INFORMATION TO USERS

Th. qurlity af thh nproducüon ir depanâmnt upon thr qwlity of th. copy wbmitted. Broken or indistinct print, cd- or poor qwlity illustraüons

and phatographr, prlnt ùiwdtt~twgh, subsîandard margins, and impmpr

alignmt un dversely affect nprodudion.

In the unlikely avant (hst the author diâ not $end UMI a ampleta rnanuraipt

and them a n misring pages, thrr MI k notecl. Ako, 1 unauthorized

cwright material Mû to k mnoved, a no@ will indiate th deletion.

Photognphr i n d u in the odginal miinusdpt M e bwn reprodwrd

x«ogmphkally in thir a p y . Highrr qu8li(y C x 9" hck and white

photognphic prints are miiabb for n y photognphs or illustrations appeaiing in tM8 eopy for an additionil cham. Contact UMI dim(ly to order.

WI & Homll InformeCion and î.euning 300 NWUI Zûûû R a . Am A M , MI 481061346 USA

800.521-

ISOLATION AND CHARACTERIZATION OF VERY EARLY OSTEOPROGENITOR CELLS

Susan Allison Steckle

A thesis submitted in conformity with the requirements for the degree of Master of Science

Graduate Departnient of Anatomy and Cell Biology University of Toronto

8 Copyright by Susan Allison Steckle (1997)

National Library Bibliotheque nationale du Canada

Acquisitions and Acquisitions et BiMiographic Ssniices services bibliographiques

The author has granted a non- L'auteur a accordé une licence non exclusive licence aiîowing the exclusive permettant à la National Libraiy of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sel1 reproduire, prêter, distribuer ou copies of this thesis in rnicrofom, vendre des copies de cette thèse sous paper or electronic formats. la forme de microfiche/fiîm, 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 fiom it Ni la thèse N des extraits substantiels may be p ~ t e d or otherwise de celle-ci ne doivent être imprimés reproduced without the author' s ou autrement reproduits sans son permission. autorisation.

ISOLATION AND CHARACTERIZATION OF VERY EARLY OSTEOPROGENITOR CELLS

Master of Science, 1997.

Susan Allison Steckle Graduate Department of Anatomy and Ce11 Biology University of Toronto

The purpose of ihis work was to use a previously established technique. poly(A)-

PCR. on freshly plated primary cells to aid in the characterization of very early

osteoprogenitor cells. Establishing a molecular profile representative of the

osteoprogenitor phenotype will help in the identification of morphologically indistinct

osteoprogenitor cells prior to their overt morphological differentiation. In the rat calvarial

(RC) cell mode1 of osteoblast differentiation. RC cells proliferate and differentiate in

vitro over a time span of 21 - 25 days, culminating in the formation of mineralized bone

nodules. Very early (day 4 to 6 aller plating) in the culture period, individual colonies

arising from single cells were identified microscopically, and a cellular sample (1 -3 cells)

was micrornanipulated fiom each colony. This cellular sample was used to produce a

poly(A) cDNA library, while the remaining cells in the colonies were left undisturbed and

allowed to differentiate. After bone-forming colonies had been retrospectively identified,

their respective cDNA libraries were used for Southem blotting for the presence of

messages for bone matrix proteins and other putative osteoprogenitor markers.

Osteoprogenitors were compared with fibroblastic cells h m similarly identified colonies

that did not differentiate to bone nodules. There was intercellular heierogeneity in the

expression profiles of the very early osteoprogenitor cells, which expressed al1 or some

combination of osteopontin, osteonectin, type 1 collagen and type III collagen; none

expressed bone sialoprotein. alkaline phosphatase or osteocalcin. Fibroblastic cells were

found to express some combination of type 1 collagen, osteopontin, and osteonectin; only

one sarnple expressed alkaline phosphatase. Results suggest that the fibroblast and

osteoprogenitor phenotypes express several cornmon markers. This technique will aid in

the fùrther molecular characterization of early cells in the osteoblast lineage.

iii

ABSTRACT ................................................................................................................................................. ................................................................................................................................ TABLE OF CONTENTS iv

LIST 01: FIGUES .......................................................................................................................................... v ........................................................................................................................... LISI- OF ABBREVIATIONS vi

OBJECTIVE ................................................................................................................................................... I ...................................................................................................................................... ~NTRODUCTION 1

I . Histology of Bone Cclls ..................................................................................................................................... I 7 .............................................................................................................................................................. Figure I -

1.. Ostroblast Diffrrentiation .............................................................................................................................. 4 .. kigure 2 ............................................................................................................................................................. 5 III . DiffcrentiotionMarkers .................................................................................................................................. 7

.................................................................................................................................................. Osteopontin 7 ................................................................................................................................................. Osreoncctin 9

........................................................................................................................................... Collagen type 1 I O Coltagen type I I 1 ...................................................................................................................................... I I Osteocalcin ............................. .. ......................................................................................................... 13

.................................................................................................................................... Bonc Sialoprotein 1 4 ................................................................................................................................. Alkaline Phosphatase 15

L32 ........................................................................................................................................................... 16 IV . Poly(A) PCR and Low Density Cultures .................................................................................................... 16

..................................................................................................................................................... V . Objectives 1 8 ............................................................................................................................................... MET~IODS 19

Ccll Culture ............................................................................................................................................... 19 Single Csll Isolaiion .................................................................................................................................. 19

............................................................................................................................................. Poly(A)-PCR 2 0 7 7 Southem Analysis ................................................................................................ ........................................ -

............................................................ ............................... cDNA Probes .. 2 3 .............................................................................................................................................. Data Analy sis 24

......................................................................................................................................... R ~ s u ~ r s -26 Figure 3 ............................................................................................................................................................ 27 Table I ............................................................................................................................................................. 29 Figurc 4 .......................................................................................................................................................... 30 Figure 5 ............................................................................................................................................................ 32 Figure 6 ............................................................................................................................................................ 33

............................................................................................................................................... Discussio~ 36 I . The Poly(A)-PCR Approach ............................................................................................................................ 37

Tcchnical Limitations ................................................................................................................................. 37 Rcproducibility and Deiection Limits ......................................................................................................... 39

..................................................................................................... . II Phenotypc of Early Osteoprogcnitor Cclls 40 .................................................................................................................. Phenoiypic Heterogeneiîy 42

..................................... III . Rctrospective Analysis of Single RC Cells .. ............................. 44 SUMMARY AND CONCLUSIONS .................................................................................................................. 47

............................................................................................................................................. REFERENCES 48

Proposed lineage diagram with differentiation steps for cells of the osteoblast lineage.

Ostcoblast lineage diagram showing known expression profiles for each stage in the lineage.

Representative RC culture dishes stained by the Von Kossa technique showing mineralized nodules and other designated colonies.

Composite of Phosphorimager data for molecular markets tcsted on each sample.

3-Dimensional graph of al1 markers (as a ratio to total cDNA), showing relative expression levels.

Graphs of individual molecular markers for each sarnple tested.

a-MEM ALP BSA BSP cDNA COLL-1 COLL- I II dATP, dA dCTP dGTP dNTP DTT dTTP, dT ECM FGF-R kd M-MLV m mRNA OB OCN OD ON OP OPN PCR PDGF-R PTH-R RC RGD RT SDS SPARC SSC TBE

a-minimum essential medium alkaline phosphatase bovine serum albumin bone sialoprotein - complementary deoxyribonuclcic acid collagen type 1 collagen typc III 2'-deoxyrib adenosine triphosphate 2'-deox yribo c ytidine triphosphate 2'-deoxyribo guanosine triphosphate 2'-deoxyribonuc leotides dithiothrtitol Z'deoxyribo thymidine triphosphate extracellular matri x fibroblast growth factor nceptor kilodalton Moloncy Murine Leukemia V i w mi llimcûe messenger ribonucleic acid ostcoblast o s t d c i n optical density units osttonectin osteoptogenitor osteopontin polymcrase chah reaction plateletderived growth factor nceptor parathymid hormone rwcptor rat calvarial argininc-gl ycine-aspartate reverse mnscription sodium docecyl sulfate secretcd pmtein acidic and rich in cystcine sodium chloridt/sodium cime b& tris-boratc/EDTA electrophorcsis bufkr

The purpose of this work was to identiQ retrospectively and characterize very

emly osteoprogenitor cells in the rat calvarial (RC) ce11 mode1 of osteoblast

di fferentiation.

During osteoblast differentiation, morphologically indistinct osteoprogenitor cells

proliferate and differentiate into morphologically identifiable preosteoblasts, then

osteoblasts and osteocytes. An increasing number of molecular markers are known to be

up- and dom-regulated during this differentiation sequence, and describing their

expression profiles aids in characterization of a phenotype for these cells as they progress

fiom less mature to more mature cells. In this Introduction, 1 will describe various

aspects, including histological and molecular features, of the osteoblast lincage.

1. Hbtology of Bone Celh

The cumnt mode1 of bone ce11 development is a linear dif'ferentiation sequence

h m the mesenchymal stem ce11 to cornmitteci osteoprogenitor to pmsteoblast to newly

diffcrentiated osteoblast to mature osteoblast to osteocyte as depicted in Figure 1 (Owen

et al, 1985; Aubin et al, 1993, 1995).

Mesenchymal E a r l ~ Late PreOB/ OB Osteocyte Stem Cell OP OP Early OB

Figure 1: A proposed lineage diagram with differentiation steps for cells of the osteoblast lineage. Adapted h m Aubin et al, 1993. (OP: osteoprogmitor; OB: osteoblast)

Although its location in vivo is uncertain and its molecular profile not yet

established, evidence for the existence of a mesenchymal stem ce11 includes the isolation

of clonal rnultipotential ce11 lines with capacity to differentiate into various mesenchymal

tissues including muscle, fat, cartilage and bone (reviewed in Aubin et al, 1993). By

definition, the osteoprogenitor is considered commitîed to the osteoblast lineage. At least

some osteoprogenitors are thought to reside in the periosteum of bone and the stroma1

compartrnent of bone marrow. The spindle-shaped cells resembling fibroblastic cells and

with plump oval nuclei and abundant cytoplasm in growing bone are thought to be

osteoprogenitors (Scott, 1967). The osteoprogenitor cells in the fibrous periosteum are

thought to have undergone little differentiation; they are morphologically

indistinguishable fiom fibroblastic cells and it is not possible to identiQ them on the basis

of their microscopic appearance alone.

Osteoblasts, on the other hand, have a distinct cuboidal or polygonal morphology

whm actively foming bone. They have a basophilie cytoplasm that is rich in rough

endoplasmic reticulum, and they line the bone surface while laying dom bone matrix.

Preosteoblasts resemble osteoblasts morphologically and are Iocated behind the layer of

osteoblasts on the bone surface (farther away h m the newly developing osteoid).

Osteocytes are the terminally diffetcntiated ce11 of the osteogenic lineage. They are

flattened, stellate-shaped cells that lie in lacunae in bone. As such, they arc completely

surroundcd by the bone maûix, and have long cytoplasmic processes that extend through

canaliculi in bont to aid in communication with other cells and ensure nourishment.

Bone lining cells are thought to be inactive osteoblasts that have flattened and

elongated dong bone surfaces where neither formation nor rcsorption is occurring. Bone

lining cells communicate amongst themselves and with neighbouring osteocytes through

gap junctions.

II. Osteoblast Differentiation

Osteoblasts are bone-fonning cells which are readily identified by their distinct

cuboidal morphology and their location on the bone surface during bone formation.

During osteogenesis, they actively synthesize a variety of bone matrix proteins such as

collagen, osteopontin, SPARCIosteonectin, osteocalcin, bone sialoprotein, as well as the

membrane-bound enzyme alkaline phosphatase (reviewed in Aubin et al, 1993, 1995).

Expression of these proteins has helped in the molecular identification of osteoblasts.

Early osteoblasts and preosteoblasts have also been extensively characterizcd recently.

For exarnple, they express many of the same proteins that mahue osteoblasts express, but

there is marked heterogeneity in the expression profiles at the single ce11 level at al1 these

nlatively mature stages (Liu et al, 1994). Individual cells stageci as early osteoblast,

pmsteoblast or osteoprogenitor have varied expression patterns of the repertoire of bone

matrix proteins as well as cytokine receptors, such as PDGF-R, FGF-R and PTH-R (Liu

et al, 1994). The known expression profiles for the osteoblast lineage are depicted in

Figure 2, and will be individually discussed below.

