You are on page 1of 10

JOURNAL OF CELLULAR PHYSIOLOGY 203:573–582 (2005)

Human Trabecular Bone-Derived Osteoblasts


Support Human Osteoclast Formation In Vitro
in a Defined, Serum-Free Medium
GERALD J. ATKINS,1* PANAGIOTA KOSTAKIS,1,2 KATIE J. WELLDON,1 CRISTINA VINCENT,1
DAVID M. FINDLAY,1 AND ANDREW C.W. ZANNETTINO2
1
Department of Orthopaedics and Trauma, University of Adelaide,
and the Hanson Institute, Adelaide, South Australia, Australia
2
Division of Haematology, Institute of Medical and Veterinary Science,
and the Hanson Institute, Adelaide, South Australia, Australia

While it has been assumed that osteoblasts in the human support osteoclast formation, in vitro evidence of this is currently lacking.
We tested the ability of normal human trabecular bone-derived osteoblasts (NHBCs) to support osteoclast formation from human
peripheral blood mononuclear cells (PBMC) in response to treatment with either 1a,25-dihydroxyvitamin D3 (1,25D) or parathyroid
hormone (PTH), using a serum-replete medium previously used to support human osteoclast formation on a stroma of murine ST-2
cells. Under these conditions, NHBC did not support osteoclast formation, as assessed by morphological, histochemical, and
functional criteria, despite our previous results demonstrating a link between induction of RANKL mRNA expression and NHBC
phenotype in these media. We next tested a defined, serum-free medium (SDM) on NHBC phenotype, their expression of RANKL
and OPG, and their ability to support osteoclast formation. SDM, containing dexamethasone (DEX) and 1,25D, induced phenotypic
maturation of NHBC, based on the expression of STRO-1 and the bone/liver/kidney isoform of alkaline phosphatase (AP). PTH as a
single factor did not induce phenotypic change. 1,25D and DEX induced the greatest ratio of RANKL:OPG mRNA, predictive of
supporting osteoclast formation. Consistent with this, co-culture of NHBC with CD14þ PBMC, or bone marrow mononuclear cell
(BMMC), or CD34þ BMMC precursors in SDM þ 1,25D þ DEX, resulted in functional osteoclast formation. Osteoclast formation
also occurred in PTH þ DEX stimulated co-cultures. Interestingly, SDM supplemented with recombinant RANKL (25–100 ng/ml)
and M-CSF (25 ng/ml), did not induce osteoclast formation from any of the osteoclast precursor populations in stromal-free cultures,
unlike serum-replete medium. This study demonstrates that under the appropriate conditions, adult human primary osteoblasts
can support de novo osteoclast formation, and this model will enable the detailed study of the role of both cell types in this
process. J. Cell. Physiol. 203: 573–582, 2005. ß 2004 Wiley-Liss, Inc.

Cells of the osteoblast lineage perform a number of model for the differentiation of human osteoclasts using
functions, including the formation of new bone during human osteoblasts as a stromal layer has, until now, not
modelling and remodelling (Parfitt, 1984), organic been described.
matrix degradation at the initiation of remodelling Models of human osteoclast formation described
(Everts et al., 2002), support of haemopoiesis (Taichman previously include those in which human osteoclast
and Emerson, 1998), and notably, osteoclastogenesis precursors are cultured together with rodent cell lines,
(Hofbauer et al., 2000). However, the way in which such as mouse ST-2 cells (Fujikawa et al., 1996, 2001)
these seemingly diverse functional states of osteoblasts or rat UMR-106 cells (Quinn et al., 1998), as stroma.
interrelate, and the phenotypic identity of the osteo- Whilst these assays are useful in examining interactions
blasts involved, has not been elucidated, particularly for between osteoclast precursors and osteoblast-lineage
human osteoblasts. Evidence linking the osteogenic and cells (Haynes et al., 1999; Atkins et al., 2000; Fujikawa
osteoclast-inductive properties of osteoblasts is sparse, et al., 2001), it is not certain that these assays can
despite the long held view that the processes of bone accurately model human osteoclastogenesis. The limita-
resorption and bone formation are intimately coupled tions of these models include the use of transformed cell
(Parfitt, 2000). Normal human trabecular bone-derived lines, as opposed to primary cells, as stromal layers and
osteoblastic cells (NHBCs) provide a potentially valu- the demonstrated differences that exist between rodent
able in vitro model system, with which to investigate the and human osteoblast differentiation (Siggelkow et al.,
relationship between these diverse functional proper-
ties. It has been shown that NHBC retain phenotype
plasticity in the absence of an osteoinductive stimulus,
and are able to differentiate into several mesenchymal
lineages given the appropriate growth stimuli (Sottile Contract grant sponsor: National Health and Medical Research
et al., 2002). However, culture in the continuous pre- Council of Australia; Contract grant sponsor: Eli Lilly Australia
sence of long-acting ascorbate (ascorbate-2-phosphate, Ltd.; Contract grant sponsor: EISAI Co. Ltd., Japan; Contract
grant sponsor: Department of Orthopaedic Surgery and Trauma,
AS2P) confers an osteoblastic phenotype and NHBC University of Adelaide.
have been extensively characterised from the viewpoint
of osteogenesis (Gronthos et al., 1999; Stewart et al., *Correspondence to: Gerald J. Atkins, Department of Orthopae-
1999; Sottile et al., 2002). In contrast, little is known dics and Trauma, University of Adelaide, North Terrace,
Adelaide, South Australia, Australia, 5000.
regarding the ability of these cells to support osteoclast E-mail: gerald.atkins@adelaide.edu.au
formation from naive precursors. In fact, the identifica-
tion of the human osteoblast phenotype that promotes Received 13 August 2004; Accepted 6 October 2004
osteoclastogenesis has been elusive and an in vitro DOI: 10.1002/jcp.20255
ß 2004 WILEY-LISS, INC.
574 ATKINS ET AL.

