Particles
All metal and polymeric particles used in this study were described
previously17,18,26,28. Particles
of commercially pure titanium with a 1 to 3-mm nominal diameter
and chromium orthophosphate (CrPO4 4H2O) (mean diameter and standard
deviation, 1.42 ± 0.83 m)28 were
purchased from Johnson Matthey (Danvers, Massachusetts). Titanium-alloy
particles (6% aluminum and 4% vanadium) (Sulzer Metco, Troy, Michigan)
were ground from 150 to 300-mm-sized grinding particles29. Conventional medical-grade ultra-high
molecular weight polyethylene (GUR 415; Hoecht-Celanese, Houston,
Texas) was pulverized in liquid nitrogen29.
Particles were sedimented and then subjected to filtration. The
particles had a comparable size distribution, with at least 90%
of the particles less than 3 m in diameter. Polystyrene particles
(mean diameter and standard deviation, 1.14 ± 0.01
mm) and polystyrene-based fluorescent particles (Fluoresbrite) (mean
diameter and standard deviation, 0.926 ± 0.027
mm) were purchased from Polysciences (Warrington, Pennsylvania).
A 0.1% (volume/volume) particle suspension contained approximately 2.2
to 6.7 108 particles per milliliter.
The particles were sterilized by irradiation with 2.2 megarad (22,000
gray) from a Cs-137 source (model 143; J.L. Shepherd Irradiator,
San Fernando, California) opsonized in 10% human type-AB serum29 and stored in sterile phosphate-buffered
saline solution, pH 7.2. Endotoxin contamination of particles was
excluded by limulus assay (E-Toxate; Sigma Chemical, St. Louis,
Missouri).
Cells and Cell Cultures
The MG-63 osteoblast cell line was purchased from the American
Type Culture Collection (ATCC, Rockville, Maryland). Cells were
cultured in monolayer in Dulbecco modified Eagle medium (GIBCO,
Grand Island, New York) containing 10% fetal bovine serum (HyClone
Laboratories, Logan, Utah) in a humidified atmosphere of 5% carbon dioxide
in air at 37C17,18,26.
Treatment of Cells with Particles, Cytokines,
and Growth Factors
Confluent cultures of cells were subjected to serum starvation
(0.3% fetal bovine serum) for twenty-four hours prior to treatment.
Culture media were then replaced with fresh media consisting of
0.3% fetal bovine serum containing particles, cytokines, or growth
factors. Proliferation and viability assays, phagocytosis analysis,
and total RNA extraction were performed on the cultured cells. Tissue
culture media were collected at various time-points, centrifuged,
filtered through a 0.22-mm polycarbonate filter (Spin-x; CoStar,
Cambridge, Massachusetts), and stored at -80C. All of the experiments
were performed in duplicate or triplicate in at least five independent
experiments.
Reagents were purchased from Calbiochem (La Jolla, California)
or R and D Systems (Minneapolis, Minnesota). Insulin-like growth
factor-I (IGF-I, 30 ng/ml) and transforming growth factor-beta1 (TGF-b1,
20 pg/ml) were used to stimulate collagen production33,34. Tumor necrosis factor-a, IL-6,
and IL-1b were added to the cultures at concentrations of 10 ng/ml, 500
pg/ml, and 30 pg/ml. Prostaglandin E2 (100 ng/ml) is an eicosanoid
that has been shown to regulate collagen synthesis35. Actinomycin D (1 mg/ml) was used
to block transcriptional events, cyclohexamide (35.5 M) was used
to inhibit protein translation and synthesis, brefeldin A (0.1 M)
was used to inhibit the transport of freshly synthesized proteins
from the endoplasmatic reticulum to the Golgi apparatus, monensin
(1.5 M) was used to block the release of newly synthesized proteins
from the Golgi apparatus, and cytochalasin D (1 M) was used to destabilize
the cytoskeleton, thus inhibiting phagocytosis. All of the concentrations
listed above were selected after serial dilutions of each compound
were tested in MG-63 cell culture.