Mesenchymal Earl~ Late Pre-OB/ OB Osteocyte Stem Ce11 OP OP Earl y OB

MARKERS

Osteopontin ?

Collagen type 1 ?

Ostcocalcin ?

Alkaline Phosphatase ?

Bone Sialoprotein ?

Figure 2: A proposed lineage diagram with differentiation steps for cells of the osteoblast lineage. Known expression profiles for each stage in the osteoblast lincage are shown. Adapted fmm Aubin et al, 1993,1995

LEGEND: 3 Not known - - - - - - - - Not preseni ----------------- Present at low or heterogenous levels Present Present at high levels

Whereas the mature osteoblast phenotype is considered reasonably well-

characterized, recently. more attention has been directed towards the delineation of the

osteoblast differentiation sequence and questions related to osteoblast development. It is

unknown how many discrete steps or stages exist as a putative mesenchymal stem ce11

diflerentiates into a cornmitted osteoprogenitor, and on to a mature osteoblast. Reasons

why this differentiation sequence has been difficult to characterize include the inability to

distinguish morphologically earlier cells in the lineage and the lack of molecular markers

for the early cells and transition points. Nevertheless, limiting dilution analysis has shown

that the osteoprogenitor is present at a very low frequency, approximately 1 in 300 cells,

in the RC ce11 population (Bellows and Aubin, 1989). one mode1 cornmonly used to study

osteoblast differentiation in vitro. The RC cell population is obtained by enzymatically

digesting rat bone tissue to yield a mass population including cells at different stages of

differentiation, such as more mature osteoblasts and fibroblasts, as well as

osteoprogenitors and other ce11 types with high proliferative potential (Bellows et al,

1990). Although their fiequency in the population is very low, the presence of the

osteoprogenitor is defined by its proliferative capacity and the fact thnt its progeny have

the ability to differentiate into bone nodule forming osteoblests in viîro. Their low

fiequency has also contributed to the difficulty in ncognizing osteoprogcnitor cells.

III. Differentiation Markers

Osteoblasts express many different molecules, but some of the better studied ones

are extracellular matrix molecules that the osteoblast lays down in the bone matrix, such

as types 1 and III collagen, osteopontin, osteonectin, bone sialoprotein, osteocalch, and

the membrane-bound enzyme alkaline phosphatase. As osteoblasts develop, more of these

molecules are upregulated, and are thus used as markers of osteoblastic differentiation.

The following sections will describe the current state of characteriuition of these markers,

their tissue distribution, putative fùnctions and their expression profiles throughout the

osteoblast lineage. The final section will address L32, not as a marker of osteoblastic

differentiation. but as a general marker of the presence of amplified cDNA fiom each of

the cells analyzed.

Osteopontin

Osteopontin (OPN) is a sialic-acid rich, 44 kd phosphorylated, ca2+-binding

glycoprotein that contains an ArgGly-Asp (RGD) adhesion motif (Oldkrg et al, 1986).

It was originally purified fiom bone as sialoprotein 1 (Franzen and Heinegard, 1985;

Fisher et al, 1987) or 44 kd bone phosphoprotein (Prince et al, 1987), and h m other

cells as 2ar (Smith and Denhardt, 1987). Although OPN was originally purified fiom

bone, it has sincc been identified in non-mineralizing tissues in such diverse otgans and

tissues as kidney (Yoon et ai, 1987; Lopn et ai, 1993; Giachelli et al, 1994; Shiraga et

al, 1992; Worcester et al, 1992). imer ear (Lopez et al, 1995; Swanson et ai, 1989),

utew, decidu, placenta and metrial glands (Young et ai, 1990; Nomura et al, 1988;

Waterhouse et al, 1992), artenal smooth muscle (Giachelli et al, 1991, 1993; Liaw et al,

1995), and epithelium of several organ systems (Brown et al, 1992). It is also expressed

in numerous tumour ce11 types and cells stimulated to proliferate by treatment with

tumour promoters (Craig et al, 1988; Khokha et al, 1991).

Since it is secreted by osteoblasts? and expressed in osteoclasts and osteocytes.

OPN is considered to be important in both mineralization and resorption of the bone

matrix (Reinholt et al, 1990; Weinreb et al, 1990; Flores et al, 1992; Tezuka et al, 1992;

Chen et al, 1993; Sodek et al, 1992) though a definitive huiction has not yet been

discovered. One possibility, suggested by the fact that OPN is found

imrnunohistochemically at the mineralization front in developing bone, is that OPN may

influence the rate of rnineralization rather than nucleation of hydroxyapatite crystal

growth (McKee el al, 1990; Sodek et al, 1992; nviewed in Denhardt and Guo, 1993). A

role in resorption has also been proposed (Reinholt et al, 1990; Ross et al, 1993). OPN is

reporteci to associate with fibronectin (Nemir et al, 1989; Singh et al, 1990), type 1

collagen (Chen et al, 1992) and osteocalcin (Ritter et al, 1992). It is also regarded as a

ceIl attachrnent factor, as it promotes the adhesion and spreading of osteoblasts,

osteoclasts, mesenchymai cells, fibroblasts, nontransformecl calvarial ce11 lines,

osteoblast-like osteosarcorna cells and many transfomiad fibroblastic ce11 lines

(Somerman et al, 1987, 1989; Oldberg et al, 1986), and is a ligand of the CD44 receptor

(Weber et al, 1 996).

OPN immunoreactivity has k e n found in osteoblasts and osteocytes of both

endochondral and membranous bones, as well as in fibroblast-shaped cells, refemd to as

preosteoblasts, which were not in contact with the bone itself but in osteogenic sites

(Mark et (11, 1987a). During the osteoblast differentiation sequence, OPN typicaily

appears prior to other bone matrix proteins, including bone sialoprotein (BSP) and

osteocalcin (Mark et al, 1 987% 1 987b, 1 988; Stein et al, 1 989; Owen et al, 1 990; Chen et

al, 199 1 a; Moore et al, 199 1 ; Pockwinse et al, 1 992; Liu et al, 1994).

Osteonectin

Osteonectin (ON) is a 32 kd phosphorylated glycoprotein that can simultaneously

bind to hydroxyapatite and type 1 collagen (Termine et al, 1981). ON is identical to

SPARC (Secreted Protein Acidic and Rich in Cysteine; Mason et <II, 1986a), to a 43 kd

protein secreted by epitheliai cells in culture (Sage et al, 1984) and to BM-40, a protein

produced by the murine Englebrah-Holm-Swarm basement membrane tumor (Dziadek et

al, 1986). While ON is expressed in osteoblasts (Holland et al, 1987) and odontoblasts

(Fujisawa et al, 1989, Reichert et al, 1992) and is abundant in bone of the axial skeleton

and skull (Termine et al, 198 1, Nomura et al, 1988, Holland et d, 1987), it is also present

in a variety of non-mineralizing tissues. SPARC was described as a product of the

parietai endoâerm cells during mouse development (Bolander et al, 1988, Holland et al,

1987, Mason et al. 1986b). ON has a widespread dishibution in otha tissues, particularly

those undergohg rapid proliferation and matrix production, as well as tissues involvcd in

the processes of rnindization and steroid production (Mundlos et al, 1992, Holland et

al, 1987).

While some investigators have proposed a role for ON in the process of

mineralization (Jundt et al, 1987, Tennine et al, 1981), its distribution in non-

mineralizing tissues (Holland et al, 1987, Nomura et al, 1988, Wasi et al, 1984, Bolander

et al, 1988, Dziadek et al, 1986, Sage et al, 1989% Mundlos et al, 1 W2), and its

inhibition of hydroxyapatite-seeded crystal formation in vitro (Doi et al, 1989, Romberg

et al, 1985) argue against a specific role for ON which is limited to the initiation of

mineralization (Sage et d. 1989a, Holland et al, 1987). Although ON is clearly linked to

proliferation in fetal tissues (Mundlos et al, 1 W2), it has also been shown to inhibit ce11

spreading (Sage et al, 1989b, Lane and Sage, 1990). Of particular interest to the present

study, ON is expressed in human marrow-derived osteoprogenitor cells (Long et al, 1995)

as well as in spindle-shaped cells, thought to be newly-recruited osteoblastic precursors,

probably at an intermediate stage between the so called "detemined osteoprogenitor

cells" (Friedenstein 1973) and already completely developed active osteoblasts (Jundt et

al, 1987).

Collagm @pe I Type 1 collagen (COLL-I) is the most abundant of the bone matrix proteins, and it

is prcvalent in other tissues such as sicin, tendon, ligament, cornea and most other

connective tissues as well. Fibrillar or fibril-forming collagens, which include both type 1

and type III, are similar in size, containhg large triple-belid domains of about 1000

amino acids per chah Type 1 collagen is a trimeric molecule composed of two a l(1)

chains and one a2(1) chain.

COLL-1 is the major structural component of the bone matrix. It can fom cross

links with other collagen types, particularly type II1 in bone (Cheung et al, 1983), and it

associates with OPN (Chen et al, 1992) and BSP (Fujisawa et al, 1995). Given its wide

distribution throughout the body, COLL-1 is not cxpected to play a rolc in the nucleation

of hydroxyapatite formation during mineralization of the bone matrix, although

hyàroxyapatite crystals do sit in the hole zone regions of collagen fibrils once the

mineralization process has begun (reviewed in Glimcher, 1984).

In some studies, expression of COLL-1 has k e n shown to be initially high, then

gradually falling to lower levels as osteoblasts mature (Stein et al, 1989; Gentenfeld el

al, 1988; Owen et al, 1990, 1991), while in others it is found to increase as osteoblasts

mature (Liu et al, 1994), more consistent with results of in situ hybridization which show

high COLL-1 expression in mature osteoblasts including both younger and older

osteoblasts (Heersche et al. 1 992).

Colfagen @pe 111 Type III collagen (COLL-III) is a homotrimer of three identical collagen al(II1)

chains which can becorne rapidly cross-linked through the formation of inttrmolecular

disulphide bonds (Cheung et al, 1983). There is some controversy regarding COLL-III

expression in bone and bone cells. While there is relatively littîe data in the literatute

describing COLL-III expression in osteoblasts and other cells in theu lincage, many have

used low or lack of COLL-III expression as indicative of osteoblast populations free of

fibroblast contamination (Robey and Termine, 1985). Indeed, some have stated that

"osteoblasts produce exclusively type 1 collagen" (Rodan and Noda, 199 1). However,

others report the pnsence of small amounts of COLL-III in the organic matrix of bone

(Niyibizi and Eyre. 1989; Keene et al, 1988), particularly in discrete fiber bundles

throughout the bone cortex, mainly concentrated at the Haversian canal surface and the

bone-periosteal interface (Keene et al, 1991), although whether of osteoblast or other ce11

origin is not clear. Mouse odontoblasts express COLL-III (Lukinmaa et al, 1993). COLL-

III is the major collagen of the fibrous matrix that foms dong the periosteal surface

(reviewed in Ashhurst, 1990) and is expressed in the upper demis, as well as in the

fibroblasts of the periosteurn and fibrous mesenchyme between bone spiculas (Sandberg

et al, 1988). Various subpopulations of osteoblast-like cells isolated fiorn chick calvaria

produced only COLL-1 in vitro, while early digestion cells with a more "tibroblast-like"

appearance produced COLL-I as well as COLL-III (Wiesîner et al, 1981). Bone nodules

formed Ni vitro fiom fetal rat calvarial cells immunolabel for COLL-III (Bellows et al,

1986). Also, COLL-III expression has been demonstrated in mouse osteoblast-like cells

(Scott et al, 1980). clonal osteoblastic lines fiom fetal rat calvaria (Aubin et al, 1982),

chick embryo osteoblests (Majmudar et al, 1991), and in cultures of 7-day old rat tibia1

ceils (Stringa et al, 1995). The alkaline phosphatase-rich rat osteosarcorna ce11 line ROS

3 712.8 was shown to synthesi~ approh te ly 98% COLL-1, while a more 'bfibroblastic"

ce11 line h m the seme tumot, ROS 291, produced 7096 COLLI and 304b COLL-III (P.

Bernstein, unpublishcd observation; discussed in Rodan and Rodan, 1983).

Osteocalcin

Osteocalcin (OCN), also known as bone gla protein, is a 6 kd vitamin-K

dependent matrix protein that contains yîarbxyglutarnic acid and binds to ca2+ and

hydroxyapatite (Gundberg et d, 1984; Hauschka et al, 1975). OCN is synthesized by

osteoblasts and odontoblasts in bone and dentin, mspectively, and, while largely

restricted to mineralizing tissues (Bronckers et al, 1 985. 1 987; 1 keda et al, 1992; Ohta et

rrl, 1989), recently Thiede et al (1994) reported osteocalcin expression in

megakaryoc ytes.