1999). The development of human stroma-driven models L-ascorbate-2-phosphate (AS2P, Novachem, Melbourne, VIC,
of osteoclast differentiation is therefore highly desir- Australia) (a-MEM-10). Single cell suspensions were obtained
able. Mbalaviele et al. (1999) have described the use of from confluent primary NHBC cultures by enzymatic digestion
undifferentiated mesenchymal stem cells and CD34þ with collagenase (3 mg/ml) (Collagenase Type I; Worthington
Biochemical Co., Freehold, NJ) and dispase (4 mg/ml) (Neutral
bone marrow (BM) cells to generate human osteoclasts Protease Grade II; Boehringer Mannheim, GMBH, Germany)
in vitro. A number of investigators have used trans- for 30 min at 378C, followed by trypsin digestion for 5 min at
formed human osteosarcoma cell lines to support human 378C.
osteoclast formation (Blair et al., 2000; Itoh et al.,
2000; Miyamoto et al., 2002). Neale et al. (2000) used Isolation of CD34þ bone marrow
NHBC as stroma to support functional osteoclast-like mononuclear cell (BMMC)
cells formation from the synovial tissue associated with
peri-prosthetic osteolysis. However, this tissue repre- BMMC were enriched from BM aspirates from normal
healthy volunteers by diluting whole marrow 1:3 in Hank’s
sents a rich source of late-stage, committed osteoclast balanced salt solution (HBSS) and layering onto Ficoll–Paque
progenitors. To our knowledge, functional osteoclast separation media (Pharmacia Biotech, Uppsala, Sweden).
formation from a source of relatively immature osteo- After centrifugation at 400g for 25 min, mononuclear cells
clast precursors, using NHBC, has not been described were retrieved and washed twice in HBSS containing 10%
previously. FCS. An aliquot of cells was removed for CD34 staining by
We recently reported (Atkins et al., 2003) a link immunofluorescence. All subsequent steps were performed on
between the induced expression by NHBC of RANKL, a ice or at 48C, unless otherwise specified, and under sterile
key factor required for osteoclast formation (Lacey et al., conditions. CD34þ cells were isolated, essentially as described
1998; Tsukii et al., 1998; Hofbauer et al., 2000; Udagawa previously (Zannettino et al., 1998). Briefly, CD34-detacha-
beads (4  107, 100 ml/ml cells, Dynal Biotech, Oslo, Norway)
et al., 2000), and the cell-surface expression of STRO-1 were washed twice in 1 ml HHF medium (HBSS, 20 mM Hepes,
and AP, phenotypic markers of osteoblast differen- 10% FCS) before addition to the BMMC suspension. The cell-
tiation (Gronthos et al., 1999; Stewart et al., 1999). detachabead suspension was mixed for 30 min with rotation.
Induction of RANKL expression in response to 1,25D Isolation buffer was added to the cells, which were then placed
and DEX correlated positively with an induced increase in a magnet apparatus for 2 min. The supernatant containing
in the percentage of cells expressing STRO-1 and nega- CD34 BMMC was removed, washed three times in HHF and
tively with the percentage of cells expressing AP (Atkins then processed for CD14, as described below. The remaining
et al., 2003). On the basis of those results, we hypo- CD34þ BMMC complexed to beads were washed three times in
thesised that support of osteoclast formation by osteo- isolation buffer and then resuspended at 4  108 cells/ml in
HHF. To this suspension was added two volumes of detacha-
blasts occurs as a function of their differentiation state, bead buffer, and the cells were then mixed using a vortex
as defined by STRO-1/AP (Gronthos et al., 1999; Stewart mixer. The mixture was incubated at 378C with rotation for the
et al., 1999). detachment of the beads from the cells, after which time the
In this study, we sought to define the conditions under volume of the tube was increased by 2 ml with HHF. The cell
which NHBC support osteoclast formation. Although we suspension was vigorously mixed and then placed in a magnet
were unable to generate osteoclasts in serum-replete apparatus for 2 min. Supernatant containing detached cells
media in co-cultures, we were able to successfully gen- was collected and the wash process was repeated three times.
erate functional human osteoclasts from precursors in The recovered cells were pooled, washed twice by centrifuga-
human peripheral blood or BM cells in a defined, serum- tion at 400g in HHF, resuspended in thaw media (HBSS con-
taining Dnase 1) and rotated at 48C until used.
free medium based on that developed for the efficient
growth of BM colony forming units-fibroblast (CFU-F)
(Gronthos and Simmons, 1995; Gronthos et al., 2003). Isolation of CD14þ BMMC and peripheral blood
mononuclear cells (PBMCs)
The ability to support osteoclast formation was in-
dependent of the addition of exogenous RANKL and PBMC from the peripheral blood of healthy donors were
M-CSF, and was associated with the rapid phenotypic isolated, as described for BMMC, and both populations were
differentiation of the NHBC. Thus, under hormonal sti- sorted by magnetic activation cell sorting (MACS) for CD14þ
cells. First, an aliquot of cells was taken to assess input CD14
mulation, NHBC appear capable of providing all of the levels by immunostaining with the CD14 mouse monoclonal
necessary factors for the support of complete osteoclas- antibody (MAb) FMC-17 (kindly provided by Prof. Peter
togenesis from CD14þ monocytic cells (Massey and McCardle, Flinders Medical Centre, Bedford Park, South
Flanagan, 1999) derived from peripheral blood or BM, Australia, Australia). All remaining steps were performed at
and from immature CD34þ BM cells. This novel model of 48C or on ice. The remaining cells were incubated in blocking
adult human primary osteoblasts driving human osteo- buffer, consisting of 10% v/v normal rabbit serum in HHF,
clast differentiation will enable the in vitro study of the for 20 min. After this time, cells were incubated in FMC-17
role of both cell types in this process. In addition, the use supernatant for 1 h, and then washed twice in HHF by centri-
of a defined, serum-free medium will allow the study of fugation at 200g. After this, a 1/50 dilution of goat anti-mouse
g-biotin (Southern Biotechnology Associates, Birmingham,
putative regulators of bone remodelling in the absence of UK) in HHF buffer was added and the cells incubated for 1 h.
the potentially complex influence of serum. Cells were then washed twice by centrifugation in MACS
buffer (Ca2þ/Mn2þ-free PBS supplemented with 1% BSA, 5 mM
MATERIALS AND METHODS EDTA, and 0.01% sodium azide). Cells were resuspended in
NHBC
0.9 ml of MACS buffer and to these was added 0.1 ml
All human donor material in this study was used after streptavidin microbeads (Miltenyi Biotec; Bergisch Gladbach,
informed donor consent, which was approved by the Ethics Germany), which were left for 15 min to attach. Cells were then
committees of the Institute of Medical and Veterinary Science washed twice and then resuspended in 0.5 ml of MACS buffer.
and the Royal Adelaide Hospital. NHBC were derived from The cell suspension was then loaded onto a mini MACS column
trabecular bone fragments and cultured, as we have described (Miltenyi Biotec), which was then washed through twice with
previously (Atkins et al., 2003). NHBC were cultured to con- 0.5 ml MACS buffer to retrieve the CD14 cells. After addition
fluence in 75 cm2 tissue-culture flasks in alpha-minimal essen- of a further 0.5 ml MACS buffer, the column was removed from
tial medium (a-MEM: Flow Laboratories, Irvine, Scotland) the magnet and the CD14þ cells aspirated using a syringe
supplemented with 10% foetal calf serum (FCS, Thermotrace, plunger. An aliquot of cells from each fraction was stained with
Noble Park, VIC, Australia), 2 mM L-glutamine and 100 mM streptavidin-FITC to assess purity by flow cytometry. By this
HUMAN OSTEOBLAST SUPPORT OF OSTEOCLASTOGENESIS 575

