MOLECULAR AND CELLULAR ANALYSIS OF THE OSTEOBLAST LINEAGE

Fina Liu

A thesis submitted in conformity with the requirements

for the degree of Doctor of Philosophy

Graduate Department of Dentisv

University of Toronto

43 Copyright by Fina Liu 1997

National Library ,,nada

Bibliothèque nationale du Canada

Acquisitions and Acquisitions et Bibliographic Services seMces bibliographiques

395 Wellington Street 395. rue Wellington OttawaON K1AON4 OttawaON K 1 A W Canada Canada

The author has granted a non- L'auteur a accorde une licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sel1 reproduire, prêter, distribuer ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or electronic formats. la forme de microfiche/film, de

reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fkom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.

MOLECULAR AND CELLULAR ANALYSIS OF THE OSTEOBLAST LINEAGE

b~

Fina Liu

Doctor of Philosophy, 1997

Graduate Department of Dentistry, University of Toronto

ABSTRACT

A challenge in analyzing transitional stages in the osteoblast lineage is to develop means by

which to rank cetls as more or less mature and to order them with respect to stage of differentiation.

Primary cultures of fetal rat calvaria cells were used to follow single osteoprogenitors and their progeny

to address in more detail sequential patterns of expression of osteoblast-related markets during

differentiation.

In initial experiments. I used immunocytochemistry and poly(A)-PCR to investigate different

stages of osteogenic differentiation in individual cells and colonies identified rnorphologically as

fibroblastic or osteoblastic. The results showed that colonies with fibroblastic morphology are

distinguishable from immature osteogenic colonies and the latter are distinguishable from mature

osteogenic colonies. As well, marked intercellular heterogeneity in expression of mRNA and protein in

individual rnorphologically indistinguishable cells from the sarne colonies was also found. These results

validated the technical approaches being used. They also confirmed some aspects of the osteoblast

differentiation sequence as seen in mass populations. but extended it to a b e l of single cell expression

and single ceIl heterogeneity not previously accessible.

1 next analyzed further expression of osteoblast-associated markers arnongst individual mature

osteoblasts. The results revealed extensive heterogeneity in the repertoire of genes expressed in clonally-

derived, mature osteoblasts of equivalent in vitro lifetime, suggesting marked plasticity in phenotype of

mature osteoblasts. These data also suggested that the most unarnbiguous marker of the mature

osteoblast remains its ability to contribute with its siblings to the production of bone rather than in an

absol ute level of production of particular macromolecules.

Finally, I validated a replica plating approach for studying primitive progenitors prior to their

acquiring morphological characteristics of preosteoblasts/osteoblasts and used it with poly(A)-PCR to

determine molecular fingerprints of immature osteoprogenitors compared to those of preosteobkists and

mature osteoblasts. The molecular profiles I established indicate that a tempo raI sequence of expression

of multiple markers can be followed in these single colonies as they undergo differentiation, and that

both discrete stages and a continuum of changing marker expression levels occur with much variation in

expression of any given marker. 1 identified a novel transition point in osteoprogenitor development not

previously seen in any mode1 system or bone in vivo. Thus. the data have given new insights into the

osteoblast differentiation sequence and have established new landmarks in early osteoblast development.

Dedicated to my parents, Patrick and Aileen

for their unfaihg love amd support

ACKNOWLEDGEMENTS

My gratitude is exrended to the folio wing wonderful people who have enriched q v life while I pursued

my graduare education:

To my supervisor, Jane Aubin, for having faith in me. She has been an ideal supervisor, mentor, and

inspiration to me. 1 wish to thank Jane for her optirnism, patience, kindness. encouragement, support.

criticism, guidance, humour, friendship. and for building my self-confidence.

To my supervisory cornmittee, Norman Iscove and Hazel Cheng, for insightful suggestions, constructive

criticisrn. and enhancing my scientific awareness.

To the University of Toronto (Open Fellowship and Life Sciences Cornmittee Graduate Student Degree

Completion Award) and Wilson G. Harron Trust Fund for financial support.

To the several wonderful people whom 1 have had the good fortune of meeting and befriending at the

Faculty of Dentistry, Department of Anatomy and Cell Biology, and Department of Zoology for making

my leaming experience enjoyable, especially: Alexa, Shiva, Usha, Harry, Lori, Ashwani, Sui-Wah,Wing

Fun, Richard, Elisa, Eva, Carl, Kursad, Agi, Tony, Linda, Maurice, Sash, Donna, Jasmine, and Alison.

Also, 1 wish to thank Gerard Brady for helpful discussions and suggestions in the poly(A)-PCR technique

at the beginning of my studies.

To my family and friends outside the scientific community for their moral support and laughter:

Jeanna and Kevin, Diana, Victor, Agatha, Judy, Christina, Julianna, Jackie, Pelino, and Todd.

To Luc Malaval for his invaluable collaborations, critical discussions, humour, friendship. and love.

This is only the beginning.

Attribution of data in chapters based on muItiauthoredpapers:

The experimental chapters (Chapters 2-4) are based on articles published or to be published; 1 am the

first author on al1 of these papers and contributed the majority of the data presented in thern. However. 1

would like to acknowledge that Luc Malaval contributed to the immunocytochemical results in Chapters

2 and 3 with the figures. antibodies, and technical expertise. Ashwani Gupta helped to prepare several

cDNA constructs and provided technical advice during my initial attempts to use the poly(A)-PCR

procedure. In addition. 1 acknowledge Dominique Modrowski, who kindty provided the antibody against

osteocalcin, and Claude Reynal and Chantal Chenu who kindly provided the antibody against bone

sialoprotein.

When a thing was new, people said "It is not true"

Later when its truth became obvious, people said "Anyway it is not important"

And when its importance could not be denied, people said "Anyway it is not new"

- Anonymous

TABLE OF CONTENTS

Page

ABSTRACT

ACKNOWLEDGEMENTS

TABLE OF CONTENTS

LIST OF FIGURES

LIST OF TABLES

ABBREVIATIONS

C HAPTER 1 Introduction

GeneraI Introduction

0s:eobiast Ontogeny

Osteoblast Phenotypic Markers

Experimentai Approaches

Thesis Objectives

CHAPTER 2 Sirnultaneuus Detection of Multiple Bone-Refated mRNAs and Protein

Expression during Osteoblasf Di//erentiation: Polyrnerase Chain Reaction

and Immunocy tochemical Sfudies at the Single Cell Level

Introduction

Material and Methods

Results

Discussion

CHAPTE R 3 Tite Mature Osteo blast Pitenotype is Characterized by Exîensive Piasf ici@

Introduction

Materiai and Methods

Results

Discussion

*.

I I

v

vii

i x

.Y

xi

vii

CHAPTER 4 Molecuiitr Finge'printing of Primitive Osteoprogenitor Celis

undergoing DiJfeentiation in vitro

Introduction

Material and Methods

Results

Discussion

CHAPTER 5 General Summaiy and Conclusions

Page

67

REFERENCES

viii

LIST OF FIGURES

Page

Figure 2.1

Figure 2.2

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 3.1

Figure 3.2

Figure 3.3

Figure 3.4

Figure 4.1

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

Photornicrographs of early and mature osteoblast colonies

Southems of whole colonies and single cells arnplified by poly(A)-PCR

Immunolabeling for COLL-1

Membrane-associated staining with anti-alkaline phosphatase (RBM 2 1 i -13)

Immunolabeling for OPN

Immunolabeling for OCN

Photomicrograph of a typical mature osteoblastic colony

Double immunolabelings of mature osteoblastic colonies

Phosphorimage cDNA expression profiles of single cells from mature osteoblast

colonies

Normalized cDNA expression profiles of cells from Fig. 3.3

Schematic of the methodology for replica plating technique

Plating efficiencies/colony forming efficiencies

Phosphorimage cDNA expression profiles of osteoprogenitor/preosteoblast, and

early and mature osteoblast colonies

Rank order profiles of normalized cDNA expression of samples from Fig. 4.3

The means and standard deviations of osteoprogenitor and osteoblast colonies

LIST OF TABLES

Page

Table 1.1 Bone matrix proteins

Table 1.2 Osteoblast regulatory factors

Table 1.3 In vivo expression of rnarkers by osteogenic cells

ALP

BSP

COLL-1

FGF-Rl

OCN

OPN

PDGF-Ra

PTH

PTHPTHrP-R

PTHrP

RC

TNAP

alkal ine phosphatase

bone sialoprotein

collagen type I

fibroblast growth factor receptor type 1

osteocalcin

osteopontin

platelet derived growth factor receptor a subunit

parathyroid honnone

parathyroid hormone/parathyroid homone-related peptide receptor

parathyroid hormone-related peptide

rat calvaria

tissue non-specific alkaline phosphatase

CHAPTER 1

Introduction

General Introduction

Bone is a remarkably complex and dynarnic tissue which remodels and repairs itself throughout

life. The basic functions performed by bone are to provide protection of vital organs, to provide an

environment for marrow (both blood forming and fat storage), to provide attachment for muscles, and to

act as a mineral reservoir for calcium homeostasis in the body. The two specialized cells of bone are the

osteoblasts and the osteoclasts. Osteoblasts are responsible for bone matrix synthesis and its subsequent

rnineralization, while osteoclasts function in resorption of mineralized tissue; both ce11 types are involved

in bone tissue maintenance. Bone ce11 functions are under the regulation of various local and systemic

factors.

This thesis focuses on bone-forming osteoblasts, with emphasis on studies addressing osteoblast

ceIl lineage. Detailed knowledge of transitional stages in osteoblast differentiation is still limited. While

the relatively late osteoblastic differentiation stages are increasingly well-characterized, earlier stages are

relatively undefined. The reasons for this include a paucity of unambiguous methods and markers to

establish the differentiation stages of individual cells and the fact that progenitors are relatively rare

within isolated osteoblastic populations, such that large members of differentiating progenitors have not

been accessible. This thesis explores the concept of osteoblastic heterogeneity vis-à-vis differentiation

stage and establishes methods for characterizing osteoblast precursor cells. The introductory chapter

presents an overview of osteoblast ontogeny, the markers which are commonly used and associated with

various osteoblastic differentiation stages. and the experimental approaches that have been used to

characterize these stages. Chapter 2 presents the establishment of a molecular approach, defines some

osteoblast transitional stages, and describes the heterogeneity that is observed among developing

osteoblasts at the single cell level. Chapter 3 focuses on the heterogeneity amongst cells ciassed as

mature osteoblasts. Chapter 4 describes the identification and initial molecular characterization of

osteoprogenitor and preosteoblast cells. Chapter 5 concludes with a general discussion of the results as a

launching point for future research.

Osteoblast Ontogeny

Mesenchymal stem cell origin

A variety of data have suggested that there exist stroma1 or mesenchyrnal stem cells which are

multipotential and have the capacity to generate several committed and restricted ce11 lineages including

the osteogenic line (Owen, 1985). Friedenstein and colleagues provided evidence for the existence of

multipotential mesenchymal stem cells in bone marrow with high proliferative ability and capacity to

form bone, cartilage, fat, and fibrous tissue in vivo (Friedenstein, 1980; Friedenstein et al., 1987). Further

evidence for multipotentiality of mesenchymal progenitors was provided by in vitro studies on several

clona1 ce11 populations isolated from different starting tissues by different labs (Taylor and Jones, 1979:

Grigoriadis et al., 1988: Ket Ierman et al., 1990; Yamaguchi and Kahn, 199 1 ; Poliard et al., 1995). These

ceIl systems are able to differentiate concurrently into one or several phenotypes, depending on

experimental conditions. For example, Grigoriadis et al. (1988) demonstrated that a clonal ceIl line (RCJ

3.1) derived from fetal rat calvaria (RC) has myogenic, adipocytic, chondrogenic, and osteogenic

potential in vitro. Individual progenitors were detectable in RCJ 3.1 with mono-, bi-, tri-? and

multilineage potentiality (Grigoriadis et al., 1990: reviewed in Aubin et al., 1990). 1 wiII not discuss

these multiple mesenchymal lineages and their relationships any further because it is beyond the scope of

this thesis. Rather. my focus is on the osteogenic lineage and whether and how differentiation stages can

be detected.

Cells of the osreobiasr iineage

On the basis of morphological and histochemical criteria, and more recently biochemical criteria,

four osteoblastic ce11 stages have been identified and chantctenzed in vivo. They are the preosteoblast,

osteoblast, osteocyte, and bone lining ce11 (reviewed in Nijweide et al.. 1986; Marks and Popoff, 1988;

Martin et al., 1988; Aubin et al., 1993). In addition, some authors have ascribed the term osteoprogenitor

to morphologically spindle-shaped cells residing in close proximity to the preosteoblast layer, but further

from bone formation surfaces (Scott, 1967; Pritchard, I972a, 1 WOb; see also below).

The preosteoblast is the immediate precursor of the osteoblast. Preosteoblasts are defined by

their position near bone formation surfaces; Le., onIy one or a couple ce11 layers distant from the active

osteoblasts that are lining bone formation surfaces. Preosteoblasts are thought to possess a limited

capacity to divide (Owen, 1963, 1967; Kember, 197 1 ). Although preosteoblasts resemble osteoblasts

histologically and ultrastructurally, and stain for alkaline phosphatase (ALP) activity (Doty and

Schofield, 1976), they have not yet acquired many of the other differentiated characteristics of mature

osteoblasts.

Osteoblasts are plump, cuboidal, polar basophilie cells lining the bone matrix at sites of active

matrix production (Holtrop, 1975) and connected to each other by gap junctions (Doty, 198 1). Reflecting

their active matrix synthesis function, they have prominent Golgi apparatus and abundant endoplasmic

reticulum (Marks and Popoff, 1988). They stain strongly for ALP activity (Doty and Schofield, 1976)

and synthesize a number of phenotype-specific macromolecules (to be discussed later).

A srnall proportion of osteoblasts becorne incorporated within the newly formed organic matrix

(osteoid) which eventuaIIy becomes calcified bone. These mature embedded cells, considered the most

mature differentiation stage of the osteoblastic lineage. are osteocytes (Jande and Belanger, 1973;

Holtrop, 1975). Osteocytes are smailer than osteoblasts and have lost many of their cytoplasmic

organelles Wijweide et al., 1986); they also have decreased ALP activity compared to osteoblasts

(Holtrop. 1975). Osteocytes situated deep in bone matrix maintain contact with newly incorporated

osteocytes in osteoid. and with osteoblasts and bone lining cells on the bone surfaces, through an

extensive network of ceIl processes (canaliculi) coupled by gap junctions (Weinger and Holtrop, 1974;

Stanka, 1975; Doty, 198 1 ; Menton et al., 1984). Compared to the osteoblast, little is known about the

functions and biochemical activities of osteocytes because of their relative inaccessibility in the

rnineralized bone matrix. However, recently a monoclonal antibody specific for chick osteocytes has

been used to isolate a relatively pure osteocyte population (van der Plas et al., 1994). These cells were

found to be morphologically stellate-shaped, relatively quiescent or post-proliferative, exhibited ALP

activity, bound parathyroid hormone (PTH), and were stimulated by PTH with an increased CAMP

production in vitro (van der Plas et al., 1994). It is also known that osteocytes maintain expression of at

least some osteoblast markers (see below). In addition, osteocytes in vivo are thought to be ideally

situated to respond to changes in physical forces upon bone and to transduce messages to the osteoblastic

cells on the bone surface, directing them to initiate resorption or formation responses (Aarden et al.,

1994). The molecular/biochemicaI messages transducing these mechanical activities are not known

although prostaglandin El (Klein-Nulend et al., 1995) and IGF-I (Lean et al., 1995) have al1 been

postulated to play roles.

In the adult skeleton, the majority of bone surfaces that are undergoing neither formation nor

resorption (i.e., not being remodeled) are lined by bone lining cells, thought to represent the i

fom of the osteoblast in terms of matrix production (Luk et al.? 1974; Miller and Jee, 1987).

cells are flat, thin, elongated cells (Menton et al.. 1984) that have few organelles (Cameron, 1

nact ive

Bone lining

968) and

are ofien joined to each other and to nearby osteocytes by gap junctions (Miller et al., 1980). Not much is

known directly of their function because most studies have used fetal bone tissues, in which bone lining

cells are not abundant (hlijweide et al., 1986). However, it has been proposed that they seme as a

selective barrier between bone and other extracellular fluid compartments. as nutritional suppon cells for

osteocytes via gap junctions, andor as regdators of crystal growth in bone (Miller et al., 1980; Miller

and Jee, 1987).

The preosteoblast, osteoblast. osteocyte, and bone lining ceIl are the more differentiated and

easily recognizable ceIl types in the osteoblast line. However, their precursors, the osteoprogenitor cell

and osteogenic stem cell, are not histologically identifiable. As mentioned above, it is believed that the

farther away from the bone surface an osteogenic cell is, the less differentiated it will be (Scott, 1967).

Thus, osteoprogenitors are thought to be the proliferating cells behind the preosteoblast/osteoblast layer

(Pritchard, 1972a, 1972b). These undifferentiated mesenchymal cells are rnorphologically fibroblast-like

cells with an elongated nucleus (Scott, 1967). Whether the committed osteogenic progenitor goes

through discrete or gradua1 differentiation stages to give rise to the formation of osteoblasts is not known

because few tools have been available for defining earlier differentiation stages (Owen, 1985; Aubin et

al., 1993).

Osteoblast Phenotypic Markers

While in vivo studies using morphological and histochemical criteria have provided much

information regarding the temporal-spatial relationships of the different types of more mature

osteoblastic cells, earlier stages in the osteogenic differentiation pathway and recognition of transitional

stages are poorly defined. Currently, attempts to provide details of osteoblast differentiation are being

actively pursued at the protein and molecular levels. The osteoblast phenotype is defined in part by what

osteoblasts synthesize; they synthesize and deposit colIagen type 1 (COLL-1), as well as membrane-

associated ALP, and a wide variety of noncollagenous proteins which together with COLL-I comprise

the extracellular matrix of bone. Although most of these macromolecuies are not unique to bone,

together they serve as useful markers with which to characterize the osteoblast phenotype. Before

summarizing how these markers may help to define stages of osteoblast maturity, 1 will first briefly

review the nature of these markers.

Aikaline Phosphatase

ALP is a ce11 surface giycoprotein that hydrolyzes monophosphate esters at alkaline pH

(reviewed in Wuthier and Register, 1985). Three major ALP isoforms have been distinguished: placental,

intestinal, and liver/bone/kidney (also called tissue non-specific ALP; TNAP). The latter isoforrn is

found in almost al1 tissues albeit abundantly expressed at a much higher level in osteoblasts (Goldstein et

al., 1982). Although the precise physiological role of ALP is unknown, ALP has been hypothesized to be

involved in the mineralization process by hydrolyzing organic phosphate to inorganic phosphate, which

increases in concentration at sites of mineralization and results in precipitation of calcium phosphate

mineral (hydroxyapatite) (Wuthier and Register, 1985). The clearest support for its role in bone

mineralization cornes from studies of patients with the disease hypophosphatasia. This inherited disorder

results in poor bone mineralization and is characterized by low levels of ALP (reviewed in Whyte, 1989).

Furthemore, a mutation in the liver/bone/kidney ALP gene that abolishes enzymatic activity results in

extreme skeletal hypomineralization (Weiss et al., 1988). Consequently, ALP activity is positively

corretated with bone formation. and a high level of ALP activity is associated with osteoblastic

differentiation. Further insight into the roie of TNAP should corne from mice nul1 for TNAP. An initial

report suggested that skeletal formation and mineralization occurred in such rnice, but fùrther detailed

analysis have not yet been reported (MacGregor et al., 1995).

Bone matrix proteins

COLL-1, with a chain composition [a I(1)Jz a2(1), is widely distributed in connective tissues,

such as skin, tendon. cornea, and bone; it is synthesized by fibroblasts, smooth muscle cells, epithelium,

and osteoblasts (Kleinrnan et al., 198 1). COLL-I is the major extracellular matrix protein of bone,

constituting approximately 90% of the total organic matrix in mature bone (Miller, 1976). It provides the

structural frarnework for extracelIular matrix formation and minera1 deposition (Katz and Li, 1973).

Since COLL-I is a primary product of osteoblasts during bone matrix formation (for reviews, see Rodan

and Noda, 199 1; Aubin et al., 1993; Stein and Lian, 1995), it is a characteristic marker for the osteoblast

phenotype.

Osteopontin (OPN) is a secreted phosphoglycoprotein originally isolated from the extracellular

matrix of bone (Franzen and Heinegard, 1985) and synthesized by osteoblastic cells (Prince et al.. 1987;

Mark et al., 1987a. 19876. 1988; Chen et al., 1991 b). OPN, however, is not exclusive to bone since it is

also expressed by many if not al1 nonskeletal tissues or cells studied thus far (see Young et al., 1993:

Rodan, 1995) and highly expressed in tnnsformed cells (Craig et al., 1989; Senger et al., 1989). The

precise function of OPN in bone is not known. OPN contains a conserved Arg-Gly-Asp (RGD) sequence

(Oldberg et al., 1 986) which mediates ce11 binding via an integrin, the vitronectin receptor (Miyauchi et

al.. 199 I), and promotes the adhesion of a variety of ceIl types including osteoblastic cells (Oldberg et

al., 1986; Sornerman et al., 1989). It has been shown to be induced by tumor promoters and growth

factors in vitro and to be associated with stimulation of proliferation (Smith and Denhardt, 1987, 1989;

Craig et al., 1989; Nose et al., 1990). Some studies demonstrate that OPN has high affinity for

hydroxyapatite (Oldberg et al., 1986; Prince et al., 1987) and others suggest that OPN inhibits calcium

phosphate formation (Boskey, 1995; GoIdberg and Hunter, 1995). Thus, OPN has been suggested to

mediate the attachment of osteoblasts and osteoclasts to the extracellular matrix and their participation in

the mineralization process (for review, see Butler, 1989; Sodek et al., 1992; Young et al., 1993).

OPN shares some structural features with another bone matrix protein, bone sialoprotein (BSP),

with which it was originally isolated from the bone matrix (Franzen and Heinegard, 1985). Like OPN,

BSP is an acidic glycoprotein that contains an RGD cell attachment motif recognized by the vitronectin

receptor and several stretches of acidic amino acids that may be involved in binding to hydroxyapatite

(Oldberg et al., 1988a, 1988b; Fisher et al., 1990). In contrast to OPN, BSP has a very restricted tissue

distribution and is almost exclusively produced by skeletal-associated cells, including hypertrophie

chondrocytes, osteoblasts, and osteocytes (Bianco et al., 199 1, 1993% 1993b; Chen et al. 199 1 a); it has

been also reported to be produced by cells associated with other rnineralizing tissues, e.g., odontoblasts

forming defitine (Chen et al., 1992, 1993). Human osteoclasts also produce BSP (Bianco et al., 199 1).

The only other cells reported to synthesize BSP are the trophoblastic cells of the placenta (Bianco et al.,

199 1). Relatively little is known about the regulation or hnction of BSP. The occurrence of the RGD

motif in its sequence suggests that it may mediate bone cell adhesion to extracellular matrix, and

consistent with this, BSP promotes ce11 attachment and spreading in vitro (Oldberg et al., l988a;

Sornerman et al., 1988). BSP has also been postulated to play a role in mineralization, since it is capable

of initiating calcification in vitro (Hunter and Goldberg, 1993). These kinds of studies led to the

suggestion that BSP functions in two physiologically relevant processes in bone, specifically, ceIl

attachment through cell-matrix interactions and promotion of mineralization (for reviews, see Sodek et

al., 1992; Young et al., 1993). Further elucidation of the function of BSP in bone will be aided by the

availability of knockout mice for BSP prepared recently. Interestingly, preliminary analyses suggest that

skeletal development and mineralization occur relatively normally in these mice, although some

modeling and remodeling defects are being detected as mice mature (Aubin et al., 1996b).

Among the bone matrix proteins that have been isolated and characterized, only BSP and

osteocalcin (OCN) are highly specific to bone tissue. OCN, or bone Gla protein (BGP), is a major

noncollagenous protein of bone (Hauschka et al., 1975; Price et al., 1976) highly expressed in mature

osteoblasts (to be discussed later) with lower levels apparently expressed in megakaryocytes and

platelets (Thiede et al., 1994). OCN is synthesized by osteoblasts and secreted into the bone matrix and

the circulation, where its levels reflect bone formation activity (Price, 1983). The protein is distinguished

by the presence of y-carboxyglutamic acid (GIa) in its primary sequence (Hauschka et al., 1975). Gla is

formed by carboxylation of glutamic acid, a reaction which is dependent on vitamin K (Stenflo and

Suttie, 1977). These Gla residues bind calcium ions and are responsible for the high affînity of OCN for

hydroxyapatite (Price et al., 1976; Poser and Price, 1979; Hauschka and Carr, 1982). OCN is currently

the most extensively studied bone protein in tems of structure, expression, and hormonal and gene

regulation, and it appears to be a highly specific osteoblastic marker produced during bone formation (for

reviews, see Hauschka et al., 1989; Price, 1989; Cole and Haniey, 1991: Lian et al., 1992; Pike et al.,

1993). Early studies suggested that OCN might serve as an inhibitor of hydroxyapatite crystal growth

(Price et al., 1976, 1982; Hauschka and Reid, 1978). When OCN levels were reduced in rats by treating

them with warfarin, a vitamin K antagonist, hypermineralization resulted, but bone formation was not

impaired (Price et al., 1982). More recently, similar results of increased bone mass due to increased bone

formation and decreased bone loss were aiso evident in OCN knockout mice, confirming a role for OCN

in regulating bone formation (Ducy et al., 1996). Other functions that have been attributed to OCN are

that it may act to define future resorptive sites on the bone surface, perhaps by acting as a matrix signai

in recruitment and/or activation of osteocIasts (Malone et al., 1982; Mundy and Poser, 1983; Lian et al.,

1984. 1986; Glowacki and Lian, 1987).

Osteoblasts also synthesize and secrete a variety of other proteins into the extracellular matrix of

bone (Table 1 . 1 ) . Most of these macromolecules are found in other tissues, but when taken together, their

expression by osteoblasts and deposition into the matrix reflect the biosynthetic repertoire of the

osteoblast phenotype, aibeit in most cases they have not yet been well established as markers indicative

of particular stages of osteogenic differentiation.

Table 1.1 Bone rnatrix proteins'

Collagens Type 1 collagen Type III collagen Type V collagen

Glycoproteins Alkaline phosphatase Osteonectin (SPARC/BM-40) Bone acidic glycoprotein (BAG-75)

RGD cell attachment proteins Bone sialoprotein (BSPhone sialoprotein II) Osteopontin (2ar, Eta- l,44-kDa bone phosphoprotein, pp69, bone sialoprotein 1, secreted

phosphoprotein 1, and uropontin) Fibronectin Vitronectin Tenascin Thrornbospondin

Gla-containing proteins Osteocalcin (bone Gla protein/BGP) Matrix Gia protein (MGP) Protein S

Proteoglycans Biglycan (CS-PG 1) Decorin (CS-PG II ) CS-PG III Large CSPG

62-kDa protein 85-kDa protein Proteases Protease inhibitors

t Surnrnarized from reviews: Heinegard and Oldberg, 1989; Sodek et al., 199 1 ; Aubin et al., 1993; Delmas and Malaval, 1993; Gehron Robey et al., 1993.

Regufatory factors

There is extensive evidence for both local and systemic regulation of bone formation.

Experimental studies, particularly with ce11 culture systems. show that bone cetls can respond to a wide

variety of systemic and locally active hormones and cytokines (Table 1.2). These agents have been

shown to affect cet 1 proliferation and differentiation, matrix synthesis, as well as other aspects of

osteoblast function (for reviews, see Heersche and Aubin, 1990; Canalis et al., 199 1; MacDonald and

Gowen, 1992; Zheng et al., 1992). One of the rnost interesting aspects of regulation currently is that at

least some actions of growth and differentiation factors appear dependent on the state of differentiation

of the responsive osteogenic cells (Canalis et al., 199 1 ; Aubin et al., 1992, 1993). The molecular

mechanisms underlying these different effects are under intense investigation. Localization of cytokine

recepton and ligands within subgroups of osteogenic cells at different developmental stages may help in

this regard.