While generally thought to be first expressed with the onset of mineralization

(Broncken et al, 1985, 1987; Groot et al, 1986; Gerstenfeld et al, 1987; Mark et al,

1 W b , 1988; Stein et al, 1989; Boivin et al, 1990; Owen et al, 1990, 1991 ; Pockwinse et

al, 1992), single ce11 poly(A)-polymerase chain reaction (PCR) (Liu et al, 1994) and

immunohistochemistry studies (Bronckers et al, 1987) have demonstrateci that young

osteoblasts are positive for OCN prior to detectable mineralization. A definitive fùnction

has not been discovered for OCN, however in vitro evidence shows thac OCN inhibits the

transition from brushite to hydroxyapatite (Romberg et al, 1986) which suggests that

OCN may act as a negative regulator of bone mineralization. Since OCN associates with

OPN in vitro (Ritter et al, 1992), and OPN may act as a ce11 attachrnent factor for

ostaoclasts, OCN may play a role in bone resorption (reviewd in G e h n Robey, 1989).

However, recent studics on mice lacking OCN expression support a role for OCN in

growth plate development, bone formation and cortical thickness (Desbois et al, 1995).

Botte Sioloprofein

Bone sialoprotein (BSP) is a 33-34 kd protein that undergoes extensive ps t -

translational modifications to make it approximately 75 kd (Oldberg et al, 1988; Ecarot-

Chamer et al, 1989; Fisher et al, 1990; Zhang et al, 1990). BSP contains an RGD-ce11

attachent motif, as well as several stretches of polyglutamic acid that may be involved

in hydroxyapatite binding (Oldberg et al, 1988) since BSP is able to nucleate

hydroxyapatite crystal formation (Hunter and Goldberg, 1993). BSP also binds to

collagen (Fujisawa et al, 1995) and cnhances attachent and spreading of gingival

fibroblasts in vitro (Somennan et al, 1988).

BSP expression was first thought to be restncted to bone tissue (Oldberg et al,

l988), but it has bem found in trophoblasts in placenta (Bianco et al, 1991) and recently

in tumour cells metastasizing to bone (Bellahcéne et uf, 1996). Its expression is

associated with new bone formation by differentiated osteoblasts (Chen et ai, 1199 b). It

has been proposeci that the appearance of the ability to produce BSP marks the boundary

between the preosteoblast and osteoblast stages of osteoblast development (Bianco et al,

1993) and its expression ciearly p d e s that of osteocalcin (Liu et al, 1994). BSP

expression increascs with the development of nascent bone nodulcs in rat bone marrow

stroma1 ce11 cultures, and declines in older mineralizing nodules (Malaval et al, 1994).

Alkaline Phosplratose

Alkaline phosphatase (ALP) is a characteristic product of osteoblasts (Rodan and

Rodan, 1983). There are three alkaline phosphatase isozymes (placenta1 (Kam et al,

1985); intestinal (Henthom et al, 1987; Berger et al, 1987); and tissue non-specific

isozyme which is highly expressed in bone, liver and kidney (Weiss et al, 1986)) that

mise From three separate genes. ALP catalyzes the hydrolysis of monoester phosphates. at

a pH optimum of 8 to 10. The enzyme is linked to the extracellular ce11 surface by a

phosphatidyl inositol linkage (Noda et al, 1987) and is removed by cleavage with

phosphatidyl inositol-specific phospholipase C (Fedde et al, 1988; Turksen and Aubin,

199 1 ). Though its physiological function in bone remains uncertain, it is thought that

ALP is cleaved from osteoblasts and circulates in the bloodstream during penods of

active osteogenesis, allowing senim ALP activity to be used as an indicator of bone

formation and turnover (McComb et al, 1979). ALP activity in bone cells has been

reported to change with the ce11 cycle, dropping during G2 and M phases, rising during G ,

and peaking during S phase (Fedarko et al, 1990). By histochemistry and biochemical

analysis of some cells lines, ALP activity is found to be high in osteoblasts and young

osteocytes and low in more mature osteocytes (Doty and Schofield, 1976; Rodan and

Rodan, 1983).

ALP expression is thought to appear quite early and clearly pnor to the onset of

mineralization in differentiating osteoblastic cells, tending to increase as osteoblasts

mature then decrease when mineralization is well progressed (Bronckers et al, 1987;

Aronow et al, 1990; Owen et al, 1990; Turksen and Aubin, 1991 ; Mark et al, 1987a).

Studies on single cells and isolated colonies confimed early onset of ALP expression;

ALP increased as colonies became osteogenic and acquired mineral, but rnRNA levels

did not decrease relative to other markers of osteoblast differentiation during the time

period investigated (Liu et al, 1994).

We sought a specific molecule to serve as a general control for the amplification

process. An appropriate control would be a molecule that is known to be present at

relatively stable levels, regardless of ce11 cycle stage or cellular maturity. L32 is a

ubiquitous protein that is an integral component of the ribosome in both prokaryotic and

eukaryotic cells, and has a poly(A) tail. making it suitable for amplification by Poly(A)-

PCR. Thus, L32 appeared to fit the criteria for a control of the amplification process.

although its signal intensity varied more markedly than anticipated in cornparison to total

cDNA signal intensity. It was thetefore not used to nomalize the data for cornparisons

(total cDNA signal intensity was used for this purpose - see Methods/Data Analysis).

IV. Poly(A) PCR and Low Density Cultures

Work already done in ow lab has characterized several early stages in the

osteoblast lineage with a novel technique. When fetal RC cells are plated at very low

densities, single isolated colonies arise from single cells, whose differentiation fate can be

followed, Le. to colonies of osteoblastic, fibroblastic, or other mesenchymal ce11 types.

Since one osteoprogenitor ce11 gives rise to one bone nodule with the characteristics of

woven bone under standard RC culture conditions (Bhargava et al, 1988), and since

colonies are derived from clonal, normal (i.e.: non-established, non-transformed) cells, it

can be concluded that each ceIl in the colony had the same osteoprogenitor ceIl as an

ancestor. By combining a replica plating technique to these low density cultures and by

using poly(A)-PCR, this lab has been able to molecularly characterize some

osteoprogenitors, preosteoblasts and early osteoblasts; osteoprogenitor stages investigated

included ones well before the cells had acquired the morphological characteristics of

osteoblastic cells and prior to upregulation of ALP and the matrix molecules discussed

above (Liu and Aubin, 1994). The technique involved placing a polyester cloth over the

forming individual colonies. Some cells fiom each colony attached to the cloth after

several days and division cycles; the cloth was then removed, placed in its own culture

dish, and incubated at 37OC. Meanwhile, the original, or master dish, was placed in an

incubator at a lower temperature (generally 30°C) to stall the growth of the cells until the

colonies growing on the replica cloth had matureci. Once bone colonies had formed

mineralized bone nodules on the cloth, their sister colonies were identified by matching

their locations on the dish with that on the cloth. Either single cells or whole colonies

were collected by trypsinizing the colonies, and the cells were subjected to molecular

analysis by poly(A)-PCR. This approach allowed characterization of stages wlier in the

osteoblast lineage than was ever pnviously possible (Liu and Aubin, 1994; Aubin et al,

1995; Liu et al, 19%). however the cells were removed h m their in vivo ages by 8-1 l

days and several divisions in vitm. It was my objective to adapt this technique to

characterize still earlier cells in the osteoblast lineage, removed fiom their in vivo ages by

only 3 to 5 divisions in vitro.

V. Objectives

The purpose of this work was to retrospectively identify and characterize very

early osteoprogenitor cells in the RC ce11 mode1 of osteoblast differentiation. In this

approach, RC cells were plated at low density and a cellular sarnple was taken fiom each

colony after 4 to 5 days of growth. This cellular sample was used to produce a poly(A)-

PCR-based cDNA library, while the remaining cells in the colonies were allowed to

proliferate and differentiate in vitro for 21 - 25 days, during which time mineraiized bone

nodules formed in some colonies. Once bone-fonning colonies were retrospectively

identified, the cDNA libraries were used for Southem blotting to establish a molecular

profile based on the presence of messages for bone matnx proteins and other putative

osteoprogenitor marken.

Cell Culture Fetal RC cells were obtained by sequential enzymatic digestion of 21-day fetal

Wistar rat calvariae as previously described (Bellows et al, 1986, 1987). Briefly,

calvariae were cleaned of skin and connective tissue, minced and sequentially digested in

collagenase (Sigma Chemical Company. Si. Louis, MO). 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 1 5% heat-inactivated fetal bovine serum (Imrnunocorp, Montreal,

QC), as well as penicillin G (100 pg/ml; Sigma), gentamycin (50 pg/ml; Life

Technologies, Grand Island, NY) and huigizone (0.3 pg/rnl; Life Technologies). After 24

hours, attached cells were washeâ with phosphate-buffered saline to remove unattached

cells and other debris, then collected by trypsinization and an aliquot counted on an

electronic Coulter Counter. Cells were then plated at very low densities (500 to 1000 cells

per 100 mm dish) and grown in a-MEM medium (as above) supplemented with ascorbic

acid (50 pglml; Fisher Scientific, Fair Lam, NI), sodium P-glycerophosphate (10 mM;

Sigma Chemical Co.) and dexarnethasone (10 nM; Sigma Chemical Co.) as described

(Liu et al, 1994). Medium was changed every 2-3 days. Dishes were incubated at 37'C in

a humidifid atrnosphere of 95% air, 5% CQ.

Single Celf Isolation Four to five days aftcr plating, incipient colonies comprising 8 to 50 cells were

located in phase conûast micmscopy and thtù locations markecî. Each colony was

assigned an arbitrary number for later identification. A Gilson P2 micropipette was used

to manually scrape a small number of cells (1-3) fiom each of the colonies; the cell

sample was drawn up into the pipette in 1 )il of medium. The remaining cells in each

colony were left intact. The micromanipulated cells were released directly into lysis

buffer in a 0.5 ml Eppendorf tube and kept on ice. Lysis buffer consisted of 50 mM Tris

HCI, pH 8.3, 75 mM KCI, 3.0 m M MgCl,, 0.50% Nonidet P-40 (Sigma), 2 FM each in

dATP, dCTP, dGTP, and d'ITP (Boehringer Mannheim), 2000 u/ml RNAguard

(Pharmacia), 1 00 dm1 Inhibitlice (5'-3' Incorporated), and 100 ng/ml [dTIz4 cDN A

primer. When al1 cells from growing colonies had been sampled, the mRNA was reverse

transcribed and amplified by poly(A)-PCR (see below), and stored at -70°C.

Cells remaining in the individual colonies were grown to maturity, generally 21 -

25 days, in medium as above, with medium changes each 2-3 days. The cells were then

fixed and stained by the Von Kossa technique to distinguish mineralized bone colonies

fkom colonies comprising fibroblastic cells.

Po&(A)-PCR First-strand cDNA synthesis and poly(A)-PCR were performed essentially as

described ( B d y et al, 1990, Liu et al, 1994). First-strand cDNA synthesis was

perfonned within one hour of ce11 micromanipulation. Samples were heated to 6S°C for 1

minute, cooled to roorn temperature for 3 minutes, then put on ice untilO.5 FI of M-MLV

reverse transcriptase (200 uni&/@; Life Technologies Ltd.) was dded. Samples were

incubatcd at 37OC for 15 minutes, thcn the enzyme was heat-inactivated a 6S°C for 10

minutes. Following first strand synthesis, a poly(dA) tail was added to the 3' end of the

cDNA with terminal transferase. 4 pl of 2x tailing buffer (consisting of 200mM

potassium cacodylate (pH 7.2), 4 m M CoCI,, 0.4 mM DTT, and 200 pM dATP) and 1 pi

(25 units) of teminal transferase enzyme (Boehnnger Mannheim) were added; samples

were incubated for 15 minutes at 37*C, then heat-inactivated for 10 minutes at 65°C. The

entire mRNA population was then amplified by PCR using primen with a poly(T)

stretch. The sequence of the PCR primer used was 5' ATG TCG TCC AGG CC0 CTC

TGG ACA AAA TAT GAA 'l'TC 3' followed by a stretch of 24 dT residues. A l x Taq

buffer mixture was preparcd, consisting of 10 mM Tris HCI (pH 8.3). 50 mM KCI, 1.5

rnM MgCl,, 100 pg/ml bovine serum albumin (BSA), 1 mM each dATP, dCTP, dGTP,

and dTTP, 0.05% Triton X-100, and a 0.1 ODZm units per sample of PCR primer. 50 pl

of this l x Taq buffer mixture was aliquoted to each sample, plus 1 pI of Taq DNA

Polymerase (5 units/pl; Sangon Limited Canada, Scarborough. Canada). Samples were

overlaid with mineral oil (Peikin-Elmet Cetus) and arnplified for 25 cycles of 1 minute at

94T, 2 minutes at 42OC, and 6 minutes at 72OC, followod by an additional 25 cycles of 1

minute at 94°C' 1 minute at 42OC. and 2 minutes at 72OC, followed by a 10 minute

extension at 72'C. Samples were then stored at -70°C. Samples useâ for Southem

blotting underwent a secondary PCR amplification. 0.5 pl of the original PCR-amplifid

sample was diluted to 100 pl with lx Taq buffer, as above but 0.2 rnM in dNTPs, and

0.06 units of PCR primer and 0.5 pl Taq DNA polymerase per sample. No mineral

oil was used for secondary amplifications. Samples werc amplified foi 25 cycles of 1

minute at 94"C, 1 minute at 42OC, and 2 minutes at 72OC, followed by a 10 minute

extension at 72°C. Samples were stored at -70°C.

