JOURNAL OF CELLULAR PHYSIOLOGY 143:420-430 (1990)

Progressive Development of the Rat Osteoblast Phenotype In Vitro: Reciprocal Relationships in Expression of Genes Associated With Osteoblast Proliferation and Differentiation During Formation of the Bone Extracellular Matrix T H O M A S A. O W E N , MICHAEL A R O N O W , VICTORIA SHALHOUB, LEESA M. BARONE, LAURENS W I L M I N G , MELISSA S. TASSINARI, MARY BETH KENNEDY, S H l R W l N POCKWINSE, JANE B. LIAN, AND GARY S. STEIN Department of Cell Biology, University of Massachusetts Medical Center, Worcester, Massachusetts 0 1655 The relationship of cell proliferation to the temporal expression of genes characterizing a developmental sequence associated with bone cell differentiation was examined in primary diploid cultures of fetal calvarial derived osteoblasts by the combined use of autoradiography, histochemistry, biochemistry, and mRNA assays of osteoblast cell growth and phenotypic genes. Modifications in gene expression define a developmental sequence that has 1 ) three principle periodsproliferation, extracellular matrix maturation, and mineralization-and 2) two restriction points to which the cells can progress but cannot pass without further signals-the first when proliferation is down-regulated and gene expression associated with extracellular matrix maturation is induced, and the second when mineralization occurs. Initially, actively proliferating cells, expressing cell cycleand cell growth-regulated genes, produce a fibronectinhype I collagen extracellular matrix. A reciprocal and functionally coupled relationship between the decline in proliferative activity and the subsequent induction of genes associated with matrix maturation and mineralization is supported by 1) a temporal sequence of events in which there i s an enhanced expression of alkaline phosphatase immediately following the proliferative period, and later, an increased expression of osteocalcin and osteopontin at the onset of mineralization; 2) increased expression of a specific subset of osteoblast phenotype markers, alkaline phosphatase and osteopontin, when proliferation i s inhibited by hydroxyurea; and 3) enhanced levels of expression of the osteoblast markers as a function of ascorbic acid-induced collagen deposition, suggesting that the extracellular matrix contributes to both the shutdown of proliferation and the development of the osteoblast phenotype.

Osteoblasts isolated f r o m t h e c a l v a r i a o f 21 day f e t a l r a t p ups d ifferentiate in c u l t u r e as reflected by t h e p r o duction and deposition of an organized collagen ext r a cellular matrix and t h e appearance of nodules cons i s t i n g of m u l t i p l e layers of cells within a mine r a l i z e d e x tra cellular matrix (Bellows e t al., 1986; Bhargava e t al., 1988; Escarot-Charrier e t al., 1983, 1988). C u l t u r e conditions that support high levels of productio n of osteoblast m a r k e r s such as t y p e I collagen and osteocalc i n h a v e been demonstrated, and, u n d e r such conditions, a t e m p o r a l sequence of expression o f t y p e I collagen, a l k a l i n e phosphatase, osteocalcin, and mine r a l deposition h a s been observed (Aronow e t al., 1990; Gerstenfeld e t al., 1987; S t e i n e t al., 1989a), r e f l e c t i n g progressive osteoblast differentiation. T h e q u e s t i o n

@> 1990 WILEY-LISS,

INC

arises o f h o w m o d i f i c a t i o n s in gene expression m e d i a te these biochemical parameters. Equally i m p o r t a n t i s d e t e r m i n a t i o n o f t h e s i g n a l l i n g mechanisms for t h e t e m p o r a l expression of genes encoding t h e osteoblast phenotype and t h e r e l a t i o n s h i p s b e t w e e n genes expressed during t h e developmental sequence o f osteoblast differentiation. In t h e present studies, w e e x a m i n e d t h e selective expression o f v a r i o u s c e l l g r o w t h and d i f f e r e n ti a ti o n associated genes as r e f l e c t e d by mRNA levels, w h i c h serve as m a r k e r s f o r cells at d i f f e r e n t stages o f matur a t i o n t h r o u g h o u t t h e osteoblast developmental se-

Received December 19, 1989; accepted February 9, 1990.