method, greater than 95% purity of CD14þ cells was routinely Preparation of total RNA and real-time reverse
achieved (data not shown). transcription PCR
Total RNA was prepared from NHBC, and RT-PCR per-
Human–human osteoclast forming co-culture formed, as we have recently reported (Atkins et al., 2000).
Briefly, cells were dissolved in Trizol reagent (Life Technolo-
Semi-confluent cultures of NHBC were digested with gies, Gaithersburg, MD). Total RNA was then prepared as per
dispase/collagenase/trypsin, and single cell suspensions were manufacturer’s instructions. RNA was reverse transcribed
then washed twice and resuspended in media for co-culture. from 1 mg of total RNA from each sample, using a cDNA syn-
Co-cultures were performed using media used previously to thesis kit, as per manufacturer’s instructions (Promega Corp.,
support human osteoclast formation on stroma of murine ST-2 Madison, WI). RANKL, OPG, and GAPDH mRNA expression
cells (Fujikawa et al., 1996; Haynes et al., 1999; Atkins et al., was analysed by real-time PCR using the SYBR GREEN-1
2000). This consisted of a-MEM supplemented with 10% FCS, (Molecular Probes, Eugene, OR) detection system. For RANKL
L-ascorbate-2-phosphate (100 mM, Novachem), DEX (10 nM), expression, the following primer sequences were used: 50 -
L-glutamine (2 mM), 1,25D (20 nM, Wako Pure Chemical TCAGCCTTTTGCTCATCTCACTAT-30 (forward); 50 -CCAC-
Industries, Osaka, Japan), and recombinant human M-CSF CCCCGATCATGGT-30 (reverse). OPG and GAPDH cDNAs
(25 ng/ml, Genetics Institute, Cambridge, MA). Co-cultures were amplified using our previously published primer sequen-
were also performed in serum-deprived medium (SDM), based ces (Atkins et al., 2000). cDNA was amplified using AmpliTaq
on a medium shown previously to support the growth of human Gold DNA polymerase (Perkin Elmer, Norwalk, CT) ampli-
CFU-F (Gronthos and Simmons, 1995). SDM consisted of fication, as described (Atkins et al., 2000), on a Rotor-Gene
a-MEM (Flow Laboratories), supplemented with 1% BSA (Cat. thermocycler (Corbett Research, Mortlake, NSW, Australia),
No. A-2153, Sigma Chemical Co., St. Louis, MO), transferrin with the addition to each reaction of SYBR GREEN-1
(100 mg/ml), insulin (10 mg/ml), b-mercaptoethanol (5  105M), (1:100,000 dilution of the commercial stock, Molecular Probes).
human low density lipoprotein (LDL, 20 mg/ml, a kind gift from All PCRs were validated by the presence of a single peak in the
A/Prof. Kerry-Anne Rye, Heart Research Institute, Sydney, melt curve analysis and amplification of a single specific
Australia), L-ascorbate-2-phosphate (100 mM, Novachem), product was further confirmed by electrophoresis on a 2.5% w/v
epidermal growth factor (EGF, 10 nM, Peprotech, Inc., agarose gel. Relative quantification of RANKL and OPG
Rocky Hill, NJ), PDGF-BB (10 nM, Peprotech), DEX (10 nM, mRNA expression between samples was calculated using the
Fauldings, Adelaide, Australia), L-glutamine (2 mM), and comparative cycle threshold (CT) method (DCT) (Anonymous,
Hepes (20 mM). NHBC were then seeded onto sterilised 4 mm2 1997). Briefly, the formula XN ¼ 2DCT was used, where XN is
slices (0.1 mm thick) of human cortical bone, or elephant tusk the relative amount of target gene in question and DCT is the
dentine in a 96-well plate, at a density of 2–4  105 cells/ml difference between the CT of the gene in question and the CT of
in 0.1 ml volumes, and cultured overnight in 5% CO2 in air the housekeeping gene, GAPDH, for a given sample.
at 378C. Following this, CD14þ BMMC, CD14þ PBMC, or
CD34þ BMMC, isolated as described above, were washed twice Flow cytometric analysis
in SDM, counted and resuspended at 6  105 cells/ml (for NHBC were analysed by flow cytometry for the expression of
CD14þ cells) or 1  105 cells/ml (for CD34þ cells), and added STRO-1 (Simmons and Torok-Storb, 1991) and AP, as des-
to wells containing NHBC or onto virgin bone/dentine. The cribed previously (Atkins et al., 2003). Analysis was performed
cells were allowed to attach for 2 h at 378C, and then four bone/ on an Epics (Beckman Coulter, Fullerton, CA) flow cytometer.
dentine slices were transferred into wells of a 24-well plate
containing 1 ml SDM, additionally supplemented with either Statistical analysis
recombinant peptide PTH1 –34 (20 nM, Auspep Pty. Ltd.,
Parkville, VIC, Australia) or 1,25D (20 nM, Roche, Uppsala, Parametric data sets were analysed using Student’s t-test.
Sweden), and recombinant human M-CSF (25 ng/ml, Genetics Multiple and non-parametric datasets were analysed for dif-
Institute), as indicated. In some experiments, media were ferences by one way analysis of variance (ANOVA) and Tukey’s
supplemented with recombinant human RANKL or OPG Multiple Comparison using GraphPad Prism V4 software
(Peprotech, Inc.) at the concentrations indicated. Cultures (GraphPad Software Ltd., San Diego, CA). P values <0.05 were
were maintained for 21 days and media were replaced every considered to be significant.
3 days.
RESULTS
Comparison of ST-2 and NHBC in supporting
ST-2 osteoclast-forming co-culture human osteoclast formation
Human osteoclasts were derived from PBMC on a stromal
support of ST-2 cells, essentially as described previously
Initial experiments examining the osteoclastogenic
(Fujikawa et al., 1996; Haynes et al., 1999; Atkins et al., potential of NHBC were closely modelled on a serum-
2000). PBMC (2  105 cells) or CD14þ PBMC (6  104 cells) replete culture system described by Fujikawa et al.
were plated onto seeded layers of 8  103 ST-2 cells on dentine (1996), except that we substituted NHBC for mouse ST-2
slices. Non-adherent cells were removed by washing three cells. Because NHBC proliferate slowly compared with
times with HBSS after 1 h at 378C, and pairs of dentine slices ST-2 cells, we titrated the number of NHBC added to
were placed in 16 mm diameter wells containing 1 ml of a- these co-cultures, starting from 8  103 per dentine slice,
MEM medium with 10% v/v FCS, 20 nM 1,25D, 10 nM DEX as used for ST-2 cells, up to 4  104 per slice. Under these
(Fauldings), and 25 ng/ml M-CSF. Media were replenished conditions, ST-2 cells supported robust osteoclast forma-
every 3 days throughout the experiment.
tion from unfractionated human PBMC, as determined
by TRAPþ multinucleated cell (MNC) and resorption pit
Tartrate resistant acid phosphatase (TRAP) staining formation on dentine. In contrast, neither TRAPþ cells
and measurement of bone resorption nor functional osteoclasts were generated using NHBC
Cultures on slices of dentine were fixed and stained for isolated from four individual donors as a stromal sup-
TRAP, as recommended by the manufacturer (Sigma Chemical port, under otherwise identical culture conditions (data
Company, Castle Hill, Australia). TRAP positive cells were not shown).
visualised by light microscopy and images taken with a Nikon
DIH digital camera. Dentine slices were then treated with 6M Comparison of RANKL and OPG expression in
ammonium hydroxide for at least 1 h, and then the fixed SDM versus serum-replete medium
adherent cells were removed with an artist’s brush. Following
this, the bone slices were washed four times in deionised water, As an alternative to the serum-replete medium used
dehydrated in 100% ethanol, air-dried and then processed for in ST-2 co-cultures, we next proposed to test osteoclast
scanning electron microscopy (Haynes et al., 1999). formation supported by NHBC in a defined SDM. Since
576 ATKINS ET AL.