Viability Tests and [3H]-Thymidine
Incorporation
The trypan blue exclusion test was used to determine the viability
of cells. Since the presence of phagocytosed titanium particles,
especially at higher concentrations, precluded a precise evaluation
of dye exclusion, cell viability was also determined with fluorescein
diacetate (Molecular Probes, Eugene, Oregon)28.
Viability tests were performed in duplicate, and at least 200 cells
were counted with transmission or fluorescent microscopy in a Microphot-FXA microscope
(Nikon, Tokyo, Japan).
Proliferation of cells was measured by the incorporation of [3H]-thymidine
(Amersham International, Arlington Heights, Illinois) into DNA in
a ninety-six-well microplate system. Trypsinized cells were harvested
(Cell Harvester; Tomtec, Orange, Connecticut) at different time-points
after a twelve-hour [3H]-thymidine (1
C [37 Bcq] of [3H]-thymidine per well)
incubation.
RNA Extraction and Northern Blot Hybridization
Total RNA samples were isolated from monolayer cultures as described
previously17,18. Approximately
10 mg of total RNA was denatured in 50% formamide and 17.5% formaldehyde,
dissolved in MOPS buffer (20 mM 3-[N-morpholino]propane-sulfonic
acid, 5 mM sodium acetate, and 1 mM EDTA at pH 7.0), separated by electrophoresis
in 1% agarose gel, and then transferred to GeneScreen Plus membranes
(New England Nuclear, Boston, Massachusetts). Blots were hybridized
with 32P-deoxycytidine-triphosphate-labeled
specific cDNA probes at a concentration of 3 106 cpm/ml
(specific activities: 2 to 6 108 cpm/g
of cDNA)18. Human-specific cDNA
probes (plasmids) were purchased from the American Type Culture
Collection. The following recombinant plasmid DNAs were used as
probes: a 1.8-kb cDNA probe for procollagen a1[I] (Hf677; ATCC 61322),
a 2.0-kb probe for osteocalcin (ATCC 86269), a 1.8-kb probe for
osteonectin (ATCC 78193), and a 1.5-kb probe for alkaline phosphatase
(ATCC 59633). Following hybridization, the blots were washed and exposed
to Kodak XAR-5 film (Eastman Kodak, Rochester, New York) at -70C
for photographic documentation or the original Northern blot was analyzed
by STORM PhosphorImiger with ImigeQuant software (both from Molecular
Dynamics, Sunnyvale, California).
Measurement of Particle Phagocytosis
To further understand particle-induced changes in osteoblast
function, we characterized the kinetics of particle uptake by MG-63
cells. In initial experiments, cells grown in monolayers were treated
with 0.1% (volume/volume) titanium particles for various time-periods,
harvested by trypsinization, washed, and allowed to attach to coverslips. Trypsinization
and repeated washing of the cells removed nonphagocytosed and/or
surface-attached particles, and cell-free areas of the coverslip
had essentially no particles. Two hundred particle-treated cells
from different areas on the coverslip were examined with light microscopy
(Nikon). Although we could easily identify cells containing particles,
it was difficult to determine the exact number of particles in a
single cell because of the intracellular aggregation of titanium
particles.
To circumvent this problem and to be able to quantify the number
of phagocytosed particles within a single cell, we used fluorescent
particles (Fluoresbrite) as an alternative source of particles.