GENE EXPRESSION DURING OSTEOBLAST DIFFERENTIATION

42 1

RNA isolation and analysis For each time point and experimental condition analyzed, cells were scraped and pooled from 3 x 100 mm dishes in phosphate-buffered saline (PBS) and pelleted by centrifugation. The PBS was decanted, and the cell pellet was immediately frozen in liquid nitrogen and stored a t - 70°C until the experiment was completed. Total cellular RNA was isolated from each cell pellet by the SDS-proteinase K method as described by Plumb et al. (1983). RNA preparations were quantitated by absorbance a t 260 nm and intactness assessed by ethidium bromide staining following separation in 6.6% formaldehyde-1% agarose gels. RNA fractionated in such gels was transferred to Zeta-Probe membrane (BioRad, Richmond, CA) in 20 x SSC by the capillary method of Thomas (1983). In other cases, RNA samples were bound to Zeta-Probe blotting membrane by slot blot under conditions described by the apparatus manufacturer (Schleicher and Schuell, Keene, NH). DNA probes used for hybridization were the following: rat H4 histone (Grimes et al., 1987), human c-fos (Curran et al., 19831, mouse c-myc (Toscani et al., 1988), rat type I collagen (Genovese et al., 19841, r a t type I1 collagen (Kohno et al., 1984), rat fibronectin (Schwarzbauer et al., 1983), human transforming growth factor p (TGF-p) (Derynck et al., 1985), r a t alkaline phosphatase (Noda et al., 19871, rat osteopontin (Oldberg et al., 1986), and rat osteocalcin (Lian et al., 1989). All DNA probes were labeled with 32P-dCTPby the random primer method (Feinberg and Vogelstein, 19831. All prehybridizations and hybridizations were performed in 50% formamide; 5 x SSC; 5 x Denhardts; MATERIALS AND METHODS 50 mM PO, buffer, pH 6.5; 1%SDS; and 250 bg/ml salmon sgerm DNA at 42°C for 18 hr. For hybridizaCell culture tions, 10 cpmiml probe and 10% dextran sulfate were Calvaria from fetal rats of 21 days gestation were added (Maniatis et al., 1982). Following hybridization, isolated and subjected to sequential digestions of 20, blots were washed twice a t room temperature and once 40,and 90 min at 37°C in 2 mg/ml collagenase A (Boeh- at 65°C in 2 x SSC/O.l% SDS (15 min per wash) and ringer-Mannheim, Indianapolis, IN)/0.25% trypsin then twice a t 65°C in 1x SSC/O.l% SDS for nonhomol(Gibco, Grand Island, NY) (Aronow et al., in press). The ogous probes or in 0.1 x SSC/O.l% SDS for homologous cells of the first two digests were discarded, and those probes (30 min per wash). The resulting autoradioreleased from the third digestion were plated in mini- graphs were quantitated by scanning laser densitomemal essential medium (MEM; Gibco) supplemented try. Each time point represents the average of tripliwith 10% fetal calf serum (FCS) in 100 mm dishes or cate assays from a t least two independent experiments. six-well dishes (Corning, Corning, NY) a t a density of 5 x lo5 or 3.6 x lo5 cells/dish, respectively. At conTransmission electron microscopy fluence (day 7), the time course of mineralization was Cultured fetal rat osteoblasts growing on glass covaccelerated in some cultures by the addition of BGJb medium (Gibco) supplemented with 10% FCS, 50 pg/ml erslips and isolated intact 21 day gestation fetal rat ascorbic acid, and 10 mM P-glycerol phosphate. In calvaria were fixed in 3% glutaraldehyde in 0.1 M soother cultures the concentration of ascorbic acid was dium cacodylate buffer, pH 7.4, with 3 mM CaC1, for 5 varied, and in yet other cultures no P-glycerol phos- h r at room temperature. Intact rat calvaria were dephate was added. In some experiments, proliferation mineralized by incubation in 10% EDTA, pH 7.4, for 5 was inhibited in actively proliferating cells (day 5 ) by days a t room temperature. Specimens were postfixed in the addition of 5 mM hydroxyurea directly to the cul- 1% osmium tetroxide in 0.1 M sodium cacodylate ture medium. Cell number was determined for points buffer, pH 7.4; dehydrated; and embedded in Poly/Bed after plating by extensively incubating three indepen- 812 (Polysciences, Warrington, PA). After sectioning, dent wells of cells with 0.25% trypsin and counting the specimens were stained with 4% uranyl acetate and cells in a hemocytometer. When calvarial osteoblasts Reynolds’ lead citrate and examined in a Jeol 100s were isolated from 21 day fetal rats, the time of max- transmission electron microscope. imal expression of the principle phenotype markers of Histochemical analysis the 35 day developmental sequence did not vary by Cells growing on 22 mm glass coverslips were rinsed more than 48 h r in independent experiments using a standardized lot of fetal calf serum (Aronow et al., twice with ice-cold PBS and fixed for 10 min in absolute methanol ( -20°C). Alkaline phosphatase activity was 1990).

quence. Our studies show not only that a temporal sequence of expression of the genes encoding osteoblast phenotypic markers occurs in culture but also that the pattern of their expression defines three distinct periods separated by two transition points. Initially, there is a period of active proliferation during which cell growth-related genes are actively expressed, and maximal levels of type I collagen mRNA are observed. Following the down-regulation of proliferation, a period of matrix maturation occurs, when the alkaline phosphatase gene is maximally expressed and the extracellular matrix is rendered competent for mineralization, the third period of the developmental sequence. Evidence is provided for two transitions between the developmental periods representing restriction points to which cells can proceed but cannot pass without other signals. The first transition occurs with completion of the proliferative period, and the second is reached when the expression of genes such a s osteocalcin and osteopontin become significantly elevated, with the onset of mineralization. In addition, our results support two concepts related to the progressive development of the osteoblast phenotype. First, the proliferative period is requisite for collagen gene expression and initial production of a collagen matrix that subsequently supports maturation of the osteoblast phenotype. Second, there exists a reciprocal and functionally coupled relationship between the down-regulation of proliferation and initiation of expression of osteoblast phenotypic markers such a s alkaline phosphatase and osteopontin.

422

OWEN ET AL.

visualized by incubating the cells for 30 min a t room temperature with shaking in 10 mM Tris HC1, pH 8.4, containing 20 mg/ml disodium napthol AS-MX phosphate and 40 mgiml fast red TR salt (Sigma Chemical Co., St. Louis, MO). Mineral deposition was assessed by von Kossa staining of the cultures (30 min in 3% AgN03) (Clark, 1981). DNA synthesis and autoradiography For determination of the rate of DNA synthesis at each point indicated, 3H-thymidine (20 Ci/mmol; Amersham, Arlington Heights, IL) was added to the medium of cells growing in three wells of a six-well dish to 10 p.Ci/ml for 1 hr. Cells were rinsed with PBS and incubated with 5% (w/v) trichloroacetic acid (TCA) for 5 min. This TCA incubation was repeated, each cell layer solubilized in 0.5 ml 10% SDS, and counted in a scintillation counter following addition of 4 ml fluor. For in situ autoradiography, 3H-thymidine was added to the medium of cells growing on 22 mm glass coverslips to 1 yCi/ml for 24 hr. Cells were then rinsed twice in ice-cold PBS and fixed in absolute methanol (-20°C). Coverslips were then air dried overnight or first stained for alkaline phosphatase as described above and then dried. Autoradiography was performed using Ilford K-5 emulsion as described by Baserga and Malamud (1969). Exposures were for 5-7 days at 4°C. Biochemical determinations For protein, collagen, and mineral content, cell layers were hydrolyzed in 6 N HC1 under vacuum at 110°C for 24 hr. Aliquots of the hydrolysates were diluted in 0.5% lanthium chloride for determination of calcium concentration by atomic absorption spectroscopy in a Perkin-Elmer model 2001 spectrograph. Other aliquots were assayed for total amino acid composition using a Beckman 121M autoanalyzer and collagen content was calculated from the hydroxyproline concentration. Other cell layers were harvested by scraping in appropriate buffers for each individual assay. Total DNA was assayed by a fluorometric procedure using diaminobenzoic acid reagent as described by Vysatek (1982). Alkaline phosphatase activity was determined spectrophotometrically as described by measuring the amount of p-nitrophenol formed at 37°C after 30 min (Lowry et al., 1954). The medium from all these cell layers was simultaneously analyzed for osteocalcin by radioimmunoassay as previously described (Gundberg et al., 1984). For all biochemical determinations, the value represents the mean of three independent samples. RESULTS AND DISCUSSION Temporal sequence of gene expression during the osteoblast developmental sequence During a 35 day period, primary cultures of calvarial-derived osteoblasts progressively develop a bone tissue-like organization consisting of multilayered nodules of cells in an ordered, mineralized collagen extracellular matrix. To address the extent to which genes associated with the develoDment of the bone cell Dhenotype in vitro are expresskd, cellular levels of mRNA for cell growth and tissue-specific genes were determined at selected time points throughout a 35 day culture period. Figure l shows the results of Northern blot

