THE RAT BONE MARROW STROMAL CELL OSTEOGENIC SYSTEM: CHARACTERIZATION OF SUBPOPULATIONS, FRACTIONATION,

AND EFFECTS OF PDGF

Alexandra knore Herbertson

A thesis submitted in conformity with the requiremen ts for the degree of Doctor of Philosophy

Graduate Department of Dentistry

University of Toronto

Q Copyright by Alexandra Herbertson 1998

National Library Bibliothéque nationale du Canada

Acquisitions and Acquisitions et Bibliograp hic Services services bibliographiques

395 Wellington Street 395, rue Wellington OttawaON K I A W Oitawa ON KiA ON4 canada Canada

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

reproduction sur papier ou sur format électronique.

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

THE RAT BONE MARROW STROMAL CELL OSTEOGENIC SYSTEM: CHARACTERIZATION OF SUBPOPULATIONS, FRACTIONATION,

AND EFFECTS OF PDGF

by Alexandra Lenore Herbertson

Doctor of Philosophy, 1998

Graduate Department of Dentistry, University of Toronto

ABSTRACT

The objective of this thesis was to better define and characterize the heterogeneous cellular

environment of the rat bone marrow (RBM) stromal ce11 osteogenic system. 1 began by

identifying and quantifying the major subpopulations present when RBM stromal cells are

grown under conditions stimulating bone formation. ie., supplementation with

dexamethasone (dex). As expected, dex significantly increased the number of alkaline

phosphatase (AP) positive colonies and von Kossa positive bone nodules, but concomitantly

stimulated the number of macrophage colonies, while reducing the number of hemopoietic

cells expressing leukocyte common antigen. Since the presence of these multiple ce11 types

rnay cornplicate determination of mechanisms of regulation of the osteoprogenitors, we next

investigated methods for purification of the osteoblast lineage cells. Flow cytometric

sorting perfotmed on the basis of AP expression resulted in A P ~ B ~ and A P ~ O W fractions of

cells: within the ~ 9 h i g h fraction. the number of AP-positive colonies and bone nodules were

signi ficantly enriched. Adipocyte and macrophage colonies were signi ficantly depieted in

the M h i g h fraction and consistently enriched in the A P ~ W fraction of cells.

Finally, 1 investigated the effect of PDGF. a Factor produced in abundance by macrophages.

i n fractionated and unfractionated RBM stromal populations. While stimulating

proliferation and differentiation of the macrophage and adipocytic lineages, PDGF had

..- 111

dose- and time-dependent biphasic effects on bone nodule formation. stimulating at low

doses and at early stages of osteoblast differentiation but inhibiting at high doses and later;

suggesting that PDGF has pleiotropic effects on bone metabolism, and that its effects on

bone might be direct on osteoblast lineage cells.

Together, my data validate methods for enrichment of osteoprogenitors in RBM stroma1

cultures and provide a stronger b a i s for investigating the nature of their regulaiion by

hormones, cytokines. and growth factors.

ACKNOWLEDGEMENTS

First and foremost, 1 wish to thank my supervisor. Dr. Jane E. Aubin, for her

unending support, guidance, patience, and understanding during my time in her laboratory.

1 consider myself privileged to have been able to work under her.

1 also wish to thank fellow laboratory mernbers Usha Bhargava, Harry Moe, and Sui-

Wah Fung for their patience in teaching and their excellent technical assistance. As well, 1

wou1d like to thank fellow students Fina Liu and Shiva Shadmand for their friendship and

support during this time.

1 also wish to acknowledge the financial support of the Medicai Research Council of

Canada through their Dental Fellowship program, and the suppon of the Harron Trust Fund,

without which this work would not have been possible.

Finally. 1 would like to extend my love and thanks to my husband, Dale Herbertson.

for al1 the years of support and understanding.

TABLE OF CONTENTS

Page

ABSTRACT

ACKNOWLEDGEMENTS

TABLE OF CONTENTS

LIST OF FIGURES

LIST OF TABLES

ABBREVIATIONS

C)IAPTER 1 Introduction

1.1 General Introduction

1.2 Non-Osteogenic Stromal Ce11 Types

1.3 Osteogenic Stromd Cells .

1 -4 S tromal CeIl Interactions

1.5 Purification of osteoblasts from stroma

1.6 Cytokines and Growth Factors in the Stromal Environment

1 -7 Platelet-Derived Growth Factor (PDGF)

1.8 Thesis Objectives

C HA PTER 2 Dexarnetliasone Alters the Subpopulution Make- Up of

Rat Bone Marro w Stromal Cell Cultures

2.1 Introduction

2.2 MateriaIs and Methods

2.3 Results

2.4 Discussion

V

vii

viii

CHAPTER 3 CeU Sorting Enriches Osteogenic Popularions in Rat Bone

Marrow Struma1 Cell Cultures 49

3.1 Introduction 50

3.2 Materiais and Methods 53

3.3 Results 55

3 -4 Discussion 69

CHAPTER 4 PDGF has Biphosic Effecfs on the Differenfiation of

Osteoprogenilon in RBM Sbomal Ce11 Populations

4.1 Introduction

4.2 Materiais and Methods

4.3 ResuIts

4.4 Discussion

CHAPTER 5 Geneml Surnrnmy and Conclusions

5.1 Conclusions and Future Experiments

vii

LIST OF FZGURES

Figure 2.1

Figure 2.2

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.7

Figure 3.1

Figure 3.2

Figure 3.3

Figure 3.4

Figure 4.1

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Overviews of 35 mm RB M strornai ce11 culture dishes

Time course of bone noduie formation in RBM stromal ce11 cultures

AP positive colonies in RBM stroma1 ce11 cultures

aNE3E colonies in RBM strornai ce11 cultures

Micrographs of the different cell types present in RBM stroma

Micrographs of immunohistochemical staining of RBM stromal cells

Flow cytometry frequency vs. fluorescence intensity histograms

Flow cytometric set up for sorting

Flow cytometry analysis

Adipocyte colony numbers for sorted RBM stromal populations

Bone noduleslaNBE colonies vs. plating density (soned populations)

Ce11 number vs. time cultured, with and without PDGF

Bone nodules vs. PDGF concentration

Number and size of aNBE colonies vs. PDGF concentration

Adipocyte colonies vs. PDGF concentration

Bone nodules vs. PDGF pulses

PDGF's effects on sorted RBM stromal populations

Page

37

38

39

40

41

42

43

63

64

66

67

84

85

86

87

88

89

LIST OF TABLES

Page

Table 1.1 PDGFs effects on bone formation in osteoblastic ce11 cultures 20

Table 2.1 Quantitation of colony types in RBM stroma1 ce11 cultures 36

Table 3.1 Percentage of AP-positive colonies in sorted RBM stromai populations 60

Table 3.2 Net changes in various colony types in ~ ~ h i g h and ~ ~ 1 0 ~ sorted cells 61

Table 3.3 Net changes in bone nodules/aNBE colonies for sorted cells 62

Table 4.1 PDGFs ffects on bone nodule, adipocyte and aNBE colonies 82

Table 4.2 Effect of continuous exposure to PDGF on RBM stroma1 populations 83

aNBE AP BM CDR CFU-F

CFU-O

CFU-ST

d,

dex

ED2

EGF GAM GM-CSF

HPR hr.

IL- 1

IGFBP

LCA

LNGFR

M

mAB

M-CSF

MRC OX- 1

MRC OX-22

PDGF

PGE2

m RBM 21 1.13

RBM

RECA- 1

TNF

a-naphthyl butynte esterase

aikaline phosphatase

bone marrow

colour development reagent

fibroblastic colony-forming unit

osteoprogenitor colony-forming unit

stroma1 colony-forming unit

day(s) dexamethasone

monoclonal antibody specific against rat macrophage cells

epidermal growth factor

goat anti-mouse

granuIocyte/rnacrophage colony-stimulating factor

horseradish peroxidase hour(s ) interleukin 1

insulin-like growth factor binding protein

leukocyte comrnon antigen

low-affinity nerve growth factor receptor

molar

monoclonal an tibody

macrophage colony stirnulating factor

monoclonal an tibody against leukocyte cornmon ant igen

monoclonal antibody against nt lymphocytes

platelet-det-ived growth factor

prostaglandin E2

parathyroid hormone

monoclonal antibody specific against rat alkaline phosphatase

rat bone marrow

monoclonal antibody specific against rat endothelial cells

tumor necrosis factor

CHAPTER 1

Introduction

1.1 Generd Introduction

Bone manow (BM) is a complex tissue consisting of many different ce11 types with distinct

functions existing together in a coordinated, regulated and inter-dependent fashion. BM can

be divided into two broad cornpartments: the hemopoietic cells. a primarily nonadherent

population in ce11 cultures, and the strornal cells, primarily adherent in ce11 cultures.

Regulation of mammalian hemopoiesis is dependent on a microenvironment composed of a

variety of hemopoietic cells and strornal cells and their extracellular rnatrix (Dexter, 1982;

Ross et al., 199 1 ; Long, 1992). it also appears that stromal ce11 growth and development.

including those of osteogenic cells. is regulated by the local microenvironment. as

hemopoietic cells c m produce cytokines known to affect the growth and differentiation of

stromal cells, and BM stromal cells can release factors affecting other strornal cells (Ogiso et

al., 1991: Yanai et al.. 1991; Zhang et al., 1991; Sasaki and Hong, 1993).

1.2 Non-Osteogenic Stroma1 Ce11 Types

It has been recognized that BM stroma contains a mixed population of ceIl types (Dexter,

1982). but many published reports remain focused on only one or a few of the ce11 types in

the culture. In mouse hemopoieticaily active long-term BM stromal cultures. the cell

populations have been identified as: 2 1 + 2% fïbroblastoid. 3 I 0.3% endothelial, 26 + 3%

adipocytic. and 49 + 1% macrophage (Hauser et al., 1993, accounting for essentially al1

(99%) of the stroma. In rat bone marrow (RBM) stromal ce11 cultures, Iittle is known about

the subpopulation make up. except that in 10 - 12 d. primary cultures the colony distribution

was found to be 85% fibroblast-like, 10.3% rnacrophage/dendritic/granulocyte, and 2.5%

endothelial (Sirnmons et al., 1991). Maniatopoulos et al. (1988) also noted in primary RBM

stromal cultures the presence of small, mononuclear (hemopoietic) cells (Maniatopoulos et

3 al., 1988); the repeated medium changes in the 7 d. pnmary culture period in Maniatopoulos'

method is used to reduce hemopoietic contamination and cellular debris while expanding the

numbers of CFU-F. Even in the culture conditions selected to maxirnize osteogenesis, only

a proportion of the cells in RBM stromal ce11 cultures are osteogenic, and the presence of

other morphologically and histochemically/irnmunohistochemically distinguishable ceIl

types has been reported (Malaval et al., 1994). The multiple ceIl types present reflect, at

least in part, the multiple l ineages in marrow stroma, including also muitipotential

mesenchymal stem cells with the ability to differentiate dong various pathways. As the

presence of multiple subpopulations complicates the elucidation of cell-ce11 interactions

within osteogenic stroma, documentation of which cells are present and in what quantities is

needed to unravel the interactions. A brief summary follows of the ce11 types that have been

reported in BM stromal ce11 cultures (of a variety of species).

This group includes definitive fibroblasts involved in forming the supporting connective

tissue of marrow. However, the designation of "fibroblast" has also been used as a general

category of BM stromal cells which are undefined or undifferentiated. The term colony

forming unit - fibrobIast(oid) or CFU-F has been used in reference to colonies which c m

include fibroblasts, adipocytes, endothelial cells, chondro- and osteoprogenitor cells, smooth

muscle cells, reticular and reticular-fibroblastoid cells, and mesenchymal stem cells often in

combination wi th macrophages. mast cells, myeloid cells, and hemopoietic stem cells

(Wang et al., 1990; Benayahu et al., 1991; Simmons et al., 1991). Another term used is

colony forming unit strornal or CFU-ST for strornal or adherent cells (McIntyre and

Bjornson, 1986). Assessrnents of results from experiments designating colonies as CN-F

can be difficult to interpret i n light of this potential variety in ce11 composition. For

example, PDGF is known to increase growth of CFU-ST (McIntyre and Bjornson, 1986),

but whether this is a response from multiple populations of cells or just one ce11 type is not

4 known. A monoclonal antibody (mAb), STRO-1, was developed which (in combination

with a glycophorin A negative screen) sorts CFü-Fs from other human BM cells (Simmons

and Torok-Storb, 199 1 ). Cells sorted on this bais (glycophorin-negative, STRO- 1 -positive)

give rise to fibroblasts, smooth muscle cells, and adipocytes which results in an adherent

layer able to support hemopoiesis (Simmons and Torok-Storb, 199 1 ). Osteoprogenitors

capable of differentiating into functional osteoblasts are also present in adult human BM-

derived STRO- 1 -positive C N - F (Gronthos et al., 1994). Taken together, the data suggest

that STRO- 1 may be associated - although not exclusively - with stroma1 mesenchymal stem

cells. More recently, Joyner et al. (1997) have also isolated a mAB (HOP-26) reactive

against human CFU-F, including osteoprogeniton (Joyner et al., 1997). Amongst a series of

mABs developed against RBM stroma, ST3 and ST4 are reactive against fibroblasts

(Sullivan et al., 1989) but ST3 labels primarily marrow fibroblasts and ST4 labels pnmarily

fibroblasts in other tissues. suggesting that the fibroblastic cells of the marrow are different

from those of non-hemopoietic tissues (Sullivan et al., 1989). Treatment with ST3 and

complement pnor to culture decreases CFU-F (and hemopoietic colony formation) (Sullivan

et al.. 1989). While the development of mAbs has aided in identification of fibroblastic cells

in BM, many cells are labelled fibroblastic for want of other identifying characteristics, and

the term undoubtedly encompasses a variety of ce11 types.

Endothelial cells

Endothelial cells have been frequently. although not consistently, identified in BM ce11

cultures. lsolated endathelial ce11 colonies can be obtained from mouse BM with low

plating numbers of less than 20 BM stroma1 cells/35mrn dish in medium supplemented with

GM-CSF (10 UII) (Wang and Wolf, 1990). Mouse endothelial cells have dendritic

cytoplasmic processes, round nuclei, and overlapping cytoplasms in confluent cultures

(Yanai et al., 199 1). Histo- and irnmunohistochemical identification of endothelial cells has

historically posed a problem. Endothelial cells, like macrophages, will take up low density

5 lipoprotein. In mice, endothelial cells bind Ulex europaeus 1 lectin (Yanai et al.. 1991).

Factor VI11 antigen may or may not be detected in mouse endotheliai cells (Wang et al.,

1990; Yanai et ai., 199 1). Polyclonal antibodies against von Willebrand factor are not

endothelial cell-specific and do not identify al1 endothelial cells (Duijvestijn et al.. 1992).

While alkaline phosphatase (AP) activity has been reported in endothelial cells, high activity

appears to be associated with isolated brain capillary endothelial cells in vivo. while

endothelial cells lining blood vessels elsewhere do not contain significant levels (Meyer et

al., 1990). The recent development of mAbs specifically recognizing endothelial cells in

mouse marrow (Hasthorpe et al., 1990) and in rats (RECA- 1 = rat endothelial ce11 antigen)

(Duijvestijn et al., 1992) has made definitive identification of these cells easier.

Reticular ceils

Reticular cell is a histomorphornetric term for the adventitial ce11 which forms the outer

layer surrounding the sinuses in the BM; the reticular ce11 is thought to send out sheet-like

extensions containing reticular fibres to fom a three dimensional network (reticulum) of

cytoplasmic extensions throughout the marrow (Westen and Bainton, 1979). The reticular

ce11 is fibroblastic in appeannce, is characterized by its membrane-bound AP. and associates

mainly with granulocytic precursors (Westen and Bainton. 1979). Antibodies against low-

affinity nerve growth factor receptor (LNGFR) label stroma1 reticular cells with dendritic

features in fresh smears and in formalin-fixed. paraffin embedded hurnan BM (Cattoretti et

al., 1991). These LNGFR positive cells are also positive for AP, reticulin. collagen type iII,

and are negative for neural, endothelial, and leukocyte markers (Cattoretti et al.. 1991).

LNGFR positive cells have an oval nucleus, a scanty cytoplasm with long dendrites that

interrningle with hemopoietic ceils, line the abluminal side of sinus endothelial cells. and

provide the scaffold for the hemopoietic marrow (Cattoretti et al.. 1991). The reticular

stroma1 cells appear in the fetal BM before the hemopoietic activity begins, originate from

the vesse1 adventitia, and radiate in the BM cavity (Cattoretti et al., 1991). Bamer ceIl is

another term for a reticular cell described by Weiss and Geduldig (199 1 ) in a microscopie

histological study of mouse BM. Bamer cells run from the bone-lining layer to subosteal

blood vessels and appear associated with al1 stages of hemopoiesis (Weiss and Geduldig,

1991). Recent works have described ce11 correlates in vitro to the reticular ce11 in vivo.

Anti-LNGFR antibodies label 7 - 11% of the fibroblastoid population in 4 week-old long-

term human BM cultures, suggesting that these cells are reticular cells (Cattoretti et al.,

199 1). In another study. LNGFR positive cells with multiple, long dendritic processes were

similarly found in long-tenn human BM cultures; while few were seen in 2 week old

cultures, their numbers increased up to 6 weeks (Wilkins and Jones, 1995). Since AP is a

comrnon cytochemical marker of both 'reticular' cells and preadipocytes, it has been

suggested that reticular cells may convert to adi pocytes when marrow cellularity abruptly

decreases (Bianco et al., 1988). In murine BM cultured in methylcelluIose/plasma dot,

white blood ceIl conditioned medium, dexamethasone (dex) and serum, extensive branching

colonies developed which were called reticulofibroblastoid and were thought to become

adipocytic by d. 14 of culture (Ross et al., 1991). It is unclear whether these cells represent

a differentiation stage in a preadipocytic lineage, or a multipotential stem ce11 lineage, or

mixed colonies derived from multiple lineages as whole marrow ceIl cultures were used.

However, the authors felt that reticulofibroblastoid ceils are a distinct lineage on the basis of

morphology, adipocyte formation, and production of collagen type IV, and with their long

branching arms rnay be important in setting up compartments for hemopoiesis in mouse BM

(Ross et al., 1991).

Adipocytes

Adipocytes are a well-known component of BM. In long-term BM cultures, the presence of

fat ceIIs is taken as a marker of a hemopoietically productive culture (Ross et al., 1991).

Adipocytes have been descnbed in combinations with many other populations of marrow

cells, e-g., reticulofibroblastoid cells as previously mentioned (Ross et al., 199 1). Bennett

7 and others ( 199 1) described stroma1 colonies from rabbit BM with fibroblastic morphology

that had differentiated in an adipocytic direction, but that subsequently reverted to a more

proliferative stage and differentiated dong the osteogenic pathway (Bennett et al., 1991).

These, and other data, have given rise to the notion of a bipotential adipocyte-osteoblast

progenitor or adipocyte-osteoblast switch. For example, long-term munne marrow culture

stromal clone (BMS2) has been produced which can differentiate into adipocytes; while

BMS2 has been used to examine the regulation of preadipocytic cells (Gimble et al., 1991),

it also demonstrates osteogenic characterstics (Dorheirn et al., 1993). The data are

interesting, but further work will be needed to fully characterize the nature of the adipocyte-

osteoblast phenotypic transformation (see, for e-g.. (Gimble et al., 1996; Aubin and

Heersche, 1997), for reviews).

Smooth muscle cells

In long-term human BM cultures. fibroblastoid cells expressing a-smooth muscle actin have

also been described (Wilkins and Jones, 1995). These cells appear at approximately 2

weeks, and form a majority of the stromal population within 4 - 6 weeks, under at least some

culture conditions (Wilkins and Jones, 1995). In long-term adult human BM cultures,

stromal myoid componenü were also present when the stromal cells were cultured withour

hydrocortisone and horse serurn: the presence of the latter two additives actually inhibited

myoid ce11 growth and development (Bonanno et al., 1994).

Hemopoietic Ceils

Hemopoietic cells exist in intimate association with stroma1 cells in BM in vivo and in vitro.

To remove the majority of well-differentiated hemopoietic cells when culturing stromal

cells, centrifugation or adherence techniques are commonly used. Hemopoietic stem cells in

rnouse BM cm be enriched by adherence to tissue culture plastic, but it is not known if the

cells are adhering to the plastic or to a stromal ce11 (Kiefer et al., 199 1). In mouse BM,

hemopoietic stem cells are present at a frequency of 1 in 25,000 in the adherent ce11 layer

(Kiefer et al., 199 1). and their subsequent differentiation is influenced by the existing culture

conditions. Hernopoietic cell types described in BM stromal cultures include macrophages

(Dexter, 1982: McIntyre and Bjomson, 1986: Wang and Wolf. 1990; Keller et al., 1993;

Wilkins and Jones, 1995)- mast cells (Mclntyre and Bjornson, 1986), dendritic cells (Giesler

et al., 199 1 ; Miyagi and Kimun. 1993, osteoclasts (Udagawa et al., 1990), granulocytes

(Simmons et al., L99 1), and megakaryocytes (Briddell et al., 1992).

Macrophages have long been known as a component of long-term BM stroma1 cultures

(Dexter, 1982). In rnice, macrophages are frequently seen in BM stromal cell cultures and

isolated macrophage colonies can be obtained at the low plating density of 20 or less BM

stroma1 cells/35mrn dish using media supplernented with GM-CSF at 10 UA (Wang and

Wolf, 1990). Mouse macrophages cm be identified with mAbs or histochemical stains such

as berberine sulphate or a-naphthyl butyrate esterase ( aNBE) (Kanemoto et al., 199 1 ). In

human BM cultures. macrophages have been identified in the adherent layer (Keller et al.,

1993; Wilkins and Jones. 1995). While one study found a rather homogeneous macrophage

population (Keller et al., 1993)- another identified two types of macrophages (Wilkins and

Jones, 1995). One type had a rounded morphology and abundant phagocytosed debris,

while the other type had cells that were elongated and flattened, dispalyed several short

cytoplasmic processes, and lacked phagocytosed debris (Wilkins and Jones, 1995). The

second type of macrophage was seen only in more mature cultures (Wilkins and Jones.

1995). The number of macrophages present varied significantly between long-term BM

culture isolates. and did not correlate with the number of hemopoietic foci present (Wilkins

and Jones, 1995).

1.3 Osteogenic Stromal Cells

Friedenstein and colleagues first demonstrated that BM stroma contains ceils with the

capacity to f o m bone when they are transplanted in vivo in diffusion chambers

(Friedenstein et al., 1968). This has been confirmed repeatedly in vivo with BM cells in

diffusion chambers (Ashton et al., 1984; Owen, 1988) and, more recently, conditions have

been established in which bone formation is observed in vitro in cultures of BM stromal

cells isolated from several species. Tibone first reported bone production in vitro by BM

strornal cells in aspirated femoral marrow from adult rabbits (Tibone and Bernard, 1982).

