression of those two promoters in vitro and in vivo within the same cell population of differentiating bone cells. RESULTS Transgene Expression in hOC-GFPtpz-Positive Mice Two mouse lines transgenic for hOC-GFPtpz (Fig. 1a) and two lines transgenic for pOBCol3.6GFPcyan (Fig. 1b) were developed and characterized. All lines produced mice that were healthy and transmitted the transgene at the expected frequency. The lines were sufficiently strong to allow transgenic mice to be genotyped at the cage side using a portable fluorescent light. Using fluorescence microscope filters optimized for either GFPcyan or GFPtpz, tail clips showed a difference in expression pattern. hOC-GFPtpz transgene is green and expressed only in vertebra of tail segments (Fig. 1c), while pOBCol3.6GFPcyan is blue and expressed through the whole tail, including the skin (Fig. 1d). Northern blot analysis of tissues of 4-week-old hOCGFPtpz heterozygous mice showed strong GFP expression in long bone and calvaria that was associated with strong OC expression, with no detectable GFP or OC transcripts in other tissues (Fig. 1f). A similar analysis of the pOBCol3.6GFPtpz line was performed by Kalajzic et al. (2002a) that showed high expression in long bone and calvaria, but also weak expression in skin, aorta, lung, bladder, fat, muscle, and tendon. Flow cytometry of 21-day-old cultured calvarial cells showed that the hOC-GFPtpz transgene is activated in 4.4% of the total cell population (Fig. 2a). In a Northern blot of sorted cells there was no evidence for RNA degradation, which means that the cells successfully survived digestion and sorting. Northern hybridization indicated that GFP and the majority of OC expression is restricted only to the GFP-positive population of cells; an extremely faint OC signal was detected in GFP-negative cells (Fig. 2b). Analysis of hOC-GFPtpz Transgene Expression In Vitro Marrow stromal cell (MSC) cultures were examined from 6–8-week-old mice from both hOC-GFPtpz transgenic

VISUALIZING LEVELS OF OSTEOBLAST DIFFERENTIATION

89

FIG. 2. Analysis of cells liberated from 21-day-old calvarial cell culture established from hOC-GFPtpz mice. a: Flow cytometry indicates that 4.4% of cells are GFP-positive; b: Northern blot analysis of GFP positive and GFP-negative cells demonstrates a strong GFP and OC signal in GFP-positive cells relative to the same signals in GFP-negative cells. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

FIG. 1. Constructs used in the generation of the transgenic mice. a: hOC-GFPtpz is composed of 3.8 kb of the human osteocalcin promoter and 10 bp of the 50 nontranslated region (thick black line), an SV40 16S splicing unit (thin black line), a multiple cloning site containing a triple stop cassette (gray box), GFPtpz with the bGH polyadenylation signal (striped box), and 3.5 kb of a downstream segment that includes the 30 untranslated region and additional flanking genomic sequence (thick black line); b: The pOBCol3.6GFPcyan construct contains 3.5 kb of 50 flanking sequence of the rat col1a1 promoter plus the first exon and first intron driving GFPcyan. c–e: Genotyping of transgenic mice using fluorescence microscopy. c: Topaz (green) expression is restricted to the vertebral bone. d: GFPcyan (blue) is expressed in all the connective tissues components of the whole tail. e: The same pattern of expression is observed in double transgenics. f: Northern blot analysis of hOCGFPtpz transgene expression in tissues. Strong GFP expression is detected in long bone and calvaria that is associated with strong OC expression. No GFP or OC transcripts were detected in other tissues. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

lines. The GFP signal was initially detected at day 15 and was limited to cells within the mineralizing area of the nodule (identified by XO staining). The number of GFPpositive cells increased over time and the intensity of transgene expression was somewhat variable (Fig. 3A). By day 21, numerous GFP-positive round-shaped cells were present within the mineralized nodule, while the GFP-positive cells adjacent to the deposited mineral exhibited a stellate morphology (Fig. 3B). This dual pattern of cell morphology in GFP-positive cells is similar to mRNA OC signal detected by in situ hybridization (Pockwinse et al., 1993). The same temporal expression pattern was found in neonatal calvarial osteoblast (mCOB) cultures derived from both hOC-GFPtpz transgenic lines. The transgene was activated at day 15 and its expression was restricted to a mineralizing nodule. The highest level of expression was observed on day 21 in the round-shaped cells within the mineral, while the stellate-shaped cells surrounding the mineral were less apparent (Fig. 3C,D). The progression of primary MSC (Fig. 4A) and mCOB (Fig. 4B) cultures was evaluated through AP expression and mineralization. RNA was harvested during the time course of osteoblastic differentiation of cultures (Fig. 4C). GFP expression driven by the human osteocalcin promoter was detected on day 15, somewhat earlier than the mRNA signal from the endogenous gene and showed an increase at later time points in association with the increase of BSP and onset of osteocalcin expression. Histological Analysis of hOC-GFPtpz Transgene Expression In situ hybridization was performed on adjacent sections of 1-week-old mandibles (Fig. 5a,b) and 1-month-old femurs (Fig. 5c,d) in order to compare localization of cells expressing the endogenous osteocalcin gene with hOC-GFPtpz transgene-positive cells. In sections of developing molar, GFP fluorescence is expressed in odontoblasts and in osteocytes in the adjacent bone (Fig. 5b). In femur sections, GFP transgene expression is

90

BILIC-CURCIC ET AL.

