Periprosthetic osteolysis remains the major complication in patients
managed with total hip replacement, frequently resulting in implant loosening
and the need for revision
surgery1-3.
While it is generally accepted that wear-generated particle debris is a
critical initiating factor in the majority of cases of osteolysis, the full
repertoire of biologic effects of these biomaterials remains to be described.
For instance, although it has been well established that wear debris particles
can induce pro-inflammatory responses in
macrophages4-7,
it is not known whether this effect is accompanied by reprogramming of other
cytokines within the periprosthetic interface that may be employed in
anti-inflammatory or anti-osteoclastogenic activities. Given the complex
cytokine milieu around loosened joint
prostheses8,9,
this latter possibility warrants careful consideration. In light of this
possibility, we have begun to analyze how wear particles might be involved in
the modulation of signaling by interleukin-6 (IL-6) and interferon-?
(IFN-?), two cytokines with demonstrated anti-osteoclastogenic
activities.
IFN-?, in addition to its well-appreciated pro-inflammatory
properties, recently has been shown to be a potent anti-osteoclastogenic
factor10 that might
protect against osteoclastogenesis and subsequent bone loss at the bone
interface in patients with osteolysis. Likewise, IL-6 can be elevated in
periprosthetic
tissue8,9,
and maybe
systemically8,11,
in patients with osteolysis. IL-6 has complex effects, both pro-resorptive and
anti-resorptive12,
and recently has been shown to suppress the differentiation of osteoclast
precursors13. The
anti-osteoclastogenic nature of both IFN-? and IL-6 can be demonstrated
in vitro, where they suppress the RANKL (receptor activator of nuclear factor
?B ligand)-induced formation of osteoclasts from cultured
macrophages13. IL-6
and IFN-? share the ability to signal via the JAK/STAT (Janus
kinase/signal transducer and activator of transcription) pathway, with IL-6
predominantly acting via STAT3 and IFN-? acting via STAT1. Inhibition of
JAK/STAT signaling has been shown to be mediated by several mechanisms,
including activation of mitogen-activated protein (MAP)
kinases14,15
and induction of the SOCS (suppressors of cytokine signaling) family of
cytokine
suppressors16. In
particular, SOCS1 and SOCS3 appear to be of central importance in regulating
the suppression of osteoclast formation by IFN-? and IL-6,
respectively13.
In the present study, we present evidence that two types of wear debris,
polymethylmethacrylate bone cement and titanium particles, can, in addition to
inducing pro-inflammatory mediators, downregulate the response of osteoclast
precursor cells to IL-6 and IFN-?. Our data implicate both MAP kinases
and SOCS induction in the ability of wear debris particles to reprogram
cellular responses to these cytokines and suggest that one mechanism of action
of wear debris in periprosthetic osteolysis may be through the inhibition of
anti-osteoclastogenic cytokine signaling.
Reagents
RPMI 1640 medium (Invitrogen Gibco, Grand Island, New York) was
supplemented with 1% penicillin-streptomycin, 1% GlutaMAX (Invitrogen Gibco),
and 10% heat-inactivated endotoxin-free Fetal Bovine Serum (HyClone, Logan,
Utah). Ficoll metrizoate (Lymphoprep) was obtained from Axis-Shield PoC AS
(Oslo, Norway). Recombinant human M-CSF (macrophage colony-stimulating factor)
and IL-6 were obtained from R and D Systems (Minneapolis, Minnesota), and
IFN-? was obtained from Roche (Indianapolis, Indiana). Human sRANKL was
obtained from PeproTech (Rocky Hill, New Jersey). Zymosan A bioparticles were
obtained from Molecular Probes (Eugene, Oregon). All primary and secondary
antibodies were obtained from Cell Signaling Technology (Beverly,
Massachusetts). The MAP kinase inhibitors SB203580, PD98059, U0126, and
SP600125 were obtained from EMD Biosciences (La Jolla, California).
Electrophoresis reagents were obtained from Bio-Rad Laboratories (Hercules,
California), and ECL/ECL Plus reagents were obtained from Amersham Biosciences
(Piscataway, New Jersey). A staining kit for tartrate-resistant acid
phosphatase (TRAP) was obtained from Sigma-Aldrich (St. Louis, Missouri) and
was used in accordance with the manufacturer's recommendations. All other
chemicals were from Sigma-Aldrich.