Since then, other nonclonal animal models of BM osteogenesis, with bone nodules as the

terminal product, have also been characterized and optimized for more reproducible and

more copious osteoblastic differentiation. Under culture conditions previously established

to support and rnaximize osteogenesis in cultures of folded chick periostea (Tenenbaum and

Heersche, 1982) and osteobtastic cells derived from calvaria (Bellows et al., 1986), bone

nodule formation in cultures of stromal cells has been seen in young-adult rat femoral BM

(Maniatopoulos et al.. 1988), chick embryo tibial and femoral BM stroma (Kamalia et aI.,

1992), neonatal pig tibial and femoral BM stromal cells (Thomson et al.. 1993). young adult

mouse femoral BM stroma (Van Vlasselaer et al.. 1993), and adult human BM (Vilamitjana-

Amedee et al., 1993). However, optimal conditions for the formation of discrete bone

nodules in these different BM strornal cultures differ, likely refiective of species or donor

age differences. For example, culture conditions similar to those used in rat stroma

(Maniatopoulos et al., 1988) result in rnineralization without nodule formation in adult

human stroma (Cheng et al., 1994). This may in part be a reflection of the age of the BM

donor. as age has been reported to affect the yield of nodules and/or minenlized matrix (see

discussion in (Vilamitjana-Amedee et al.. 1993)), and discrete bone nodules have been

reported in young adult human BM stroma (Vilamitjana-Amedee et al., 1993). The

importance of species differences is obvious when looking at dex as an additive: 10-8 M dex

10 increases the apparent number of osteoprogenitor cells which give rise to bone nodules by 5 -

50-fold in RBM stromd cells (Aubin et ai., 1990) while 10-7 M dex has been detennined to

be optimal for minerdization in a human system (Cheng et ai.. 1994), and concentrations up

to 10-10 M significantiy reduce bone nodule formation in mouse BM stroma (Falla et al..

1993). As well, mouse BM stroma needs to be plated at higher densities to obtain CFU-F

numbers comparable to those in the rat system (unpublished observations), resulting in at

least some attempts to develop strategies for enrichment of osteoprogenitors after general

isolation protocols (Falla et al., 1993).

The mineralized bone nodules formed in these mixed stromai population systems have

morphological characteristics similar to authentic woven bone with its associated cells.

Specifically. in M M stromal cultures, a bone nodule has rows of cuboidal cells with oval

nuclei and prominent nucleoli, well-developed rough endoplasmic reticulum, and prominent

Golgi (Satomura and Nagayarna, 199 1 ) on the surface, separated from the mineralized

material by an osteoid-like matrix; cells resernbling osteocytes are embedded in the

rnineralized matrix (Maniatopoulos et al.. 1988) and sorne cells protmde ce11 processes

toward neighbouring cells through the extracellular matrix (Satomura and Nagayama, 199 1 ).

X-ray microanalysis and X-ray diffraction respectively have been used to show the presence

of Ca and P in the matrix. in the fom of hydroxyapatite (Maniatopoulos et al.. 1988). RBM

stromal cells express PTH receptors (Zhang et al.. 1995), and the cells associated with

nodules exhibit high AP activity (Maniatopoulos et al.. 1988) and produce type 1 collagen

and SPARC/osteonectin, dong with the non-collagenous bone proteins osteocalcin and bone

sialoprotein (Maniatopoulos et al., 1988; Kasugai et al.. 199 1 ; Malaval et al.. 1994).

1.4 Stroma1 Ce11 Interactions

General cell-cell interactions

While different BM stromal cells are known to interact, in most cases the mechanism of the

interaction is not clear, although multiple cytokines and growth factors produced by

different ce11 types present are likely to play roles (see Section 1.6 below). For example,

adipocytes can stimulate the growth of endothelial cells. while endothelial cells can inhibit

the growth and developrnent of preadipocytes (Yanai et al.. 1 99 1 ). Rat marrow stromal

fibroblast-like cells produce a number of osteoiropic factors that at low concentrations

stimulate, or at high concentrations inhibit, calvarial osteoblast growth and there is evidence

for the involvement of insulin-like growth factors (Zhang et al., 199 1). Rat skin fibroblasts

and periodontal ligament fibroblasts release soluble factors into their conditioned medium,

including prostaglandins, that inhibit osteoblast differentiation in BM stroma (Ogiso et al..

1991). On the other hand, bone formation is increased in stroma1 cell cultures when

conditioned medium from rat calvaria cells is added, suggesting that mature osteoblasts

produce a paracrine growth factor that can stimulate the differentiation of osteoblasts from

precursor cells (Hughes and McCulloch, 1991). It has also been suggested that factors

produced by BM cells may affect other BM stromal cells that are distant from the factor

source; healing RBM produces a growth factor activity with a preferential effect on

osteogenic cells (Bab et al., 1988). As well. marrow ablation in rat long bones induces an

increase in osteogenesis in distant skeletal sites and the systemic osteogenic response is

thought to be rnediated by circulating factors produced by the healing marrow (Bab et al.,

1988; Gazit et al., 1990; Bab et al., 1992).

Limiting dilution studies

In rat calvaria and RBM stromal bone mode1 systems, the nurnber of nodules or colonies

capable of forming bone can be counted for an assessment of osteoprogenitor numbers

12 recoverable from the tissue and able to undergo differentiation in vitro under certain test

conditions (e-g.. rat calvaria (Bellows and Aubin, 1989); RBM stroma (McCulloch et al.,

1991)). Limiting dilution analysis was used to investigate the differentiation process of

these restricted monopotential osteoprogenitor cells in these primary cultures. In rat calvaria

ce11 populations, limiting dilution indicated a "single-hit" phenornenon, i.e. that only one

ce11 type was limiting for nodule formation; the data are consistent with the lirniting cell type

being the osteoprogenitor itself, and with such cells being present at a measurable but low

frequency ( 4 % ) of the total population under standard isolation and culture conditions

(Bellows and Aubin, 1989). In contrast, in RBM stromai secondary populations expression

of neither total CFU-F nor the osteoprogenitor subfraction (Cm-O) follows a linear

relationship in limiting dilution analysis until very high ce11 densities are reached, suggesting

cooperativity of ce11 types within the population and a multi-iarget phenornenon (Aubin et

al., 1990). In multi-target events. the response (production of a bone nodule) is observed

only when at least one ce11 of each of rn categories is present in the sample (Lefkovits and

Waldmann. 1979). i.e., that at least two ce11 types are limiting for bone nodule formation in

RBM stroma. When a constant number of non-adherent (hemopoietic) BM cells was added

back to the limiting dilution analyses of the stroma1 cells, about 11100 CFU-O were

measured and their expression now followed a Iinear, single-hit relationship (Aubin et al.,

1990). Endothelial cells or endothelial ce11 conditioned medium had similar effects on CFU-

O expression (Aubin et al.. 1990). Similarly. formation of bone nodules in vitro from

murine marrow CN-F plated nt low density requires the presence of irradiated feeder celis,

and PDGF. L-3. and EGF appeared not to be able to substitute for feeder cells. leading to

the suggestion that another as yet unidentified factor produced by the imdiated cells was

required (Friedensrein et al.. 1992). These data suggest that stroma1 osteogenesis in RBM

may be under the regulation of other cells in the BM which can exert either stimulatory or

inhibitory activities.

1.5 Purification of osteoblasts from stroma

Given the multiple subpopulations reported to be present in stroma, methods for enriching or

separating the osteogenic subpopulations is desirable. Several strategies have been used to

produce enriched osteoblast populations from mixed stromal populations, but these can be

grouped into adherence techniques or antibody-based techniques.

Adherence characteristics

While nonadherence to plastic has been reported to be characteristic of the 'majonty' of

osteoprogenitor cells in mouse (Falla et al., 1993) and human (Long et al., 1990; Long et al.,

1995) BM, most BM stromal nodule systems rely on (slow) adherence to plastic for isolation

of osteoblastic cells. There is comparable osteoprogenitor plating efficiency in control and

nonadherent (1 hr. at 37" C in a tissue culture flask) rnouse marrow cultures (Falla et al.,

1993) and in control and nonadherent (24 hr. at 37' C in a tissue culture flask) primary RBM

cultures (Scutt and Bertram, 1995). In the original descriptions of osteogenic rat

(Maniatopoulos et al.. 1988) and chick (Kamalia et al., 1992) BM, stromal cells were

allowed to attach for 24 hr. prior to the first medium change (and medium is changed

cornpletely 3 times per week thereafter). The frequent medium changes are thought to

reduce hemopoietic contamination (Maniatopoulos et al., 1988; Kamalia et al.. 1992) while

expanding the CFU-F (Maniatopoulos et al., 1988). of which osteoprogenitors are a

subgroup; however, significant numbers of hemopoietic cells continue to be present into the

secondary culture even with this protocol (Herbertson and Aubin. 1995). In protocols

described for adult human BM (Cheng et al., 1994). mouse BM (Van Vlasselaer et al.,

1993). and porcine BM (Thomson et al.. 1993). longer adherence times are used. and the

medium is changed only after 5 - 7 d. of primary cultures. Scutt and colleagues have

reported recently that PGE2 can induce cells in the nonadherent population in RE3M to fom

adherent CFU-F, a proportion of which are osteoprogenitors which give rise to osteoblasts

14 which synthesize bone-like tissue (Scutt and Bertram. 1995). In these studies, the action of

PGE2 was shown to be dependent on the glucoconicoid (exogenously added dex) levels

present in the cultures. suggesting that both time and cytokine conditioning are factors to be

considered in the establishment of protocols for enrichment of osteoblastic populations.

An tibodies

The finding that cells at different stages of the osteoblast lineage express antigenic

determinants unique or distinctive to these stages is beginning to be useful for identification

and isolation of osteoblastic cells at various stages of differentiation (Aubin and Turksen,

1995). Thus, antibodies may be useful not only for identification of diverse ce11 lineages

within stroma1 populations, but also to charactenze differentiation stages within the

osteoblast lineage and to fractionate populations based on epitope expression. Of antibodies

raised against osteoblastic cells and reported to date, those recognizing the more mature cells

in the lineage (osteoblasts and osteocytes) have been more commonly found than those

recognizing less mature cells. Enrichment of relatively mature osteoblastic cells has been

reported. Van Vlasselaer and colleagues used irnmunopanning with mAbs to first deplete

populations of hernatopoietic cells, then used flow sorting with Sca-1 and wheat germ

agglutinin binding to purify osteoblastic cells from the 5-fluorouracil treated femoral BM of

young adult mouse (Van Vlasselaer et al., 1994). Two mAbs (OB 7.3 and SB-5) have been

nised that label the surface of chick osteocytes specifically and OB 7.3-coupled magnetic

beads have been used to isolate relatively pure osteocytes from chick mixed bone ceIl

populations (Van der Plas and Nijweide, 1992). Several an tibodies are available that

recognize early as well as late stage osteoblastic cells, e.g. RCC455.4. which we recently

isolated (Turksen et al., 1992) and found by expression cloning to recognize galectin 3

(Aubin et al., 1996). lmmunopanning of human BM cells with osteocalcin and osteonectin

enriches for osteoblastic cells (Long et al., 1995). Antibodies against AP have been raised

by several labs (in Our labontory, RBM 2 1 1.13 was raised against RBM stroma (Turksen et

15 al., 1992)); these label preosteoblastic cells as well as osteoblasts. In the rat calvarial

system, AP expression has been used successfully to fractionate osteoblastic populations

into AP-positive and -negative populations (Turksen and Aubin, 1991). Most of the

osteoprogenitors giving rise to osteoblasts making bone in vitro in the absence of added

glucocorticoids (dex) reside in the AP-positive pool. while osteoprogenitors present in the

AP-negative pool require dex to differentiatiate. AP expression has also been used for flow

cytometric sorting in female Rl3M stromal ce11 cultures, but production of a mineralized

matrix after sorting was only descibed in cells which were AP negative at the time of sorting

(Rickard et al., 1994). There are few reports that describe antibodies recognizing ceils

earlier than the preosteoblast. Haynesworth et al. (1992) irnmunized with human marrow

stromal cells to isolate antibodies that react on a subset of marrow stromal cells and a variety

of tissues in vivo, but do not label osteobiasts or osteocytes (Haynesworth et al., 1992). The

authors concluded that the antibodies may recognize molecules present on early progenitors

or stem cells and that the epitope(s) is lost during osteogenic differentiation, but conclusive

support for this awaits further characterization (Haynesworth et al., 1992). STRO-1

recognizes mesenchymal progeniton in human marrow stromal cells with the ability to form

cells with the phenotype of fibroblasts, adipocytes. smooth muscle cells. and osteoblasts

(Simmcns and Torok-Storb. 199 1 ; Gronthos et al.. 1994). Recently. HOP-26 was described

as an antibody strongly reactive with cells in marrow stromal colonies at early stages of

differentiation, before the induction of AP, and HOP-26 does not bind to mature osteoblasts;

immunopanning with HOP-26 selected the marrow stroma1 fibroblastic colony-forming

units (CN-F) (Joyner et al., 1997). Bruder and colleagues (Bmder et al., 1997) have also

recently reported an antibody which they believe recognizes human mesenchymal stem

cells. As yet, similar antibodies to early progenitors have not been reported for the RBM

stromal system.

1.6 Cytokines and Growth Factors in the Stroma1 Environment

The rich cytokine milieu resulting from production by both hernatopoietic and stromd celIs

creates a complex environment in which both immature osteoprogeni tors and mature

osteoblasts reside. For example, G-CSF (Felix et al., 1988; Greenfield et al., 1993;

Taichman and Emerson. 1994), GM-CSF (Felix et al., 1988; Horowitz et al., 1989; Horowitz

et al., 1989; Weir et al., 1989: Felix et al., 1991; Benayahu et al.. 1992; Greenfield et al.,

1993), M-CSF (Sato et al., 1 986; Shiina-Ishimi et al.. 1986; Elford et al., 1987; Sato et al.,

1987; Felix et al.. 1989; Greenfield et al., 1993; Weir et al.. 1993; Lacey et al.. 1994), IL- la

(Greenfield et al., 1993). ILIP (Birch et al.. 1993; Dodds et al., 1994; Bilbe et al., 1996;

Shadmand and Aubin, 1997), IL-3 (Felix et al., 1988; Birch et al., 1993), IL-5 (Greenfield et

al., 1993), IL-6 (Ishimi et al., 1990; Birch et al.. 1993; Greenfield et al., 1993; Dodds et al.,

1994; Lacey et al., 1994; Bilbe et al., 1996; Malaval et al., 1997; Xiong et al., 1997). IL-7

(Greenfield et al.. 1993). lL-8 (Birch et al., 1993; Chaudhary and Avioli, 1994; Bilbe et al..

1996), IL- 1 1 (Bilbe et al., 1996), LIF (Greenfieid et al., 1993; Bilbe et al., 1996; Malaval et

al.. 1997), ONC-M (Bilbe et al., 1996). PDGF-u (Graves et al., 1989: Zhang et al.. 1991:

Bilbe et al., 1996), PDGF-P (Bilbe et al., 1996: Canalis and Rydziel, 1996). SCF (Bilbe et

al., 1996), TGF-P (Birch et al., 1993; Greenfield et al., 1993; Dodds et al., 1994; Bilbe et al.,

1996). TNF-a (Birch et al., 1993; Greenfield et al., 1993; Bilbe et al.. 1996), and TNF-8

(8 ilbe et al.. 1 996) are produced by osteoblastic cells. Glucocorticoids are known important

regulators of cytokine and growth factor production in many cell types. and this laboratory

has been particularly interested in glucocorticoid regulation of osteoprogenitor

differentiation in both calvarial ce11 cultures and in the RBM stroma1 populations. In RBM

stromal cultures, there is little bone formation in the absence of dex. Work reported in this

thesis will address whether the effect of dex is direct on osteoprogenitor cells or indirect

through its ability to alter the numbers and kinds of other lineages present which would

concomitantly alter the cytokine and growth factor milieu of the stromal cultures. In

17 particular. the ability of dex to stimulate the maturation of monocyte lineage cells into

macrophages (see Chapter 2 and (Herbertson and Aubin. 1995)) would be expected to alter

the levels and kinds of cytokines and factors produced by cells of this lineage, including

PDGF, whose effects on osteoprogenitor differentiation 1 have investigated in detail

(Chapter 4).

1.7 Platelet-Derived Growth Factor (PDGF')

PDGF - General Notes

PDGF is well-known as a mitogen for mesenchymal cells and appears to have broad roles in

tissue development. PDGF exists as 30 kDa disulphide bonded homo- or hetero-dimen of A

and B chains which interact with the two PDGF receptors, a and P. PDGF-B, the normal

counterpart of the oncogene v-sis of the simian sarcoma virus (Westermark and Heldin,

199 l) , has been shown to have stimulatory effects on primitive pluripotential hemopoietic

cells in mouse BM (Yan et al., 1993) and overexpression of PDGF-B in murine hemopoietic

cells induces a lethal myeloproliferative syndrome in vivo with anemia, neutrophilia and

monocytosis (Yan et al.. 1994). PDGF and its receptors show appositional expression in

epitheliallstromal tissues. and appear to be involved in stromal-mesenchymal interactions

during normal embryonic development and in basal ce11 carcinomas (Ponten et al.. 1994).

Measurements of transcript levels during mouse development have shown that PDGF A

chain and PDGF a-receptor are expressed early, while PDGF B chain and PDGF fi-receptor

appear to becorne important later in development (Bowen-Pope et al., 1991). In normal

embryos, PDGF a-receptor is expressed in both mesodermal and neural crest derived

mesenchyme (Schatteman et al.. 1992). The PDGF a-receptor gene is deleted in the Patch

mutant mouse (Bowen-Pope et al., 199 1 ): in hornozygous mutants many mesodermal

derivatives are poorly developed and the embryos do not survive until birth (Bowen-Pope et

18 al., 1991). Specifically in terms of skeletai effects, homozygous embryos demonstrate

severe vertebral malformations. spina bifida and cleft face. and distoriions of other bones

formed by endochondral ossification (Schatteman et al., 1992). The Patch heterozygotes

express half the wild-type level of PDGF a-receptor transcripts and protein and suivive with

minimal defects: an increase in the width of the prefrontal bone and absence of functional

melanocytes (Bowen-Pope et ai., 1991). The Patch mutant mice suggest that PDGF is very

important in mesodemal and skeletal development.

PDGF in Non-Os feogenic Stroma

Fibroblasts and macrophages are sources of PDGF (Bowen-Pope et al.. 1991) and we have

previously reported that both ce11 types are present in RBM stroma1 cells under conditions

which prornote bone formation (Herbertson and Aubin. 1995). PDGF appears to be stored

in the bone matrix. and the trapping or sequestering of PDGF could affect its biological

action through complex modes of release and presentation to responding cells (Hauschka et

al., 1986).

PDGF in Osteoblasts

Normal human osteoblastic cells produce PDGF-A (Graves et al., 1989; Zhang et al.. 1991),

PDGF a- and p-receptors (Zhang et al., 199 1). PDGF B is reportedly not made by human

osteoblastic cells (Graves et al., 1989; Zhang et al.. 199 1: Bilbe et al.. 1996). although there

is a recent report of its production by rat osteoblasts (Canalis and Rydziel, 1996) and mRNA

for PDGF B have been detected in several human osteoblastic ceil lines (MG-63 ce11 line,

SaOs-2. TE-85) (Bilbe et al., 1996). The effect of PDGF on osteogenic cells in vitro has

been investigated by several groups. PDGF increases proliferation (fetal rat calvaria organ

cultures (Canalis and Lian, 1988; Pfeilschifter et al., 1990). fetal rat calvaria ce11 cultures

(Centrella et al., 1989; Centrella et al., 199 1 ; Pfeilschifter et al., 1992). human osteoblastic

celis (Graves et al.. 1989; Gilardetti et al.. 199 1 ; Zhang et al., 199 1 ; Abdennagy et al., 1992).

19 mouse trabecular explant cells (Abdennagy et al., L 992) and the murine osteoblastic ce11 line

MC3T3-E 1 (Davidai et al., 1992)), consistent with PDGFs well known activity as a rnitogen

for mesenchymal cells including fibroblasts. glial cells, and srnooth muscle cells. However,

the evidence in terms of osteoblast proliferation is not conclusive, as most experiments with

nontransformed cells involve possibly mixed cultures, with fibroblasts and other stromai

cells present which are dso known to respond mitogenically to PDGF (Hirata et al., 1985)-

and it has been shown that in fetal rat calvarial tissue culture, PDGF selectively stirnulated

Fibroblast replication and function (Hock and Canalis, 1994).

Effects of PDGF on Osteoblacric Cells In Vitro

The reported effects of PDGF on osteoblast differentiation have been controversial (no effect

- (Canalis and Lian, 1988; Cassiede et al., 1996); positive effect - (Centrella et al., 1989;

Pfeilschifter et al., 1990); negative effect - (Centrella et al., 1991 ; Pfeilschifter et al.. 1992;

Hock and Candis, 1994)). With respect to the effect of PDGF on differentiation, there are

several possible sources of variation, including the PDGF isoform used. Although the two

PDGF receptors are structurally and functionally related, there are important differences

between them in ligand binding: the a-receptor binds al1 three PDGF isoforms (PDGF-AA,

-AB, -BB) with similarly high affinities, but the p-receptor binds only PDGF-BB with high

affinity and binds PDGF-AB with a 10-fold lower affinity (Hart, 1990; Grotendorst et al.,

199 1). The proportions of PDGF receptor types can Vary between different ce11 types (Hart,

1990). In fetal rat calvaria ce11 populations, osteoblasts express primarily @ receptor

cornpiexes, with 30% pp receptor and 10% aa receptor complexes (Pfeilschifter et al., 1992).

There also appear to be important differences between PDGF receptor types in responses

elicited (Pfeilschifter et al., 1992) and the two receptors initiate traverse of the ce11 cycle by

distinct mechanisms (Coats et al.. 1992). In human fibroblasts the PDGF p-receptor, and not

Table 1.1 The Effects of PDGF on Proliferation and Differentiation in

Osteobiastic Ce11 Cui tures

S tudy Culture Prolifn Dserentiation (#)

Candis and Lian (1988) FRC* organ culture + no effect (collagen/osteocalcin)

Graves et al. ( t 989) adult human ceIl cultures + not determined

Centrella et al. (1989) FRC ce11 cultures + + (collagen/totai protein)

Pfeilschifter et al. (1 990) FRC organ cultures + + (collagen)

Zhang et al. (1991) explant human bone cells + not deterrnined

Centralla et al. ( 199 1) FRC ce11 culture + - (ml

Gilardetti et al. ( 199 1) adult human ce11 culture + not determined

Pfeilschifter et al. (1992) FRC ce11 culture + + (collagen)

- Hock and Candis ( 1994) FRC organ cultures + - (collagen)

- (bone rnatrix formation)

Cassiede et al. ( 1996) RBM stroma1 ce11 cultures + - (Ap)

no effect (osteogenesis)

*FRC = fetal rat calvaria

# = markers used as an indicator of bone differentiation are included in parentheses

21 the a-receptor, mediated induction of actin reorganization and membrane mffling

(Westermark and Heldin, 199 1). U-2 OS cells migrated in the presence of PDGF AB and BB

dimers but not in the presence of PDGF AA dimers (Allam et al., 1992). suggesting the

chernotactic effect is rnediated through the $-receptor. Thus the use of a particular isoform,

or a mixture of isoforms, may yield different results depending on the levels and kinds of

PDGF receptors a particular ceIl type expresses.

The fact that different culture systems (e-g., fetnl versus adult, tissue versus cell, clonal

versus mixed population) and different culture conditions (e-g., semm supplementation or

not) are being used to monitor osteoblast responsiveness to PDGF exacerbates the dificulty.

