Viral interleukin-10 gene inhibition of inflammation, osteoclastogenesis, and bone resorption in response to titanium particles

Authors


Abstract

Objective

To evaluate the potential of viral interleukin-10 (vIL-10) gene therapy as an approach to prevent wear debris–induced inflammation, osteoclastogenesis, and bone resorption as it relates to periprosthetic osteolysis in patients with total joint replacements.

Methods

Replication-defective adenovirus vectors expressing vIL-10 (AdvIL-10) or LacZ (AdLacZ) target genes were used to transduce fibroblast-like synoviocytes (FLS) in vitro, and the effects of these cells on wear debris–induced proinflammatory cytokine production and receptor activator of nuclear factor κB ligand + macrophage colony-stimulating factor splenocyte osteoclastogenesis were assessed by enzyme-linked immunosorbent assay and tartrate-resistant acid phosphatase assay. The effects of AdvIL-10 administration on wear debris–induced osteolysis in vivo were analyzed using the mouse calvaria model, in which AdLacZ was used as the control.

Results

In the presence of AdLacZ-infected FLS, titanium particle–stimulated macrophages exhibited a marked increase in secretion of tumor necrosis factor α (TNFα) (6.5-fold), IL-6 (13-fold), and IL-1 (5-fold). Coculture with AdvIL-10–transduced FLS suppressed cytokine secretion to basal levels, while addition of an anti–IL-10 neutralizing antibody completely blocked this effect. The vIL-10–transduced FLS also inhibited osteoclastogenesis 10-fold in an anti–IL-10–sensitive manner. In vivo, titanium implantation resulted in a 2-fold increase in osteoclasts (P < 0.05) and in a 2-fold increase in sagittal suture area (P < 0.05). This increase over control levels was completely blocked in mice receiving intraperitoneal injections of AdvIL-10, all of whom had measurable serum vIL-10 levels for the duration of the experiment. Immunohistochemistry demonstrated reduced cyclooxygenase 2 and TNFα expression in AdvIL-10–infected animals.

Conclusion

This study demonstrates that gene delivery of vIL-10 inhibits 3 processes critically involved in periprosthetic osteolysis: 1) wear debris–induced proinflammatory cytokine production, 2) osteoclastogenesis, and 3) osteolysis.

More than 400,000 total joint arthroplasties are performed annually in the US, among ∼1.3 million performed worldwide to treat debilitating end-stage arthritis (1). It is anticipated that the number of arthroplasties performed will rise dramatically over the next several decades, due to increasing numbers in the elderly population. Unfortunately, arthroplasty is not without complication, since up to 20% of the patients develop bone loss around the prosthesis within 15–20 years of the initial surgery, often leading to revision surgery (2–4). Presently, there are no proven drug therapies for aseptic loosening.

Over time, billions of microscopic debris particles from the prosthesis are released into the joint space surrounding an implant and become imbedded into the surrounding tissues (5, 6). Polyethylene is the major source of particles, but methylmethacrylate bone cement and metal also contribute (7–9). As the particles become imbedded in the tissues surrounding the implant, they are phagocytosed by macrophages and thereby stimulate an inflammatory reaction leading to the secretion of proinflammatory cytokines (10, 11). Tumor necrosis factor α (TNFα), interleukin-6 (IL-6), and IL-1 are all found in abnormally high levels in synovial fluid and tissue surrounding failed implants (10–12). These cytokines stimulate osteoclast differentiation and activation through a mechanism that appears dependent on the induction of receptor activator of nuclear factor κB ligand (RANKL), a critical factor in osteoclastogenesis (13, 14). Thus, periprosthetic osteolysis depends on 3 important events: 1) inflammation, 2) osteoclast formation, and 3) osteoclastic bone resorption. Factors that inhibit these processes are potential therapeutic agents for the treatment and prevention of periprosthetic osteolysis.

IL-10 is an antiinflammatory cytokine produced by macrophages and lymphocytes. IL-10 functions in a negative-feedback loop, in which it suppresses the release of inflammatory cytokines and dampens the acute inflammatory response (15). Mice deficient in IL-10 develop a number of autoimmune inflammatory diseases (16). In vitro studies have shown that IL-10 reduces TNFα, IL-1, and IL-6 messenger RNA levels and protein production following exposure to a number of stimuli, including titanium (Ti) wear debris (17). IL-10 also has potent inhibitory effects on osteoclastogenesis (18). Clinically, IL-10 has been identified in interfascial membranes (19), and has recently been shown to correlate with a protective effect on joint destruction (20).