Two different methods were applied to determine the number of engulfed
particles in osteoblasts. First, the number of Fluoresbrite particles
in cells was counted directly in epifluorescence mode with use of
a Microphot-FXA microscope. However, this method could only be applied
to cells treated for short time-periods (less than twelve hours),
as the large number of phagocytosed particles in cells treated for
longer periods formed intracellular aggregates, precluding accurate
particle counts. The second method involved preparing serial dilutions
of Fluoresbrite particles. The particles were counted in a hemocytometer,
and the fluorescence intensity of each concentration was determined with
a fluorescent plate reader (Victor 1420 multilabel counter; Wallac,
Gaithersburg, Maryland). The particle numbers and the corresponding
fluorescence intensities were plotted to generate a standard curve,
which then was used to determine the number of fluorescent particles
(on the basis of the fluorescence intensities) in cell lysates from
osteoblasts treated with Fluoresbrite. Cells containing phagocytosed
Fluoresbrite were trypsinized, washed, counted, and lysed by ultrasonication
(VirTis, Gardina, New York) at 20 kHz for two minutes on ice.
Measurement of Cytokines and Osteoblast-Specific Proteins
in Culture Media
Cytokine concentrations in supernatants of osteoblast cultures
were measured by sandwich enzyme-linked immunosorbent assays (ELISAs)
in ninety-six-well plates. High-sensitivity assay kits for TNF-a
(range, 0.5 to 32 pg/ml), IL-1b (range, 0.12 to 8.0 pg/ml), IL-6
(range, 0.12 to 8.0 pg/ml), and TGF-b1 (range, 31 to 2000 pg/ml)
were purchased from R and D Systems. Secreted osteocalcin was measured
by NovoCalcin and type-I collagen was measured by Procollagen-C
ELISAs purchased from Metra Biosystems (Mountain View, California).
Statistical Analysis
Descriptive statistics were used to determine group means and
standard deviations. The Pillai trace (similar to Wilks lambda or
the Hotelling-Lawley trace) criterion was used to detect multivariate
significance. Subsequently, paired Student t tests were performed
between groups of interest. The level of significance was set at
p < 0.05. All statistical analyses were performed with use of
computer-based statistical software (SPSS/PC+, version 4.0.1; SPSS,
Chicago, Illinois).
Particle Phagocytosis by Osteoblasts
Osteoblasts phagocytosed particles in a time-dependent fashion
(Fig. 1-CFig.
1-C). Determining the number of phagocytosed particles, however,
is difficult because of the indiscernible location of phagocytosed,
partially engulfed, or surface-attached particles and the intracellular aggregation
of phagocytosed particles. The use of fluorescent particles (Fluoresbrite)
is a method for determining the number of engulfed particles. The mean
number of particles per cell (and standard deviation), measured
with fluorescence intensity in MG-63 cell lysates, was 0.4 ± 0.6 (range, zero to two) after one hour, 5 ± 2 (range, three to twelve) after two hours, 13 ± 3 (range, seven to nineteen) after six hours,
23 ± 2 (range, sixteen to twenty-eight) after
twelve hours, 53 ± 5 after twenty-four hours,
60 ± 6 after forty-eight hours, and 58 ± 5 after seventy-two hours (Fig. 1-CFig. 1-C).
A maximum of forty to sixty Fluoresbrite particles phagocytosed
within twenty-four hours seems to be the saturation level for MG-63
cells. The number of engulfed particles could be precisely determined
only when fluorescent-labeled particles were used. Cytochalasin
D, which destabilizes the cytoskeleton and inhibits phagocytosis,
significantly reduced (p < 0.01) the phagocytosis of particles
(4 ± 3 particles per cell after forty-eight
hours) but did not completely abolish it (Figs. 1-AFigs. 1-A,
panel A2, and 1-B1-B,
panel B2).
Effect of Particulate Wear Debris on Cell Viability and
Proliferation
Next, we addressed how the phagocytosed particles influenced
cell functions and whether different compositions of particles could
initiate different cell responses. Particulate wear debris had no
effect on the viability of MG-63 cells, which remained higher than
95% even in long-term experiments (seventy-two to ninety-six hours)
over a wide range of particle compositions (titanium, titanium alloy, chromium
orthophosphate, polyethylene, polystyrene, and Fluoresbrite) and
concentrations (0.0125% to 0.2% volume/volume). In contrast to viability,
cell proliferation was reduced when compared with that in untreated
cultures, and the effect was dose-dependent (Fig. 2Fig. 2, A).