The maturational status or differentiation of progenitor populations can also lead to changes

in receptor levels: although PDGF a-receptor appears to be present throughout osteoblast

development, the levels appear to increase with osteoblast differentiation (Liu and Aubin,

1996). The duration of exposure to PDGF may also be important; most investigations have

used short exposure times (24 hr.) (Candis and Lian, 1988; Centrella et al., 1989; Zhang et

al., 199 1 ) or 48 hr. (Pfeilschifter et al., 1990) to 72 hr. (Hock and Canalis, 1994). Given

earlier data with other factors such as EGF (Antosz et al., 1987). which show biphasic

effects depending upon the time and duration of exposure. longer exposures rnay have

significantly different effects from shorter exposures. The timing of both the pulse and the

phenotypic assessments are also important, since recentIy it was reported chat assessrnents of

proliferation and AP activity of marrow stroma1 cells immediately after exposure to PDGF

were different from assessments done later (Cassiede et al., 1996). Another possible reason

for reported differences is that different markets of osteoblast development have been used

in different studies. For example, total protein synthesis, collagen synthesis, AP and

osteocalcin production have al1 been used as differentiation markers. While PDGF has been

shown to increase collagen synthesis (Centrella et al., 1989; Pfeilschifter et al.. 1990), the

collagen may be incorporated into a fibrous tissue matrix instead of bone matrix

22 (Pfeilschifter et al., 1990). Experiments which use the more specific bone markers such as

AP (although not unique to the osteoblast lineage) or the highly specific osteocalcin suggest

that PDGF inhibits bone formation in vitro (Centrella et al., 1992; Pfeilschifter et al., 1992),

consistent with reports that in calvaria tissue culture, PDGF inhibits bone matrix formation

(Hock and Canalis, 1994). This may be in keeping with the effect of PDGF on smooth

muscle cells: PDGF-BB increases proliferation while inhibiting expression of the

differentiated smooth muscle ce11 phenotype (Ho1 ycross et al., 1992).

Effects of PDGF on Osteobiastic Crlk In Vivo

Similarly to the in vitro studies, there are discrepancies in terms of the effect of PDGF on

bone in vivo. PDGF (A and 8) is expressed at sites of normal human fracture repair,

suggesting a role in regulation of this process (Andrew et al.. 1995). Local administration of

PDGF-BB at the time of osteotomy in rabbits had a stimulatory effect on bone fracture

healing in terms of increased callus density. increased mechanical bone strength. and

osteogenic differentiation (Nash et al.. 1994). In contrast, local administration of PDGF-BB

in rat craniotorny defects was found to inhibit osteogenin induced bone regeneration and

stimulated a soft tissue repair wound phenotype (Marden et al., 1993). Systemic

administration of PDGF-BB to adult fernale rats significantly increased osteoblast number.

bone density and strength, although premature closure of the growth plate, decreased body

fat. and extraskeletal collagen deposition were also noted (Mitlak et al., 1996). The differing

results in the in vivo studies may also be due to the different systems and methods of

administration of PDGF. PDGF is not normally found in plasma and plasma appears to

contain PDGF binding protein(s) that would serve to inactivate PDGF released into plasma

(Bowen-Pope et al.. 1984). Intravenously injected PDGF is rapidly cleared frorn the plasma

in baboons (t 112 = 2 min. (Bowen-Pope et al.. 1984)) and in rodents (90% removed after I

hr. (Cohen et al.. 1990). These results suggest that release of PDGF at sites of vascular

injury would greatly increase the local concentration of PDGF but PDGF not localized to the

23 site of injury would be rapidly cleared frorn the circulation (Bowen-Pope et al., 1984).

Thus, systemic administration of PDGF likely results in only short pulse exposures. In

summary, while PDGF is thought to have critical roles in proliferation during developrnent

and during fracture healing, its role in osteoprogenitor differentiation and bone formation is

stiI1 poorl y understood.

1.8 Thesis Objectives

The objective of this thesis w u to better define and characterize the heterogeneous cellular

environment of the rat bone marrow (RBM) stromal cell osteogenic system. To this end, I

f i n t used histochemical and immunohistochemical methods to identify and quantify the

main subpopulations present in RBM stroma and determine their tirne course of

development under conditions supporting osteogenesis (Chapter 2). I then applied methods

to separate and enrich the osteogenic population from the other stromal populations (Chapter

3). Finally, 1 investigated the effects of PDGF on RBM stromal osteogenesis using both the

mixed populations and the enriched populations (Chapter 4).

CHAPTER 2

Y

Dexamethasone Allers the Subpopulation Make-Up of Rat Bone

Marro w Stroma1 Ce11 Cultures

This chapter has been published in Journal o f Bone and Mineral Research ( 1995) 10.285- 94 as an original paper with the above title (Alexandra Herbertson and Iane E. Aubin)

25

2.1 Introduction

The stromal system of BM refers to the connective tissue elernents which provide structural

and functional suppon for hemopoiesis. Osteogenic cells are known to be present in the BM

stroma of various animal models (Maniatopoulos et al.. 1988; Benayahu et al., 1989;

Friedenstein. 1990; Leboy et al.. 199 1) including human marrow (Haynesworth et al.. 1992).

and production of a bone-like mineralized tissue from marrow cells has been demonstrated

both in vivo, ie.. in diffusion charnkrs loaded with BM cells (Friedenstein. 1968), and in

vitro, where bone-like tissue is synthesized by marrow stromal cells cultured in medium

containing ascorbic acid, 8-glycerophosphate and the glucocorticoid dex (Maniatopoulos et

a 1988; Kasugai et al., 199 1; Satomura and Nagayama. 199 1 ; Malaval et al.. 1992).

However, in addition to osteogenic cells, BM stroma contains other ce11 types: those widely

acknowledged include fibroblasts. adipocytes, and endothelial cells (Dexter. 1 982; McIntyre

and Bjornson, 1986: Owen, 1988; Hasthorpe et al., 1990: Wang and Wolf, 1990; Benayahu

et al., 1991; Bennett et al.. 1991). Other ce11 types identified as components of BM stroma

include srnooth muscle cells (Peled et al., 1 99 1 ; S immons and Torok-S torb, 1 99 1 ; Galmiche

et al., 1993), reticular (Dexter. 1982; Benayahu et al.. 199 1 ) and reticulo-fibroblastoid cells

(Ross et al., 1991).

Hernopoietic cells exist in intimate association with stromal cells in BiM in vivo and in vitro.

Long tenn BM cultures have been used to unravel the differentiation process of many white

blood cell lineages, as well as the role of stromal lineages in this process (Dexter, 1982).

When stromal cells are cultured, the majority of well-differentiated hemopoietic cells are

thought to be removed by centrifugation or. as in our system. by adherence techniques.

However, hemopoietic stem cells are present at a frequency of 1 in 25,000 adherent rnarrow

cells (Kiefer et al.. 199 1 ). and culturing the adherent BM fraction results in the culturing of

early hemopoietic cells whose development is guided by the existing culture conditions. In

27 mice, macrophages are frequently seen in cultures of BM stromal cells (Dexter, 1982) and

isolated macrophage colonies c m be obtained at low plating density (Wang and Wolf,

1990). Dendritic cells, which belong to the myeloid lineage, have also been described in

stromal cultures (Giesler et al., 199 1 : Inaba et al,, 1992).

Although there are some data describing the different ce11 types in RBM stroma, there is

little quantitative information on the subpopulation make-up. One study of pnmary RBM

cultures after 10 d. growth in vitro reponed largely fibroblast-tike (85%) colonies, and by

immunoperoxidase staining 10.3% macrophages-dendritic and granulocytic cells, and 2.5%

endothelial cells (Simmons et al., 199 1 ). Since it is known that the culture conditions have a

significant effect on what lineages proliferate and differentiate, the subpopulations present

need to be determined for each system. There are scant data on the subpopulation make-up

when stromal cells are grown under conditions stimulating bone formation. When RBM

stromal cells were cultured under such conditions (Maniatopoulos et al.. 1988), fibroblast

colony forming units (CFU-F) were found to comprise about 11100th of the adherent ce11

fraction, while osteoprogenitor cells giving rise to bone nodules comprise about 1/300th of

the same plated population i n the presence of dex (Aubin et al., 1990).

Immunocytochemistry, in combination with Northern analysis and morphological

description, confirmed that only a portion of the cells contnbute to osteogenesis (Malaval et

al.. 1994). Neither CN-F nor osteoprogenitor expression followed a linear relationship in

limiting dilution analysis suggesting cooperativity of ce11 types within the population and a

multi-target phenornenon (Aubin et al., 1990). In support of interacting ce11 types, when a

large constant number of nonadherent (hemopoietic) BM cells was added back to adherent

stromal cells, osteoprogenitor cells were measured at a frequency of lllOOth adherent cells

and their expression followed a linear. single-hit relationship. Endothelial cells or

endothelial cell-conditioned medium aiso increased expression of osteoprogenitors giving

rise to bone nodules (Aubin et al.. 1990). These data and others (Hughes and McCulloch.

28 199 1 ; Ogiso et ai., 199 1 ; Zhang et al., 199 1) indicate that the expression of osteogenesis by

BM osieoprogenitor ceils may be under the replation of oîher cells in the BM, which cm

exert either stimulatory or inhibitory activities. As a step toward understanding further the

cell-ce11 interactions affecting bone formation in the RBM stromal systern, we have analyzed

quantitatively the subpopulations over a temporal sequence during which osteoprogenitors

differentiate and bone nodules forrn.

2.2 Materials and Methods

Cell culture and pluring

BM stromal cells from the femora of young 40 - 43-day-old, 1 10 - 120 gram, male Wistar

rats were cultured essentially as described (Maniatopoulos et al., 1988). The rats were

sacrificed by cervical dislocation, and the femora were dissected under aseptic conditions

and placed in medium (a-modified essential medium, a-MEM) containing antibiotics ( 1000

uglml penicillin G (Sigma Chernical Co.. St. Louis, MO), 500 ug/ml gentamycin sulfate

(Sigma). and 3.0 @ml Fungizone (Flow Laboratories, McLean, VA)) (lOxAB). The

adherent connective tissue and muscle were removed from the exterior of the femora, the

femora were pIaced in fresh antibiotic medium as above, and their ends were cut off with a

scalpel. Each fernur was flushed with a-MEM until the bone appeared blanched (about six

or eight times). This suspension was passed through a syringe several times to produce a

largely single ce11 suspension; cells recovered from two femora were added to a T-75 tissue

culture flask (Falcon. Becton Dickinson, Oxnard, CA) and incubated for one week in a 37'C

hurnidified 95% air/5% CO2 incubator. Growth medium. consisting of a-MEM containing

10% fetal calf serum. antibiotics (100 ug/ml pen G, 50 ug/ml gentamycin, and 0.3 uglml

Fungizone), 50 ug/ml ascorbic acid, was changed every 2 - 3 d. After 7 d., each T-75 ce11

culture flask was washed with 15 ml warrn PBS and adherent cells were recovered with a

29 mixture of 10 ml of 0.25% trypsin (w/v in citrate saline) and 5 ml of collagenase (Rao et al.,

1977). Recovered ceils were passed through a syringe with a 22 gauge needle to ensure a

singletell suspension. Cells were counted electronicaily on a Coulter Counter then plated

at densities between 8 x 103 and 5 x 104 cells per 35mm dish (Falcon), or at a range of these

densities to obtain a balance between colony isolation and adequate colony numben for

quantitation (see Results section). For flow cytometry. cells were plated in 100 mm dishes

(Falcon). Cells were cultured in a-MEM supplemented as above, and where indicated, IO-^

M dex (Sigma) for various penods of time from 7 - 35 d. To promote mineralization. 10

m M Na-$-glycerophosphate (Sigma) was added either continuously or for the 2 d. of culnire

prier to fixation. Cultures were used to determine total ce11 number, stained, or prepared for

flow cytometry as below.

Ce11 Counrs

Cells were recovered from a minimum of triplicate dishes by trypsinizing as above and

counted on a Coulter Counter.

Stn in ing

AP/von Kossa/toluidine blue

The histochemical stain for AP is a modification of Pearse's (1960) method (Drury and

Wallington. 1967). Cells were rinsed once with cold PBS and fixed in 10% coid neutral

buffered forrnalin for 15 min., rinsed with distilled water, and left in distilled water for 15

min. Fresh substrate (10 mg Naphthol AS MX-PO4 (Sigma) dissolved in 400 ul N.N-

dirnethylformarnide, added to 50 ml distilled water and 50 ml Tris-HCl (0.2 M. pH 8.3) and

60 mg Red Violet LB salt (Sigma)), filtered through Whatman's No. 1 filter immediately

before use, was added and incubated for 45 min at 20°C. The dishes were then rinsed in

distilled water, drained and stained with 2.5% silver nitrate for 30 min. at 20°C (von Kossa

stain). The dishes were again rinsed in distilled water and drained, toluidine blue was

30 applied for 2 seconds and the dishes were then rinsed 3 times with tap water. The dishes

were finally air dried.

ccNBE

The procedure from Sigma Diagnostics Kit 181 (St. Louis. USA) for aNBE was

followed, except that the counterstaining was omitted. The kit describes the staining

patterns for monocytes and T lymphocytes, but macrophage staining with this stain has been

described in several studies (Giesier et al., 199 1 ; houe et al., 199 1 ; Kanemoto et aI., 199 1 ).

Sudan IV

Sudan IV staining method is modified frorn Clark (Clark, 198 1). Medium was suctioned

from the culture dishes and celIs were fixed in 10% neutral buffered formalin for 1 hr. After

rinsing with distilled water, followed by rinsing twice rapidly with 70% alcohol, Sudan IV

solution (0.1 gm Sudan IV, 50 ml acetone, 50 ml alcohol) was added for 5 - 10 minutes.

Differentiation of the stain was done with 70% alcohol for 2 - 5 seconds, then the dishes

were rinsed with water and allowed to air dry.

Horseradish Peroxidase Coniueated Antibodv S taininq

Medium was suctioned from the culture dishes, ceils were fixed in 3.7% formalin in TBS

(10 mM TRIS pH 7.5, 150 mM NaCI) for 5 minutes, followed by -20' methanol for 5

minutes if permeabilization was needed. Endogenous peroxidase activity was blocked with

3% Hz02 in TBS. The first antibodies, either RBM 2 1 1.13, anti-rat AP (Turksen et al.,

1992) (x 1000 dilution) or ED2 anti-rat macrophage (Dijkstra et al., 1985) (Serotec Ltd..

Toronto, Canada, diluted x500) were diluted in TBS containing 3 9 BSA prior to addition,

and ce11 cultures incubated at 37' for 45 - 60 minutes. After washing, horseradish

peroxidase-conjugated (HPR) CAM (diluted x300) was added and incubation con tinued at

37'C for 30 - 45 minutes. HPR Colour Development Reagent (0.3 g. BioRad CDR in 100

3 1 ml of -20' methanol plus 60 pl Hz02 (30%) in 100 ml TBS, mixed in equal proportions

immediately before using) was added at 20' for one hr.. cells were washed in Hz0 and

ailowed to air dry.

Quan ritarion

Quantitation of colony types involved identification of AP positive colonies by staining as

above, bone nodules by morphology and von Kossa staining, macrophages by aNBE and

ED2 staining, and adipocytes by rnorphology and Sudan IV staining. For quantitation, only

foci containing >10 cells (approximately three population doublings) were classified as

colonies; both colony counts and grid point counts (percentage of grid cross points covered

by colonies) for proportion of the culture dish covered were done. In some experiments.

colonies were graded according to their size (10 - 25 ce11 foci, 25 - 100 cells, or > LOO cells).

For statistical analysis. the Student t test (unpaired, double-sided) was used to determine

levels of significance.

Flow cytornetry

Cells were recovered from dishes by a mixture of trypsin-collagenase as above. passed

through a syringe with a 22 gauge needle, and centrifuged. The cells were resuspended in

Fresh growth medium as above and incubated at 37'C for 30 - 60 minutes to allow the cells

to recover from trypsinization. Cells were centrifuged then resuspended in sterile primary

mouse anti-rat antibodies (mouse myeloma Ig's (Organon Teknika Co.. West Chester, PA)

as a control, RBiM 2 L 1.13 for AP(Turksen and Aubin, 199 1). ED2 (Serotec) for

macrophages (Dijkstra et al.. 1985). MRC OX- 1 (Cedarlane Laboratories Ltd.. Canada) for

leukocyte common antigen (Sunderland et ai.. L979) or MRC OX-22 (Serotec) for leukocyte

common antigen restricted to T and B lymphocytes (Woollett et al.. 1985) diluted in 3%

BSA in PBS, and incubated at 37°C for 30 minutes. After incubation, cells were washed.

incubated with FITC-conjugated sheep anii-mouse antibodies (Serotec) for 30 minutes at

3 2 37'C. and then washed with PBS. Immediately before analysis. cells were filtered through a

fine mesh to remove any ceIl clumps. Propidium iodide was added to determine ceIl vitality.

Sarnples were mn on a Coulier EPICS Profile Analyzer equipped with an Argon laser

operated at 488 nm for fluorescence excitation. Fluorescence was detected with a 525 nm

band pass filier (+/- 10 nrn) and samples gated for size and low propidium iodide staining.

RBM stroma1 cells cultured for 21 d. in the presence of 10% FCS, ascorbic acid, P-

glycerophosphate. and dex, produced mineralized bone nodules similar to those previously

characterized as bone-like (Maniatopoulos et al., 1988; McCulloch et al., 199 1). When

marrow stroma1 cells were cultured as above but without dex, bone nodules were also seen,

but at a much lower frequency. The difference in bone nodule formation in cultures with

and without dex is apparent macroscopically (see Figure 2.1. A and B). Microscopic

assessrnent of bone nodule nurnbers showed significantly greater numbers in the cultures

with dex. compared to those without, in six out of six separate experiments. three of which

are shown in Table 2.1. An increase in the numbers of bone nodules in cultures with dex

was accompanied by a significant decrease in total cell number in these cultures (data not

shown). consistent with previous studies (Maniatopoulos et al.. 1988). Cells associated with

early bone nodules exhibited typical polygonal osteoblast rnorphology and stained positively

for AP (see Figure 2.5A). Bone nodules formed in these cultures over several days: the

number of bone nodules formed plateaued by d. 2 1 (Figure 1.2) but mineralization detected

by von Kossa staining continued to increase up to at least 35 d. when P -glycerophosphate

was present continuously. Because of the variability of the frequencies of different ce11

types within each BM isolation. each separate experiment used a minimum of three different

plating densities in order ro obtain bone nodules in numben satisfactory for both counting

33 and statistics (ie.. 10-150 nodules per 35 mm dish). In Table 2.1. the plating density for

each separate experiment is noted. Both tirne at which new nodule formation ceased and

especially the onset of mineralization were plating ce11 density-dependent: both occurred

earlier when higher densities of bone nodules were present (data not shown).

In cultures in which bone nodules formed, many other ce11 morphologies and colony types

were also evident. Using the same ce11 cultures as those in which bone nodules were

analyzed and the same temporal sequence, we identified and quantitated these other marrow

stroma1 populations with a variety of histochernical stains. In six out of six experiments, a

large percentage of the colonies present were positive for AP, with significantly higher

nurnbers of AP positive colonies in cultures grown with dex (three of which are shown in

Table 2.1). Similarly to the time course of bone nodule formation, the percentage of a

culture dish covered by AP positive colonies (as determined by grid point count) plateaued

after 21 d. However. oniy a proportion of the AP positive colonies were coïncident with and

corresponded to bone nodules (Figure 2.3A. 2.5A). A micrograph of an AP positive colony

with a nonbone appearance is shown in Figure 2-58 . Although the total number of AP

positive colonies appeared to decrease after d. 14 (Figure 2.3A), this is likely due to an

increase in numbers of large AP positive colonies which merge with and become

indistinguishable from the srnaller colonies (see Figure 2-38).

Pronounced differences in aNBE positive (monocyte/macrophage) colonies (Figure 2.5D) in

cultures with and without dex was also apparent. The number of aNBE positive colonies

was significantly higher in dex-containing cultures in six out of six experiments, three of

which are shown in Table 2.1. In contrast to the time course of formation of AP positive

colonies and bone nodules, al1 sizes and total number of aNBE positive

(monocyte/macroph~ge) colonies continued to increase up to ai least d. 28 of culture, the last

34 time point analyzed (see Figure 2.4). There was no apparent consistently close physical

association of bone nodules and aNBE positive colonies (Figure 2.5D).

Fat colonies (Figure 2 . X ) were present at very low frequency under Our culture conditions,

and dex significantly increased the numbers of fat colonies in only one out of six of these

experiments, and only slightly in that particular experiment (see Table 2.1).

Immunolabelling was used to identify more specifically the hemopoietic colonies. ED2

(macrophage) imrnunohistochemically stained colonies (Figure 2.6B) were of similar ce11

size and morphology as those strongly positive with aNBE. However, ED2 positive colonies

were present at similar or higher frequencies to strongly positive aNBE colonies, probably

reflecting the increased sensitivity of the immunohistochemical stain over the histochemical

stain. Quantitation of ED2 positive colonies indicated that macrophage number increased in

cultures grown with dex compared with those without dex in four out of four separate

experiments, three of which are shown in Table 2.1.

We also used flow cytometry to further quantify the stromal ce11 make-up under these

growth conditions. The stromal population exhibited a broad distribution of ce11 sizes

consistent with the many subpopulations present. The number of cells positively labeled

were compared with control cells incubated with nonspecific mouse myeloma Ig's. Similar

to the results from the colony counts. there were more AP positive cells and more

macrophages in the cultures with dex compared with the cultures without dex. For example,

in the experiment shown in Figure 2.7. in the cultures with dex compared to those without,

29.7% venus 12.5%. respectively, of total cells were labeled positively for AP, and 8.3%

versus 2.5%. respectively, of total cells were positive with ED2 antibody. Flow cytometry

was more sensitive than immunohistochemical staining in that it identified relatively more

macrophages in the cultures without dex than were apparent by colony counting (the results

35 from Figure 2.7 are from the same experiment shown as experiment 3 in Table 2.1).

consistent with the cells seen in culture dishes (compare Figure 2.6C to 2.6A), flow

cytometry identified more hemopoietic cells (MRC OX-1 positive) in the cultures without

dex compared to those with dex (Figure 2.7D versus 2.7H). While the majority of the

hemopoeitic cells in the cultures with dex are macrophages (8.3% macrophages out of

1 1.5% of cells identified as hemopoietic cells), a much smaller proportion of the

hemopoietic cells in the cultures without dex are macrophages (2.5% macrophages out of

36.9% hernopoietic cells). The bulk of the hemopoeitic cells in the cultures without dex are

as yet unidentifieci, but do not appear to be lymphocytes as determined by lack of labeling

with MRC OX-22 (data not shown). They do not appear to be of the granulocytic lineage as

they lack the typical morphology and previous histochemical staining trials with naphthol

AS-D chloroacetate esterase (Sigma) and myeloperoxidase (Sigma) for neutrophils and their

precursors were also negative (data not shown).

Table 2.1 Quantitation of colony types identified in RBM stroma1 ce11 cultures grown for 21

d. in vitro with ascorbic acid. P-glycerophosphate, and with or without dex (10-8 M)

Experiment Plating density (+) or # bone % covered with # fat # aNBE+ # ED2+ (35mm dishl (4 dex nodules AP+ colonies colonies colonies colonies

1 5 x 104 + 3444 89.1 +2-1 4 1 2 73k10 91kIO

5 x 104 - O k O 82.9 k 1.1 lkl 3+,1 O+O

2 5 x 1 0 4 + 112k7 98.4i0.2 O k O 119+5 259+1I

5 x 104 * 1 + 1 95.8 f 0.2 1+1 7 + 2 O I O

3 2 x 104 + 1 1 + 2 92.4 k0.7 O+,O 52k5 81+4

2 x 104 4 O f O 87.4 + 1.2 O f O 1 1 + 4 0.4st0.2

Experirnents with + dex > - dex 3/3*** 3/3** 1/3* 3/3*** 3/3***

T-Test of (+) dex venus (-) dex: * p 5 .OS; ** p < -01; *** p 1 .O01

Figure 2.1 Overviews of 35 mm culture dishes grown for 21 d. in medium containing

ascorbic acid and 8-glycerophosphate. Dishes were stained with AP/von Kossa/toluidine

blue; cells cultured with dex (A) or culnired without dex (B). Black von Kossa positive

nodules are apparent in A, while AP positive colonies appear gray in both (magnification

I -6x).