The Epstein-Barr virus protein BCRF1 is a viral homolog of mammalian IL-10 (vIL-10), but only possesses the immunosuppressive/antiinflammatory effects (21, 22). Following the current paradigm that explains inflammatory disease as an imbalance between positive and negative mediators (23–25), investigators have evaluated vIL-10 as a candidate for arthritis gene therapy. These studies have demonstrated that vIL-10 can inhibit the invasiveness of human rheumatoid arthritis fibroblast-like synoviocytes (FLS) (26) and prevent collagen-induced arthritis (27, 28). Based on these studies, we hypothesized that vIL-10 would be an effective inhibitor of wear debris–induced osteolysis and a potential gene therapy for aseptic loosening. To this end, we examined its effect on the production of proinflammatory cytokines by Ti-stimulated macrophages, as well as its direct effect on osteoclastogenesis in vitro. We also examined the ability of vIL-10 to inhibit osteoclast formation and activity in an animal model (29).

MATERIALS AND METHODS

Ti particles

Pure Ti particles were obtained from Johnson Matthey Chemicals (Ward Hill, MA) and prepared as previously described (30). Briefly, particles 1–3 μm in diameter were suspended in phosphate buffered saline (PBS) at a concentration of 1 × 108 particles/ml. Particle size was confirmed with a Coulter Channelizer (Beckman Coulter, Fullerton, CA) which determined 90% of the particles to be <5 μm in diameter. A Limulus assay (BioWhittaker, Walkersville, MD) was used to show that the suspension was free of endotoxin.

Adenovirus preparation and vIL-10 production

We repropagated and titered high-titer stocks (∼1010 plaque-forming units/ml) of replication-incompetent adenoviral IL-10 (AdvIL-10; a gift from Dr. C. Evans) and adenoviral LacZ (AdLacZ) as described previously (28). To determine the transduction efficiency and production of vIL-10, the murine FLS L1.1198 cells (31) were seeded at a density of 3 × 105 cells/well in a 6-well plate and grown overnight at 37°C. Cells were infected at different multiplicities of infection (MOI; 0–100), and media were collected 24 hours later and analyzed for vIL-10 protein production by enzyme-linked immunosorbent assay (ELISA) as previously described (27), with purified rat anti–vIL-10 antibodies (PharMingen, San Diego, CA).

Cell-culture models

The murine macrophage ANA-1 cell line was kindly provided by Dr. George Cox (National Cancer Institute, Frederick, MD) and was maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin. ANA-1 cells were derived from bone marrow macrophages of C57BL/6 mice by infection with the J2 recombinant retrovirus expressing v-myc/v-raf (32). The cells express markers of the differentiated macrophage, including Ly-5, Mac-1, and Fcγ receptor, but do not express either B or T lymphocyte cell markers (32). The FLS (L1.1198 FLS cells) were obtained from the knee joint of CBA/B6 mice (33). The cells express type I collagen and vascular cell adhesion molecule 1, consistent with a synovial fibroblast phenotype. The cells have remained stable over >50 passages in culture (33).

ANA-1 cell–conditioned media and ELISA

To examine the effects of different vectors on cytokine production by Ti-stimulated macrophages, ANA-1 cells were plated at a density of 5 × 105 cells/well in 24-well plates and treated with 1 × 107 Ti particles. Using transwell membranes (purchased from VWR, Bridgeport, NJ) which were 6.5 mm in diameter, with a pore size of 0.4 μm, uninfected control cells or FLS infected with either AdvIL-10 or AdLacZ were cocultured with the macrophages at the same time as Ti stimulation. For the coculture experiments, FLS were infected at an MOI of 100 for 18 hours in 6-well plates. FLS were removed by pipetting and recultured on the transwell membranes (500,000 cells/transwell membrane). After 18 hours, coculture supernatants were collected and centrifuged at 1,500 revolutions per minute for 3 minutes to remove Ti particles, and TNFα, IL-6, IL-1, and vIL-10 production was analyzed by ELISA (PharMingen) as we have previously described (30). In some experiments, the cocultures were performed in the presence of neutralizing antibodies that recognize both IL-10 and vIL-10 (catalog number 18550D; PharMingen). The antibody was used at a concentration of 40 ng/ml.