the ratio of RANKL to OPG mRNA is a major determi-


nant of the propensity to support osteoclast formation,
we initially tested the expression of RANKL and OPG
mRNA by NHBC cultured in the absence of PBMC,
in media containing combinations of the components
in SDM. NHBC were cultured for 72 h in basal SDM
(B-SDM), comprising a-MEM supplemented with in-
sulin, transferrin, BSA, and human LDL, being the
basic requirements for growth in the absence of serum
(Gronthos and Simmons, 1995). Other components of
the previously published formula for SDM, EGF, and
PDGF-BB (Gronthos and Simmons, 1995), and factors
that were pro-osteoclastogenic in the serum replete co-
culture medium used with ST-2 cells, DEX and 1,25D
(Fujikawa et al., 1996), were then added alone or in com-
bination. These media were compared with the serum-
replete medium, used previously to examine changes in
NHBC gene expression and phenotype (Atkins et al.,
2003). Of the conditions tested, only culture in media
containing both EGF and PDGF-BB (i.e., complete SDM),
complete SDM containing DEX, and those additionally
supplemented with 1,25D, gave appreciable increases in
RANKL mRNA expression (Fig. 1A). Conditions that
favoured RANKL mRNA expression were also associat-
ed with lower OPG mRNA expression (Fig. 1B), and
thereby a higher RANKL:OPG mRNA ratio (Fig. 1C).
Significantly, the RANKL:OPG mRNA ratio in SDM
media was markedly higher than that seen in serum-
replete medium containing DEX and 1,25D (Fig. 1C).
Based on these data, complete SDM with the addition of
1,25D and DEX was predicted to be the medium most
favourable for osteoclast formation.
Comparison of NHBC phenotype in SDM
versus serum-replete medium
We previously showed a link between osteoblast phe-
notype, as defined by STRO-1 and AP expression, and
RANKL expression when experiments were performed
in serum-replete medium (Atkins et al., 2003). The
phenotype of NHBC cultured in SDM has not been
previously reported. Thus, we next tested the effect
on osteoblast phenotype of culture in media consisting
of various combinations of the components of SDM.
NHBC were cultured for 72 h, as for the gene expression
studies described above, and analysed for STRO-1 and
AP expression by flow cytometry. The expression of
STRO-1/AP in response to culture in B-SDM is shown in
Figure 2A. All additions to this basal medium signifi-
cantly affected the phenotypic profile. EGF decreased Fig. 1. Real-time RT-PCR analysis of (A) RANKL and (B) OPG
mRNA expression, and (C) the resulting RANKL:OPG mRNA ratio, in
the STRO-1þ populations and increased the AP popula- normal human trabecular bone-derived osteoblasts (NHBC) cultured
tion, consistent with a differentiative effect (Fig. 2B). in the absence of peripheral blood mononuclear cells (PBMC), under
Culture in PDGF-BB as a single factor increased the serum-free conditions for 72 h in various components of serum-
proportion of osteoprogenitor cells (%STRO-1þ/AP) deprived medium (SDM). Data are expressed as the DCT values
relative to the expression of GAPDH (A and B) or the ratio of these
(Fig. 2C). Culture in B-SDM with EGF and PDGF-BB values (C), and are means of triplicate reactions  standard errors of
together (complete SDM) increased the %STRO-1APþ the mean (SEM). Similar results were obtained in two independent
and %STRO-1/AP, consistent with a net osteogenic experiments.
effect of these factors (Fig. 2D). DEX in complete SDM
dramatically enhanced the %STRO-1APþ (Fig. 2E).
The addition of 1,25D in complete SDM increased the phenotype over the 72 h period at 20 nM (compare
proportion of STRO-1þ/APþ and the STRO-1/APþ Fig. 2D with 2I), nor was the phenotype affected further
populations, compared with complete SDM alone (com- by a combination of PTH and DEX in (Fig. 2J). Together,
pare parts D and F). Culture of NHBC in complete SDM these data are consistent with a stronger osteogenic
with the addition of DEX and 1,25D resulted in a further effect and more rapid phenotypic maturation in SDM
increase in the percentage of STRO-1/APþ cells than in serum-replete media.
(%STRO-1APþ) (Fig. 2G), and greatly enhanced the
differentiation-associated progression into the STRO- Osteoclast formation in SDM
1/APþ population when compared with culture in DEX Having established the composition of the medium
and 1,25D-supplemented standard serum-replete med- that resulted in the greatest ratio of RANKL:OPG
ium (Fig. 2H). PTH had no additional effect on NHBC mRNA, and which would therefore most likely favour
HUMAN OSTEOBLAST SUPPORT OF OSTEOCLASTOGENESIS 577

Fig. 2. Flow cytometry dot plots showing the expression of STRO-1 combination with 1,25D (20 nM); (H) serum-replete a-MEM contain-
(Y-axis) and AP (X-axis) by NHBC cultured for 72 h in (A) basal SDM ing 1,25D (20 nM) and DEX (10 nM); (I) complete SDM containing
(B-SDM) (SDM without added EGF or PDGF-BB); (B) B-SDM con- PTH (20 nM); (J) complete SDM containing PTH (20 nM) and DEX
taining EGF (10 mg/ml); (C) B-SDM containing PDGF-BB (10 mg/ml); (10 nM) in combination. Percentages of cells in each quadrant are
(D) B-SDM containing EGF and PDGF-BB together (complete SDM); indicated. Staining patterns shown are representatives of those ob-
(E) complete SDM containing DEX (10 nM); (F) complete SDM con- tained using two independent donors’ cells.
taining 1,25D (20 nM); (G) complete SDM containing DEX (10 nM) in