7 14 2 1 28 35

Days in culture

Figure 2.2 Time course of formation of bone nodules in RBM stroma1 cells cultured with

10-8 M dex. Bone nodules first appeared during the second week of u.dture and their

number appeared to plateau during the third week of culture; the proportion of nodules

mineralized appeared to increase with time in culture. In this experirnent, p -

glycerophosphate was present continuously. W mineralized nodules;. total bone nodules

O 10 20 30 40

Days in culture

Days in culture

Figure 2.3 (A) Cornparison of the nurnber of AP positive colonies (u ), the area of

the culture dish they cover (i), and the number of bone nodules (+ ) in RBM

stroma1 cells cultured for various periods of time with dex. Only a proportion of the AP

positive colonies are coincident with bone nodules. (B) Size distribution of AP positive

colonies seen in (A) with time. While the number of small colonies and foci decrease with

time, the number of large colonies increase with time. Colony size distributions: 61 10 - 25

cells; 25 - 100 cells; . > 100 cells

7 14 21 28

Days in cuiture

Figure 2.4 Al1 size classes of strongly positive aNBE colonies in RBM stroma1 cells,

cultured with dex, continued to increase at least up to d. 28, the latest point analyzed.

Colony size: u 10 - 25 cells; h-- 25 - I û û cells; > 100 cells. Total

colonies: &

Figure 2.5 Photomicrographs of the different cell types present in RBM stromal cells

cultured for 21 d. with dex. (A) AP/von Kossaltoluidine blue staining of a bone nodule

where * marks mineralized von Kossa positive nodule and AP positive (appear gray in this

black and white photomicrograph) cells with typical polygonal morphology can be seen at

the periphery of the mineralized nodule (magmfkation ~100) . (B) AP positive cells (appear

gray) (mawcation x100). (C) Sudan N positive (fat) cells showing a colony with two

foci; stained lipid droplets appear black in this black and white photomicrograph

(magnification x30). Inset shows an adipocyte with typical morphology (magmfication

x300). @) aNBE positive (monocytelmacrophage) cells (appear black) not associated with

a bone nodule (magnification x 170)..

1 S . . - .

f -. \ - m. . : -- - a . * - - \ .* -- -- O * . &? * - .'i ,* . . - t - . O -

N .. \

Y;. . 4 8 - - - - * . -. . 0 I - \ -. \ - .- , . - = . i * b

r . * \ . : . ' b . \ . r - - . ' u- 4 - * . - t 0 -. L\- - . *

Figure 2.6 Photomicrographs of irnmunohistochemical staining of RBM stromal cells

grown for 21 d. (A) Control cells of cultures with dex, stained with non-specific myelorna

IgGs. The black in the upper left corner is a portion of a bone nodule. (B) The same

cultures as in (A), but stained with ED2 mAb; a positively stained macrophage colony can

be seen. (C) Control cells for the same experiment as above, but grown without dex. stained

with non-specific myeloma IgGs. In addition to the stromal ceiis, many small unstained

hemopoietic ceiis can be seen - these appear dark when stained with control antibodies (non-

specific IgGs) andor second antibody alone or without any antibodies. (D) The same

cultures as in (C), but stained with ED2 mAb; no positive colonies are seen. Al1

magnifications x 120.

Aw - Dex, conirol 'i 1

B. - Dex. RBM 211.13 D. - Dex. MRC OX-1

E. + Dex. control F. + Dex. RBM 211.13 G. + Dex. ED2 Hw + Dex. MRC OX-1

Figure 2.7 Flow cytometry: frequency histogrm vs fluorescent intensity. (A) and (E) Controls stained with non-specific

myeloma IgGs, for cells grown without dex and with dex, respectively, in the sarne experiment. (B) and (F) Staining with

RBM 2 1 1.13 for AP in crlls grown without dex and with dex, respectively. (C) and (G) Staining with ED2 (macrophages) in

cells grown withoui dex and with dex, respectively. (D) and (H) Staining with MRC OX- 1 (leukocytes) in cells grown

without dex and with dex, respectively. Al1 cells were collected for analysis at d. 21 of culture.

2.4 Discussion

In this paper, we provide quantitative evidence that RBM stromai populations grown under

conditions prornoting osteogenesis and bone formation comprise a heterogeneous mixture of

ce11 types. We confirmed that the addition of dex, similar to its effects in other osteogenic

models (Candis, 1983; Bellows et al., 1986) stimulates osteogenesis and the formation of

bone-like tissue or bone nodules in vitro from adult RBM stroma1 cells (Maniatopoulos et

al., 1988; Kasugai et al., 199 1 ; Leboy et al., 199 1). The presence of other morphologically

distinguishable ceIl types in stromal cultures making bone has been noted on occasion but

not carefully investigated. For example, previously we reportecl that the predominant cells

under similar culture conditions with dex were of spindle-shaped rnorphology (fibroblastic)

or were pleomorphic and the majority did not srain for AP activity (McCulloch et al., 199 1).

More recently, in a molecular and imrnunohistochemical study of osteoprogenitor

differentiation, we confirmed that only a proportion of the cells in RBM stromal cultures are

osteogenic (Malaval et al., 1994). In this paper. we characterized and quantified the

presence of cells of the monocyte-macrophage lineage and adipocytes, both lineages known

to be present in stromal cultures of mouse and human (Dexter. 1982; Wang and Wolf.

1990), but not previously identified in adult RBM stromal cultures undergoing osteogenesis.

Dex significantly increased the numbers of AP positive colonies and cells in RBM stromal

cultures, a finding consistent with reports that dex increased AP enzyme levels in the mouse

B M stromal ce11 line, LMBA- 15 (Benayahu et al., 1989), and in other marrow stromal-derived

ce11 lines (Benayahu et al., 199 1). Interestingly, we found that both the percentage area of

the culture dish covered by AP positive colonies and bone nodule production plateaued in

the RBM stromal cultures at similar times, but only a small proportion (about 1/4) of AP

positive colonies were coincident with bone nodules. Likewise in chick BM stromal

cultures, only a proportion, albeit higher than in rat (55% venus 25%) of the AP positive

45 ceIl population contnbutes to osteogenesis (Kamalia et al.. 1992). This low proportion of

bone AP to total AP colonies could be due to AP positive osteoblastic cells being present,

but either lacking the ability to proliferate sufficiently to form a bone nodule andor being at

stages of differentiation which they cannot form bone in vitro (see aiso discussion in

Malaval et al. (Malaval et al., 1994)). Alternatively, while AP is a well-known marker of

bone cells, it is not exclusive to them. and the presence of other, non-bone, AP positive

lineages is possible. For example. several preadipose ce11 lines are AP positive (Udagawa et

al., 1989; Benayahu et al., 199 1; Dorheim et al., 1993).

Only low numbers of adipocyte colonies were seen under Our culture conditions. This is in

contrat to other systems where production of adipocytes is stimulated by the addition of

glucocorticoids (Dorheim et al., 1993; Bellows et al., 1994). Bennett et al. (Bennett et al.,

199 1) have suggested that there is a bipotential progenitor in rabbit BM stroma which can

differentiate in an adipocytic direction or along the osteogenic pathway. When Beresford et

al. (Beresford et al., 1992) grew adult rat marrow stroma1 cells in the presence of FCS and

dex, both adipocytic and osteogenic cells differentiated, but an inverse relationship between

adipocytic and osteogenic cells was seen. The latter authors showed that with continuous

exposure to dex in both the primary and secondary cultures. the differentiation of osteogenic

cells predominated, whereas the differentiation of adipocytes predominated when dex was

present in secondary culture only. Although dex was present only in the secondary cultures

in our experiments, few adipocyte colonies were seen, suggesting that other factors in our

culture conditions favor differentiation of osteogenic cells over adipocytes, whether there is

a bipotential adipo-osteogenic ce11 present or not (for funher discussion, see Bellows et al.

(Bellows et al., 1994). For example. Beresford et al. (Beresford et al., 1992) used female

rats (we used male), an initial period in the primary where the medium was not changed for

5 - 6 d. (we changed medium 3 tirnes/week), and a shorter secondary culturing time of 8 - 10

d. (versus our tirne of 21 d.), al1 conditions that could markedly affect the subpopuiation

46 make-up of the cultures and influence the ability of multiple celi populations to interact

andfor differentiate.

As previously mentioned, Simmons et al. (Sirnrnons et ai.. 199 1) reported that ptimary RBM

stromal cultures contained 10% "residual marrow components" of macrophagidgranulocytic

leukocytes. Maniatopoulos et al. (Maniatopoulos et al.. 1988) also noted the presence of

small, mononuclear round cells in their primary cultures, and attributed them to the

hemopoietic component of BM. but did not comment on them in the secondary cultures, in

which they analyzed osteogenesis. However, the repeated medium changes in the 7 d.

primary culture penod adapted from Maniatopouios et al. (Maniatopoulos et al.. 1988) and

used here is thought to reduce hemopoietic contamination while expanding the numbers of

CFU-F and decreasing cellular debtis. Clearly, the hemopoietic population recognized by

MRC 0x4 (Cedadane). which detects leukocyte common antigen, is present in the stromal

layer of cells in cultures grown bath in the absence and presence of dex. However, dex

reduced the fraction of hemopoietic cells that express this antigen in these cultures. Of

interest, the majority of the hernopoietic cells present in dex-containing cultures are

macrophages, while a much srnaller proportion are macrophages in the absence of dex.

These other MRC OX- 1 positive cells appear not to label with MRC OX-22, suggesting that

they are not lymphocytes, and by histochemical stains appear not to be neutrophils. They

may be monocytes or more primitive hemopoietic cells, however, the paucity of specific

mAbs to rat hemopoietic cells, especially when compared with those available for mouse

and human, currently hinders their identification.

The finding that macrophage production is stimulated under conditions also stimulating

bone formation is interesting. While macrophages have been reported in murine BM

secondary cultures (Wang and Wolf, 1990), we are unaware of any definitive identification

of hemopoietic cells in secondary cultures of RBM stroma grown under conditions favoring

47 bone formation. Notably, clearly concomitant with stimulating the differentiation of

osteoprogenitors, dex stimulated the maturation of the monocytic lineage into macrophages

in our cultures, possibly while inhibiting proliferation of more primitive precunors or

altering the proportion of other lineages in the hemopoieitic fraction. An aitemate

explanation could be that dex stimulates osteogenic cells or other strornal cells to produce

factors which indirectly stimulate macrophage production. Murine osteoblastic cells are

known to produce M-CSF and GM-CSF (Felix et al., 1988; Horowitz and Jilka, 1992), and

recent reports descnbe the detection of rnRNA transcripts for IL- 1 P , IL-6, IL-8, and TGF-

p 1,2,3 in normal human bone (Birch et al., 1993) and for IL-5, IL-6, IL-7, M-CSF, GM-CSF.

and TGF- P 1 in the MC3T3-E 1 ce11 line and rat primary osteoblastic cultures (Greenfield et

al., 1993). As well, rnouse BM stromal cells produce stimulators such as M-CSF and GM-

CSF (Benayahu et al., l992), and constitutively produce mRNAs for CSF-I . GM-CSF. kit

ligand, and IL-6 (Kittler et al., 1992). Thus it appears that osteoblasts or other stromal cells

secrete many factors that could affect the hemopoietic lineages, including macrophage

production. Of course, the inverse is also possible: with dex altenng hernopoietic lineages,

stimulating production of mature macrophages and altering cytokine Ievels in such a way

that stimulation of bone nodule production results. As well, the differentiation of monocytes

into macrophages could result in the loss of production of inhibitory factors. For example,

IL-IP, which has biphasic effects on bone formation but is inhibitory when present

continuously during the differentiation sequence (Ellies and Aubin, 1990), is secreted by

monocytes, but its secretion decreases with differentiation into macrophages (Kreutz et al.,

1993). In any case. it appears crucial to identify the ce11 populations present in this system,

as the effecrs of any factor on bone formation in this system may be direct or indirect and

mediated through the other ce11 lineages present.

In RBM stromal populations, the culture conditions used to promote growth and

differentiation of AP positive colonies, a subpopuIation of which includes osteoprogenitor

48 cells that differentiate to form bone nodules (colonies of bone), concornitantly reduce the

proportion of hemopoietic cells expressing leukocyte common antigen, while promoting the

growth and differentiation of macrophages. These data show conclusively for the first time

the presence of macrophages in RBM stroma1 cultures grown under conditions favounng

bone formation, extend the information on the subpopulation rnake-up and heterogeneity

under these culture conditions, and contribute to an understanding of the multitarget

phenornenon observed in osteogenic differentiation in RBM stroma.

CHAPTER 3

Ce11 Sorting Enriches Osteogenic P o p lations in Rat Bone Marro w Stroma1

Cell Cultures

This chapter has ken accepted for publication in Bone (December, 1997) as an onginal paper with the above tide (Alexandra Herbertson and Jane E. Aubin)

49

3.1 Introduction

While a number of clonai, established and irnrnortdized bone-derived ce11 lines have been

isolated. many of these appear to show anomalies in aspects of their phenotype and

regulation by hormones and growth factors (for review, see (Aubin et al.. 1993)). It is

therefore of considerable importance to identify the normal osteoprogenitors in populations

freshly isolated from tissue. Cultures of whole marrow stromal cells were first shown to

contain cells with the capacity to fonn bone when they were transplanted in vivo in diffusion

chambers by Friedenstein and colleagues (Friedenstein et al.. 1968). This has been

confirmed repeatedly in vivo with BM cells in diffusion chambers (for example, see (Ashton

et al., 1984; Owen, 1988)) and, more recently. conditions have been established in which

bone formation is observed in vitro in cultures of BM strornal cells from diverse species

including aduk rabbit (Tibone and Bernard, 1982), adult male (Maniatopoulos et al.. 1988)

and female (Leboy et al.. 1991) rat, chick embryo (Karnalia et al.. 1992). neonatal pig

(Thomson et al., 1993). adult mouse (Van Vlasselaer et al., 1993). and adult human

(Vilamitjana-Amedee et al.. 1993). In stroma (McCulloch et al., 199 I), as in calvariai

populations (Bellows and Aubin, 1989; Bellows et al., 1990). the bone nodule assay can be

used as an indirect functional assay of the osteoprogenitor. the end product being a nodule of

osteoblasts/osteocytes embedded in a mineraiized matrix characterized as bone on the buis

of light and phase contrast microscopy, histochemistry including von Kossa staining.

scanning electron microscopy. and transmission electron rnicroscopy (Tibone and Bernard,

1982; Maniatopoulos et al., 1988). The RBM stroma1 ce11 system has been well

characterized in terms of morphological characteristics of differentiating osteoblasts and

their matrix (Maniatopoulos et al., 1988; Satomura and Nagayama. 199 1), in terms of

molecular and irnmunohistochemical details of the expression of rnarkers associated with

bone formation (Kasugai et al., 199 1; Malaval et al.. 1994). and with respect to the effect of

exogenous agents on bone formation (Ogiso et al.. 1991; Zhang et al., 199 1; Notoya et al..

1994; Tenenbaum et al., 1995). However, in RBM stromal ce11 cultures. only a small

5 1 proportion of the cells are osteogenic, and the presence of other morphologicaily

distinguishable ce11 types has been reported (Malaval et al., 1994; Herbertson and Aubin,

1995). For example, in these culture conditions, while some osteoprogenitors can be

assayed in the absence of dex. the assayable number increases markedly in dex,

concomitantly with the growth and differentiation of macrophage colonies (Herbertson and

Aubin, 1995). The presence of multiple ce11 types in RBM stroma1 populations and their

varying proportions from one isolation to another despite ngorous attention to isolation

protocol may account for the considerable variation in the absolute number of bone nodules

formed in different experiments, and may complicate interpretation of cytokine and hormone

effects as direct or indirect on the osteoprogenitors present. The presence of these multiple

subpopulations suggests a need for a method which can punfy the osteogenic component,

including osteoprogenitors, from other lineages in the RBM stroma. To remove the majority

of well-differentiated hematopoietic cells when culturing, centrifugation or adherence

techniques are commonly used. Maniatopoulos et al. (1988) noted in their primary cultures

the presence of small hematopoietic cells and the repeated medium changes in the 7 d.

primary culture period were used to reduce their presence (Maniatopoulos et al.. 1988;

Kamalia et al.. 1992) and cellular debris while expanding the C N - F (Maniatopoulos et al.,

1988), of which osteogenic colonies are a subgroup. However. it is also known that

hematopoietic stem cells are present at low frequency in the adherent marrow ceIl fraction

(Kiefer et al.. 1991) and their subsequent development is guided by the existing cuiture

conditions. Agents for hernolysis and density gradients are other commonly used strategies

in other systems for decreasing hematopoietic contamination (Falla et al., 1993). Nij weide

and collragues reported selec tive cyanide killing of monocyte-macrophage lineage ce1 1s in

mouse marrow stroma1 ce11 cultures (Modderman et al.. 1994); in Our hands, no specific

killing effects were achievable with this method in RBM stroma1 cultures (unpublished

observations). Macrophage and endothelid ce11 cornponents c m also be separated from the

fibroblast-like population by their phagocytic properties (Stein and Stein, 1980; Pitas et al.,

52 198 1 ; Yanai et ai., 199 1 ) and magnetic bead separation techniques based on phagocytosis

(Zhang et al., 1995) have been reported. Row cytometric analysis based on phagocytosis of

fluorescently labeled acetylated low density lipoprotein was found to be Iess effective in

identifying macrophages ccjmpared to mAB labelling (unpublished observations).

In addition to the strategies used to decrease the hematopoietic population, there has also

been investigation into methods to enrich selectively for osteoblast populations from stromal

populations. Selective adherence has been one approach, as sumrnarized in Chapter 1,

Section 1 .S. Increasingly, the availability of mAbs recognizing subpopulations of

osteoblasts (Aubin and Turksen. 1995) is providing means for antibody-based rnethods for

purification. AP expression has been reported previously to be useful in fractionation of

osteoblastic populations into an AP-positive population and an AP-negative population in

the rat calvaria system; interestingly amongst the former were dex-independent progenitors

and in the latter were progeniton which required dex to differentiate (Turksen and Aubin,

1991). AP expression has also been used for flow cytometric sorting in female RBM

stromal ce11 cultures, but production of a mineralized matrix after sorting was only descibed

in cells which were AP-negative at the time of sorting (Rickard et al., 1994). In this paper

we describe flow cytometnc sorting of RBM stroma based on AP expression, with resultant

enrichment for osteoprogenitors capable of dividing and differentiating to f o m osteoblasts

which synthesize bone.

3.2 Materiais and Methods

Ceil culture und plating

BM stroma1 cells from the femora of young 40 - 43-day-old, 1 10 - 120 gram. male Wistar

rats were cultured essentially as described previously ((Maniatopoulos et al., 1988), see

Chapter 2). After 7 d., adherent cells were recovered in a singIe ceIl suspension, then either

counted electronically on a Coulter Counter (Coulter Electronics. Hiaieah, FL) and plated at

densities between 2 x 104 and 4 x lo4 cells per 35 mm dish (Falcon) (unsorted controls), or

labelled and processed through the sorter (see below) prior to plating. After plating. cells

were cultured for an average of 21 d., then stained (APIvon Kossa/toluidine blue, N E , and

Sudan IV) as previously described in Chapter 2.

Flow cytometry

Cells were recovered from flasks or dishes were passed through a syringe with a 22 gauge

needle, and centrifuged. The cells were resuspended in fresh growth medium and incubated

at 37°C for 30 - 60 minutes to allow the cells to recover from the trauma of the procedure.

Cells were centrifuged then resuspended in stenle primary mouse anti-rat antibodies diluted

in 1.5% BSA in PBS: mouse myeloma Ig's (Organon Teknika Co., West Chester. USA) for

controfs, RBM 2 1 1.13 for AP (Turksen et al., 1992), MRC OX- 1 (Cedarlane Laboratories

Ltd., Hornby, Canada) for leukocyte common antigen (Sunderland et al., 1979). ED2

(Serotec Canada, Toronto, Canada) for macrophages (Dijkstra et al.. 1985), or RECA

(Medac GmgH, Hamburg, Germany) for rat endothelial cells (Duijvestijn, 1992), and

incubated at 37°C for 30 minutes. After incubation, celIs were washed twice then incubated

with FITC-conjugated sheep anti-mouse antibodies (Serotec) diluted in 1.5% BSA in PBS

for 30 minutes at 37'C and then washed with PBS twice. Ce11 suspensions were then placed

on ice and flow cytometric sorting was begun immediately; the sorting process took on

average 2 - 4 hr. Immediately before analysis, cells were passed through 22 gauge needles

54 to remove any ce11 clumps. For flow cytometric counting. propidium iodide was added and

populations were dou ble-gated for low propidium iodide staining and size. For ceIl sorting,

samples were also gated for size but propidium iodide was added only to a small control

sample to determine cell vitality and plating numbers were varied accordingly (one sorting

trial with small amounts of propidium iodide - 5 Wrnl cell suspension - left in demonstrated

its toxicity in these populations). Samples were run on a EPICS Ceil Sorter (Coulter)

equipped with an Argon laser operated at 488 nm for fluorescence excitation. Fluorescence

was detected with a 525 nm band pass filter (f 10 nm). Sorting regions were set a minimum

of 3% away from the border of the 1% negative control region to exclude overlapping

populations. Percentage of positive cells (FITC versus Count) was monitored throughout

the sort as a check for antigen capping, although this was not observed to be a problem

dunng the sorting periods. Sorted cells were collected in tubes with a-MEM containing

30% fetal calf serum and antibiotics as above; they were then diluted in medium as above

and plated accordingly. For son controls, cells were stained as above. one gate expanded to

include 100% of the population, and cells were passed through the machine prior to

collection.

Quantitation

Quantitation of colony types involved identification of AP-positive colonies by

staining (red) as above, identification of bone nodules by black von Kossa stained mineral

deposits surrounded by cuboidal AP-positive stained cells. monocytes/rnacrophages by

aNBE staining (dark brown). and adipocytes by (red) Sudan IV staining and morphology.

For quantitation, only foci containing >10 positively stained cells (approximately three

population doublings) were classified as colonies; both colony counts and gnd point counts

for proportion of the culture dish covered were done. For statistical analysis, the Student t

test (unpaired. double-sided) was used to determine levels of significance. unless the

standard deviations differed significantly. in which case a Welch t test was used.

When RBM strornal cells were analyzed by flow cytometry after either 7 d. primary culture

or after 7 d. primary and 18 d. secondary culture. they exhibited a broad distribution of ce11

sizes consistent with the multiple hematopoietic and mesenchymal subpopulations present

(Herbertson and Aubin, 1995). necessitating the use of both log forward scatter (related to

ce11 size) and log side scatter (related to ce11 granulaiity) in analysis and sorting parameters.