Osteoclastogenesis assay in vitro

Mouse splenocytes were prepared as described previously (34), and seeded at 1 × 106 cells/well in a 24-well plate in complete osteoclast media consisting of phenol red–free α–minimal essential medium with nonessential amino acids (10 nM; Gibco, Grand Island, NY), L-glutamine (200 mM; Gibco), 10% heat-inactivated FBS, and penicillin–streptomycin at pH 7.25. The osteoclast medium was supplemented with macrophage colony-stimulating factor (M-CSF; 30 ng/ml) and RANKL (100 ng/ml). Using transwell membranes, uninfected FLS or FLS expressing either AdvIL-10 or AdLacZ were placed in coculture with the splenocytes in a manner identical to that described above. A 50% fresh-medium change was performed every 2 days, and the cultures were incubated at 37°C for 6 days. The spleen-cell cultures were stained using the Leukocyte Acid Phosphatase kit (Sigma, St. Louis, MO) according to the manufacturer's directions, and osteoclast formation was quantified by counting multinucleated tartrate-resistant acid phosphatase (TRAP)–positive cells.

In vivo mouse calvarial resorption model

Bone resorption and osteoclast numbers were examined in vivo as previously described (29). Briefly, 5 healthy male CBA × B6 mice (Jackson Laboratories, Bar Harbor, ME) were used in each group. All animals were housed and treated according to University-approved guidelines. Mice were anesthetized with 70–80 mg/kg of ketamine and 5–7 mg/kg of xylazine by intraperitoneal (IP) injection. A 1 cm × 1 cm area of calvarial bone was exposed by making a midline sagittal incision over the calvaria. Thirty milligrams of Ti particles (Johnson Matthey Chemicals) was spread over the area and the incision was closed. Animals in the sham group had the surgical procedure performed without implantation of particles. Two hundred microliters of either AdvIL-10 or AdLacZ containing a total of 2 × 108 infectious viral particles was administered via IP injection on day 0. Serum for the measurement of vIL-10 was obtained from anesthetized mice at 0, 2, 3, 4, 5, 7, 9, and 10 days postsurgery by retroorbital venous puncture, and ELISA was performed as described above. Five animals were assessed in each group, but 1 of the AdvIL-10–infected mice died of an anesthetic complication on day 3. Data are provided on this animal through day 3.

Ten days after surgery, the mice were killed by cervical dislocation and the calvaria were harvested, fixed in 10% formalin, decalcified in 10% EDTA, dehydrated in graded alcohol, and embedded in paraffin. Sections of calvarial bone centered on the midline suture and just anterior to the occipital suture were obtained. The sagittal suture area and osteoclast numbers were quantified using 3-μm sections stained with hematoxylin and eosin and TRAP (Diagnostics Acid Phosphatase kit; Sigma), respectively. Five sections from each of 5 animals were digitally photographed at 40× magnification, and the image was oriented with the midline suture in the center of the 1.3-mm diameter field as previously described (29). The area of the sagittal suture and surrounding bone loss was traced by hand and quantified with Osteometrics software (Atlanta, GA). Four sections per mouse were analyzed in this manner. The number of osteoclasts in each calvarial section was quantified from a TRAP-stained section by 3 independent and blinded observers.