In addition to being able to exhibit a positive reaction for ALP and to synthesize collagenous and

noncollagenous proteins of the extracellular matrix. the osteoblast phenotype includes the capacity to

respond to hormones such as PTH. Many osteobIast populations, derived from either normal tissues or

osteosarcornas. have been characterized as osteoblastic based on their having a PTH response (i.e., a

PTH-stimulatable adenylate cyclase). In general, PTH affects osteoblast function by decreasing the

expression of genes involved in bone matrix formation (reviewed in Partridge et al., 1994). PTH

treatment of primary fetal RC cells in vitro inhibits, but reversi bly, osteoprogen i tor differentiation

(Bellows et al., 1990). Osteoblastic cells possess receptors for PTH, but parathyroid hormone-related

peptide (PTHrP) has been Found to compete with PTH for binding to the same PTWPTHrP receptor

(Juppner et al., 199 1 ). PTHrP shares amino-terminal amino acid homology with PTH in 8 of the first 13

arnino acids, but is much larger than PTH, and has been shown to have similar biological activities to

those described for PTH (Nissenson et al., 1988; Rodan et al., 1988; Heersche and Aubin, 1990; Moseley

and Gillespie, 1992). Unlike PTH, PTHrP is not a product of the parathyroid gland, but is rather

expressed in multiple tissues (Martin et al., 199 1; Moseley and Gillespie, 1992), including osteoblasts in

bone tissue, and rnay locally regulate a variety of cellular functions as an autocrine/paracrine factor.

Table 1.2 Osteoblast regdatory factors'

Systemic factors Parathyroid hormone (PTH) Vitarnin D3 Glucocortico ids Retinoids Estrogens Thyroid hormones Insulin Growth Hormone

Local Factors Parathyroid hormone-related peptide (PTHrP) Fibroblast growth factors (FGFs) Platelet-derived growth factors (PDGFs) Epidermal growth factor/Transforming growth factor alpha (EGFITGFa) Insulin-Iike growth factors (IGFs) Transfonning growth factor betas (TGFPs) Bone rnorphogenetic proteins (BMPs) Interferon-gamma ( F N - y j Tumour necrosis factors (TNFs) Interleukins Leukemia inhibitory factor (LIF) Prostagland ins (PGs)

t Summarized from reviews: Canalis, 1983; Martin et al., 1988; Heersche and Aubin, 1990; MacDonald and Gowen; 1992; Zheng et al., 1992.

While mitogenesis is ofien the focus of interest in growth factors, the potential action by which

specific differentiated functions of nonproliferating ce11 populations may be regulated is of equal

importance. Growth factors and other cytokines modulate such osteoblastic functions as collagen and

noncollagen protein synthesis, and ALP activity, as well as proliferation and differentiation. Various

systemic and local factors that have been s h o w to regulate the osteoblast population are listed in Table

1.2, but it has been dificult to determine the responsive target cells in the osteoblast differentiation

pathway. The heterogeneity (stimulatory, inhibitory, no effect) of the responses may be a resutt of

different mode1 systems studied which comprise mixtures of osteoblastic cells at different stages of

differentiation, and may be dependent on timing, concentration and whether other agents are present

(Heersche and Aubin, 1990; Aubin et al., 1993). An approach to sort out sorne of these discrepancies is

to define more precisely the subpopulation makeup of the cells or tissues being used and to sort the

heterogeneity observed into physiologically rneaningful classes by determining the relationship of

cytokine receptor expression to the state of osteoblastic differentiation. These investigations are only

beginning to be explored in detai l by immunolocalization. in situ h ybridization and ligand binding

studies (see Table 1.3).

Other potential osteobiastic markers

Osteo blast ic cells synthesize and express col lagenous and noncol lagenous proteins, and cytokine

receptors and ligands. However, the molecular mechanisms responsible for the coordinate expression and

regulation of these genes during osteoblast differentiation and development are just beginning to be

deciphered (Siddhanti and Quarles, 1 994). Transcription factors are DNA- bind ing proteins that play

central roles in regulating ce11 growth, development, and differentiation by interacting with DNA sites

that regulate gene transcription. Evidence is beginning to emerge from recent research of the

involvernent of transcription factors in gene regulation during osteoblast development and differentiation

and some of these may themselves prove useful as markers of osteoblastic cells. I will not attempt to

summarize this area exhaustively, but rather restrict the discussion to several recent examples of interest.

Sorne of these transcriptional regulators include hXBP- 1 (a member of the basic region-Ieucine zipper

r o o o o b l \ O \ O \ O \ O

I L-6 TGF-P IGF-1 IGF-II IGF-RI Estrogcn-R -1- 8 CD44 - -t 29 a, Integrin +/- + t / - 25 a , lntegrin + t-

a,P, integrin t t

P3/P5 integrin 1- - t Throm bospondin -t/-- .t

Fibroneciin +/-- i

Biglycan Decorin c-fos

Porcine COLL-1 OPN + t 1. 20 BSP - -t 1- 30 OCN Osteonectin t t .t- 20 Estrogeri-R -1. 8

Cliickcn C0LL-I 4- 39 ALP OPN Tenascin t -4

bFGF t /- IW-iiPT HrP-R i t

NCAM + t

93-2, SB-3 - t

SB-5, OB7.3 - -

+, positive expression; -, iiegative exprcssioii; +/-. weok exprcssioii; -/+, iiegetivc or positive expressioii depeiidiiig on study; *, iodinatcd ligand binding

family) which has been found to be expressed by in situ hybridization in osteobIasts and preosteoblasts in

areas of ossification (Clauss et al., 1993); a novel transcriptional coactivator termed 1.9.2 which

specifically inmunolabels osteoblasts (Yotov et al., 1995); Msx-2 protein (homeodomain gene family

member) which has been suggested to be expressed in undifferentiated but not in differentiated mouse

calvarial osteoblasts (Dodig et al., 1995). and has been observed to be a transcriptional suppressor of the

rat OCN promoter (Towler et al., 1994); and c-fos (proto-oncogene) which has been reported to label

osteoprogenitors but not mature osteoblasts by immunohistochemistry (Machwate et al., 1995).

Knowledge of the nature of the role of these transcription factors in osteoblast cornmitment and

regulation of differentiation is still sparse, but they are known to regulate at least certain osteoblast-

associated genes, and their own expression profiles may provide landmarks as osteobiastic cells mature.

Methods aimed at generating specific markers and tools for isolating cells of the osteogenic

lineage include the generation of monoclonal antibodies directed against differentiation stage-specific

cellular or matrix-derived antigens (reviewed in Aubin and Turksen, 1996). Nijweide and Mulder (1986)

prepared a monoclonal antibody (OB 7.3) reacting only with chick osteocytes. Although the epitope or

macromolecule recognized is still unknown, OB 7.3 is being used to isolate and characterize relatively

pure populations of chick osteocytes (see above and van der Ptas and Nijweide, 1992; van der Plas et al..

1994). El I is another monoclonal antibody that labels rat osteocytes and a proponion of hlly

differentiated osteoblasts (Wettenvald et al., 1992. 1996). Recently it was reported to recognize OTS-8

(Nose et al.. 1990). to belong to the mucin family, and to mediate shape changes in osteoblasts (Sprague

et al., 1996). Bruder and Caplan ( 1989, 1990a, 1990b) and Turkscn et al. ( 1992) reported a series of

antibodies directed against different subpopulations of chick or rat osteoblastic cells, respectively. Ones

characterized include ones recognizing ALP (Bruder and Caplan, 1989; 1 !Boa, 1990b; Turksen and

Aubin. 1991), and galectin 3 (Aubin et al., 1996a). The latter antibodies, and antibodies against some of

the bone matrix macromolecules (e-g., COLL-1, ALP, OPN, OCN) (reviewed in Aubin and Turksen,

1996; to be discussed later) and transcription factors (see above), are being used as tools to attempt to

del ineate discrete stages during osteogenic development and di fferentiation.

Experimental Approaches

In vivo studies

Toois for investigating osteoblasts in vivo such as immunocytochemistry and in situ

hybridization have been utlilized for associating the expression of several proteins and their messages,

respect ively, at the ce1 IuIar IeveI with the four morphologically recognizable stages of osteoblast

devetopment (Table 1.3). Not al1 proteins and mRNAs have been associated with al1 of the individual

types of osteogenic cells, helping to add biochemical features to their histological features. Most of the

available markers recognize the more differentiated osteoblast and osteocyte, such that it has been

difficult to recognize ceIls earlier than the preosteoblast. Nevertheless, these results have provided some

important insight into the phenotype of osteogenic celis, at least at late stages of osteoblast development

(see Chapter 2, Introduction). For exampie, ALP is expressed in preosteobiasts (Doty and Schofield,

1976; Weinreb et al., 1990) and is an earlier differentiation marker than OPN (Mark et al., 1 988;

Weinreb et al., 1990: Chen et al., 1993a), followed by BSP and OCN in mature osteoblasts (Bronckers et

al., 1987; Mark et al., 1988; Chen et al., 1993a). Sorne variations in levels of expression are associated

with different morphology and relationship of the cells to the bone matrix and their synthetic lifetirne

(Martin et al., 1988; Gehron Robey et al., 1993). For example, differences in amounts of extractable bone

proteins have been reported for trabecular vs. cortical bone (Ninomiya et al., 1990); differences in bone

matrix protein mRNA expression have been observed osteoblasts in young vs. adult rats (Ikeda et al..

1992; 1999, between bones undergoing intramernbranous (calvaria) vs. endochondral (tibia) bone

formation with respect to development (Chen et al., 1992, 1993), and between newly differentiated vs.

more mature osteoblasts (Heersche et al., 1992). These variations may refiect the existence of osteoblast

subpopulations at different stages of differentiation and at different sites/microenvironments in bone

(Martin et al., 1988; Heersche and Aubin, 1990; Heersche et al., 1992; Gehron Robey et a[., 1993). Other

variations may be attributable to the species examined, the age of the specimens, the anatomical site of

the bone (e.g., calvaria, tibia, vertebra, mandible) or the process of ossification followed

(intramembranous vs. endochondral), the part of the bone studied (trabecular vs. cortical), and even

differences in the immunocytochemical techniques applied. I will not be addressing most of these sorts

of variations in this thesis, but rather focus on one cell model (RC cells) undergoing differentiation and

maturation in vitro.

In vitro models

Cell culture models have proved to be invaluable tools for study of the proliferation,

differentiation, and function of osteoblastic cells. Characterization of cells as osteoblast-like has been

based on three general criteria: production of osteoblast-associated products, expression of characteristic

hormone receptordres ponsiveness as outlined above. and ability to make a bone-1 ike tissue (Majeska and

Rodan, 1985; Rodan and Noda, 1991). Major models investigated include prirnary cultures of tissue-

derived cells (e.g., rat, mouse, calf or chick calvaria; rat, rnouse, bovine or human trabecular bone), and

clonal cell lines derived either from osteosarcorna tumours (e.g., rat ROS 1712.8 or UMR 106; human

MG-63 or SaOS-2) or from prirnary bone cell cultures (e-g., MC3T3-El, UMR 20 1, and RCJ ceIl lines).

Based on their different expression profiles of osteoblastic characteristics, and what is expected

vis-à-vis morphologically recognizable osteoblastic cells in vivo, MC3T3-El cells have been described

as relativeiy immature, UMR 106 as preosteoblastic cells, and ROS 17/23 as more differentiated

osteoblasts (reviewed in Rodan and Noda, 199 1 ). In addition to heterogeneity between clones,

phenotypic variations exist within clonal osteoblastic populations whether of tumorigenic (Grigoriadis et

al., 1985; Majeska and Rodan, 1985) or non-tumorigenic origin (Aubin et al., 1982). ALP activity,

detected by histochemical staining in ROS 1712.8 cultures has been repo~ed to be heterogeneous and this

may represent the presence in ROS 1712.8 cultures of cells at various stages of phenotypic maturation

(Majeska and Rodan, 1985). Consistent with this, elevated levels of ALP and other osteoblast markers

develop with time post-confluence in al1 of the above ce11 lines (Rodan and Noda, 1991) and the MC313-

E 1 line at least appears to be able to tenninally differentiate to form mineralized bone in vitro in at Ieast

some labs (Sudo et ai., 1983; Franceschi and Iyer, 1992). However, at least some of the phenotypic

variations observed both between and within clones does not appear consistent with this interpretation

and instead may represent the aberrant, transformed nature of the lines or the fact that they are

established to long term growth (Aubin et al., 1993). In addition, clonai ceIl lines appear inherently

unstable to a greater or Iesser degree (Heersche et al., 1985; Bellows et al., 1986b; Majeska and Rodan,

1985).

Osreoblast phenotype develûpmenr in vitro

The advantages of primary cultures of bone cells are that studies rnay be carried out on

nontransformed populations, and that potentially they contain al1 the different stages of osteoblastic cells

known to be present in bone. Prirnary cultures of fetaI RC-derived cells, obtained by enzymatic

digest ion, have contri buted much to our understanding and definition of the osteoblastic phenotype,

although full understanding of the cultures is still evolving. These populations of cells are heterogeneous

and contain a number of cell types including osteoblastic cells at different stages of differentiation and

fibroblastic cells (Aubin et al., 1982; Rodan and Rodan, 1983; Heersche et al., 1985; Hodge and Kream,

1988; Rodan et al., 1988; Guenther et al., 1989; Heersche and Aubin, 1990). These isolated osteobtastic

ce11 populations have been extensively characterized and s h o w to express ALP activity, to synthesize

COLL-I and noncollagenous bone proteins, and to respond to various hormones and cytokines (for

reviews, see Canalis et al., 1989; Aubin et al., 1990, 1992, 1993; Stein and Lian, 1993). Of particular

note for this thesis work. when calvarial-derived ce11 populations are maintained for extended periods in

the presence of serurn, ascorbic acid and kglycerophosphate, well defined three-dimensional bone-like

nodules f o m with the histolagical, ultrastructural and imrnunohistochemical appearance of

embryonic/woven bone (Nefussi et al., 1985; Bellows et al., I986a; Bhargava et al., 1988). In cross-

section, these nodules comprise a continuous layer of cuboidal cells having the histological and

rnorphological characteristics of osteoblasts with cells resembling osteocytes embedded within the

matrix of the nodule, and completety surrounded by collagenous matrix. The rnatrix of nodules contains

COLL-1, osteonectin, and OCN; cells associated with the nodules label intensely for ALP and the

mineral of the colfagenous matrix is hydroxyapatite (Nefussi et al., 1985; Bellows et al., 1986a;

Bhargava et al., 1988).

It was determined that the number of nodules formed was linearly related to the number of cells

plated over a wide range of plating densities (Bellows et al., 1986% 1987). Limiting dilution analysis

hrther showed that only one ce11 type was Iimiting for nodule formation in fetal RC populations,

apparently the osteoprogenitor itself, that they were present at less than 1 % of the total isolated

population and that one osteoprogenitor ceIl develops into one bone nodule under standard culture

conditions (BeIIows and Aubin, 1989). The addition of dexamethasone, a synthetic gIucoconicoid.

increased the number of nodules formed suggesting that dexamethasone is required in order for some

osteoprogenitor cells to form bone (Bellows et al., 1987, 1990), suggesting that subpopulations of nodule

forming cells may exist in RC populations. When a monoclonal antibody against ALP was used to

fractionate RC cells into ALP' and ALP- cells. bone nodules formed in the ALP' population in the

absence of dexamethasone. whereas nodules formed in the ALP- populations only in the presence of

dexamethasone, suggesting that an immature ALP- osteoprogenitor (requiring dexamethasone) and

mature ALP' osteoprogenitor were distinguished (Turksen and Aubin, 199 1). Hence, a spectrum of

osteogenic cells. including osteoprogenitor cells at different maturational stages, are present in calvarial

ceIl populations.

Primary cultures of fetal RC cells with osteoprogenitors undergoing bone nodule formation and

mineralization provide a useful in vitro mode1 system for investigating the differentiation and

development of functional osteoblasts. Bone nodules form in RC populations afier populations reach

confluence and begin to multilayer (Nefussi et al., 1985; Bellows et al.. 1986a). The process of nodule

formation was further defined as comprising three time periods: (i) proliferation. (ii) extracellular matrix

developrnent and maturation, and (iii) mineralization (reviewed in Stein and Lian, 1993). With Northern

blot analysis and in situ hybridization, a temporal pattern of gene expression reflecting stages of

progressive bone formation was described. Stein, Lian and colleagues have reported that COLL-1 is

expressed in the proliferative phase and in gradually downregulated as differentiation proceeds; the

second phase is characterized by an increase in the expression of ALP, and progressive expression of

OPN and OCN; finally, when cultures are heavily mineraiized ALP was reported to decrease (reviewed

in Stein and Lian, 1993). Others have found somewhat different expression profiles, in which, e.g.,

COLL-1 expression continues to increase markedly as nodules form (Lee et al., 1992; Malaval et al.,

1994), more consistent with in vivo observations. While these mass population studies address some

aspects of the temporal relationship of osteoblast differentiation to the expression of specific bone-

related macromolecules, many questions and arnbiguities remain about the expression of osteoblast

phenotype markers and osteoblast differentiation, as will be discussed in more detail in Chapter 2.

Thesis Objectives

The overall objective of my thesis was to establish new methods for investigating the osteoblast

lineage, and in particular, to provide new insights into transitionat stages in osteoblast differentiation,

including cells more primitive than preosteoblasts, osteoblasts, and osteocytes. Given that the RC cell

system contains normal osteoprogenitors that undergo proliferation and differentiation steps leading to

terminal differentiation, it was chosen as my mode1 of choice. The fact that their number is low,

however, and that they are not morphologically or molecularly identifiable, meant that I had to develop

approaches by which to find them and follow their differentiation process, without confounding aspects

resulting frorn the presence of other cells and osteobiastic cells themselves at other stages of

differentiation.

To these ends, 1 first applied the recently described technique of poIy(A)-PCR which allows

expressed gene repertoires to be analyzed, and immunocytochemistry, which provides data on protein

expression coupled to ceII rnorphological characteristics, to determine the temporal expression of bone-

related macromolecules at the single cell level among osteoprogenitor colonies of clonal origin

undergoing differentiation in vitro (Chapter 2). 1 then analyzed in more detail the heterogeneity 1

observed amongst cells classed as mature osteoblasts (Chapter 3 j. To identie unambiguously cells

earlier than osteoblasts, i.e., osteoprogenitors and preosteoblasts, 1 established a replica technique,

validated it, and then used it with poly(A)-PCR to analyze early stages in osteoprogenitor differentiation

(Chapter 4).

Simultaneous Detection of Mulripe Bone-Related mRNAs and Protein Expression during

Osteublasî Dîjferentiation: Poiyrnerase Chain Reaction and Imnrunocytochernicrtl S M e s at

the Single Cell Level

This chapter has been published in Developmental Biology ( 1 994) 166,220-234 as an original paper

with the above title (Fina Liu, Luc Malaval, Ashwani K. Gupta, and Jane E. Aubin)

Introduction

Relativeiy little is known about the number of steps leading fiom mesenchymal stem ce11

through committed osteoprogenitor to fully differentiated osteoblast and related cells, nor is it clear how

many discrete stages can be identified in the osteoblast differentiation pathway and how cells at each

stage differ. Osteoblasts, defined by their site on bone and their ability to synthesize mineralized matrix,

have a few known properties, Le., ability to synthesize COLL-1. SPARC/osteonectin, OPN, BSP, and

OCN. and high ALP activity; other properties such as responsiveness to hormones (e.g.. PTH) and

growth factors (e-g.. epidermal growth factor) are also characteristic of osteoblast populations (for

reviews, see Rodan and Rodan. 1983: Heersche et al., 1985; Majeska and Rodan, 1985; Rodan et al.,

1988: Wong, 1990).

Changes in levels of expression of these osteoblast-associated molecules and different

combinations of several of the properties shown to be associated with osteoblast-like cells may be

characteristic of cells at different developmental or maturational stages (Yoon et al., 1 987; Ng et al..

1988: Heath et al., 1989; Stein et al., 1989; Aubin et al., 1990. 1992; Owen et al., 1990, 1991; Turksen

and Aubin. 199 1 : Heersche et al.. 1992; Bianco et al.. 1993). For example, over a time course of

osteoblast differentiation in some model systems in vitro. or when osteoblasts are analyzed in vivo,

expression of COLL-I is relatively high and then decreases (Gerstenfeld et al.. 1988: Stein et al., 1989;

Owen et al., 1990. 1991); ALP increases but then decreases when mineralization is weIl progressed

(Bronckers et al., 1987; Mark et al., I987b: Stein et al.. 1989; Owen et al., 1990, 199 1 ; Zernik et al.,

1990; Turksen and Aubin; 199 1 ); OPN appears prior to certain other matrix proteins including BSP and

OCN (Mark et al., i987a, 1987b, 1988; Stein et al., 1989; Owen et al., 1990; Chen et al., 199 1 b; Moore

et al., 1991; Pockwinse et al., 1992); BSP is first detected in differentiated osteoblasts forming bone

(Bianco et al.. 1991; Chen et al., 1991 b); and OCN appears with mineralization (Bronckers et al., 1985,

1987; Groot et al., I986; Gerstenfeld et al., 1987; Mark et al., 1987a, 1988: Stein et al., 1989: Boivin et

al., 1990; Owen et al., 1990, 199 1 ; Pockwinse et al., 1992). In other analyses, COLL-1 mRNA was

reported to be approximately the same in newly differentiated osteoblasts as in more mature osteoblasts,

while OCN mRNA was undetectable in newly differentiated osteoblasts but clearly detectable in more

mature osteoblasts and, relative to COLL-1, was higher in old flat osteoblasts compared to younger

cuboidal osteoblasts (Heersche et al.. 1992). These and other studies suggest that expression of at least

some osteoblast markers varies during the differentiation or maturation cascade, however, depending on

the approach used (e.g., mRNA detection versus protein expression) and/or on the ceIl system used and

culture conditions. temporal aspects of the difierentiation process and ability to ascertain a sequential

expression of markers varies markedly and sometimes is contradictory. One of the complicating factors

is that despite attempts to isolate populations of relativeIy homogeneous composition, the populations are

still a mixture of different celi types or differentiation stages and most of the availabte osteoblast-

associated markers are not restricted to osteoblast cells (Nijweide et al., 1988; Aubin et al., 1990, 1992;

Heersche and Aubin. 1990). Even with the availability of clona1 cell lines of normal or tumor origin, it

can be difficult to see an unambiguous pattern of expression of osteoblast markers or to attribute the

extreme heterogeneity observed to stage of differentiation. age. or consequence of passaging (Wong,

1980: Aubin et al.. 1982. 1993). Thus. for example. the negative feedback of mineralization on matrix

synthesis reported earlier (Tenenbaum. 1987: Tenenbaum et al., 1989) was not found when the

mineralization phase of cultures was controiled and looked at independently of other stages (Lee et al..

1992). When immunocytochemistry was coupled with Northern blotting in analysis of rat bone marrow

stroma1 cultures. the imrnunolabeling revealed small numbers of OCN- mature osteoblast cells. which

were not detectable by Northem analysis of the whole population. present transientty early in cultures

before new osteogenesis (bone nodules) occurred (Malaval et al., 1994). Finally. there are discrepancies

between observations by immunolabeling (Silve et al., 1982; Rao et al.. 1 983) and in situ hybridization

(Abou-Sarnra et al.. 1992; Urena et al., 1992) which show mRNA for PTH receptors on the osteoblast

and its immediate precursors versus other studies with iodinated hormone which suggest that the highest

number of receptors is on a relatively undifferentiated cell, with relatively few on the mature osteoblast

itself (Rouleau et al., 1988, 1990). These latter data are also dificult to reconcile with data from some

studies in vitro in which a PTH-stimuiated adenylate cyclase response was acquired along with ALP in a

clonal, osteoprogenitor-like cell line, RCT- 1. induced to mature with retinoic acid (Heath et al., 1989).

These exarnples point to the discrepancies and voids in Our understanding of when and on what celis a

variety of markers first appear and/or disappear during osteoblast differentiation and maturation (for

further discussion, see Aubin et al., 1993).

When cells enzymatically isolated from fetal RC are grown in medium supplemented with

ascorbic acid and ~glycerophosphate, discrete three-dimensional mineralizing nodules form which have

the histological. ultrastructural and imrnunohistochemical appearance of woven bone (Nefussi et al.,

1985: Bellows et al., I986a; Bhargava et al., 1988). However, RC ce11 populations are heterogeneous and

contain osteoblast cells at different stages of differentiation, osteoprogenitor cells and other lineages

(Aubin et al., 1982. 1992; Rodan and Rodan, 1983; Heersche et al., 1985: Hodge and Kream, 1988;

Rodan et al., 1988: Guenther et al., 1989: Heersche and Aubin, 1990). Limiting dilution analysis has

indicated that less than 1% of the celIs in the isolated RC population are osteoprogenitor cells capable of

dividing and differentiating to form bone nodules in vitro and that one osteoprogenitor cell gives rise to

one bone nodule under standard culture conditions (Bellows and Aubin. 1989). Thus. by piating fetal RC

celIs at very low densities. it is possible to obtain single isolated colonies which have arisen from

individual cells that have attached and proliferated and a few of these colonies represent the

osteoprogenitor cells and their progeny, yielding a spectrum of differentiated osteogenic cells including

osteoblasts that f o m bone which mineralize (Bellows and Aubin, 1989; Aubin et al., 1990). Other

colonies may represent other lineages. cg., fibroblastic. adipocytic, or other osteoblastic stages not

capable of forming bone nodules in vitro. Single cells can be isolated from individual colonies over a

time course at different stages of differentiation. Therefore, primary cultures of single colonies arising

from fetal RC cells provide an in vitro model system in which to address experimentally the relationship

of osteoblast difierentiation to the expression of specific bone-related macromolecules in a normal, non-

transformed population that progresses to recapitulate a sequence from proliferation to terminal

differentiation.

Of the toois available with which to assess the differentiation stage of cells in vitro, most study

populations of living cells or dead, fixed cells, and are not suitable for single living cells or small

numbers of cells. Recently, the entire spectrum of polyadenylated mRNA present in a single ce11 has

been simultaneously amplified using poly(A)-PCR and probed many times for the presence of different

sequences (Brady et al., 1990; Brady and Iscove, 1993). We have employed this poly(A)-PCR technique

and cornbined it with immunocytochemistry to investigate the simultaneous expression of multiple bone-

related macrornolecules in single colonies and single cells from individual discrete fibroblastic and

osteoblastic colonies undergoing differentiation in vitro.

Materials and Methods

Cell Culture

Cells were enzymatically isolated from the cafvariae of 2 1 d Wistar rat fetuses by sequential

digestion with collagenase as described (Rao et al., 1977; Aubin et al., 1982; Bellows et al., 1986a). Cells

obtained from the last four of the five digestion steps (populations II-V) were pooled and plated in T-75

flasks in a-MEM containing 15% heat-inactivated FBS (Flow Laboratories, McLean, VA) and

antibiotics comprising 100 pg/ml penicillin G (Sigma Chemicîl Co., St. Louis, MO), 50 pglml

gentamycin (Sigma), and 0.3 pg/rnl fungizone (Flow Laboratories). After 24 h incubation, attached cells

were washed with PBS to remove nonviable cells and other debris, and then collected by trypsinization

using 0.01% trypsin in citrate saline. Aliquots were counted with a Coulter Counter, and the remaining

cells were resuspended in the standard medium described above supplemented with 50 pg/rnl ascorbic

acid, I O m M sodium fl-glycerophosphate, and 10 nM dexarnethasone (Merck, Sharp, and Dohme,

Canada, Ltd., Kirkland, PQ). The resuspended cells were plated into 100 mm tissue culture dishes at 10-

35 cells/cm2. Medium was changed every 2-3 days. AH dishes were incubated at 37°C in a hurnidified

atmosphere of 95% air/5% CO2 incubator.