FIG. 3. The expression of hOC-GFPtpz transgene in MSC and mCOB cultures. A: Transmitted (upper panel) and fluorescent images (lower panel) of MSC cultures established from mice harboring hOC-GFPtpz were obtained from day 7 to day 21 of culture (magnification 103). The red color that develops in the transmitted image is mineral that stains with XO. hOC-GFPtpz transgene was activated at day 15 and by day 21 numerous GFP-positive cells were present in the mineralized area. B: High-power image of day 21 culture that shows two population of cells, one with round-shaped cells (red arrow) in the mineral and the other with a stellate shape (yellow arrow) localized adjacent to the deposited mineral. C: mCOB cultures grown under identical conditions as the MSF culture. The expression pattern is the same as in MSC with the exception that the stellate-shaped cells are not as apparent in the high-power magnification (D). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

limited to osteoblasts and osteocytes in cortical and trabecular bone as well as vascular regions within cortical bone (Fig. 5d). In situ hybridization of adjacent sections probed with OC riboprobe demonstrated the same expression pattern as the hOC-GFP transgene (Fig. 5a,c). Expression of the hOC-GFPtpz transgene in vivo was analyzed in whole-length frozen decalcified sections of femurs from 7-day, 1-month, 3-month, and 6-month-old animals (Fig. 6). In 7-day-old femurs GFP expression was restricted to endosteal and periosteal osteoblasts and osteocytes throughout the cortical bone. With the exception of a few scattered positive cells, there was a striking absence of GFP-positive cells in the trabeculae of the primary spongiosum, such that most of the GFP signal came from cortical bone (Fig. 6, day 7). Transgene activity reached a peak in 1-month-old animals. Osteoblasts expressing the transgene were present on the surface of the trabecular bone within the secondary spongiosum and on the periosteal and endosteal surface of cortical bone. The pattern of GFP-positive osteocytes became limited to those cells immediately beneath the site of GFP-positive surface osteoblasts. The osteoblasts within the primary spongiosa remained GFP-negative, although some of the hypertrophic chondrocytes exhibited GFP activity. These observations are consistent with previous reports using in situ hybridization (Sims et al., 1997; Weinreb et al., 1990) or immunohistochemis-

try (Frenkel et al., 1997; Kesterson et al., 1993) (Fig. 6, 1 month). The GFP activity was less evident in 3-month and 6month-old femurs and was detectable only on the bone surface of cortical and trabecular bone. The GFP-positive osteocytes present at a younger age were no longer visible. Instead, small groups of positive cells were located within cortical bone in association with a blood vessel (Fig. 6c, white square). The positive cells were lining bone presumably undergoing remodeling and were not marking endothelial or perivascular cells. Expression of hOC-GFPpositive cells could be detected in these intracortical vascular areas in younger animals, but they were less obvious because of the numerous GFP-positive osteocytes. An unusual new area of positive cells emerged along the periosteal surface leading up to the proximal and distal articular surface of the femur. The line of GFP-positive cells began abruptly beyond the articular cartilage, extending across the metaphyseal region, and ending before the diaphyseal region. Unlike similar hOC-GFP-positive cells elsewhere in bone, this region was not associated with subsurface GFP-positive osteocytes (Fig. 6, 3 and 6 months). Frozen undecalcified sections of 6-week-old femurs were prepared from animals injected with XO 2 days before sacrifice to correlate mineral deposition and GFP expression. XO deposition was concomitant with osteocalcin expression in the remodeling active regions, includ-


91

FIG. 4. Evaluation of osteoblastic differentiation in MSC (A) and mCOB (B) cultures. ALP staining was performed on days 7, 11, 15, and 21 and mineralization was assessed with von Kossa staining on day 21. C: Northern blot analysis of GFP expression and markers of osteoblastic lineage in cultures derived from hOC-GFP transgenic mice. In both cultures GFP expression was detected on day 15 and showed an increase at later time points. This upregulation was concomitant with the differentiation of the cell culture and the increase of BSP and osteocalcin expression. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

FIG. 5. Colocalization of cells expressing endogenous OC mRNA (red color) and hOC-GFP (green)-positive cells in 1-week-old mandibles (a,b) and 1-month-old femurs (c,d). Fluorescent and brightfield images of adjacent sections were taken (magnification 103). a,b: Sections of the second molar show GFP fluorescence and OC riboprobe in odontoblasts and in osteoblasts in the developing mandible. c,d: Sections of the femur shows that the GFP transgene OC riboprobe expression is limited to surface osteoblasts and osteocytes of cortical bone. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

ing the positive cells within the cortical bone that were associated with blood vessels. (Fig. 7A). The only regions expressing the hOC-GFP transgene not associated with XO labeling were the periosteal cells in the metaphyseal region (Fig. 7B). Transgene Expression Pattern of pOBCol3.6GFPcyan 3 hOC-GFP Transgenic Mice in MSC and mCOB Cell Cultures Double transgenics were obtained by crossing heterozygous hOC-GFPtpz with pOBCol3.6GFPcyan-positive heterozygous animals. Genotyping was performed on the