Immunohistochemistry

The paraffin-embedded tissues were prepared for immunohistochemistry. Blocking was done with a 1:20 dilution of normal goat serum (Vector, Burlingame, CA) for 20 minutes at room temperature. The slides were drained and the appropriate dilution of cyclooxygenase 2 (COX-2; 1:200) rabbit polyclonal anti-mouse primary antibody was obtained (Cayman, Ann Arbor, MI) and applied. For the negative controls, the primary antibody was eliminated. Following incubation overnight at 4°C, a 1:200 dilution of biotinylated goat anti-rabbit secondary antibody (Vector) was applied for 30 minutes at room temperature. The slides were washed with PBS, and a 1:250 dilution of horseradish peroxidase–streptavidin (Zymed, San Francisco, CA) was applied for 30 minutes at room temperature. The slides were washed with PBS and then rinsed with deionized water before applying the chromagen. AEC chromagen (Biocare Medical, Walnut Creek, CA) was applied for 5 minutes at room temperature. Slides were drained, washed with distilled water, then counterstained with Tachas hematoxylin (Biocare Medical). Immunostains for TNFα used a TNFα rabbit polyclonal anti-mouse primary antibody (1:500 dilution; Genzyme, Cambridge, MA) and subsequent steps were identical to the protocol described for COX-2 immunohistochemistry.

Statistical analysis

Comparisons were made between the various groups by one-way analysis of variance. Statistical significance was considered present when the P value was less than or equal to 0.05.

RESULTS

Transduction of L1.1198 FLS, and vIL-10 protein production. Since FLS adjacent to the prosthesis represent ideal target cells for gene therapy for aseptic loosening, we determined the transduction efficiency and vIL-10 production with L1.1198 FLS in vitro. FLS were infected at MOI ranging from 0 to 100, and the culture supernatants were collected 24 hours after infection and analyzed for the levels of vIL-10 protein by ELISA (Figure 1). Viral IL-10 production was dose dependent and maximal at an MOI of 100, at which point a mean of ∼600 ng/ml of vIL-10 protein was produced. β-galactosidase staining of cell cultures infected with AdLacZ at an MOI of 100 demonstrated infection in essentially 100% of cells (data not shown), but there was no production of vIL-10 protein by these cultures.

Figure 1.

Viral interleukin-10 (vIL-10) production by transduced fibroblast-like synoviocytes. L1.1198 cells were grown to 80% confluence and infected at different multiplicities of infection (MOI). The culture media were collected 24 hours later and analyzed for vIL-10 production by enzyme-linked immunosorbent assay. The graph depicts the mean and SEM of 4 experiments for each group. Infection with adenoviral LacZ served as the control.

In subsequent experiments, FLS were infected with AdvIL-10 or AdLacZ at an MOI of 100, and after 18 hours were subcultured on transwell membranes (500,000 cells/well) in 24-well culture plates. ELISA demonstrated a vIL-10 concentration of 278 ± 6 ng/ml (mean ± SEM) in these cultures after 18 hours, whereas levels in AdLacZ-infected and uninfected control cultures were undetectable.

Effect of vIL-10 on the production of TNFα, IL-6, and IL-1 by macrophages stimulated with Ti particles. IL-10 has antiinflammatory properties and has been shown to inhibit cytokine release by macrophages (23, 35). To determine the effect of vIL-10 on wear debris–induced proinflammatory cytokine production by macrophages, uninfected (control) FLS or FLS infected with adenoviruses expressing either vIL-10 or LacZ at an MOI of 100 were cultured on transwell membranes. The transwell membranes were placed in coculture with ANA-1 macrophages at the time of exposure to a maximally stimulatory concentration of Ti particles (1 × 107 particles/ml). In the presence of AdLacZ-infected FLS, Ti-stimulated macrophages exhibited a marked increase in secretion of TNFα (6.5-fold), IL-6 (13-fold), and IL-1 (5-fold) as compared with unstimulated macrophages (Figure 2). The effect was similar between the cultures containing the uninfected controls and those containing the AdLacZ-infected FLS, indicating that adenoviral infection of the FLS did not influence the cytokine secretion in the cocultures of stimulated macrophages. In contrast, coculture with AdvIL-10–infected FLS resulted in complete suppression of TNFα, IL-6, and IL-1 secretion by Ti-stimulated macrophages. In these cultures, cytokine levels did not increase substantially above baseline following Ti stimulation.

Figure 2.

Viral interleukin-10 (vIL-10) inhibition of proinflammatory cytokine production by titanium (Ti)–stimulated macrophages. ANA-1 cells were grown and stimulated with Ti, with and without anti–vIL-10 antibodies (Ab), in cocultures with L1.1198 fibroblast-like synoviocytes (FLS) alone or infected with adenoviral vIL-10 or adenoviral LacZ. The culture media were collected 18 hours later and analyzed for A, tumor necrosis factor α (TNF-α), B, IL-6, and C, IL-1 production, by enzyme-linked immunosorbent assay. Control (Ctrl) cultures received no Ti stimulation. The graphs depict the mean and SEM of 4 independent experiments for each group. P < 0.001 compared with controls (*) or compared with cultures with adenoviral vIL-10–infected FLS cells without neutralizing antibody (#).