osteoclast formation, we next tested the ability of NHBC and most frequently, the edge of the bone or dentine slice
to support complete osteoclast formation in SDM. To (e.g., Fig. 3B,C, and E). Significantly, cultures grown
more stringently define the system, we used MACS- simultaneously in media containing 10% FCS failed to
purified CD14þ PBMC, shown previously to contain the generate osteoclast-like cells in the presence of either
osteoclast precursor population (Massey and Flanagan, 1,25D or PTH (not shown). In experiments where both
1999). NHBC were co-cultured with purified CD14þ peripheral blood and BM precursors were isolated from
PBMC on dentine slices in SDM containing DEX, with or the same donor, functional osteoclasts were derived
without 1,25D. After culture for 21 days, large TRAPþ successfully from CD14þ PBMC (Fig. 3D), CD14þ
cells were detected in co-cultures stimulated with 1,25D BMMC, with similar growth kinetics and appearance
(Fig. 3A,B). It has been our experience with human (Fig. 3E), and from immature CD34þ BMMC. In the case
cells that TRAP positivity and/or multinucleation does of CD34þ BMMC, the resulting resorption sites
not always equate to functional osteoclast formation appeared as rosette-like clusters (Fig. 3F), perhaps
(unpublished data). It was therefore important in these resulting from discrete colonies of mononuclear pro-
cultures to demonstrate the resorbing capability of these geny. While the addition of recombinant human M-CSF
cells. SEM analysis revealed the presence of typical is a necessary addition to the ST-2 co-cultures because of
resorption lacunae underlying regions of TRAPþ cells the lack of species cross-reactivity, exogenous M-CSF in
(Fig. 3C). It was noted that resorption was frequently this human-human system was not a requirement
associated with pre-existing topological features of the (Fig. 3G) most likely due to its endogenous expression
substrate, such as the edges of haversian canals when by NHBC (e.g., Yao et al., 1998 and confirmed in our
human cortical bone was used, small (approximately laboratory; data not shown). Similar resorption data
1 mm) pores, likely to be microvessels, or alternatively, were obtained from experiments, in which PTH1–34 was
578 ATKINS ET AL.

Fig. 3. Human–human osteoclast forming co-cultures in serum-free culture of the same CD14þ BMMC, as in (E), except in (G) the absence
conditions after 21 days of culture. Unless otherwise stated, co- of added M-CSF and (H) the absence of added M-CSF and 1,25D. In all
cultures included 1,25D (20 nM) and DEX (10 nM). A: A transmission cases, the white bar represents 200 mm. I: Quantitation of resorption
image collected on a confocal microscope of tartrate resistant acid area of experiment depicted in D–H. Average resorption  SEM was
phosphatase (TRAPþ) multinucleate cells (arrows) derived from calculated using Imagequant software (Molecular Dynamics) of mul-
CD14þ PBMC. B: Light microscopic image of typical ‘‘edge’’ staining tiple.TIF images encompassing the entire surface area of four 4 mm2
of osteoclasts, in this case from CD14þ BMMC and (C) the same region cortical bone slices per co-culture. One-way analysis of variance anal-
analysed by scanning electron microscopy for resorption pits. D–F: ysis (ANOVA) revealed no significant difference existed between co-
Showing resulting electron microscopic images of co-cultures of (D) cultures, except in the case of no added M-CSF and 1,25D (*P < 0.05).
CD14þ PBMC, (E) CD14þ BMMC and (F) CD34þ BMMC, all isolated Similar resorption was obtained using NHBC isolated from three
from the same donor. Resorption lacunae were also derived from a co- donors.
HUMAN OSTEOBLAST SUPPORT OF OSTEOCLASTOGENESIS 579

utilised as a stimulus in place of 1,25D (data not shown). Recombinant RANKL/M-CSF do not support
Occasionally, osteoclasts formed in co-cultures where no stroma-free osteoclast formation in SDM
M-CSF, 1,25D, or PTH was added, although the extent All osteoclastogenesis assays included, as a control,
and number of resorption pits was always small the respective osteoclast precursor populations cultured
compared with hormone treated cells (Fig. 3H,I). Under in the absence of stroma with the addition of recombi-
stimulation with 1,25D, the extent of resorption result- nant RANKL (20–30 ng/ml) and M-CSF (25 ng/ml).
ing from all of CD14þ PBMC and BMMC, and CD34þ PBMC cultured in SDM under these conditions gave rise
BMMC populations, under the conditions described, only to few, weakly TRAPþ, cells. In addition, these cells
was found to be similar (Fig. 3I). Together, these are the did not form resorption lacunae. However, the same cells
first data to demonstrate that, given the appropriate cultured in serum-replete a-MEM containing RANKL
stimuli, NHBC can support de novo osteoclast formation and M-CSF produced large numbers of functional
from relatively mature (CD14þ PBMC and BMMC) and osteoclasts (data not shown). To investigate this pheno-
immature (CD34þ BMMC) precursors. menon further, we added FCS to SDM containing 1,25D
and DEX, and tested the ability of this medium to sup-
Effect of exogenous recombinant RANKL port osteoclastogenesis. As seen in Figure 5, culture
on co-cultures in SDM containing 0, 25, or 100 ng/ml recombinant
Our previous experiments indicated that OPG was human RANKL, in the presence of 25 ng/ml recombi-
secreted by NHBC in the range of 2–5 ng/ml over a 24 h nant human M-CSF, did not result in the formation of
period, and 10–15 ng/ml after 72 h, in sub-confluent functional osteoclasts. Consistent with our co-culture
cultures (Atkins et al., 2002). We hypothesised that the results, some (atypical) TRAPþ MNC were observed in
production of OPG, as well as other potential inhibitors, SDM cultures, however, these cells did not arise in a
could account for the observed variability in the ability RANKL-dependent fashion, nor did they resorb dentine.
of NHBC from different donors to support osteoclast Addition of FCS at either 2% v/v or 10% v/v to SDM did
formation in this system. To test the influence of endo- not induce TRAPþ MNC formation and did not give rise
genous OPG production, co-cultures were established in to functional osteoclast formation at concentrations of
SDM, as above, in which sub-optimal concentrations RANKL up to 100 ng/ml. In contrast, CD14þ PBMC
of recombinant human (rh) RANKL (1–10 ng/ml) were
added, and the resulting osteoclast formation examin-
ed. This was compared with co-cultures to which no
rhRANKL was added, or to RANKL-treated mono-
cultures of PBMC. In all cases, 25 ng/ml rhM-CSF,
10 nM 1,25D, and 10 nM dexamethasone were added
to the cultures. Recombinant human RANKL induc-
ed a dose-dependent increase in osteoclast formation in
co-cultures to a significantly greater extent than in
cultures of PBMC alone, at all concentrations tested
(Fig. 4). These results suggest that endogenous OPG is
an inhibitory influence on osteoclast formation in this
model.