A.

I

I

z

Pln ln W

EX W

9E 9 I-

z W

V

a W

a

B. z 0 v) W v)

U 0. X

100 -

80 -

W

60 -

X

z

40 -

5

20 -

sz I-

W

0.

04 0

10

20

30

40

0

10

20

30

JC

C. z 11. v) v)

W

a a X W

9R

; c W z

0

a W

0.

DAYS IN CULTURE

Fig. 1. Temporal expression of cell growth-, extracellular matrix-, and osteoblast phenotype-related genes during the development of the osteoblast phenotype in vitro. Primary isolated cells were cultured after confluence in BGJb medium supplemented with 10% FCS, 50 Fg/ml ascorbic acid, and 10 mM B-glycerol phosphate. Cellular RNA was isolated at the times indicated (3, 5, 7, 10, 12, 14,16, 20, 28, and 35 days) during the differentiation time course and assayed for the steady state levels of various transcripts by Northern blot or slot blot analysis. The resulting blots were quantitated by scanning densitometry and the results plotted relative to the maximal expression of each transcript. A DNA synthesis ( 0 ) and cell growth-related genes: H4 histone ( 0 ) ; c-myc (w); c-fos (0). B Extracellular matrix-associated genes: type I collagen (0); fibronectin (0); TGF-B ( 0 ) . C: Osteoblast phenotype marker genes: osteopontin ( 0 ) ; osteocalcin (0); alkaline phosphatase (0). Note the induction of alkaline phosphatase at the end of proliferation and the induction of osteocalcin and osteopontin with the onset of calcium deposition (day 12). Calcium accumulation is represented in Figure 2

GENE EXPRESSION DURING OSTEOBLAST DIFFERENTIATION

analysis of RNA transcribed from three different classes of genes. Immediately following isolation of the cells from calvaria, the protooncogene c-fos was expressed at its highest level, and c-myc was maximally expressed 1 day later (Fig. 1A). The expression of the cell growth-associated H4 histone gene (Fig. 1A) reflects its well-known coupling to DNA synthesis (Plumb et al., 1983), and showed that the isolated rat osteoblasts undergo a n initial period of proliferation during the first 10-12 days in culture. The down-regulation of the expression of H4 histone and the protooncogenes c-fos and c-myc marks the end of the proliferative period, with their mRNA levels less than 20%of maximum by day 15. Figure 1B shows that several genes associated with formation of the extracellular matrix, type I collagen, fibronectin, and TGF-P were maximally expressed during the period of proliferation and were gradually down-regulated during the subsequent stages of the cultures. Type I1 collagen mRNA, a phenotypic marker for cartilage (Kohno et al., 19841, was not detectable a t any time in the osteoblast cultures, indicating the absence of chondrocytes either coisolated with the osteoblasts from calvaria or chondrocytes derived from potential progenitor cells during the 35 day culture period. The expression of several genes associated with the bone cell phenotype is shown in Figure 1C. The expression of alkaline phosphatase mRNA increases greater than ten-fold immediately following the down-regulation of proliferation during the period from 12 to 18 days. Then, as the cultures progressed into the mineralization stage, cellular levels of alkaline phosphatase mRNA declined. Two other bone-related genes, osteopontin and osteocalcin, exhibited a different pattern of expression. Osteocalcin was not detectable prior to day 12 in culture and did not reach a significant level of expression until 16-20 days after isolation when its expression rose coordinately with the increase in total mineral accumulation in the cultures. Osteopontin similarly reached its peak expression during the mineralization period (days 16-20). It was also expressed during the period of active proliferation (25% of maximal levels), at days 5 and 7 (Fig. 10. The expression of alkaline phosphatase mRNA and the onset of enzyme activity (one of the bone phenotype markers), preceded any significant mineralization of the osteoblast cultures, as measured either by mineral accumulation or by the increase in osteocalcin levels. These results suggest that alkaline phosphatase may be involved in preparation of the extracellular matrix for the ordered deposition of mineral and that the coexpression of other genes such as osteocalcin and osteopontin may support the onset and progression of extracellular matrix mineralization. Alternatively, the induction of expression of these genes may reflect a n acquisition of osteoblast properties associated with signalling in bone turnover. The expression of at least two of these genes follows the pattern of expression of their proteins, which were determined biochemically. Alkaline phosphatase enzyme activity, measured both by histochemical staining and spectrophotometrically, closely followed the expression of its mRNA (compare Fig. 1C and Fig. 2). Similarly, the increase in osteocalcin mRNA paralleled the levels of osteocalcin detect-