tail snips by fluorescence microscopy in which both colors could be distinguished (Fig. 1e). MSC cultures were established to define spatial and temporal expression of the two different transgenes (Fig. 8). Low-level pOBCol3.6GFPcyan started to appear at day 7 in spindleshaped cells and its expression grew stronger as the multilayered nodules developed (Fig. 8, day 12). The cells within the nodule became cuboidal in shape and acquired a much stronger level of expression, so there were two levels of expression, one of low intensity surrounding the nodule, and one with stronger intensity, which would subsequently mineralize. The expression of the hOC-GFPtpz transgene first developed at day 13 in cells that were already strongly pOBCol3.6GFPcyan-positive and were localized within the mineralizing nodule (Fig. 8, day 13, red arrow). By day 16, another pattern of co-GFP expression was evident. Faint pOBCol3.6GFPcyanexpressing cells with elongated processes situated at the periphery of the mineral became strongly hOC-GFPtpzpositive (yellow arrow). Thus, two different populations of double-positive cells became apparent as the culture progressed. One subpopulation of stellate-shaped cells adjacent to the mineralizing region of the nodule expressed faint GFPcyan fluorescence and was strongly positive for hOC-GFPtpz. The second population expressed both transgenes very prominently in the cuboidal cells within the mineralized area of the nodule. In mCOB cultures the population of doubleexpressing stellate-shaped cells adjacent to the mineral were not as apparent as in MSC cultures, which is consistent with the previous observation (Fig. 3B). However, the temporal pattern of expression in the population of cuboidal double-positive cells within the mineralizing nodule was the same as in MSC cultures (data not shown). Histological Analysis of pOBCol3.6GFPcyan 3 hOC-GFPtpz Transgenics One-month-old femur from a pOBCol3.6GFPcyan 3 hOC-GFPtpz transgenic animal was sectioned for histo-

92

BILIC-CURCIC ET AL.

FIG. 6. Histological analysis of hOC-GFPtpz transgene expression. hOC-GFPtpz transgene expression was analyzed in frozen sections of femurs from 7-day, 1-month, 3-month, and 6-month-old animal (upper panel magnification 103; lower panel magnification 203). In 7-dayold femur, GFP expression was restricted to osteoblasts and osteocytes throughout the cortical bone, with a few positive osteoblasts in the growth plate. One-month-old animals show strongest GFP expression. The transgene is active in subsurface osteocytes (OC), in osteoblasts (OB) lining the surface of cortical and trabecular bone, and in hypertrophic chondrocytes (HC) in primary spongiosum. The GFP activity is regressing in 3-month and 6-month-old femurs and is visible only on the surface of the bone, around blood vessels (BV) in the cortical bone (white square in 3-month image, magnification 403), and in the periosteal cells adjacent to the articular surface (PCS, 6-month image). At all developmental stages, absence of positive osteoblasts in primary spongiosum is apparent except in a few scattered cells. High-resolution images of these femur sections can be seen and downloaded from our Web site: http//skeletalbiology.uchc.edu (click image center, then click title). OB, osteoblasts; HC, hypertrophic chondrocytes; OC, osteocytes; BV, blood vessels; PCS, para-articular cortical cells or periosteal cells adjacent to the articular surface. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

logical analysis. Figure 9A is a low-power view of a 103 scan of a whole-length femur section. Regional differences can be appreciated in the expression patterns of pOBCol3.6GFPcyan and hOC-GFPtpz transgenes. pOBCol3.6GFPcyan expression (blue) is evident only in the primary spongiosum. In contrast, hOC-GFPtpz (green) is expressed primarily on the endosteal surface of the diaphysis and femoral neck. Both transgenes are expressed by endosteal osteoblasts in the metaphyseal region (light blue). At higher power these regions show more heterogeneity in cellular composition than the low-power image conveys. Three examples are illustrated. Figure 9B shows trabecular bone in the secondary spongiosum in which pOBCol3.6GFPcyan-only and hOC-GFPtpz-only cells are interspersed between light blue cells along the bone surface. Figure 9C shows the preponderance of pOBCol3.6GFPcyan-only cells in the primary spongio-

sum and scattered hOC-GFPtpz-only cells in the hypertrophic zone. However, a few double-positive cells were present in the lower border of the spongiosum. The full extent of the cellular heterogeneity was found in the diaphyseal region (Fig. 9D), in which all three cell types were seen on the endosteal surface and in the vascular regions within the cortical bone.

DISCUSSION The success in defining levels of differentiation within the hematopoietic lineage is based on the empiric association of cell surface markers with biological properties of the cells indicating their capacity as a progenitor or for differentiated function (Cotta et al., 2003; Kondo et al., 2003). A similar strategy has not been possible for the osteoprogenitor lineage, in part due to the require-


FIG. 7. Histological analysis of an undecalcified femur section obtained from 6-week-old transgenic animal injected with xylenol orange (XO) prior to sacrifice (magnification 103). Colocalization between hOC-GFP expression (green) and XO deposition (red) was analyzed in a femur section. A: Endosteal surface of cortical bone. B: Trabecular bone in the metaphyseal region. Note that the periosteal cells adjacent to the articular surface (PCS) is not associated with the XO staining. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

ment for a dense connective tissue network for late osteoblast/osteocyte differentiation. GFP, when driven by a promoter that is activated at a particular level of cellular differentiation, may provide a strategy for identifying and isolating subpopulations of cells at increasing levels of osteoblast development. When examined in primary osteoblast cultures, pOBCol3.6GFP activates at low intensity concomitant with the onset of alkaline phosphatase (ALP) activity in cells that form the initial stages of a bone nodule (Kalajzic et al., 2002a). Microarray analysis of the fluorescent activated cytometry (FAC) separated cells show that the GFP-positive cells have lost expression of genes characteristic of a pericyte/myofibroblast, indicating that early progenitor cells within the GFP-negative population have committed to the osteoblast lineage (Kalajzic, 2005). Histological analysis of bones show fibroblastic-