Single Cell lsola~ion and cDNA P r e p m i o n

Individual colonies with morphoIogies of interest and well-separated fiom other colonies were

marked. The ceIls were rinsed with PBS, and a cloning ring was placed around each marked colony. The

cells from the colony were released with either 0.0 1% trypsin (when matrix was devoid of mineral) or a

1 : 1 mixture of 0.0 1% trypsin and collagenase containing mixture (Rao et al., 1977) (when osteoid was

mineralizing) and the enzymes were neutralized after cell release by the addition of a-MEM containing

15% FBS. The cells were split into two portions. One portion was resuspended in O. 1% methyl

cellulose/PBS and from this portion cells were drawn into a g las micropipette with the help of a

micromanipuIator (Narishige. Japan); suction was applied by mouth through a connected plastic tubing.

After withdrawing the pipette containing a single cell, the ce11 was ejected into a Terizaki well containing

4 pl of prechilled lyçis/lst strand cDNA buffer (Brady et al., 1990; Brady and Iscove, 1993). The ce11 and

buffer were then transferred into a PCR reaction tube. The second portion consisting of the remaining

cells was Iysed with vortexing in a solution of 5 M guanidine thiocyanate, 0.5% sarkosyl, 25 m M sodium

citrate pH 7.0, 20 m M dithioerythritol: FUVA was precipitated ovemight with 0.5X volume of 7.5 M

ammonium acetate and 800 &ml glycogen (Boehringer Mannheim Canada, Laval, PQ) and 3X volume

ethanol. The pellet was washed twice with 75% ethanol and resuspended in a buffer containmg 0.5% NP-

40 (Sigma) and 1 U Inhibit Ace (5'-3' Inc., West Chester. PA). One microliter of this resuspension was

placed directly in 4 pl of prechilled lysis/ 1 st strand buffer (Brady et al., 1990; Brady and Iscove, 1993)

for amplification of whole colonies as "populations". In both cases, the mRNA was reverse transcribed

into cDNA to about 300-700 bases in size. which was then poly(A)-tailed, and amplified using oligo(dT)

and PCR as described previously by Brady et al. (1990). Samples were initially amplified for 25 cycles

of 1 min at 94OC, 2 min at 42°C. and 6 min at 72OC in a thermal cycler (Perkin Elmer Cetus, La Jolla,

CA), then linked to an additional 25 cycles of 1 min at 94"C, 1 min at 4Z0C, and 2 min at 72°C.

Hybr idizat ion

Southern blots were prepared by running 5 pl aliquots of amplified cDNA on OSX Tris-

boratefEDTA agarose (1 -5%) gels followed by transfer to BioTrans nylon membrane (0.2 pm pore size;

ICN, Costa Mesa, CA) in 2OX SSC by the capillary method of Sambrook et al. (1989). Complementary

DNA probes were labeled with ['*P]~cTP using an oligolabeling kit (Pharmacia, Uppsala, Sweden).

Total cDNA probe was prepared as described by Sambrook et ai. (1989) from poly(A)+ mRNA isolated

(Aufiay and Rougeon, 1980) from mass populations of fetal RC cells grown in the presence of

dexamethasone in which bone nodules were beginning to mineralize. Labeled probes were used at an

activity of 106 cpm/ml. Al1 prehybridizations and hybridizations were performed in 50% formamide, 5X

SSC, 5X Denhardt's, 50 m M sodium phosphate buffer pH 6.5,0.1% SDS, and 250 rng/rnl salmon sperm

DNA at 42OC for overnight. After hybridization, the blots were washed once for 30 min each at 42°C and

50°C in 2X SSC/O. 1% SDS, at 55°C and 60°C in 1 X SSC/O. 1% SDS, and at 6S°C in 0.5X SSCIO. 1%

SDS. The blots were exposed to Kodak X-OMAT film at -70aC using intensiGing screens.

cDNA Probes

In keeping with the size restriction to 300-700 bases of 3' sequence, optimal detection of specific

sequences required probes which included sequence at or close to the extreme 3' ends of the native

transcripts. Rat a 1 COLL-1 (pa 1 R2) (Genovese et al., 1984; kindly provided by Dr. D. Rowe,

Farmington, CT) was a 900 bp cDNA Psrl fragment containing the entire 3' non-coding region and one

half of the C-terminal of the propeptide of the a 1 chain of type 1. Rat bone/liver/kidney ALP (Noda et

al., 1987; gifi of Dr. G.A. Rodan, Merck Sharpe, and Dohme Research Laboratories, West Point, PA)

was a 600 bp cDNA EcoRi fragment O btained by digesting pRAP54 with BssHZI-NroI to remove 1.8 kb

of 5' region and religating the blunt ends. Rat OPN (kindly provided by Dr. R. Mukherjee, Montreai, PQ)

was a 700 bp cDNA BamHI-EcoKi fragment obtained by digesting full length cDNA with Pvtd to

remove 800 bp of 5' region and ligating the blunt ended fragment into Smakut pGEM-7Zf(+) vector

(Promega, Madison, Wt). Rat BSP (A.K. Gupta and J.E. Aubin, unpublished data) was a partial cDNA

containing 500 bp of 3' region isolated with BSP-specific primers from a hgtI 1 library prepared from RC

cells forming bone nodules. Rat OCN (A.K. Gupta and J.E. Aubin, unpublished data) was a partial cDNA

containing 350 bp of 3' region isolated with OCN-specific primers from a lgtl 1 library prepared fiom

ROS 17f2.8 cells.

Imrnunocy~ochernis~ry

Immunocytochemistry was donc essentially as described (Turksen et al., 1992) but with

modification for single colonies. Medium was rernoved from culture dishes and they were rinsed three

times with PBS. Cells were fixed with 3.7% formaldehydePBS for 5 min and permeabilized with

methanol at -20°C. After rinsing, the dishes were air-dried until the buffer had retracted to form a drop

on each isolated colony, leaving the surrounding surface dry. Selected colonies were then isolated by

surrounding them with a ring of wax, and antibody solutions and rinsing buffer were added and removed

over each colony with a pipette. The colonies were incubated for 1 h at room temperature with

appropriate dilutions of primary antibodies in Tris/caIcium buffer (50 m M Tris pH 7.4, 500 m M NaCI, 2

mM CaC12, 3% BSA). The anti-rat OCN antiserum was kindly provided by Dominique Modrowski

(WSERM U349, Hôpital Lariboisière, Paris, France), and used at 1/200 dilution. The anti-OPN

monoclonal antibody was purchased from the Hybridoma Bank (Iowa City, IA) and used at a 11800

dilution. A sheep polyclonal antiserum raised against COLL-1 extracted and purified from pig (Connor et

al.. 1983) was used at a III O dilution. The production and characterization of monocIonal antibody RBM

3 1 1-13. directed to rat bonelliver/kidney type ALP, have been described elsewhere (Turksen and Aubin,

199 1 ; Turksen et al., 1992): it was used at a 1/100 dilution of purified ascites fluid. Controls contained

non-immune rabbit or sheep serum (1/100 in buffer) or a non-specific mouse immunoglobulin (mouse

IgG3 (KAPPA), Cappel, West Chester, PA; 100 pg/ml in buffer). The colonies were rinsed in buffer and

incubated with anti-mouse, anti-rabbit or anti-sheep secondary antibodies conjugated either with

fluorescein (Dupont, Boston, MA; 1/50 dilution) or with CY-3 (Jackson Immunoresearch Lab, West

Grove, PA; 11200 final dilution) for 30 min at room temperature. For double labeling experiments, the

samples were incubated successively in each prirnary antibody, then in a mixture of the secondary

antibodies. After final rinsing of the whole dishes, the preparations were mounted in Moviol (Hoescht

Ltd., Montreal, PQ). The dishes were observed by epifluorescence microscopy on a Zeiss

Photomicroscope III (Zeiss, Oberkochen, Gerrnany). For photography and printing, equal exposure times

were used for experimental and control specimens.

Results

Co/ony Morphology

By plating fetal RC cells at very low densities ( 1 0-35 cells/cm2), it was possible to obtain single

isotated colonies which had arisen from individual cells that had attached and proliferated. A few of

these colonies represent osteoprogenitor cells and their progeny (Fig. 2.1 ); on differentiation, these yield

a spectrum of differentiated osteogenic cells, form bone nodules and rnineralize (Bellows and Aubin,

1989: Aubin et al., 1990). Other colonies represent fibroblastic cells; still others have been shown to be

adipocytic (data not shown). Single colonies or single cells from individual colonies were isolated over a

time course at different stages of differentiation. Colonies selected were morphologically identifiable and

classed as fibroblastic (monolayer of spindle-shaped cells) or osteoblastic (packed cuboidal cells).

Osteoblastic colonies were further subdivided into: (i) early osteoblastic (monolayer of cuboidal cells

with minimal deposited matrix) (Fig. 2.1 A, B); (ii) osteobiastic (multilayer of cells with abundant

osteoid matrix but no detectable mineral); and (iii) mature osteoblastic (multilayer of cuboidal cells with

abundant osteoid and well-progressed mineralization) (Fig. 2.1 B, D). As evident in the micrographs

(Fig. 2. 1). in growing colonies, the morphologically most mature cells in the colony reside in the central

zone, followed by an intermediate zone and less differentiated cells at the colony periphery. The

expression of specific bone-related macromolecules was examined in single colonies and single cells

from these individual discrete colony types using poiy(A)-PCR and imrnunocytochemistry.

PohfA) -PCR

Individual cells from colonies categorized into morphologically recognizable cet1 types were

micrornanipulated into lysis/l st strand cDNA buffer, processed individually to yield amplified material.

a fraction of which was electrophoresed, transferred and probed for the expression of various bone-

related genes. Expression of single cells was cornpared to that of whole colonies or the "population" as

represented by al1 the rest of the cells from the same colony. At the time of analysis, colonies derived

from single cells had been growing in vitro for 2-3 weeks and colony sizes ranged fiom approximately

Figure 2.1 Isolated discrete colonies designated as osteogenic on the basis of ceII morphology and

colony characteristics. (A) Early osteoblastic colony with monolayered cuboidal celts and little osteoid.

(B) Mature osteoblastic colony with multilayered cuboidal cells and mineralized osteoid; the outer edge

of the colony is slightly out of focus as a result of the three dimensional nature of the bone nodule. (C. D)

Higher magnifications of a portion through approximately the center of the above early (C) and mature

(D) osteoblastic colonies, respectively. Bar, (A, B) 275 lm; (C, D) 200 p.

500-2000 cells per coiony. Total colony and individuai cell analyses have been repeated at least five

times, with comparable results fiom al1 experiments. The Southern blots shown (Fig. 2.2) are

representative of the usual results.

ln control experirnents, we verified that the cDNA was size restricted as expected to 300-700

bases and amplification conditions for single cells were such that amplification was exponential (data not

shown). We also found expression of mRNAs for so-called housekeeping genes such as actin, @tubulin,

or glyceraldehyde-6-phosphate dehydrogenase to Vary substantially in individual cells (data not shown),

consistent with other reports on poly(A)-PCR-amptified single cells (Trurnper et al., 1993). Thus, a total

cDNA probe, which comprised the total cDNA repertoire prepared from poly(A)+ mRNA of mass

populations of fetal RC cells, was used as control and total cDNA signai strength to quanti@ relative

abundance. Use of total cDNA indicated that amplification was achieved from 100% of single colonies

and more than 95% of single micromanipulated cells. We have also included the hybridization results

from whole colonies to show the average level of expression of al1 cells from the same colony. The

averaged data give semi-quantitative results similar to Northern analysis, whereas the single ce11

expression patterns show degree of intercellular variabil ity and the repertoire and levels of individual

mRNAs expressed simultaneously in individual cells.

PCR amplification of cDNA from individual colonies of different morphologies was

reproducible and showed that the repertoire of expression of particular messages was distinct in different

colony types (Fig. 2.2, whole colonies). For example, COLL-1, ALP and OPN were detectable in

virtually al1 colonies, but were less highly expressed by colonies morphologically classed as fibroblastic

compared to ones classed as osteoblastic. COLL-1, ALP and OPN were most highly expressed in

colonies representative of relatively mature stages of osteoblast differentiation (rnultilayered to

rnineralized colonies). BSP and OCN message levels were not detectable in colonies of fibroblastic

morphology but were clearly detectable in the osteoblastic colonies, and were most highly expressed in

the osteoblastic colonies in which osteoid was already detectably mineralized.

Having verified that the amplification procedure worked reproducibly and with fidelity

documented a different repertoire of mRNAs highly expressed in osteoblastic versus fibroblastic

Early OB Mature OB

Total cDNA

Collsgen Type 1

Alkallne Phosphatase

Osleopontin

Figure 2.2 Southerns of samples ainplified by poly(A)-PCR. Whole colonies (or "populations") and single cells froni each whole colony type were

exainined by poly(A)-PCR at various stages of differentiation as deterinined inorphologically froin Figure 2.1 : fibroblastic (FB- 1 , FB-2). early

osteoblastic (Early OB), osteoblastic (OB), and mature osteoblastic (Mature OB). The mRNA from each colony type or single ceIl was reverse

transcribed into cDNA and ainplified by tlie poly(A)-PCR; tlie control sainple consisted of no cells. The resulting cDNA amplified froni each colony

or ceIl was tlien probed for the expression of total cDNA and the individual iiiessages froin various boiie-related proteins. Each vertical lane is the

cDNA froni the same colony or the same individual cell.

colonies, we next sought to ask whether individual cells within each colony type expressed al1 or only

some of the mRNAs and whether cells were heterogeneous or uniform. Of the five specific cDNAs

(COLL-1, ALP, OPN, BSP and OCN) that we analyzed simultaneously in each cell, some individual cells

in osteoblastic colonies expressed al1 five; other cells expressed diverse combinations of these molecules

(Fig. 2.2, Single Cells). First, within particular colony types, both the number of cells in a colony which

expressed, and the level of expression of, a particular message varied between different single cells

isolated from an individual colony, from not detectable to highly expressed. Second, the repertoire of

mRNAs expressed and the level of expression by individual cells was different. Third, if ten cells were

compared from each of several different colony types, however, larger differences in the overall

frequency and level of expression of a message were detectable between cells from different colony

types. For example, COLL-I and OPN mRNAs were detected in most cells from al! colony types and

frequently at high levels. ALP was detectable in some cells classified as fibroblastic as well as in

osteoblastic colony types, but was highest in cells from osteoblastic colonies, especially those in which

mineraIization was extensive. BSP mRNA was detectable in celk of the unmineralized as well as the

mineralized osteoid colonies. OCN mRNA was most highly expressed in osteoblast colonies in which

osteoid was mineralized, although we have detected it in some cells in colonies in which osteoid did not

yet contain detectable mineral.

Variations in Colonies Classified as Fibroblastic

It should be noted that varied results were obtained from colonies classified morphologically as

fibroblastic. Strikingly, OCN mRNA has consistently been detected at very low levels in most PCR

experiments in a small number of cells from colonies classed as fibroblastic (Fig. 2.2), a classification

consistent with the fact that these cells were not expressing other bone proteins, but it has not been

detected when "populations" (whoIe colonies) of fibroblastic cells were analyzed. Also, in one

experiment BSP mRNA was detected along with OCN in a few cells within a fibroblastic colony (e.g.,

Fig. 2.2, sample FB-2), a colony in which BSP mRNA was also detectable at the "population1' level.

Imrnunocytochemistty

The PCR results indicated marked heterogeneity in expression of mRNAs for bone-related

macromolecules in different colony types and between individual fibroblastic and osteoblastic cells in

the same coiony. Therefore, we sought to determine whether heterogeneity at the level of protein

expression also existed. Imrnunocytochemistry was used to examine individual celIs in the sarne kinds of

colony types classed morphologically as above. The antibodies available were directed against COLL-1,

ALP. OPN and OCN.

The majority of cells in al1 colony types labeled for COLL-I (Fig. 2.3). The labeling was not very

intense in cells of fibroblast rnorphology, Le., in either cells of fibroblastic colonies or of spindle-shaped

cells at the periphery of osteoblastic colonies. Nevertheless, sorne fibroblastic cells clearly labeled more

intensely than others; this was especially easy to detect given that the majority of the cytoplasmic stain

for COLL-i was localized to the perinuclear Golgi region. Virtually al[ of the multilayering cuboidal

cells in the center of osteoblastic colonies showed intense labeling, also in the Golgi region; however,

again variation in staining intensity from cell-to-ceIl was discernible. Staining for COLL-1 was also

evident in the matrix of some colonies. e-g., note the labeling of the rnatrix of osteoblastic colonies (Fig.

2.3D).

Anti-ALP (RBM 2 1 1.13) gave a typical diffuse membrane-associated staining, but with marked

variations in intensity behveen cells in different colonies and between cells in the same colony types. In

osteoblastic colonies (Fig. 2.4C). the staining was most intense in those cells at the center of the nodules

cornpared to the periphery, consistent with the rnorphologically detectable more mature phenotype of

celis in the central zone. Cells in fibroblastic colonies labeled not detectably (Fig. 2.4A) or much less

intensely (Fig. 2.4B).

Imrnunolabeling for OPN was present in cells of al1 of the colony types, but again inter- and

intra-colony variation was striking (Fig. 2.5). Few cells were labeled detectably in fibroblastic colonies

(Fig. 2.5A) or at the periphery of nonrnineralized multilayering colonies (Fig. 2.58). The number of cells

staining for OPN increased, however. frorn the periphery towards the center of the latter colony type

(Fig. 2 .X) . Particularly notable was the intense label in the intermediate zone of osteoblast colonies

Figure 2.3 Immunolabeling for COLL-1 in (A) a fibroblastic colony; (B) the peripheral and (C) central

zones of an osteoblastic colony, and (D) a mature osteoblastic colony. The labeling was not very intense

in cells of fibroblast morphology, i.e., in either cells of fibroblastic colonies or of spindle-shaped cells at

the periphery of osteoblastic colonies. The majority of the cytoplasmic stain for COLL-I was localized to

the perinuciear Golgi region, and variation in staining intensity from cell-to-ce11 was discemible. Bar,

(A-C) 15 pm; (D) 50 pm.

Figure 2.4 Membrane-associated staining with anti-alkaline phosphatase (RBM 2 1 1.13). Cells in

fibroblastic colonies labeled (A) not detectably or (B) intensely. (C) In osteoblastic colonies, the staining

was most intense in those cells at the center of the nodules compared to the periphery, consistent with the

morphologically detectable more mature phenotype of cells in the central zone. Bar, 20 Pm.

Figure 2.5 Imrnunolabeling for OPN showed striking inter- and intra-colony variation. (A) Few cells

were labeled in a fibroblastic colony. In an osteoblastic colony, (B) fewer number of cells were labeled

in the periphery of the colony compared to (C) the center. The most intense staining was seen in the

intermediate zone of mature osteoblastic colonies with (D) partially and (E) heavily mineralized osteoid.

(F) The central zone of a mature osteoblastic colony showing cell-to-ce11 variation in staining. Bar, (A-C,

F) 15 Fm; ( D E ) 50 Pm.

(Fig. 2.5D-E). Osteoblastic colonies with partially and heavily mineralized osteoid showed similar

patterns of labeling, but labeling was less intense and fewer cells were labeled (Fig. 2SF).

OCN did not label any ceils in fibrobhstic colonies (Fig. 2.6A) or with fibroblastic morphology

at the periphery of osteoblastic colonies. Thus, only cuboidal celIs in the intermediate and central zones

of al1 colonies classed as osteoblastic were labeled (Fig, 2.6B-E), even in colonies in which minerai

deposition was not yet detectable. Variations in staining intensity from celi-to-cell, even between

adjacent rnorphoIogica1ly indistinguishable cells, waç striking. When the osteoid of bone nodules was

heavily mineralized, the number of cells labeled intensely for OCN decreased.

Variatiot~s in Colonies Classtfied as Fibro blast ic

As with the PCR analysis, irnmunocytochemical results have also indicated that there were

differences between colonies classi fied morphologicai ly as fibroblastic. These differences were not

apparent with COLL-1, OPN and OCN immunolabeling; in contrast to the PCR results summarized

above, labeling for OCN was not detected in any colonies classified as fibroblastic. Differences were

noticeable when we labeled for ALP. Some fibroblastic colonies did not stain detectably for ALP (Fig.

2.4A), whereas others contained a few very lightly stained cells, and still other sirnilar colonies contained

cells al1 of which stained but with marked variation in intensity (Fig. 2.48).

Discussion

In this paper, we have combined low density plating of RC bone forrning cells (Bellows and

Aubin, 1989) with a nonspecitic PCR amplification technique (Brady et al., 1 990; Brady and Iscove,

1993) and with immunocytochemistry to investigate different stages of osteogenic differentiation and

cell-to-ce11 heterogeneity in individual colonies arising from single cells. The results of several

independent experiments have shown that the PCR amplification of cDNA from single colonies is

Figure 2.6 Irnmunolabeling for OCN. (A) Fibroblastic colonies showed only background labeling. For

a11 osteoblastic colony types, labeling was observed in cells in the intermediate and central zones. (B)

Few celIs were intensely stained in an osteoblastic colony. (C) Variations in staining intensity from cell-

to-ce11 was striking. Greater num bers of cells were intensely stained in mature osteoblastic colonies with

(D) partially and (E) heavily mineralized osteoid: as mineralization progressed, fewer ce1 ls were

intensely labeled. Bar, 20 Hm.

reproducible in ternis of the overall level of expression of particular messages between colony types,

such that colonies with fibroblastic morphology are distinguishable from immature osteogenic colonies

and the latter are distinguishabie fiom mature osteogenic colonies; immunocytochemistry for levels of

protein expression confirmed the PCR results. In addition to the detection of different developmental

stages, marked intercellular heterogeneity in expression of mRNA and protein in individual

morphoIogicalfy indistinguishable ceils frorn the same colony was also found. Our data confirm some

aspects of the differentiation sequence seen in mass populations with several well-known markers, but

extend it to a level of single ceIl expression and single ceIl heterogeneity not previously accessible.

We used the procedure descri bed by Brady et al. ( 1990) to generate microgram amounts of

cDNA representative of the entire spectnim of polyadenylated mRNAs in single colonies and single

cells. Relative abundance relationships in the original sample were kept intact in the arnplified product

by fixing conditions for the reverse transcriptase reaction that limited the size of the first cDNA strand to

300-700 bases, rninimizing selection against longer cDNAs during amplification with consequent

distortion of relative sequence abundance. PCR is therefore initiated on cDNA that is relatively uniform

in size and within the range for which amplification is most efficient (Brady et al., 1990; Brady and

Iscove, 1993). The arnplified material from single colonies and single cells was then probed with

multiple probes allowing each colony and each single cell to be analyzed for the particular repertoire of

messages expressed at a particular point in time. Because there is significant fluctuation in expression of

so-called "housekeeping genes" at the single cell level (Trumper et al., 1993, we used a total cDNA

probe to confirm success of the first strand synthesis and PCR amplification steps and to use for

comparison of the hybridization signals with specific probes.

Earlier, in situ hybridization, immunocytochemistry, and Northem analysis have been used to

study the developmental expression of various osteoblast-associated molecules. These approaches have

shown that over a time course of osteoblast differentiation in vivo and in vitro, the expression of

macromolecules associated with osteoblast cells changes; however, am biguities in data from different

approaches and in populations representative of cells at different developrnental stages are extant

(summarized in Introduction). Our poly(A)-PCR and immunocytochemical data on the overall levels of

expression of COLL-1, ALP, OPN, BSP and OCN by single colonies during osteoblastic colony

development are consistent with the results obtained by others using these different experimental

approaches. OPN appeared prior to certain other matrix proteins including BSP and OCN (Mark et al.,

1987% 1987b, 1988; Stein et al., 1989; Owen et al., i990; Chen et al., 199 1 b; Moore et al., 199 1;

Pockwinse et al., 1992). BSP was first detected in differentiated osteoblasts forming bone (Bianco et al.,

199 1 ; Chen et al.. 199 1 b). However, differences are also evident. The expression of COLL-1 was found

to be high through al1 stages of colony maturation; at no time did we find it to be down-regulated as

suggested by others (Gerstenfeld et al., 1988; Stein et al., 1989; Owen et al., 1990, 199 1 ), even when

colonies contained significant mineral. ALP increased as colonies becarne osteogenic and acquired

mineral, but again we did not find it to decrease relative to other mRNAs as described in some other

studies (Bronckers et al., 1 987; Mark et al., 1987b; Stein et al., 1989; Owen et al., 1990, 1 99 1 ; Zernik et

al., 1990). We found OCN to be detectable in osteogenic colonies prior to any detectable mineralization,

an observation that agrees with some studies (Bronckers et al., 1987; Gerstenfeld et al., 1987; Mark et

al., 1987a, 1988 ), but contrasts with those suggesting that OCN appears later at the onset of, or after,

initiation of mineralization (Bronckers et al. 1985; Groot et al., 1986; Stein et al., 1989; Boivin et al.,

1990; Owen et al., 1990, 199 1 : Pockwinse et al., 1992). Some of the differences may be ascribable to the

fact that we used individual discrete colonies or "populations" of cells of clonal origin, rather than mixed

mass population cultures, allowing unarnbiguous assessrnent of expression from particular ce11 types, a

feature of importance given that several of the markers for which high expression is associated with

osteoblast are not, however, osteoblast-specific. Also, for establishment of the system described here, we

did not attempt to isolate cells from colonies totally obscured by mineral, because lengthy digestion

times are required to isolate even a proportion of the cells. Thus, it is possible that we have not sampled

the terminal stages of maturation.

Our data support the hypothesis that individuai fibroblastic and osteobtastic cells are

heterogeneous in expression of matrix-associated proteins (Aubin et al., 1982; Nijweide et al., 1986;

Heersche and Aubin, 1990; Heersche et al., 1992; Aubin et al., 1993). Some of the heterogeneity is

clearly reflecting different differentiation or maturation stages as noted earlier above. However, some of

the differences in message and protein expression between individual cells within a colony more likely

reflect stage of cell cycle, regional differences in the time at which cells reach particular stages of

maturation (i.e., cells from the center versus those fiom the periphery of a colony), temporally

asynchronous di fferentiation, or true ce1 1-to-ce11 heterogeneity. Fedarko et al. ( 1 990) have suggested that

osteoblast activities may be correlated with specific ceIl cycle events based on the obsewation that

variations in ALP activity appear to reflect the distribution of ceils throughout the ce11 cycle. Several

aspects of the proliferative status of the cells and osteoblast marker expression have been analyzed also

in detail in vivo (Bianco et al.. 1993) and in vitro (Stein et al., 1989; Owen et al., 1990). Previously, our

histological, immunocytochemical, and ultrastructural investigations of cross sections of noduIes forrned

from RC cells showed that the cells that existed in a mineralized nodule are heterogeneous. Relatively

fibrobfast-like cells compt-ised the bottom portion of the nodule, whereas the upper surface of the nodule

was covered by a layer of more cuboidal osteoblast-like cells, and osteocyte-like cells were found within

the nodule (Bellows et al., 1986a; Bhargava et al.. 1988). A drawback to single cell PCR analysis of

adherent cells as we report it here was that it was not possible to detemine from which zone of a colony

a single ce11 was removed. However, morphologically, it was possible to stage the colony and its zones

of maturity as indicated above. Concomitantly, considering the known osteoblastic markers used, it was

possible to compare whether a cell was more or less mature. Moreover, imrnunocytochemistry was used

to help localize the protein expression of some of these same markers to particular zones. Thus, while

some of the heterogeneity is expected to reflect the stage a particular ce11 has reached in the

differentiation sequence, this explanation cannot account for the marked heterogeneity of intensity of

expression between adjacent cuboidal cells expressing the mature osteoblast marker OCN. Thus, the

results suggest a true variation in expression of osteoblast-associated markers even between adjacent

morphologically indistinguishable cells, supporting a concept of osteoblast heterogeneity (see also,

Malaval et al., I994). In keeping with this, so-called fibroblastic colonies were also found to be

heterogeneous for expression of the markers, while being morphologically indistinguishable from each

other (see below).