In both counting and sorting experiments, the populations were first gated for size: using log

forward scatter, the gated area was set to remove smdl fragments and debris or discard any

ohvious ce11 clumps, with a lower limit consistent with platetelet size (approximately 2 pm)

and an upper limit consistent with large cultured cells (approximately 30 pin). On the basis

of fluorescence, control cells incubated with mouse myeloma Ig's as the ptimary antibody

were gated to 1 .O% (see Figure 3.2A) to eliminate 99% of the background autofluorescence

(a normal characteristic of a11 viable mammalian cells (Aubin, 1979) and marked in sorne

BM stromal cells (Keller et al., 1993)) and any non-specific staining (see Figures 3.1 A and

3. IB. and control in Figure 3.2). With these conditions, RBM populations labelled with the

anti-rat AP antibody (RBM 21 1.13) displayed a broad range of labelling intensity (see

Figure 3.1 C).

In RBM stromal cells grown for 7 d. primary and then 18 d. secondary culture. multiple cell

lineages are represented, with subpopulations varying according to the presence or absence

of dex, as we previously reported (Herbertson and Aubin, 1995). Typically. more cells

labeled with RBM 2 11.13 (for AP) when RBM strornal cells were grown in the presence

(74.7%) than in the absence of dex (13.9%) (see Figure 3.2A and 3.2B). In contrast, cells

staining for MRC OX I (for leukocyte common antigen) were present at lower frequency in

cultures with dex (2.8%) than without dex (21.2%). A small percentage of cells stained

positively with ED2 (anti-macrophages) only in the cultures with dex (2.0%). RECA, an

56 antibody specific for rat endothelial cells (Duijvestijn et al.. 1992), stained few cells in either

cultures with (1.1 %) or without dex (0.9%). representing < 1% of cells over backgound. [n

multiple experiments, by either flow cytometry or immunohistochemical staining (data not

shown). negligible numbers of ceils stained with RECA, suggesting that endothelial cells are

present in Iow numbers in RBM stroma1 ce11 cultures under these growth conditions.

For ce11 fractionation after 7 d. primary culture, sorting gates were set to capture cells either

~ ~ h i s h (cells intensely labeled with RBM 2 11.13) or A P ~ O W (clearly negative for RBM

2 1 1.13 labeling) as indicated in Figure 3.1 ; in this experirnent 3 1.2% of cells were in the

former category and 32.7 % in the latter (see Figures 3.1B and 3. ID). A P ~ ~ ~ , A P ~ O ~ , or

control cells were plated to determine osteoprogenitor distribution and frequency. Similar to

results we reported previously with flow cytometrk sorting of fetal rat calvarid populations

(Turksen and Aubin. 1991), and consistent with reports that murine osteoblastic cells (Van

Vlasselaer et al., 1994) and stroma initiating cells are depleted by passage through a flow

cytometer (Deryugina and Muller-S ieburg, 1 993; Deryugina et al., 1995)- approximately

50% (on a per plated ce11 basis) of nodule forming cells are lost during processing/sorting

when compared to unsorted controls. In some experiments, we therefore added an

additional control, measuring ceIl viability with propidium iodide at the end of labelling

manipulations and then adjusting plating densities to compensate for reduced ceIl viability.

The nurnber of AP-positive colonies was enriched in the ~ ~ h k h fraction of RBM cells

cornpared to either the unfractionated cells or the AP'OW fraction. whether or not the sorted

cells were cultured in the presence or absence of dex (see Table 3.1). However, since

numbers of bone nodules and macrophage colonies are low i n cultures without dex

(Herbertson and Aubin, 1995) and even in the A P ~ @ fraction bone nodule numbers were

significantly higher (p 1 -05) when cultured with dex versus without dex, and since trends in

bone nodule or macrophage colony numbers were not signifiant in minus-der cultures,

subsequent results are reported only for dex-containing cultures. Table 3.2 summarizes 5

57 separate expenments. It is clear that within the A P ~ ~ ~ ~ fraction the number of AP-positive

colonies and osteoprogenitors (giving rise to bone nodules) were signifkantly enriched

compared to control, but as we reported earlier. bone colonies comprise only a srnall

proportion of the total AP-positive colonies (Herbertson and Aubin, 1995). The increase in

osteoprogenitor frequency ranged from approximately 2 to 100 fold (see Table 3.3). The

fold increase in osteoprogenitor frequency did not appear to correlate with either the initial

percentage of cells positive for AP in the culture at the time of sorting, nor did it appear to

correlate with any fold change in macrophage colonies (as determined by N E staining)

(see Table 3.3). In addition, the fold increase in osteoprogenitor frequency did not appear to

correlate with the post-sorting purity check; both experiments B and C in Table 3.3 had fold

increases of 1 . 7 ~ but different post-sorting purïties (B post-sorting purity was 95.8%. C

post-sorting purity was 96.5%). while in experiment D in Table 3.3 the post-sorting purity

was lower, 93.5%. but the fold increase was higher at 3.5~.

Concomitantly with the significant enrichment of osteoprogenitors, the ~ P h i g h fraction of

cells was usually significantly depleted for macrophages (in 415 experiments listed in Tables

2 and 3). In the anomalous experiment. experiment C in Table 3.3. macrophage numbers

were very high in the unfractionated control; while the total number of macrophage colonies

increased in the A P @ ~ fraction, in this experiment the number of large (A00 cells)

macrophage colonies in the AP hieh fraction (3.3 k1.5 colonies) significantly decreased (p 5

.O\) cornpared to the unfractionated control (13.0 t 2.0 colonies). The apparent increase in

macrophage colonies is likely due to the decrease in number of large macrophage colonies.

which tend to overgrow and obscure srnaller colonies that may also be present. Thus.

although total colony number suggests othenvise. in experiment C the actual number of

individual macrophages probably decreased. In contrast. macrophages appear to be

consistently ennched in the APIOW fraction of cells, a fraction with no consistent changes

from control in either AP-positive colonies or osteoprogenitors (see Table 3.2).

58 Very low numbers of adipocyte colonies fonn in RBM stromal ce11 cultures under these

growth conditions (Herbertson and Aubin. 1995). Of the 3 experirnents in which adipocyte

colonies were quantitated, 2 had fat colonies present in large enough numbers (> 4 colonies

per dish in the unfractionated cells) to be analyzed statistically. Within the fraction

of cells. adipocyte colonies were significantly depleted, while they were significantly

enriched in the AP~OW fraction (see Table 3.2 and Figure 3.3).

While enrichment of bone nodules and AP-positive colonies is seen within the ~ ~ h i g h

fraction, it is also apparent thar this fraction is still a mixed population. For example. culture

dishes are not completely covered with AP-positive colonies (see Table 3.1). reflecting the

presence of AP-negative colonies. and while the macrophage colony number is generally

decreased, some are still present (see Table 3.3). As mentioned earlier, post-sort purity

checks (re-analyzing the sorted population on the flow cytometer) of the A P ~ @ cells

showed that this fraction was minimally 93 - 97% pure; thus some of the non-AP-positive

colonies may reflect the 3 - 7 9% "contaminating" cells. Post-sort purity checks of the AP~OW

cells showed that this population was minimally 99% pure, reflecting the relative ease in

exciuding a highly stained cell.

MRC OX-1 is a mAB reactive against leukocyte common antigen, which is known to be

present on > 95% of thymocytes, BM cells, and thoracic duct lymphocytes (Sunderland et

al., 1979). Two sorting trials with this antibody were perfomed to see if they yielded

"cleaner" RBM stromal populations, ie.. populations better depleted of hematopoietic cells.

While the removal of the majority of the hematopoietic ce11 fraction increased AP-positive

colony and bone nodule numbers, macrophages were also present in numbers equal to or

greater than those in unfractionated control populations (results not shown).

59 Previous work in our labontory has shown that in RBM stroma1 populations osteoprogenitor

expression, as determined by bone nodule formation. does not follow a linear relationship in

limiting dilution analysis until very high ce11 densities are reached, suggesting cooperativity

of ce11 types within the population and a multi-target phenomenon (Aubin et al., 1990)

(Aubin, 1997. submitted). When we analyzed osteoprogenitor frequencyhodule number

versus plating density in control versus A P ~ @ and A P ~ O W populations, osteoprogenitor

expression still followed a non-linear relationship (Figure 3.4A). Similarly, but not to the

same degree, the aNBE positive macrophage colonies appear to follow a non-linear

relationship with decreasing plating densities (Figure 3.4B).

Table 3.2

Net clianges (enrichnient/depletion) in various coloiiy types in AP high and AP low fractions, compured io unfractionated controls.

Al1 cultures contüined dex üfter plating. Changes indicated by the ürrows were significant at minimally p 5 .O5 as determined by t

tests, or the trend seen i s indicated.

I

Al kaline Phosphatase

Bone Nodules

aNBE i- Colonies

Fat Colonies

A P High

1' 515

AP L o i v

$315 (trend & 1/5) ?1/5

'r 5/5

4 315 (trend 4 115) 1' 115

5 212

1 3/5 215

1' 415 (trend 7 115)

7' 2/2

Figure 3.1

Flow cytometric sorting. (A) and (C) are frequency (count) histograms versus ce11

fluorescence (FITC) intensity for control (incubated with mouse myeloma I, 0's as the

primary antibody) and RBM 2 1 1.13 (AP) stained RBM stroma1 populations respectively. In

(A), the control cells have the lower cut-off of cursor 'a' set to I .O%. 1 n (C) , the number of

cells in the area 'a' represents those positively stained for RBM 2 1 1.13 (38.8%). (B) and

(D) are forward scatter log (FS log; representing size) venus FITC intensity contour maps

for control and RBM 2 1 1.13 stained populations respectively. The sorting gates were set to

capture cells with either high (gate 'c') or low (gate 'b') expression of RBM 21 1.13. In (B)

gate 'b' contains 66.4% and gate 'c' contains 1 % of the total population, whereas in (D) gate

'b' contains 32.7% and gate'c', which has been shifted right relative to the control, contains

3 1.28 of the total population.

65 Figure 3.2 Legend

Flow cytometnc analysis. (A) and (B) are frequency (count) histograms versus ce11

fluorescence (FITC) intensity in RBM stromal cell cultures grown in secondary cultures for

18 d. without or with dex, respectively. From top to bottorn: RBM stromal cells are stained

with control mAb (incubated with rnouse myeloma Ig's as the primary antibody) and RBM

21 1.13 (anti-AP), MRC 0x4 (anti-LCA), ED2 (anti-macrophage), and RECA (anti-rat

endothelid cell antigen).

ControI High Low

RBM 211.13 Expression

Figure 3.3

Adipocyte colony numbers for RBM stroma1 populations sorted according to RBM 2 1 1.13

(AP) expression. Data is frorn the experiment designated "D" in Table 3. T tests show that

adipocyte colonies are significantly lower (p 1 -01) in AP high (high RBM 21 1.13 binding),

and are significantly higher (p 5 -01) in AP low (low RBM 2 1 1.13 binding) populations.

Control Low High

RBM 211.13 Expression

# cells platedn5 mm dish

O 2 x lW4)

0 IxlO(4)

5 x l q 3 )

2 5 x lO(3)

R 1 x W 3 )

Y cdls p l a t e d / 3 ~ m dish 2 x lO(4)

1 x lO(4)

5 x lO(3)

2.5 x 10(3)

1 x W 3 )

Control Low High

RBM 21 1.13 Expression

68 Figure 3.4 Legend

Bone nodules (A) or aNBE colonies (B) versus plating density for RBM stroma1 populations

sorted according to RBM 2 1 1.13 (AP) expression. Data are from the experiment designated

"C" in Table 3.3. In (A), t tests show that osteoprogenitor (capable of giving rise to a bone

nodule) numbers are significantly lower (p S .O 1) in the AP low (low RBM 2 1 1.13 binding),

and are significantly higher (p 5 -01) in the AP high (high RBM 21 1-13 binding) at al1

densities from 2 x 104 to 2.5 x 103. In (B), t tests show that aNBE positive colony numbers

are significantly lower (p 5 .O 1) in AP low (low RBM 2 1 1.13 expression) populations only

at the 2 x 104 density, and are significantly higher (p < .05) in AP high (high RBM 21 1.13

expression) populations at densities from 2 x 104 to 2.5 x 103.

3.4 Discussion

Previous work from Our lab has shown the presence of multiple ce11 types in RBM stroma1

popul;itions, including cells of the monocyte-macrophage lineage, fibroblasts, and

adipocytes (Herbertson and Aubin. 1995). Notably the addition of dex, concomitant with

stimulating the differentiation of osteoprogenitors to bone nodule forming cells, stimulated

the maturation of the monocytic lineage into macrophages in our cultures, possibly while

inhibiting proliferation of more primitive precurson (Herbertson and Aubin. 1995). In

cultures with dex, macrophages appear to make up the majority of the hematopoietic cells

present (Herbertson and Aubin, 1995). Despite rigorous attention to ce11 isolation

techniques, the proportions of hematopoietic versus mesenchymal subpopulations present in

RBM stroma appear to Vary from one isolation to another and rnay account for the

considerable variation in the absolute number of bone nodules seen in different experiments.

Consistent with the hypothesis that heterotypic cell-cell interactions influence osteogenesis

in the RBM system (Aubin et al., 1990; Aubin, 1997, submitted), the presence of these

multiple subpopulations also complicate the interpretation of direct versus indirect cytokine

effects in BM systems (Thomson et al., 1993; Herbertson and Aubin, 1995). Notably, in Our

fractionated populations, more than one cell type remains limiting in the populations,

suggesting that cell-ce11 interactions still appear to influence osteoprogenitor differentiation

and nodule formation; in addition, i t is possible that homotypic ce11 interactions, i.e.

"community effects" may also play a role as hypothesized recently based on limiting

dilution and other frequency analyses (Aubin, 1997, submitted).

Clearly, in the ~ ~ h i g h fraction of cells we ennched for osteoprogenitors which divide and

differentiate to give rise to bone nodules, although the resultant cultures still comprised a

mixture of lineages. This mixture of lineages may reflect the 3 - 7 8 "contaminating" cells

in the ~ ~ h k h fraction, and the 1% "contaminating" cells in the A P ~ O W fraction. Although

70 flow cytometric sorting can yield populations of high purity, 100% purity is only ever

approximated unless single ce11 cloning methods are added to the isolation step. Given the

large range of ce11 sizes seen in the RBM stromal populations, Our sort gates were

accordingly set wider, which would also tend to reduce the stringency of the sort.

Nevertheless, the 93 - 978 purity obtained in our sorting experiments are sirnilar to those

achieved by an improved immunopanning technique applied to isolate CD 34+ cells from

human BM (93.54 i 3.4 % purity). and better than the 57 - 80% purities obtained in some

other immunopanning procedures (Cardoso et al., 1995). Multiple antibody sorting steps

and an imposition of single ce11 collection methods may help to reduce the presence of the

contarninating cells further.

The mixed nature of the ~ ~ h ~ g ~ population could result not only from limitations in the

efficacy of sorting, but also from cells that have reverted from AP-positive osteoblastic cells

to a more immature, AP-negative phenotype, or the presence of other, non-bone, AP-

positive lineages. The presence of AP-positive colonies and osteoprogenitors (giving rise to

bone nodules) in the AP**W fraction is likely representative of the presence of immature

progenitors which, at the time of sorting, are not yet expressing AP (see also (Turksen and

Aubin. 1991)). While AP is a well known marker of the mature osteoblast lineage. AP is

also a cytochemical marker of 'reticular' cells in marrow which make up a large proportion

of the marrow stroma (Westen and Bainton, 1979; Cattoretti et al., 1991): these cells have

often been postulated to be preadipocytes (Bianco et al., 1988; Ross et al., 1991) which are

known to be AP-positive (Udagawa et al.. 1989; Benayahu et al., 1991; Dorheim et al.,

1993). However, in our stromal ce11 cultures with the highest numbers of fat colonies

present in the controls, fat colony number was decreased in the ~ ~ h g h and increased in the

AP'OW fraction of cells. Although isolated brain capillary endothelial cells express high AP

activity which decreases with ce11 culturing (Meyer et al., 1990). bone endothelial cells have

been reported to express oniy low levels of AP (Rickard et al., 1995). We would thus expect

71 these not to be found in the ~ ~ h i g h fraction of cells. especially since direct quantitation of

endothelial cells by endotheliai specific antibodies suggested that they do not form a

significant proportion of the RBM stromal ce11 cultures under our culture conditions.

While the finding that cells at different stages of the osteoblast lineage express antigenic

determinants unique or distinctive to these stages is beginning to be useful for identification

and isolation of osteoblastic cells at various stages of differentiation (Aubin and Turksen,

1995), we are still limited by a lack of specific antibodies definitely recognizing cells earlier

than the preosteoblast. Haynesworth et al. (1992) reported isolation of antibodies that react

on a subset of human rnarrow stromal cells and a variety of tissues in vivo, but do not label

osteoblasts or osteocytes (Haynesworth et al.. 1992). The authon concluded that the

antibodies may recognize molecules present on early progeniton or stem cells and that the

epitope(s) is lost during osteogenic differentiation (Haynesworth et al., 1992). STRO- 1

recognizes human marrow stroma1 mesenchymal progenitors with the ability to form cells

with the phenotype of fibroblasts. adipocytes. srnooth muscle cells. and osteoblasts

(Simmons and Torok-Storb. 199 1 ; Gronthos et al.. 1994). Very recently. HOP-26. a mAE3

against hurnan marrow stromal osteoprogeniton, has been described (Joyner et al.. 1997).

HOP-26 labels an epitope present on osteoprogenitors before the induction of APT but dues

not bind to osteoblasts, and its presence on the ce11 surface allowed enrichment of human

marrow CN-F including osteoprogenitors (Joyner et al .. 1997). As ye t, similar antibodies

to early progenitors do not exist for the rat nor have the macromolecules recognized by these

antibodies been identified to allow preparation of rat-specific reagents. Enrichment of

relatively mature osteoblastic cells has also been reported. Immunopanning of human BM

cells with osteocalcin (normally considered a marker of mature osteoblasts) and osteonectin

has yielded sorne enrichment for cells with osteoprogenitor properties (Long et al., 1995).

Van Vlasselaer and colleagues used immunopanning with mAbs to first deplete populations

of hematopoietic cells, then used flow soning with Sca- 1 and wheat germ agglutinin binding

72 to enrich for osteoblastic cells from the 5-fluorouracil treated femoral BM of young adult

mouse (Van Vlasselaer et al.. 1994). Two mAbs (OB 7.3 and SB-5) that label the surface of

chick osteocytes specifically have been used (with magnetic beads) to isolate relatively pure

osteocytes from chick mixed bone ceIl populations (Van der Plas and Nijweide, 1992).

MAbs against AP have been raised by several labs (in Our lab, RBM 2 1 1.13 was raised in an

injection of RBM stroma (Turksen et al., 1992)); these label preosteoblastic celis as well as

osteoblasts. In the rat calvarial system, AP expression has been used successfully to

fractionate osteoblastic populations into AP-positive (highest 10% of cells) and AP-negative

populations (Turksen and Aubin, 1991) and this report extends that work into the RBM

stromal system. In the rat calvarial system, the fractionation resulted in osteoprogenitors

being separated into two "classes": one capable of expressing bone formation without

exogenous dex stimulation (high AP-positive). and one expressing bone formation only in

its presence (AP-negative) (Turksen and Aubin. 199 1). These data are consistent with the

hypothesis that these two populations of progenitor cells represent different stages of

differentiation. with the latter representing a less mature ceIl than the former and requiring

different regulatory signals for its expression than the latter (Turksen and Aubin, 199 1).

Given that few bone nodules f o m in RBM stromal ce11 cultures in the absence of dex, and

the marked stimulation of bone nodule formation noted with dex. it is tempting to speculate

that the majority of osteoprogenitors in stromal cultures. especially as assayed in the

secondas, cultures. may be of the relatively immature phenotype. In our experiments, we

Found few assayable osteoprogenitors (little bone nodule production) in the APL& fraction

in the absence of dex. suggesting that RBM stroma contains largeiy dex-dependent

osteoprogenitor populations and the possibility that dex is required later than when AP is

first expressed.

73 Flow cytometric sorting of RBM stroma1 populations according to high or low AP

expression is an effective technique for ennchment of AP colonies and osteoprogenitors

which c m divide and differentiate to give rise to bone nodules, and should prove useful for

further studies of the nature of the osteoprogeniton and their regdation in this tissue.

CHAPTER 4

PDGF has Biphasic Effects on the Differentiation of Osteoprogenitors

in Rat Bone Marrow Stroma1 Ce11 Populations

This chapter has been submitted for publication with the above title (Aiexandra Herbertson and Iane E. Aubin)

7 4

4.1 Introduction

PDGF, 30 kDa disulphide bonded homo- or hetero-dimers of A and B chains which interact

with the two PDGF receptors, a and P, appears to have broad roles in tissue development

and repair (Bowen-Pope et ai., 199 1; Schatteman et al.. 1992; Yan et al., 1993; Ponten et al.,

1994; Yan et ai., 1994). With respect to bone, normal human osteoblastic cells produce

PDGF-A (Graves et al., 1989; Zhang et ai., 199 1) and PDGF a- and fi-receptors (Zhang et

al ., 1 99 1 ). PDGF B is reportedly not made by human osteoblastic cells (Graves et al ., 1 989;

Zhang et al., 1991; Bilbe et al.. 1996), although there is a recent report of its production by

rat osteoblasts (Candis and Rydziel. 1996) and mRNA for PDGF B has been detected in

several human osteoblastic celi lines (MG-63 ce11 line, SaOs-2, TE-85) (Bilbe et al., 1996).

PDGF has been reported to affect both proliferation and aspects of the differentiated

phenotype of osteoblastic populations in vitro. It increases proliferation in many

osteoblastic mode1 systems (fetal rat calvaria organ cultures (Canalis and Lian, 1988;

Pfeilschifter et al., 1 WO), fetal rat calvaria ceIl cultures (Centrella et al., 1989; Centrella et

aI., 199 1; Pfeilschifter et al., 1992), human osteoblastic cells (Graves et al., 1989; Gilardetti

et al.. 1991; Zhang et al., 1991; Abdennagy et al., 1992), mouse trabecular explant cells

(Abdennagy et al., 1992) and the murine osteoblastic cell line MC3T3-El (Davidai et al.,

1992)), consistent with its well known activity as a mitogen for other mesenchymal cells

inctuding fibroblasts, glial cells, and smooth muscle cells. However. whether it is

osteobiastic cells alone responding in most of these models is not clear, as most experiments

with primary ce11 cultures involve mixed cet1 populations. with fibroblasts and other stroma1

cells present which are also known to respond mitogenically to PDGF (Hirata et al., 1985).

In this regard. PDGF was found to stimulate seiectively fibroblast replication and function in

fetal rat calvarial organ culture (Hock and Candis, 1994).

76 The reported effects of PDGF on parameters of osteoblast differentiation have k e n diverse

(no effect - (Canalis and Lian, 1988; Cassiede et al., 1996); positive effect - (Centrella et al..

1989; Pfeilschifter et al., 1990); negative effect - (Centrella et al.. 199 1 ; Pfeilschifter et al.,

1992; Hock and Candis. 1994)). These conflicting results may mise from several different

sources, including the PDGF isoform used (there are isoform differences in ligand binding

(Hart. 1990; Grotendorst et al., 199 l)), and differing proportions of ceIl subpopulations with

different PDGF receptor types (Hart, 1990). which mediate different responses (Westermark

and Heldin. 1991; Allam et al., 1992; Coats et al., 1992; Pfeilschifter et al., 1992). The fact

that different culture systems (e-g.. fetal versus adult. orgadtissue versus ceil. clonal venus

mixed populations) and different culture conditions (e-g., serum supplementation or not) are

being used to monitor osteoblast responsiveness to PDGF exacerbates the difficulty. The

maturational status or differentiation of progenitor populations can also lead to changes in

receptor levels: although PDGF a -receptor appears to be present throughou t osteoblast

development. the levels appear to increase with osteoblast differentiation (Liu and Aubin.