Interestingly, suppressed proliferation returned to normal by forty-eight
hours at low particle concentrations (0.0125% to 0.05%) (Fig. 2Fig. 2, A).
Effect of Exogenous Cytokines on Cell Viability and
Proliferation
Exogenous cytokines (IL-1b and IL-6), prostaglandin E2, and growth
factors TGF-b1 and IGF-I had no effect on cell viability over a
wide range of concentrations. Only TNF-a significantly reduced cell viability
when higher concentrations (more than 100 ng/ml) were used (p < 0.01),
especially in long-term experiments (more than seventy-two hours). As
shown in Figure 2Figure
2, B, TNF-a (at a nontoxic concentration of 10
ng/ml) and prostaglandin E2 decreased cell proliferation, while
TGF-b1 and IGF-I increased it. Neither IL-b1 nor IL-6 affected cell
proliferation.
Cytokine Release in Osteoblasts Induced by
Either Titanium Particles or Exogenous Cytokines
The earliest cytokine release in culture media of titanium-stimulated
MG-63 cells was detected after twelve hours, and it was restricted
to IL-6 (Fig. 3Fig.
3, A). There were undetectable amounts of IL-1
and TNF-a in culture media of either untreated or particle-challenged
MG-63 cells. A basal TGF-b1 secretion was enhanced in particle-treated
cultures after seventy-two hours (Fig. 3Fig. 3, B).
As found in titanium-stimulated MG-63 cultures, only IL-6 and
TGF-b1 secretion was modified by exogenous mediators. Only TNF-a
at a nontoxic concentration (10 ng/ml) had a significant effect
on IL-6 release (p < 0.01) (Fig. 4Fig. 4, A), whereas
the basal TGF-b1 secretion was increased (p < 0.05) by exogenous
TNF-a, IGF-I, or prostaglandin E2 (Fig. 4Fig. 4, B).
Suppression of Osteoblast-Specific Gene Expression
We reported a 40% to 60% suppression of procollagen a1[I] mRNA
expression in MG-63 osteoblasts exposed to titanium particles17,25. To further characterize the
effect of particles on osteoblast-specific gene expression and protein synthesis,
MG-63 cells were exposed to titanium, titanium-alloy, chromium-orthophosphate,
polystyrene, polyethylene, and Fluoresbrite particles for forty-eight
hours. Particles, regardless of composition, significantly suppressed
procollagen a1[I] mRNA expression in MG-63 osteoblasts (p < 0.05) (Fig. 5Fig. 5). This
downregulation of collagen gene expression was accompanied by reduced
type-I collagen protein synthesis. In contrast to procollagen gene expression,
none of the particles significantly altered the expression of osteocalcin
or other osteoblast-specific genes such as osteonectin or alkaline phosphatase.
These data demonstrate that particles differentially affect gene
expression in osteoblasts (for example, procollagen a1[I] compared
with osteocalcin) (Fig. 5Fig. 5), and the gene-specific effect
is a general response to particles and is not specific to particles
of a particular composition.
Effect of Exogenous Cytokines on Procollagen a1[I]
Gene Expression
An increased IL-6 secretion in titanium-stimulated osteoblasts
(Fig. 3Fig.
3, A) correlated inversely with the suppression
of procollagen a1[I] mRNA and reduced collagen synthesis in all
particulate-stimulated osteoblast cultures. Neither exogenous IL-6
(Fig. 6Fig.
6, A) nor neutralizing antibodies to IL-6 (data
not shown) altered the collagen gene expression in the presence
or absence of titanium particles, indicating that procollagen a1[I]
gene regulation was independent of IL-6 or IL-6-induced transcription factors
in osteoblasts. In contrast, the effect of exogenous TNF-a on procollagen
a1[I] gene expression was highly comparable with the effect of titanium
particles (Fig. 6Fig.