Preparation of Wear Particles
Commercially pure titanium and polymethylmethacrylate particles were
obtained from Johnson Matthey (catalog #00681; Alfa-Aesar, a Johnson Matthey
Company, Ward Hill, Massachusetts) and Polysciences (catalog #19130,
Warrington, Pennsylvania), respectively. According to the manufacturer, the
average diameter of the polymethylmethacrylate particles was 6.06 µm.
Histologic analysis revealed that 75% of the polymethylmethacrylate particles
were <5 µm in diameter and that 86% of the titanium particles were
<10 µm in diameter. Polymethylmethacrylate particles were sterilized by
treatment with 70% ethanol for forty-eight hours. Titanium particles were
sterilized by baking at 180°C for six hours, followed by treatment with
70% ethanol for forty-eight hours. Sterilized particles were suspended in
complete RPMI medium. Only endotoxin-free particles as determined by means of
Limulus assay (E-TOXATE; Sigma-Aldrich) were used in the present study. Such
commercially available particles have been shown to effectively mimic wear
particles retrieved from periprosthetic tissue when used in cell-culture
experiments in
vitro5,6.
Isolation of Osteoclast Precursor Cells and Cell-Particle
Coculture
Peripheral blood mononuclear cells were isolated by means of density
gradient centrifugation with Lymphoprep from buffy coats isolated from healthy
volunteers17. CD14+
myeloid cells were further purified from peripheral blood mononuclear cells by
means of positive selection with use of anti-CD14 magnetic beads as
recommended by the manufacturer (Miltenyi Biotec, Auburn, California) and were
shown to be >98% pure as determined by means of flow cytometry, with CD3+ T
cells and CD19+ B cells each representing <1% of the final cell population.
The cells were cultured overnight in sterile Petri dishes in the presence of
M-CSF (10 ng/mL), harvested, and recultured overnight in fresh RPMI with M-CSF
in six-well plates (4 × 106 cells per well) prior to
treatment with zymosan, titanium, or polymethylmethacrylate particles.
Consistent with previous
reports18-20,
these M-CSF-differentiated monocyte-derived cells are referred to as
osteoclast precursor cells by virtue of their ability to differentiate into
osteoclasts when cultured with
RANKL21. For some
experiments (see Figure 6),
these cells were further differentiated with 25-ng/mL human M-CSF and
100-ng/mL human sRANKL for forty-eight hours (to form "late osteoclast
precursors") or fourteen days (to form mature, multinucleated,
TRAP-positive osteoclasts). Zymosan was used at a cell-to-particle ratio of
1:15. For polymethylmethacrylate and titanium, a cell-to-particle ratio of
1:30 was used throughout. This ratio was chosen because it supported optimal
cellular response without significant toxicity (unpublished data).
For determination of the effects of wear particles on cytokine signaling,
cells were treated with either recombinant human IL-6 (25 ng/mL) or
IFN-? (1.5 U/mL) for twelve minutes after culture with or without
particles (to test for STAT activation) or with IFN-? (1.5 U/mL) for one
hour after culture with or without titanium particles (to test for expression
of IFN-?-inducible genes). This low dose of IFN-? (1.5 U/mL) is
sufficient for activation of STAT1 in CD14+
cells17 but is
insufficient to substantially activate STAT1 in the small proportion of
contaminating T cells (<1%, as noted above) present in the purified CD14+
cell population22.
To determine the role of MAP kinases in cellular responses to wear particles,
cells were pretreated with either specific inhibitors of p38 (SB203580, 10
µM), MEK (U0126, 20 µM; PD98059, 10 µM), and JNK (SP600125, 20 µM)
MAP kinases, either individually or in combination, for forty-five minutes
before challenge with particles and/or cytokines. These concentrations and
incubation times were shown to confer specific and complete inhibition of the
respective MAP kinases (unpublished data).
RNA Extraction Real-Time Quantitative Reverse
Transcription-Polymerase Chain Reaction
RNA was extracted (RNeasy Mini Kit; QIAGEN, Valencia, California) and
assessed for concentration and purity by means of optical density measurement.