Fig. 5. Stromal-free osteoclastogenesis assays of CD14þ PBMC


cultured in the presence of M-CSF (25 ng/ml) and the presence or
Fig. 4. Augmentation of osteoclast formation in co-cultures by exo- absence of rhRANKL (25 or 100 ng/ml). Cells were either cultured in
genous rhRANKL. NHBC-CD14þ PBMC co-cultures or CD14þ PBMC SDM, SDM supplemented with 2% v/v FCS (SDM-2), SDM supple-
cultured alone were performed in the presence or absence of rhRANKL mented with 10% v/v FCS (SDM-10), or a-MEM supplemented with
ranging from 0 to 10 ng/ml, all in the presence of rhM-CSF (25 ng/ml), 10% FCS (a-MEM-10). Upper part: TRAPþ multinucleate cell (MNC)
as described. Average resorption  SEM was calculated using Image- formation in wells containing cells cultured on plastic. Data shown are
quant software (Molecular Dynamics) of multiple.TIF images encom- mean  SEM of quadruplicate wells of a 96-well plate. Lower part:
passing the entire surface area of four 4 mm2 cortical bone slices per Resorption pit formation on dentine slices (mm2 per 2  106 mm2). Data
co-culture. Significance was determined using one-way ANOVA and shown are mean  SEM of four representative fields taken at 50
Tukey’s multiple comparison of non-parametric data. *P < 0.05. Simi- magnification. Similar results were obtained from two independent
lar data were reproduced in two independent experiments. experiments. *P < 0.03.
580 ATKINS ET AL.

cultured in serum-replete a-MEM under otherwise iden- 1999). The culture medium for that assay contains both
tical culture conditions gave rise to TRAPþ MNC ascorbate and DEX, potent osteoblast differentiation
capable of extensive resorption of dentine, in a strictly factors. We have shown that under conditions sup-
RANKL-dependent manner (Fig. 5). portive of human osteoclast formation, in the additional
presence of 1,25D, the expression of RANKL mRNA
DISCUSSION relative to OPG increases during the co-culture period
To date, little is known about which cells of the (Atkins et al., 2000).
osteoblast lineage are capable of supporting osteoclast For all donors’ NHBC tested, the extent of osteo-
formation. Although various reports have demonstrated clast formation in our co-culture system appeared to be
the ability of human osteosarcoma cell lines to support tightly regulated, evidenced by the highly focal nature
human osteoclast formation (Blair et al., 2000; Itoh et al., of osteoclast generation, often associated with a pre-
2000; Miyamoto et al., 2002), understanding the existing physical feature of the bone or dentine, such as
requirements for normal human osteoclast formation microvascular pores or the cut edge of the material. This
has proved elusive. Fujikawa et al. (1996) adapted the is in contrast to the widespread osteoclast formation
use of the pre-adipocytic cell line ST-2 model to sup- observed in the Fujikawa assay (Fujikawa et al., 1996),
port human osteoclast formation from PBMC. While where, in our hands, resorption events routinely exceed
our laboratory has successfully reproduced this model 500 per 105 input PBMC (data not shown), compared
system (Haynes et al., 1999; Atkins et al., 2000), our with 1–300 resorption events per 105 CD14þ PBMC in
numerous attempts to transfer those culture condi- NHBC co-cultures. The reasons for NHBC donor vari-
tions to an all-human osteoclastogenesis model, where ability and lower efficiency of osteoclast formation in the
NHBC are substituted for ST-2 cells as the stromal all-human co-culture are not yet known. Unlike ST-2
element, were unsuccessful. However, the use of a de- cells, which are clonal in origin, the NHBC cultures
fined SDM, shown previously to support the growth of represent a mixed population of cells with likely dif-
CFU-F (Gronthos and Simmons, 1995) and which has ferential potential for either osteoclast support or bone
been used in our laboratory to grow NHBC routinely, formation. Our previous study identified that 1,25D-
was tested and found to give enhanced expression of induced RANKL mRNA expression was positively asso-
RANKL relative to OPG mRNA, particularly with the ciated with an induced change in the percentage of
addition of 1,25D and DEX. SDM supplemented with cells expressing STRO-1, implying that in the hetero-
1,25D and DEX was found to promote fully functional geneous population of NHBC only a proportion of the
human osteoclast formation from CD14þ PBMC and cells would likely positively influence osteoclast forma-
BMMC, and from CD34þ BMMC, using NHBC as a tion (Atkins et al., 2003). In that study also, the ex-
stromal layer. pression of OPG mRNA did not appear to be regulated
The finding that SDM rather than serum-replete in a sub-population dependent fashion, implying that
a-MEM is permissive for osteoclast formation is in- the majority of osteoblasts potentially contribute to the
triguing. One explanation may simply be the greater active suppression of osteoclast formation under normal
induction of RANKL relative to OPG expression in conditions (Atkins et al., 2003). In the co-culture system
NHBC when exposed to 1,25D and DEX in SDM. Based described here, the endogenous production of OPG, or
on our previous study (Atkins et al., 2003), we hypo- other factors, appeared to be readily overcome by the
thesise that this property is intrinsically related to addition of sub-optimal concentrations of exogenous re-
osteoblast differentiation. NHBC are heterogeneous combinant RANKL, which were significantly less effec-
and display a range of phenotypes representing various tive in inducing osteoclast formation from cultures of
stages of osteogenic maturation (Gronthos et al., 1999; PBMC in the absence of stromal cells. We cannot rule out
Stewart et al., 1999). In our previous study, we showed that as well as overcoming the effect of endogenous OPG,
that when grown in medium containing FCS, NHBC the effect of rhRANKL in these experiments was to
differentiation followed a distinct pattern of phenotypic synergise with other co-stimulators of osteoclastogen-
change upon exposure to 1,25D and DEX, with respect to esis, such has been shown for TNF-a (Fuller et al., 2002).
the expression of the phenotypic markers STRO-1/AP In addition, we have shown that NHBC express abund-
and RANKL expression (Atkins et al., 2003). Whilst ant mRNA for the newly described osteoclast inhibitory
positive changes in the RANKL:OPG ratio were ob- lectin (OCIL) (Hu et al., 2004), and this, too, was expres-
served in that study, driven mainly by 1,25D induction sed by all sub-populations of NHBC (data not shown). It
of RANKL, it now seems that these changes are not should be noted that variability also exists in the ability
adequate to allow osteoclastogenesis to proceed. Simi- of various donors’ cells to mineralise in vitro (Findlay
larly, the phenotypic changes observed in that study et al., 2004); the possibility exists that the apparent dif-
were mild when compared to the dramatic and rapid ferences in both the bone forming and osteoclast sup-
progression to the mature (STRO-1APþ) differentia- porting abilities of NHBC reflect inherent differences in
tion state observed in a culture background of SDM, as the bone metabolism of these individuals.
shown in this study. From the studies to date, it is not To date, we have derived human osteoclast precursors
possible to conclude which is the phenotype of osteoblast from peripheral blood or BM cells in response to 1,25D
that best supports osteoclast formation, an issue made and to PTH. While the expression of RANKL and OPG
difficult to resolve because of the transient nature of in response to 1,25D was reflected in the effects of this
NHBC phenotype. However, progression through the agent on the phenotype of NHBC, the same relation-
osteoblast differentiation pathway does occur and is ships were not seen with PTH. It should be noted that
likely necessary to provide a window in which osteoclast while in all donors’ cells tested, RANKL expression
formation can proceed. Interestingly, differentiation of increased to some extent in response to 1,25D, PTH did
ST-2 cells also occurs during the osteoclast-forming co- not consistently up-regulate RANKL expression. Con-
culture. ST-2 cells represent an immature osteoblastic sistent with this, we have found that the PTH/PTHrP
cell line, and are capable of osteoblastic differentia- receptor is not always detectable in NHBC cultures by
tion under stimulation by bone morphogenetic proteins RT-PCR (unpublished data). Nevertheless, in some ins-
(BMPs) or ascorbate (Ogawa et al., 1988; Otsuka et al., tances donors’ cells did respond positively to PTH in the
HUMAN OSTEOBLAST SUPPORT OF OSTEOCLASTOGENESIS 581