In parallel cultures, we added anti–vIL-10 neutralizing antibody in order to confirm that this effect was due to the secretion of vIL-10 protein by the infected FLS. In the presence of the neutralizing antibody, the inhibitory effect of the AdvIL-10–infected fibroblast coculture was completely reversed. Thus, the cytokine levels in the culture medium 18 hours after Ti stimulation were similar to those observed in control and AdLacZ-infected fibroblast cocultures. These findings 1) confirm the production of functional vIL-10 in adenovirus-infected fibroblast cultures at a relatively low MOI, and 2) demonstrate the potent effect of vIL-10 as a suppressor of the macrophage inflammatory response to wear-debris particles.

Next, experiments were performed to determine if FLS expressing vIL-10 could inhibit cytokine secretion after initiation of the inflammatory response to wear debris (Figure 3). Transwell membranes containing either control or AdvIL-10–infected FLS were added at 0, 12, or 24 hours after exposure of ANA-1 macrophages to Ti particles. While maximal inhibition occurred when AdvIL-10–infected FLS were added to the cultures at the time of Ti exposure (0 hours pretreatment), significant (P < 0.001), but less, inhibition occurred when vIL-10–expressing FLS were added at later time points after the inflammatory response had been established. The suppression of cytokine secretion was intermediate when AdvIL-10–infected FLS were added after 12 hours, and least when coculture was initiated after 24 hours of Ti exposure. Compared with the Ti-stimulated control cultures, treatment with vIL-10–expressing FLS after 24 hours resulted in a 48%, 46%, and 66% reduction in TNFα, IL-6, and IL-1 production, respectively (Figure 3). Thus, vIL-10 is able to inhibit an established inflammatory response in a macrophage population exposed to Ti particles.

Figure 3.

Viral interleukin-10 inhibition of inflammatory cytokine production after an inflammatory response has been established. ANA-1 cells were grown and stimulated with titanium (Ti) as described in Materials and Methods. Coculture with fibroblast-like synoviocytes (FLS) was initiated either at the time of Ti stimulation (0 hours) or after 12 hours or 24 hours following exposure, using either infected (cross-hatched bars) or noninfected (solid bars) FLS on transwell membranes. Culture media were collected 18 hours after initiation of the coculture, and analyzed for A, tumor necrosis factor α (TNF-α), B, interleukin-6 (IL-6), and C, IL-1. Control cultures received no Ti stimulation. The graphs depict the mean and SEM of 4 independent experiments for each group. P < 0.001 compared with noninfected controls (*) and compared with noninfected control cultures at the same time point (#); P < 0.05 compared with infected controls (@).

Ability of vIL-10 to inhibit splenocyte osteoclastogenesis. Having shown that vIL-10 inhibits the secretion of proosteoclastogenic cytokines by Ti-stimulated macrophages, experiments were performed to examine its direct effects on osteoclastogenesis. Osteoclastogenesis was assessed in cultures of murine splenocytes treated with RANKL and M-CSF in the presence of cocultures of uninfected control FLS or FLS infected with AdvIL-10 or AdLacZ at an MOI of 100. M-CSF– and RANKL-treated spleen cultures had ∼950 TRAP+ multinucleated cells per well in cocultures with control and AdLacZ-infected FLS (Figure 4). In contrast, there was an ∼10-fold decrease in the number of TRAP+ multinucleated cells in cultures containing AdvIL-10–infected FLS. Addition of an anti–vIL-10 neutralizing antibody to these cultures resulted in a complete recovery of osteoclastogenesis, and the number of TRAP+ multinucleated cells was similar to those observed in cultures containing uninfected and AdLacZ-infected FLS. These findings indicate that vIL-10 inhibits the ability of osteoclast precursors to undergo differentiation in response to M-CSF and RANKL.

Figure 4.