One-microgram aliquots of total cellular RNA were reversely transcribed with
use of random hexamers and Superscript II Reverse Transcriptase/RNase Out
recombinant RNase inhibitor (Invitrogen Gibco) as recommended by the
manufacturer. Real-time quantitative polymerase chain reaction was carried out
in triplicate with use of the iCycler iQ thermal cycler detection system
(Bio-Rad Laboratories). Reactions included iQ SYBR Green Supermix reagent
(Bio-Rad Laboratories), 10-ng cDNA, and forward and reverse primers, each at a
concentration of 250 nM (for human TNF-a, IL-6, or HPRT [hypoxanthine
phosphoribosyltransferase]) or 500 nM (for human SOCS1, SOCS3, MIG, or IP-10),
in a total volume of 25 µL. mRNA amounts were normalized relative to the
housekeeping gene HPRT. Generation of only the correct amplification products
was confirmed with use of melting-point curve analysis of the products.
Results were representative of at least three independent experiments. The
sequences of the oligonucleotide primers used were as follows.
TNF-a sense: 5'-ACACCATCAGCCGCATCG-3'TNF-a anti-sense: 5'-AGTCGGTCACCCTTCTCC-3'HPRT sense: 5'-TGAGGATTTGGAAAGGGTG-3'HPRT anti-sense: 5'-GAGGGCTACAATGTGATGG-3'IL-6 sense: 5'-TGCTCCTGGTGTTGCCTGCT-3'IL-6 anti-sense: 5'-AGCCACTGGTTCTGTGCCTGC-3'SOCS3 sense: 5'-CACTCTTCAGCATCTCTGTCGGAAG-3'SOCS3 anti-sense: 5'-CATAGGAGTCCAGGTGGCCGTTGAC-3'SOCS1 sense: 5'-AGAGGTAGGAGGTGCCAGT-3'SOCS1 anti-sense: 5'-TGTTGTAGCAGCTTAACTGTATC-3'MIG sense: 5'-TTGGGCATCATCTTGCTGGTTCT-3'MIG anti-sense: 5'-TGGCTGACCTGTTTCTCCCACTT-3'IP-10 sense: 5'-TTGCTGCCTTATCTTTCTGACTC-3'IP-10 anti-sense: 5'-ATGGCCTTCGATTCTGGATT-3'
TNF-a sense: 5'-ACACCATCAGCCGCATCG-3'
TNF-a anti-sense: 5'-AGTCGGTCACCCTTCTCC-3'
HPRT sense: 5'-TGAGGATTTGGAAAGGGTG-3'
HPRT anti-sense: 5'-GAGGGCTACAATGTGATGG-3'
IL-6 sense: 5'-TGCTCCTGGTGTTGCCTGCT-3'
IL-6 anti-sense: 5'-AGCCACTGGTTCTGTGCCTGC-3'
SOCS3 sense: 5'-CACTCTTCAGCATCTCTGTCGGAAG-3'
SOCS3 anti-sense: 5'-CATAGGAGTCCAGGTGGCCGTTGAC-3'
SOCS1 sense: 5'-AGAGGTAGGAGGTGCCAGT-3'
SOCS1 anti-sense: 5'-TGTTGTAGCAGCTTAACTGTATC-3'
MIG sense: 5'-TTGGGCATCATCTTGCTGGTTCT-3'
MIG anti-sense: 5'-TGGCTGACCTGTTTCTCCCACTT-3'
IP-10 sense: 5'-TTGCTGCCTTATCTTTCTGACTC-3'
IP-10 anti-sense: 5'-ATGGCCTTCGATTCTGGATT-3'
Electrophoretic Mobility Shift Assay
Whole-cell extracts corresponding to 4 × 106 cells were
prepared by means of cell lysis in buffer containing 20-mM HEPES (pH 7.0),
300-mM NaCl, 10-mM KCl, 1-mM MgCl2, 0.1% Triton X-100, 0.5-mM DTT
(dithiotreitol), 200-mM phenylmethylsulfonyl fluoride (PMSF), 20% glycerol,
Pefablock SC (50 ng/mL), and 1X protease-inhibitor cocktail (Roche). Total
protein concentrations were determined with use of Bradford assays (Bio-Rad
Laboratories). For preparation of nuclear extracts, cells were washed with
cold phosphate-buffered saline solution, suspended in hypotonic lysis buffer
(10-mM HEPES, 10-mM KCl, 1.5-mM MgCl2, 0.5-mM DTT, 1X
protease-inhibitor cocktail, 1-mM Pefablock), and incubated at 4°C for
fifteen minutes, after which NP-40 was added to a final concentration of
0.64%. Nuclei were pelleted by centrifugation at 14,000 rpm for thirty seconds
at 4°C and were resuspended in nuclear extraction buffer (20-mM HEPES [pH
7.8], 420-mM NaCl, 1.2-mM MgCl2, 0.2-mM EDTA, 25% glycerol, 0.5-mM
DTT, 0.1% Triton X-100, 200-µM PMSF). The samples were rotated horizontally
at 4°C for thirty minutes, centrifuged at 14,000 rpm at 4°C for five
minutes, and the supernatants were transferred to fresh microfuge tubes. The
nuclear protein concentration was determined with use of Bradford assays.