1996). The duration of exposure to PDGF may also be important; most investigations have

used short exposure times (24 hr. (Canalis and Lian, 1988; Centrella et al., 1989; Zhang et

al., 199 1 ), 48 hr. (Pfeilschifter et al., 1990) or 72 hr. (Hock and Canalis. 1994)). Given

earlier data with other factors such as EGF (Antosz et al., 1987; Aubin et al., 1992). which

elicit biphasic effects depending upon the time and duration of exposure. and when during

the proliferation/maturation sequence they were given, the effects of PDGF might be

expected to Vary depending on the treatment protocol. This was found to be the case. for

example. when the effects of PDGF on proliferation and AP activity of marrow stroma1 cells

were compared immediateiy after exposure to PDGF versus later (Cassiede et al.. 1996).

Finally. different parameters to measure osteoblast development have been used in different

studies. For example. total protein synthesis. collagen synthesis. AP andor os teocalcin

production have variously been used as differentiation markers. While PDGF has been

shown to increase collagen synthesis (Centreth et al., 1989; Pfeilschifter et al.. 1990), the

77 collagen may be incorporateci into a fibrous tissue matrix rather than a bone matnx

(Pfeilschifter et al., 1990). Consistent w ith this. experiments which use more specific bone

markers. such as AP (although not unique to the osteoblast lineage) and osteocalcin, suggest

that PDGF inhibits bone formation in vitro (Centrella et al., 1992; Pfeilschifter et ai., 1992;

Hock and Candis, 1994).

Osteoprogenitor cells present in the stromal population of young-adult RBM, cultured in

medium containing ascorbic acid. P-glycerophosphate and dex, divide and differentiate to

give nse to osteoblasts which synthesize bone-like tissue (Maniatopoulos et al., 1988): the

number of bone nodules is quantifiable and provides a very specific, terminal differentiation

product by which to measure the effects of PDGF on osteoblast differentiation. Given the

established proliferation-differentiation sequence charactenstic of the RBM mode1 (see, e.g.,

(Malaval et al., 1994)), this system provides an opportunity to target specific events by

varying the exposure times to PDGF. Further, BM stromal cells may mimic the in vivo

environment, with its mu1 tiple subpopulations and different lineages present. For example,

fibroblasts and macrophages, which we have previously reported to be present in RBM

stromal cells grown under conditions which promote bone formation (Herbertson and

Aubin, 1995). are also sources of and responders to PDGF (Bowen-Pope et al.. 199 1). Thus,

while BM stromal ce11 cultures rnay recapitulate the complex cellular interactions that occur

in vivo. establishing underlying mechanisrns as direct or indirect on particular

subpopulations can be difficult. Flow cytometnc soning of RBM stroma1 cells on the basis

of AP expression, however. allows separation of the stroma into A P ~ ~ and A P ~ O W

populations. Within the A P ~ ~ ~ ~ fraction, the number of AP-positive colonies and

osteoprogenitors (giving rise to bone nodules) are significantly enriched and adipocyte and

macrophage colonies depleted compared to the unfractionated control (Herbertson and

Aubin, 1997). In contras, within the M l o w fraction of cells. adipocyte and macrophage

colonies are consistently enriched (Herbertson and Aubin, 1997). We have used the RBM

78 stromal system, in combination with flow cytometric sorting, as a mode1 for investigating

the role that PDGF plays in osteoprogenitor differentiation.

4.2 Materials and Methods

S e m

Levels of PDGF in semm Vary depending on how semm is prepared and the species

used (e-g.. PDGF levels in whole blood semm from normal humans (17.5 t 3.1 ng/ml),

baboons (2.7 ng/rnl) (Bowen-Pope et al., 1984) and fetal bovine serum (0.929 f 0.23 ng/mI)

(Hyclone Defined Fetal Bovine Serum)). In these experiments. we used fetal bovine serum

at a final concentration of 10%; even with the higher concentrations of PDGF reported (17.5

ng/ml for human whole blood serum), this would yield basal PDGF levels of approximately

1 - 2 ng/ml. In experiments in which Hyclone "defined fetal bovine serum" was used, the

serurn would contribute only ,093 ng/ml PDGF, negligible compared to the additional

concentrations (0.25 - 10 ng/ml) added. PDGF-BB is the major isoform seen in semm from

most non-primate species (Bowen-Pope et al.. 1989); accordingly. the porcine PDGF (R &

D S ystems. Inc.) used in these experiments contains only PDGF-BB homodimers.

Cell culttlrr and plnting

BM stromal cells from the femora of young 40 - 43-day-old. 1 10 - 120 gram, male

Wistar rats were cdtured essentially as previously described ((Maniatopoulos et al.. 1988).

see Chapter 2). After 7 d.. adherent cells were recovered in a single ce11 suspension, and

either counted electronically o n a Coulter Counter (Coulter Electronics, Hialeah, FL) and

plated at densities between 2 x 104 and 4 x 104 cells per 35 mm dish (Falcon) or stained and

processed through the flow cytornetric ce11 sorter (see Chapter 3) prior to plating. After

plating. cells were cultured in growth medium plus porcine PDGF (1 - 10 @ml) in pulses as

79 indicated in the Figures or continuously as indicated for an average of 21 d., then stained

(APIvon Kossa/toluidine blue, nNBE, and Sudan IV) as previously described in Chapter 2.

Quantitation

Quantitation of colony types was as prevously described in Chapter 3. In addition,

for analysis of multiple sets of data, the Wilcoxon signed rank test was used.

As we reported previously (Herbertson and Aubin. 1995), dex markedly affects both the

subpopulation make-up and osteogenic differentiation in RBM stromal ce11 cultures. While

a small number of bone nodules form in cultures without dex, dex increases significantly the

number of bone nodules and the number of macrophage colonies present, and decreases the

proportion of cells expressing leukocyte common antigen. In al1 experiments reported here,

PDGF was added to cultures also supplemented with dex.

Cunfin~ious expasure to PDGF

The continuous presence of PDGF (4 ng/ml) was mitogenic in RBM stromal ce11 cultures,

significantly increasing late log phase growth rate and saturation density (see Figure 4.1).

PDGF at the same concentration decreased the number of bone nodules, but concomitantly

increased macrophage and adipocyte colony number (see Table 4.1). suggesting that the

decrease in bone nodules is not due to generalized ce11 toxicity. PDGF dose response

analysis was then undertaken for bone nodule (Figure 4.2), macrophage (Figure 4.3), and

adipocyte (Figure 4.4) colony types. Multiple experiments were done over a three year

period in regular and defined serum and data from al1 experiments are summarized in Table

4.2. PDGF had biphasic effects on bone nodule formation: the number of bone nodules was

80 stimulated by very low doses of PDGF (I L ng/ml in regular or defined semm experiments)

and inhibited by higher doses (2 4 ng/ml); in multiple repeat experiments the intermediate

concentrations either had no affect over control (Figure 4.2) or showed a sharper transition

between stimulatory and inhibitory doses (e.g., Table 4.1). In three experiments in which

complete dose response curves were available. the ID50 for bone nodule formation was

approximately 3 ng/ml PDGF (range in al1 experiments was approximately 2 - 10 nghl) . In

contrast, the continuous presence of PDGF consistently stimulated aNBE positive

(previously confimed to be macrophage colonies by specific antibody (ED2) labelling

(Herbertson and Aubin. 1995)) colonies and adipocyte colonies at al1 concentrations tested

in al1 experiments (see Tables 4.1. 4.2). Total macrophage colony number appeared to be

maximally stimulated a PDGF concentrations of 2 - 2.5 ng/ml. however. the number of

large macrophage colonies continued to increase dose dependently (see Table 4.1 and Figure

4.3) and may have overgrown and obscured smaller colonies. Likewise, adipocyte colonies

appear to increase in a dose-dependent manner in al1 experiments (see Tables 4.1. 4.2 and

Figure 4.3): in three experiments in which a plateau of adipocyte stimulation was seen

(corresponding to the same experiments as for bone nodules above). the EDso was

approximately 1.5 ng/rnl.

Pulse exposrrre to PDGF

The development of bone nodules is characterized by a proliferative period followed by a

period of morphologically and molecularly defined osteoblast differentiation and matrix

deposition and mineraiization (Liu et al., 1994; Lian and Stein, 1995; Aubin and Liu, 1996).

To discriminate the developmental stages at which the biphasic effects of PDGF occurred.

we pulse-treated cultures at different stages of the nodule formation process. 2 d. pulses

with PDGF (4 nglml or 6 nglml) early during log phase growth (d. 1 - 3. d. 3 - 5)

significantly stimulated bone nodule numbers. PDGF pulses during late log phase growth

(d. 5 - 7 or d. 5 - 8) either increased (2/4 experiments) or did not alter (2/4 experiments)

8 1 bone colony formation. PDGF pulses later in the culture penod (d. 7 - 9 onwards), when

saturation density was reached and nodules were forming, tended to decrease or significantly

decreased (d. 13 - 15) nodule formation (see Figure 4.5).

The effects of PDGF on RBM cells fractionaied on the busis of AP expression

Flow cytometric soning of 7 d. primary RBM strornal ceIl cultures was performed on the

basis of AP expression with a mAb against AP (RBM 2 1 1.13). The resultant ~ ~ h i g ~ . APiow,

or control cells were plated to determine osteoprogenitor (giving rise to bone nodules),

macrophage. and adipocyte distribution and frequency . As descri bed in detai 1 elsew here

(Herbertson and Aubin, 1997). wi thin the ~ p h i g h fraction, the number of AP-posit ive

colonies and osteoprogenitoa were significantly enriched while adipocyte and macrophage

colonies were depleted compared to the unfractionated control. In contrast. within the

APiow fraction of cells, adipocyte and macrophage colonies were consistently enriched

(Figure 4.6). When these sorted populations were cultured with PDGF (4 ng/ml), results

similar to those with unfractionated control populations were obtained. Specifically, in two

separate experiments. the same treatment with PDGF (4 ng/ml. continuous exposure)

significantly decreased bone nodules while significantly increasing macrophage and

adipocyte colonies in the fractions (see Figure 4.6). In the A P ~ O W fraction of cells, 4

ng/ml PDGF significantly decreased bone nodules in one experiment (see Figure 4.6) (no

statistically significant difference was seen in the other). significantly increased adipocyte

colonies, and significantly increased large macrophage colonies (often to an extent that

colonies merged making counting of total colonies difficult) while slightly decreasing or not

significantly affecting total macrophage colony nurnbers (see Figure 4.6).

4 Table 4.1 RBM stroma1 cells plated at 2x 10 B5 mm dish were grown for 2 Id., with dex and

the PDGF concentrations indicated present continously from d. 1 - 21. PDGF had dose-

dependent biphasic effects on bone nodule formation. while over the same concentration

range both adipocyte and aNBE positive colonies increased. Results are means f one

standard deviation.

** t tests venus PDGF O ng/ml: 'p I .OS, *p 5 .O 1, p I -00 1.

1 Condition 1 Bone nodules 1 Fat colonies Total aNBE

PDGF O ng/ml

PDGF 1 ndml

PDGF 2.5 ng/rnl

PDGF 5 ng/mi

PDGF 10 nghl

25.7 (+ 10.2)

57.3 (k 16.4)#

42.5 (f 9.1)#

4.0 (k 7.4)#

0.3 (1- 0.5) *

1 .O (I 0.0)

9.0 (f 0.0)

59.7 (+ 13.1)"

63.7 (f 1 1.5) **

74.3 (ir4.0) **

Table 4 2 Effect of continuous exposure to PDGF on RBM stromd populations

Net changes (stimulation = up arrow. inhibition = down arrow) with continuous exposure to

PDGF at the doses indicated, compared to vehide controls in RBM stroma1 ce11 cultures

with dex. Changes indicated by the arrows were signifiant at minimally p 5 .O5 as

determined by t tests, or the trend seen is indicated.

Adi~ocvte colonies 3DGF n g / d

0.25 (not deterrnined)

Bone nodules

7. 111

* Wilcoxon signed rank test of al1 experiments at 4 ng/ml PDGF show that overail patterns

are significant, ie., bone nodule numbers decrease (p = .0020), aNBE positive (macrophage)

colony numbers increase (p = .0020), fat colony numbers increase (p = .O 156).

&BE + colonies

T 111

0.50

1

2 - 2.5

4*

T 1/3 (trend 7 2/3)

't 1/6 (trend T 416) (trend 1/6)

k 5/8 ? 1 /8 (trend 1' 2/8)

J6110 (trend & 4/10)

2/3 (trend f in)

?' 5/6 (trend f 1/6)

? 6/'7 (trend T In)

? 8/10 (trend f 2/10)

7' 111

f 414

7- 414

T 7n

Days cultured

Figure 4.1 Ce11 number (log scale) versus time cuitured. The continuous presence of 4

ng/rnI of PDGF (PDGF 4) significantly increased ceIl numbers present in RBM stroma1

cultures at days 4, 6, and 8 (late log phase growth) and the final saturation density when

cornpared to the vehicle control (PDGF O). t test: #p 5 -05, *p I .O 1.

O 0.25 O 5 1 2 4 6 10

PDGF (ng/mI)

p p p p p p p p p p p p p p - - - - - - - - - - - - - - - -

- - - - -

Figure 4.2 The number of bone nodules present in RBM stroma1 ce11 cultures grown for

2 Id. in the continuous presence of dex and PDGF concentratiocs as indicated. PDGF 0.25

nglml significantly increased the number of bone nodules. while PDGF concentrations > 2

ng/rnI dose-dependently decreased the number of bone nodules setn in dex treated cultures.

t test versus vehicle control (PDGF O): #p 5 -05, *p < .O 1 , **p S .O0 1.

....... Coloay size:

msma11( 10-25 cells) medium (25- 100 cells)

' ~large(>100celIs) """'

1 . Totai number .......

PDGF concentration (ng/ml)

Figure 4.3 The number and size of aNBE positive (macrophage) colonies present in RBM

stroma1 ce11 cultures grown for 21d. in the continuous presence of dex and PDGF

concentrations as indicated. The total colony number increases start with PDGF 0.25 ng/ml

and appear to plateau at PDGF concentrations 2 2 ng/rnl. however. large colony numbers

continue to increase. t test versus vehicle control (PDGF 0) for the corresponding size or for

total number: #p 1 -05, *p I .O 1. **p S -00 1 .

O 0.5 1 2 4

PDGF concentration (ng/ml)

Figure 4.4 The number of adipocyte colonies present in RBM stroma1 cell cultures grown

for 21d. in the continuous presence of dex and PDGF concentrations as indicated. PDGF

concentrations > 1 ng/ml dose-dependently increased the number of adipocyte colonies.

Since vehicle control (PDGF 0) had a rnean and standard deviation of 0, one column tests

versus a theoretical mean of O were used: #p 5.05, *p I .O 1. **p 2 .ûû 1.

O 1 - 3 3-5 5 - 7 7 - 9 9- 11 11-13 13-15

PDGF pulse (days exposed)

Figure 4.5 The number of bone nodules present in RBM stroma1 ce11 cultures grown for

21d. in the continuous presence of dex and PDGF 4 n g h l pulses as indicated. 2 d. PDGF

pulses at d. 1 - 3, d. 3 - 5, and d. 5 - 7 significantly increased the number of bone nodules,

while PDGF pulse at d. 13 - 15 showed a (srnaIl) decrease in the number of bone nodules. T

test versus control (O days exposed): #p 5.05, *p S .O 1, **p S -001.

controi controi + PDGF A$.'* A@& + PDGF

90 Figure 4.6 Legend

The number of bone nodules present in RBM stroma1 cell cultures grown for 7d. primary

culture, then sorted according to alkaline phosphatase (AP) expression (high expression =

~ ~ k g h , low or negative expression = AP~OW, unfractionated cells = control) and grown for

21d. in the continuous presence of dex and PDGF 4 @ml. Bone nodules are decreased

significantly by PDGF in both ~ p h i g h and AP~OW fractions; adipocyte colonies are

significantly increased in al1 populations; total aNBE (macrophage) colonies increased in

control and A P ~ ~ ~ ~ populations. While the total uNBE colony number decreased in the

AP~OW fraction in ihis experiment, the number of large colonies increased significantly,

likely obscuring srnaller colonies. t test versus same condition without PDGF: #p < -05, *p I

.Ol, **p 1.001.

Our data suggest that PDGF affects both proliferation and differentiation in RBM stromal

ce11 cultures. m i l e the main purpose of our study was to investigate an effect of PDGF on

osteoprogenitors and bone fomation, we also monitored other lineages which are present in

these cultures, regulated by some of the same factors and hormones (e-g., dex (Herbertson

and Aubin, 1995)), and whose presence may modulate osteoblast lineage ce11 expression

(Aubin, 1997, submitted). The mitogenic effects of PDGF on late log phase growth rate and

saturation density of RBM stromal cultures. together with its ability to stimulate the

numben of macrophage and adipocytic lineage colonies, suggest that PDGF stimulates

proliferation and differentiation in these lineages. In contrat. continuous exposure to PDGF

decreased AP-positive colony and bone nodule number. suggesting that PDGF inhibits

osteoprogenitor differentiation to mature bone-forming osteoblasts. The ability of pulse

treatment with PDGF in early to late log phase growth to increase bone nodule numbers

while pulse treatment late in the differentiation sequence inhibits bone nodule numbers

suggests, however, that PDGF is stimulatory for progenitors at an early stage of osteoblast

differentiation and is inhibitory at later stages. This is consistent with data on calvaria organ

cultures in which the cells proliferating in response to PDGF appeared to be in the periosteal

fibroblast and osteoprogenitor ce11 zones (Hock and Canalis, 1994), and with the suggestion

that PDGF may stimulate a fibrous tissue, rather than bone tissue, phenotype (Marden et al.,

1993). The biphasic nature of the effects of PDGF was also evident over different

concentration ranges. with low doses stimulating and high doses inhibiting bone nodule

fomation. These in vitro studies may shed light on the conflicting data reported from

PDGF treatments in vivo. Continuous exposure to PDGF may be considered analogous to

studies with local administration of PDGF-BB (e.g.. in rat craniotomy defects), which was

found to inhibit bone regeneration (Marden et al.. 1993). Since PDGF is rapidly cleared in

rats (90% within 1 hr. (Cohen et al.. 1990)). pulse exposures rnay mimic systemic

92 administration which has been shown to significantly increase osteoblast number (Mitlak et

al., 1996)). as with early PDGF pulses in our culture system.

Previous studies reported that 24 - 48 hr. pulses with PDGF-BI3 (approximately 10 - 100

ng/ml) decreased AP activity in fetal rat calvaria ce11 cultures (Centrella e t al., 1991;

Pfeilschifter et al.. 1992). Cassiede and colleagues found that a 48 hr. pulse of 5 ng/ml

PDGF-BB on preconfluent rat marrow stromal cells had no significant effect on

osteochondrogenic potential (as measured by bone nodule formation) and observed that a

decrease in AP activity directly after pulse treatment is not predictive of subsequent

osteochondrogenic potential on removal of the growth factor (Cassiede et al.. 1996). We

cannot account for why our data are different from Cassiede and colleagues, but since their

pulse window (d. 4 - 6) falls between our d. 3 - 5 pulses (which were consistently

stimulatory) and Our late log phase growth (d. 5 - 7 or 5 - 8) pulses (which showed mixed

increaselno difference result). it is possible that they treated at a particular developmental

tirne/differentiation stage which is insensitive to a PDGF effect. Taken together with Our

biphasic PDGF concentration effects and the exquisitely sensitive timing required to target

particular differentiation stages, the data argue for cornplex. multiphasic effects of PDGF on

this bone ceIl lineage.

While dex. concomitantly stimulated bone nodule formation and macrophage development in

RBM stromal ce11 cultures (Herbertson and Aubin. 1995), continuous exposure to PDGF

stimulated production of macrophage colonies but inhibited bone nodule formation. Both

macrophages (Yan et al.. 1993) and osteoblasts (Zhang et al.. 1991) are known to express

PDGF p-receptors. and thus both lineages should respond to porcine PDGF which is

primarily PDGF-BB. Macrophages produce cytokines and growth regulating factors (eg.,

PGEz, TNF, IL-6 (Horowitz and Gonzaieç, 1996); IGFBP-4 (Li et al.. 1996)) which may

affect osteoblastic cells. and inhibit stromal development (Goliaei et al., 1995). Since PDGF

93 strongly up-regulates IL- 1 mRNA levels in munne macrophages (Yan et al., 1993), and the

continuous presence of both IL-if3 and IL-la are known to inhibit bone formation (Eliies

and Aubin, 1990), it seemed possible to us that the inhibitory effect of PDGF on bone

nodules might be indirect and mediated through macrophages and IL- 1 production. IL- 1 has

been shown to have biphasic effects on bone formation in RBM stroma1 ce11 cultures, and

the dose response curve varies according to the endogenous IL4 levels (Shadmand and

Aubin, 1995), similar to the biphasic response we measured here with PDGF and the small

shifts in dose responsiveness seen between independent RBM isolations. On the other hand,

the fact that osteoblasts express PDGF-P receptors suggests that PDGF could have direct

effecü on osteoblast iineage cells. To test this hypothesis, we sorted RBM stroma1 cells on

the bais of expression of AP. We showed previously that this resulted in populations 93 -

97% pure in the M h i g h fraction, a fraction significantly enriched for osteoprogenitors and

depleted in adipocytes and macrophages, while the AP[OW fraction was significantly

enriched in macrophages and fat cells (Figure 4.6, and Herbertson and Aubin, 1997). In two

separate experiments, the effects of PDGF on osteoprogenitors (giving rise to bone nodules)

in the M h i g h fractions appeared quantitatively sirnilar to what we observed in the

unfractionated control. suggesting a direct effect of PDGF on the osteoblast lineage. Of

course, we cannot nile out that the fewer remaining monocyte-macrophage lineage cells in

the ~ ~ h i g h fraction are sufficient to effect a cytokine cascade rnediating the PDGF response,

but this seems unlikely given the similar fold-changes in bone in response to PDGF in

populations with markedly different initial numbers and fold-changes in monocyte-

macrophage cells. The reciprocal question of whether the effects of PDGF on macrophages

are a11 direct is also of interest since osteoblastic cells are known to produce some of the

factors thought to be involved in hematopoiesis including M-CSF (Shiina-Ishimi et al.,

1986; Elford et al., 1987: Benayahu et al., 1992; Greenfield et al., 1993)). Again, there

appears to be no direct relationship between the number of bone nodules and the number of

macrophage colonies presen t in unfractionated or fractionated populations, suggesting that

94 the stimulatory effects of PDGF on macrophages are direct. In this regard, other authors

have reported that PDGF-BB has a stimulatory effect on primitive pluripotential

hematopoietic cells (preCFCmulti), mediated by up-regulation of U-1 in marrow

macrophages. and showed conclusively the presence of PDGF $-receptors on normal

marrow macrophages (Yan et al., 1993).