423

able by radioimmunoassay. For example, osteocalcin synthesis on day 12 measured 0.8 nglyg DNA, increased to 3.7 nglyg DNA by day 16, and to 9.5 nglpg DNA by day 20. These increases occur within the period of mineral accumulation in these cultures (see Fig. 2 for calcium deposition). Taken together, the patterns of expression of these genes demonstrate that a temporal sequence of gene expression exists during the culture period associated with development of the extracellular matrix and reflects maturation of the osteoblast phenotype in vitro. It is of interest to note that osteopontin mRNA exhibits a biphasic pattern of expression during the osteoblast developmental sequence, with mRNA levels during the period of cell proliferation that are about 25% those observed during the period of mineralization. This is not a n unexpected result for two reasons. First, osteopontin expression during the proliferative period is consistent with its increased level of expression during the prereplicative phase of the cell cycle following serum stimulation of quiescent fibroblasts (Nomura et al., 1988). Second, its induced expression coincident with mineralization may be related to physical properties of the protein. Osteopontin is a 60 kd acidic glycoprotein (Oldberg et al., 1986) containing o-phosphoserine, thereby possessing several putative calcium binding sites, which are known to be importa n t both for cell proliferation and also for mineralization of the extracellular matrix in bone (Glimcher, 1985). It therefore appears th a t expression of the osteopontin gene early and late in the osteoblast developmental sequence may be mediated by alternative regulatory mechanisms. The possibility should also be considered that the lower levels of osteopontin mRNA observed during the proliferative period may in part reflect mRNA transcribed in vivo in osteoblasts undergoing matrix mineralization prior to their isolation from the fetal calvaria. Role of the extracellular matrix in the shutdown of proliferation and expression of the osteoblast p h e n o ty p e Although the temporal patterns of expression of these representative cell growth- and osteoblastrelated genes define discrete periods within a sequence of events that supports or promotes development of the osteoblast phenotype in vitro, they do not provide a direct indication of specific mechanisms by which expression of these genes mediates progressive bone cell phenotype development. Here, a particularly relevant question is whether these temporal events are functionally coupled or causally related. Functional relationships between expression of genes and events associated with osteoblast phenotype development are suggested by their patterns of expression. It can be inferred from the high levels of type I collagen mRNA expression during osteoblast proliferation that the proliferative period supports synthesis and deposition of collagen required for subsequent events th a t are necessary to render the collagen matrix competent for the ordered deposition of mineral. Experimental results that are consistent with a functional a s well as a temporal relationship between proliferation and collagen synthesis are shown in Figure 2. The effect of various ascorbic acid concentrations (0,

424

OWEN ET AL.

300] U 0

A

7

Y I

a

gls

w

m

5z 1 UI 1

20 30: 10

U 0

10

20

200

C W z

0

4-1

8 c

0

0

30

p::::::i-l 10

20

10

20

D

30

7

100

30

DAYS AFTER ISOLATION

0

0

10

20

30

DAYS AFTER ISOLATION

Fig. 2. Effect of ascorbic acid and collagen accumulation on osteoblast proliferation (cell number) and differentiation parameters (collagen content, alkaline phosphatase activity, and calcium deposition). Cells were maintained during a 30 day culture period, first in MEM medium (until day 7) and then in BGJb medium supplemented with 10% FCS, 10 mM p-glycerol phosphate and containing 0 (*L 25 (n), or 50 (W pg/ml ascorbic acid. Values plotted represent the mean of three sample wells per time point. Cell number (A) was determined after trypsinization. Alkaline phosphatase (B) is expressed as nM p-nitro-

phenol per microgram DNA. Percent collagen (C) was determined from the hydroxyproline concentration in total amino acid analysis. Calcium (D) was determined by atomic absorption spectroscopy in the same samples hydrolyzed for the amino acid analysis. With increasing ascorbic acid concentrations, the cells reached confluency at a lower density and there was an increase in collagen accumulation associated with a parallel increase in alkaline phosphatase activity. In the absence of a collagen matrix, no accumulation of calcium occurred, reflecting the absence of mineralization.

25, or 50 pglml), a requirement for collagen synthesis, on parameters of the osteoblast developmental sequence was examined. It can be seen in Figure 2A that cultures incubated in medium with 50 pglml ascorbic acid plateaued with the lowest cell number (105 x lo4 cells on day 28) whereas cells cultured in the presence of 25 p,g/ml ascorbic acid plateaued at a higher density (150 x lo4 cells). In contrast, cells maintained in the absence of ascorbic acid exhibited continued growth throughout the culture period, reaching the highest cell number on day 28 (220 x lo4 cells). Total collagen accumulation was increased as a function of ascorbic acid concentration (Fig. 2 0 . The greater amounts of accumulated collagen were paralleled by higher levels of alkaline phosphatase activity (Fig. 2B) and higher levels of mineral accumulation in the cell layer (Fig. 2D).In the absence of ascorbic acid, mineralization was not observed, suggesting the requirement of a collagen matrix for mineral deposition. It appears that either the total amount of collagen in the cultures or a modification of the collagen matrix such as the association of noncollagenous proteins contributes to the down-regulation of proliferation, thereby accounting for the lower total cell number seen when higher amounts of collagen accumulate in the osteoblast cell layer. These data, taken together with the pattern of gene expression observed during the developmental sequence (Fig. l),suggest that the collag-

enous extracellular matrix is produced and subsequently modified in stages leading to its competency to incorporate mineral. Reciprocal relationship between growth and differentiation is apparent in individual cells Since the final result of this developmental sequence in cultured rat osteoblasts is a multilayered nodule of cells surrounded by a mineralized extracellular matrix having a bone tissue-like organization (Bellows et al., 1986; Bhargava et al., 1988; Escarot-Charrier et al., 19881, we addressed the relationship between events in this developmental sequence within the context of the individual cells in the population. The extent of proliferation and its down-regulation with time in culture as the extracellular matrix develops was first studied a t the single cell level by 3H-thymidine incorporation followed by in situ autoradiography. The results of these experiments (Fig. 3) showed that, from day 2 to day 5 following isolation, 95-100% of the calvarial-derived osteoblasts were proliferating, as seen by 3H-thymidine labeled nuclei. From day 7 to day 10 in culture, proliferation was decreased, and a t 12 and 14 days after plating discrete multilayered regions of nonproliferating cells were apparent, surrounded by a limited number of proliferating cells in the internodular monolayer region. The nonproliferating area shown in Figure 3 (day 14) is one of many such areas in this culture