Ten micrograms of total cell or nuclear protein extracts were incubated
with 0.5 ng of 32P-labeled double-stranded high-affinity
oligonucleotide probe (hSIE: 5'-GTCGACATTTCCCGTAAATCGTCGA-3') for
fifteen minutes at room temperature in a 15-µL binding reaction containing
40-mM NaCl and 2 µg of Poly (dl-dC) (Amersham Biosciences) as
described17, and
the complexes were resolved on nondenaturing 5% polyacrylamide gels, followed
by autoradiography to visualize the bands. Results were representative of at
least three independent experiments.
Immunoblotting
Total cell lysates for immunoblotting were prepared by cell lysis with use
of sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample
buffer (150 µL per 4 × 106 cells), followed by boiling for
five minutes. Lysates were cleared by centrifugation (14,000 rpm), and
15-µL aliquots of protein extracts (equivalent to approximately 30 µg of
total cell protein) were fractionated on 10% SDS-PAGE gels and transferred to
polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica,
Massachusetts). Membranes were air-dried overnight and were incubated for one
hour at room temperature in blocking solution (5% skim milk prepared in
phosphate-buffered saline solution containing 0.1% Tween-20) to prevent
nonspecific binding. After washing with phosphate-buffered saline
solution/Tween-20 buffer (three times, for fifteen minutes each, at room
temperature), membranes were incubated overnight at 4°C with primary
antibodies, washed with phosphate-buffered saline solution/Tween-20 buffer (as
described above), incubated for one hour at room temperature with horseradish
peroxidase-conjugated secondary antibodies, and finally rewashed with
phosphate-buffered saline solution/Tween-20 buffer (as described above). The
antibody dilutions used were 1:1000 (for antibodies to
phospho-[Thr180/Tyr182]-p38, phospho-[Thr202/Tyr204]-ERK,
phospho-[Tyr701]-STAT1, phospho-[Tyr705]-STAT3, p38, ERK, STAT1, and STAT3),
1:1500 (for anti-phospho-[Thr183/Tyr185]-JNK), and 1:3000 for all secondary
antibodies. Detection was performed with ECL (for p38, ERK, JNK, STAT1, and
STAT3) and ECL Plus (for the phosphorylated forms of these proteins), in
accordance with the manufacturer's directions.
Enzyme-Linked Immunosorbent Assay
The concentrations of TNF-a and IL-6 in supernatants from osteoclast
precursor cultures were determined with use of enzyme-linked immunosorbent
assay kits specific for human TNF-a and IL-6 (BD Pharmingen, San Jose,
California).
Activation of Multiple MAP Kinase Pathways Mediates
Polymethylmethacrylate and Titanium Induction of Cytokine Expression in Human
Osteoclast Precursor Cells
Human osteoclast precursors, generated by culturing CD14+ monocytes in
M-CSF over two nights, were used throughout (see Materials and Methods
section). These conditions have been shown to provide a stable, relatively
quiescent cell
population23. As
expected, treatment with both polymethylmethacrylate and titanium particles
resulted in induction of pro-inflammatory TNF-a mRNA
(Fig. 1, A) and
protein (Fig. 1, C).