osteoclast-forming assay, indicating heterogeneity be- ACKNOWLEDGMENTS


tween different donors’ cells in this respect also. Having The authors wish to thank the surgeons and nursing
developed this model, it will now be of interest to further staff of the Department of Orthopaedic Surgery and
explore the effects of other agents, including PTH, in Trauma, Royal Adelaide Hospital, for the provision of
promoting human osteoclast formation in a defined surgical bone samples.
medium.
An interesting aspect of this study was the attempt
to generate osteoclasts from PBMC precursors in a LITERATURE CITED
stromal- and serum-free system. Recombinant RANKL Anonymous. 1997. User Bulletin #2. ABI PRISM 7700 Sequence Detection
System.
in combination with M-CSF has been widely used to Atkins GJ, Haynes DR, Geary SM, Loric M, Crotti TN, Findlay DM. 2000. Co-
generate osteoclasts from PBMC in stromal-free sys- ordinated cytokine expression by stromal and hematopoietic cells during
human osteoclast formation. Bone 26(6):653–661.
tems but always in serum-replete media. We were un- Atkins GJ, Bouralexis S, Evdokiou A, Hay S, Labrinidis A, Zannettino AC,
able to generate osteoclasts from PBMC or BMMC in Haynes DR, Findlay DM. 2002. Human osteoblasts are resistant to Apo2L/
stromal-free cultures in SDM, or in two commercially TRAIL-mediated apoptosis. Bone 31(4):448–456.
Atkins GJ, Kostakis P, Pan B, Farrugia A, Gronthos S, Evdokiou A, Harrison
available serum-free media (AIM V (Invitrogen, Carls- K, Findlay DM, Zannettino AC. 2003. RANKL expression is related to the
bad, CA) and Nutridoma (Roche), data not shown), even differentiation state of human osteoblasts. J Bone Miner Res 18(6):1088–
1098.
when concentrations of up to 100 ng/ml rhRANKL were Blair HC, Sidonio RF, Friedberg RC, Khan NN, Dong SS. 2000. Proteinase
used in combination with 25 ng/ml rhM-CSF. Further- expression during differentiation of human osteoclasts in vitro. J Cell Biochem
78(4):627–637.
more, the addition of FCS to SDM did not recover the Everts V, Delaisse JM, Korper W, Jansen DC, Tigchelaar-Gutter W, Saftig P,
ability of RANKL and M-CSF to stimulate functional Beertsen W. 2002. The bone lining cell: Its role in cleaning Howship’s lacunae
osteoclast formation. TRAPþ MNC that did form in SDM and initiating bone formation. J Bone Miner Res 17(1):77–90.
Findlay DM, Welldon K, Atkins GJ, Howie DW, Zannettino AC, Bobyn D. 2004.
were atypical in appearance, did not resorb dentine, did The proliferation and phenotypic expression of human osteoblasts on tantalum
not form in a RANKL-dependent fashion, and were metal. Biomaterials 25(12):2215–2227.
Fujikawa Y, Quinn JM, Sabokbar A, McGee JO, Athanasou NA. 1996. The
reminiscent of TRAPþ macrophage polykaryons human osteoclast precursor circulates in the monocyte fraction. Endocrinology
described by Guicheux et al. (1998). The reasons why 137(9):4058–4060.
Fujikawa Y, Sabokbar A, Neale SD, Itonaga I, Torisu T, Athanasou NA. 2001. The
RANKL-mediated osteoclastogenesis is not observed in effect of macrophage-colony stimulating factor and other humoral factors
SDM are currently unclear. We can conclude, however, (interleukin-1, -3, -6, and -11, tumor necrosis factor-alpha, and granulocyte
that complete SDM used for the generation of osteo- macrophage-colony stimulating factor) on human osteoclast formation from
circulating cells. Bone 28(3):261–267.
clasts in our human co-culture contains, in addition to Fuller K, Murphy C, Kirstein B, Fox SW, Chambers TJ. 2002. TNFalpha potently
osteogenic factors, anti-osteoclastogenic factors that activates osteoclasts, through a direct action independent of and strongly
synergistic with RANKL. Endocrinology 143(3):1108–1118.
cannot be overcome by the addition of exogenous Gronthos S, Simmons PJ. 1995. The growth factor requirements of STRO-1-
RANKL/M-CSF. Also, NHBC cultured in SDM must positive human bone marrow stromal precursors under serum-deprived
conditions in vitro. Blood 85(4):929–940.
produce factors that overcome the inhibitory influence of Gronthos S, Zannettino AC, Graves SE, Ohta S, Hay SJ, Simmons PJ. 1999.
the medium on the osteoclast precursor, allowing Differential cell surface expression of the STRO-1 and alkaline phosphatase
osteoclast formation to proceed. It is likely that the antigens on discrete developmental stages in primary cultures of human bone
cells. J Bone Miner Res 14(1):47–56.
other serum-free media used in this study (AIM V and Gronthos S, Zannettino AC, Hay SJ, Shi S, Graves SE, Kortesidis A, Simmons PJ.
Nutridoma) both contain combinations of growth factors 2003. Molecular and cellular characterisation of highly purified stromal stem
cells derived from human bone marrow. J Cell Sci 116(Pt 9):1827–1835.
that divert differentiation of precursors from the Guicheux J, Heymann D, Gouin F, Pilet P, Faivre A, Daculsi G. 1998. Growth
osteoclastogenic pathway. FCS, on the other hand, hormone stimulates multinucleated cell formation in long-term bone marrow
cultures. Eur J Cell Biol 75(1):59–65.
contains factors supportive of osteoclast formation in Haynes DR, Atkins GJ, Loric M, Crotti TN, Geary SM, Findlay DM. 1999.
stromal-free systems but with the required presence of Bidirectional signaling between stromal and hemopoietic cells regulates in-
recombinant RANKL/M-CSF. Our results indicate that teleukin-1 expression during human osteoclast formation. Bone 25(3):269–
278.
FCS must also contain factors that negatively influence Hofbauer LC, Khosla S, Dunstan CR, Lacey DL, Boyle WJ, Riggs BL. 2000. The
the ability of human osteoblasts to support osteoclast roles of osteoprotegerin and osteoprotegerin ligand in the paracrine regulation
of bone resorption. J Bone Miner Res 15(1):2–12.
formation in co-culture, highlighting both the complex- Hu YS, Zhou H, Myers D, Quinn JM, Atkins GJ, Ly C, Gange C, Kartsogiannis V,
ity of this common additive to cell culture systems Elliott J, Kostakis P, Zannettino AC, Cromer B, McKinstry WJ, Findlay DM,
Gillespie MT, Ng KW. 2004. Isolation of a human homolog of osteoclast
(Leffert, 1974) and the desirability for the establishment inhibitory lectin that inhibits the formation and function of osteoclasts. J Bone
of defined serum-free assays. Elucidation of these Miner Res 19(1):89–99.
phenomena will require a more substantive breakdown Itoh K, Udagawa N, Matsuzaki K, Takami M, Amano H, Shinki T, Ueno Y,
Takahashi N, Suda T. 2000. Importance of membrane- or matrix-associated
of SDM and testing the effects of individual components forms of M-CSF and RANKL/ODF in osteoclastogenesis supported by SaOS-4/3
alone and in combination on osteoclast formation. This is cells expressing recombinant PTH/PTHrP receptors. J Bone Miner Res 15(9):
1766–1775.
the first report, to our knowledge, that shows RANKL Lacey DL, Timms E, Tan HL, Kelley MJ, Dunstan CR, Burgess T, Elliott R,
and M-CSF do not always stimulate osteoclast forma- Colombero A, Elliott G, Scully S, Hsu H, Sullivan J, Hawkins N, Davy E,
Capparelli C, Eli A, Qian YX, Kaufman S, Sarosi I, Shalhoub V, Senaldi G, Guo
tion and that other growth factors may divert precursors J, Delaney J, Boyle WJ. 1998. Osteoprotegerin ligand is a cytokine that regu-
from this pathway. lates osteoclast differentiation and activation. Cell 93(2):165–176.
In summary, we report the development of an ‘‘all Leffert HL. 1974. Growth control of differentiated fetal rat hepatocytes in pri-
mary monolayer culture. V. Occurrence in dialyzed fetal bovine serum of macro-
human’’ culture system in which fully functional human molecules having both positive and negative growth regulatory functions. J Cell
osteoclasts can be formed by co-culture of normal, Biol 62(3):767–779.
Massey HM, Flanagan AM. 1999. Human osteoclasts derive from CD14-positive
phenotypically-defined osteoclast peripheral blood- and monocytes. Br J Haematol 106(1):167–170.
BM-derived precursors with normal human trabecular Mbalaviele G, Jaiswal N, Meng A, Cheng L, Van Den Bos C, Thiede M. 1999.
Human mesenchymal stem cells promote human osteoclast differentiation
bone-derived osteoblasts as a stromal layer. Further- from CD34þ bone marrow hematopoietic progenitors. Endocrinology 140(8):
more, it is the first report of human osteoclast formation 3736–3743.
in a serum-free system. This model system provides Miyamoto N, Higuchi Y, Mori K, Ito M, Tsurudome M, Nishio M, Yamada H, Sudo
A, Kato K, Uchida A, Ito Y. 2002. Human osteosarcoma-derived cell lines
a basis for the study of both the osteoclast and its produce soluble factor(s) that induces differentiation of blood monocytes to
precursor, and the normal osteoblast that likely carries osteoclast-like cells. Int Immunopharmacol 2(1):25–38.
Neale SD, Fujikawa Y, Sabokbar A, Gundle R, Murray DW, Graves SE, Howie
out this function in vivo. We contend that the con- DW, Athanasou NA. 2000. Human bone-derived cells support formation of
tributions of both cell types need to be taken into human osteoclasts from arthroplasty-derived cells in vitro. J Bone Joint Surg
Br 82(6):892–900.
account to fully elucidate the intricacies of human bone Ogawa M, Nishikawa S, Ikuta K, Yamamura F, Naito M, Takahashi K. 1988. B
remodelling. cell ontogeny in murine embryo studied by a culture system with the monolayer
582 ATKINS ET AL.