Another interesting finding in this study was the expression of low levels of OCN mRNA by a

small number of single cells from fibroblastic colonies which did not express any of the other osteoblast-

associated mRNAs. Yet, no cells fiom fibroblastic colonies were labeled for OCN protein. One

possibility is that OCN mRNA may be transiently expressed by some fibroblastic cells, in response to

unknown cues, suggesting a currently unrecognized role for OCN in programmes other than

osteogenesis. In this regard, OCN mRNA has been identified using the PCR technique in non-osteoid

tissues (Fleet and Hock, 1992) and OCN mRNA and immunoreactive protein in rat platelets (Thiede et

al., 1993). The variations in immunolabeling for ALP in fibroblastic colonies, and the presence of mRNA

for ALP, OPN, BSP and OCN in some cells of colonies with fibroblastic rnorphology may indicate these

cells to be osteoprogeniton or preosteoblasts. with marker expression preceding overt morphological

changes and prior to copious matrix deposition. We are currentIy testing this hypothesis using replica

plating techniques that will allow osteoprogenitors to be identified retrospectively and their cDNA

repertoire to be studied.

In vitro studies of osteoblast development described to date have utilized mass, heterogeneous

populations of cells, whether of clonal origin or not. However, to understand the relationship between

bone-associated protein expression and osteoblast differentiation. it would be advantageous to follow

single osteoprogenitors and their progeny to address in more detail sequential patterns of expression of

osteoblast-related phenotypic properties. The PCR technique enhances the sensitivity of detecting

differentially expressed mRNAs. The poly(A)-PCR procedure used here offen a further unique

advantage because it is now possible to identify patterns of CO-expression of several genes within the

same cells, whereas other techniques such as in situ hybridization and immunocytochemistry limit the

number of genes which can be detected concomitantly within a single cell. Moreover, we have adapted

the poiy(A)-PCR procedure, previously described for hemopoietic cells, for general use on cells which

grow adherent to a substrate, indeed for cells that require adherence for differentiation. Thus, individual

osteoblast cells and colonies analyzed by poly(A)-PCR can be used in lieu of mass populations to extend

investigation of stages in the progression of osteoblast differentiation. Further, while making

contributions to our understanding of the osteoblast lineage, the methodology and biological conclusions

extend beyond osteoblasts themselves to any system of differentiating cells.

CHAPTER 3

The Mature Osteoblast Plienotype is Characterized by Extensive Plusticiîy

This chapter has been submitted for publication with the above title (Fina Liu, Luc Malaval,

and Jane E. Aubin)

Introduction

Osteoblasts oriyinate from rnesenchymal stem cells and through a series of cornmitment and

differentiation steps acquire the ability to synthesize the extracellular matrix of bone and regulate its

mineralization. The relatively late stage osteoblastic ceIl types, the preosteoblasts, osteoblasts, lining

cells, and osteocytes, are recognized and characterized by morphological and histochemical criteria (for

review, see Aubin et al.. 1993; Aubin and Liu, 1996). In addition. immunocytochemical and

irnmunohistochemical markers, combined with moIecular analysis of mRNA expression levels, have

provided an increasingly detailed understanding of differentiation-maturation events that appear to

fol Iow an ordered temporal sequence: acquisition and increase in ALP activity; increase in COLL-1

synthesis; and acquisition of expression of several non-collagenous bone matrix proteins, e.g., OPN. BSP

and OCN, deposition and subsequent mineralization of these matrix components to form bone (for

review. see Aubin et al.. 1993; Stein and Lian, 1995; Aubin and Liu, 1996). However, recently emerging

evidence has suggested that there is heterogeneity of bone at the tissue level as well as amongst mature

osteoblasts. For exarnple, differences in amounts of extractable bone proteins including OCN have been

reported in trabecular versus cortical bone (Ninomiya et al., 1990). Relative levels of OCN and COLL-1

mRNA Vary as a function of the secretor). lifetime of osteoblasts in vivo, e.g., OCN is undetectable by in

situ hybridization in newly differentiated osteoblasts, but detectable in mature osteoblasts. and is higher

in older flat osteoblasts than cuboidal osteoblasts (Heersche et al.. 1992). Ikeda et al. (1 995b) coutd not

detect OPN in osteoblasts at formation sites in the distal portion of rat femur. and expression of COLL-1,

OCN, and OPN markedly decreased at this site in old versus younger rats. In a recent

immunocytochemica1 study of bone formation in rat bone rnarrow stroma1 ce11 cultures (Malaval et al.,

1994), we found that even adjacent, morphologically indistinguishable cuboidal osteoblasts in bone

nodules stained with strikingly different intensities with antibodies against two different molecules: one

against OCN, and one against the IgE and laminin-binding protein (EBP or galectin 3) (Aubin et al.,

1996a). Other recent descriptions of expression levels for mRNAs of osteoblast-associated markers by in

situ hybridization of bone nodules also support the idea that heterogeneity of expression is associated not

only with differentiation stage but amongst individual cells at the same differentiation stage (Beilows et

ai., 1995; Pockwinse et al., 1995).

In this report, we analyze further the extent of heterogeneity of phenotype amongst mature

osteogenic cells in bone-derived primary cultures. We used low density cultures in which discrete

isoiated bone colonies f o m and are easily distinguis hable from fibroblastic colonies (Chapter 2).

Imrnunocytochernistry was done on osteoblasts in mature bone colonies with a battery of antibodies

against osteoblast-associated proteins including anti-COLL-1, anti-ALP, anti-OPN, anti-BSP, and anti-

OCN, in cornbinations of single or double labels. In addition, we used poly(A)-PCR and a panel of 3'

probes against these and other osteoblast markers including the El 1 and PTWPTHrP-R to measure the

simultaneous CO-expression profiles of these seven mRNAs in ouenty single osteoblasts. Both the protein

analysis and the molecular fingerprints revealed extensive heterogeneity in the repertoire of genes

expressed in clonally-derived, mature osteoblasts of equivalent in vitro lifetime, suggesting marked

plasticity in phenotype of mature osteoblasts.

Materials and Methods

Cell culrure

Cells were isolated by sequential digestion of 21 d fetal Wistar rat calvariae with a collagenase

enzyme mixture (Rao et al., 1977; Aubin et al., 1982; Bellows et al.. 1986a). Populations II-V were

pooled and plated in T-75 flasks containing a-MEM, 15% FBS, 100 &ml penicillin G, 50 @ml

gentamycin, and 0.3 pg/ml fungizone. After 24 h, cells were trypsinized (0.01% trypsin in citrate saline),

counted, replated at 10-35 cells/crn2 in 100 mm culture dishes, and cultured in medium as above but also

supplemented with 50 pg/ml ascorbic acid, 10 mM sodium pglycerophosphate, and 10 nM

dexamethasone (Bellows et al., 1987). These dishes were incubated at 37OC in a humidified atmosphere

of 95% air/5% COz.

Single ceIl isolation and cDNA prepurafion

Single isolated osteoblast colonies containing plump cuboidal cells with osteoid that was

mineralizing were marked in RC populations cultured as above and viewed under phase contrast

microscopy. Cells were rinsed with PBS, a cloning ring was placed around the marked colonies, and cells

frorn each colony were released with a 1 : 1 mixture of trypsin (0.0 1% trypsin in citrate saline) and

collagenase solution (Rao et al., 1977); enzymes were neutralized afier ce11 release by adding a-MEM

containing 15% FBS. Recovered ceIIs were resuspended in 0.1 % methyl cellulose/PBS and individually

rnicromanipulated into prechilled lysid1 st strand cDNA bufier (Brady et al., IWO; Brady and Iscove,

1993). cDNA was synttiesized by oligo(dT) priming, poly(A) tailed, and amplified by PCR using

oligo(dT) primer (Brady et al., 1990: Brady and Iscove, 1993; Chapter 2).

Hybridkat ion

Amplified cDNA (5 pl) was run on 1.5% agarose gels. transferred onto 0.2 prn pore size nylon

membrane (ICN. Costa Mesa, CA), and immobilized by baking at 80°C for 2 h. Prehybridizations,

hybridizations, and washes were performed as described in Chapter 2. After hybridization, the blots were

washed once for 30 min each at 42°C and 50°C in 2X SSC/O. 1% SDS, at 55°C and 60°C in 1X

SSC/O.I% SDS. and at 65°C in 0.5X SSC/O. 1% SDS. The blots were then exposed to Phosphorimager

screens (Molecular Dynarnics. Sunnyvale, CA) and digitized images obtained and quantified using the

IPLab Gel prograrn (Signal Analytics Corp., Vienna. VA). Labeled probes were used at an activity of 1 o6

cpm/ml. cDNA probes were labeled with ["PI~CTP using an oligolabeling kit (Pharmacia Uppsala,

Sweden). Total cDNA probe was prepared as described by Sarnbrook et al., (Sambrook et al., 1989) from

poly(A)+ mRNA isolated (Auffray and Rougeon. 1980) from mass populations of fetal RC cells grown

in the presence of dexamethasone in which bone nodules were beginning to mineralize. The rat cDNA

probes used were for COLL-1 (Genovese et al., 1984; provided by Dr. D. Rowe, Farmington, CT),

bone/tiverkidney ALP (Noda et al., 1987; provided by Dr. G.A. Rodan, West Point, PA), OPN (provided

by Dr. R. Mukherjee, Montreal, PQ), BSP and OCN (prepared by A.K. Gupta and J.E. Aubin, by

generating specific probes by PCR, screening osteoblast libraries and then confirming cDNAs by

sequencing; see also Chapter 2), E 1 1 (Wettenvald et al., 1996; provided by A. Wetterwald and H.

Fleisch, Basel, Switzerland and M. Atkinson, Munich, Germany), and PTH/PTHrP-R (Abou-Samra et

al., 1992; provided by Dr. G.V. Segre, Boston, MA). For optimal detection of PCR amplified cDNA, the

cDNA probes required sequences at or close to the extreme 3' ends of the native transcripts. The 3' cDNA

probes for COLL-1, ALP, OPN, BSP, and OCN were generated as described previously (Chapter 2). E l 1

was a 400 bp cDNA fragment obtained by digesting the fù l l length cDNA with StuI-XbaI to obtain the

extreme 3' end of the clone; and PTWPTHrP-R was an 800 bp cDNA HindII-Ba1 fragment.

Immunocytochernist~

Irnmunocytochemistry was done essentially as described previously (Chapter 2). Briefly, dishes

were rinsed with PBS; cells were fixed with 3.7% formaldehyde/PBS and permeabilized with methanol.

Mature osteoblast colonies were surrounded by a ring o f wax, and antibody solutions and rinsing buffer

were added and removed over each colony. Antibodies used were against pig COLL-1 (Connor et al.,

1983), rat ALP (RBM 2I 1.13; Turksen and Aubin, 199 1 ; Turksen et al., 1992), and rat OCN (kindly

provided by Dominique Modrowski, Paris, France; see also, Malaval et al., 1994). The monoclonal

antibody MPIIIB 1 O,, recognizing rat OPN, was obtained from the Developmental Studies Hybridoma

Bank maintained at the Department of Pharmacology and Molecular Sciences, Johns Hopkins University

School of Medicine, Baltimore, MD, USA, and the Department of Biology, University of Iowa, Iowa

City. IA, USA, under contract NO 1 -HD-6-29 15 from NICHD. The antiserum directed against BSP was

obtained from rabbits immunized with a synthetic peptide from the published sequence of mouse BSP

(CYDNENGEPRGDTYRAYED) coupled to keyhole Limpet's hemocyanin (Raynai et al., 1996).

HD33258 (Hoescht Ltd., Montreal, PQ, Canada) was used at a final concentration of 1 mg/rnl to label

chromatin in cells of interest.

Results

As described previously (Bellows and Aubin, 1989; Chapter 2), osteoprogenitor cells present in

primary cultures of fetal RC cells, plated at low density and grown in appropriately supplemented

medium, proliferate and differentiate to fom discrete colonies of osteoblasts foming bone. Previously.

we noted that expression levels of the bone matrix proteins determined by immunocytochemistry or by

cDNA Southern analysis were reflective of the differentiation stage of the cells (Brady et al., 1990;

Chapter 2). However, marked variations were also noted in cells in close physical proximity and judged

by morphological criteria to have reached the mature osteoblast stage. To confirm that mature osteoblasts

could express non-identical phenotypes and assess the extent of intercellular variation, we analyzed

numerous mature osteoblast colonies. i.e., ones in which cells were cuboidal in shape and surrounded by

refractile rnatrix and matrix mineralization was well progressed (Fig. 3.1).

By double-label immunocytochemistry, mature cuboidal osteoblasts were found to express

diverse levels of al1 of the osteoblast-associated proteins analyzed. COLL-1 staining was mainly

cytoplasmic, localized in the Golgi region and varied somewhat in intensity from cell-to-ceIl (Fig. 3.2A);

the sarne cells double-labeled for OPN showed marked differences in staining intensity even in adjacent,

morpho logical ly cornparab te ce1 1s (Fig. 3 2B). Sim ilar variation was seen with every other combination

of antibodies used, Le., ALP and BSP (Fig. 3.2C-D), and ALP and OCN (Fig. 3.2E-F). For example, the

ce11 surfaces of al1 cells of mature osteoblast colonies labeled for ALP (Fig. 3.2C, E); the same cells

double-labeled for BSP (Fig. 3.2D) and OCN (Fig. 3.2F), respectively, showed considerable

heterogeneity in staining intensity from cell-to-cell. The variation included not only whether a particular

marker was detected over background or not, but also level of expression (labeling intensities) for al1

markers. In zones of the colonies in which mineralization was very well advanced, OPN, BSP, and OCN

labeling intensities were ofien reduced in al1 or most cells.

With immunocytochemistry, we were limited to double-labeling analysis of only two markers at

a tirne. To address further the extent of heterogeneous marker expression, we wished to analyze a11

available markers simultaneously in individual osteoblasts. Individual mature osteoblasts ( 1 0 fiom each

Figure 3.1 Photomicrograph of a typical mature osteoblastic colony or single bone nodule. Note the

cuboidal osteoblast cells and the mineralized osteoid in the center zone of the colony. Bar: 285 Fm.

Figure 3.2 Double immunolabeling of mature osteoblastic colonies for: (A) COLL-1 and (B) OPN; (C)

ALP and (D) BSP; (E) ALP and (F) OCN; and ( G ) HD33258 and (H) OCN. Bar: (A, B, E, F) 25 Pm; (C,

D. G, H) 20 Pm.

of two mineralited bone nodules/colon ies) were isolated by trypsin ization and micromanipulation, and a

random amplification poly(A)-PCR performed on the mRNA from each single cell. The amplified cDNA

from the individual cells was then probed to determine whether each ceII expressed or not COLL-1, ALP,

OPN, BSP, OCN. E 1 1 and PTH/PTHrP-R messages; a total cDNA probe was used as control on the

amplification procedure and al1 other signals were compared and quantified against it (Fig. 3.3). When

individual cells were compared for overall pattern or repertoire of mRNAs expressed, marked difierences

were evident (Figs. 3.3,3.4). As expected based on the immunocytochemistry (Fig. 3.2), al1 individual

cells for which good amplification was achieved (i.e., al1 except cell 16 from colony 2) expressed COLL-

1, but levels varied from very high (maximal: 100%) to intermediate (080% of maximal) to low (20%

of rnzuimal) (Figs, 3.3,3.4). Levels of other osteoblast markers varied over even larger extremes, i.e.,

frorn essentially background to very high (Figs. 3.3, 3.4). Given that OCN is the most osteoblast-specific

marker of those available, and is the latest stage marker of the mature differentiated osteoblast phenotype

being expressed afier proliferation ceases (Owen et aI., 1990: Owen et al., 199 1 ; Pockwinse et al., 1992:

Malaval et al., 1994, 1996), we asked which mRNAs are CO-expressed in those mature cells expressing

OCN. Diverse CO-expression profiles were seen. For exampie, some OCN-high ceIIs also concomitantly

expressed the mRNAs for COLL-1, ALP, OPN, and BSP in moderate to high levels (e.g., ceIl 7), whereas

others in which OCN was high expressed low levels of at least one or several of these ( e g , cells 10, 1 1

and 20). Altematively, in some cells in which al1 other messages were moderately to highly expressed,

OCN was weakly detectable or undetectable (e.g.. cells 3, 5, 8,9, and 13). Intermediate levels of OCN

were also associated with diverse expression levels of other markers. E 1 1 is a marker that has been

shown to be expressed by cells in transition frorn osteoblast to osteocyte, i.e., a subset of mature

osteoblasts and newly formed osteocytes (Wetterwald et al., 1992; Wetterwald et al., 1996). Some cells

expressed OCN and no detectable El 1 messages (e.g., cells 7 and 1 1) or vice versa (e-g., cells 3, 13, and

1 8), or both messages (e-g., cells 1, 14, and 17). Arnongst these, other osteoblast markers again appeared

to Vary in a random fashion.

Colony 1 Colony 2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

ALP

OPN

OCN

Figure 3.3 Phosphorimage cDNA expression profiles of ten single ceils fiom each of two mature

osteoblastic colonies. The mRNA from each ce11 was reverse transcribed and mplified by the poly(A)-

PCR technique. The resutting cDNA from each cell was then probed for the expression of total cDNA

and the individual messages from various bone-associated proteins. Each vertical lane is the cDNA From

the same individual ce11 and each horizontal row has been exposed for the same period of time.

Figure 3.4 Normatized cDNA expression profiles of the 20 single cells characterized in Figure 3.3 and

illustrated as percent maximal expression. Phosphorimage signal intensities for each probe were

quantitated and standardized against total cDNA. Note that ce11 16 was not included in the Results

becayse the sample did not ampli@ well as shown in Figure 3.3.

ALP

OPN

BSP

OCN

E i l

The apparently mdom expression profiles of the seven markers analyzed suggested that the

variations observed could not be ceIl cycle related. However, to address this more explicitly, a mature

osteoblast colony was double-labeled for HD33258 (a fluorescent DNA stain for chromatin) (Fig. 3.2G)

and the post-proliferative marker OCN (Stein and Lian, 1995: Malaval et al., 1996) (Fig. 3.2H). Clearly,

remarkable variation in staining intensity for OCN fiom undetectable to intermediate to highly intense

was observed even in adjacent morphologically indistinguishable cuboidal osteoblasts with similar

interphase staining with HD332.58.

Discussion

In this study. we analyzed protein and mRNA expression levels for osteoblast-associated

markers in individuai cells cfassed as mature osteoblasts in colonies making osteoid that was

mineralizing. Since the fetal RC cells were plated at very low densities, each isolated colony that foned

was derived from a single ce11 that had attached and proliferated and. thus. the cells of the mature

osteoblast colonies were of clonai origin (Bellows and Aubin, 1989: Chapter 2). Brady and coworken

( 1990. 1993) validated the poly(A)-PCR technique for maintenance of relative abundance information of

multiple mRNAs in single cells and we have shown the utility of the approach for analyzing expressed

gene repertoires in osteoblasts (Chapter 2). By cornbining this PCR technique to determine the repertoire

of seven osteoblast-associated mRNAs expressed concom itantly in individual ce! 1s with analysis of

protein expression (detectable or not. and levels of expression) detected by immunocytochemistry, we

have found strikingly diverse phenotypes amongst mature osteoblasts.

While both the poly(A)-PCR and immunocytochernistry document marked heterogeneity in

phenotype expression in individual mature osteoblasts, the spatial information on ce11 location within

colonies is lost by the trypsinization procedure required to collect cells for the PCR analysis. However,

the immunolabeling retains the spatial information about ce11 location; thus, whether a cet1 is surrounded

by other cuboidal cells and is adjacent to mineralized osteoid, or is at a colony periphery, is clearly

evident. The immunolabeling demonstrated that even adjacent morphologically indistinguishable

cuboidal osteobIasts in the zone of mineralized matrix have differences in protein expression, especially

in relation to OPN. BSP, and OCN expression. Initially, we speculated that at least some of the

differences in protein expression between individual cells might refiect stage of ce11 cycle (Chapter 2).

However, OCN is an osteoblast marker expressed only afier osteoblasts have ceased proliferative activity

(Stein and L i a , 1995: Malaval et al., 1996). When we doubIe-labeled ceIls for Hl333258 and OCN, it

was clear that adjacent interphase (post-proliferative) cells Vary rnarkedly in OCN staining intensity and

the variable expression levels appear to have no direct relationship to ceIl cycle, a finding consistent with

other detailed analyses combining double-label imrnunocytochemistry and an observation we believe

extends to other markets as well (Malaval et al., 1996), and with other recent studies in other ceIl

1 ineages (see below).

While immunocytochernistry has the advantage of maintaining spatial information with regard to

protein expression, it is limited technically in the number of fiuorochrornes which can be detected

sirnultaneously within a single cell. The method of poly(A)-PCR has the advantage of allowing

detemination of CO-expression profiles for many messages simultaneously, while maintaining the

relative abundance relationships, in samples as srnaIl as a single ceIl (Brady et al.. 1990). Using this

technique to ampli@ al1 the messages present in hventy individual cells from two different mature

osteoblast colonies, we were able to extend analysis of differences in mRNA expression of various bone-

related markers to many messages in single mature osteoblasts. Clearly. heterogeneity in the repenoire of

genes expressed exists among single mature osteoblasts at the rnRNA level for the seven osteoblast

markers analyzed (Fig. 3.4). Immunocytochemistry documented that OCN labeling intensity decreases

overall in the central area of the nodules as mineralization becomes well-progressed (see also Chapter 2;

Malaval et al., 1996). At the mRNA level, poly(A)-PCR results also indicated that there exist single

mature osteoblasts that express very low levels of OCN, suggesting that some of these cells might be a

terminal differentiation or osteocyte stage at which OCN has been down-regulated. However, this cannot

account for al1 of the heterogeneity observed because immunocytochemistry showed that even adjacent

cells in other portions of the nodule are different.

Since nodules are complex three-dimensional sûuctures in which cell focalization can be

approximately but not completely determined by differential focusing at the light microscope, we sought

to confirm whether we could distinguish between mature cuboidal osteoblasts on osteoid and osteocytes

embedded within mineralizing osteoid. To do this, we looked at the CO-expression of E 1 1 mRNA along

with the other bone-related markers, particularly O W . The El 1 antibody was descrikd to label cells in

transition from osteoblast to osteocyte (Wetterwald et al., 1992. 1996). and it was suggested that El 1 is a

later stage marker for osteobhsts differentiating into osteocytes than OCN which seems to be a marker of

osteoblast maturation. Le.. immunolabeling showed E 1 1 protein to be highly expressed on the ceIl

membrane of newly forrned osteocytes. and less intensely on mature osteoblasts (Wetterwald et al.. 1992.

1996). E 1 1 cDNA was isolated recently from an ROS 17!3.8 expression library (Wetterwald et al..

1996): with the 3' probe we prepared from this cDNA. we were able to distinguish three types of mature

osteoblast ceIls: E 1 1 Iow/OCN hi&. E l 1 high/OCN hi&. and E 1 1 high/OCN low. In accordance with

El I king a later marker than OCN for mature oneoblasts (Wetterwald et al.. 1992). the first ceIl type

would be designated mature osteoblasts. the third ceIl type newly formed osteocytes. and the second cell

type an intermediate stage or cell in transition from osteobtast so osteoc>.te. Taking into account these

developmental stages d l 1 cannot account for the extensive variation observed in the other osteoblast

markers (Fig. 3.3).

Thus. ue are lefi to conciude that marked heterogeneity exists in both protein and mRNA

expression levels of various osteoblast-related markers. especially OPN. BSP. OCN. and E 1 1 in mature

osteoblasts. Therefore. the mature osteoblast phenotlpe shoufd not be regarded as a single unique

phenotype but as one that encornpasses a plasticity or flexibility in panern of rnRhiA and protein

expression of markers. and some of this variation may be nochastic in nature. Heterogeneity amon@

single cells in clona1 cell lines or subclones derived fiom hem has k e n described. including for

osteoblastic cell lines [e.g.. ROS 17/2 (Grigoriadis et al.. I985)] amongt others reg., pituiiary ceII lines

(Boockfor et al.. 1985: Kinernan and Frawley. 1994jl. However. the differences arnongst cells in these

clonal iines is often attributed to there king aberrant expression from transformation and/or long term

culturing with resultant genomic alterations. Heterogeneity in single ceIl expression profiles was reporteci

amongst morphologically similar single live neurons fiom hippocampus, with a technique with similar

features to the PCR approach we have iised here (Ebenvine et al., 1985). Ciearly, like in the latter study,

what we observe amongst mature OBs in primary culture is representative of expression at one time, the

time of sarnpling. As we raised above, although further systematic ce11 cycle studies may be informative,

it appears untikely that ce11 cycIe differences can account for the differences observed in post-

proliferative neuronal ceil expression profiles or in the patterns we have observed here, since we have

included analyses of markers expressed only in post-proiiferative osteoblastic cells. Sim ilar conclusionc

were reached afier an extensive analysis of hemopoietic ceils of different developmental fates and

subjected to the sarne technique of global amplification of m k l A transct-ipts as we used here (Brady et

al., 1995). In the latter analysis. sorne sibling cell pairs were positioned within minutes cf each other in

the cet1 cycle and displayed heterogeneous profiles: others separated in tirne from late G2 to eariy G 1 did

not display corresponding correlations with expression. Brady and coworkers (1995) concluded that the

marked variation in transcript abundance might be an inherent aspect of the transcription process.

Whether cells cycle through the different repenoires observed in some regimented fashion or the levels

reflect a stochastic process is not 'et known. but a stochastic element to the timing of transcriptional

activation is likeiy (Michaelson. 1993: Ross et al.. 1994: Brady et al.. 1995). The biological significance

or consequence of plasticity in mRNA and protein repertoire expression is not yet known. but it suppom

the notion. for example. that not al1 mature OBs may be functionally identical. Since the heterogeneip

exxends to at least one hormone receptor, PTWPTHrP-R it points to the possibility that some regulatory

molecules. i.e.. PTH and PTHrP. also rnay elicit their effects on only a subset of cells at any one time.

These data suggest that the most unarnbiguous marlier of the mature OB remains its abiliry to contribute

with iw siblings to the production of bone rather than in an absolute level of production of particula.

macromolecules at al1 points in time.

Molecular Fingerprinting of Primitive Osteoprogenitor Ce& undergoing

Differentiution in vitro

This chapter has been submitted for publication with the above title (Fina Liu and Jane E. Aubin)

67

Introduction

According to current understanding, multipotential mesenchymal stem cells give rise to difierent

lineages, one of which is the osteogenic lineage (for reviews, see Aubin et al., 1993; Aubin and Liu,

1996). Osteogenic cells originate from committed osteoprogenitors that proliferate and differentiate into

cells which by morphologica1, histochemical, immunohistochemical and rnolecular criteria are

unam biguously identifiable as preosteobIasts, mature osteoblasts, quiescent bone lining ceils, and

terminally-differentiated osteocytes. To identiQ more primitive cells in the lineage has been difficult

because their morphological characteristics are not known to be different from fibroblasts, unique

rnolecular markers are not yet known, and they are relatively rare compared to their more differentiated

progeny, Some clonai osteoblastic ce11 lines have been isolated whose phenotypes are consistent with

their being relatively immature, but whether any of these faithfully represent authentic osteoprogenitors

is uncertain (for review, see Rodan and Noda, 199 1 ; Aubin et al., 1993). Further, while they have been

useful rnodels in which to investigate some aspects of regulation of osteoblast maturation and osteoblast

macromo lecular products, their apparently aberrant expression of at least some markers has suggested

that other models are required for detailed characterization of the molecular nature of the more primitive

osteoprogenitors. To this latter end, we devised an alternate strategy which relies on use of primary

cultures of fetal RC ceIl populations. Freshly isolated RC populations contain osteoprogenitors that can

divide and differentiate in vitro to f o m bone nodules (Nefussi et al., 1985; Bellows et al., 1986a),

although their nurnber is low, Le., 4% of an RC population under standard culture conditions (Beltows

and Aubin. 1989). In this system, the osteoprogenitors present are morphologically indistinguishable

from other pleiomorphic or tibmblastic cells making up the majority of the population, but they can be

identified indirectly or retrospectively by their ability to give rise to more differentiated progeny, up to

and including mature osteoblasts forming the easily identifiable colony type of a mineralized bone

nodule (multilayer of cuboidai cells with abundant osteoid and deposited hydroxyapatite; Nefussi et a[.,

1985; Bel lows et al., 1 986a; B hargava et al., 1988). Expression levels of known osteoblast-associated

markers, i.e., COLL-1, ALP, OPN, BSP, and OCN, are increased in a well-established temporal sequence

as osteoblasts develop and bone foms in this model (Owen et al., 1990, 199 1; Pockwinse et al., 1992;

Malaval et al., 1994; Chapter 2). Briefly, by immunocytochemistry, Northem or cDNA Southern

analysis, the differentiation sequence is defined by: increase in COLL-I synthesis; acquisition and

increase in ALP activity; and acquisition of expression of several non-collagenous bone matrix proteins,

e.g., OPN, followed by BSP and finally OCN as the latest marker of the mature osteoblast (for reviews,

see Rodan and Noda, 1991; Aubin et al., 1993; Stein and Lian, 1995; Aubin and Liu, 1996).