TNF-a mRNA was induced within one hour and secreted protein by six
hours. The magnitude of induction of TNF-a was comparable between
polymethylmethacrylate and titanium and between wear debris and zymosan (a
particulate preparation of yeast cell walls that is used as a model
inflammatory macrophage activator and that was used as a positive control in
these experiments). In addition, IL-6 mRNA
(Fig. 1, B) and
protein (Fig. 1, D)
were induced by polymethylmethacrylate, titanium, and zymosan. Interestingly,
polymethylmethacrylate repeatedly induced higher levels of IL-6 protein than
titanium did, whereas the IL-6 mRNA levels induced by these two types of
particles were similar. The mechanisms behind this difference are unknown but
may involve differential post-transcriptional regulation of IL-6 expression in
response to different wear particles.
The possible involvement of MAP kinase signaling in the induction of
TNF-a expression was assessed in two ways. First, activation of the
three MAP kinase families (p38, ERK, and JNK) was assessed in human osteoclast
precursors after treatment with polymethylmethacrylate and titanium wear
debris. Immunoblotting with antisera against the phosphorylated (active) forms
of these kinases revealed that p38, ERK, and JNK kinases were all activated
within fifteen minutes by titanium and within thirty to forty-five minutes by
polymethylmethacrylate (Fig.
2). Activated MAP kinases were generally undetectable in untreated
cells. Blotting with control antisera against phosphorylated and
nonphosphorylated forms of each MAP kinase confirmed that the abundance of
p38, ERK, and JNK remained fairly constant throughout the experiment. Second,
cells were pre-incubated with inhibitors specific for p38 (SB203580), MEK/ERK
(U0126), and JNK (SP600125) prior to particle challenge. Consistent with the
activation of all three MAP kinase pathways by each type of particle, all
three inhibitors reduced the ability of polymethylmethacrylate and titanium to
induce the expression of TNF-a mRNA
(Fig. 3). Moreover, paired
combinations of these inhibitors further reduced TNF-a and IL-6 mRNA
levels, and pre-incubation with all three virtually abolished the ability of
wear debris to induce TNF-a mRNA expression in the human osteoclast
precursor cells. Cell viability was preserved during the time-course of these
experiments. These results show that wear debris particles activate all three
MAP kinase pathways and that these three pathways contribute to inflammatory
cytokine production in an additive manner.
Polymethylmethacrylate and Titanium Inhibition of IL-6 Induced STAT3
Activation
To assess the ability of polymethylmethacrylate and titanium to modulate
IL-6 signaling, STAT3 activation was determined in cells that were stimulated
with IL-6 with and without particle pretreatment. With use of electrophoretic
mobility shift assays (EMSA), IL-6 alone showed robust STAT3 activation, as
expected (Fig. 4, A).
However, pretreatment with polymethylmethacrylate or, to a lesser extent,
titanium reduced this activation. Total STAT3 levels were unchanged by any of
the treatments. Inhibition of IL-6 signaling by polymethylmethacrylate and
titanium was confirmed by immunoblotting with antisera against the tyrosine
phosphorylated (active) form of STAT3. In agreement with the results of the
electrophoretic mobility shift assay, STAT3 activation was induced by IL-6 and
was inhibited by pretreatment with either polymethylmethacrylate or titanium
(Fig. 4, B, top panel,
lanes 3 and 8 compared with lane 2).
Involvement of MAP Kinases in Wear Debris Inhibition of IL-6
Signaling
One described mechanism for the downregulation of JAK/STAT signaling is
through the activation of the p38 MAP kinase
pathways15. In
addition, as detailed above, wear debris particles activate multiple MAP
kinases. We therefore sought to determine the involvement of MAP kinase
activation in particle inhibition of cytokine signaling. As shown in
Figure 4, B,
inhibition of p38 substantially reduced the ability of polymethylmethacrylate
to inhibit IL-6 signaling (top panel, lane 3 compared with 4), whereas
inhibition of ERK or JNK had no discernible effect, as assessed by
immunoblotting for phosphorylated STAT3. A similar pattern was observed for
titanium inhibition of IL-6 signaling, with substantial reversal of inhibition
only seen with inhibition of p38 alone or of all three MAP kinase families.
The predominant role for p38 in polymethylmethacrylate inhibition of IL-6
signaling was confirmed by electrophoretic mobility shift assay analysis
(Fig. 4, B, bottom).
Thus, in contrast to inflammatory cytokine production, which was mediated by
all three MAP kinase pathways, inhibition of IL-6 signaling was mediated
primarily by p38.