Southern Ana&sis Following the secondary poly(A)-PCR amplification, selected samples were used

for Southern blots. Before preparing Southem blots, jpi of each of the poiy(Aj-PCR

arnplified cDNA samples were stained with ethidium bromide, electrophoresed through a

1.5% agaroseI0.5X TBE gel and viewed under ultraviolet light. Those samples that were

subjectively judged to have very little to no cDNA present relative to other samples were

excluded From further analysis (results not show). Next, Southem blots were prepared

from the remaining selecied samples. 5 pl of secondary PCR products were

electrophoresed approximately 3cm in a 1.5% agarose gel in 0 . 5 ~ TBE. The gels were

denatured in a solution of 1.5 M NaCVO.5 N NaOH, neutralized in a solution of 1 M

Tris/1 .5 M NaCI, and DNA transferred to a nylon membrane (0.2 Fm; ICN Biotrans) by

capillary transfer. Membranes were baked at 80°C for 2 hours and blots were probed with

specific cDNA probes produced from the 3' end of cDNA clones for various mRNA

messages (sec below), as previously described (Liu et al, 1994). Briefly, cDNA probes

were labeled with [ 3 2 ~ ] d ~ ~ ~ using an oligolabelling kit (Phannacia, Uppsala, Sweden).

All prehybridizations and hybridizations were perfomed at 42OC. AAer hybridization, the

blots were washed once for 30 minutes each at 42°C and 50°C in 2x SSC/O.i% SDS, and

once for 15 minutes each at 55°C and 60°C in lx SSC/O.l% SDS. Results of Southern

blotting were visualized by Phosphorimager or x-ray film (Kodak X-OMAT).

Phosphorimager data were quantitated and analyzed by IPLab Gel Software.

cDNA Probes The cDNAs comprising the 3' sequences were prepared and provided by F. Liu in

the lab. Rat al COLL-1 (palR.2) (Genovese et d.. 1984; kindly provided by Dr. D.

Rowe, Farmington, CT) was a 900-bp cDNA Pst1 fragment containing the entire 3'

noncoding 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; gift of Dr. G. A. Rodan, Merck Sharpe,

and Dohme Research Laboratories, West Point. PA) was a 600-bp cDNA EcoRi fragment

obtained by digesting pRAP54 with BssHII-XhoI to remove 1.8 kb of the 5' region and

religating the blunt ends. Rat OPN (kindly provided by Dr. R. Mukhejee, Montréai,

Qutbec) was a 700-bp cDNA BomHI-Ecoiü hgrnent obtained by digesting full length

cDNA with PvuI to remove 800 bp of the 5' region and ligating the blunt ended fragment

into Sniol cut pGEM-7Zf(+) vector (Promega, Madison, W?). Mouse SPARCION (Mason

et al, 1986a; kindiy pmvided by Dr. B. Hogan, London, üK) was a 600-bp cDNA

EcoRVHindII fragment obtained by digesting pG43 with AvaI and SmuI to remove 1400

bp of the 5' ngion and nligating the blunt ended fragment of the cut pGEM-1 vector

(nomega). Rat COLL-III (Glurnoff et al, 1994; kindly provided by Dr. E. Vuorio, Turku,

Finland) was a 500-bp cDNA fragment obtained by digesting pRGRS with BamHI to

mnove 1700 bp of îhe 5' region. Rat BSP was a partial cDNA containing 500 bp of 3'

region isolated with BSP-specific pnmcts h m a Agt 1 1 library pmpared h m RC ails

forming bone nodules (prepared by A. Gupta and J.E. Aubin). Rat OCN was a partial

cDNA containing 350 bp of 3' region isolated with OCN-specific primers from a hgtl 1

library prepared from ROS 17/2.8 cells (prepared by A. Gupta and J.E. Aubin). Rat L32

was a partial cDNA containing 400 bp of 3' region isolated with a mouse L32 probe

(Meyuhas and Peny. 1980; kindly provided by Dr. N.N. Iscove, Ontario Cancer Institute,

Toronto, Canada) from a Agt 1 1 library prepared from RC ceils forminy bune nodules.

Total cDNA probe was prepared as described by Sambrook et al (1989) from poly(~) '

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.

Data Analysis Phosphorimager data were recorded and images analyzed by IPLabGei Software

for Macintosh. A rectangle (37x30 pixels) was drawn on the most intensely labeled

portion of each lane of the blots and the intensity of the image inside the rectangles was

digitally recorded. A background value was taken from the negative (cell-free) control

lane. This background value was subtracted from the values obtained for sample lanes.

Data were expressed graphically as a ratio of the southem blot labeling intensity to the

total cDNA labeling intensity for each marker examined. Ratios were plotted in arbitrary

Expression data were subjectively judged for each marker examined such that

visi bl y discernible signals were designated as king expressed for that particular sample.

In order to have a consistent lower limit of detection, the ratio of the signal intensity of a

marker to the total cDNA value for a sample had to be greater than or equal to 10% of the

corresponding mature OB sarnple ratio (which was the highest ratio for al1 markers

except OPN and by far the most intense signal for each marker). Those markers whose

ratios were less than 10% of the mature OB ratio were designated as not expressed, or not

significantiy difkrent fiom background. Cornparison of du subjective visual inspection

and the values deteminrd in this manner showed that they agreed well.

When RC cells were plated at low density, nascent colonies were easily identified

at day 4 or 5 of the culture period, and their locations recorded. Samples comprising I to

3 cells were micromanipulated fiom the colonies and subjected to poly(A)-PCR. When

the remaining cells in each colony were allowed to continue growth for 21 to 25 days, a

few bone colonies were identified aller staining by the Von Kossa technique and non-

bone colonies aller staining with Bromophenol Blue. Two representative culture dishes

stained by the Von Kossa technique, each with one bone colony, are shown in Figure 3;

circles on the bottom of the dish mark the locations of other colonies fiom which cells

were also sampled. Samples 2 and 4 through 9 were recovered h m colonies that

subsequently formed mineralized bone nodules; samples 3 and 10 through 14 were

recovered frorn colonies that also formed bone which had not mineralized by the time of

fixation. Samples 15, 17. and 18 were recovered h m larger (greater than 500 cells)

colonies comprising cells of fibroblastic morphology, while samples 16, 19, and 20 were

fiom smaller (under 100 cells) colonies of non-osteogenic cells. In three cases, two

samples were taken h m the same colony (samples 10 and 1 1; 12 and 13; 19 and 20).

Southem blots were prepared with poly(A)-PCR amplified cDNA h m both the

samples repmenting early osteoprogenitor cells and other early precursor cells (here

designated as fibmprogenitors based on morphology of the resuliing cells). Controls that

were also included on the blots were a sarnple that was processed without any ceils

prcsent, and a sample amplifieà h m mRNA h m a mature, mineralid osteoblast

Figure 3: Two represmtative culture dishes stained by the Von Kossa technique, each with one bone colony (arrow). Circles on the bottom of the dish mark the locations of other colonies fiom which cells were sarnpled.

colony (kindly provided by F. Liu). The lane with no ce11 present for the poly(A)-PCR

amplification was used as a negative control lane, and was subtracted as the background

reading for quantitation by the IPLabGel program. The mature mineralized osteoblast

colony lane served as a positive control for al1 of the markers tested by Southem blotting.

Relative levels of cDNA loading on Southem blots were compared by ethidium bromide

staining, L32 expression, and a total cDNA 3' probe prepared as in Mcthods. The

expression level of specific markers was qwtified by detennining the ratio of the

intensity of the scanned gel for that marker over the intensity of total cDNA labeling

(data show in Table 1).

The Southem blot (Figure 4) and comsponding nomalized data from the

Phosphorimager (Figures 5 and 6) indicate that OCN, ALP, and BSP were either

undetectable or expressed at very low levels in ail samples except for the positive control,

the sample arnplified fiom a mineralized, mature osteoblast colony. Only one sample

(sample 16) showed expression of any of these markers (ALP) at a level above

background. Levels of expression of other marken were variable. OPN and ON were

expressed at moderate levels in most samples tested, with high expression in a few

sarnples. For example, OPN was detectable in al1 osteoprogenitor samples tested, and at

quitc high levels in some samples (samples 6,7,9, 1 1, 13 and 14). Expression was lower,

but detectable, in most fibroblastic sampks as well. Expression of OPN and ON were, on

average, higher in osteoprogenitor cells than in fibmprogenitor cells. COLL-1 was

expressed at moderate to high levels in a few osteoprogenitors (3 ,S to 7,9 and 1 1). and at

low levels in several other samples (samples 8, 13 and 14). COLL-III was e x p t e d at

high levels in 2 osteoprogenitor samples (samples 7 and 4). and at low levels in two other

Table I : Ratio of marker signal intensity to total cDNA signal intensity for each sample. Numbers in bold print indicate sarnples with expression levels above background (designated as greater than or equal to 10% of the ratio for sample 1 ). These data were used to plot Figures 5 and 6.

Fi yre 4: Composite of al1 Phosphorimager data. Each column represents the same sample; data in each row have the same exposure time to the Phosphorimager cassette. Sample 1 : mature osteoblast colony; samples 2- 14: osteoprogenitor samples; sarnples 1 5- 20: non-osteogenic samples.

COLL- III

SPARC ON ~~~ . .. ..; , , ,*A.

umpie v i l w / total CONA

Figure S: 3-dimensional representation of nomaiized Southern bloning data showing I

expression levels of dl markers. Sample 1: mature osteoblast colony; samples osteoprogenitor samples; samples 1 5-20: non-osteogenic samples.

dative 2- 14:

Figure 6: Individual graphs of normalized Southem blotting data for each marker. Data is expressed as a ratio of marker expression to total cDNA for the same sample.

Figure 6 conthued ...

osteoprogenitor samples (samples 5 and 8). No fibroprogenitor samples had COLL-III

expression at a level above baseline.

These data indicate a great deal of intercellular heterogeneity in both

osteoprogenitor and fibroprogenitor cells. However, five general osteoprogenitor

phenotypes emerged: cells with (1) OPN expression only (samples 10 and 12); (2) OPN

and ON expression (sample 2); (3) OPN, ON and COLL-1 expression (samples 3, 6, 9,

11. 13, 14); (4) OPN, ON, and COLL-III expression (sample 4); and (5) OPN, ON,

COLL-1 and COLL-III expression (samples 5, 7 and 8). Four general fibroprogenitor

phenotypes also emerged: cells with (1) ON expression only (samples 19 and 20); (2) ON

and OPN expression (sarnple 18); (3) OPN, ON and COLL-I expression (samples 15 and

1 7); and (4) ON. OPN, COLL-I and ALP expression (sample 16).

This is the first report of the isolation and partial characterization of determined,

normal (i.e., non-established, non-transfomieci), early (recovered at day 4-5, division

number 4 to 6 hom in vivo lifetime) osteoprogenitor cells.