Exposure to PDGF-BB has been reported to inhibit expression of differentiated phenotypes

in certain other mesenchymal ceil types, e.g., smooth muscle (Holycross et al., 1992). While

chronic exposure to high concentrations inhibits osteogenesis, it is clear that low doses and

puisatile exposure of early progenitors leads to stimulation of differentiation to mature

matrix-synthesizing osteoblasts. In addition, we consistently found a stimulation of

adipocyte colony numbers between 1 - 10 ng/ml PDGF. which appears to be a novel

observation. Indeed, PDGF has been reported to inhibit adipocyte differentiation in human

female adipose tissue cultures, although the concentrations required to see inhibition were 2

10 ng/ml (Hauner et al., 1995). concentrations higher than we used here. It is not clear why

Hauner et al. did not see a stimulatory effect although differences in the cultures and culture

conditions used (including species differences - human versus rat: serum-free venus serum

containing medium, and differences in anatomic sites (rneduflary versus non-medullary

adipocytes)) are al1 possibilities. It is also interesting that PDGF is stimulatory to both

adipocyte precursors and osteoblast precursors under at least some conditions, e.g., very low

dose treatment, but that the responsiveness of the lineages diverges at higher concentrations

such that PDGF stimulates adipogenesis white concomitantly inhibiting osteogenesis in the

same cultures. Since we found that PDGF is mitogenic at early stages of culture during log

phase growth, it is possible that progenitors for both lineages, including, e.g.. a putative

bipotential adipocyte-osteoblast progenitor (reviewed in (Gimble et al., 1996)), are

comparably stimulated to undergo division, but that subsequent differentiation steps are

modulated quite differently by this growth factor. In this regard, the differences in

95 sensitivity to PDGF (EDso for adipocyte colony stimulation, 1.5 n g / d PDGF; ID50 for bone

nodule inhibition, 3 ng/ml PDGF, in the sarne experiments) of these lineages are notable and

together with the marked differences in absolute nurnber of adipocyte and osteoblast

colonies, suggest that the stimulation in adipocyte formation is not at the expense of

osteoblast formation. as suggested with some hormone treatment protocols, including

1,25(OH)2D:, treatment (reviewed in (Gimble et al., 1996; Aubin and Heersche, 1997)).

Our data suggest that the effects of PDGF on the differentiation of osteoprogenitors to

mature osteoblasts in RBM stromal populations are biphasic, with both stimulation and

inhibition of differentiation and bone nodule formation being observed, probably through

the growth factor targeting different events early and late in the differentiation sequence.

These complex effects, which occur concomitantly with modulation of other Iineages

present in the stroma1 environment, suggest that PDGF has pleiotropic effects on bone

metabolism.

CHAPTER 5

Conclusions and Future Experirnents

97 5.1 Conclusions and Future Experiments

The studies presented in this diesis were aimed at furthering knowledge on whether cell-ce11

interactions affect osteogenesis in the heterogeneous cellular environment of RBM ce11

cultures. To this end, RBM stromal ceils were grown under conditions stimulating bone

nodule formation, then histochemical and immunohistochemical staining, and flow

cytometric analysis and sorting were used for description and analyses. Finally, the effect of

PDGF was examined in this system using these techniques.

In Chapter 2, 1 began to elucidate and quantify some of the subpopulations present when

RBM stromal cells are grown with or without dex under conditions favoring bone formation.

Culture dishes were analyzed for ce11 counts, or stained with either histochemical or

immunohistochemical stains, and colony types were quantitated. or cells were processed for

flow cytometry. Dex significantly increased the number of AP positive colonies and von

Kossa positive bone nodules, aNBE positive colonies, and ED2 positive (macrophage)

colonies. The number of adipocyte foci was largely unaffected in these experiments. Flow

cytornetry confirmed colony counts and showed stimulation by dex of AP positive cells and

macrophages, and in addition, the reduction of hemopoietic cells expressing leukocyte

common antigen. These data show that when RBM stromal populations are grown under

conditions stimulating osteoprogenitor differentiation and bone formation, the stroma1

subpopulation rnake-up, including expression of hemopoietic lineages, is markedly altered.

Specifically, dex stimulated osteoprogenitor differentiation concomitant with promoting the

growth and differentiation of macrophages. These data also showed for the first tiine the

presence of macrophages in RBM stroma1 cultures grown under conditions favouring bone

formation, and extended the information on the subpopulation make-up and heterogeneity

under these culture conditions.

98 Since the presence of multiple ceIl types in RBM stromai populations rnay modulate or even

mask, and complicate interpretation of, cytokine and hormone effects on the

osteoprogenitors present. 1 looked for a method for purification of the osteoprogenitor

population. In Chapter 3, 1 described experiments in which flow cytometric sorting of 7 d.

primary RBM stromal ce11 cultures was performed on the basis of AP expression with an

antibody against AP (RBM 21 1.13 (Turksen and Aubin, 1991)). The resuitant ~ ~ h i p h .

AP'OW, or control cells were plated to determine osteoprogenitor. macrophage, and adipocyte

distribution and frequency. Approximately 50% of osteoprogenitor cells (capable of giving

rise to bone nodules) were lost during processing/sorting when compared to unsorted

controls. Nevertheless. within the kiPhigh fraction. the number of AP positive colonies and

bone nodules were signi ficantly enriched compared to the unfractionated control; the

increase in osteoprogenitor frequency ranged from approximately 2 to 100 fold. There were

few assayable osteoprogenitors in either the ~ ~ h i g h or APlow fractions in the absence of dex,

suggesting that RBM stroma contains largely dex-dependent osteoprogenitor populations,

and that dex rnay regulate osteoprogenitors subsequent to the upregulation of AP. Bone

nodule numbers in either the or A P I ~ W fraction did not follow a linear relationship

w i th decreasing plating densi ty . Adipocyte and macrophage colonies were depleted in the

M b g h fraction of cells, but were enriched in the AP~OW fraction. I thus showed that flow

cytometric sorting of RBM stromal populations according to high or low Ai? expression is

an effective technique for enrichment of AP positive colonies and osteoprogenitors capable

of dividing and differentiating into bone nodule-forming cells, and should prove useful for

further studies of the nature of the osteoprogenitors and their regulation in this tissue.

In Chapter 4. 1 combined the techniques of the previous two chapters to investigate the

effects of PDGF in RBM stroma1 ceil populations, again grown under conditions promoting

growth and differentiation of osteoprogenitors to mature bone-forming osteoblasts. PDGF

was mitogenic during late log phase growth and increased the saturation density of RBM

99 stromal cultures, as well as dose-dependently stimulating the numbers of macrophage and

adipocytic lineage colonies, suggesting that PDGF stimulated proliferation and

differentiation in these lineages. The continuous presence of PDGF over the same

concentration range had biphasic effects on bone nodule formation, increasing bone nodules

at low doses (6 1 ng/ml) and dose-dependently decreasing formation at 2 4 ngfml. When

present as a short-duration pulse in early to Iate log phase only, PDGF increased bone

nodule numbers at al1 doses tested, but inhibited bone nodule formation late in the

differentiation sequence, suggesting that PDGF is mitogenic during an early stage of

osteoblast differentiation but inhibitory later. Comparison of controVmixed RBM stroma1

cultures with populations fractionated on the bais of AP expression suggested that the

inhibitory effects of PDGF may be elicited by a direct action on osteoblast lineage cells.

These complex effects on bone. which occur concomitantly with modulation of other

lineages (macrophages and adipocytes) present in the stromal environment, suggest that

PDGF has pleiotropic effects on bone metabolism.

The histochemical and immunohistochemical techniques described. dong with the ability to

fractionate RBM stroma1 populations on the basis of AP expression, have facilitated the

characterization of the RBM stromal system, and will be of use when the effects of other

cytokines or growth factors are analyzed in this system. We are currently limited in the rat

system by a lack of other specific antibodies to early osteoprogenitors as recently described

for human osteoblastic cells (Sirnmons and Torok-Storb, 199 1 ; Haynesworth et al., 1992;

Gronthos et al.. 1994; Joyner et al., 1997), especially given that the specific macromolecules

recognized by these antibodies are only beginning to be identified (see, e.g.. (Bruder et al.,

1997)) to allow preparation of rat-specific reagents. As other specific reagents are identified

by antibody or molecular approaches, future sorting experiments can be designed to achieve

even better enrichment for cells earlier in the osteoblast lineage. While the 93 - 97% purity

obtained in Our sorting experiments is comparable to other sorting and irnmunopanning

1 0 0 techniques, it has left open some arnbiguity and necessitated some caution in interpretations

of the data as discussed in Chapter 3 and Chapter 4. Nevertheless, the approaches have been

established and they lead to reasonable ce11 yields, so that future experiments can al1 be

airned at improving the purity of the osteoblast lineage rather than development of

techniques. Methods for improving purity could include double sorting with one specific

antibody. use of both positive and negative immunoselection in sorting, andlor a

combination of sorting with irnmunopanning techniques. Likewise, as antibodies become

available recognizing other rat early hemopoietic cells. it will become possible to identify

more precisely and manipulate the leukocyte antigen positive cells, whose numben are also

markedly altered by dex in osteogenic RBM cultures, to determine more clearly their

roIe/interactions with osteoprogenitors in this system. The ability to identify and then

fractionate the subpopulations present in RBM stroma1 ce11 populations represents an

important step towards understanding the cell-ce11 interactions which affect bone formation

in BM.

REFERENCES

Abdennagy, B., M. Hott, et al. (1992). "Effects of PDGF on human and mouse osteoblastic

cells isolated from the trabecular bone surface." Ce11 Bi01 Int Reo 16(3): 235-47.

Allam. M., N. Martinet, et al. (1992). "Differential migratory response of U-2 OS

osteosarcorna ce11 to the various forms of PDGF." Biochimie 74(2): 183- 186.

Andrew, I. G., J. A. Hoyland, et al. (1995). "PDGF expression in nonnally heaiing human

fractures." Bone 16(4): 455-60.

Antosz, M. E., C. G. Bellows, et al. (1987). "Biphasic effects of epidemal growth factor on

bone nodule formation by isolated rat calvaria cells in vitro." J Bone Miner Res 2(5): 385-

393.

Ashton. B. A., C. C. Eaglesom, et al. (1984). "Distribution of fibroblastic colony-forming

cells in rabbit bone rnarrow and assay of their osteogenic potential by an in vivo diffusion

chamber method." Calcif Tissue Int 36: 83-86.

Aubin, J. E. (1979). "Autofluorescence of viable cultured mammalian cells." J Histochem

Cvtochem 27( 1 ): 36-43.

Aubin, J. E., C. Bellows, et al. (1992). Analysis of the osteoblast lineage and regdation of

differentiation. Chemistrv and Bioloev of Mineralized Tissues. H. Slavkin and P. Price.

Amsterdam, Elsevier Science Publishers.

103 Aubin, J. E., S.-W. Fung, et al. (1990). 'The influence of non-osteogenic hemopoietic cells

on bone formation by bone marrow stromal populations." J Bone Miner Res S(SUPPL. 2):

S8 1 (Abstract 3 1).

Aubin, J. E., A. K. Gupta, et al. (1996)- "Expression and regulation of galectin 3 in rat

osteoblastic cells." J CeII Phvsiol 169: 468-80.

Aubin, J. E- and J. M. N. Heersche (1997). Vitamin D and osteoblasts. Vitamin D. D.

Feldman, F. H. Glorieux and J. W. Pike. San Diego, Acadernic Press.

Aubin, 1. E. and F. Liu (1996). The osteoblast lineage. Principles of Bone Bioloev. L. G.

Raisz, G. A. Rodan and J. P. Gilezikian. San Diego, Acadernic Press: 5 1-67.

Aubin, J. E. and K. Turksen (1995). "Monoclona1 antibodies as tools for studying the

osteoblast lineage." Microscopv Research and Technique 33: 128- 140.

Aubin, I. E., K. Turksen, et al. (1993). Osteoblastic ceIl Iineage. Cellular and Molecular

Biolom of Bone. M. Noda. New York, Academic Press, Inc.: 1-45.

8abT I., D. Gazit. et al. (1992). "Histone H4-related osteogenic growth peptide (OPG): a

novel circulating stimuiator of osteoblastic activity." EMBO J 11(5): 1867-1 873.

Bab. 1.. D. Gazit. et al. (1988). "Regenerating bone marrow produces a potent growth-

promoting activity to osteogenic cells." Endocrinoioay 123(1): 345-352.

Bellows, C. G. and J. E. Aubin ( 1989). "Determination of numbers of osteoprogenitors

present in isolated fetal rat calvaria cells in vitro." Dev Bi01 133: 8- 13.

Bellows, C. G-, 1. E. Aubin, et ai. (1986). "Mineralized bone nodules formed in vitro from

enzymatically released rat ciilvaria ceil populations." Calcif Tissue Int 38: 143- 154.

Bellows. C. G., J. N. M. Heersche, et al- (1990). "Determination of the capacity for

proliferation and differentiation of osteoprogenitor cells in the presence and absence of

dexamethasone." Dev Bi01 140: 132- 138.

Bellows, C. G., Y .-H. Wang, et al. ( 1994). " 1.25 Dihydroxyvitarnin D3 stimulates adipocyte

differentiation in cultures of fetal rat calvarial cells: Cornparison with the effects of

dexamethasone." Endocrinology 134(5): 222 1-9.

Benayahu, D., A. Fried, et al. (1991). "Subpopulations of rnarrow stromal cells share a

variety of osteoblastic markers." Calcif Tissue Int 49: 202-207.

Benayahu, D., M. Horowitz, et al. (1992). "Hemopoietic functions of marrow-derived

osteogenic cells." Calcif Tissue Int 51: 195-20 1.

Benayahu, D., Y. Kletter, et al. (1989). "Bone marrow-derived stromal ce11 line expressing

osteoblastic phenotype in vitro and osteogenic capacity in vivo." J Ce11 Phvsiol 140: 1-7.

Bennett, J. H., C. J. loyner, et al. (1991). "Adipocyte cells cultured from marrow have

osteogenic potential." J Cell Sci 99: 13 1- 139.

Beresford, J. N., J. H. Bennett, et al. (1992). "Evidence for an inverse relationship between

the differentiation of adipocytic and osteogenic cells in RBM stromal ceIl cultures." J Ce11

Sci 102: 34 1-35 1. -

Bianco, P.. M. Costantini, et al. ( 1988). "Alkaline phophatase positive precursors of

adipocytes in the human bone marrow." Br J Haematol68: 40 1-3.

Bilbe, G.. E. Roberts, et al. ( 1996). "PCR phenotyping of cytokines. growth factors, and

their receptors and bone matrix proteins in human osteoblast-Iike ce11 lines." Bone 19(5):

43745.

Birch, M. A., A. F. Ginty. et al. (1993). 9 C R detection of cytokines in normal human and

pagetic osteoblast-like cells." J Bone Miner Res 8(10): 1 155- 1 162.

Bonanno, E., L. Ercoli, et al. (1994). "Homogeneous stroma1 ce11 population from normal

human adult bone marrow expressing a-smooth muscle actin filaments." Lab Invest 71(2):

308- 15.

Bowen-Pope, D. F., C. E. Hart. et al. (1989). "Sera and conditioned media contain different

isoforms of PDGF which binds to different classes of PDGF receptor." J Biol Chem 264:

2502-8.

Bowen-Pope, D. F.. T. W. Malpass. et al. (1984). "Platelet-derived growth factor in vivo:

Levels, activity, and rate of clearance." Blood 64(2): 458-69.

Bowen-Pope, D. F.. A. Van Koppen, et al. (1991). "1s PDGF really important? Testing the

hypothesis." TIG 7(11/12): 4 13-4 i 8.

Briddell, R. A.. J. E. Brandt, et al. (1992). "Role of cytokines in substaining long-term

human megakaryocytopoiesis in vitro." Blood 79(2): 332-337.

Bruder. S. P., N. S. Ricalton. et al. (1997). bbPurification, molecular cloning. and the

functional role of a human osteoprogenitor ce11 surface antigen." J Bone Miner Res 12(Suppl

1): S 141-

Canalis, E. (1983). "Effect of glucocorticoids on type 1 collagen synthesis, alkaline

phosphatase activity. and deoxyribonucleic acid content in cultured rat calvaria."

Endocnnology 112: 93 1-939.

Canalis, E. and J. B. Lian (1988). "Effects of bone associated growth factors on DNA,

collagen, and osteocaicin synthesis in cultured fetal rat calvaria." Bone 9: 243-246.

Canalis. E. and S. Rydziel (1996). Platelet-derived growth factor and the skeleton. Princioles

of bone biologv. J. P. Bilezikian, LI G. Raisz and G. A, Rodan. Toronto, Academic Press:

6 19-626.

Cardoso, A. A., S. M. Watt. et al. (1995). "An improved panning technique for the selection

of CD 34+ human bone marrow hematopoietic cells with high recovery of early

progenitors." EXD Hematol 23: 407- 12.

Cassiede, P.. I. E. Dennis. et al. (1996). "Osteochondrogenic potential of rnarrow

mesenchymal progenitor cells exposed to TGF-B 1 or PDGF-BB as assayed in vivo and in

vitro." .J Bone Miner Res 1 l(9): 1264-73.

Cattoretti, G.. R. Schiro. et al. ( 1 99 1). "Bone marrow stroma in humans: Anti-nerve growth

factor receptor antibodies selectively stain reticular cells in vivo and in vitro." Blood 81(7):

1726-38.

Centrella. M., T. L. McCarthy, et al. (1989). "Platelet-derived growth factor enhances

deoxyribonucleic acid and collagen synthesis in osteoblast-enriched cultures from fetal rat

parietal bone." Endocrinology 125( 1): 13- 19.

Centrella, M., T. L. McCarthy, et al. (1991). "Relative binding and biochemical effects of

heterodimeric and homodimenc isoforms of platelet-derived growth factor in osteoblast-

enriched cuItures from fetai rat bone." J Cell Ph~siol 147: 420-426.

Centrella, M., T. L. McCarthy, et al. (1992). "Isoform-specific regulation of PDGF activity

and binding in osteoblast-enriched cultures from fetal rat bone." J Clin Invest 89: 1076-84.

Chaudhary, L. R. and L. V. Avioli (1994). "Dexamethasone regulates IL4 beta and TNF-

alpha-induced IL-8 production in human bone rnarrow stroma1 and osteoblast-like cells."

Calcif Tissue Int 55( 1): 16-20.

Cheng, S.-L., J. W. Yang, et al. (1994). "Differentiation of human bone marrow osteogenic

st romal cells in vitro: Induction of the osteoblast phenotype by dexamethasone."

Endocrinology 134( 1): 277-86.

Clark, G. ( 198 1). Staining Procedures. Baltimore, Williams and Wilkins.

Coats, S. R., J. E. Olson, et al. (1992). "Rapid induction of cornpetence formation is PDGF-

isoform specific." J CeIl Biot 48: 242-247.

Cohen. A. M., C. Soderberg, et al. (1990). "Plasma clearance and tissue distribution of

recombinant human PDGF (B-chain homodimer) in rats." .J Sure Res 49: 447452.

Davidai, G., A. Lee. et al. (1992). "PDGF induces tyrosine phosphorylation in osteoblast-

like ceils: relevance to mitogenesis." Am J Phvsiol 263(2): E205-9.

Deryugina E. 1. and C. E. Muller-Sieburg (1993). "Stroma1 cells in long-term cultures: Keys

to the eiucidation of hematopoietic development?" Cri t Rev Immun01 13(2): 1 15- 150.

Deryugina, E. I., B. I. Ratnikov, et al. ( 1995). "Identification of a growth factor for primary

murine stroma as M-CSF." Blood 86(7): 2568-78.

Dexter, T. M. (1982). ''Stroma1 ce11 associated haemopoiesis." J Ce11 P h ~ s i o l SUDDI 1: 87-94.

Dijkstra, C. D.. E. A. Dopp, et al. (1985). 'The heterogeneity of mononuclear phagocytes in

lymphoid organs: distinct macrophage subpopulations in the rat recognized by monoclonal

antibodies ED 1, ED2. ED3." I m m u n o l ~ 54: 589-599.

Dodds. R. A.. K. Merry, et al. (1994). "Expression of mRNA for ILI P. IL6. and TGFP1 in

developing human bone and cartilage." J Histochern Cvtochem 42(6): 733-44.

Dorheim. M.-A.. M. Sullivan, et al. (1993). "Osteoblastic gene expression during

adipogenesis in hematopoietic supporting murine bone marrow stroma1 cells." J Ce11 Phsiol

154: 3 17-328.

Drury, R. A. B. and E. A. Wallington (1967). Carleton's Histolo~ical Technique. Toronto.

Oxford University Press.

IO9 Duijvestijn, A. M., H. Van Goor, et al. (1992). "Antibodies defining rat endothelial cells:

RECA- 1, A pan-endothelial cell-specific monoclonal antibody." Lab Invest 66(4): 459-466.

Elford, P. R., R. Felix, et al. (1987). "Murine osteoblastlike cells and the osteogenic cell

MC3T3-El release a macrophage colony-stimulating activity in culture." Calcif Tissue Int

41: 15 1-6.

Ellies, L. G. and J. E. Aubin ( 1990). "Temporal sequence of interleukin 1 a-mediated

stimulation and inhibition of bone formation by isolated fetal rat caIvarial cells in vitro."

Cvtokine 2(6): 430-437.

Falla, N., P. Van Vlassalaer, et al. (1993). "Characterization of a 5-Fluorouracil-enrïched

osteoprogenitor population of the murine bone marrow." Blood 82(12): 3580-359 1.

Felix, R., M. G. Cecchini, et al. ( 199 1). "Production of GM-CSF and G-CSF by rat clonai

osteoblastic cell population CRP 10/30 and the immortalized cell line IRC10/30-mycl

stimulated by TNFa." Endocrinolo~v 128(2): 66 1-7.

Felix, R., P. R. Elford, et al. (1988). "Production of hemopoietic growth factors by bone

tissue and bone cells in culture." J Bone Miner Res 3( 1): 27-36.

Felix, R., H. Fleisch, et al. (1989). "Bone-resorbing cytokines enhance the release of

macrophage colony-stimulating activity by the osteoblastic ce11 MC3T3-E 1 ." Calcif Tissue

Int 44: 356-60. -

110 Friedenstein, A. J. (1990). Osteogenic stem cells in the bone marrow. Bone and Minera1

Research/7. J. N. M, Heersche and J. A. Kanis, Elsevier Science Publishers B. V.

(Biomedicd Division) : 243-270.

Friedenstein. A. J., N. V. Latzinik. et al. (1992). "Bone marrow stroma1 colony formation

requires stimulation by haemopoietic cells." Bone Miner 18(3): 199-2 13.

Friedenstein, A. J., K. V. Petrakova, et al. (1968). "Heterotopic transplants of bone marrow.

Analysis of precursor cells for osteogenic and hematopoietic tissues." Transolantation 6(2):

230-47.

Galmiche, M. C., V. E. Koteliansky. et al. (1993). "Stroma1 cells from human long-term

marrow cultures are mesenchymal cells chat differentiate following a vascular smooth

muscle differentiation pathway." Blood 82( 1 ): 66-76.

Gazit, D.. M. Karmish. et al. (1990). "Regenerating marrow induces systemic increase in

osteo- and chondrogenesis." Endocrinoloqy 126: 2607-26 1 3.

Giesler, R. K. H., R. Rober, et al. (1991). "Dendritic accessory ceIls derived from RBM

precursors under chemically defined conditions in vitro belong to the myeloid lineage."

J Cell Bi01 54: 171-181.

Gilardetti, R. S., M. S. Chaibi. et al. (1991). "High-affinity binding of PDGF-AA and

PDGF-BB to normal human osteoblastic cells and modulation by interleukin-1." Am J

Ph~sioI 261: C980-C985.

111 Gimble, J. M., J. Hudson, et al. (1991). "Regulation of interleukin-6 expression in munne

bone marrow stroma1 cells." Exp Hematol 19: 1055- 1060.