Reduction of osteoclastogenesis in splenocytes treated with macrophage colony-stimulating factor (M-CSF) and receptor activator of nuclear factor κB ligand (RANKL) by viral interleukin-10 (vIL-10) produced by infected fibroblast-like synoviocytes (FLS). Splenocytes were seeded at 1.5 × 106 cells per well in a 24-well plate and cocultured with FLS using transwell membranes for 6 days, with and without anti–vIL-10 antibodies (Ab). Control (Ctrl) cultures received no M-CSF/RANKL. The mean and SEM number of tartrate-resistant acid phosphatase–positive (TRAP+) multinucleated cells (osteoclasts) is shown for each group. P < 0.001 compared with controls (*) and compared with cultures with adenoviral vIL-10–infected FLS without neutralizing antibody (#).

Effects of vIL-10 on wear debris–induced osteolysis and osteoclastogenesis in vivo. The in vitro findings suggest that vIL-10 may be a particularly potent inhibitor of particle-mediated bone resorption, since it suppressed 2 important independent processes involved in periprosthetic bone loss: inflammatory cytokine secretion and osteoclast differentiation. The effect of vIL-10 on osteoclastogenesis and bone resorption in vivo was examined using a well-established murine model of particle-induced osteolysis whereby surgical placement of 30 mg of particles onto the calvaria results in a reproducible pattern of inflammatory bone loss over 10 days (29). Initial studies assessed the expression of vIL-10 in adenovirus-infected mice. Serum concentrations of vIL-10 were assessed in mice following IP injection of adenovirus. IP administration of AdvIL-10 resulted in an increase in vIL-10 within 2 days, with peak levels by 5 days (mean ± SEM 700 ± 100 ng/ml), and a persistent increase for at least 10 days (Figure 5A). In contrast, vIL-10 was not detected in either uninfected or AdLacZ-infected mice. These findings demonstrate that administration of AdvIL-10 resulted in the production of detectable levels of vIL-10 protein for the duration of the experiment.

Figure 5.

Adenoviral interleukin-10 (AdvIL-10) gene inhibition of titanium (Ti)–induced osteolysis in the mouse calvaria model. Mice were injected with AdvIL-10 or adenoviral LacZ, intraperitoneally, on day 0, and serum concentrations of vIL-10 were determined on the indicated days by enzyme-linked immunosorbent assay (A). The graph depicts the mean concentration in 3 serum samples obtained on each of the days. Mice were killed on day 10, and osteoclast numbers (B) and bone resorption (C) were quantified as described in Materials and Methods. Sham-operated animals did not receive Ti implantation during surgery; control (Ctrl) animals received Ti stimulation without viral injection. ∗ = P < 0.05 compared with sham-operated control animals. TRAP+ = tartrate-resistant acid phosphatase positive.

The number of osteoclasts and the amount of bone resorption were determined in mouse calvaria sections harvested 10 days following Ti implantation or sham surgery. Compared with the sham-operated mice, Ti implantation onto the calvarial surface resulted in an ∼2-fold increase in the number of osteoclasts adjacent to the sagittal suture (Figure 5B). However, the number of osteoclasts was markedly reduced in Ti-implanted animals that received a systemic injection of AdvIL-10. In contrast, AdLacZ-infected mice treated with Ti particles had an increased number of osteoclasts compared with the Ti-treated controls. Thus, the decrease in osteoclasts was not due to adenoviral infection, which, instead, enhanced osteoclast formation.

We and other investigators have previously demonstrated that the sagittal suture area can be used as a quantitative measure of bone resorption in response to particles (29, 36, 37). In control animals, Ti particles caused a 2-fold increase in the mean sagittal suture area (Figure 5C), consistent with our previous findings (29, 36). However, in animals infected with AdvIL-10, the mean sagittal suture area remained at basal levels, demonstrating a complete block of inflammatory bone resorption in response to Ti particles. Consistent with the effect observed on osteoclast formation, Ti-treated animals infected with AdLacZ had levels of bone resorption that were greater than those observed with Ti stimulation alone. Collectively, these findings demonstrate that vIL-10 inhibits osteoclast formation and bone resorption in response to Ti wear debris in vivo.