The population of cells isolated fiom 21-day fetal rat calvaria is a mixed

rnesenchymal population which consists of cells at many varying stages of

differentiation. There are osteoprogenitor cells, cells more advanced or differentiated in

the osteoblast lineage, as well as cells associated with colonies designated here as

fibroblastic based on their spindle-shaped to pleiomorphic morphologies and the fact that

they did not fom bone. By enzymatically digesting fetal rat calvaria, the extracellular

matrix is effectively removed, and, once the cells are plated, any remaining debris and

non-adherent cells are removed as well. This makes it much easier to select a cellular

sample fiom the population for molecular analysis than if the cells were housed in vivo,

although, when isolated and plated in this manmr, al1 spatial information about the origin

of the cells is lost. In this regard, the procedure intmduced hem is a way to obtain cells

removed h m their in vivo age by only 4 to 6 ce11 divisions in vitro, and to study low

kquency osteoprogenitor cells whose pnçisc in viw location is unlnown. Since discnte

colonies arise from single cells, the sibling cclls to the ones that are selected for analysis

will be of clonal origin, and their fate can be followed to detexmine the lineage of the

early cells.

1. The Poly(A)-PCR Approach

Both the duration of the reverse transcription reaction and the concentration of the

deoxyribonucleoddes used for the poly(A)-PCR approach used were limited to restrict the

length of the resulting cDNA to approximately 300 to 800 base pairs. This allows the

relative sequence abundances to be preserved in the subsequent PCR amplification by

preventing selection against long sequences or those with extensive secondary structures.

Although it is now possible to amplify sequences of up to 35 kb in length (Bames. 1994),

the current poly(A)-PCR protocol was employed for the following reasons: it has been

established to work on rat calvarial cells in this laboratory (Liu et al, 1994); it is more

economical considering the large numbets of samples that are processed at each time; and

it is sufficient in that it produces a 3' cDNA library which can be used for Southem

blotting to assess expression patterns in carly osteoprogenitor cells. The technique has

been used to obtain representative cDNA libraries for analysis of hematopoetic cells

(Brady et al, 1990, 1995), pre- and post-implantation stage embryonic tissues (Rambhatla

et al, 1995, Varmuza and Tate, 1992), and recently, in our lab, for the analysis of both

early and mature osteoblasts and preosteoblasts (Liu et al. 1994) and osteoprogeniton

identified by replica plating (Liu and Aubin, 1994).

Tech nicul Limitations It has been documented that the poly(A)-PCR technique is very sensitive to slight

variations in amplification conditions (Brady and Iscove, 1993.). This sensitivity is due in

part to the minute quantities of RNA serving as a template for the amplification reaction,

and the non-specificity of the technique. Several factors may have contributed to the

observed variations in ampli fication efficient y. The material being examined was

amplified from proliferating single cells, which results in varying message levels

depending on what stage of the ce11 cycle the cells are in at the time of isolation. This is

one variable that is beyond our control as we are unable to determine before selecting an

individual ce11 what stage it is at. Another variable that is dificuit to control is c d

adhesion. When cells were picked from sparsely populated young colonies, generally

only ceIl was isolated at a time. However, in more confluent colonies cells had often

made very strong contacts amongst themselves, and it is very likcly that more than one

ce11 was selected for analysis. This may also affect the amplification reaction efiiciency

since there is more RNA to serve as a template for the reaction. Several samples were

excluded from the final analysis because they did not ampli@ well. as assessed by

ethidium bromide staining, total cDNA and L32 signal intensities. If the total cDNA and

L32 signal intensities were very low to not discemible when subjectively comparing them

relative to other samples on the same Southern blot, the sample was excluded from

further analysis. Since some cellular samples were more efficient1 y amplified than others

by the poly(A)-PCR technique, uneven arnounts of cDNA, for each sample relative to

others, were loaded on Southem blots. Also, the material king examined was amplified

from proliferative single cells and we had found earlier that no known message including

several genes considered housekeeping was present at a steady-state in al1 sarnples (Liu

and Aubin, unpublished data) to allow direct cornparison of cDNA quality between

samples. Therefore, we used total cDNA as a standard against which al1 samples were

normalized as a ratio to compare relative expression levels between samples.

Reproducibiliîy and Detection Limits

The results presented in the current study are based on digital Southem blotting

data obtained with a phosphorimager system. Ideally, one would prepare one Southem

blot to be repeatedly probed with the individual marken in succession, in order to ensw

that amounts of cDNA on the blot were precisely consistent for al1 markers tested.

However, one limitation of the technique is that despite stringent stripping methods.

small amounts of radioactive probe remain on the blots after attempts to refiesh the blots

for subsequent probing; these minute amounts of radioactivity are detectable by the

extremely sensitive phosphorimager system. As a result, a separate Southem blot was

used for each individual marker that was examined. To assess the reproducibility of

sample loading in these individual blots, four separate blots were prepared and probed for

COLL-III. A consistent pattern of COLL-III expression was observed when these

multiple blots were compared. Similarly, OPN was repeatedly tested on four separate

blots, also yielding a consistent pattern of expression among the samples examined.

The data obtained when testing markers multiple times were not averaged for

statistical analysis since the raw values varied greatly fiorn blot to blot, while relative

differences between samples remained consistent. Factors that affected this variability

include the cassette exposure time. the "fieshness" of the radioisotope at the time of

labeling the blots, and the background labeling of the blots.

As introduced in Methods, marker signals were subjectively judged by comparing

signals with the overall background of the blots and with other samples. This subjective

determination of expression was followed with a numerical analysis of one blot for each

market. The lower limit of signal intensity that designated a signal as representing

expression of a marker was set arbitrarily to 10% of the positive control (mature OB)

sample, afier first normalking the signal to standardizc: difkrences in the aiiiount of

cDNA loaded per sample. Due to the subjective nature of the analysis, there are some

samples whose expression pattern may be different had the baseline expression level been

set to 5%, for exarnple, rather than 10% of the control. However, the 10% limit chosen

gave agreement between visuai inspection and the Phosphorimager values.

The data generated from the blots already have the background (the signal

intensity of the negative control lane, the cell-fiee reaction products) subtracted From each

sarnple value. It is of note that al1 sample lanes resulted in positive values, indicating that

every lane. even those with no expression of the marker of interest, had some background

level of signal detectable by the highly sensitive phosphorimager system.

II. Pbenotype of Early Osteoprogenitor Cells

My data indicate that early osteoprogenitor cells express combinations of COLL-1

and COLL-III, OPN and ON. That these cells did not express BSP, OCN or ALP,

markers characteristic of more mature osteoblasts, is consistent with and confirms their

immature stage. Notably, however, there is considerable heterogeneity in the expression

profiles of the markea that are expressed by the early osteoprogenitor cells, from cells

expressing al1 of COLL-I, COLL-III, OPN and ON (Le.: sarnple 7). to ones expressing

different combinations of one, two or three markers at diverse levels. This heterogeneity

will be discussed further below.

Of particular note is the expression of COLL-III. Among non-osteogenic or

fibroprogenitor cells examined, COLL- I II was not expressed. However, severai

osteoprogenitor sarnples showed moderate to high levels of COLL-III expression

(samples 4, 5, 7 and 8). Reports that COLL-III mRNA is detected by in situ hybridization

and imrnunohistochemistry in fibroblastic cells of the periosteum (Sandberg et al, 1988).

and that COLL-III is the major collagen of the fibrous matrix that forms dong the

periosteal surface (Keene, 1991; reviewed in Ashurst 1990) lend credence to the

designation of COLL-III as a marker of early osteoprogenitor cells. The high level of

COLL-III expression found in the mature osteoblast sarnple is consistent with a few

reports (see Introduction) of COLL-III in bone nodules formed by fetal RC cells (Bellows

et al, 1986), and in certain clonal osteoblastic lines h m fetal RC (Aubin et al, 1982) and

adult rat tibia1 osteoblastic populations (Strhga et al, 1995). It also suggests that

detection of COLL-III expression in sorne isolated osteoblastic populations is as

consistent with the presence of immature osteoprogenitoa as with fibroblastic

contamination, to which it has oAen been atûibuted (Robey and Termine, 1985).

The various phenotypes that have emergcd here for both osteoprogenitor and

fibroprogenitor cells are compriscd of different combinations of the 4 markers used

(COLL-1, COLL-III, OPN, ON). However, it is of note that only two phenotypes are

common to cells of both colony types; coexpression of OPN and ON and of OPN, ON

and COLL-1 was seen in both some osteoprogenitors and some fibroprogeniton. These

cornmon phenotypes may be representative of cells earlier or more primitive (less

differentiated) than tlie other cells exarnined here. This possibility is consistent with the

fact that the colonies resulting from these particular fibroprogeniton (samples 15, 17 and

18) became very large, indicative of extensive proliferative capacity and that few of the

bone colonies from these particular osteoprogenitors had yet mineralized by the time of

fixation (only samples 2, 6 and 9 had mineralized; samples 3, 11, 13, 14 had not),

suggesting that they may have arisen from more immature osteoprogenitors.

Phenorypic Heterogenee A gteat deai of intercellular heterogeneity is evident in both osteoprogenitor and

fibroprogenitor colonies. Duplicate samples pickeâ fiom the same colony, such as

samples 10 and 1 1, and 12 and 13 (these pairs h m two separate bone colonies), and

samples 19 and 20 h m a colony that did not fom a bone nodule, are informative. One

explanation for the phenotypic heterogeneity observeâ here is non-synchronous

differentiation progression for sibling cells within colonies, in other words, cells wen

picked at different stages in their progression through the diffeientiation sequence. A

second possibility is that even cells at the same stage of differentiation may have some

flexibility in both the repertoin of genes expressed and their levels of expression. Liu et

al, (1994; Liu and Aubin, 1994) found support for this by both the poly(A)-PCR

technique and by imrnunocytochemistry with sntibodies against bone markers, including

ALP, BSP and OCN. The technique 1 have used for ce11 isolation is biased to examine

cells at an early osteoprogenitor stage, i.e. afkr only 4 to 6 divisions in vitro. Within the

specific window on which these techniques focus, there are discrete developmental stages

as well. Some colonies, when picked, had as few as 5 to 15 cells while others had as

many as 50 to 80 cells, despite having been in vitro for the same length of time and

despite having the same end resuit. bone nodule formation. These differences, whiçh may

reflect progenitor cells of different developmental or proliferative lifetimes, could also

account for some of the heterogeneity in the expression pattern observed.

One further point that should be made with regard to intercellular heterogeneity is

that of sample size. When, for exarnple, 10 cells are selected fiom a colony for molecular

analysis by poly(A)-PCR. al1 of the messages expressed in those 10 cells will be shown in

the phenotype, regardless of whether al1 10 cells or just one ce11 is expressing a particular

message. On the other hand, during single ce11 analysis, it is not an average phenotype

that is s h o w but a specific phenotype of a cell at a particular stage in its differentiation

sequence, which could account for intercellular differences. Some of the observed

heterogeneity may be a result of this averaging effect, as more than one ceIl was obtained

by micromanipulation in some cases.

My data are limited by the relatively small number of ce11 samples available for

analysis. It would be of interest to repeat the selection of cells in multiple independent

experiments and determine whether the apparently separate phenotypes are reproducible

in every ce11 isolate examined.

III. Retrospcctive Analysis of Single RC Cells

Osteoprogenitor cells proliferate in vitro for several rounds of cell division

resulting in bone nodules whose sizes Vary. Cells fiom such developing bone colonies

(sarnples 2-14), the focus of this study, express COLL-1 and -III, OPN and ON to varying

degrees. Arnongst possible cxplanations for thc intcrccllular variations observed (see

above) some rnay result from osteoprogenitor cells being more or less mature in the

osteogenic lineage relative to othen at the time of sampling. While al1 osteoprogenitor

samples examined here ultimately give rise to bone nodules in vitro, some bone colonies

at the time of fixation were very heavily mineralized (samples 2, 5, 6, 8 and 9), others

were sparsely mineralized (samples 3, 4, 7, 10, 1 l), and othen had no detectable

mineralization (samples 1 2- 14). These devcloprnental differences could reflect that these

osteoprogenitor cells were at varying stages of differentiation when isolated. However,

not al1 bone nodules are identical in size. Whether size reflects relative developmental

maturity (Le. large colonies result fiom very immature progenitors) is not currently

known. Data h m Liu et al (in pnparation) and Aubh (in preparation) suggest this may

not be so and that colony sizc and differentiation progression rnay instead reflect a

stochastic process.

Thm are also more mature cells of the osteoblast lineage in the RC population.