GENE EXPRESSION DURING OSTEOBLAST DIFFERENTIATION

425

Fig. 3. Down-regulation of proliferation during the initial period of the development of the osteoblast phenotype in vitro reveals the initial cessation of proliferation in multilayered regions of the cultures. Primary cultures of osteoblasts growing on glass coverslips in MEM supplemented with 10% FCS and 50 pg/ml ascorbic acid (added at each feeding after day 7) were incubated with 'H-thymidine, fixed, and autoradiographed in situ as described in Materials and Methods. The days after initial plating are indicated under each photomicrograph. ~ 1 0 0 .

and demonstrates that, as proliferation is down-regulated, the extent of multilayering of cells increases. By day 21, the entire culture was multilayered, with little evidence of proliferative activity (Fig. 1). The temporal relationship between proliferation and the formation of multilayered regions within the cultures reflects the down-regulation of the cell growth genes and the concurrent accumulation of an extracellular matrix (Figs. 1, 3). As indicated in Figure 1, alkaline phosphatase mRNA was not induced until completion of the proliferative period. To identify at the single cell level the relationship between proliferation and initiation of alkaline phosphatase expression, we combined the techniques of 3H-thymidine labeling and autoradiography with alkaline phosphatase histochemistry. Figure 4 directly demonstrates that it is the nonproliferating multilayered regions (seen in Fig. 3, day 12) that initially express alkaline phosphatase. The results shown in Figure 4 also confirm on an individual cell basis that a temporal sequence of gene expression occurs and that at least some events (proliferation and alkaline phosphatase expression) appear to be sequential, mutually exclusive events in the same cell. That is, proliferation must be down-regulated prior to the expression of alkaline phosphatase. Figure

Fig. 4. Histochemical changes during the rat osteoblast developmental sequence. A The combination of alkaline phosphatase histochemistry and autoradiography following 3H-thymidine incorporation shows that it is in multilayered regions of the cultures that proliferation is first down-regulated, and i t is these regions that first become intensely alkaline phosphatase positive. Day 12. x 100.B: The entire culture then becomes alkaline phosphatase positive. Day 16. x 10. C: An ordered deposition of mineral in nodules as extracellular matrix develops within the multilayered regions of cells in the cultures. Day 35. x50.

4B shows that, following the initial expression of alkaline phosphatase in the heavily multilayered nodules where proliferation was first down-regulated, by day 21 all cells in the culture have ceased proliferation and exhibit alkaline phosphatase activity. This transition from a proliferating cell to one that can express an

426

OWEN ET AL.

with osteoblast differentiation was also found at the molecular level. A specific change in proteidDNA interactions in the proximal promoter of a cell growthregulated gene, H4 histone, has previously been shown both in this rat osteoblast differentiation model system (Owen et al., in press; Holthuis et al., in press) and following the induction of differentiation of HL60 promyelocytic leukemia cells by phorbol esters (Stein et al., 1989b). The down-regulation of proliferation is accompanied first by a loss of H4 histone mRNA followed by a down-regulation of transcription and concomitant loss of the specific interaction of a nuclear promoter binding factor, HiNF-D, with a region of the H4 histone gene proximal promoter that influences both specificity and level of transcription of the gene. It i s this loss of H4 histone gene transcription and of HiNF-D binding that accompanies differentiation of the cells. Other molecular markers for the transition from proliferation to the onset of differentiation are expression of a cell cycle-independent histone gene that encodes a highmolecular-weight poly-A RNA (Shalhoub et al., 1989) and specific modifications in the composition of nuclear matrix proteins (Dworetzky, et al., in press). A more direct demonstration that the down-regulation of proliferation results in the induction of expression of some genes that normally are expressed later in the osteoblast developmental sequence is derived from experiments in which DNA synthesis was inhibited in actively proliferating (subconfluent) osteoblasts by hydroxyurea treatment (5 mM). As demonstrated by a >90% decrease in DNA synthesis and H4 histone mRNA levels by 1 hr after hydroxyurea treatment (Fig. 61, cell proliferation is rapidly and selectively inhibited under these experimental conditions. Similarly, by 1 hr following inhibition of DNA synthesis, a fourfold increase in alkaline phosphatase mRNA levels was observed, indicating that the down-regulation of proliferation prematurely induced the expression of an early marker for the extracellular matrix maturation period of the osteoblast developmental sequence. With inhibition of DNA synthesis by hydroxyurea, the levels of osteopontin mRNA increased to levels approximating those present at the late stage (day 22) of the mineralization period of the cultured osteoblasts (Fig. 6 ) . These data suggest a direct functional coupling of the down-regulation of proliferation and the preferential expression of genes whose induction is normally associated with the appearance of the mature osteoblast phenotype in a mineralized matrix. The absence of osteocalcin induction is consistent with the concept that there is at least a second set of genes whose expression is coupled not directly to the down-regulation of proliferation but rather to development of the more differentiated osteoblast in a mineralized matrix. These experiments in which premature differentiation has been promoted by hydroxyurea inhibition of proliferation reveal a second transition point in the developmental sequence of osteoblast differentiation, since inhibition of proliferation supports progression of the developmental sequence only to the stage where mineralization is initiated. Mineral deposition may be required to signal expression of a subset of osteoblast phenotype genes, such as osteocalcin. Other genes expressed in relation to the developing extracellular matrix (which are not induced by hydroxyurea inhibition of prolifer+

Fig. 5. Transmission electron micrographs comparing cross sections of a 21 day fetal rat calvarium (columnA) with those of a mineralized nodule formed in a 35 day osteoblast culture (column B). The top micrographs in each column show that the overall patterns of mineralization of bone extracellular matrix in vivo and in vitro are indistinguishable, with mineralized extracellular matrix enveloping the cells. The second row of micrographs shows early stages of mineralization around the fringes of regions of heavy mineral deposition. Note the mineral deposition associated with collagen fibrils and the absence of intracellular calcification. The lower micrographs demonstrate identical orthogonal organization of the extracellular collagen matrix in both the intact calvarium and in the mineralized nodule formed in vitro. The lower micrograph in column A was made following sectioning of a demineralized calvarium. Bars in the lower micrographs represent 2.5 pm; others represent 5 pm.

early marker of the osteoblast phenotype (alkaline phosphatase) represents a restriction point where cessation of proliferation appears to be required for initiation of tissue-specific gene expression. The completed developmental sequence is shown in Figure 4C, which demonstrates intense von Kossa staining of the mineralized nodules reflecting hydroxyapatite deposition. The development of in uiuo-like bone-tissue organization in these cultures is further shown by comparison of the ultrastructure of the mineralized regions of the culture with sections through an intact 21 day fetal rat calvarium (Fig. 5). Note the similar type of ordered deposition of crystals within and between the orthogonally organized bundles of collagen fibrils. No evidence for cell necrosis or intracellular calcification is indicated in the cultures, particularly where mineralized matrix has enveloped the osteoblasts.