Effect of vIL-10 transgene expression on tissue inflammation in vivo. Prior work in our laboratory has established that transgenic mice with deficient TNFα signaling and reduced COX-2 expression have reduced inflammatory bone resorption in response to Ti particles (37, 38). Thus, these markers of inflammation are functionally important in this pathologic process. Immunohistochemical stains for COX-2 (Figures 6A–D) and TNFα (Figures 6E–H) were performed on histologic sections obtained from the 4 treatment groups. The expression of COX-2 was essentially absent in the sham controls and in the AdvIL-10–treated calvaria (Figures 6A and D). In contrast, in the absence of vIL-10, Ti particles induced COX-2 expression in inflammatory tissues, with the greatest level of expression noted in the animals receiving AdLacZ treatment (Figures 6B and C). Similarly, TNFα staining was scant in both the sham- and AdvIL-10–treated animals (Figures 6E and H), whereas the immunodetection of TNFα was robust in the inflammatory tissue in the calvaria of the Ti-treated animals (Figures 6F and G). Again, the staining was most intense in the animals receiving AdLacZ treatment (Figure 6G). Thus, the decreased bone resorption in AdvIL-10–treated animals is associated with a decreased local inflammatory response to Ti particles.

Figure 6.

Adenoviral interleukin-10 (AdvIL-10) gene inhibition of the expression of cyclooxygenase 2 (COX-2) and tumor necrosis factor α (TNFα) proteins in calvarial tissues following titanium (Ti) exposure. Calvarial sections were prepared for immunohistochemistry and were stained with antibodies directed against COX-2 (A–D) or TNFα (E–H) 10 days after surgery. Mice received different treatments, including sham operation (A and E), Ti implantation (B and F), adenoviral LacZ (AdLacZ) infection and Ti implantation (C and G), and AdvIL-10 infection (D and H). The AdvIL-10 or AdLacZ was injected intraperitoneally on day 0. The arrows demonstrate regions of increased expression in representative sections of mice calvaria. (Original magnification ×100 in A–D; ×400 in E–H.)

DISCUSSION

Aseptic loosening of orthopedic implants, secondary to periprosthetic osteolysis, remains a serious clinical problem that has no prophylactic or therapeutic solution other than revision surgery. Considering the currently accepted etiology of this condition, factors with inhibitory effects on inflammation, osteoclastogenesis, and bone resorption have potential therapeutic value. A number of studies support the concept that IL-10 might be an effective agent, including 1) the demonstration that IL-10 inhibits several important steps in the inflammatory process, including the secretion of cytokines, chemokines, and angiogenesis (17); 2) observations that IL-10 inhibits osteoclastogenesis (18); 3) a relative decrease in IL-10 is believed to favor a more intense inflammatory response (15); 4) higher IL-10 levels have been shown to correlate with a protective effect on joint disease in humans (20); and 5) vIL-10 gene therapy has been shown to inhibit FLS invasion (26) and ameliorate collagen-induced arthritis in mice (27, 28).

To evaluate the potential of vIL-10 gene therapy for aseptic loosening, we have studied its effects on wear debris–induced inflammation, osteoclastogenesis, and osteolysis in vitro and in vivo. In these studies we found that AdvIL-10–infected FLS produced high levels of active protein in vitro and inhibited the inflammatory response to Ti particles. The induction of IL-1, IL-6, and TNFα by Ti particles in ANA-1 murine macrophages was completely inhibited in cocultures containing AdvIL-10–infected FLS. Similarly, cocultured FLS infected with AdvIL-10 prevented osteoclastogenesis in primary murine splenocyte cultures stimulated with RANKL and M-CSF. The role of vIL-10 in these experiments was confirmed by using an anti–vIL-10 neutralizing antibody, which completely blocked the effect of the cocultured cells. Finally, in vivo experiments were performed using a well-characterized murine model of wear debris–induced osteolysis. These experiments demonstrated 1) successful adenoviral infection in vivo with synthesis of measurable levels of vIL-10, and 2) the ability of vIL-10 gene therapy to inhibit particle-induced inflammatory bone loss in the animal model. It is likely that the potent inhibitory effects of vIL-10 are due to the unique ability of this factor to interfere with inflammatory bone resorption through multiple mechanisms, involving direct effects both on the early events involved in inflammation and on the late events of osteoclast formation and activation.