FIG. 8. Transgene expression pattern of pOBCol3.6GFPcyan 3 hOC-GFPtpz transgenics in MSC cultures. Fluorescent images of MSC cultures were obtained on a daily basis (magnification 103). The first sign of hOC-GFP transgene expression (red arrow) was observed at day 13 in a cell previously expressing strong cyan transgene. With the progression of mineralization, two types of two color expressing cells are visible by 17 days of culture. Cells inside a mineralized nodule are expressing both transgenes very strongly (red arrow), but there is a subpopulation of stellate-shaped cells surrounding the nodule that express faint cyan signal and strong osteocalcin transgene (yellow arrow). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

93

94

BILIC-CURCIC ET AL.

FIG. 9. Analysis of pOBCol3.6GFPcyan 3 hOC-GFPtpz expression pattern in vivo. A: One-month-old decalcified femur was sectioned for histological analysis (magnification 103). Three regions expressing different transgenes can be appreciated. hOC-GFPtpz only transgene (green) is mainly expressed on the surface of cortical and trabecular bone and in the osteocytes. pOBCol3.6GFPcyan only (deep blue) is expressed in primary spongiosum and in some osteoblasts lining the bone surface. Double-positive cells are present as light blue areas on the endosteal surface of metaphyseal bone. B–D: Regions of this bone, shown as 203 images, reveal the intermingling of the three colors of osteoblastic cells; green arrows, GFPtpz cells; red arrows, GFPcyan cells; yellow arrows, doublepositive cells. Each is present on trabecular and cortical bone surface (B,D) and around blood vessels within cortical bone (D). In primary spongiosum, few osteoblasts express both colors (C). In addition, scattered hypertrophic chondrocytes express hOC-GFP only. Images of these femur sections can be seen on our Web site: http// skeletalbiology.uchc.edu (click image center, then click title).

shaped cells located in the periosteum, with a low level of GFP expression. Based on the bone cell culture and bone histology data, an association can be made between low-level expression of pOBCol3.6GFP and a preosteoblast-level of differentiation. More firmly established is the association of strong pOBCol3.6GFP and pOBCol2.3GFP expression with a mature osteoblast (Kalajzic et al., 2002a). These cells are always associated with a mineralizing nodule in vitro and mineralizing matrix in vivo. Microarray data from

FAC isolated pOBCol2.3GFP-positive cells show strong expression of osteoblast-specific genes relative to the GFP-negative cells from the same culture dish (Kalajzic, 2005). One drawback of the pOBCol2.3GFP marker gene is that its expression extends into the osteocytes so that this unique cell cannot be distinguished from mature osteoblasts. However, that problem appears to have been resolved with the development of a DMP1-GFP mouse whose expression pattern is strongly linked and restricted to osteocytes (Kalajzic et al., 2004). One reason for developing the hOC-GFP mouse line was the hope that it could further subdivide the population of pOBCol2.3GFP-positive cells. Specifically, do bone surface cells exhibit different levels of osteoblast differentiation that cannot be appreciated either with standard light histology or even with a broad osteoblast GFP marker gene? The two hOC-GFPtpz mouse lines that were developed with this human 3.8 kb promoter fragment seem to have the strength and fidelity for a visual marker of an osteocalcin-expressing osteoblast. Mature osteoblasts appear to be marked by the construct because its expression is limited to cells within the well-developed bone nodule in vitro and the FAC isolated cells strongly express the endogenous osteocalcin gene relative to the GFP-negative cell population from the same culture dish. Furthermore, in situ hybridization of bone and tooth sections demonstrated colocalization of cells positive for osteocalcin mRNA and the hOC-GFPtpz transgene. Because a strong GFP signal is retained in both decalcified and nondecalcified cryosections of bone, it is possible to correlate in vitro expression pattern to the pattern in intact bone. As anticipated, there is no hOCGFPtpz expression in the immature osteoblasts of the primary spongiosum, although cells covering the endosteal surface of the diaphyseal bone exhibit a strong fluorescence. There is a strong correlation between bone surface hOC-GFPtpz-positive cells and recently deposited bone matrix, suggesting that they are matrix-producing cells. The age of the animal is particularly important to the pattern of expression. In the 7-day-old mouse, strong GFPtpz-positive cells are present on the endosteal surface and throughout the cortical bone, while by 1 month of age only the most recently formed osteocytes are positive. By 3–6 months of age, no osteocytes within the cortical bone are labeled except for the foci of cells surrounding vascular regions. pOBCol3.6GFPcyan transgenic animals also demonstrate an age-dependent pattern. Unlike hOC-GFPtpz, its highest level of expression is observed in the primary spongiosum in 8-day pups. However, at later time points both transgenics have a similar expression pattern that reflects new bone formation in young animals and bone remodeling in older mice. Thus, in 1-month-old animals strong transgene expression is found on the bone surface and within bone in both trabecular and cortical bone. By 3–5 months the activity is restricted to bone surface cells in regions of endosteal and trabecular bone and does not extend into matrix embedded cells.