Functional coupling of the stages of osteoblast development Support for the concept that proliferation must be down-regulated prior to the onset of events associated

GENE EXPRESSION DURING OSTEOBLAST DIFFERENTlATION

z

-

"

C

.

developmental sequence, we modified the onset and extent of mineral deposition in the cultures. To accomplish this, the cultures were maintained in medium containing ascorbic acid after day 7 but lacking the organic phosphate source, p-glycerol phosphate (pGP). Figure 7A demonstrates that, under accelerated mineralizing conditions (in the presence of pGP), osteocalcin mRNA and synthesis increased steadily beginning at day 15 in parallel with calcium accumulation in the cell layer. In contrast, when cultures were maintained in the absence of PGP, calcium did not begin to accumulate in the cell layer until approximately day 25. Notably, osteocalcin was not detectable until this point, and its low level of synthesis reflects the slower rate of mineralization. It is important to note that the presence or absence of PGP had no effect on the proliferative period (as measured by rate of DNA synthesis) or on the onset of alkaline phosphatase expression. These experiments provide additional evidence to support the existence of the second transition point. The cells can progress through the maturation sequence to the onset of mineralization but cannot initiate expression of genes related to the mineralization stage unless mineral accumulation occurs. It appears that genes such as osteocalcin are not only temporally expressed but also "coupled" to deposition of hydroxyapatite.

@%-z-d 1 4 M

0

cn cn W a

P

X W

w

>

2

4

b

I

427

M

w U

D.

Fig. 6 . Coupling of alkaline phosphatase and osteopontin but not osteocalcin expression to the down-regulation of proliferation. Proliferation was inhibited in actively growing osteoblasts (day 5)by addition of 5 mM hydroxyurea (HU). Following HU addition, cells were harvested at 1 and 4 hr, examined for DNA synthesis (A),and cellular RNA prepared and analyzed for H4 histone (A), alkaline phosphatase (B), and osteopontin (C). The Northern blots of these transcripts and osteocalcin is also shown (D). Note that osteocalcin, which is not present in 5 day cultures (0,is also not induced following inhibition of proliferation by HU. For comparison, the relative expression of these genes in non-HU-treated control ( C ) and in mineralized (MI cultures (30 days after plating) is also shown.

ation) may be required to render the matrix competent for mineralization. To address further the relationship between mineralization and genes expressed during the osteoblast

T h e in vitro osteoblast developmental sequence is a reinitiation of events occurring in vivo Since our primary cultures of osteoblasts are derived from developing fetal bone, a key question is whether we are observing in culture the initiation of development of the osteoblast phenotype from progenitor cells or whether the osteoblast developmental sequence observed in vitro is a reinitiation and reoccurrence of the fetal developmental sequence in vivo. Therefore, we examined the representation of mRNA transcripts from osteoblast phenotype marker genes in cells of the third trypsin-collagenase digestion used to establish the cultures. High levels of osteocalcin and osteopontin mRNA were present at isolation and were then down-regulated very rapidly. Osteocalcin mRNA was not detectable, and osteopontin levels were lower in cells cultured for 2 days. Since active expression of these genes is not restored until formation and maturation of the extracellular matrix (Figs. 1, 31, it appears that the in vitro temporal sequence is a recapitulation of events that occur in vivo. Indeed, the biological relevance of the developmental sequence we observe in vitro is supported by a similar sequential expression of genes in fetal rat calvaria during gestation (Yoon et al., 1987). In light of the high levels of osteocalcin and osteopontin transcripts found in the third digest cells and the removal of the osteoprogenitor cells due to scraping of the perichondrium and discarding of the first and second enzymatic digests, we believe that the cells expressing the developmental sequence in vitro are the ones th a t have expressed the osteoblast phenotype in vivo. The similarity of the levels of bone phenotype-associated transcripts in vivo and in vitro and the appearance of mineral organization in the electron micrographs support the physiological significance of this model system a s a n appropri-

1

OWEN ET AL.

428 A.

B.

300

200

4 0100

40

4w3w

j:

200

40

4

20 30

._k

10

40

0

0

DAYS AFTER ISOLATION

Fig. 7. Conditions that delay mineralization of the osteoblast cultures do not effect the earlier events of proliferation and extracellular matrix maturation. Cultures maintained in BGJb medium supplemented with 50 pg/ml ascorbic acid and 10 mM p-glycerol phosphate (A) show an earlier accumulation of calcium in the extracellular matrix and earlier osteocalcin expression than cultures maintained in

ate one in which to study molecular mechanisms associated with bone formation. Consistent with a model in which the extracellular matrix surrounding the cells controls the differentiation process, enzymatic removal of the in vivo synthesized extracellular matrix when the primary calvarialderived osteoblast cultures are established supports the reinitiation of the developmental sequence of proliferation, extracellular matrix maturation, and mineralization. These cultures can be maintained for up to 120 days and possibly longer, with the cells retaining bone-like features. The reinitiation of proliferationand differentiation-related events also occurs in subcultivated cells. Trypsinization and replating of mature mineralized cultures of osteoblasts (after day 35) results in a rapid down-regulation of osteocalcin and osteopontin mRNAs. As is shown in Figure 8, cells passaged four times were capable of reinitiating expression of osteoblast phenotype markers and mineralizing under the appropriate culture conditions. However, although subcultivated cells remain competent to express the developmental sequence, we observed a variable lengthening of the time course required to develop a mineralized extracellular matrix. Whether cells in vivo are capable of reinitiating a similar developmental sequence under the appropriate circumstances remains to be established, but the ability of the calvarial cells, once freed of the extracellular matrix, to differentiate in culture certainly suggests that this is possible.

lo lo 400

10

20

30

DAYS AFTER ISOLATION

the same medium without p-glycerol phosphate (B).However, cultures maintained under either culture regimen had essentially identical periods of proliferation ("H-thymidine incorporation) and extracellular matrix development (alkaline phosphatase activity) (top panels in A and B). DNA synthesis (01, alkaline phosphatase activity (m), calcium (n), osteocalcin (m).