6, A), and this correlated with reduced type-I
collagen synthesis25. Prostaglandin
E2 also inhibited procollagen a1[I] gene expression (data not shown).
Exogenous IL-1b significantly reversed titanium-induced procollagen
a1[I] gene suppression (p < 0.05) (Fig. 6Fig. 6, A). Growth
factors IGF-I and TGF-b1 significantly increased collagen gene expression
(p < 0.05) and could completely reverse the titanium-induced
suppression of procollagen a1[I] mRNA (Fig. 6Fig. 6, B).
Effect of Transcriptional, Translational, and
Protein Transport Inhibitors on Cytokine Release and Procollagen
a1[I] Gene Expression
To determine whether an effect of a freshly synthesized cytokine
was involved in the titanium-induced procollagen a1[I] gene suppression
and to distinguish the mechanism of titanium-induced IL-6 production
from procollagen a1[I] gene suppression at the molecular level,
MG-63 cells were treated with various inhibitors prior to stimulation with
titanium particles. Actinomycin D, a potent inhibitor of transcriptional
events by inhibiting RNA polymerase II, was used as a positive control. This
compound blocked both procollagen a1[I] mRNA transcription and IL-6
production. Translational (protein-synthesis) inhibitor cyclohexamide, protein-transport
inhibitor brefeldin A, and monensin (a nonselective inhibitor of
the release of newly synthesized protein from the Golgi apparatus)
uniformly blocked the release of IL-6. In contrast, none of these
chemicals modified titanium-particle-induced suppression of the
procollagen a1[I] mRNA level (Fig. 7Fig. 7), confirming our recent observation
that particle phagocytosis has a direct effect on procollagen gene
expression through the activation of the protein tyrosine kinase-NF-kB
pathway25. Taken together, particle
phagocytosis has a direct effect on procollagen a1[I] mRNA, whereas
particle-induced cytokine release requires factors for protein synthesis
and intracellular trafficking.
Particulate wear debris from prosthetic components is continuously
generated and phagocytosed by cells of the periprosthetic soft tissue.
Phagocytosis is a strong signal for cells, first inducing a series
of upstream events of cell stimulation through the activation of
protein tyrosine kinases19,23,25,36,37.
The activation of protein tyrosine kinases leads to the activation
of nuclear transcription factors. These nuclear transcription factors
then are translocated into the nucleus, resulting in the upregulation of
various genes, including proinflammatory cytokines. In a broader
sense, all cell types of the periprosthetic soft tissue (macrophages,
fibroblasts, osteoclasts, and osteoblasts) are able to phagocytose
particulate wear debris, and virtually all cells can reach an activated
state. These cells produce a number of cytokines, chemokines, and
prostaglandins, which may further affect the function of cells in
either an autocrine or a paracrine manner with use of distinct signaling
mechanisms.
Bone is a dynamic tissue with a well-balanced homeostasis preserved
by both formation and resorption of bone. Normal turnover of bone,
however, can be unbalanced by either increased osteoclast activity
or decreased osteoblast function; either mechanism or both mechanisms
may result in a net loss of bone. Both osteoclasts and osteoblasts
phagocytose particles in vitro, and it is assumed
that this process may occur in vivo as well. Osteoblasts,
which phagocytose particles, become activated and produce IL-625 and prostaglandin E231, simultaneously losing their capacity
to synthesize type-I collagen17,25.
The secreted IL-638-42 and prostaglandin
E238,43 then activate osteoclasts
in a paracrine fashion, which are assumed to already be in an activated state
because of phagocytosed particles at the interface16,44. Other cytokines such as IL-1b
and TNF-a, secreted by particle-stimulated macrophages/monocytes6,11,15,24,26-30, are also present
in the periprosthetic tissue. Tumor necrosis factor-a can activate
osteoblasts to secrete IL-6 and suppress type-I collagen synthesis25,45,46, an effect similar to that
described for particle phagocytosis (Fig. 3Figs. 3, A;Fig. 44, A; and Fig. 66, A).