These cells, while they have a roughly fibroblastic or pleiomorphic appearance when

initially isolated i~ vitro, would be expected to prolifemte to only a limited extent in viîro

but to express markers consistent with a relatively ma- phenotype. An example of this

type of cell may be sample 16. The colony from which this sample came formed a colony

of only approximately 80 cells by day 21 of culture. Yet, the cell isolated at 4 days in

vitro already expressed the more mature osteoblast marker ALP to a small degree, more

than any other immature cellular sample examined during these experiments. In fact, it

was the only sample, apart from the mature osteoblast colony which served as a positive

control, to have any of the more mature matkers (ALP, BSP and OCN) detectably

expressed. The expression of this more mature marker, along with the very small size of

the resulting colony, suggest that sarnple 16 may have been an early preosteoblast at the

time of sampling. This is in accordance with other work currently king performed in this

lab, in which some cells isolated by replica plating express ALP and other more mature

markers, but do not yet have the morphologid characteristics of osteoblasts (Liu et al.

manuscript in preparation).

A third ce11 type that cm be described in the RC population is one in which cells

are undifferentiated or uncomrnitted to the osteoblast limage or immature enough that

they will be very proliferative in vitro, i.e. they give rise to very large so-called

fibroblastic colonies. Examples of thesc are samples 15, 17 and 18. None of these cells

expressed detectable levels of ALP, BSP, OCN or COLL-III, but they did express ON,

OPN and COLL-1 to varying degrces.

These studies have provided somc initial steps to allow molecular chanicterization

of early osteoptogenitor cells and preparation of corresponding cDNA libraries chat can

be used to investigate novel genes and markers of this stage in the osteoblast

differentiation scqucnce. The techniques can be extended to investigate expression of any

markers whose 3' sequences are known by repmbing the Southem blots or n-ampliQing

the cDNA to prepare new blots. ln faci, the potential application of the technique in this

regard is limited only by the number of probes that can be obtained with a full 3'

sequence, since cDNA material can be reamplified many times.

SUMMARY AND CONCLUSIONS

1. I t is possible to isolate early osteoprogenitor cells by plating RC cells at low density

and manually micromanipulating cells fiom the culture dish.

2. Within the limitations set by the small sample size available, early osteoprogenitor

celis isolated in this marner are morphologically fibroblastic or pleiornorphic and

exhibit one of five molecular phenotypes:

0 expression of OPN only

a expression of ON and OPN

expression of ON, OPN and COLL-1

a expression of ON, OPN and COLL-III

a expression of ON, OPN, COLL-1 and COLL-III

3. Within the limitations set by the small sample size available, early cells associated

with colonies that did not fom bone nodules in vitro that are morphologically

fibroblastic or pleiomorphic also exhibit one of four molecular phenotypes:

expression of ON only

a expression of ON and OPN

0 expression of ON, OPN and COLL-1

a expression of ON, OPN, COLL-1 and ALP

4. It is possible to examine cells in the osteoblast lineage at earlier cimes in vitro than

previously possible. This work will make a signifiaint contribution to our cumntly

limited knowledge of very eatly osteoprogenitor cells.

Aronow MA, Gerstenfeld LC, Owen TA, Tassinari MS, Stein GS and Lian JE3 Factors that promote progressive development of the osteoblast phenotype in cultured fetal rat calvaria cells. J. Cell. Physiol. 143:2 13-22 1 , 1990.

Ashhurst DE. Collagens synthesized by healing fractures. Clin Orthop. 255:273-83, 1990.

Aubin JE, Liu F, Malaval L, and Gupta AK. Osteoblast and chondroblast differentiation. Bone. 1 7:77s-83s, 1 995.

Aubin JE, Turksen K. and Heersche M M . Osteoblastic cell lineage. in. of Bo@ (M. Noda. Ed.), pp 1-45. Academic Press. San Diego.

1993.

Aubin JE, Heersche JN, Merrilees MJ and Sodek J. Isolation of bone ce11 clones with differences in growth, hormone responses, and extnicellular matrix production. J. Cell. Biol. 92:452-46 1, 1 982.

Auffiay C and Rougeon F. Purification of mouse immunoglobulin heavy-chain messenger RNAs h m total myeloma tumor RNA. Eur. J. Biocheni. 107:303-3 14. 1980.

Bames WM. PCR amplification of up to 35-kb DNA with high fidelity and high yield from lambda bacteriophage templates. Proc. Natl. Acad. Sei US. A. 91 :22 16- 2220,1994.

Bellahcène A, Kroll M, Liebens F and Castronovo V. Bone sialoprotein expression in primary human breast cancer is associated with bone metastases dcvelopment. J. Bone. Miner. Res. 1 1 :665-670, 1996.

Bellows CG, Heersche MM, and Aubin JE. Determination of the capacity for proliferation and differentiation of osteoprogenitor cells in the presence and absence of dexarnethasone. Dev. Bio. 140: 1 32- 1 38, 1 990.

Bellows CG and Aubin JE. Detemination of numbers of osteoprogeniton present in isolated fetal rat calvaria cells in vitro. Dev. Bio. 133:8-13, 1989.

Bellows CG. Aubin JE, and Heersche JNM. Physiological concentrations of glucocorticoids stimulate formation of bone nodules fiom isolated rat calvaria cells in vitro. Endminology. 121:1985-1992, 1987.

Bellows CG, Aubin JE, Heersche JNM, and Antosz ME. Mineralized bone nodules formed in vitro from enzymatically released rat calvaria ce11 populations. Calcif Tissue. Int. 38: 143- 1 54, 1986.

Berger J, Garattini E, Hua JC and Udenfnend S. Cloning and sequencing of hurnan intestinal aikaline phosphatase cDNA. Proc. Natl. Acad. Sci U.S.A. 84:695-698, 1987.

Bhargava U, Bar-Lev M, Bellows CG and Aubin JE. Ultrastructural analysis of bone nodules formed in vitro by isolated fetal rat calvaria cells. Bone. 9: 155- 163, 1988.

Bianco P. Riminucci M, Bonucci E, Termine JD and Robey PG. Bone sialoprotein (BSP) secretion and osteoblast differentiation: relationship to bromodeoxyuridine incorporation, alkaline phosphatase, and matrix deposition. J. Histochem. Cytochem. 41:183-191, 1993.

Bianco P, Fisher LW, Young MF, Termine JD and Robey PG. Expression of bone sialoprotein (BSP) in developing human tissues. Caicij: Tiss. Int. 49:42 1-6, 199 1 .

Boivin G, More1 G, Lian JB, Anthoine-Therrier C, Dubois PM, and Meunier PJ. Localization of endogenous osteocalcin in neonatal rat bone and its absence in articular cartilage: E ffect of warfarin treatment. Vircho ws. A rch. A. Pathol. Anat. Histophathoi. 4 17:505-5 12, 1990.

Bolander ME, Young MF, Fisher LW, Yamada Y, and Termine JD. Osteonectin cDNA sequence reveals potential binding regions for calcium and hydroxyapatite and shows homologies with both a basement membrane protein (SPARC) and a serine proteinase inhibitor (ovomucoid). Proc. Natf. Acad. Sci LI. S. A. 85(9):29 1 9-23. 1988.

Brady G, Billia F, Knox J, Hoang T, Kirsch IR, Voura EB, Hawley RG, Curnrning R, Buchwald M, Siminovitch K, Miyarnoto N, Boehmelt G, Iscove NN. Analysis of gene-expression in a complex differentiation hierarchy by global amplification of cDNA from single cells. Cur. Bio. 5: 909-922, 1995.

Brady G and Iscove NN. Construction of cDNA libraries from single cells. Meth. Enzymol. 225:611-623, 1993.

Brady G, Barbara M, and lscove NN. Representative in vitro cDNA amplification fiom individual hemopoietic cells and colonies. Merhods. Mol. Cell. Biol. 2: 17-25, 1990.

Bronckers ALJJ, Gay S, Finkelman RD, and Butler WT. Developmental appearance of Gla proteins (osteocalcin) and alkaline phosphatase in tooth germs and bones of the rat. Bone. Miner. 2:36 1-73, 1 987.

Bronckers ALJJ. Gay S, Dimuzio MT, and Butler WT. Immunolocalization of gamma- carboxyglutamic acid containing proteins in developing rat bones. Coll. Relar. Res. 5:273-81, 1985.

Brown LF, Berse B, Van de Water L, Papadopoulos-Sergiou A, Pemizzi CA, Manseau EJ, Dvorak HF, Senger DR. Expression and distribution of osteopontin in hurnan tissues: widespread association with luminal epitheiiai surfaces. Mol. Biul. Crll. 3:1169-1180. 1992.

Chen J, Singh K, Mukhejee BB, Sodek J. Developmental expression of osteopontin (OPN) mRNA in rat tissues: evidence for a role for OPN in bone formation and resorption. Matrix. 13: 1 13-1 23. 1993.

Chen J, Zhang Q, McCulloch CAG, and Sodek J. Immunohistochemical localization of bone sialoprotein in foetal porcine bone tissues: Cornparisons with secreted phosphoprotein 1 (SPP- 1, osteopontin) and SPARC (osteonectin). Hisrochem. J 23:28 1.289, 199 1 a.

Chen JK, Shapiro HS, Wrana IL, Reimers S, Heersche JN and Sodek J. Localization of bone sialoprotein (BSP) expression to sites of minerdized tissue formation in fetal rai tissues by in situ hybridization. Matrix. 1 1 : 133- 143, 1991 b.

Chen Y, Bal BS, Gorski JP. Calcium and collagen binding properties of osteopontin, bone sialoprotein, and bone acidic glycoprotein-75 from bone. J. Biol. Chem. 267:%87 1-24878, 1992.

Cheung DT, Dicesare P, Benya PD, Libaw E and Nimni ME. The presence of intermolecular disulfide cross-links in type III collagen. J. Biol. Chem. 258:7774- 7778,1983.

Craig AM. Nemir M. Mukhejee BB. Chambers AF. Denhardt DT. Identification of the major phosphoprotein secreted by many rodent cell lines as 2adosteopontin: enhanced expression in H-ras-transfonned 3T3 cells. Biochem. Biophys. Res. Commun. 1 57: 166-73, 1988.

Denhardt DT, Guo X. Osteopontin: a protein with diverse hctions. FASEB. J. 7:1475- 1482,1993.

Desbois C, Hagihara K, Pinero G, Boyce B, Bonadio J, Tseng KF, Goulet R, Goldstein S, Loyer E, Bradley A and Karsenty G. Increased bone formation in osteocalcin deficient mice. J Bone. Miner. Res. 10:s 1 39, 1995.

Doi Y, Okuda R, Takezawa Y, Shibata S, Moriwaki Y, Wakamatsu N, Shimizu N. Monyma K, and Shimokawa H. Osteonectin inhibiting de novo formation of apatite in the presence of collagen. Caïcif: Tissue. Int. 44:200-8, 1989.

Doty SB and Schofield BH. Enzyme histochemistry of bone and cartilage cells. Prog. Histochem. Cytochem. 8: 1-38, 1976.

Dziadek M, Paulsson M, Aurnailley M, and Tirnpl R. Purification and tissue distribution of a small protein (BM-40) extracted from a basement membrane tumor. Eirr. 1 Biochem. 16 1 :W-64, 1986.

Ecarot-Charrier B, Bouchard F and Delloye C. Bone sialoprotein II synthesized by cultured osteoblasts contains tyrosine sulfate. J. Biol. Chem. 264:20049-20053, 1989.

Fedarko NS, Bianco P, Vetter U and Robey PG. Human bone ce11 enzyme expression and cellular heterogeneity: correlation of alkaline phosphatase enzyme activity with ce11 cycle. J Cell. Physiol. 144: 1 15- 12 1, 1990.

Fedde KN, Lane CC and Whyte MP. Alkaline phosphatase is an ectoenzyme that acts on micromolar concentrations of natural substrates at physiologic pH in human osteosarcorna (SAOS-2) cells. Arch. Biochem. Biophys. 264:400-409, 1988.

Fisher LW, McBride OW, Termine .ID and Young MF. Human bone sialoprotein. Deduced protein sequence and chromosomal localization. J. Biul. Chem. 265:2347-235 1, 1990.

Fisher LW, Hawkins GR, Tuross N, Termine JD. Purification and partial characterization of small proteoglycans 1 and II, bone sialoproteins I and II, and osteonectin fiom the mineral cornpartment of developing human bone. J. Biol. Chem. 262:9702- 9708, 1987.

Flores ME, Norgarâ M, Heinegard D, Reinholt FP, Andersson G. RGD-directed attachent of isolated rat osteoclasts to osteopontin, bone sialoprotein, and fibronectin. Exp. Cell. Res. 20 1 526-530, 1992.