When the two marker genes are multiplexed, three different levels of osteoblast differentiation can be appreciated. First are the cells that in primary culture begin as weak expressing pOBCol3.6GFPcyan-positive cells and acquire strong pOBCol3.6GFPcyan activity followed by the onset of hOC-GFPtpz activity. These cells are roundshaped and are located within the mineralizing region of the bone nodule. Similar cells within intact bone that express both transgenes can be distinguished from those which express each transgene singly, and these cells (light blue) are located in regions of active bone formation both on the bone surface and within the vascular areas of cortical bone. Second are cells that acquire strong pOBCol3.6GFPcyan expression but do not activate the hOC-GFPtpz transgene. These cells (strong blue) are located in and around the mineralizing bone nodule in primary MSC and mCOB culture and are found at sites of new osteoblast differentiation primarily in the metaphyseal regions. Third are cells that begin as weak pOBCol3.6GFPcyan-positive cells that do not progress to strong expression at the time that hOC-GFPtpz is activated. These cells (strong green) are located within the bone nodule adjacent to the region of mineralization and have a stellate shape. Temporal studies of the same cells within a developing nodule suggest that these cells do not arise from cells that have previously shown strong pOBCol3.6GFPcyan activity, nor do they progress to cells with strong pOBCol3.6GFPcyan activity. It is not clear whether these cells become incorporated within the mineral, at which time they would acquire strong pOBCol3.6GFPcyan expression and lose the stellate shape. Within bone it is likely that they are represented as hOC-GFPtpz-positive-only cells because the weak pOBCol3.6GFPcyan expression cannot be appreciated. These cells are found within the endosteal surface of the diaphysis and the femoral neck region. However, they could also be represented by the band of hOC-GFPtpzpositive cells located in the metaphyseal region of the periosteum that extends to the articular cartilage. Mature osteocytes do not appear to be labeled by either marker gene. This study illustrates the heterogeneity of cell populations that by standard histological and molecular properties are grouped as a single level of osteoblast differentiation. Despite the wide use of osteocalcin as a marker of osteoblast differentiation in cell culture, the number of OC-expressing cells within the total cell population is probably less than 5%. Most of the genes that are associated with differentiated osteoblasts will not be osteoblast-specific and the relative importance of these commonly expressed genes will only be appreciated when the 5% of cells that have achieved full osteoblast differentiation are examined as an isolated population. Similarly, all cells covering bone surfaces do not have the same GFP profile, indicating that they have different levels of osteoblast differentiation. It can be expected that many of the transgenic and knockout murine models that affect bone mass accretion act at different levels of osteoblast development. The use of the transgene driven

95

marker genes for different levels of osteoblast development may be helpful in mapping where in the lineage a mutation is interfering with bone cell maturation, as has been done for mutations in the hematopoietic lineages (Li et al., 2003; Nishimura et al., 2002; Passegue et al., 2003). Because the expression patterns of a transgenic construct can be influenced by its integration site, it is necessary to use the same transgenic reporter mouse line to ensure a similar expression profile in different laboratory settings. The assignment to a level of differentiation, which is based on multiple biological criteria and may or may not reflect the activity of the endogenous gene from which it was derived, will remain reproducible within the specific transgenic line. Thus, the availability of a commonly shared repertoire of characterized marker genes that reflects the broad range of development along the osteoprogenitor lineage should help to better define the molecular and cellular determinants of normal bone biology. MATERIALS AND METHODS Generation of Transgenic Mice The GFP expression construct hOC-GFPtpz was cloned by inserting an expression cassette composed of a segment of the SV40 large intron containing a splice donor and acceptor site and a three-frame stop signal, the cDNA for the topaz variant of GFP and bovine growth hormone polyadenylation sequence. This segment was inserted into Xho1 and EcoR1 sites of the hOC plasmid (Sims et al., 1997) which consists of 3.8 kb of the human osteocalcin promoter, the first 10 bp of a nontranslated region, a multiple cloning site, and 3.5-kb fragment of DNA that encompasses part of the last exon, the 30 untranslated region, and additional downstream genomic sequence. Unique Hind III sites flank the osteocalcin construct and allow its excision from a pUC18 backbone. The construct was then isolated for microinjection into C57Bl/6 ova to generate transgenic animals. Two founders were identified by fluorescent microscopy of skinned tail segments and confirmed by polymerase chain reaction (PCR) using oligos directed against GFP. The transgene was maintained on the C57Bl/6 background as a homozygous breeder and was crossed into CD1 to increase the yield of experimental mice. Mice were genotyped with portable fluorescent goggles (Biological Laboratory Equipments Maintenance and Service, Budapest, Hungary). pOBCol3.6GFPcyan transgenic mice were produced using a strategy similar to the generation of pOBCol3.6GFPtpz transgenics previously characterized (Kalajzic et al., 2002a). The GFPtpz reporter was exchanged with eCFP (BD Biosciences, ClonTech, Palo Alto, CA) and renamed GFPcyan, which was inserted in the SmaI-HindIII sites of a Cla adapter plasmid containing bovine growth hormone polyadenylation signal (bPA). The GFPcyan-bPA was excised from the adaptor plasmid and inserted into the recipient collagen 3.6 plas-