Titanium, But Not Polymethylmethacrylate, Inhibited
IFN-?-Induced STAT1 Activation and Gene Expression
Electrophoretic mobility shift assays were similarly used to assess STAT1
activation by IFN-? with or without pretreatment of osteoclast
precursors with wear debris. Interestingly, the two types of wear debris under
investigation had very different effects on IFN-? signaling, with
titanium pretreatment abolishing STAT1 activation but with
polymethylmethacrylate having no effect
(Fig. 5, A). Total
STAT1 levels were unaffected by any treatment. Immunoblotting with antisera
against phosphorylated STAT1 confirmed this finding, with titanium (but not
polymethylmethacrylate) inhibiting the IFN-?-induced activation of STAT1
(Fig. 5, B). Titanium
was also able to downregulate IFN-?-induced expression of mRNAs encoding
the STAT1-dependent genes MIG and IP-10, demonstrating the functional
consequences associated with the inhibition of IFN-? signaling (data not
shown). In contrast to wear debris inhibition of IL-6 signaling
(Fig. 4, B), no
evidence was found for involvement of MAP kinases in titanium inhibition of
IFN-?-signaling, with inhibition of p38, JNK, and ERK, either
individually or together, having no significant impact on the inhibitory
effects of titanium (Fig. 5,
C). These results indicate that titanium-mediated
inhibition of IFN-? signaling is independent of MAP kinases and suggest
that the mechanisms by which IL-6 and IFN-? signaling are inhibited
differ.
Wear Debris Inhibition of Cytokine Signaling Persists in Late
Osteoclast Precursors
It has become conventional to categorize M-CSF-dependent hematopoietic
cells, such as those described here, as osteoclast precursor cells (or
pre-osteoclasts)18-20.
This convention is centered on the observation that the addition of RANKL is
sufficient to differentiate these cells to mature
osteoclasts21. To
verify that our cells were osteoclast precursors, they were incubated with
RANKL and M-CSF (or M-CSF alone) for fourteen days and then were stained for
the osteoclast marker TRAP. As shown in
Figure 6, A, abundant
multinuclear, TRAP-positive cells were generated in the presence, but not in
the absence, of RANKL, confirming that the cells used in this and previous
experiments were genuine osteoclast precursors. We then asked whether the
observed effects of wear debris on anti-osteoclastogenic cytokine signaling
are preserved after the initiation of RANKL-induced differentiation. To this
end, osteoclast precursors were incubated with M-CSF and RANKL for forty-eight
hours prior to treatment with cytokine and wear debris to generate, consistent
with the terminology of Huang et
al.18, late
osteoclast precursors. As shown in Figure
6, B, incubation of osteoclast precursors with RANKL for
forty-eight hours diminished neither the ability of IL-6 and IFN-? to
activate STAT3 and STAT1, respectively, nor the inhibition of these signaling
pathways by wear debris. Identical results were obtained following incubation
of osteoclast precursors with M-CSF and RANKL for twenty-four hours (data not
shown). Thus, inhibition of anti-osteoclastogenic signaling by wear debris
occurs in both early and late osteoclast precursor cells.
Involvement of SOCS Proteins in Wear Debris Inhibition of Cytokine
Signaling
In addition to p38 activation, a second mechanism identified for the
downregulation of JAK/STAT signaling involves induction of members of the SOCS
family of cytokine suppressors. Specifically, SOCS1 downregulates IFN-?
signaling, and SOCS3 downregulates IL-6 signaling. With this in mind, we
studied the effects of wear debris particles on SOCS1 and SOCS3 mRNA
expression. As shown in Figure 7,
A, SOCS3 mRNA was substantially induced following
incubation of osteoclast precursors with polymethylmethacrylate or titanium
particles for one hour, whereas SOCS1 mRNA levels remained relatively
unaffected. This finding suggests that SOCS3 contributes to the inhibition of
IL-6 signaling by both polymethylmethacrylate and titanium (see above);
however, the failure of titanium to significantly induce SOCS1 suggests that
the drastic downregulation of IFN-? signaling by titanium particles was
not mediated by this suppressor. Finally, MAP-kinase inhibition reduced the
induction of SOCS3 mRNA (Fig. 7,
B) by wear debris in much the same pattern as seen for
TNF-a mRNA (Fig. 3). This
result suggests that MAP kinases inhibited IL-6 signaling, at least in part,
by inducing expression of SOCS3.