Franzen A, Heinegard D. Isolation and characterization of two sialoproteins present only in bone calcified matrix. Biochem. A 232:715-724, 1985.

Friedenstein AJ. Determined and inducible osteogenic precursor cells. in: T h . . m ( C 1 B A Foundation Symposium), Blackwell, London, pp 169- 185, 1973.

Fujisawa R, Nodasaka Y, and Kubki Y. Further characterization of interaction between bone sialoprotein (BSP) and collagen. Calcif: Tissue. Int. 56: 140-144, 1995.

Fujisawa R and Kuboki Y. Changes in levels of osteonectin in bovine dentine during tooth development. Arch. OraZ. Biol. 34:89-92, 1989.

Gehron Robey P. The biochemistry of bone. Endocrinol. Metab. Clin. North. Am. 18:858-902, 1989.

Grnovese C, Rowe D, and Kream B. Construction of DNA sequences complementary to rat al and a2 collagen mRNA and their use in studying the regdation or type 1 collagen synthesis by 1,25-dihydroxyvitamin D. Biochemistry. 23 :62 10-62 16. 1984.

Gerstenfeld LC, Chipman SD, Kelly CM, Hodgens KJ, Lee DD and Landis WJ. Collagen expression, ultrastructural assembly, and mineralization in cultures of chicken embryo osteoblasts. J. Cell. Biol. 1 O6:979-989, 1988.

Gerstenfeld LC. Chipman SD, Glowacki J and Lian IB. Expression of differentiated function by mineralizing cultures of chicken osteoblasts. Dev. Biol. 122:49-60. 1987.

Giachelli CM. Pichler R. Lombardi D, Denhardt DT, Aipers CE, Schwartz SM, Johnson RI. Osteopontin expression in angiotensin II-induced tubulointerstitial nephritis. Kidney. Int. 4 5 5 15-524, 1994.

Giachelli CM, Bae N, Almeida M. Denhardt DT, Alpers CE. Schwartz SM. Osteopontin is elevated during neointima formation in rat arteries and is a novel component of human atherosclerotic plaques. J. Clin. Invest. 92: 1686- 1696, 1993.

Giachelli C, Bae N, Lombardi D, Majesky M, and Schwartz S. Molecular cloning and characterization of 2B7, a rat mRNA which distinguishes smooth muscle ce11 phenotypes in vitro and is identical to osteopontin (secreted phosphoprotein 1, 2a.R). Biochem. Biophys. Res. Commun. 1 77:867-73, 199 1

Glimcher MJ. Recent studies of the mineral phase in bone and its possible linkage to the organic matrix by protein-bound phosphate bonds. Phil. Trans. R. Soc Lond. B. 304:479-508, 1984.

Glumoff V, Makela JK, and Vuorio E. Cloning of cDNA for rat proal(lI1) collagen mRNA. Different expression patterns of type 1 and type III collagen and fibronectin genes in experimental granulation tissue. Biochem. Biophys. Acta. 1217: 41-48, 1994.

Groot CG, Danes JK, Blok JT Hoogendijk A and Hauschka PV. Light and electron rnicroscopic demonstration of osteocalcin antigenicity in embryonic and adult rat bone. Bone. 7:379-385, 1986.

Gundberg CM, Hauschka PV, Lian JB, and Gallop PM. Osteocalcin: isolation, characterization, and detection. Meth. Enzymol. 1 075 1 6-44, 1 984.

Hauschka PV, Lian JB and Gallop PM. Direct identification of the calcium-binding amino acid, gamma-carboxyglutamate, in minedized tissue. Proc. N d . Acad. Sci. ü. S. A. 72:3925-3929, 1975.

Heersche JN, Reimers SM, Wrana JL, Waye MM and Gupta AK. 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. 88(S 1 ): 1 73- 182, 1992.

Henthom PS, Raducha M. Edwards YH, Weiss MJ, Siaughter C, Lafferty MA and Harris H. Nucleotide and amino acid sequences of human intestinal alkaline phosphatase: close homology to placental alkaline phosphatase. Proc. Nail. Acud. Sci. U.S.A. 84: 1234- 1238, 1987.

Holland PW, Harper SJ, McVey JH, and Hogan BL. In vivo expression of mRNA for the Ca++-binding protein SPARC (osteonectin) revealed by in situ hybridization. J. CeII. Biof. t 05:473-82, 1987.

Huang YH, Ohsaki Y, and Kurisu K. Distribution of type 1 and type II1 collagen in the developing periodontal ligament of mice. Matrix. 1 1 :25-35, 199 1.

Hunter GK and Goldberg HA. Modulation of crystal formation by bone phosphoproteins: role of glutamic acid-rich sequences in the nucleation of hydroxyapatite by bone sialoprotein. Biochcm. J . 302: 175-1 79, 1994.

Ikeda T, Nomura S, Yamaguchi A, Suda T, and Yoshiki S. In situ hybridization of bone matrix proteins in undecalcified adult rat bone sections. J. Histochem. Cytochem. 40: 1079-88, 1992.

Jundt G, Berghauser KH, Termine Jû, and Schulz A. Osteonectin-a difierentiation marker of bone cells. Cell. Tissue. Res. 248:409-4 15, 1987.

Kam W. Clauser E, Kim YS, Kan YW and RutterWJ. Cloning, seqwncing, and chromosomal localidon of human tenn placental alkaline phosphatase cDNA. Proç. Natl. Acad. Sci, U.S.A. 823871 5-8719,1985.

Keene DR, Sakai LY, and Burgeson RE. Human bone contains type 111 collagen, type V1 collagen, and fibrillin: type II1 collagen is present on specific fibers that may

mediate attachment of tendons, ligaments, and periosteum to calcified bone cortex. d Histochem. Cytochern. 3959.69, 1 99 1 .

Keene DR, Engvall E, Glanville RW. Ultrastructure of type VI collagen in human skin and cartilage suggests an anchoring function for this filamentous network. J. Cell. Biol. 107: 1995-2006, 1988.

Khokha R, Waterhouse P, Lala P, Zimrner Ml Denhardt DT and Khokha R. Increased proteinase expression during tumor progression of ceIl lines down-modulated for TIMP levels: a new transformation paradigm? J. Cancer. Res. Clin. Onc. 1 17:333-8, 199 1.

Lane TF and Sage EH. Functional mapping of SPARC: peptides fiom two distinct Ca+(+)-binding sites modulate ceIl shape. J. Cell. Biol. 1 1 1 :3065-76, 1990.

Liaw L. Skinner MP, Raines EW, Ross R, Cheresh DA, Schwartz SM, Giachelli CM. The adhesive and migratory effects of osteopontin are mediated via distinct cell surface integrins. Role of alpha v beta 3 in smooth muscle cell migration to osteopontin in vitro. J. Clin. Invest. 927 13-724, 1995.

Liu F, Malaval L. Gupta AK, and Aubin JE. Simultaneous detection of multiple bone- related mRNAs and protein expression during osteoblast differentiation: polymerase chah reaction and immunocytochemical studies at the single cell level. Dev. Bio. 166:220-234, 1994.

Liu F and Aubin JE. Identification and molecular characterization of preosteoblasts in the osteoblast di fferentiation sequence. J. Bone. Miner. Res. 9:s 1 25, 1 994.

Long MW, Robinson JA, Ashcraft EA, and Mann KG. Regulation of human bone marrow-derived osteoprogenitor cells by osteogenic growth factors. J. Clin. Invesî. 95881-7, 1995.

Lopez CA, Olson ES, Adams JC, Mou K, Denhardt DT, Davis RL. Osteopontin expression detected in adult cochleae and inner ear fluids. Hearing. Res. 85:2 10- 22, 1995.

Lopez CA, Hoyer JR, Wilson PD, Waterhouse P, Denhardt DT. Heterogeneity of osteopontin expression among nephrons in mouse kidneys and enhanced expression in sclerotic glomenili. La&. Invest. 69:355-363, 1 993.

Lukinmaa PL, Vaahtokari A, Vainio S, Sandberg M, Waltimo J, and Thesleff 1. Transient expression of type III collagen by odontoblasts: developmental changes in the distribution of pro-alpha l(II1) and pro-alpha 1(1) collagen mRNAs in dental tissues. Matrix. 1 3 :5O3- 1 5, 1 993.

Majmudar G, Bole D, Goldstein SA, and Bonadio J. Bone ce11 culture in a three- dimensional polymer bead stabilizes the differentiated phenotype and provides evidence that osteoblastic cells synthesize type III collagen and fibronectin. J. Bone. Miner. Res. 6:869-8 1 , 1991.

Malaval L, Modrowski D, Gupta AK, and Aubin JE. Cellular expression of bone-related proteins during in vitro osteogenesis in rat bone marrow stroma1 ce11 cultures. J. Cell. Physiol. 1 58555-72. 1994.

Mark MP, Butler WT, Prince CW. Firielman RD, and Ruch J-V. Developmental expression of 44-kDa bone phosphoprotein (osteopontin) and bone gamma- carboxyglutamic acid @a)-containing protein (osteocalcin) in calcifying tissues of rat. Differentiation 37: 123-1 36, 1988.

Mark MP, Prince CW, Oosawa T, Gay S, Bronckers AL, and Butler WT. Immunohistochemical demonstration of a 44-KD phosphoprotein in developing rat bones. J. Histochem. Cytochem. 35:707- 1 5 , 1 987a.

Mark MP, Prince CW, Gay S, Austin RL. Bhown M, Finkelrnan RD and Butler WT. A comparative imrnunocytochemical study on the subcellular distributions of 44 kDa bone phosphoprotein and bone gamma-carboxyglutarnic acid (g1a)- containing protein in osteoblasts. J. Bonc. Miner. Res. 2:3 37-346, 1987b.

Mason IJ. Taylor A, Williams JG, Sage H, and Hogan BLM. Evidence from molecular cloning that SPARC, a major product of mouse embryo parietal endoderm. is related to an endothelial ce11 "culture shock" glycoprotein of M, 43 000. EMBO. J. 5:1465-1472, 1986a.

Mason IJ, Murphy D, Munke M. Francke U, Ellion RW, Hogan BL. Developmental and transformation-sensitive expression of the Sparc gene on mouse chromosome 1 1. EMBO. J S(8): 183 1-7, l986b.

McComb RB, Bowers GN Jr., and Solomon P. Al--. Plenum Press, New York, 1979.

McKee MD, Nanci A, Landis WJ, Gotoh Y, Gerstenfeld LC, Glimcher MJ. Developmental appearance and ultrastructural immunolocalization of a major 66 kDa phosphoprotein in embryonic and pst-natal chicken bone. Anat. Rec. 228:77-92, 1990.

Meyuhas O and Perry RP. Construction and identification of cDNA clones for mouse ribosomal proteins: Application for the study of r-protein gene expression. Gene. 10:113-129, 1980.

Moore MA, Gotoh Y, Rafidi K, and Gerstenfeld LC. Characterization of a cDNA for chicken osteopontin: Expression during bone development, osteoblast differentiation, and tissue distribution. Biochemistry. 30:2501-2508, 199 1 .

Mundlos S, Schwahn B. Reichert T, Zabel B. Distribution of osteonectin mRNA and protein during huinan embryonic and fetal development. J. Histochem. Cytochem. 4Q:283-29 1. t 992.

Nemir M. DeVouge MW, Mukhejee BB. Normal rat kidney cells secrete both phosphorylated and nonphosphorylated forms of osteopontin showing different physiological properties. J. Bb i . Chem. 264: 18202- 18208, 1989.

Niyibizi C and Eyre DR. Identification of the cartilage al(X1) chain in type V collagen frorn bovine bone. FEBS. Lett. 242:3 14-3 18, 1989.

Noda M, Yoon K, Thiede M, Buenaga R. Weiss M, Henthorn P. Harris H and Rodan G. cDNA cloning of alkaline phosphatase from rat osteosarcorna (ROS 1 7/28) cells. J. Bone. Miner. Res. 2: 16 1 - 164, 1987.

Noda M. Yoon K. Rodan GA and Koppel DE. High lateral mobility of endogenous and transfected al kaline phosphatase: a phosphatidy linositol-anchored membrane protein. J. Cell. Biol. 105: 167 1 - 1677, 1987.

Nomun S. Wills AJ, Edwards DR. Heath JK, and Hogan BL. Developmental expression of 2ar (osteopontin) and SPARC (osteonectin) RNA as revealed by in situ hybridization. J. Cell. Biol. 1 O6:M 1-50, 1988.