Gimble, J. M., C. E. Robinson, et al. (1996). "The function of adipocytes in the bone

marrow stroma: An update." Bone 19(5): 421-8.

Goliaei, B.. 2. Soheili, et al. (1995). "Hematopoiesis in the presence of macrophages in

long-term bone marrow cultures." Exp Hematol 23: 1 115-20.

Graves, D. T., A. Valentin-Opran, et al. (1989). "The potential role of platelet-derived

growth factor as an autocrine or paracrine factor for human bone cells." Connect Tissue Res

23: 209-2 1 8.

Greenfield. E. M., S. A. Gomik, et al. (1993). "Regulation of cytokine expression in

osteoblasts by parathyroid hormone: Rapid stimulation of interleukin-6 and Ieukemia

inhibitory factor mRNA." J Bone Miner Res S(10): 1 163- 1 17 1.

Gronthos, S., S. E. Graves, et al. (1994). "The STRO- I + fraction of adult human bone

marrow contains the osteogenic precursors." B lood 84( 12): 4 164-73.

Grotendorst, G. R., A. Igarashi, et al. ( 199 1 ). "Differential binding, biological, and

biochernical actions of recombinant PDGF AA, AB, and BB molecules on connective tissue

celIs." J Ce11 Phvsiol 149: 235-243.

Hart, C. E. (1990). "PDGF recptor: Current views of the two sub-unit model." J Invest

Dermatol94: 53s-57s.

112 Hasthorpe, S., J. Rogerson, et al. (1990). "Use of bone marrow somatic ce11 hybrid lines to

generate rare monoclonal antibodies specifically reactive with rare marrow cells." EXD

Hematol 18: 223-227.

Hauner. H., K. Rohrig, et ai. ( 1995). "Effects of epidermal growth factor (EGF), platelet-

derived growth factor (PDGF and fibroblast growth factor (FGF) on human adipocyte

development and function." Eur J Clin Invest 25: 90-6.

Hauschka, P. V., E. M. Athanasios, et al. (1986). "Growth factors in bone matrix." J Biol

Chem 261(27): 12665- 12674.

Hauser, S. P., 1. A. Waldron, et al. (1995). "Morphological characterization of stromai ceIl

types in hematopoieitically active long-term murine bone marrow cultures." J Histochem

Cvtochem 43(4): 37 1-9.

Haynesworth, S. E., M. A. Baber, et al. (1992). "Cell surface antigens on hurnan marrow-

derived mesenchymal cells are detected by monoclonal antibodies." Bone 13(1): 69-80.

Haynesworth, S. E., J. Goshirna, et al. (1992). "Characterization of cells with osteogenic

potential from hurnan rnarrow." Bone 13(1): 8 1-88.

Herbertson. A. and J. Aubin (1995). "Dexarnethasone alters the subpopulation make-up of

R8M strornal ceII cultures." I Bone Miner Res lO(2): 285-94.

Herbertson. A. and I. E. Aubin (1997). T e l l sorting enriches osteogenic populations in

RBM strornal ce11 cultures." Bone in press.

113 Hirata, J., S. Kaneko, et ai. (1985). "Effect of platelet-derived growth factor and bone

marrow-conditioned medium on the proli feration of human bone marrow-derived

fibroblastoid colony forrning cells." Acîa Haematol74: 189- 194.

Hock, J. M. and E. Canalis (1994). "PDGF enhances bone ce11 replication. but not

differentiated function of osteoblasts." Endocrinoloqy 134(3): 1423-8.

Holycross, B. J., R. S. Blank, et al. (1992). "PDGF-BB-induced suppression of smooth

muscle ce11 differentiation." Circ Res 71: 1525- 1532.

Horowitz, M. C., D. L. Coleman, et al. (1989). "PTH and LPS induce munne osteoblast-like

cells to secrete a cytokine indistinguishable from GM-CSF." I Clin Invest 83: 149-57.

Horowitz, M. C., D. L. Coleman. et al. (1989). "Osteotropic agents induce the differential

secretion of GM-CSF by the osteoblast ce11 line MC3T3-E 1." I Bone Miner Res 4(6): 9 1 1-9.

Horowitz, M. C. and R. L. Jilka (1992). Colony stirnulating factors and bone remodeling.

Cvtokines and Bone Metabolism. M. Gowen. Ann Arbor, CRC Press : 186-2 12.

Horowitz, S. M. and J. B. Gonzales ( 1 996). "Inflammatory response to implant particdates

in a macrophage/osteoblast coculture model." Calcif Tissue Int 59: 392-6.

Hughes, F. J. and C. A. G. McCulloch (1991). "Stimulation of the differentiation of

osteogenic RBM stroma1 cells by osteoblast cultures." Lab Invest 64(5): 6 17-622.

Inaba, K., M. Inaba, et al. (1992). "Generation of large numben of dendritic cells from

mouse bone marrow cultures supplemented with GM-CSF." J Exp Med 176: 1693- 1702.

Inoue, K., B. Wongsasant, et al. (199 1). "Human c-fgr induces a monocyte-specific enzyme

in NM 3T3 cells." MOI Ce11 Bi01 ll(I2): 6279-6284.

Ishimi, Y.. C. Miyaura. et ai. (1990). "IL6 is produced by osteoblasts and induces bone

resorption." J Immunol 145( 10): 3297-303.

Joyner, C. J., A. Bennett, et al. (1997). "Identification and enrichment of human

osteoprogenitor cells by using differentiation stage-specific monoclonal antibodies." Bone

21(1): 1-6.

Kamalia, N., C. A. G. McCulloch, et al. (1992). "Dexamethasone recruitment of self-

renewing osteoprogenitor cells in chick bone marrow stroma1 ce11 cultures." B lood 79(2):

320-326.

Kanemoto, T., K. Tohya, et al. (1991). "Effects of culture matrix on differentiation of

munne mast celis." EXP Hematol 19: 288-293.

Kasugai, S., R. Todescan, et al. (1991). "Expression of bone matrix proteins associated with

mineralized tissue formation by adult RBM cells in vitro: Inductive effects of

dexamethasone on the osteoblastic phenotype." I Cell Phvsiol 147: 1 1 1 - 120.

Keller. R., R. Keist, et al. (1993). "Mononuclear phagocytes from human bone marrow

progenitor cells; morphology, surface phenotype, and functional properties of resting and

activated cells." Clin Exp Immunol 91: 176- 182.

115 Kiefer. F., F. W. Erwin. et al. (1991). "Fractionation of mouse bone marrow by adherence

separates primitive hematpoietic stem cells from in vitro colony-forming cells and spleen

colony-forming cells." Blood 78( 10): 2577-2582.

Kittler, E. L. W.. H. McGrath, et al. (1992). "Biological siginificance of constitutive and

sublimai growth factor production by bone marrow stroma." Blood 79(12): 3 168-3 178.

Kreua, M.. R. Andreesen, et al. ( 1993). " 1.25-Dihydroxyvitamin D3 production and vitamin

D3 receptor expression are developmentally regulated during differentiation of human

monocytes into macrophages." BIood 82(4): 1300- 1 307.

Lacey, D. L., J. M. Erdmann, et al. (1994). "Interleukin 4 enhances osteoblast M-CSF, but

not interleukin 6, production." Calcif Tissue Int 55( 1): 2 1-8.

Leboy, P. S.. I. N. Beresford. et al. (1991). "Dexamethasone induction of osteoblast mRNAs

in rat marrow stroma1 ceIl cultures." J Cell Phvsiol 146: 370-378.

Lefkovits, 1. and H. Waldmann (1979). Limitinc dilution anal~sis of celis in the immune

svstem. New York, Cambridge University Press.

Li. Y. M., S. Arkins. et al. ( 1996). "Macrophages synthesize and secrete a 25kD protein that

binds insulin-like growth factor-1." J Immun01 156: 64-72.

Lian, J. B. and G. S. Stein (1995). "Development of the osteoblast phenotype: Molecular

mechanisms rnediating osteoblast growth and differentiation." Iowa Ortho~ J 15: 1 1 8-40-

116 Liu, F. and J. E. Aubin (1996). "Molecular fingerprinting of primitive osteoprogenitor cells

undergoing di fferentiation in vitro." submitted.

Liu, F., L. Malaval, et al. (1994). "Simultaneous detection of multiple bone-related rnRNAs

and protein expression during osteoblast differentiation: Polymense chain reaction and

immunocytochemical studies at the single ceIl level." Dev Bi01 166: 220-34.

Long, M. W. ( 1992). "Blood cell cytoadhesion molecules." Exp HematolZO(3): 288-30 1.

Long, M. W., J. A. Robinson, et al. (1995). "Regulation of human bone marrow-derived

osteoprogenitor cells by osteogenic growth factors." J Clin Invest 95(2): 88 1-7.

Long, M. W.. J. L. Williams, et al. (1990). "Expression of human bone-related proteins in

the hematopoietic microevironment." I Clin Invest 86: 1387-95.

Malaval. L.. A. Gupta, et al. (1997). "LIF, but not IL-6, reguiates osteoprogenitor

differentiation in rat calvarial ce11 cultures - modulation by dexamethasone." J Bone Miner

Res in press. -

Malaval, L., D. Modrowski, et al. (1994). "Cellular expression of bone-related proteins

during in vitro osteogenesis in RBM stromal ce11 cultures." J Cell Phvsiol 158: 555-72.

Malaval, L., D. Modrowsky, et al. ( 1992). "Immunodetection of bene related proteins during

in vitro osteogenesis in RBM cultures." J Bone Miner Res 7 SI: S 220.

Maniatopoulos. C., J. Sodek. et al. (1988). "Bone formation in vitro by stromal cells

obtained from bone marrow of young adult rats." Ce11 Tissue Res 254: 3 1 7-330.

Marden. L. J., R. S. Fan. et al. (1993). "PDGF inhibits bone regeneration induced by

osteogenin, a bone morphogenetic protein, in rat craniotomy defects." J Clin Invest 92:

2897-905.

McCulloch, C. A. G., M. Strugurescu. et al. (1991). "Osteogenic progenitor cells in RBM

stromai populations exhibit self-renewal in culture." Blood 77(9): 1906- 19 1 1.

McIntyre. A. P. and B. Bjornson (1986). "Human bone rnarrow stroma1 ce11 colonies:

response to hydrocortisone and dependence on platelet-derived growth factor." Exo Hematol

14: 833-839,

Meyer, J., U. Mischeck, et ai. (1990). "Blood-brain barrier characteristic enzyme properties

in cultured brain capillary endothelial cells." Brain Res 514: 305-309.

Mitlak. B. H., R. D. Finkelman, et al. (1996). "The effect of systemicaily administered

PDGF-BB on the rodent skeleton." J Bone MIner Res 11(2): 238-47.

Miyagi, T. and H. Kimura (1995). "Phenotype and function of dendritic cells derived from

RBM cell cultures." Transplant Proc 27(2): 1568-70.

Modderman. W. E., T. Vrijheid-Lammers, et al. ( 1994). "Removal of hematopoietic cells

and macrophages from mouse bone marrow cultures: isolation of fibroblastlike stroma1

cells." Exp Hematol 22: 194-20 1.

Nash, T. J., C. R. Howlett, et al. (1994). "Effect of PDGF on tibia1 osteotomies in rabbits."

Bone lS(2): 203-8. -

Notoya, K.. K. Yoshida, et al. (1994). "Effect of ipriflavone on expression of markers

charactenstic of the osteoblast phenotype in M M stromal cell culture." J Bone Miner Res

9(3): 395-400.

Ogiso. B.. F. J. Hughes, et al. (1991). "Fibroblasts inhibit mineralized bone nodule

formation by RBM stromal cells in vitro." J Ce11 Phvsiol 146: 442-450.

Owen, M. ( 1988). "Marrow stroma1 stem cells." J CeII Sci SUDDI 10: 63-76.

Peled, A., D. Zipori, et al. (199 1). "Expression of a-smooth muscle actin in murine bone

marrow stromal cells." Blood 78(2): 304-309.

Pfeilschifter, J., R. Krernpien, et al. (1992). "Differential effects of PDGF isoforms on

plasmingen activator activity in fetal rat osteoblasis due to isoform-specific receptor

functions." Endocrinology 130(4): 2059-2066.

Pfeilschifter, J.. M. Oechsner, et al. (1990). "Stimulation of bone matrix apposition in vitro

by local growth factors: A cornparison between insulin-like growth factor 1, platelet-derived

growth factor, and transfoming growth factor P ." Endocrinoloey 127( 1 ): 69-75.

Pitas. R. E., T. L. Innenrity, et al. (198 1). "Aceto-acetylated lipoproteins used to distinguish

fibroblasts from macrophages in vitro by fluorescence rnicroscopy." Aneriosclerosis 1: 177-

185.

Ponten, F., 2. Ren, et al. (1994). "Epithelial-stroma1 interactions in basal ceil cancer: The

PDGF system." J Invest Dermatol 102(3): 304-9.

Rao, L. G., B. Ng, et al. (1977). "Parathyroid hormone and prostaglandin El response in a

selected population of bone cells after repeated subculture and storage at -80C."

Endocrinoloq 100: 1233- 124 1.

Rickard, D. J., 1. Kazhdan, et ai. (1995). "lmportance of 1,25-dihydroxyvitamin D3 and the

nonadherent cells of marrow for osteoblast differentiation from rat rnarrow stromal cells-"

Bone 16(6): 67 t -8. -

Rickard, D. J., T. A. Sullivan, et al. (1994). "Induction of rapid osteoblast differentiation in

RBM stromal ce11 cultures by dexamethasone and BMP-2." Dev Biol 161: 2 18-228.

Ross, W. M., J. Peeke, et al. (1991). "Reticulo-fibroblastoid stromal ce11 progenitors (CFU-

RF') in murine bone rnarrow." Exp Hematol 19: 1069- 1074.

Sasaki, T. and M.-H. Hong (1993). "Endothelin- 1 localization in bone ceils and vascular

endothelial ceIls in RBM." Anat Rec 237: 332-7.

Sato, K-, Y. Fujii, et al. (1986). "Recombinant human ILla and b stimulate mouse

osteoblast-like cells (MC3T3-El) to produce M-CSA and PGEZ." Biochem Biophvs Res

Commun 141( 1): 285-9 1.

Sato, K., K. Kasono, et al. (1987). "TNFa (cachectin) stimulates mouse osteoblast-like cells

(MC3T3-El) to produce macrophage-colony stimulating activity and PGE?." - Biochem

Biophvs Res Commun 145(l): 323-9.

120 Satomura, K. and M. Nagayama ( 199 1). "Ultrastructure of mineralized nodules formed in

RBM stroma1 ceIl culture in vitro," Acta Anat 142: 97-104.

Schatteman, G. C.. K. Morrison-Graham, et al. (1992). "Reguiation and role of PDGF

receptor a-subunit expression during embryogenesis." Development 115: 123- 13 1.

Scutt. A. and P. Bertram (1995). "Bone marrow cells are targets for the anabolic actions of

PGE2 on bone: Induction of a transition from nonadherent to adherent osteoblast

precursors." J Bone Miner Res lO(3): 474-87.

Shadmand, S. and J. E. Aubin (1995). "Interleukin 1-a has biphasic effects on bone

formationby RBM stromal cells." J Bone Miner Res 10(S 1): S409.

Shadrnand, S. and J. E. Aubin (1997). "IL4 expression by osteoblasts in vivo is

heterogeneous and varies with developmental age." J Bone Miner Res 12(Sl): Abstract

S319-

Shiina-Ishimi. Y.. E. Abe, et al. (1986). "Synthesis of colony-stimulating factor (CSF) and

differentiation-inducing factor (D-factor) by osteoblastic cells, clone MC3T3-E 1 ." Biochem

Bio~hvs Res Commun 134( 1): 400-6.

Simmons, D. J., P. Seitz, et al. (199 1). "Partial characterization of rat marrow stroma1 cells."

Calcif Tissue Int 48: 326-334.

Simmons, P. J. and B. Torok-S torb ( 199 1 ). "Identification of stromal ceIl precursors in

human bone marrow by a novel monoclonal antibody, STRO-1." Blood 78( 1): 55-62.

121 Stein, 0. and Y. Stein (1980). "Bovine aortic endothelial cells display macrophage-like

properties towards acetylated [I- 1251-low density lipoprotein." Biochim Bio~hvs Acta 620:

63 1-5.

Sullivan. A. K., D. Claxton, et al. (1989). "Cellular composition of RBM stroma."

Invest 60(4): 667-676.

Sunderland, C. A., W- R. McMaster, et aI. (1979)- "Purification with monocIonal antibody

of a predominant leukocyte-common antigen and glycoprotein frorn rat thymocytes." Eur J

Immunol 9: 155- 159.

Taichman, R. S. and S. G. Emerson (1994). "Human osteoblasts support hematopoiesis

through the production of G-CSF." J Exp Med 179: 1677-82.

Tenenbaum, H. C. and J. N. M. Heersche (1982). "Differentiation of osteoblasts and

formation of mineraIized bone in vitro." Calcif Tissue Int 34: 76-79.

Tenenbaum, H. C.. N. Kamalia, et al. (1995). "Probing glucocorticoid-dependent

osteogenesis in rat and chick cells in vitro by specific blockade of osteoblastic

differentiation with progesterone and RU38486." Anat Rec 242: 200-10.

Thomson, B. M., I. Bennett, et al. (1993). "Preliminary characterization of porcine bone

marrow stroma1 cells: Skeletogenic potential. colony-forming activity, and response to

dexamethasone, transforming growth factor B, and basic fibroblast growth factor." I Bone

Miner Res 8( 10): 1 173-83.

122 Thomson, B. M., V, Dean, et al. (1993). "Visualization and selective elimination of a

subpopulation of mitogen-responsive bone marrow stromal cells using bromodeoxyundine

and photo-induced ce11 killing." Bone 14: 779-86.

Tibone, K. W. and G. W. Bernard (1982). A new in vitro mode1 of intramembranous

osteogenesis from adult bone marrow stem cells. Factors and Mechanisms In fluencin e Bone

Growa. A. D. Dixon and B. G. Samat. New York, Alan R. Liss Inc. 101: 107-123.

Turksen, K. and J. E. Aubin (1991). "Positive and negative immunoselection for ennchment

of two classes of osteoprogenitor cells." J Cell Biol 1 l4(2): 373-384.

Turksen, K., U. Bhargava, et al. (1992). "Isolation of monoclonal antibodies recognizing rat

bone-associated molecules in vitro and in vivo." J Histochem Cvtochern 40(9): 1339-52.

Udagawa, N., N. Takahashi, et al. (1989). "The bone marrow-derived stroma1 ce11 lines

MC3T3-G2/PA6 and ST2 support osteoclast-li ke ce11 differentiation in cocultures with

mouse spleen cells." Endocnnology 125: 1805- 1 8 1 3.

Udagawa, N., N. Takahashi. et al. (1990). "Origin of osteoclasts: Mature monocytes and

macrophages are capable of differentiating into osteoclasts under a suitable

microenvironment prepared by bone marrow-derived stromal cells." J Ce11 Bi01 87: 7260-

7264.

Van der Plas, A. and P. J. Nijweide (1992). "Isolation and purification of osteocytes." J

Bone Miner 7: 389-96.

123 Van Vlasselaer, P., B. Borremans, et ai. (1993). "Interleukin-10 inhibits the osteogenic

activity of mouse bone marrow." Blmd 82(8): 236 1-70.

Van Vlasselaer, P., N. Falla, et al. ( 1994). "Characterization and purification of osteogenic

celis from murine bone marrow by two-color ce11 sorting using anti-Sca-1 monoclonal

antibody and wheat germ agglutinin." Blood 84(3): 753-63.

Vilamitjana-Amedee, J., R. Bareille, et al. (1993). "Human bone marrow stroma1 cells

express an osteoblastic phenotype in culture." In Vitro Ce11 Dev Biol 29(A): 699-707.

Wang, Q., Z. Yan, et al. (1990). "Dissecting the hematopoietic microenvironment. VI. The

effects of several growth factors on the in vitro growth of murine bone rnarrow CN-F." EXP

HematoI 18: 34 1-347.

Wang, Q.-R. and N. Wolf ( 1990). "Dissecting the hernatopoietic microenvironment.VIII.

Clonal isolation and identification of cell types in murine CN-F colonies by limiting

dilution." Exp Hematol 18: 355-359.

Weir, E. C., M. C. Horowitz, et ai. (1993). "M-CSF release and receptor expression in bone

cells." J Bone Miner Res 8( 12): 1507- 15 18.

Weir. E. C., K. L. Insogna, et al. ( 1989). "Osteoblast-like cells secrete GM-CSF in response

to PTH and LPS." Endocrinology - 124(2): 899-904.

Weiss, L. and U. Geduldig (199 1). "Barrier cells: Stroma1 regulation of hematopoiesis and

blood cell release in normal and stressed murine bone marrow." Blood 78(4): 975-990.

1 24 Westen, H. and D. P. Bainton (1979). "Association of alklaline-phosphatase-positive

reticulum cells in bone marrow with granulocyte precursors." I EXD Med 150: 9 19-37.

Westermark, B. and C. Heldin ( 199 1 ). "Platelet-derived growth factor in autocrine

transformation." Cancer Research Sl(0ctober): 5087-5092.

Wilkins, B. S. and D. B. Jones (1995). "Immunohistochemical characterization of intact

stroma1 layers in long-tem cultures of human bone marrow." Br J Haematol 90: 757-66.

Woollett, G. R., A. N. Barclay, et al. (1985). "Molecular and antigenic heterogeneity of the

rat leukocyte-common antigen from thymocytes and T and B lymphocytes." Eur J Immun01

15: 168-173.

Xiong, X., F. Liu, et al. (1997). "Expression of IL-4, IL-6. and GP130 as a function of

osteoblast differentiation in bone colonies in fetal rat calvarial ce11 population." Second

Regional Svmposium on Skeletal Biotogv and Metabolic Bone Disease London, Ontario.

Yan, X.-Q., G. Brady, et al. ( 1993). "PDGF activates primitive hematopoietic precursors

(pre-CFCrnulti) by up-regulating IL- 1 in PDGF receptor-expressing macrophages." 1

Immun01 150(6): 2440-8,

Yan, X.-Q., G. Brady, et al. (1994). "Overexpression of PDGF-b in murine hematopoietic

cells induces a lethal myeloproliferative syndrome in vivo." Oncogene 9: 163-73.

Yanai, N.. T. Satoh, et al. (1991). "Endothelial cells create a hematopoietic inductive

microenvironment preferential to erythropoiesis in the mouse spleen." Ce11 Stmct Funct 16:

87-93.

Zhang. L.. E. Leeman. et al. ( 1 99 1) . "Human osteoblasts synthesize and respond to platelet -

denved growth factor." Am J Phvsiol 261: C348-C354.

Zhang, R. W.. D. J. Simmons. et al. (1991). "Contribution of marrow stroma1 cells to the

regulation of osteoblast proliferation in rats: evidence for the involvement of insulin-like

growth factors." Bone Miner 13: 20 1-2 15.

Zhang, R. W., S. C. Supowit, et ai. (1995). "Expression of selected osteogenic markers in

the fibroblast-like cells of rat marrow stroma." CaIcif Tissue int 56: 283-91.

IMAGE EVALUATION TEST TARGET (QA-3)

APPLIED - IWGE . lnc = 1653 East Main Street - -. . - Rochester. NY 14609 USA -- - - = Phone: 7'1 W4l8~-O3OO -- --= Fax 71 â1286-5989

O 1993. Appid Image. Inc.. All ALgtim RBseiued