Earlier, while investigating changes in levels of expression of the specific osteoblast-related

rnacromoIecules in RC colonies classified morphologicalIy as comprising early or more mature

osteoblasts. compared to colonies designated "fibroblastic", we found evidence to suggest that

developmental stages more primitive than those recognizable by the presence of cuboidal cells and prior

to up-regulation of any of the known osteoblast markers (Le., COLL-1, ALP, OPN, BSP, and OCN) were

accessible for analysis in this model (Chapter 2). Our solution to the problem of identiQing definitiveiy

the low frequency primitive osteoprogenitors for further molecular characterization was to use the

technique of replica plating on dishes plated at low density and sampled early in the development of

colonies. Replica plating has been found to allow screening of large numbers of individual mammatian

ce11 clones for the phenotype of interest while still rnaintaining a master copy of the colonies: use of

polyester cloth, in particular, has been found to provide high fidelity copies for a variety of ceIl types

(Raetz et al., 1982; Esko, 1989). The replica technique allowed us to identiSr the less than 1%

osteoprogenitors present on the master plates, which then served as the basis of samples for molecular

analysis. Since the ceIl number was limited in each colony sampled, we applied poly(A)-PCR (Brady et

al., IWO; Brady and Iscove, 1993) to detennine simultaneous expression profiles of the mRNAs of

interest. These included both the osteoblast-related mRNAs and mRNAs for potential regulatory

molecules, i.e., cytokine receptor messages (Le., FGF-R 1 and PDGF-Ra). We now report that this

approach allows rnolecular fingerprinting of definitive primitive osteoprogenitors and characterize novel

transient developmental stages of such cells as they progress through their differentiation sequence.

Materials and Methods

Ce Il culture

Cells were isolated by sequential digestion of 2 1 d Wistar rat calvariae with a collagenase

enzyme mixture (Rao et al., 1977; Aubin et al., 1982; Bellows et a!., 1986a). Populations II-V were

pooled and plated in T-75 flasks containing a-MEM, 15% FBS, IO0 pg/ml penicillin G, 50 pg/ml

gentamycin, and 0.3 pg/ml fungizone, and incubated at 37°C in a humidified atrnosphere of 95% air and

5% CO?. After 24 hl cells were trypsinized (0.01% trypsin in citrate saline), counted, replated at 10-25

cellslcrn' per 100 mm culture dish, and cultured in standard medium described above supplernented with

50 m g h l ascorbic acid. 10 mM sodium pglycerophosphate, and 10 nM dexamethasone (Bellows et al.,

1987; Chapter 2).

Replica plat ing

Replica plating (Fig. 4.1) was done essentially as described by Raetz et al. ( 1 982) and Esko

( 1 989) but with minor modifications. Briefly, a disc of 1 prn pore size polyester cloth (HD7- 1 ; B&SH

Thompson, Scarborough, ON) was fioated above the cells and weighted down by a monolayer of 4 mm

glass beads; replica cloths were placed on cells at day 1 or day 4 and were removed on day 5 or day 1 1,

respectively, and transferred into a new dish. The master dish and polyester disc were each rinsed with

PBS and fed with supplemented medium as above. The replica disc was incubated at 37°C; whereas, the

master dish was incubated at either 25°C or 30°C to stall the proliferative and differentiation activities of

colonies. The medium was changed every 2-3 days.

In prelirninary experirnents to determine the transfer efficiency and fidelity of transfer for

colonies of fetal RC cells, replica cloths were fixed on day 25 with 10% neutral buffered formalin and

stained with the von Kossa technique (Bellows et al., 1986a). Also on day 25, master dishes were

transferred from the lower temperature incubators to a 37OC incubator. After 14 days at 37OC, these

master dishes were fixed and stained for the presence of bone nodules as described above. For statistical

analysis (Fig. 4.2), the data were expressed as means and standard deviations.

Individuaf colonies

Master dish

Master dish Dish with replica cloth

Figure 4.1 Schematic of the methodology for the replica plating technique to identifi osteoprogenitor

(OP) and preosteoblast colonies by retrospective analysis.

Osteo b lasric colony isolation and cDNA preparat ion

Afier two weeks at 37OC, the replica cloth was fixed in 10% neutral buffered formalin and

stained with the von Kossa technique to identify bone nodules. The master dish was transferred to 37OC

for 5-9 h. After this time, the replica disc was matched up with the master dish to LocaIize primitive

osteoblast colonies, which were then rnarked. In addition. single isolated osteoblast colonies containing

plump cuboidal cells with unmineralized osteoid or mature osteoblast colonies containing mineralizing

osteoid were marked and collected fiom dishes that were not cultured with polyester discs. Dishes were

rinsed with PBS and a cloning ring was placed around the marked colonies. The cells from each colony

were released with 0.0 1% trypsin (when matrix lacked mineral) or a 1 : 1 mixture of 0.0 1 % trypsin and

collagenase (Rao et al., 1977) (when osteoid was mineralizing), and the enzyme(s) neutralized after ceIl

release by adding a-MEM containing 15% FBS. Total RNA was extracted using a mini-guanidine

thiocyanate method as described previously (Brady and Iscove, 1993; Chapter 2). cDNA was synthesized

by oligo(dT) priming, poly(A)-tailed, and amplified by PCR with oIigo(dT) primer (Brady et al., I W O ;

Brady and Iscove, 1993; Chapter 2). The 1 O8 colonies reported here were collected from 12 independent

ceIl isolations and replica plating experirnents.

Southern blors and hybridization

Amplified cDNA (5 pl) was run on 1.5% agarose gels, transferred ont0 0.2 prn pore size nylon

membrane (ICN. Costa Mesa, CA). and immobilized by baking at 80°C for 2 h. Prehybridization and

hybridization were performed as described in Liu et al. (1994). After hybridization, the blots were

washed at 65°C for I h each in 2X SSC/O.lOh SDS and in 0.5X SSCIO. 1% SDS. The blots were then

exposed to phosphorimager screens (Molecular Dynamics, Sunnyvale, CA), and digitized images

obtained and quantified with the IPLab Gel program (Signal Analytics Corp., Vienna, VA). Afier

quantiQing the data, the signal intensity for each probe was standardized against total cDNA (see

below). For preparation of comparative histograms of relative expression profiles, we next determined

maximal expression value for each message; the largest value was divided by 5 to obtain five ranges of

values or categories. Sarnples were then given a rank of 1,2,3,4, or 5 depending on where their values

fell within each range or were given a rank of O if the intensity of signal was not detectable. For

statistical analysis of relative expression levels (Fig. 4.9 , the actual standardized expression levels for

each probe in each sample were used to calculate means and standard deviations within categories of

populations. Levels of statistical significance were calculated by the Welch t-test.

Labeled probes were used at an activity of 106 cpm/ml. cDNA probes were labeled with

[ 3 ' ~ ] d ~ ~ ~ using an oligolabeling kit (Phannacia, Uppsala, Sweden). Total cDNA probe was prepared as

described by Sambrook et al. (1989) from poly(A)+ mRNA isolated (Auffray and Rougeon, 1980) from

mass populations of fetal RC cells grown in the presence of dexamethasone in which bone nodules were

beginning to mineralize. The rat cDNA probes used were for COLL-1 (Genovese et al.. 1984; gifi of Dr.

D. Rowe, Farmington, CT), bone/liver/kidney ALP (Noda et al., 1987; gift of Dr. G.A. Rodan, West

Point, PA), OPN (gifi of Dr. R. Mukherjee, Montreal, PQ), BSP and OCN (prepared by generating

specific probes by PCR, screening an osteoblast library (Aubin et al.. 1996b) and then confiming

isolateci cDNAs by sequencing; see also Chapter 2). For optimal detection of PCR amplified cDNA, the

cDNA probes required sequences at or close to the extreme 3' ends of the native transcripts. Labeled

probes for the cDNAs listed above were generated as described previously (Chapter 2). Rat PDGF-Ra

(Lee et al., 1990; gift of Dr. R.R. Reed, Baltimore. MD) was a 400 bp cDNA Pst1 fragment obtained by

digesting full length cDNA with HindIlI to remove 6 kb of the 5' region and religating the sticky ends.

Mouse FGF-RI (Mansukhani et al., 1990: gifi of Dr. C. Basilico: New York. NY) was a 400 bp cDNA

HincI 1- PsA fragment.

Results

Replica Piuting

To our knowledge, the replica technique as applied to mammaiian cells has been limited to use

with rapidly growing, and often transformed, ceil lines. To validate it for identification of

osteoprogenitor cells in primary RC populations, we first did a quantitative analysis of colony formation

on rnaster dishes and replicas. We compared three main parameters: total population plating efficiencies

and bone colony plating eficiencies on RC cells plated at low density; the efficiency of faittifid colony

replication and bone nodule formation when replica cloths were added to cultures on day lhernoved on

day 5 or added on day 4/removed on day 1 1; and the best low temperature for stalling colony growth on

master dishes whik maintaining osteoprogenitor viability, relative immaturity, and subsequent recovery

with ability to differentiate intact. Bone and non-bone colonies were identified by well-established

rnorphologicaI characteristics and confirmed by immunocytochemical and molecular characteristics as

shown elsewhere (see Chapter 2). Consistent with our eartier limiting dilution analyses (Bellows and

Aubin, 1989) and other low density plating experiments (Chapter 2), total population plating

efficiency/colony forming efficiency of primary RC populations was approximately 1 O%, and of these

the fraction of colonies that were bone was 1 O%, i.e., approximately 1 % of the starting ceIl population at

37°C (Fig. 4.2). Control experiments in which both replica cloth and master dish were maintained at

37°C confirmed that replication of both bone and non-bone colonies worked with essentially 100%

efficiency in these low density cultures on both master and replica. Stall temperatures of 4°C or lower

gave extremely poor cell recovery from the master dish so this was not used further.

Replica cloths placed over colonies on day 4 (many cells entering log phase growth) and

removed a week tater on day 1 1 allowed sufficient numbers of cells to attach to cloth and master dish to

achieve approximately a 95% transfer eficiency when the master dish was stalled at 30°C (Fig. 4.2). In

control experiments, we found that a few colonies did not replicate faithfully either because the whote

colony transferred to the cloth or little or none of the colony transferred. Considering that such high

transfer effkiencies were attained, we next detennined whether it was possible to isolate

osteoprogenitors at even earlier stages by either stalling the master dishes at 25°C to maintain colonies at

even lower growth rates or by placing replicas for shorter periods, i.e., placing them over colonies at day

1 after plating (cells adhered, some dividing) and removing them at day 5 (during log phase growth)

followed by incubation at either 30°C or 25°C. These latter manipulations frequently led to considerabIy

lower replication efficiencies. Maintenance of master dishes at stall temperatures of 25°C led to

30°C 25°C 30°C 25°C 37°C

day 4/day 1 1 day l/day 5

Figure 4.2 Plating eficiencies/colony foming efficiencies of bone and non-bone colonies in controls

(Co), master dishes (M) and replica cloths (R) under different replication conditions. Bars are means and

standard deviations of colony counts from a range of 4-34 100 mm dishes depending on the condition.

The days indicated correspond to when the replica cloths were placed ovedremoved from the colonies in

master dishes as described in Materials and Methods.

approximately a 50% recovery in faithfully replicated bone and non-bone colonies (Fig. 4.2), mainly due

to poor recovery of cells on the rnaster dish at this temperature. When replica cioths were placed on day

1 and removed on day 5, we also achieved only a 50% or lower transfer and replication eficiency, even

when the 30°C stall temperature was used for the master dish (Fig. 4.2). Visual inspection of dishes

treated in this manner indicated that cells in many colonies in this first five days of growth divided

relatively slowiy as compared to an accelerated rate between days 4 and 7. consistent with the notion that

at least some progenitor populations (Potten et al., 1979) including osteoprogenitors (Bellows et al..

1990a; Friedenstein, 1990) may be relatively quiescent at early times in vitro. In these cases. few cells

were present on master dish or replica cloth, leading to overall lower colony efficiencies on replica and

master dish (Fig. 4.2).

By using optima1 conditions and selecting for fingerprinting only cotonies in which cells had a

robust appearance. and ones discrete/well-separated from and so not contaminated by cells from other

colony types, we were able to isolate over 80 definitive osteoprogenitor colonies representing a spectrum

of differentiation stages as we outline below. Visual inspection indicated that during the 14 days required

for differentiatiodbone colony formation to occur on the replica cloths. on average the osteoprogenitors '

on the stalled master dishes were relatively quiescent and had undergone only 0-1 additional population

doublings; a few which had a senescent or dying appearance were discarded from the analysis. All

osteoprogenitor colonies selected for analysis comprised cells with a pleiomorphic morphology

indistinguishable from non-bone colonies on the same dishes. indicating that none had yet acquired the

cuboidal shape characteristic of differentiated osteoblasts. Prior to collecting cells for the molecular

fingerprinting, the master dishes were transferred back to 37°C from the lower stall temperature for

several hours. That celIs were well-recovered prior to mRNA isolation is indicated by the fact that they

were relatively synchronized mitotically after 5 h and 9 h from dishes originally stalled at 30°C and

25"C, respectively. Estimated colony size judged by microscopic viewing on a grid ranged from 50-3000

cells per colony, colony size did not directly correlate with any specific phenotypes observed by

fingerprinting.

Poly(A) - K R

Previously, we reported that the technique of poly(A)-PCR was a powerful and discriminating

tool to establish molecular profiles for osteoblastic colonies and single cells that had differentiated to

different extents as recognized morphologically (Le., early, intermediate, and mature osteoblast)

(Chapter 2). Because we wished to extend the fingerprinting beyond the osteoblast-associated probes

(i.e., COLL-1, ALP, OPN, BSPt and OCN) used previously, and to compare later (more mature)

deveiopmental stages with osteoprogenitor stages, we included in the present analysis a few colonies

already advanced to the stage of being morphologicalIy recognizable as osteoblasts. These (labeled OB

and Mature OB; 24 colonies total) were subjected to the same poly(A)-PCR manipulations and Southern

blotting as colonies identified by replica plating (Iabeled OP/Pre-OB; 84 colonies total) (Fig. 4.3). A total

cDNA probe was used as a control to assess first strand synthesis and amplification amongst the 108

colonies (Fig. 4.3); this proved relatively constant and its signal strength was used to standardize relative

abundance of all other messages for comparison amongst colonies. Of al1 messages probed. OPN was

most uniformly and abundantly present in virtually al1 colonies analyzed. FGF-RI and PDGF-Ra

mRNAs were also detectabIe in virtually al1 colony types but their abundance was lower and

heterogeneous from one colony to another (Fig. 4.3). In keeping with Our earlier analysis of co-

expression profiles for osteoblast-associated messages at later differentiation stages, most of the

osteoblast colonies expressed al1 of COLL-1, ALP, OPN, BSP, and OCN, but to various extents; a few

osteoprogenitor colonies also expressed al1 of these messages to some degree, but most expressed none

or only one. two or three messages to a limited level (Fig. 4.3).

Immature progenitor colonies or more mature osteoblastic colonies analyzed in the Southern blot

shown in Fig. 4.3 are in random order, reflective only of the order of processing. To determine whether

immature progenitors could be placed in a rank order of more primitive or less primitive cells, we

developed a paradigm for their comparison. Relative expression levels of al! messages in al1 colonies

sarnpled were determined by norrnalization of their signal strengths against that for total cDNA; as

described in the Materials and Methods section, they were then assigned categories of relative expression

from low (0-1) to high (5). This classification smoothed out the small but not the large variations in

OPIPre-OB OB Mature OB

Figure 4.3 Phosphorimage cDNA expression profiles of ostcoprogeiiitor/preosteoblast (OP/Pre-OB) coloiiics ideiitified by the replica plating techiiique,

and carly and mature osteoblast coloiiies identificd inorpliologicnlly by the cuboidal shupe of osteoblasts aiid deposiiioii of osteoid (OB) and

mineralization of osteoid (Mature OB), respectively. 'fhe inRNA froiii ench sainple was reverse tra~iscribed into cDNA and aniplified by the poly(A)-PCR

method. The resulting cDNA amplified froin eacli colony was theii probed for tlic expression of total cDNA (a coiitrol for the aniplification procedure),

the various bone-related proieiiis (COLL-1. ALP, OPN, DSP, alid OCN), aiid cytokiiie receptor messages (FGF-R 1 and PDGF-Ra) as specified. Eacli 1

vertical lane is the cDNA from the saine coloiiy. Tlie cxposure is the saine for al1 coloi~ies for each probe. Thcse colonies arc in rriiidom ordcr, rcflective

only of the order of processing.

expression levels such that overall patterns of gene expression became more obvious. Based on

Northems, immunùcytochemistry, and our earlier poly(A)-PCR results, it is known that ALP expression

rises as osteoblasts mature and then d e c h e s as osteoid becornes heavily mineralized (Bronckers et al.,

1987; Mark et al., l987b; Zernik et al., 1990; Owen et al., 1990, 199 1 ; Turksen and Aubin, 199 1 ;

Malaval et al., 1994: Chapter 2). Therefore, we rank-ordered the early osteobIast and mature osteoblast

colonies manually based on their expression levels of this marker. such that ALP increased in the early

osteoblast colonies and decreased in the mineralized colonies (Fig. 4.4): this imposed a profile for al1

other genes probed in those colonies which at least for the known osteoblast messages (i.e., COLL-1,

OPN, BSP, and OCN) "fit" with their known expression patterns (for reviews, see Rodan and Noda,

199 1 ; Aubin et al.. 1993; Stein and Lian, 1995; Aubin and Liu, 1996). Colonies selected by replica

plating were then also rank-ordered with respect to a variety of markers. Al1 replica colony samples

expressing OCN, the Iatest of welI-established osteoblast markers to be acquired as the cells mature,

were assigned positions to the mature end (extreme right; COLL-I'IALP'IOCN~), followed in order to

the left by colonies expressing fewer and fewer markers in the order COLL-I'/ALP' colonies, then

COLL-I'IALP- colonies, then COLL-1-/ALP-. based on current information of the maturational sequence

of osteoblastic cells undergoing differentiation (for reviews, see Rodan and Noda, 199 1 ; Aubin et al..

1993; Stein and Lian. 1995; Aubin and Liu, 1996). Among the COLL-I&/ALP- colonies, a few were

BSP*. S ince a few COLL-1-/ALP- colonies were also BsP', we gmuped al1 these BSP' colonies together

at the COLL-I'/ALP--COLL-1-!ALP- border. Colonies expressing no osteoblastic rnarken were placed

on the extreme left of the graph. Using categories established in Fig. 4.4, Le., COLL-1-IALP-, COLL-

~ ' I ~ L P - , COLL-I'/ALP+. and OBs (OB and Mature OB combined; COLL-I'/ALP~/BSP'/OCN'), we also

calculated the means and standard deviations of corrected (standardized against total cDNA) expression

levels for each probe within categories of populations and looked for statistically significant changes

during the osteogenic differentiation sequence (Fig. 4.5).

These analyses allowed us to discern whether patterns existed in the growth factor receptors

during the differentiation sequence and generated several other patterns of interest (Figs. 4.4,4.5).

Figure 4.4 Rank order profiles of normalized cDNA expression by the colonies characterized in Figure

4.3. Phosphorimage signal intensities for each probe were quantified and standardized against total

cDNA. Immature osteoprogenitor coIonies identified by replica plating that did not express detectable

levels of ALP were classed as immature or primitive osteoprogenitors (OP) compared to more mature

osteoprogenitors/preosteoblasts (Pre-OB) that did express low levels of ALP message; OBs and Mature

OBs are defined as in Figure 4.3. All colonies were rank-ordered according to the paradigm outlined in

Results.

ALP

OPN

BSP

Figure 4.5 The means and standard deviations (bars on the left) of corrected (standardized against total

cDNA) expression levels for each probe within categories of populations were calculated and subjected

to the Welch t-test for statistically significant changes as cells differentiate (table on the right).

Categories were defined as COLL-17ALP- colonies, COLL-I'/ALP- colonies, COLL-I'/ALP' colonies,

and OB colonies (COLL-I+/ALP'/BSP'/OCN~). (nt, not testable)

COLL-I'IALP- COLL-I'IALP' 0 8 s COLL-TIALP' nt nt nt

ALP

OU) - 020 - 010 - 000

COLL-I'IALP* p0.05 ~ 0 . 0 5 COLL-I'IALP' p0.05

COLL-I'IALP' COLL-I'IALP' 0 8 s COLL-1-IALP- ~ ' 0 . 0 5 ~ ~ 0 . 0 5 p<O.OOOl COLL-I'IALP- v0.05 p<O.OOOl COLL-I'IALP' vo.001

O 70 OCN

O60 - 050 - COLL-I'IALP- COLL-I'IALP' OBS

COLL-I'IALP- nt nt nt

COLL-I'IALP- COLL-I'IALP* OBS COLL-I'IALP' pw0.05 p>O.OS v0.01

First, there is a recognizable cohort of cells captured by the replica method and considered very primitive

(lefi; no ostecblast-specific mRNAs expressed); these are not quiescent since they express low-

intermediate levels of rnRNAs for OPN, FGF-R 1 and PDGF-Ra. Second, a group of relatively primitive

(expressing no other osteoblast-associated mRNAs) progenitors transiently expressing BSP emerges as a

distinct developmental stage. Third, during osteoprogenitor differentiation, BSP mRNA expression

undergoes a second significant up-regulation prior to that of OCN, and relatively early after up-

regulation of ALP. Fourth, FGF-RI mRNA is significantly up-regulated prior to that of PDGF-Ra and

earlier than up-regulation of ALP; a second significant up-regulation occurs late in the differentiation

sequence as cells acquire other features of differentiated osteobiasts including OCN. Fifth, PDGF-Ra

mRNA is maintained at a relatively constant Ievel until it is up-regulated similarly to FGF-RI and

osteoblast-associated markers late in the differentiation sequence. Sisth, arnongst the more mature cells

in the Iineage (expressing a11 osteoblast-associated markers) both FGF-R 1 and PDGF-Ra mRNAs tended

aiso to decrease in relative concert with the osteoblast-specific markers. but FGF-RI appeared to be

down-regulatec! to a greater extent than PDGF-Ra in most of these mature cells. Seventh, on average,

BSP expression remained higher than did other osteoblast markers in the rnost mature colonies in the

analys is.

Discussion

Our data provide the first molecular characterization of definitive normal osteoprogenitors

committed to terminal differentiation and bone formation in vitro. The profiles established to date are

only a first step towards understanding primitive cells in this lineage, bved as they are on analyzing the

simultaneous expression profiIes of seven mRNAs - five known osteoblast-associated markers (COLL-1,

ALP, OPN, BSP, and OCN) and two receptors for potential regulatory factors (FGF-RI and PDGF-Ra).

Our data indicate that the approach of replica plating combined with poly(A)-PCR has allowed us to gain

new insights into the differentiation sequence within the osteoblast lineage, establishing landmarks for

new early developmental stages. We have extended characterization of osteoprogenitor cells to cells

more primitive than those yet expressing the earliest established marker (Le., ALP) ofcommitted

osteoprogenitors, have identified a cohort of early progenitors transiently expressing BSP, and have

characterized distinct stages of up-regulation of two different growth factor receptors, al1 of which

contribute new understanding of early developmental stages in this skeletal lineage.

The replica technique we applied to the RC population worked with the fidelity required for the

studies undertaken, i.e., to identi@ retrospectively osteoprogenitors early in their developmental lifetime.

The repIicas were made in low density cultures from discrete colonies well-separated from other

contaminating ce1 I and colony types and early in their developmental sequence while progenitor cells

were still protiferative. The progenitor celk stalled at lower than physiological temperature on the master

dishes were essentially quiescent or very slowly proliferating, but were able to resume their proliferation

and differentiation capabilities ultimately to form mineralized bone nodules when transferred back to

37°C as we determined both in pretiminary control experiments and by visual inspection of those

colonies chosen for anatysis here. To date, we have used this replica approach to identiq and isolate over

80 osteoprogenitor and preosteoblast colonies, al1 of which cornprised cells with pleiomorphic

morphology and none of which displayed the cuboidal morphology of differentiated osteoblasts

depositing and mineralizing osteoid. Our reason for selection of large numbers of colonies was to

confirm the validity of the approach. Le.. each phenotype was expected to be sampted more than once.

and to acquire information on the differentiation process itself, i.e., whether cells traversed a continuum

of changing expression profiles or traversed the differentiation sequence in quantal leaps. Our data

suggest that both patterns are characteristic of and define the osteoblast developmental process (see

be low ).

Previously, we showed that the technique of poly(A)-PCR (Brady et al., 1990; Brady and Iscove,

1993) kvas a powerful and discriminating tool to estabIish molecular profiles for osteobiastic cotonies

and single cells that had differentiated to different extents as recognized morphologically (Le., early,

intermediate, and mature osteoblast) (Chapter 2). When mRNA expression in the replica colonies

analyzed was quantified by this technique and CO-expression profiles of a variety of messages for bone-

related macrornolecules and growth factor receptors detemined, we found marked differences amongst

colonies. Elsewhere we have discussed heterogeneity of marker expression amongst single cells at the

same developmental stages in tems of a stochastic process contributing to osteoblast plasticity (see

Chapters 2 and 3). In the present analysis to seek transition points and landmarks representative of celts

as much more or less primitive, we have analyzed expression across Iarger developmental boundaries

and averaged amongst small cohorts of sibling cells (colonies) in which single ceil stochastic variations

would be averaged and thus minimized as a contribution to the profiles achieved. A few colonies

displayed features consistent with their having already reached a preosteoblast or early osteoblast

phenotype, Le.. simultaneously expressing al1 osteoblast-associated markers analyzed (COLL-1, ALP,

OPN. BSP, and OCN), as did the few morphologically defined later stage osteoblast colonies used here

for comparison and validation. In marked contrast, other replica colonies expressed none of these

mRNAs to detectable levels, while still others expressed different combinations of these messages.

Expression levels for al! markers analyzed with the exception of OPN, which was relatively uniformly

and highly expressed in virtually al1 colonies (see below), covered a range of expression levels from

undetectable (off) to detectable (on) but very low to intermediate to very high. This suggests that to sorne

extent. and at least as represented by those osteoblast markers analyzed, there is a quantal change (off-

on), but that once a marker is odacquired its expression as populations of osteoprogenitor cells

differentiate represents a continuum from detectable expression through gradually higher expression

until mature osteoblasts express most markers analyzed at very high levels.