Ohta T, Mori M, Ogawa K, Matsuyama T, and Ishii S. lmmunocytochemical localization of BGP in human bones in various developmentai stages and pathological conditions. Virchows. Arch. A. Pathol. Anat. Histopathol. 41 5459-66, 1989.

Oldberg A, Franzen A and Heinegard D. The primary structure of a cell-binding bone sialoprotein. J. Biol. Chem. 263: 19430- 19432, 1988.

Oldberg A, Franzen A, and Heinegard D. Cloning and sequence analysis of rat bone sialoprotein (osteopontin) cDNA reveals an Arg-Gly-Asp cell-binding sequence. Proc. Natf. Acud. Sei. US. A. 83 :88 19-23, 1986.

Owen TA, Aronow M. Barone LM, Bettencourt B, Stein GS and Lian JB. Pleiotropic effects of vitamin D on osteoblast gene expression are related to the proliferative and differentiated state of the bone ceIl phenotype: Dependency upon basal levels of gene expression, duration of exposure, and bne matrix competency in normal rat osteoblast cultures. Endocrinology. 128: 1496- 1 504, 199 1 .

Owen TA, Aronow M, Shalhoub V, Barone LM, Wilming L, Tassinari MS, Kennedy MB, Pockwinse S. Lian JB, and Stein GS. 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, 1990.

Pockwinse SM, Wilming LG, Conlon DM, Stein GS and Lian JB. Expression of cell growth and bone specific genes at single ce11 resolution during development of bone tissue. J. Cell. Biochem. 49:3 10-323, 1992.

Prince CW, Oosawa T, Butler WT, Tumana M, Bliown AS, Bliown hi, and Schrohenloher RE. Isolation, characterization, and biosynthesis of a phosphorylated glycoprotein from rat bone. J. Biol. Chcrn. 262:2900-7, 1987.

Rarnbhatla L, Patel B, Dhanasekaran N, and Latharn KE. Analysis of G protein a subunit mRNA abundance in preimplantation mouse embryos using a rapid, quantitative RT-PCR approac h. Mol. Reprod. Dev. 4 1 :3 1 4-324, 1 995.

Reichert T, Storkel S, Becker K, and Fisher LW. The role of osteonectin in human tooth development: an imrnunohistological study. Calcif: Tissue. In[. 50:468-72, 1992.

Reinholt FP, Hultenby K, Oldberg A, and Heinegard D. Osteopontin-a possible anchor of osteoclasts to bone. Proc. N d . Acud. Sci. US. A. 87:4473-4475, I 990.

Ritter NM, Farach-Carson MC. Butler WT. Evidence for the formation of a complex between osteopontin and osteocalcin. J. Bone. Miner. Res. 7:877-885, 1997.

Robey PG and Termine JD. Human bone cells in vitro. Calcif Tiss. In!. 37:453-60, 1985.

Rodan GA and Noda M. Gene expression in osteoblastic cells. Crit. Rev. Euk. Gene. Exp. 1 :85-98, 199 1.

Rodan GA and Rodan SB. Expression of the osteoblastic phenotype. in. I M, Annual 2 (WA Peck, Ed.), pp.244-285. Elsevier Science

Publishers, 1 983.

Romberg RW, Werness PG, Riggs BL and Mann KG. Inhibition of hyàroxyapatite crystal growth by bone-specific and other calcium-binding proteins. Biochemistry. X:I 176-1 180, 1986.

Romberg RW, Wemess PG, Lollar P, Riggs BL, and Mann KG. Isolation and characterization of native adult osteonectin. J Biol. Chern. 26032728.36, 1985.

Ross FP, Chappel J, Alvarez JI, Sander D, Butler WT, Farach-Carson MC, Mintz KA, Robey PG, Teitelbaum SL, Cheresh DA. Interactions between the bone matrix proteins osteopontin and bone sialoprotein and the osteoclast integrin alpha v beta 3 potentiate bone resorption. J. Biol. Chem. 268:990 1-7, 1993.

Sage H, Vernon RB, Decker J, Funk S, and Iruela-Arispe ML. Distribution of the calcium-binding protein SPARC in tissues of embryonic and adult mice. J. Histochem. Cytochem. 37:8 19-29, 1989a.

Sage H, Vernon RB, Funk SE, Everin EA, and Ange110 J. SPARC, a secreted protein associated with celiuiar proliferation, inhibits ceil spreading in vitro and exhibits Ca+2-dependent binding to the extracellular matrix. J. Cell. Biol. 1 O9:M 1 -56. l989b.

Sage H, Johnson C, and Bornstein P. Characterization of a novel serum albumin-binding glycoprotein secreted by endothelial cells in culture. J. Biol. Chem. 259:3993- 4007, 1984.

Sambrook J, Frisch EF. and Maniatis T. "Molecular Cloning: A Laboratory Manual." Cold Spnng Harbor Laboratory, Cold Spring Harbor, NY, 1989.

Sandberg M, Autio-Harmainen H, and Vuorio E. Localization of the expression of types 1, III, and IV collagen, TGF-beta 1 and c-fos genes in developing human calvarial bones. Dev. Biol. 1 30:324-34, 1988.

Scott BL. Thymidine-3H electron microscope radioautography of osteogenic cells in the fetal rat. J , Cell. Biol. 35:115-126, 1967,

Scott DM, Kent GN, and Cohn DV. Collagen synthesis in cultured osteoblast-like cells. Arch. Biochem. Biophys. 20 1 :384-39 1 , 1980.

Senger DR, Pemui CA, Papadopoulos A, Tenen DG. Purification of a human milk protein closely similar to tumor-secreted phosphoproteins and osteopontin. Biochim. Biophys. Acta. 996:43-48, 1989.

Shiraga H, Min W, VanDusen WJ, Clayman MD, Miner D, Terrell CH, Sherbotie JR, Foreman JW, Prysiecki C, Neilson EG. and Hoyer JR. Inhibition of calcium oxalate crystal growth in vitro by uropontin: another member of the aspartic acid- rich protein superfamily. Proc. Nat/. Acad Sci. Lr.S.A. 89:426-430, 1992.

Singh K, DeVouge MW, Mukhe jee BB. Physiological properties and differential glycosylation of phosphorylated and nonphosphorylated forms of osteopontin secreted by normal rat kidney cells. J . Biol. Chent. 265: 18696- 1870 1 , 1990.

Smith JH and Denhardt DT. Molecular cloning of a tumor promoter-inducible mRNA found in JB6 mouse epidermal cells: induction is stable at high, but not at low, cell densities. J. Cell. Biochem. 34: 1 3-22, 1987.

Sodek J, Chen J, Kasugai S. Nagata T, Zhang Q, McKee MD, and Nanci A. Elucidating the fùnctions of bone sialoprotein and osteopontin in bone formation. in:

Bi- o f m d T-. (Slavkin H and Price P. eds.) pp 297-306, Elsevier, New York, 1992.

Somerman MJ, Prince CW, Butler WT, Foster RA, Moehring JM, Sauk JJ. Cell attachment activity of the 44 kilodalton bone phosphoprotein is noi resiricted io bone cells. Matrix. 9:49-54, 1989.

Somerman MJ, Fisher LW, Foster RA, Sauk JJ. Human bone sialoprotein 1 and II enhance fibroblast attachment in vitro. Calcif: Tiss. Int. 4350-53. 1988.

Someman MJ, Prince CW, Sauk JJ, Foster RA, and Butler WT. Mechanism of fibroblast attachent to bone extracellular matrix: role of a 44 kilodalton bone phosphoprotein. J. Bone. Miner. Res. 2:259-65, 1987.

Stein GS, Lian JB, Gerstenfeld C, Shalhoub V, Aronow M, Owen T, and Markose E. The onset and progression of osteoblast differentiation is functionally related to cellular proli feration. Connect. Tissue. Res. 2O:3 - 1 3, 1989.

Stringa E, Filanti C. Giunciuglio D, Albini A, and Manduca P. Osteoblastic cells from rat long bone. 1. Characterization of their differentiation in culture. Bone. 16:663-670, 1995.

Swanson GJ, Nomura S, Hogan BL. Distribution of expression of 2AR (osteopontin) in the embryonic mouse inner ear revealed by in situ hybridisation. Hearing. Res. 41:169-177. 1989.

Termine JD, Kleinrnan HK, Whitson SW, COM KM, McGarvey ML, and Martin GR. Osteonectin, a bone-specific protein linking mineral to collagen. Cell. 26:99- 105, 1981.

Tezuka K, Sato T, Karnioka H, Nijweide PJ, Tanaka K, Matsuo T, Ohta M, Kuihara NT Hakeda Y, Kumegawa M. Identification of osteopontin in isolated rabbit osteoclasts. Biochem. Biophys. Res. Commun. 1 86:9 1 1 -9 1 7, 1 992.

Thiede MA, Smock SL, Petenen DN, Grasser WA, Thompson DD and Nishimoto SK. Presence of messenger ribonucleic acid encoding osteocalcin, a marker of bone turnover, in bone marrow megakaryocytes and peripherai blood platelets. Endocrinology. 1 3 5 :929-3 7, 1994.

Tung PS, Domenicucci C, Wasi S, Sodek I. Speci fic immunohistochemical localization of osteonectin and collagen types 1 and II1 in fetal and adult porcine dental tissues. J. Histochem. Cytochem. 33 5 3 1-40, 1985.

Turksen K and Aubin JE. Positive and negative imrnunoselection for enrichment of two classes of osteoprogenitor cells. J. Cell. Bol. 1 14:373-384, 19%.

Varmuza S and Tate P. Isolation of epiblast specific cDNA clones by differential hybridization with PCR arnplified probes derived from single embryos. Mol. Reprod. Dev. 32:339-348, 1992.

Wasi S, Otsuka K, Yao KL, Tung PS, Aubin JE, Sodek J, and Termine ID. An osteonectinlike protein in porcine periodontal ligament and its synthesis by periodontal ligament fibroblasts. Can. J. Biochem. Cell. Biol. 62:470-8, 1984.

Waterhouse P. Parhar RS, Guo X. Lala PK, Denhardt DT. Regulated temporal and spatial expression of the calcium-binding proteins calcyclin and OPN (osteopontin) in mouse tissues during pregnancy. Mol. Reprod. Dev. 32:3 1 5-323, 1992.

Weber GF, Ashkar S, Glimcher MJ and Cantor H. Receptor-ligand interaction between CD44 and osteopontin. Science. 27 1 509-5 12. 1996.

Weinreb M, Shinar D, Rodan GA. Different pattern of alkaline phosphatase, osteopontin, and osteocalcin expression in developing rat bone visualized by in situ hybridization. J . Bone. Miner. Res. 5 :83 1-42, 1990.

Weiss MJ, Henthom PS, Lafferty MA, Slaughter C, Raducha M, and Harris H. Isolation and characterization of a cDNA encoding a human liver/bone/kidney-type alkaline phosphatase. Pruc. N d . Acud. Sci U.S.A. 83:7 182-7 186, 1986.

Wiestner M, Fischer S, Dessau W, and Muller PK. Collagen types synthesized by isolated calvariurn cells. Exp. Cell. Res. 133: 1 15- 125, 198 1.

Worcester EM, Blumenthal SS, Beshensky AM, Lewand DL. The calcium oxalate crystal growth inhibitor protein produced by mouse kidney cortical cells in culture is osteopontin. J. Bone. Miner. Res. 7: 1029-1 036, 1992.

Yoon K, Buenaga R and Rodan GA. Tissue specificity and developmental expression of rat osteopontin. Biochem. Biophys. Res. Commun. 148: 1 129- 1 1 36, 1 987.

Young MF, Ken JM, Termine .ID, Wewer UM, Wang MG, McBride OW, and Fisher LW. cDNA cloning, mRNA distribution and heterogeneity, chromosomal location, and RFLP analysis of human osteopontin (OPN). Genomics. 7:491-502, 1990.

Zemik J, Trawog K and Upholt WB. Regulation of alkaline phosphatase and alpha 2 (1) procollagen synthesis during early intramembranous bone formation in the rat rnandible. Drfferentiation. 44:207-2 1 5 , 1990.

Zhang Q, Domenicucci C, Goldberg HA, Wrana JL and Sodek J. Characterization of fetal porcine bone sialoproteins, secreted phosphoprotein 1 (SPPI, osteopontin). bone sialoprotein, and a 23-kDa glycoprotein. Demonstration that the 23-kDa glycoprotein is derived from the carboxyl terminus of SPPI. J. Biol. Chem. 265: 7583-7589, 1990.