96

BILIC-CURCIC ET AL.

mid described previously (Kalajzic et al., 2002a). Two lines were developed in a CD1 background that showed strong expression and good reproductive fitness when maintained as a homozygous breeder stock. Genotyping was performed using fluorescent goggles. Mice that are heterozygous for the GFP reporter were generated for the present study by crossing the breeders for either transgene with a nontransgenic (one color) or crossing both breeders (two colors). All of the studies reported here were performed on C57Bl/6 3 CD1 F1 offspring. Histological Evaluation of GFP Expression in Mouse Bone Femurs from 7-day, 1-month, 3-month, and 6-month-old hOC-GFPtpz transgenic mice and 1-month-old femurs from hOC-GFPtpz 3 pOBCol3.6GFPcyan animals were fixed in 4% paraformaldehyde for 1–2 days followed by decalcification in 15% EDTA for 1–5 days. The bones were then soaked in 30% sucrose in phosphate-buffered saline (PBS) for 1 day. All the processes of fixation, decalcification, and cryoprotection were done at 48C under constant agitation. Samples were embedded in frozen embedding medium (Thermo Shandon, Pittsburgh, PA) at –708C and 5-lm-thick full-length sections were made with the assistance of the CryoJane tape system (Instrumedics, Hackensack, NJ). The sections were examined by fluorescence microscopy using a Zeiss Axioplan 200 inverted microscope (Carl Zeiss MicroImaging, Thornwood, NY) with dual filter cube optimized for GFPtpz (Chroma 51004v2) or GFPcyan (Chroma 51018). Images were taken with Zeiss Axiocam MR camera (Carl Zeiss MicroImaging). No influence of one transgene over the other was observed in histology sections and their detection was not compromised due to a different emission and excitation profiles of these two specific GFP isomers. Composite high-power images of the bone were generated by Improvision software (Lexington, MA) controlled mechanical stage, which recorded adjacent images taken separately with either the topaz or cyan filter cube. The original composite image files used to produce the images here can be downloaded for viewing at http//skeletalbiology.uchc.edu (click image center and select this article). Undecalcified femurs for sectioning were obtained from 6-week-old animals injected with 0.09 mg/g body weight of xylenol orange (XO) (Sigma Chemical, St. Louis, MO) 48 h prior to sacrifice. Bones were processed as previously described with the exception of the decalcification step. Primary Osteoblast Cultures Bone marrow stromal cell (MSC) cultures obtained from 6–8-week-old transgenic mice and primary mouse calvarial osteoblast cell (mCOB) culture were prepared using the method previously described (Kalajzic et al., 2002a). At days 15, 18, and 21 xylenol orange (Sigma Chemical) was added (20 lM) 24 h prior to taking images to iden-

tify newly deposited mineral in the cell culture (Wang et al., 2005). The following day, a-MEM was changed and images were obtained. All images were taken using a Zeiss Axioplan 200 inverted microscope with single filter cube optimized for GFPtpz (Chroma 41028-29), GFPcyan (Chroma 31044v2), or Tritc (Chroma 31002) using the Improvision software. Detection of two transgenes was based on the same principle as in histology sections. In Situ Hybridization The patterns of expression of hOC-GFP in the developing facial bones and adult femurs were examined by in situ hybridization of tissue sections using 33P-labeled riboprobes as previously described (Braut et al., 2003). A 500-bp fragment of mouse OCcDNA in Bluescript was digested with EcoRI or PstI and transcribed with T3 or T7 RNA polymerase for antisense and sense probes, respectively. For comparative analyses of the patterns of expression, serial sections were used so that expression patterns for hOC-GFP and mouse mRNA OC could be related within the same animal. Slides were left to develop between 2 days and 2 weeks. Following hybridization and development of the emulsion, sections were counterstained. Photographs were taken under darkfield or brightfield illumination with a Zeiss Axioplan 200 inverted microscope. The silver grains in the darkfield image were selected, false-colored red, and superimposed onto the brightfield image. Histochemical Analysis of Cell Cultures Histochemical staining for ALP activity was performed using a commercially available kit (86-R Alkaline Phosphatase; Sigma Diagnostics, St. Louis, MO) according to the manufacturer’s instructions. Mineralization was assessed using a modified von Kossa silver nitrate staining method (Kalajzic et al., 2002a). RNA Extraction and Northern Analysis Animals were sacrificed by CO2 asphyxiation followed by cervical dislocation and excised soft tissues were frozen immediately in liquid nitrogen in a 15-ml polypropylene tube (Falcon Cat 2059; Becton Dickinson, Franklin Lakes, NY). Frozen samples were suspended in 3 ml of TRIzol Reagent (Gibco, Invitrogen, Carlsbad, CA) and immediately homogenized with a 5-mm Polytron probe (Brinkman, Westbury, NY) for 30 s. Total RNA was prepared from mouse tissues and cultured cells using TRIzol Reagent according to the manufacturer’s instructions. RNA was separated on a 2.2 M formaldehyde / 1% agarose gel and transferred onto a nylon membrane (Maximum Strength Nytran; Schleicher & Schuell, Keene, NH). Membranes were probed with the radiolabeled 0.7-kb GFP fragment, a 900-bp PstI rat Col1a1 fragment (Genovese et al., 1984), a 1,000-bp EcoRI mouse BSP fragment (Young et al., 1994), and a 400-bp murine OC cDNA fragment (Celeste et al., 1986).