of a stromal cell clone, ST2: B cell progenitor develops first in the embryonal Stewart K, Walsh S, Screen J, Jefferiss CM, Chainey J, Jordan GR, Beresford JN.
body rather than in the yolk sac. EMBO J 7(5):1337–1343. 1999. Further characterization of cells expressing STRO-1 in cultures of adult
Otsuka E, Yamaguchi A, Hirose S, Hagiwara H. 1999. Characterization of osteo- human bone marrow stromal cells. J Bone Miner Res 14(8):1345–1356.
blastic differentiation of stromal cell line ST2 that is induced by ascorbic acid. Taichman RS, Emerson SG. 1998. The role of osteoblasts in the hematopoietic
Am J Physiol 277(1 Pt 1):C132–C138. microenvironment. Stem Cells 16(1):7–15.
Parfitt AM. 1984. The cellular basis of bone remodeling: the quantum concept Tsukii K, Shima N, Mochizuki S, Yamaguchi K, Kinosaki M, Yano K, Shibata O,
reexamined in light of recent advances in the cell biology of bone. Calcif Tissue Udagawa N, Yasuda H, Suda T, Higashio K. 1998. Osteoclast differentiation
Int 36(Suppl 1):S37–S45. factor mediates an essential signal for bone resorption induced by 1 alpha,25-
Parfitt AM. 2000. The mechanism of coupling: A role for the vasculature. Bone dihydroxyvitamin D3, prostaglandin E2, or parathyroid hormone in the micro-
26(4):319–323. environment of bone. Biochem Biophys Res Commun 246(2):337–341.
Quinn JM, Neale S, Fujikawa Y, McGee JO, Athanasou NA. 1998. Human osteo- Udagawa N, Takahashi N, Yasuda H, Mizuno A, Itoh K, Ueno Y, Shinki T,
clast formation from blood monocytes, peritoneal macrophages, and bone Gillespie MT, Martin TJ, Higashio K, Suda T. 2000. Osteoprotegerin produced
marrow cells. Calcif Tissue Int 62(6):527–531. by osteoblasts is an important regulator in osteoclast development and func-
Siggelkow H, Rebenstorff K, Kurre W, Niedhart C, Engel I, Schulz H, Atkinson tion. Endocrinology 141(9):3478–3484.
MJ, Hufner M. 1999. Development of the osteoblast phenotype in primary Yao GQ, Sun B, Hammond EE, Spencer EN, Horowitz MC, Insogna KL, Weir EC.
human osteoblasts in culture: Comparison with rat calvarial cells in osteoblast 1998. The cell-surface form of colony-stimulating factor-1 is regulated by
differentiation. J Cell Biochem 75(1):22–35. osteotropic agents and supports formation of multinucleated osteoclast-like
Simmons PJ, Torok-Storb B. 1991. Identification of stromal cell precursors in cells. J Biol Chem 273(7):4119–4128.
human bone marrow by a novel monoclonal antibody, STRO-1. Blood 78(1):55– Zannettino AC, Buhring HJ, Niutta S, Watt SM, Benton MA, Simmons PJ. 1998.
62. The sialomucin CD164 (MGC-24v) is an adhesive glycoprotein expressed by
Sottile V, Halleux C, Bassilana F, Keller H, Seuwen K. 2002. Stem cell charac- human hematopoietic progenitors and bone marrow stromal cells that serves as
teristics of human trabecular bone-derived cells. Bone 30(5):699–704. a potent negative regulator of hematopoiesis. Blood 92(8):2613–2628.

You might also like