Fig. 8. Mineralization of osteoblast cultures following four passages of the cells. Osteoblasts isolated as described were allowed to progress through the in vitro developmental sequence by addition of BGJb medium supplemented with 50 pglml ascorbic acid and 10 mM pglycerol phosphate and then passaged by trypsinization after development of the mineralized extracellular matrix. This sequence was repeated four times. The cells shown developed mineralized nodules 66days after plating following the fourth passage and were stained by von Kossa. x 10.

Conclusions and model of osteoblast differentiation in vitro The results we have presented based on dekrminations of molecular, biochemical, and histochemical pa-

GENE EXPRESSION DURING OSTEOBLAST DIFFERENTIATION

CGUMITMENT PERIODS AND RESTRICTION POINTS

A,

429

, DAYS

1 1

I

I

Fig. 9. Model of the relationship between proliferation and differentiation during the rat osteoblast developmental sequence. This relationship is schematically illustrated within the context of modifications in expression of cell cycle- and cell growth-regulated genes as well as genes associated with the maturation, development, and mineralization of the osteoblast extracellular matrix. The three principle periods of the osteoblast developmental sequence are designated within broken vertical lines (proliferation, matrix development and maturation, mineralization). Commitment periods and restriction points, which are indicated in the lower portion of the figure, were experimentally established. A functional relationship between the

down-regulation of proliferation and the initiation of extracellular matrix maturation and development is based on stimulation of alkaline phosphatase and osteopontin gene expression when proliferation is inhibited, but the developmental sequence is induced only to the second transition point. Growth of the osteoblast under conditions that do not support mineralization confirms the day 20 restriction point since the developmental sequence proceeds through the proliferation and the extracellular matrix development/maturation periods, but not further. DNA, DNA synthesis; col I, collagen Type I; AP, alkaline phosphatase; OP, osteopontin; OC; osteocalcin; mineral, total accumulated calcium.

rameters are consistent with a reciprocal and functional relationship between proliferation and the sequential development of the osteoblast phenotype, which is schematically shown in Figure 9. The progressive and interdependent series of biochemical events that characterizes the osteoblast developmental sequence reflects the selective expression initially of cell growth and subsequently of a series of tissue-specific genes. Such modifications in gene expression define a developmental sequence with three distinct periods: proliferation, extracellular matrix maturation, and mineralization. Our experimental results indicate two principle restriction points in the osteoblast developmental sequence to which the cells can progress but cannot pass without further appropriate signals, the first when proliferation is down-regulated and gene expression associated with extracellular matrix maturation is initiated and the second when mineralization occurs. Our working hypothesis is that genes involved in the production and deposition of the extracellular matrix must be expressed during the proliferative period for the onset and progression of differentiation to occur. We postulate that proliferation is functionally related to the synthesis of an organized bone-specific extracellular matrix and that the maturation and organization of the extracellular matrix contributes to the shutdown of proliferation, which then renders the matrix competent for the mineralization that is essential for complete expression of the osteoblast phenotype. This

working model provides the basis for addressing whether particular stages of osteoblast differentiation exhibit selective responsiveness to actions of hormones and other physiological factors that influence osteoblast activity and other such questions related to the molecular mechanisms associated with bone formation.

ACKNOWLEDGMENTS We thank Marie Giorgio for photographic assistance. These studies were supported by grants from the National Institutes of Health (GM32010, GM32381, AR35166, AR39122, HD22400), the National Science Foundation (DCB88-96116), the March of Dimes Birth Defects Foundation (1-813), and the International Life Sciences Institute (Washington, DC). LITERATURE CITED Aronow, M.A., Gerstenfeld, L.C., Owen, T.A., Tassinari, M.S., Stein, G.S., and Lian, J.B. Factors that promote progressive development of the osteoblast phenotype in cultured fetal rat calvaria cells. J. Cell. Physiol., 143:213-221. Baserga, R., and Malamud, D. (1969) Autoradiography. Hoeber, New York. Bellows, C.G., Aubin, J.E., Heersche, J.N.M., and Antosa, M.E. (1986) Mineralized bone nodules formed in vitro from enzymatically released rat calvaria populations. Calcif. Tissue Int., 38:143-154. Bhargava U., Bar-Lev, M., Bellows, C.G., and Aubin, V. (1988) U1trastructural analysis of bone nodules formed in vitro by isolated fetal rat calvarial cells. Bone, 9:155-163. Clark, G. (1981) Miscellaneous stains. In Staining Procedures, 4th Ed. G. Clark, ed. Williams and Wilkins, Baltimore, p. 187.

430

OWEN ET AL.