Flow Cytometry Primary neonatal calvarial cell cultures grown in differentiation media for 21 days were washed in PBS and then digested in 2.5% trypsin (Gibco, Invitrogen) with 0.2% of collagenase A (Roche, Indianapolis, IN) and 0.2% of hyalouronidase (Sigma Chemical) for 20 min at 378C. DMEM/10%FCS was added to neutralize the enzyme and the cell layer was disrupted with repetitive pipetting and transferred into 50-ml Falcon tubes. Visual inspection performed after this procedure showed quantitative removal of mineralized areas. Cells were centrifuged, resuspended in PBS, filtered through a 70-lm nylon cell strainer (BD Biosciences, Bedford, MA), centrifuged again, and resuspended in PBS/ 2%FCS. Cell sorting was done using a FACS vantage sorter (Becton Dickinson) using an Argon laser with excitation at 488 nm and emission 530/30 FITC bandpass filter. Due to the large size of the cells separation was done at a speed of 3,000–5,000 cells/sec using a 100-lm nozzle. Cells were collected into DMEM/ 30%FCS media. Prior to, during, and following sorting, the cell suspension was kept on ice to minimize potential changes in gene expression.

LITERATURE CITED Aubin JE, Turksen K. 1996. Monoclonal antibodies as tools for studying the osteoblast lineage. Microsc Res Tech 33:128–140. Aubin JE, Liu F, Malaval L, Gupta AK. 1995. Osteoblast and chondroblast differentiation. Bone 17:77S–83S. Braut A, Kollar EJ, Mina M. 2003. Analysis of the odontogenic and osteogenic potentials of dental pulp in vivo using a Col1a1-2.3-GFP transgene. Int J Dev Biol 47:281–292. Byers PH, Starman BJ, Cohn DH, Horwitz AL. 1988. A novel mutation causes a perinatal lethal form of osteogenesis imperfecta. An insertion in one alpha 1(I) collagen allele (COL1A1). J Biol Chem 263:7855–7861. Celeste AJ, Rosen V, Buecker JL, Kriz R, Wang EA, Wozney JM. 1986. Isolation of the human gene for bone gla protein utilizing mouse and rat cDNA clones. EMBO J 5:1885–1890. Chen XD, Qian HY, Neff L, Satomura K, Horowitz MC. 1999. Thy-1 antigen expression by cells in the osteoblast lineage. J Bone Miner Res 14:362–375. Clemens TL, Tang H, Maeda S, Kesterson RA, Demayo F, Pike JW, Gundberg CM. 1997. Analysis of osteocalcin expression in transgenic mice reveals a species difference in vitamin D regulation of mouse and human osteocalcin genes. J Bone Miner Res 12:1570– 1576. Cotta CV, Zhang Z, Kim HG, Klug CA. 2003. Pax5 determines B- versus T-cell fate and does not block early myeloid-lineage development. Blood 101:4342–4346. Dacic S, Kalajzic I, Visnjic D, Lichtler AC, Rowe DW. 2001. Col1a1driven transgenic markers of osteoblast lineage progression. J Bone Miner Res 16:1228–1236. Erlebacher A, Filvaroff EH, Gitelman SE, Derynck R. 1995. Toward a molecular understanding of skeletal development. Cell 80:371– 378. Frenkel B, Capparelli C, Van Auken M, Baran D, Bryan J, Stein JL, Stein GS, Lian JB. 1997. Activity of the osteocalcin promoter in skeletal sites of transgenic mice and during osteoblast differentiation in bone marrow-derived stromal cell cultures: effects of age and sex. Endocrinology 138:2109–2116. Genovese C, Rowe D, Kream B. 1984. Construction of DNA sequences complementary to rat alpha 1 and alpha 2 collagen mRNA and