Multiple lines of evidence support the hypothesis that an inflammatory
response to wear debris is an important factor in the development of
periprosthetic osteolysis. Most notably, several different types of
particulate debris have been shown to initiate pro-inflammatory cytokine
synthesis, both when cocultured in vitro with isolated
macrophages4-7
and when introduced to in vivo animal models of wear debris-induced
osteolysis6,24,25.
These observations provide a very simple model for osteolysis, in which the
pro-inflammatory cytokines released from macrophages in response to wear
debris, particularly TNF-a and IL-1, support the genesis and activation
of osteoclasts both directly and through the induction of elevated levels of
RANKL, the key mediator of osteoclast formation. This generation and
activation of osteoclasts then leads to the unchecked bone resorption seen in
osteolysis.
However, this model almost certainly reflects an over-simplification of the
cellular and molecular complexities of periprosthetic osteolysis. For
instance, particulate debris can stimulate cells other than macrophages in the
periprosthetic tissue, such as
fibroblasts26,27,
mesenchymal stem
cells28, and
osteoblasts29. In
addition, the involvement of lymphocytes, while controversial, remains a
distinct
possibility30-32.
Another area that has received insufficient attention is the possibility that
wear debris might be involved in "reprogramming" cellular
responses to cytokines within the periprosthetic space. A full understanding
of the excess osteoclastogenesis that is seen in osteolysis thus requires an
assessment of the influence of particulate debris on pro-osteoclastogenic and
anti-osteoclastogenic cytokine signaling in osteoclast precursors.
We addressed this issue by evaluating the ability of polymethylmethacrylate
and titanium debris to modulate signaling by IFN-? and IL-6, cytokines
that recently have been shown to engage in anti-osteoclastogenic
signaling10,12,13.
The antiresorptive actions of IFN-? are well understood, involving
downregulation within osteoclast precursor cells of TRAF6, an essential
mediator of RANKL-induced
osteoclastogenesis10.
Characterization of this ability to antagonize RANKL signaling has led to
widespread acceptance of IFN-? as an anti-osteoclastogenic cytokine,
although there may be rare exceptions to this rule, such as a recently
described experimental periodontitis model in which IFN-? may have
pro-resorptive
activity33. IL-6
presents a more complicated picture. The pleiotropic nature of IL-6, as
elucidated in the "signal orchestration
model"34,
which describes how this cytokine can generate multiple signals that oppose
each other, is epitomized by the observations that IL-6 can mediate both
pro-resorptive and anti-resorptive
signaling12.
Although it has been reported that IL-6 can induce osteoclastogenesis in a
RANKL-independent
manner35, the
observed effect was dramatically less than that of RANKL and is unlikely to be
of significance under normal circumstances, where RANKL is essential for
osteoclastogenesis36,
or in models of osteolysis, where RANKL blockade alleviates
disease37,38.
IL-6 also promoted osteoclastogenesis of the monocytic cell line U937 in
cocultures with osteoblasts, but the role of RANKL in this setting was not
established39.
Careful analysis of this complex situation has indicated that the
pro-resorptive actions of IL-6 most likely can be explained by the induction
of RANKL synthesis by osteoblasts, whereas the direct actions of IL-6 on
osteoclast lineage cells appear to be principally
anti-resorptive12,13.
Thus, for the osteoclast precursor cells used in our experiments, IL-6, like
IFN-?, can be considered to act by inhibiting osteoclastogenesis.
IFN-? and IL-6 both signal through the JAK/STAT
pathway40, a
feature shared by many anti-osteoclastogenic cytokines, suggesting that
JAK/STAT signaling may be a general mechanism for cytokine inhibition of
osteoclastogenesis. Both polymethylmethacrylate and titanium substantially
inhibited IL-6-induced STAT3 activation, showing that wear debris does indeed
possess the ability to suppress JAK/STAT cytokine signaling. This inhibition
was seen in both early osteoclast precursors (prior to RANKL addition) and
late osteoclast precursors (after differentiation with RANKL for forty-eight
hours). To investigate the mechanism of inhibition of IL-6 signaling by wear
debris, we evaluated two pathways known to be involved in the downregulation
of IL-6 signaling (namely, the induction of SOCS3 and the activation of p38
MAP kinase), and we found evidence for involvement of both.