To detemine whether it was possible to define categories or discrete stages through which the

progenitors pass, we devised a paradigm by which to rank the expression profiles based on information

available by a variety of techniques including in situ hybridization, Northerns, and immunocytochemistry

of osteoblasts forrning bone nodules in vitro and bone tissue (for reviews, see Rodan and Noda, 199 1;

Aubin et al., 1993; Stein and Lian, 1995; Aubin and Liu, 1996) and ordered them according to these

parameters (see results and Fig. 4.4). First, more differentiated (mdtilayered cuboidal cells with

abundant osteoid) and mature (osteoid with deposited mineral) osteoblast colonies were ranked based on

increasing levels of ALP expression followed by decreasing leveIs of ALP, respectively, since ALP

levels decrease when osteobtasts become less synthetically active (Doty and Schofield, 1976) or becorne

osteocytes (Holtrop, 1975); this imposed an order on al1 other rnarkers in these colonies. Second, we

defined primitive or immature osteoprogenitors as ones in which ALP was not yet expressed, to

distinguish them from mature or later osteoprogenitors which do express ALP. Such a designation is

consistent with previous observations in which immature osteoprogenitors (ALP-) are functionally

distinguishable from mature (ALP+) osteoprogenitors based on the finding that the former require a

stimulus (e-g., glucocorticoids; dexamethasone) to undergo the differentiation sequence in vitro whereas

the latter do not (Turksen and Aubin, 199 1). Since we performed the low density and replica plating

technique in the presence of dexamethasone, we expected and did capture both categories of

osteoprogenitor for Our molecular analysis. We next ranked those colonies in the p r o g e n i t o r / ~ ~ ~ '

category according to increasing ALP expression, since ALP activity is present in late stage

osteoprogenitors (Turksen and Aubin, 199 1 ), in preosteoblasts and osteoblasts (Doty and Schofield,

1976), and is known to increase as osteoblasts mature (Rodan and Rodan, 1983). Within such

osteoprogenitor-preosteoblslst colonies in which ALP was detectable along with COLL-1 and OPN, BSP

was detected in rnost colonies, and both BSP and OCN were detectabte at low levels in a few other

colonies. These results support Our observations made earlier at the single ce11 level in which we detected

the presence of mRNAs for ALP. COLL-1, OPN, BSP, andior OCN in a few cells in fibroblastic colonies

and confirm the hypothesis for which we had no detinitive proof then that these were probably

representative of comrnitted osteoblast lineage cells that had not yet acquired the rnorphological

characteristics of the lineage, rather than cells with leaky or prorniscuous expression of mRNAs (Chapter

2). It is also notable that al1 these markers can be and are expressed in cells that have not yet assumed the

cuboidal shape of preosteoblasts and osteoblasts; while these studies have not addressed the presence of

deposited protein, they do suggest the possibility that none of these, including the RGD-containing BSP

and OPN, may be directly responsible for the cuboidal shape detemination characteristic of mature

osteoblastic cells, and vice versa that overt cuboidal shape is not itself required for detectable

transcription levels of these particular genes.

By aligning colonies as described, general expression patterns were evident for the various bone-

associated and cytokine receptor markers (Figs. 4.4,4.5). In keeping with previous observations, as late

stage progenitor colonies develop into more mature osteoblast colonies. there is a sequential up-

regulation of expression of bone-related macromolecules beginning with COLL-1, followed by ALP and

then BSP, and finally by OCN (sec summaries in Rodan and Noda, 1991; Aubin et al., 1993; Stein and

Lian, 1995; Aubin and Liu, 1996). However. in contrast to some experimental models (see, e.g., Stein

and Lian, 1995) but in keeping with others and including in situ hybridization of bone in vivo (see, e.g.,

Aubin and Liu, 1996), we also found that COLL-1, BSP, and OCN messages displayed increasing levels

and were most highly expressed by osteoblast colonies with the highest ALP levels, then their levels

dropped in most cases relatively sharply when osteoid was mineralizing in mature osteoblast colonies.

FGF-Rl and PDGF-Ra mRNAs followed the same trends as the bone matrix molecule messages at these

differentiation stages. However, the most primitive colonies expressed only OPN, FGF-RI, and PDGF-

Ra , and a proportion of them expressed BSP (sec below). That virtually ail early osteoprogenitor and

preosteoblast colonies expressed moderate to high levels of OPN is not surprising, since as outlined

above these colonies had high numbers of cycling cells; OPN is known to be high in proliferating cells,

and it was originally cloned as a ceIl cycle related rnolecule induced in cells in vitro by tumor promoters

and growth factors (Smith and Denhardt, 1987; Nomura et al., 1988).

Both FGF-R1 and PDGF-Ra play a role in normal skeletal development. It was recently shown

in FGF-RI knockout mice that FGF-RI is essential for ce11 proliferation and axial patterning in mouse

developrnent (Deng et al., 1994) and mutations in FGF-RI are associated with skeletal abnonnalities in

hurnans (Muenke et al., 1994). In homozygous PDGF-Ra nuIl mouse mutants, mesenchymal ce11

proliferation is affected (Bowen-Pope et al., 199 1), resulting in growth retardation and deficiencies in

mesodermal structures (Schatteman et al., 1992). While early mesodermal cells are affected, cetls

traversing the osteoblast lineage specifically also respond to these growth factors. Although one study

has reported that FGF stimulates osteoprogenitor proliferation and di fferentiation in the rat bone marrow

culture system in vitro (Pitaru et al., 1993), most results have suggested that while FGF increases DNA

synthesis and stimulates osteoblast proliferation (Rodan et al., 1987; Canalis et al., 1988; Globus et al.,

1988; McCarthy et al., 1989; Berrada et al., 1999, it inhibits differentiation, i.e., decreasing osteoblast

differeqtiation related markers such as ALP (McCarthy et al., 1989; Shen et al., 1989; Rodan et al., 1989;

Berrada et al., 1995), COLL-1 (Shen et al., 1989; Hurley et al., 1993; Berrada et al., 1995), and OCN

production (Rodan et al., 1989; Berrada et al., 1995). PDGF has also been shown to enhance proliferation

of osteoblastic cells (Centrella et al., 1989; Zhang et al., 199 1; Pfeilschifier et al., 1992; Hock and

Canalis, 1994), but its effects on differentiation have been inconsistent in different models in vitro,

although in general inhibition is observed (Centrella et al.. 1992; Pfeilschifter et al., 1992; Hock and

Canalis, 1994). While these studies have s h o w that osteoblastic populations respond to FGF and PDGF

and others have shown that fetal RC cultures express PDGF-Ra (Pfeilschifier et al., 1992), there is little

information on which subpopulations of osteoblastic cells express the receptors and whether receptor

numbers change as a function of differentiation stage. Our results show that mRNAs for both growth

factor receptors are expressed continuously through the lineage from early progenitor to mature

osteoblast and that both tend to peak as osteoblasts reach maturity and then decline concomitant with

down-regulation of the osteoblast-specific messages. However, they are differentially regulated in the

earlier progenitors in the lineage. FGF-RI rnRNA, in particular, is fint significantly up-regulated prior to

that of PDGF-Ra, earlier than up-regulation of the early marker of committed progenitors, ALP. and

approximately concomitant with up-regulation of COLL-1; a second significant up-regulation occurs

later as the cells acquire other features of mature osteoblasts including OCN. Our finding that FGF-RI

goes through the first of its two stages of up-regulation relatively early in the differentiation sequence

and prior to up-regulation of ALP suggests that amongst target populations for FGF spanning multiple

developmental stages. FGF may have a special role in very early committed and proliferatively active

osteoprogenitors. The second special target stage for both receptors which peak at late developmental

stages concomitant with other markers such as BSP and OCN appears to be the matrix synthesizing

mature osteoblast cell. There is also a trend towards more rapid down-regulation of FGF-RI than PDGF-

R a in mineralization phase (Fig. 4 4 , suggesting that the latter may play a unique role at this terminal

differentiation stage (see also BSP below). These data provide evidence that the growth factor receptors

can be used as additional markers for osteoprogenitors at different developrnental stages and provide

some clues as to target genes for their activities.

One striking result from these analyses of primitive progenitor cells was that a srnall group of

colonies expressed low but detectable levels of BSP; a few of these colonies also expressed COLL-1, but

no other osteobiast-associated molecules. We (Chapter 2) and others (Bianco et al., 1991, 1993; Chen et

al., 199 1 b) have shown that BSP is up-regulated as osteoblasts mature in vitro and in vivo at sites of de

novo bone formation. While it is generally considered a relatively late stage marker, expressed

concomitantly with OCN and just prior to mineralization, and can seed hydroxyapatite crystal formation

in vitro (Hunter and Goldberg, 1993), Our analysis of clonogenic bone nodules showed that it is up-

regulated prior to OCN and well before detectable mineralization can be observed (Chapter 2). We have

now detected transient BSP expression even earlier, prior to the onset of COLL-I and ALP expression,

and osteogenesis. Recently, we have also confirmed the presence of primitive ~ r d L J ' ( c ~ c l i n ~ ) / ~ ~ ~ -

IBSP' cells by double-label immunocytochemistry in clonogenic bone nodules in vitro (Malaval et al.,

1996) indicating that the mRNA results reported here are meaningful also at protein translational level. It

is interesting to speculate that this early and apparently transient expression of BSP rnay reflect its role as

a ceIl adhesion molecule through its RGD recognition site for integrins' notably the a& vitronectin

receptor (Oldberg et al., 1988b); BSP has also been reported to mediate osteoblast ce11 attachment in

vitro (Mintz et al., 1993). This role for BSP may be functionally separate from its role during its second

round of up-regulation during the later stages of the differentiation sequence when osteoblasts are

actively synthesizing other matrix molecules and depositing osteoid. Late during this second

developmental window, Le.. when matrix is mineralizing and cells achieve osteocyte and/or lining ceIl

status, it is also interesting that BSP expression remains on average higher than that of other bone matrix

molecules and ALP.

CIearlyo the combined approach of single colony isolation and poly(A)-PCR appears to offer a

means by which to detemine a molecular fingerprint of normal (Le., non-established, non-transformed)

committed osteoblasts and early preosteoblasts, through to differentiating osteoblasts and mature

osteoblasts. The observations suggest that within the osteoblast differentiation sequence both discrete

stages and a continuum of changing marker expression levels occur with much variation in expression

for any given marker. We have identified novel developrnental stages characterized by expression of

known osteoblast-associated markers such as BSP. We have also demonstrated that the mRNA

expression for certain growth factor receptors is modulated during osteoblast differentiation and the

sequential expression of different receptors appears to provide markers for cells earlier in the tineage

than those already expressing ALP. Bone is a heterogeneous tissue that contains a mixed cell population.

and it becomes necessary to understand when and which cells in the osteoblast lineage express receptors

for various factors to sort the heterogeneity observed among osteoblasts into physiologically meaningful

classes. By studying the expression of various cytokines and their receptors during the osteoblast

differentiation sequence, we will not only have available more marken to help define transitional stages,

but also shed light on the cellular targets rnediating the diverse effects of over-expression or under-

expression of these farni lies of molecules. Finally, the poly(A)-PCR approach used has allowed us to

eenerate cDNA libraries of these multiple osteogenic stages. spanning primitive osteoprogenitor to L

mature osteoblast. from which we are now isolating novel osteoblast lineage genes (Candeliere et al.,

1996).

General Sumrnaty and Conclusions

The studies presented in this thesis were aimed at furthering knowledge about osteoblast lineage

and differentiation. The fetal RC bone noduIe system was used in combination with poly(A)-PCR,

immunocytochernistry, and replica plating techniques to investigate patterns of expression of repertoires

of osteoblast-associated markers during transitional stages of the osteoblast differentiation sequence

from osteoprogenitor through to mature osteoblast.

In Chapter 2, I validated the use of poly(A)-PCR in the RC osteoblast mode1 in both whole

discrete colonies and single cells. Both poly(A)-PCR and immunocytochemistry revealed that different

colony types present in the RC population were reproducibly distinguishabie in their expression of either

general (COLL-1) or bone-associated (ALP, OPN, BSP, and OCN) macromolecules, such that

fibroblastic colonies were distinguishable fiom osteoblastic colonies, and the latter could be subdivided

into less mature or more mature osteoblastic colonies. While our results confirmed some aspects of the

osteoblast differentiation sequence as seen in mass populations, they also extended it to a level of single

ce11 expression and single ce11 heterogeneity not previously accessible, and thereby provided new insight.

First, different repertoires of osteoblast-associated markers were expressed in different cells, suggesting

variation in the switch-on or maintenance of the osteoblast differentiation program. There was marked

intercellular heterogeneity in expression of mRNA and protein within colonies; while some of this

appeared to be related to the differentiation stage of the cells, notably differences were also seen in

adjacent morphologically indistinguishable cells from the same colonies. Second. amongst colonies

classified as fibroblastic on the basis of rnorphology, heterogeneity was also evident and there were some

cells expressing features consistent with their being osteoprogenitor cells, but for which we had no

definitive proof. 1 followed up what we considered the two most important aspects of the observations:

heterogeneity and the accessibility of osteoprogenitors with this method.

One manifestation of osteoblast heterogeneity observed in Chapter 2 was that osteoblasts in close

proximity and morphologically identical appeared to be different in expression of a variety of osteoblast-

associated markers. Chapter 3 explored further the validity of the hypothesis that heterogeneity of

phenotype exists amongst such prirnary culture mature osteoblasts of clonal origin. To discriminate

heterogeneous expression frorn overt differences in differentiation stage or proliferative status of cells,

we concentrated on mature osteoblasts, Le.. post-proliferative ce11 that make osteoid that was in the

process of mineralizing. 1 found that extreme cell-to-ce11 heterogeneity exists in sibling cells for the

repertoire of genes expressed. Both double immunolabeling with Hoechst 33258 and OCN and the

diverse patterns observed amongst expression of the 7 genes analyzed suggested that the variation in

both mRNA levels and antibody labeling intensity/protein expression appeared independent of variations

in cell cycle. This rnarked heterogeneity suggests that the mature osteoblast phenotype is not a singIe

unique phenotype but rather encompasses a plasticity or flexibility in pattern of protein and mRNA

expression from the repertoire of osteoblast-associated markers. However, the biological significance of

this plasticity is not yet known. Whether cells cycle through the different repertoires observed in some

regimented fashion or the levels refiect a stochastic process is yet to be deterrnined. Since the

heterogeneity extends to at least on hormone receptor, it will be of interest to investigate the possibility

that some regulatory molecules may elicit their effects on only a subset of ceIls at any point in time. The

data further suggest that the most unarnbiguous marker of the mature osteoblast remains its ability to

contribute with its siblings to the production of bone rather than in an absolute level of production of

particular macromolecules.

In Chapter 4, 1 identified and established a molecular profile for the more primitive and Iargely

uncharacterized osteoprogenitor cells in the lineage. A strategy was devised in which the relatively rare

osteoprogenitors present in primary cultures of fetal RC cell populations were identified retrospectively

by a replica plating technique and the repertoire of genes expressed anaIyzed by poly(A)-PCR. 1 am

unaware of any previous reports of the replica plating approach for such analyses of rare progenitors in

primary cell populations. but my data suggest it is a valuable tool applicable to other tissue vpes. A

molecular fingerprint and a developmental sequence were established based on simultaneous analysis for

both known osteoblast-associated markers (COLL-1, ALf, OPN, BSP, and OCN) and potential

regulatory molecules, i.e., cytokine receptor messages (FGF-RI and PDGF-Ra). By analysis of

osteoprogenitor and osteoblast colonies captured at different developmental stages, I found a

recognizable cohort of cells considered very primitive but not quiescent since they express low-

intermediate levels of mRNAs for OfN, FGF-RI, and PDGF-Ra; a distinct developrnental stage of

relatively primitive ceils transiently expressing BSP, Le., a novel transition point in osteoprogenitor

development not previously seen in any mode1 system or bone in vivo; and evidence for both discrete

stages and continua of changing marker expression Ievels occur with much variation in expression for

any given marker within the osteoblast differentiation sequence. The combined approach of replica

plating and poly(A)-PCR allowed rnolecular fingerprinting of definitive primitive osteoprogenitors, gave

new insights into the osteoblast differentiation sequence and established new landmarks in early

osteoblast development, that could clearly be extended fùrther by analysis of the available pools of

sarnples with other probes.

Through these studies, several cDNA libraries, spanning from osteoprogenitor to preosteoblast

through to mature osteoblast, are now collectively available for future investigations of osteogenic

cornmitment and differentiation. One useful approach will be to use these libraries to address whether

there exists specifichovel genes which are activated or inactivated during osteoblast

commitment/differentiation processes. Recently, we have been able to identiQ a novel marker

(Candeliere et al., 1996) using these libraries and a new method, so-called cDNA fingerprinting,

suggesting that this is a viable approach and that new markers of transitional events in osteoblast

differentiation await detection. There are currently several antibodies available that irnrnunolabel

osteoblastic cells, although most detect relatively mature osteoblasts (Chapten 1-3). The availability of

antibodies prepared by expressing new cDNAs would be of particular interest for further analysis of

earlier osteoprogenitor cells. The new gene isolated recently by a postdoc in the lab appears to code a

putative membrane protein and an antibody to it may be usefûl in sorting or panning for enrichment of

osteoblast subpopulations. Together, my work has established new information on cells of the osteoblast

lineage and is a useful Iaunching place for generation of other novel tools for further elucidation of both

primitive progenitor and terminal ly differentiated cells.

REFERENCES

Abou-Samra, A.-B., Juppner, H., Force, T., Freeman, M.W., Kong, X.-F., Schipani, E., Urena, P.,

Richards, J., Bonventre, J.V., Potts, J.T., Kronenberg, HM., and Segre, G.V. (1992). Expression cloning

of a comrnon receptor for parathyroid hormone-related peptide from rat osteoblast-like cells: A single

receptor stimulates intracellular accumulation of both CAMP and inositol triphosphates and increases

intracellular free calcium. Proc. Natl. Acad. Sci. USA 89, 2732-2736.

Aubin, J.E., Heersche, J.N.M., Merrilees, M.J., and Sodek, J. (1982). Isolation of bone ceII clones with

differences in growth, hormone responses, and extracellular matrix production. J. Ce11 Biol. 92,452-46 1.

Aubin, J.E., Heersche, J.N.M., Bellows, CG., and Grigoriadis, A.E. (1990). Osteoblast Iineage analysis

in fetal rat calvaria cells. In Calcium Regulation and Bone Metabolism. D.V. Cohn, F.H. Glorieux, and

T.J. Martin, eds. (Amsterdam: Elsevier Science), pp. 362-370.

Aubin, J.E.. Bellows, CG., Turksen, K., Liu, F., and Heersche, J.N.M. (1992). Analysis of the osteoblast

lineage and regulation of differentiation. In Chernistry and Biology of Mineralized Tissues. H. Slavkin

and P. Price, eds. (Amsterdam: Elsevier Science), pp. 267-276.

Aubin, J.E., Turksen, K., and Heersche, J.N.M. (1993). Osteoblastic ce11 lineage. In Cellular and

Molecular BioIogy of Bone. M. Noda, ed. (San Diego: Academic Press), pp. 1-45.

Aubin, J.E.. Gupta, A.K., Bhargava, U., and Turksen, K. (1 996a). Expression and regulation of galectin 3

in rat osteoblastic cells. J. Cell. Physiol. (In press).

Aubin, J.E., Gupta. A.K.. Zirngbl, R.. and Rossant. J. (1996b). Knockout mice lacking bone sialoprotein

expression have bone abnorrnalities. J. Bone Miner. Res. 1 I(SI), S 1 02.

Aubin, J.E. and Liu. F. (1 996). The osteoblast lineage. In PrincipIes of Bone Biology. L.G. Raisz, G A .

Rodan, and J.P. Bilezikian, eds. (San Diego: Academic Press), pp.5 1-67.

Aubin, J.E. and Turksen, K. (1996). Monoclonal antibodies as tools for studying the osteoblast lineage.

Microsc. Res. Tech. 33, 128- 140.

Auffky, C. and Rougeon, F. (1980). Purification of mouse immunoglobulin heavy-chain messenger

RNAs trom total myeloma tumor RNA. Eur. J. Biochem. 107, 303-3 14.

Becker, J., Schuppan, D.. Benzian. H., Bals, T., Hahn, E.G., Cantaluppi, C., and Reichart, P. (1986).

Immunohistochemical distribution of collagens types IV, V, and VI and of pro-collagens types 1 and III

in human alveolar bone and dentine. J. Histochem. Cytochem. 34, 14 i 7- 1429.

Bellows, C.G., Aubin, J.E., Heersche, J.N.M., and Antosz, M.E. ( I986a). Mineralized bone nodules

forrned in vitro from enzymatically reIeased rat calvaria ceIl populations. Calcif. Tissue Int. 38, 143- 154.

Bellows, C.G., Sodek, J., Yao, K.-L., and Aubin, LE. (1 986b). Phenotypic differences in subclones and

long-tenn cuItures of clonally derived rat bone ceIl lines. J. Cell. Biochem. 31, 153- 169.

Bellows, C.G., Aubin, J.E., and Heersche, J.N.M. (1987). Physiological concentrations of glucocorticoids

stimulate formation of bone nodules from isolated rat calvaria cells in vitro. Endocrinology 121, 1985-

1992.

Bellows, CG., Heersche, J.N.M.. and Aubin, J.E. (1990a). Determination of the capacity for proliferation

and differentiation of osteoprogenitor cells in the presence and absence of dexamethasone. Dev. Biol.

140, t 32-138.

Bellows, C.G., Ishida, H., Aubin, J.E.. and Heersche, J.N.M. (1990b). Parathyroid hormone reversibly

suppresses the differentiation o f osteoprogenitor cells into functional osteoblasts. Endocrinology 177,

3111-3116.

Bellows, C.G., Reimers. S., and Heersche, J.N.M. (1995). Expression of mRNA for osteopontin,

osteocalcin, bone sialoprotein and type I collagen at different stages of osteoblastic differentiation and its

regulation by I,2S(OH)2D3. J. Bone Miner. Res. 10(SI), S414.

Bellows, C.G. and Aubin, J.E. (1989). Determination of nurnben of osteoprogenitors present in isolated

fetal rat calvaria celfs in vitro. Dev. Biol. 133, 8-13.

Berrada, S., Lefebvre, F., and Harmand, M.F. (1995). The effect of recombinant human basic fibroblast

growth factor rhFGF-2 on human osteoblast in growth and phenotype expression. In Vitro Cell. Dev.

Biol. Anim. 31,698-702.

Bhargava, U., Bar-Lev, M., Bellows, CG., and Aubin, J.E. ( 1988). Ultrastructural analysis of bone

nodules fonned in vitro by isolated fetal rat calvaria cells. Bone 9, 155- 163.

Bianco, P., Fisher, L. W., Young, M.F., Termine, J.D., and Robey, P.G. ( 1990). Expression and

localization of the two smalI proteoglycans bigiycan and decorin in developing human skeletal and non-

skeletal tissues. J. Histochem. Cytochem. 38, 1 549- 1563.

Bianco, P., Fisher, L. W., Young, M.F.. Termine, J.D., and Robey, P.G. ( 199 1 ). Expression of bone

sialoprotein (BSP) in developing human tissues. Calcif. Tissue Int. 49,42 1-426.

Bianco, P., Riminucci. M., Bonucci, E.. Termine, J.D.. and Gehron Robey, P. (1993). Bone sialoprotein

(BSP) secretion and osteoblast differentiation: Relationship to bromodeoxyuridine incorporation,

alkaline phosphatase, and matrix deposition. J. Histochem. Cytochem. 4!, 183- 19 1.

Boivin, G., Mesguich, P., Pike, J. W.. Bouillon. R., Meunier, P.J., Haussier, M.R., Dubois, P.M., and

More[, G. (1987). Ultrastructural immunocytochemical localization of endogenous 1,25-

dihydroxyvitamin D3 and its receptors in osteoblasts and osteocytes from neonatal mouse and rat

calvaria. Bone Miner. 3, 125- 136.

Boivin, G., Morel, G., Lian, J.B., Anthoine-Therrier, C., Dubois, P.M., and Meunier, P.J. (1990).

Localization of endogenous osteocalcin in neonatal rat bone and its absence in articular cartilage: Effect

of warfarin treatment. Virchows Archiv. A Pathol. Anat. Histopathol. - / I 7, 505-5 12.

Boivin, G., Anthoine-Terrier, A., and Morel, G. (1994). Ultrasmictural localization of endogenous

hormones and receptors in bone tissue: An immunocytological approach in frozen samples. Micron 25,

15-27.

Boockfor, F.R., Hoeffler. J.P., and Frawley, L.S. (1985). Cultures of GH3 cells are functionally

heterogeneous: Thyrotropin-releasing hormone, estradiol and cortisol cause receiprocal shifis in the

proportions of growth hormone and proIactin secretors. Endocrinology 1 17, 4 1 8-420.

Boskey, A.L. (1995). Osteopontin and related phosphorylated siatoproteins: Effects on rnineralization. In

Osteopontin: RoIe in Cell Signalling and Adhesion. D.T. Denhardt, W.T. Butler, A.F. Chambers, and

D.R. Senger, eds. (New York; New York Academy of Sciences), pp. 249-256.

Bowen-Pope, D.F., Van Koppen, A., and Schatteman, G. (1991). 1s PDGF really important? Testing the

hypothesis. Trends Genet. 7,4 13-4 t 8.

Brady, G. and Iscove, N.N. (1993). Construction of cDNA libraries from single cells. Methods Enzymol.

225, 6 1 1 -623.

Brady, G., Barbara, M., and Iscove, N.N. (1990). Representative in vitro cDNA amplification from

individual hemopoietic cells and colonies. Methods Mol. CeIl. Biol. 2, 17-25.

Brady, G., Billia, F., Knox, J., Hoang, T.. Kirsch. I.R., Voura, E.B., Hawley. R.G., Cumming, R.,

Buchwald, M., Sirninovitch. K., Miyarnoto, N., Boehmelt, G., and Iscove. N.N. (1 995). Analysis of gene

expression in a coinplex differentiation hierarchy by global amplification of cDNA from single cells.

Curr. Biol. 5, 909-922.

Braidman, I.P.. Davenport, L.K., Carter, D.H., Selby, P.L.. Mawer, E.B., and Freemont, A.J. (1995).

Preliminary in situ identification of estrogen target cells in bone. J. Bone Miner. Res. 10, 74-80.

Bronckers, A.L.J.J., Gay, S., Dimuzio, M.T., and Butler, W.T. (1985). Immunolocalization of garnma-

carboxyglutarnic acid-containing proteins in developing rat bone. Coll. Relat. Res. 5,273-28 1.

Bronckers, A.L.J.J., Gay, S., Finkelman, R.D., and Butler, W.T. (1 987). Developmental appearance of

the Gla proteins (osteocalcin) and alkaline phosphatase in tooth gems and bone of the rat. Bone Miner.

Res. 2,361-373.

Bruder, S.P. and Caplan, A.I. (1989a). First bone formation and the dissection of an osteogenic lineage in

the embryonic chick tibia is revealed by monocIona1 antibodies against osteoblasts. Bone IO, 359-375.

Bruder, S.P. and Caplan, A.I. (1989b). Discrete stages within the osteogenic lineage are revealed by

alterations in the ce11 surface architecture of embryonic bone cells. Connect. Tissue Res. 20, 73-79.

Bruder, S.P. and Caplan, A.I. ( 1 990). Terminal differentiation of osteogenic ceIls in the embryonic chick

tibia is revealed by a monoclonal antibody against osteocytes. Bone II, 1 89- 198.

Butler, W.T. ( 1989). The nature and significance of osteopontin. Connect. Tissue Res. 23, 123- 136.

Carneron, D.A. ( 1 968). The Golgi apparatus in bone and cartilage cells. Clin. Orthop. 58, 19 1-2 1 1.

Canalis, E. (1983). The hormonal and local regulation of bone formation. Endocr. Rev. 4, 62-77.

Canalis, E., Centrella, M., and McCanhy, T. (1988). Effects of basic fibroblast growth factor on bone

fonnation in vitro. J. Clin. Invest. 81, 1572-1577.

Canalis, E., McCarthy, T.. and Centrella. M. (1989). The regulation of bone fonnation by local growth

factors. Bone Miner. Res. 27-56.

Canalis, E.. McCarthy, T.L., and Centrella, M. (1991). Growth factors and cytokines ir, bone ceIl

metabolism. Annu. Rev. Med. 42. 17-24.

Candeliere, G.A., Liu. F., and Aubin, J.E. ( 1996). IdentiQing new marken for transitional stages in

osteoblast maturation by cDNA fingerprinting of differentiating rat calvaria cultures. J. Bone Miner. Res.