Curran, T., MacConnell, W.P., van Straaten, F., and Verma, I.M. (1983) Structure of the FBJ Murine Osteosarcoma Virus genome: Molecular cloning of its associated helper virus and the cellular homolog of the v-fos gene from mouse and human cells. Mol. Cell. Biol., 37914-921. Derynck, R., Jarrett, J.A., Chen, E.Y., Eaton, D.H., Bell, J.R., Assoian. R.K.. Roberts. A.B.. SDorn. M.B.. and Goeddel. D.V. (1985) Human transforming growih factor+ complementary DNA sequence and expression in normal and transformed cells. Nature, 316:701-705. Dworetzky, S.I., Fey, E.G., Penman, S., Stein, J.L., Lian, J.B., and Stein, G.S. Progressive changes in the protein composition of the nuclear matrix during rat ostoeblast differentiation. Proc. Natl. Acad. Sci. USA, in press. Escarot-Charrier, B., Glorieux, F.H., van der Rest, M., and Pereira, G. (1983) Osteoblasts isolated from mouse calvaria initate matrix mineralization in culture. J . Cell Biol., 96:639-643. Escarot-Charrier, B., Shepard, N., Charette, G., Grynpas, M., and Glorieux, F.H. (1988) Mineralization in osteoblast cultures: a light and electron microscopic study. Bone, 9:147-154. Feinberg, A.P., and Vogelstein, B. (1983) A technique for radiolabeling DNA restriction endonuclease fragments to a high specific activity. Anal. Biochem., 132:6-13. Genovese, C., Rowe, D., and Kream, B. (1984) Construction of DNA sequences complementary to rat a1 and a 2 collagen mRNA and their use in studying the regulation of type I collagen synthesis by 1,25-dihydroxy D. Biochemistry, 23:6210-6216. Gerstenfeld, L.C., Chipman, S.D., Glowacki, J., and Lian, J.B. (1987) Expression of differentiated function by mineralizing cultures of chicken osteoblasts. Dev. Biol., 122:49-60. Glimcher, M.J. (1985) Role of collagen and phosphoproteins in the calcification of bone and other collagenous tissues. In Calcium in Biological Systems. R.P. Rubin, G.B. Weiss, and J.W. Putney, ed. Plenum Press, New York, pp. 607-616. Grimes, S., Weisz-Carrington, P., Daum, H. 111, Smith, J., Green, L., Wright, K., Stein, G., and Stein, J. (1987) A rat histone H4 gene closely associated with the testis-specific H l t gene. Exp. Cell Res., 173534445, Gundbera. C.M., Hauschka, P.V., Lian, J.B., and Gallop, P.M. (1984) Osteocalcin isolation, characterization, and detection: Methods Enzymol., 107:516-544. Holthuis, J., Owen, T.A., van Wijnen, A.J., Wright, K.L., RamseyEwing, A., Kennedy, M.B., Cosenza, S.C., Carter, R., Soprano, K.J., Lian, J.B., Stein, J.L., and Stein, G.S. Tumor cells exhibit deregulation of the cell cycle histone gene promoter factor HiNF-D. Science, in press. Kohno, K., Martin, G.R., and Yamada, K. (1984) Isolation and characterization of a cDNA clone for the amino-terminal portion of the pro-al(I1) chain of cartilage collagen. J . Biol. Chem., 259.1366813673. Lian, J., Stewart, C., Puchaz, E., Mackowiak, S., Shalhoub, V., Collart, D., Zambetti, G., and Stein, G. (1989) Structure of the r a t osteocalcin gene and regulation of vitamin D-dependent expression. Proc. Natl. Acad. Sci. USA. 86.1143-1147. '

Lowry, O.H., Roberts, N.R., Wu, M., Hixen, W.S., and Crawford, D. (1954) The quantitative histochemistry of brain 11. Enzyme measurements. J. Biol. Chem., 207:13-19. Maniatis, T., Fritsch, E., and Sambrook, J . (1982) Molecular Cloning, A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Noda, M., Yoon, K., Thiede, M., Buenga, R., Weiss, M., Henthorn, P., Harris, H., and Rodan, G. (1987) cDNA cloning of alkaline phosphatase from rat osteosarcoma (ROS 17/2.8) cells. J. Bone Mineral Res., 2:161-164. Nomura, S., Wills, A.J., Edwards, D.R., Heath, J.K., and Hogan, B.L.M. (1988) Developmental expression of 2ar (osteopontin) and SPARC (osteonectin) RNA as revealed by in situ hybridization. J. Cell Biol., 106:441-450. Oldberg, A,, Franze, A,, and Heinegard, D. (1986) Cloning and sequence analysis of rat bone sialoprotein (osteopontin)cDNA reveals a n Arg-Gly-Asp cell-binding sequence. Proc. Natl. Acad. Sci. USA, 83:8819-8823. Owen, T.A., Holthuis, J., Markose, E., van Wijnen, A.J., Lian, J.B., and Stein, G.S. Modifications of protein-DNA interactions in the proximal promoter a t a cell growth regulated histone gene during the onset and progression of osteoblast differentiation. Proc. Natl. Acad. Sci. USA, in press. Plumb, M., Stein, J., and Stein, G. (1983) Coordinate regulation of multiple histone mRNAs during the cell cycle in HeLa cells. Nucleic Acids Res., 11:2391-2410. Sehwarzbauer, J.E., Tamkin, J.W., Lemischka, I.R. and Hynes, R.O. (1983) Three different fibronectin mRNAs arise by alternative splicing within the coding region. Cell, 35.421-431. Shalhouh, V., Gerstenfeld, L.C., Collart, D., Lian, J.B., and Stein, G.S. (1989) Downregulation of cell growth and cell cycle regulated genes during chick osteoblast differentiation with the reciprocal expression of histone gene variants. Biochemistry, 28.5318-5322. Stein, G.S., Lian, J.B., Gerstenfeld, L., Shalhoub, V., Aronow, M., Owen, T., and Markose, E. (1989a) The onset and progression of osteoblast differentiation is functionally related to cellular proliferation. Connective Tissue Res., 20.3-13. Stein, G., Lian, J., Stein, J., Briggs, R., Shalhoub, V., Wright, K., Pauli, U., and van Wijnen, A (1989b) Altered binding of human histone gene transcription factors during the shutdown of proliferation and onset of differentiation in HL60 cells. Proc. Natl. Acad. Sci. USA, 86:1865-1869. Thomas, P.S. (1983) Hybridization of denatured RNA transferred or dotted to nitrocellulose paper. Methods Enzymol., 100:255-266. Toscani, A., Soprano, D.R., and Soprano, K.J. (1988) Molecular analysis of sodium butyrate induced growth arrest. Oncogene Res., 3: 223-238. Vysatek, R. (1982) A sensitive fluorometric assay for the determination of DNA. Anal. Biochem., 48:243-248. Yoon, K., Buenaga, R., and Rodan, G.A. (1987) Tissue specificity and developmental expression of osteopontin. Biochem. Biophys. Res. Commun., 148:1129-1136.