In addition, both IL-1b and TNF-a induce osteoclast differentiation
from precursors and activate differentiated osteoclasts in
vitro38,42,43. Taken
together, phagocytosis directly affects and phagocytosis-induced
cytokine release indirectly affects the bone turnover negatively
by altering osteoblast and osteoclast functions.
From the osteoblast side, a phagocytosis-induced direct signal
and exogenous TNF-a (paracrine effect) seem to be the most potent
inducers of diminished type-I collagen synthesis. However, the particle-induced
and TNF-a-induced signaling mechanisms must be independent because
(1) we were unable to detect TNF-a in either particle-treated MG-63
(Fig. 3Fig.
3) or bone-marrow-derived primary human osteoblast cultures25, (2) neutralizing anti-TNF-a antibody
could abolish the exogenous TNF-a effect but did not modify the
osteoblast response to titanium particles (data not shown), and
(3) protein synthesis inhibitors, while blocking cytokine release,
had no influence on particle-induced procollagen a1[I] gene suppression
(Fig. 7Fig.
7).
While neither exogenous nor endogenous IL-6 can affect osteoblast-specific
functions, IL-6 along with the IL-6 soluble receptor could enhance
different osteoblast responses47-49.
The IL-6 receptor complex consists of two transmembrane proteins:
a ligand-binding chain (IL-6 receptor) and a non-ligand-binding
signal transducer, glycoprotein 130 (gp130). Interleukin-6 binding
to the ligand-binding chain triggers heterodimerization of the two
chains, and then the cytoplasmic domain of the gp130 chain transduces the
signal. The soluble form of the ligand-binding chain (soluble IL-6
receptor) is also able to activate gp130 when IL-6 binds to it.
It is likely that MG-63 osteoblasts express gp130 but not the ligand-binding
chain47,49. Therefore, neither
secreted nor exogenous IL-6 can bind to the IL-6 receptor. As a
result, no signal can be transferred from the cell surface to the
cell41,47-50.
Among a number of cytokines and growth factors, IGF-I and TGF-b1
were able to completely reverse the suppressive effect of particles
on procollagen a1[I] gene expression. These growth factors, when used
alone, significantly upregulated the procollagen a1[I] gene expression
(Fig. 6Fig.
6, B) and type-I collagen synthesis. Furthermore,
these growth factors increased osteoblast proliferation without
affecting cell viability or inducing substantial IL-6 secretion.
Thus, IGF-I and TGF-b1 seem to be potent inducers of bone matrix
formation.
One of the most important findings of the present study is that,
in addition to a direct effect of particles on osteoblast functions17,25, the proinflammatory cytokine
TNF-a also exhibits a massive and substantial effect on procollagen a1[I]
gene expression, cell proliferation, cell viability, and IL-6 secretion
in osteoblasts. While this proinflammatory cytokine induces bone
resorption via osteoclast activation, it also contributes to bone loss
via reduced bone formation by osteoblasts. Since TNF-a, IL-1b, and
IL-6 are present and are continuously secreted by particle-stimulated
cells in the periprosthetic space, their long-term in vivo autocrine
and paracrine effects are critical in the pathogenesis of the periprosthetic
osteolysis. Eventually, local delivery of certain growth factors
(IGF-I or TGF-b1), protein tyrosine kinase, or NF-kB inhibitors25-all of which can reverse the suppressive
effect of either proinflammatory cytokines or wear particles on
type-I collagen synthesis in osteoblasts-may have the therapeutic
potential to prevent or treat periprosthetic bone loss.
Note: The authors thank József Dobai, MD, Ramesh Narayanan, PhD,
David Gerard, BA, and Sonja Velins, MLA, for their technical help.
We also appreciate our valuable discussions with Rick Sumner, PhD,
Tom Turner, DVM, and a number of visitors to the Section of Biochemistry
and Molecular Biology of the Department of Orthopedic Surgery at
Rush University (Chicago, Illinois).