I I (SI).

Carmada, A.J., Butler, W.T., Finkelman, R.D., and Nanci, A. ( 1 987). lmmunocytochem ical localization

of gamma-carboxyglutamic acid-containing proteins (osteocalcin) in rat bone and dentin. Calcif. Tissue

Int. 40,349-355.

CentreIIa, M., McCarthy, TL., and Canalis, E. (1989). Platelet-derived growth factor enhances

deoxyribonucleic acid and collagen synthesis in osteoblast-enriched cultures from fetai rat parietal bone.

Endocrinology 125, 1 3- 19.

Centrella, M., McCarthy, T.L., Kusmik, W.F.. and Canalis, E. (1992). Isoform-specific regulation of

PDGF activity and binding in osteoblast-enriched cuItures fiom fetal rat bone. J. Clin. Invest. 89, 1076-

1084.

Chen, J., Shapiro, H.S., Wrana, J.L., Reimers, S., Heersche. J.N.M., and Sodek, J. ( 1 99 1 a). Localization

of bone sialoprotein (BSP) expression to sites of mineralized tissue formation in fetal rat tissues by in

situ hybridization. Matrix I I , 133- 143.

Chen, J., Shang, Q., McCulloch, C.A.G., and Sodek, J. (1991b). Immunohistochemical localization of

bone sialoprotein in foetal porcine bone tissues: Cornparisons with secreted phosphoprotein 1 (SPP- 1,

osteopontin) and SPARC (osteonectin). Histochern. J. 23,28 1-289.

Chen, J., Shapiro, H.S., and Sodek, J. (1992). Developrnental expression of bone sialoprotein mRNA in

rat mineralized connective tissues. J. Bone Miner. Res. 7,987-997.

Chen, J.. McCulloch, C.A.G., and Sodek, J. (1993a). Bone sialoprotein in developing porcine dental

tissues: Cellular expression and cornparison of tissue localization with osteopontin and osteonectin.

Archs. Oral Biol. 38, 24 1-249.

Chen, J., Singh, K., Mukherjee, B.B., and Sodek, J. (1993b). Developrnental expression of osteopontin

(OPN) mRNA in rat tissues: Evidence for a role for OPN in bone formation and resorption. Matrix 12,

113-123.

Chen, J., McKee, MD., Nanci, A., and Sodek, J. (1994). Bone sialoprotein mRNA expression and

ultrastructural localization in fetal porcine calvarial bone: Cornparisons with osteopontin. Histochem. J.

26,67-78.

Clauss, LM., Gravallese, E.M., Darling, J.M., Shapiro, F., Glimcher, M.J., and Glimcher, L.H. (1993). In

situ hybridization studies suggest a role for the basic region-leucine zipper protein hXBP- 1 in exocrine

gland and skeletal development during mouse embryogenesis. Dev. Dyn. 197, 146- 156.

Cole, D.E.C. and Hanley, D.A. ( 199 1 ). Osteocalcin. In Bone. Volume 3: Bone Matrix and Bone Specific

Products. B.K. Hall. ed. (Boca Raton, FL: CRC Press), pp. 239-294.

Connor, N.S., Aubin, LE., and Sodek, J. (1983). Independent expression of type 1 collagen and

fibronectin by normal fibroblast-like cells. J. Ce11 Sci, 63, 233-244.

Craig, A.M., Smith, J.H., and Denhardt. D.T. (1989). Osteopontin, a transformation-associated ce11

adhesion phosphoprotein, is induced by 12-O-tetradecanoylphorbol- 13-acetate in mouse epidermis. J.

Biol. Chem. 264,9682-9689.

Delmas. P.D. and Malaval, L. ( 1 993). The proteins of bone. In Handbook of Experimental

Pharmacology. G. Mundy and T.J. Martin, eds. (New York: Springer Verlag), pp. 673-723.

Deng, C.-X., Wynshaw-Boris. A., Shen. M.M.. Daugherty, C., Ornitz, D.M., and Leder, P. (1994).

Murine FGFR-1 is required for early postimplantation growth and axial organization. Genes Dev. 8.

3045-3057.

Dodds. R.A., Men-y, K., Littlewood, A., and Gowen, M. ( 1 994). Expression of mRNA for IL 1 b, IL6 and

TGF bl in developing human bone and cartilage. J. Histochem. Cytochem. 42, 733-744.

Dodig, M., Kronenberg, M.S., Krearn. B.. Pan, Z., Upholt, WB., and Lichtler, A.C. (1 995). Msx-2

inhibits COL 1 A 1 promoter-CAT construct expression in osteoblastic cells. J. Bone Miner. Res. I O ( S I ) ,

S42 1.

Doty. S.B. (1981). Morphological evidence of gap junctions between bone cells. Calcif. Tissue Int. 33,

509-5 12.

Doty, S.B. and Schofield, B.H. (1976). Enzyme histochemistry of bone and cartilage cells. Prog.

Histoc hem. Cytochem. 8, 1 -3 8.

Ducy, P., Desbois. C., Boyce, B., Pinero. G., Story, B., Dunstan, C.. Smith. E., Bonadio, J., Goldstein, S..

Gundberg, C., Bradley, A., and Karsenty, G . ( 1996). Increased bone formation in osteocalcin-deficient

mice. Nature 382.448-452.

Eberwine, J., Yeh, H., Miyashiro, K., Cao, Y., Nair, S., Finnell, R., ZetteI, M., and Coleman, P. (1985).

Analysis of gene expression in single live neurons. Proc. Natl. Acad. Sci. USA 89, 30 10-30 14.

Esko, J.D. (1989). Replica plating of animal cells. Methods Ce11 Biol. 32,387-422.

Fedarko, N.S., Bianco, P., Vetter, U., and Gehron Robey, P. ( 1 990). Human bone ce11 enzyme expression

and cellular heterogeneity: Correlation of aIkaline phosphatase enzyme activity with ce11 cycle. I. Cell.

Physiol. 1-14, I 1.5- 12 1.

Fisher, L. W.. McBride. O. W.. Termine, J.D., and Young, M.F. ( 1990). Human bone sialoprotein:

Deduced sequence and chromosomal localization. J. Biol. Chem. 265,2347-235 1.

Fleet. J.C. and Hock, J.M. ( 1994). Identification of osteocalcin mRNA in non-osteoid tissues of rats and

humans by reverse transcription-polymerase chah reaction. J. Bone Miner. Res. 9, 1565- 1573.

Franceschi, R.T. and Iyer. B.S. (1992). Relationship between collagen synthesis and expression of the

osteoblast phenotype in MC3T3-E 1 cells. J. Bone Miner. Res. 7,235-246.

Franzen. A. and Heinegard. D. ( 1985). Isolation and characterization of two sialoproteins present only in

bone calcified matrix. B iochem. J. 232. 7 1 5-724.

Friedenstein, A.J. ( 1980). Strornal mechanisms of bone marrow: Cloning in vitro and retransplantation in

vivo. In Immunobioiogy of Bone Marrow Transplantation. S. Theinfelder, ed. (Berlin: Springer-Verlag),

pp. 19-29.

Friedenstein, A.J., Chailakhyan, R.K., and Gerasimov, U.V. (1 987). Bone marrow osteogenic stem cells:

In vitro cultivation and transplantation in diffusion chambers. CeIl Tissue Kinet. 20,263-272.

Friedenstein, A.J. (1990). Osteogenic stem cells in the bone marrow. Bone Miner. Res- 7. 243-272.

Funakoshi, Y., Matsuda, S., Uryu, K., Fujita. H., Okumura, N., and Sakanaka, M. (1993). An

immunohistochemica1 study of basic fibroblast growth factor in the developing chick. Anat. Embryol.

187.415423.

Gehron Robey, P., Fedarko, N.S., Hefferan, T.E., Bianco. P., Vetter, U.K.. Grzesik, W., Friedenstein, A.,

van der Plum, G., Mintz, K.P., Young, M.F., Ken; J.M., Ibaraki, K., and Heegaard, A. (1993). Structure

and molecular regulation of bone rnatrix proteins. J. Bone Miner. Res. 8(S& S483-S.187.

Genovese, C.. Rowe, D.. and Kream. B. (1984). Construction of DNA sequences complementary to rat

a, and ai collagen mRNA and their use in studying the regulation of Q Q ~ 1 collagen synthesis by 1,25-

dihydroxyvitamin D. Biochemistry 23.62 10-62 16.

Gerstenfeld, L.C.. Chipman, S.D., Glowacki. J., and Lian, J.B. ( 1 987). Expression of differentiated

fiinction by mineralizing cultures of chicken osteoblasts. Dev. Biol. IZ,49-60.

Gerstenfeld. L.C.. Chipman. S.D., Kelly. CM.. Hodgens. K.J., Lee, D.D.. and Landis. W.J. (1988).

Collagen expression, ultrastructural assembly. and mineralization in cultures of chicken embryo

osteoblasts. J. Cell Biol. 106, 979-989.

Globus. R.K.. Panerson-Buckendahl. P.. and Gospodarowicr D. (1988). Regulation of bovine bone ce11

proliferation by fibroblast growth factor and transforming g r o h factor p. Endocrinology 123,98- 105.

Glowacki, J. and Lian, J.B. (1987). lmpaired recruitment and differentiation of osteoclast progenitors by

osteocalcin-deplete bone implants. Cell Differ. 21,247-254.

Glurnoff, V.. Savontaus, M., Vehanen, J., and Vuorio, E. (1994). Analysis of aggrecan and tenascin gene

expression in mouse skeletal tissues by Norther and in situ hybridization using species specific cDNA

probes. Biochim. Biophys. Acta 1119,6 13-622.

Goldberg, H.A. and Hunter. G.K. (1995). The inhibitory activity of osteopontin on hydroxyapatite

formation in vitro. In Osteopontin: Role in Cell Signalling and Adhesion. D.T. Denhardt. W.T. Butler,

A.F. Chambers. and D.R. Senger, eds, (New York: New York Academy of Sciences), pp. 305-308.

Goldstein, D.J., Rogers, C., and Hanis, H. ( 1 982). A search for trace expression of placental-Iike alkaline

phosphatase in non-malignant human tissues: Demonstration of its occurrence in lung, cervix. testis and

thymus. Clin. Chem. Acta 125.63-75.

Grigoriadis, A.E.. Petkovich. P.M., Ber. R.. Aubin, J E . and Heersche, J.N.M. (1985). Subclone

heterogeneity in a clonally-derived osteobiast-like cell line. Bone 6, 249-256.

Grigoriadis. A.E.. Heersche, J.N.M.. and Aubin. J.E. (1988). Differentiation of muscle. fat. cartilage, and

bone from progenitor cells present in a bone derived clona1 cell population: Effects of dexamethasone. J.

Ce11 Biol. 106.2 139-2 15 1.

Grigoriadis. A.E.. Heersche. J.N.M., and Aubin, J.E. (1990). Continuously growing bipotential and

monopotential rnyogenic. adipogenic. and chondrogenic subciones isolated fiom the multipotential RCJ

3.1 clonal cell line. Dev. Biol. 142. 3 13-3 18.

Groot. C.G.. Danes. J.K.. Blok. J. . Hoosendijk. A.. and Hauschka P.V. (1986). Light and electron

microscopie demonstration of osteocalcin antigenicity in embryonic and adult rat bone. Bone 7. 379-385.

Grzesik, W.J. and Gehron Robey. P. (1994). Bone matrix RGD glycoproteins: Immunolocalization and

interaction with human primary osteoblastic bone cells in vitro. J. Bone Miner. Res. 9.487-496.

Guenther, H.L.. Hofstetter, W., Stutzer, A.. Muhlbauer. R., and Fleisch. H. (1989). Evidence for

heterogeneity of the osteoblastic phenotype detennined with clonal rat bone cells estabiished fiom

transforming growth factor-b-induced ce11 colonies grown anchcrage independently in semisolid

medium. Endocrinology 125,2092-2 102.

Hamada, K.. Okawara, Y.. Fryer. LN., Tomonaga, A., and Fukuda, H. (1995). The detection of the

mRNAs of procollagens types 1, II and III in human fetal fingers by in situ hybridization using

digoxigenin-labelled oligonucleotide probes. Histochem. J. 27, 309-3 17.

Hauschka, P.V.. Lian, J.B., and Gallop, P.M. (1975). Direct identification of the calcium-binding amino

acid gamma-carboxyglutamate in mineralized tissue. Proc. Natl. Acad. Sci. USA 72,3925-3929.

Hauschka, P.V., Lian, J.B., Cole, D.E.C., and Gundberg, C.M. ( 1 989). Osteocalcin and matrix gIa

protein: Vitamin K-dependent proteins in bone. Physiol. Rev. 69, 990- 1047.

Hauschka, P.V. and Carr, S.A. ( 1982). Calcium dependent a-helical structure of osteocalcin.

Biochemistry 2 1, 2538-2547.

Hauschka, P.V. and Reid, ML. (1978). Timed appearance of calcium binding protein containing garnma-

carbo'ryglutamic acid in developing chick bone. Dev. B iol. 65,426-434.

Heath. J.K., Rodan, SB., Yoon, K.G., and Rodan, G.A. (1989). Rat calvarial ce11 lines immortalized by

SV-40 large T-antigen -Constitutive and retinoic acid-inducible expression of osteoblastic features.

Endocrinology 124, 3060-3068.

Heersche, J.N.M., Aubin, J.E., Grigoriadis, A.E., and Moriya, Y. ( 1 985). Hormone responsiveness of

bone ceIl populations: Searching for answers in vivo and in vitro. In The Chemistry and Biology of

Mineralized Tissues. W.T. Butler, ed. (Birmingham. AL: Ebsco Media), pp. 286-295.

Heenche, J.N.M., Reimers, S., Wrana, J.L., Waye, M.M.Y.. and Gupta, A.K. (1 992). Changes in

expression of alpha 1 type 1 collagen and osteocalcin mRNA in osteoblasts and odontobiasts at different

stages of maturity as shown by in situ hybridization. Proc. Finn. Dent. Soc. W S I ) , 173-1 82.

Heersche. J.N.M. and Aubin, J.E. (1990). Regulation of cellular activity of bone-forming cells. In Bone.

Volume 1 : The Osteoblast and Osteocyte. B.K. Hall, ed. (Caldwell, NI: Telford Press), pp. 327-350.

Heinegard, D. and Oldberg, A. (1 989). Structure and biology of cartilage and bone matrix

noncollagenous macromolecules. FASEB J. 3,2042-205 1.

Helder, MN. , Bronckers, A.L., and Woltgens, J.H. (1993). Dissimilar expression patterns for the

extracellular matrix proteins osteopontin (OPN) and collagen type 1 in dental tissues and alveolar bone of

the neonatal rat. Matrix 13,4 15-425.

Hock, J.M. and Canalis, E. ( 1994). Platelet-derived growth factor enhances bone cell replication, but not

differentiated function of osteoblasts. Endocrinology 134, 1423- 1428.

Hodge. B.O. and Kream, B.E. ( 1 988). Variable effects of dexamethasone on protein synthesis in clona1

rat osteosarcorna cell. Endocrinology 122, 2 127-2 133.

Holtrop, M.E. ( 1975). The ultrastructure of bone. Ann. Clin. Lab. Sci. 5,264-27 1.

Hughes, F.J., Aubin, J.E., and Heersche, J.N.M. ( 1992). Differential chernotactic responses of different

populations of fetal rat calvaria cells to platelet-derived growth factor and transfonning growth factor B. Bone Miner. 19,63-74.

Hunter, G.K. and Goldberg, H.A. (1993). Nucleation of hydroxyapatite by bone sialoprotein. Proc. Natl.

Acad. Sci. USA 90. 8562-8565.

Hurley, M.M., Abreu, C., Harrison, J.R., Lichtler, AC., Raisz, L.G., and Krearn, B.E. (1993). Basic

fibroblast growth factor inhibits type 1 coltagen gene expression in osteoblastic MC3T3-El cells. J. Biol.

Chem. 268,5588-5593.

Ikeda, T., Nomura, S., Yamaguchi, A., Suda, T., and Yoshiki, S. (1992). In situ hybridization of bone

matrix proteins in undecalcified adult rat bone sections. J. Histochem. Cytochem. 40, 1079- 1088.

Ikeda, T., Miyaura, C., Ichikawa, A., Narumiya, S., Yoshiki, S., and Suda, T. ( 1995a). In situ localization

of three subtypes (EP,, EP3 and EP4) of prostaglandin E receptors in embryonic and newborn mice. J.

Bone Miner. Res. I OfSI), S 172.

Ikeda, T., Nagai, Y., Yamaguchi, A., Yokose, S., and Yoshiki, S. (1995b). Age-related reduction in bone

matrix protein mRNA expression in rat bone tissues: Application of histomorphometry to in situ

hybridization. Bone 16, 1 7-23.

Ishidou, Y., Kitajuma, 1.. Obama, H., Maruyama, I., Murata, F., Imamura, T., Yarnada, N., Ten Dijke, P..

Miyazono, K., and Sakou, T. (1995). Enhanced expression of type 1 receptors for bone morphogenetic

proteins during bone formation. J. Bone Miner. Res. 1 O, 165 1 - 1659.

Jamal, H.H. and Aubin, J.E. (1 996). CD44 expression in fetal rat bone: In vivo and in vitro analysis. Exp.

CeIl Res. 223,467-477.

Jande, S.S. and Belanger, L.F. ( I 973). The life cycle of the osteocyte. Clin. Orthop. 94, 28 1-305.

Karmali, R., Schiffmann, S.N., Vandenvinden, LM. , Hendy, G.N., Nys-DeWolf, N., Corvilain, J.,

Bergmann, P., and Vanderhaighen, J.-J. ( 1992). Expression of mRNA of parathyroid hormone-related

peptide in fetal bones of the rat. Cet1 Tissue Res. 270, 597-600.

Kasai, R., Bianco, P., Robey, P.G., and Kahn, A.J. (1994). Production and characterization of an

antibody against the human bone GLA protein (BGP/osteocalcin) propepetide and its use in

immunocytochemistry of bone celis. Bone Miner. 25, 167- 182.

Kasugai, S., Oida, S.. Iimura, T., Arai, N., Takeda, K., Ohya, K., and Sasaki. S. ( 1995). Expression of

prostglandin-E receptor subtypes in bone. Bone 17, 1-4.

Katz, E.P. and Li, S.-T. ( 1 973). Structure and function of bone collagen tibrils. J. Mol. Biol. 80, 1- 15.

Kelleman, O.. Buc-Caron, M.H.. Marie, P.J., Lamblin. D., and Jacob, F. (1990). An immortalized

osteogenic ce11 line derived from mouse teratocarcinorna is able to minerahze in vivo and in vitro. J. Cell

BioI. 110, 123-132.

Kember, N.F. (1 97 1). Ce11 population kinetics of bone growth: The first ten years of autoradiographic

studies with tritiated thymidine. Clin. Orthop. 76, 213-230.

Kineman, R.D. and Frawley, L.S. (1994). Secretory characteristics and phenotypic plasticity of growth

hormone- and prolactin-producing ce11 lines. J. Endocrinol. 140,455-463.

Klein-Nulend, J., van der Plas, A., Semeins, C.M., Ajubi, NE., Frangos, J.A., Nijweide, P.J., and Burger.

E.H. (1995). Sensitivity of osteocytes to biomechanical stress in vitro. FASEB 1. 9, 441-445.

Kleinman, H.K., Klebe, R.J., and Martin, G.R. (i98 1). Roie of collagenous matrices in the adhesion and

growth of cells. J. Cell Biol. 88,473-485.

Lazowski, DA., Fraher, L.J.. Hodsman, A., Steer, B., Modrowski, D., and Han, V.K. (1994). Regional

variation of insulin-Iike growth factor4 gene expression in mature rat bone and cartilage. Bone 15, 563-

5 76.

Lean, J.M., Jagger, C.J., Chambers, T.J., and Chow, J.W. (1995). lncreased insulin-like growth factor 1

mRNA expression in rat osteocytes in response to mechanical stimulation. Am. J. Physiol. 268, E3 18-

E327.

Lee, K., Deeds. J.D., Bond, A.T., Juppner, H., Abou-Samra, A.-B., and Segre, G.V. (1993). In situ

localization of PTH/PTHrP receptor mRNA in the bone of fetal and young rats. Bone 14, 34 1-345.

Lee, K., Deeds, J.D., Chiba, S., Un-No, M., Bond, A.T., and Segre, G.V. (1 994). Parathyroid hormone

induces sequential c-fos expression in bone cells in vivo: In situ localization of its receptor and c-fos

messenger ribronucleic acids. Endocrinology 134,44 1-450.

Lee, K.-H-, Bowen-Hope, D.F., and Reed. R.R. (1990). Isolation and characterization of the a platelet-

derived growth factor receptor from rat olfactory epithelium. Mol. CeIl. Biol. 10, 2237-2246.

Lee, K.-L., Aubin, J.E., and Heersche, J.N.M. (1992). Beta-glycerophosphate-induced mineralization of

osteoid does not alter expression of extracellular matrix components in fetal rat calvaria ce11 cultures. J.

Bone Miner. Res. 7, 1 2 1 1 - 12 1 9.

Lee, Y.-S. and Chuong, C.-M. (1992). Adhesion molecules in skeletogenesis: 1. Transient expression of

neural cell adhesion molecules (NCAM) in osteoblasts during endochondral and intmembranous

ossification. J. Bone Miner. Res. 7, 1435- 1446.

Lian, J.B., Tassinari, M., and Glowacki, J. (1984). Resorption of implanted bone prepared from normal

and warfarin-treated rats. J. Clin. Invest. 73, 1223- 1226.

Lian, J.B., Dunn, K., and Key, L.L. (1986). In vitro degradation of bone particles by human monocytes is

decreased with the depletion of the vitamin-K dependent bone protein from the matrix. Endocrinology

118, 1636-1642.

Lian, J.B., Stein, G.S., Owen, T.A., BorteIl, R., Shalhoub, V.. Bidwell, J., Breen, E., Barone, L.M., van

Wijnen, A., and Aronow, M. ( 1 992). Regulation of osteocaIcin gene expression during growth and

differentiation of the bone ceIl phenotype. In The Biological Mechanisms of Tooth Movement and

CraniofaciaI Adaptation. 2. Davidovitch, ed. (Birmingham, AL: EBSCO Media), pp. 159- 179.

Liu, F., Malaval, L., Gupta, A.K., and Aubin, J.E. (1994). Sirnultaneous detection of multiple bone-

related mRNAs and protein expression during osteoblast di fferentiation: Polymerase chain reac tion and

immunocytochemical studies at the single ceIl level. Dev. Biol. 166,220-234.

Luk, SC., Nopajaroonsri, C., and Simon, G.T. ( 1 974). The ultrastructure of endosteum: A topographic

study in young adult rabbits. J. Ultrastruct. Res. 46, 165- 183.

MacDonald, B.R. and Gowen, M. (1992). Cytokines and bone. Br. J. Rheumatol. 31, 149- 155.

MacGregor, G.R., Zambrowicz, B.P., and Soriano, P. (1995). Tissue non-specific alkaline phosphatase is

expressed in both embryonic and extraembryonic lineages during mouse embryogenesis but is r?ot

required for migration of primordial gerrn cells. Development I Z I, 1487- 1496.

Machwate, M., Jullienne, A., Moukhtar, M., and Marie, P.J. (1995). Temporal variation of c-fos proto-

ooncogene expression during osteoblast differentiation and osteogenesis in Geveloping rat bone. J. Cell.

Biochem. 57,62-70.

Mackie, E.J. and Tucker, R.P. (1992). Tenascin in bone morphogenesis expression by osteoblasts and

cell-type expression of spIice variants. J. Ce11 Sci. 103, 765-771.

Majeska, R.J. and Rodan, G.A. (1 985). Culture and activity of osteoblasts and osteoblast-like cells. In

The Chernistry and Biology of Mineralized Tissues. W.T. Butler, ed. (Birmingham, AL: Ebsco Media),

pp. 279-285.

Malaval, L., Modrowski, D., Gupta, A.K., and Aubin, J.E. ( 1994). Cellular expression of bone related

proteins during in vitro osteogenesis in rat bone marrow stroma1 ce11 cultures. J. Cell. Physiol. 158, 555-

572.

Malaval, L., Liu, F., and Aubin, J.E. (1996). Correlation of proliferative status and expression of bone-

related proteins during osteoblast differentiation and the development of bone nodules in vitro.

(Subm itted for publication).

Malone, J.D., Teitelbaum, S.L., Grifin, G.L.. Senior, R.M., and Kahn, A.I. ( 1982). Recruitment of

osteocIast precursors by purified bone matrix constituents. J. Celi Biot. 92,227-230.

Mansukhani. A., Moscatelli, D., Talarico, D., Levytska, V., and Basilico, C. ( 1990). A murine fibroblast

growth factor (FGF) receptor expressed in CHO cells is activated by basic FGF and Kaposi FGF. Proc.

Natl. Acad. Sci. USA 87,4378-4382.

Mark, M.P., Prince, C.W., Gay, S., Austin, R.L., Bhown, M., Finkelman, RD., and Butter, W.T. (1987a).

A comparative immunocytochernical study on the subcellular distributions of 44 kDa bone

phosphoprotein and bone gamma-carboxyglutamic acid (g1a)-containing protein in osteoblasts. J. Bone

Miner. Res. 2,337-346.

Mark, M.P., Prince, C.W., Oosawa, T., Gay, S., Bronckers, A.L.J.J., and Butler, W.T. (1987b).

Immunohistochemical demonstration of a 44 kDa phosphoprotein in developing rat bones. J. Histochem.

Cytochem. 35, 707-7 1 5 .

Mark, M.P., Butler, W.T., Prince, C.W., Finkelman, R.D., and Ruch, J.-V. (1988). Developmental

expression of 44-kDa bone phosphoprotein (osteopont in) and bone gamma-carboxyglutarn ic acid (g1a)-

coritai~ing protein (osteocalcin) in calciQing tissues of rat. Differentiation 37, 123- 136.

Marks, S.C. and Popoff, S.N. (1 988). Bone cell biology: The regulation of development, structure, and

function in the skeleton. Am. J. Anat. 183, 1-44.

Martin, T.J., Ng, K.W., and Nicholson, G.C. ( 1 988). Cell biology of bone. In Clinical Endocrinology and

Metabolism. T.J. Martin. ed. (London: Bailliere TindaII), pp. 1-29.

Martin, T.J., Moseley, J.M., and Gillespie, M.T. (199 1). Parathyroid hormone-related protein:

Biochemistry and molecuIar biology. Crit. Rev. Biochem. Mol. Biol. 26, 377-395.

Martineau-Doize. B., Lai. W.H., Warshawsky, H., and Bergeron, J.J.M. ( 1988). In vivo demonstration of

cell types in bone that harbor epidemal growth factor receptors. Endocrinology 123, 84 1-858.

McCarthy, T.L.. Centrella, M., and Canalis, E. (1989). Effects of fibroblast growth factors on

deoxyribonucleic acid and collagen synthesis in rat parietal bone cells. Endocrinology 1 5 , 2 1 18-2 126.

McKee, M.D., Nanci. A., Landis, W.J., Gotoh, Y., Gerstenfeld, L.C., and Glimcher, M.J. (1990).

DevelopmentaI appearance and ultrastructural immunoIocalization of a major 66 kDa phosphoprotein in

embryonic and post-nata1 chicken bone. Anat. Rec. 228, 77-92.

Menton. D.N.. Simmons, D.J., Chang, S.-L.. and Orr, B.Y. (1984). From bone lining ceIl to osteocyte -

An SEM study. Anat. Rec. 209, 29-39.

Metsaranta, M., Young, M.F., Sandberg, M., Termine, J., and Vuorio, E. (1989). Localization of

osteonectin expression in human fetal skeIetal tissues by in situ hybridization. Calcif. Tissue Int. 45, 146-

152.

Michaelson, J. (1993). Cellular selection in the genesis of multicellular organization. Lab. Invest. 69.

136-151.