Polymethylmethacrylate and titanium strongly induced the expression of
SOCS3 in osteoclast precursor cells. SOCS3, which inhibits IL-6 signaling by
direct interaction with the gp130 receptor, recently was identified as a key
regulator of the anti-osteoclastogenic activity of
IL-613. SOCS3 is
induced in osteoclast precursors by the pro-resorptive cytokines RANKL and
TGF-ß, and it can directly promote osteoclastogenesis and inhibit the
actions of anti-osteoclastogenic cytokines when overexpressed in precursor
cells41. Thus,
SOCS3 is emerging as an important determinant of osteoclast formation, and we
can interpret the induction of SOCS3 seen in our studies as contributing to
the wear debris-induced suppression of anti-osteoclastogenic signaling by
IL-6.
In addition to SOCS3 induction, our results also showed involvement of
activation of p38 MAP kinase in wear debris downregulation of IL-6 signaling.
Polymethylmethacrylate and titanium each were shown to activate p38, ERK, and
JNK MAP kinases, and, while inhibition of either ERK or JNK activation was
without effect, inhibition of p38 specifically prevented the ability of wear
debris to inhibit IL-6-induced STAT3 activation. This finding suggests that
p38, but not ERK or JNK, is also involved in mediating the effects of wear
debris on IL-6 signaling.
In contrast to its effects on IL-6 signaling, polymethylmethacrylate had no
apparent effect on STAT1 activation by IFN-?. However, titanium
particles did cause a dramatic inhibition of IFN-?-induced STAT1
activation and expression of the IFN-?-inducible mRNAs MIG and IP-10. As
was the case for the inhibition of IL-6, the inhibition of IFN-?
signaling by titanium was observed in both early and late osteoclast
precursors. Mechanistically, titanium inhibition of IFN-? signaling did
not appear to involve activation of MAP kinases. This finding contrasts with
wear debris inhibition of IL-6 signaling, which involved p38. A second
difference related to the inhibition of IL-6 signaling was the absence of
evidence implicating induction of SOCS expression in wear debris inhibition of
IFN-? signaling. SOCS1, responsible for downregulation of STAT1
signaling, was not induced to any great extent by titanium (or
polymethylmethacrylate). Future experiments will be required to determine the
molecular basis for titanium inhibition of IFN-? signaling.
MAP kinase activation mediated not only the downregulation of
anti-osteoclastogenic cytokine signaling but also the induction of
pro-inflammatory cytokine expression by both polymethylmethacrylate and
titanium wear debris. As detailed above, p38, ERK, and JNK were all activated
by both polymethylmethacrylate and titanium treatment of osteoclast precursor
cells, consistent with the involvement of all three of these MAP kinase
pathways in
osteoclastogenesis42.
While inhibition of each of these MAP kinase families individually reduced
wear debris-induced TNF-a expression moderately, the inhibition of
multiple MAP kinases was much more suppressive, suggesting redundancy between
the different MAP kinase pathways in this pathway. Simultaneous inhibition of
p38, ERK, and JNK effectively abolished TNF-a induction. This finding
contrasts with the wear debris suppression of cytokine signaling, which was
mediated by p38 alone in the case of IL-6 and independently of MAP kinase
activation in the case of IFN-?.
SOCS3 expression was inhibited by MAP kinase inhibitors in a pattern
similar to that seen for TNF-a. Thus, the inhibition of p38 alone caused
only a moderate decrease in wear debris-induced SOCS3 expression, suggesting
that the dramatic dependence on p38 activation of wear debris-induced
inhibition of IL-6 signaling cannot be fully explained by p38-dependent
induction of SOCS3 expression. SOCS-independent mechanisms by which MAP
kinases can inhibit JAK/STAT signaling have been previously
reported14,15,
and further experiments will be needed to fully detail the roles of the
various MAP kinase and SOCS pathways in wear debris-induced pro-inflammatory
signaling and inhibition of anti-osteoclastogenic cytokine signaling.
In summary, we have demonstrated that wear debris can potently inhibit
signaling by IFN-? and IL-6. Since these cytokines, like many other
cytokines that signal via JAK/STAT activation, are involved in
anti-osteoclastogenic signaling, this represents a novel mechanism for wear
debris-induced osteolysis. ?