97

their use in studying the regulation of type I collagen synthesis by 1,25-dihydroxyvitamin D. Biochemistry 23:6210–6216. Gronthos S, Zannettino AC, Graves SE, Ohta S, Hay SJ, Simmons PJ. 1999. Differential cell surface expression of the STRO-1 and alkaline phosphatase antigens on discrete developmental stages in primary cultures of human bone cells. J Bone Miner Res 14:47–56. Jiang X, Kalajzic Z, Maye P, Braut A, Bellizzi J, Mina M, Rowe D. 2005. Histological analysis of GFP expression in murine bone. J Histol Cytometry 53:593–602. Kalajzic I, Kalajzic Z, Kaliterna M, Gronowicz G, Clark SH, Lichtler AC, Rowe D. 2002a. Use of type I collagen green fluorescent protein transgenes to identify subpopulations of cells at different stages of the osteoblast lineage. J Bone Miner Res 17:15–25. Kalajzic Z, Liu P, Kalajzic I, Du Z, Braut A, Mina M, Canalis E, Rowe DW. 2002b. Directing the expression of a green fluorescent protein transgene in differentiated osteoblasts: comparison between rat type I collagen and rat osteocalcin promoters. Bone 31:654–660. Kalajzic I, Braut A, Guo D, Jiang X, Kronenberg MS, Mina M, Harris MA, Harris SE, Rowe DW. 2004. Dentin matrix protein 1 expression during osteoblastic differentiation, generation of an osteocyte GFP-transgene. Bone 35:74–82. Kalajzic I, Staal A, Yang WP, Wu Y, Johnson SE, Feyen JHM, Krueger W, Maye P, Yu F, Zhao Y, Kuo L, Gupta RR, Achenie LEK, Wang H-W, Shin D-G, Rowe DW. 2005. Expression profile of osteoblast lineage at defined stages of differentiation. J Biol Chem 280:24618– 24626. Kerner SA, Scott RA, Pike JW. 1989. Sequence elements in the human osteocalcin gene confer basal activation and inducible response to hormonal vitamin D3. Proc Natl Acad Sci U S A 86:4455–4459. Kesterson RA, Stanley L, DeMayo F, Finegold M, Pike JW. 1993. The human osteocalcin promoter directs bone-specific vitamin D-regulatable gene expression in transgenic mice. Mol Endocrinol 7:462– 467. Kondo M, Wagers AJ, Manz MG, Prohaska SS, Scherer DC, Beilhack GF, Shizuru JA, Weissman IL. 2003. Biology of hematopoietic stem cells and progenitors: implications for clinical application. Annu Rev Immunol 21:759–806. Li X, Plett PA, Yang Y, Hong P, Freie B, Srour EF, Orschell CM, Clapp DW, Haneline LS. 2003. Fanconi anemia type C-deficient hematopoietic stem/progenitor cells exhibit aberrant cell cycle control. Blood 102:2081–2084. Liu F, Malaval L, Gupta AK, Aubin JE. 1994. Simultaneous detection of multiple bone-related mRNAs and protein expression during osteoblast differentiation: polymerase chain reaction and immunocytochemical studies at the single cell level. Dev Biol 166:220– 234. Manolagas SC. 2000. Birth and death of bone cells: basic regulatory mechanisms and implications for the pathogenesis and treatment of osteoporosis. Endocr Rev 21:115–137. Marie PJ, de Vernejoul MC, Connes D, Hott M. 1991. Decreased DNA synthesis by cultured osteoblastic cells in eugonadal osteoporotic men with defective bone formation. J Clin Invest 88:1167– 1172. Morrison NA, Shine J, Fragonas JC, Verkest V, McMenemy ML, Eisman JA. 1989. 1,25-Dihydroxyvitamin D-responsive element and glucocorticoid repression in the osteocalcin gene. Science 246:1158– 1161. Nishimura J, Ware RE, Burnette A, Pendleton AL, Kitano K, Hirota T, Machii T, Kitani T, Smith CA, Rosse WF. 2002. The hematopoietic defect in PNH is not due to defective stroma, but is due to defective progenitor cells. Blood Cells Mol Dis 29:159–167. Oreffo RO, Bord S, Triffitt JT. 1998. Skeletal progenitor cells and ageing human populations. Clin Sci (Lond) 94:549–555. Passegue E, Jamieson CH, Ailles LE, Weissman IL. 2003. Normal and leukemic hematopoiesis: are leukemias a stem cell disorder or a reacquisition of stem cell characteristics? Proc Natl Acad Sci U S A 100(Suppl 1):11842–11849. Pockwinse SM, Lawrence JB, Singer RH, Stein JL, Lian JB, Stein GS. 1993. Gene expression at single cell resolution associated with development of the bone cell phenotype: ultrastructural and in situ hybridization analysis. Bone 14:347–352.

98

BILIC-CURCIC ET AL.

Sims NA, White CP, Sunn KL, Thomas GP, Drummond ML, Morrison NA, Eisman JA, Gardiner EM. 1997. Human and murine osteocalcin gene expression: conserved tissue restricted expression and divergent responses to 1,25-dihydroxyvitamin D3 in vivo. Mol Endocrinol 11:1695–1708. Turksen K, Aubin JE. 1991. Positive and negative immunoselection for enrichment of two classes of osteoprogenitor cells. J Cell Biol 114:373–384. Van Vlasselaer P, Falla N, Snoeck H, Mathieu E. 1994. Characterization and purification of osteogenic cells from murine bone marrow by two-color cell sorting using anti-Sca-1 monoclonal antibody and wheat germ agglutinin. Blood 84:753–763. Wang YH, Liu Y, Buhl K, Rowe DW. 2005. Comparison of the action of transient and continuous PTH on primary osteoblast cultures

expressing differentiation stage-specific GFP. J Bone Miner Res 20:5–14. Weinreb M, Shinar D, Rodan GA. 1990. Different pattern of alkaline phosphatase, osteopontin, and osteocalcin expression in developing rat bone visualized by in situ hybridization. J Bone Miner Res 5:831–842. Young MF, Ibaraki K, Kerr JM, Lyu MS, Kozak CA. 1994. Murine bone sialoprotein (BSP): cDNA cloning, mRNA expression, and genetic mapping. Mamm Genome 5:108–111. Zannettino AC, Harrison K, Joyner CJ, Triffitt JT, Simmons PJ. 2003. Molecular cloning of the cell surface antigen identified by the osteoprogenitor-specific monoclonal antibody, HOP-26. J Cell Biochem 89:56–66.