Middteton, J., Arnott, N., Walsh, S., and Beresford, J. (1995). Osteoblasts and osteoclasts in adult human

osteophyte tissue express the mRNAs for insulin-like growth factors 1 and II and the type 1 IGF receptor.

Bone 16,287-293.

Miller. E.J. (1 976). Biochernical characterization and biological significance of the genetically distinct

collagens. Mol. Cell. Biochem. 13. 165- 192.

Miller. S.C.. Bowman, B.M., Smith. J.M.. and Jee, W.S.S. (1980). Characterization of endosteal bone-

lining cells from fatty marrow bone sites in aduIt beagles. Anat. Rec. 198. 163-173.

Miller. S.C. and Jee. W.S.S. (1987). f h e bone Iining ceIl: A distinct phenotype? Calcif. Tissue Int. 4I, 1-

Mintz. K.P., Grzesik W.J.. Midura, R.J.. Robey. P.G.. Termine. J.D.. and Fisher, L W . (1993).

Purification and fragmentation of nondenatured bone sialoprotein: Evidence for a cryptic. RGD-resistant

ce11 attachment dornain. J. Bone Miner. Res. 8. 985-995.

Miyauchi. A.. Alvarez. J., Greenfield, E.M.. Teti, A.. Grano. M., Colucci. S., Zarnboriin-Zal!oae. A.,

Ross. F.P.. Teitelbaum. S.L.. Cheresh. D.. and Hruska. K.A. ( 1 991). Recognition of osteopontin and

related peptides by an a& integrin stimulates immediate ce11 signals in osteoclasts. J. Biol. Chem. 266.

20369-103 74.

Moore. M.A.. Gotoh. Y.. Rafidi. K.. and Gerstenfeld. L.C. ( 199 1 ). Charsrcterizriticn of a cDNA for

chicken osteopontin: Expression during bone development osteoblast diflerentiation. and tissue

distribution. Biochemistry JO. 250 1-2508.

.Morel. G.. Boivin. G.. David. L.. Duboid. P.M.. and Meunier. P.J. (1985). Imrnunoc~ochemicai

evidence for endogenous calcitonin and parathyroid hormone in osteoblasts from the calvaria of neonatal

mice: Absence of endogenous estradiol and estradiol receptors. Ce11 Tissue Res. 240. 89-93.

Muenke. M.. Schell. U.. Hehr. A.. Robin. N.H.. Losken. H.W.. Schinzel, A.. Pulleyn. LJ.. Rutland. P.,

Reardon. W.. Malcolm. S., and Winter. R.M. (1994). A common mutation in the fibroblast growth factor

receptor 1 gene in Pfeiffer syndrome. Nat. Genet. 8.269-274.

Mundlos. S.. Engel. H., Michel-Behnke. 1 .. and Zabel. B. (1 990). Distribution of type 1 and type II

collagen gene expression during the development of human long bones. Bone 11,275-279.

Mundlos, S., Schwahn, B.. Reichert. T., and Zabel, B. ( 1 992). Distribution of osteonectin mRNA and

protein during human embryoinc and fetal development. J. Histochem. Cytochern. 40,283-29 1.

Mundy. G.R. and Poser, J.W. ( 1983). Chernotactic activity of the gamma-carboxyglutamic acid

containing protein in bone. Calcif. Tissue [nt. 35, 164- 168.

Nakamura, H., Kenmotsu, S.. Sakai. H.. and Ozawa, H. (1995). Locaiization of CD44. the hyaluronate

receptor. on the plasma membrane of osteocytes and osteoctasts in rat tibiae. Ce11 Tissue Res. 280, 225-

333.

Nefussi. J.-R.. Boy-Lefevre, M.L.. Boulekbache. H., and Forest, N. (1985). Mineralization in vitro of

matrix formed by osteoblasts isolated by collagenase digestion. Differentiation 29. 160-1 68.

Ng, K.W.. Gumrner, P.R-, Michelangeli, V.P.. Baterman, J.F., Mascara, T., Cole. W.G., and Martin, T.J.

( 1988). Regulation of alkaline phosphatase expression in a neonatal rat calvarial ce11 strain by retinoic

acid. J. Bone Miner. Res. 3, 53-6 1.

Nijweide. P.J., Burger. E.H., and Feyen, J.H.M. ( 1986). Cells of bone: Proliferation, differentiation. and

hormonal regulation. Physiol. Rev. 66, 855-886.

Nijweide, P.J.. van der Plas, A., and Olthof. A.A. (1988). Osteoblastic differentiation. In Ce11 and

Molecular Biology of Vertebrate Hard Tissues. D. Evered and S. Harnett, eds. (Chichester, UK: John

Wiley & Sons), pp. 61-77.

Nijweide. P.J. and Mulder. R.J.P. ( f 986). Identification of osteocytes in osteoblast-like cultures using a

monoclonal antibody specificalty directed against osteocytes. Histochemistry 84, 343-350.

Ninomiya, J.T., Tracy. R.P., Carore, J.D.. Gendreau, M.A., Kelm, R.J., and Mann, KG. (1990).

Heterogeneity of human bone. J. Bone Miner. Res. 5,933-938.

Nissenson, R.A., Diep. D.. and Strewler, G.J. (1988). Synthetic peptides comprising the amino-terminal

sequence of a parathyroid hormone-like protein from human malignancies. Binding to parathyroid

hormone recepton and activation of adenylate cyclase in bone cells and kidney. J. Biol. Chem. 263,

12866- 1287 1.

Noda, M., Yoon, K., Thiede, M., Buenaga, R., Weiss. M., Henthorn, P., Harris, H.. and Rodan, G.

( 1987). cDNA cloning of alkaline phosphatase from rat osteosarcoma (ROS lV2.8) cells. 1. Bone Miner.

Res. 2, 161-164.

Nomura, S., Wills, A.J., Edwards, D.R., Heath, J.K., and Hogan, B.L.M. (1 988). Developmental

expression of 2ar (osteopontin) and SPARC (osteonectin) RNA as revealed by in situ hybridization. J.

Cell Eiol. 106.44 1-450.

Nose, K., Saito, H., and Kuroki, T. ( 1 990). Isolation of a gene sequence induced later by turnor-

promoting 12-O-TetradecanoylphorboI- 1 3-acetate in rnouse osteoblastic cells (MC3T3-E 1) and

expressed constitutively in ras-transformed ce1 1s. Cell Growth Differ. 1, 5 1 1-5 1 8.

Oldberg, A., Franzen. A., and Heinegard. D. (1986). Cloning and sequence analysis of rat bone

sialoprotein (osteopontin) cDNA reveals an Arg-Gly-Asp cell-binding sequence. Proc. Natl. Acad. Sci.

USA 83,88 19-8823.

Oldberg A.? Franzen, A., and Heinegard, D. ( l988a). The primary structure of a cell-binding bone

sialoprotein. J . Biol. Chem. 263, 19430- 19432.

Oldberg, A., Franzen. A., Heinegard, D., Pierschbacher. M., and Ruislahti. E. (1988b). Identification of a

bone sialoprotein receptor in osteosarcoma cells. J. Biol. Chem. 263. 1 9433- 1 9436.

Owen, M. (1963). Cell population kinetics of an osteogenic tissue. J. Cell Biol. 19, 19-32.

Owen, M. (1967). Uptake of ['HI-uridine into precursor pools and RNA in osteogenic cells. J. Cell Sci.

2,39-56.

Owen, M. (1985). Lineage of osteogenic cells and their relationship to the stroma1 system. Bone Miner.

Res. 3, 1-25.

Owen, T.A., Aronow, M., Shalhoub, V., Barone, L.M., Wilming, L., Tassinari, M.S., Kennedy, M.B.,

Pockwinse, S.. Lian, J.B., and Stein, G.S. (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. J. Cell. Physiol. 143,

420-430.

Owen, T.A., Aronow, M.S., Barone, L.M., Bettencourt. B., Stein, G.S., and Lian, J.B. (1991). Pleiotropic

effects of vitamin D on osteoblast gene expression are related to the proliferative and differentiated state

of the bone ce11 phenotype: Dependency upon basal levels of gene expression, duration of exposure, and

bone matrix competency in nomai rat osteoblast cultures. Endocrinology 128, 1496- 1504.

Partridge, N.C., Bloch, S.R., and Pearman, A.T. (1994). Signal transduction pathways mediating

parathyroid hormone regulation of osteoblastic gene expression. J. Cell. Biochem. H , 3 2 1-327.

Pfeilschifter, J., Krempien, R.. Naumass. A., Gronwald, G.K., Hoppe, J., and Zielger, R. (1992).

Differential effects of PDGF isofoms on plasminogen activator activity in fetal rat osteobIasts due to

isoform-specific receptor functions. Endocrinology 130, 2059-2066.

Pike. J.W., Sone, T., Ozono, K., Kesterson, R.A.. and Kemer, S.A. (1993). The osteocalcin gene as a

molecular mode1 for tissue-specific expression and 1,25-dihydroxyvitamin D3 regulation. In Cellular and

Molecular Biology of Bone. M. Noda, ed. (San Diego: Academic Press), pp. 235-256.

Pitaru, S., Kotev-Emeth, S., Noff, D., Kafhler, S., and Savion, N. (1993). Effect of basic fibroblast

growth factor on the growth and differentiation of aduit stroma1 bone marrow cells: Enhanced

development of mineralized bone-like tissue in culture. J. Bone Miner. Res. 8, 9 19-929.

Pockwinse, S.M., Wilming, L.G., Conlon, D.M., Stein, G.S., and Lian, J.B. (1992). Expression of cell

growth and bone specific genes at single ce11 resolution during development of bone tissue. J. Cell.

Biochem. 49, ; 10-323.

Pockwinse, S.M., Stein, J.L., Lian, J.B., and Stein, G.S. (1995). Developmental stage-specific cellular

responses to vitamin D and glucocorticoids during differentiation of the osteoblast phenotype:

Interrelationship of morphology and gene expression by in situ hybridization. Exp. Ce11 Res. 216,244-

260.

Poliard, A., Nifuju, A., Larnblin, D., Plee, E., Forest, C., and Kellerman, 0. ( 1995). Controlled

conversion of an immortalized mesodermal progenitor ce11 towards osteogenic, chondrogenic, or

adipogenic pathways. J. Ce11 Biol. 130, 146 1- 1472.

Poser, J.W. and Price, P.A. (1979). A method for decarboxylation of gamma-carboxyglutarnic acid in

proteins. 1. Biol. Chem. 254,43 1-436.

Potten, C.S., Schofield, R., and Lajtha, L.G. (1979). A comparison of ceIl replacement in bone marrow,

testis and three regions of surface epithelium. Biochim. Biophys. Acta 560,28 1-299.

Price, P. (1983). Osteocalcin. Bone Miner. Res. 1, 157- 190.

Price, P.A., Otsuka, A S . Poser, J. W., Kristaponis, J., and Raman, N. ( 1976). Characterization of a

gamma-carboxyglutamic acid-containing protein from bone. Proc. Natl. Acad. Sci. USA 73, 1447- 145 1.

Price, P.A., Williamson, M.K., Haba. T., Deli, R.B., and Lee, W.S.S. (1 982). Excessive mineralization

with growth plate closure in rats or chronic warfarin treatrnent. Proc. Natl. Acad. Sci. USA 79, 7736-

7738.

Price, P.A. (1989). Gla-containing proteins in bone. Connect. Tissue Res. 21, 5 1-60.

Prince, C.W., Oosawa. T., Butler, W.T., Tomana. M., Bhown, A.J., Bhown, M., and Schrohenloher, R.E.

( 1 987). Isolation, characterization and biosynthesis of a phosphorylated gl ycoprotein from rat bone. J.

Biol. Chem. 262,2900-2907.

Pritchard, J.J. (1972a). The osteoblast. In The Biochemistry and Physiology of Bone. G.H. Bourne, ed.

(New York: Acadernic Press), pp. 2 1-43.

Pritchard, J.J. ( I972b). Growth and di fierentiation of bone and connective tissue. In Differentiation and

Growth of Cells in Vertebrate Tissues. G. Goldspink, ed. (London: Chapman and Hall), pp. 101-1 28.

Raetz, C.R.H., Wermuth, M.M., McIntyre, T.M., Esko, J.D., and Wing, D.C. (1982). Somatic ce11

cloning in polyester stacks. Proc. Natl. Acad. Sci. USA 79,3223-3227.

Rao, L.G., Ng, B., Brunette, D., and Heersche, J.N.M. (1977). Parathyroid hormone and prostaglandin

E 1-response in a selected population of bone cells after repeated subculture and storage at -80°C.

Endocrinology 100, 1233- 124 1.

Rao, L.G., Kalliecharan, R., Wang, H.-M., Heersche, J.N.M., and Sodek, J. (1979). Specific

immunohistochemical localization of type 1 collagen in porcine periodontal tissues using the peroxidase-

labclled antibody technique. Histochem. J. 11, 73-82.

Rao, L.G., Murray, T.M., and Heersche, J.N.M. (1983). Irnmunohistochemical demonstration of

parathyroid hormone binding to specific cell types in fixed rat bone tissues. Endocrinology 113, 805-8 10.

Raynal, C., Delmas. P.D., and Chenu, C. (1996). Bone sialoprotein stimulates in vitro bone resorption.

Endocrinology 13 7,2347-2354.

Rodan, G.A., Heath, J.K., Yoon, K., Noda, M., and Rodan, S.B. ( 1 988). Diversity of the osteoblastic

phenotype. In Ce11 and Molecular Biology of Vertebrate Hard Tissues. D. Evered and S. Harnett, eds.

(Chichester, UK: John Wiley & Sons), pp. 78-9 1.

Rodan, G.A. (1995). Osteopontin overview. In Osteopontin: Role in Cell Signalling and Adhesion. D.T.

Denhardt, W.T. Butler, A.F. Chambers, and D.R. Senger, eds. (New York: New York Academy of

Sciences), pp. 1-5.

Rodan, G.A. and Noda, M. (1991). Gene expression in osteoblastic cells. Crit. Rev. Eukaryot. Gene

Expr. 1,85-98.

Rodan, G.A. and Rodan, S.B. (1983). Expression of the osteoblastic phenotype. Bone Miner. Res. 2,244-

285.

Rodan, S., Wesolowski, G., Thomas, K., and Rodan, G.A. ( 1987). Growth stimulation of rat calvaria

osteoblastic cells by acidic fibroblast growth factor. Endocrinology 12 1, 19 17- 1923.

Rodan, SB., Wesolowski, G., Thomas, G.A., Yoon, K., and Rodan, G.A. (1 989). Effects of acidic and

basic fibroblast growth factors on osteoblastic cells. Connect. Tissue Res. 20,283-288.

Ross, I L . , Browne, C.M., and Hume, D.A. (1994). Transcription of individual genes in eukaryotic cells

occurs randomly and infrequently. Immunol. Ce11 Biol. 72, 1 77- 1 85.

Rouleau, M.F., Mitchell, J., and Goltzman, D. (1988). In vivo distribution of parathyroid hormone

receptors in bone: Evidence that a predominant osseous target ceIl is not the mature osteoblast.

Endocrinology 123, 1 87- 19 1.

Rouleau, M.F., Mitchell. J., and Goltzman, D. (1 990). Characterization of the major parathyroid hormone

target ceIl in the endosteal metaphysis of rat long bones. J. Bone Miner. Res. 5, 1043- 10%.

Sambrook. J., Frisch, E.F.. and Maniatis, T. ( 1 989). Motecular Cloning, A Laboratory Manual (Cold

Spring Harbor, NY: Cold Spring Harbor Laboratory).

Sandberg, M.. Autio-Hamainen, H., and Vuorio, E. (1988a). Localization of the expression of types 1.

III, and IV collagen, TGF-PI and c-fos genes in developing human calvarial bones. Dev. Biol. 130, 324-

334.

Sandberg, M., Vuorio, T., Hirvonen, H., Alitalo, K., and Vuorio, E. (1988b). Enhanced expression of

TGF-P and c-fos rnRNAs in the growth plates of developing human long bones. Development 102,46 1-

470.

Sandberg, M. and Vuorio, E. (1987). Localization of types 1, II, and II! collagen rnRNAs in developing

human skeletal tissues by in situ hybridization. J. Cell Biol. 101, 1077- 1084.

Schatteman, G.C., Morrison-Graham, K., Van Koppen, A., Weston, J.A., and Bowen-Pope, D.F. (1 992).

Regulation and role of PDGF receptor a-subunit expression during embryogenesis. Development 115,

123-131.

Scott, B.L. (1967). Thymidine 3~ electron microscope radioautography of osteogenic cells in the fetal

rat. J. Ce11 Biol. 35, 1 15- 126.

Senger, D.R., Perruui, C.A., and Papadopoulos, A. (1989). Elevated expression of secreted

phosphoprotein 1 (osteopontin, Zar) as a consequence of neoplastic transformation. Anticancer Res. 9,

1291-1300.

Shen, V., Kohler, G.. Huang, J., Huang, S.S., and Peck, W.A. (1989). An acidic fibroblast growth factor

stimulates DNA synthesis, inh ib its collagen and alkaline phosphatase synthesis and induces resorption in

bone. Bone Miner. 7, 205-2 19.

Siddhanti, S.R. and Quarles, L.D. (1994). Molecular and pharmalogic control of osteoblast proliferation

and differentiation. J. Cell. Biochem. 55.3 10-320.

Silve. C.M., Hradek, G.T., Jones, A.L., and Arnaud. C.D. (1982). Parathyroid hormone receptor in intact

embryonic chicken bone: Characterization and ce1 Mar localization. J. Ce11 Biol. 9 4 3 79-386.

Smith, J.H. and Denhardt, D.T. (1 987). Molecular cloning of a tumor promoter-inducible mRNA found

in JB6 rnouse epidermal cells: Induction is stable at high, but not at low, cell densities. J. Cell. Biochem.

34, 13-32.

Smith, J.H. and Denhardt, D.T. (1989). Evidence for two pathways of protein kinase C induction of 2ar

expression: Correlation with mitogenesis. J. CeIl. Physiol. 139, 189-195.

Sodek, J., Zhang, Q., Goldberg, H.A., Domenicucci, C., Kasugai, S., Wrana, J.L., Shapiro, H., and Chen,

J. (199 1). Non-collagenous bone proteins and their role in substrate-induced bioactivity. In The Bone-

Biomaterial Interface. J.E. Davies, ed. (Toronto: University of Toronto Press), pp. 97- 1 10.

Sodek, J., Chen, J., Kasugai, S., Nagata, T., Zhang, Q., McKee, M.D., and Nanci, A. (1992). Elucidating

the functions of bone sialoprotein and osteopontin in bone formation. In Chemistry and Biology of

Mineralized Tissues. H. Slavkin and P. Price, eds. (Elsevier), pp. 297-306.

Somerman, M.J., Fisher, L.W., Foster, R.A., and Sauk, J.J. (1988). Human bone sialoprotein I and [I

enhance fibroblast attachment in vitro. Calcif. Tissue [nt. 43, 50-53.

Somerman, M.J., Prince. C.W., Butler, W.T., Foster, R.A., Moehring, J.M.. and Sauk, J.J. (1989). Cell

attachment activity of the 44 kilodalton bone phosphoprotein is not restricted to bone cells. Matrix 9,49-

54.

Sprague, L., Wettewiald, A., Heinzman, U., and Atkinson, M.J. (1996). Phenotypic changes following

over-expression of sense or antisense E 1 1 cDNA in ROS 1 712.8 cells. J. Bone Miner. Res. I IfSI), S 1 32.

Stanka, P. ( 1975). Occurrence of cell junctions and microfilaments in osteoblasts. Celi Tissue Res. 159,

4 13-422.

Stein, G.S., Lian, J.B., Gerstenfeld, L.C., Shalhoub, V.. Aronow, M.. Owen, T., and Markose, E. (1989).

The onset and progression of osteoblast di fferentiation is hnctionally related to cellular pro1 iferation.

Connect. Tissue Res. 0 , 3 - 13.

Stein, G.S. and Lian, J.B. (1 993). Molecular mechanisms mediating developmental and hormone-

regulated expression of genes in osteoblasts: An integrated relationship of cell growth and

differentiation. In Cellular and Molecular Biology of Bone. M. Noda, ed. (San Diego: Academic Press),

pp. 47-95.

Stein, G.S. and Lian, J.B. (1995). Molecular mechanisms mediating proliferatioddifferentiation

interrelationships during progressive development of the osteoblast phenotype: Update 1995. Endocr.

Rev. 4,290-297.

Stenflo, J. and Suttie, J.W. (1977). Vitamin K-dependent formation of gamma-carboxyglutamic acid.

Annu. Rev. Biochem. 46, 157- 172.

Sudo, H., Kodama, H.-A., Arnagai, Y., Yamamoto, S., and Kasai, S. (1983). In vitro differentiation and

calcification in a new clonsl osteogenic cell line derived from newborn mouse calvaria. J . Cell Biol. 96,

191-198.

Taylor, S.M. and Jones, P.A. ( 1979). Multiple new phenotypes induced in 10T1/2 and 3T3 ceIls treated

with hzacytidine. Cell 17, 77 1-779.

Tenenbaurn, H.C. ( 1 987). Levarnisole and inorganic pyrophosphate inhibit b-glycerophosphate-induced

rnineralization of bone formed in vitro. Bone Miner. 3, 13- 16.

Tenenbaum, H.C., McCulloch, C.A.G., Fair, C., and Birek, C. (1989). The regulatory effect of phosphate

on bone metabolkm in vitro. Cell Tissue Res. 257, 555-563.

Thiede, M.A., Smock, S.L., Petersen, D.N., Grasser, W.A., Thompson, D.D., and Nishimoto, S.K.

(1994). Presence of messenger ribonucleic acid encoding osteocalcin, a marker of bone turnover, in bone

marrow rnegakaryocytes and peripheral blood platelets. Endocrinology 135,929-937.

Towler. D.A., Rutledge. S.J., and Rodan, G.A. (1994). Msx-2/Hox 8.1: A transcriptional regulator of the

rat osteocalcin promoter. Mo!. Endocrinol. 8, 1484- 1493.

Trurnper, L.H., Brady, G., Bagg, A., Gray. D., Loke, S.L., Griesser, H., Wagman, R., Braziel, R.,

Gascoyne, R.D., Vicini, S., Iscove, N.N., Cossman, J.. and Mak, T.W. (1993). Single-ce11 analysis of

Hodgkin and Reed-Sternberg cells: Molecular heterogeneity of gene expression and p53 mutations.

Blood 81,3097-3 1 15.

Turksen, K.. Bhargava, U., Moe, H.K., and Aubin, J.E. (1992). Isolation of monoclonal antibodies

recognizing rat bone-associated molecules in vitro and in vivo. J. Histochem. Cytochem. JO, 1 339- 1 3 52.

Turksen, K. and Aubin, J.E. (1991). Positive and negative immunoselection for enrichment of two

classes of osteoprogenitor cells. J. Cell Biol. 114,373-384.

Urena, P., Lee, K., Weaver, D., Kong, X.F., Brown, D., Bond, A.T., Abou-Samra, A.B., and Segre, G.V.

( 1992). PTWPTHrP receptor mRNA expression as assessed by northem blot and in situ hybridization. J.

Bone Miner. Res. 7(Si), S 1 19.

van der Plas, A., Aarden, E.M., Feijen, J.H.M., De Boer, A.H., Wiltink, A., Alblas, M.J., De Leij, L., and

Nijweide, P.J. (1994). Characteristics and properties of osteocytes in culture. J. Bone Miner. Res. 9,

1697- 1 704.

van der Plas, A. and Nijweide, P.J. ( 1992). Isolation and purification of osteocytes. J. Bone Miner. Res.

7'. 389-396.

Wang, E., Wang, J.. Chin, E., Zhou, J., and Bondy, C.A. (1995). Cellular patterns of insulin-like growth

factor system gene expression in murine chondrogenesis and osteogenesis. Endocrinology 136,274 1-

275 1.

Weinger, J.M. and iioltrop, M.E. (1974). An ultrastructural study of bone cells: The occurrence of

microtubules, microfilaments and tight junctions. Calcif. Tissue Res. 14, 15-29.

Weinreb, M., Shinar, D.. and Rodan, G.A. (1990). Different pattern of alkaline phosphatase, osteopontin,

and osteocalcin expression in developing rat bone visualized by in situ hybridization. J. Bone Miner. Res.

5. 83 1-842.

Weiss, M.J.. Cole, D.E.C., Ray, K., Whyte, M.P.. Lafferty, M.A., Mulivor, R.A., and Harris, H. (1988).

A missense mutation in the human liverhonekidney alkaline phosphatase gene causing a lethal fonn of

hypophosphatasia. Proc. NatI. Acad. Sci. USA 85. 7666-7669.

Wetterwald, A., Hofstetter, W., Cecchini, M.G., and Fleisch, H. (1992). The rat caudal vertebrae as a

mode1 to study bone formation: Temporal expression of osteopontin, osteocalcin and of the E 1 1 antigen.

In Calcium Regulating Hormones and Bone Metabolism. D.V. Cohn, C. Gennari, and A.H. Tashjian, eds.

(Amsterdam: Elsevier), pp. 203-205.

Wettenvald, A., Hofstetter, W., Cecchini, M.G., Lanske, B., Wagner, C., Fleisch, H., and Atkinson, M.

(1 996). Characterization and cloning of the E 1 1 antigen, a marker expressed by rat osteoblasts and

osteocytes. Bone 18, 125- 132.

Whyte, M.P. (1 989). Hypophosphatasia. In The Metabolic Basis of Inherited Disease. C.R. Scriver, A.L.

Beaudet, W.S. Sly, and D. Valle, eds. (New York: McGraw-Hill), pp. 2843-2856.

Wong, G. ( 1980). Bone ceil cu[tures as an experimental model. Arthritis Rheum. 23, 108 1 - 1 086.

Wuthier, R.E. and Register, T.C. (1985). Role of alkaline phosphatase, a polyfbnctional enzyme, in

rnineralizing tissues. In The Chemistry and Biology of Mineralized Tissues. W .T. Butler, ed.

(Birmingham, AL: EBSCO Media), pp. 1 13- 124.

Yamaguchi, A. and Kahn. A.J. (199 1). Clonal osteogenic ceIl Iines express myogenic and adipocytic

developmental potential. Calcif. Tissue Int. 49, 22 1-225.

Yoon, K., Buenaga, R., and Rodan, G.A. (1987). Tissue specificity and developmental expression of rat

osteopontin. Biochem. Biophys. Res. Commun. 148, 1129-1 136.

Yotov, W.V., Glorieux, F.H., and St-Arnaud, R. (1995). A novel transcriptional coactivator linking the

basal transcriptional machinery to activating transcription factors is expressed specifically in bone during

embryogenesis. J. Bone Miner. Res. 1O(SI/, S 17 1.

Young, M.F., [baraki, K., Kerr, J.M., and Heegaard, A. (1993). Molecular and cellular biology of the

major noncollagenous proteins in bone. In Cellular and Molecular Biology of Bone. M. Noda, ed. (San

Diego: Academic Press), pp. 19 1-234.

Zemik, J., Twarog, K., and Upholt, W.B. (1990). Regulation of alkaline phosphatase and alpha 2 ( I )

procollagen synthesis during early intramembranous bone formation in the rat mandible. Differentiation

44,207-2 1 5.

Zhang, L., Leeman, E., Carnes, DL, and Graves, D.T. (1991). Human osteoblasts synthesize and

respond to platelet-derived growth factor. Am. J. Physiol. 261, C348-C3 54.

Zheng, M.H., Wood, D.J., and Papadimitriou, J.M. (1 992). Whatfs new in the role of cytokines on

osteoblast proliferation and differentiation? Pathol. Res. Pract. 188, 1 104- 1 12 1.

IMAGE EVALUATION TEST TARGET (QA-3)

APPLIED 2 i W G E , lnc - = 1653 East Main Street - -. - - Rocheçter. NY 14609 USA -- -- - - Phone: 71 6/482-0300 -- -- - - Fax: 71 6/288-5989

Q 1993. Appiied Image. lnc., Ail Rights Rese~ed