The effects of both IL-10 and vIL-10 are dependent on the IL-10 receptor (IL-10R), which is a member of the interferon receptor family (39, 40). However, although IL-10 exhibits a high binding affinity, the binding affinity of vIL-10 is ∼1,000-fold lower (40–42). Thus, the relatively high levels of vIL-10 observed both in vivo and in vitro represent physiologically relevant concentrations. Since the goal of the current experiments was to demonstrate proof of the concept that vIL-10 gene therapy could successfully inhibit inflammatory bone resorption, an attempt was not made to determine the lowest concentrations capable of inhibiting this process. Thus, defining the optimal dose of AdvIL-10 remains a goal for future investigation.

The IL-10 receptor is composed of 2 subunits, with ligand binding occurring with the type 1 IL-10R (IL-10R1), which also associates with Jak1 (43). In contrast, type 2 IL-10R is solely involved in mediating downstream signaling events, and associates directly with the Jak kinase signaling molecule Tyk2. Ligand binding to the receptor complex results in phosphorylation and activation of the transcription factors Stat-1 and Stat-3 in monocytes (43). Stat-3, in particular, has been associated with the immunomodulatory effect of IL-10, since Stat-3−/− animals develop inflammatory diseases and are unresponsive to IL-10–mediated signaling events (44). The important direct downstream targets of Stat-3 are not yet clearly defined, although an additional signal, mediated by the carboxy-terminal region of the IL-10R1, is required for the antiinflammatory effects of IL-10 signaling (44). Antiinflammatory targets include the down-regulation of a variety of cytokines, chemokines, and inflammatory molecules, including IL-1, IL-6, and TNFα, as well as COX-2 in macrophages (17, 45–48). The IL-10 effect on these target molecules involves decreased transcription, as well as increased catabolism of the transcripts (47–50).

The expression of inflammatory mediators was suppressed both in vitro and in vivo by vIL-10. In vivo we examined the expression of TNFα and COX-2 by immunohistochemistry, both of which were inhibited in the setting of vIL-10 gene therapy. We selected these 2 genes as relevant biologic markers since prior work in our laboratory has demonstrated that transgenic animals with deficient signaling of either TNFα or COX-2 have a marked reduction in inflammatory bone resorption in response to Ti particles (37, 38). These findings demonstrate that vIL-10 results in the suppression of inflammatory mediators that have been demonstrated to be causatively related to particle-mediated bone resorption.

Although these studies support further investigation directed toward the clinical use of vIL-10 gene therapy for aseptic loosening, such an approach remains dependent on the development of safe and effective methods of DNA delivery. The host-response problems associated with the use of recombinant adenoviruses have been well documented (35, 51), and it is unlikely that this vector will be useful in humans (52). Consistent with this, we have observed that recombinant adenovirus induces a systemic inflammatory process that results in calvarial bone resorption in our mouse model, which was not observed in Nude mice (53). Similar observations were made in the current study, in which Ti + AdLacZ resulted in a level of bone resorption that was greater than that observed in Ti-treated control animals. The inhibition of bone resorption by vIL-10, therefore, overcame both the Ti-mediated stimulus and the inflammatory stimulus due to adenoviral infection.

Other methods of gene delivery, including the use of gene-activated matrices (54) and the adeno-associated virus vector (55), may offer safer and more practical solutions for both short-term and longer-term gene therapy. Provided that methods of gene delivery can be optimized, the problem of inflammatory bone loss adjacent to prosthetic implants offers an ideal situation for the application of gene therapy. This is due to the localized nature of this problem which theoretically permits localized delivery of the transgene to the joint capsule, with minimal systemic administration or adverse effects. The cells that line the effective joint space are synovial cells, the cell type used here to express high levels of vIL-10 in our cell-culture studies.

A major requirement of any approach to prevent the onset of prosthetic loosening is safety, particularly since this condition is both chronic and localized. IL-10–mediated antiinflammatory effects have been studied in a number of animal models, including adjuvant-induced arthritis, pulmonary diseases, myocardial diseases, and endotoxemia (48, 56–60). While these studies confirm the efficacy of vIL-10 for the treatment of inflammatory diseases, systemic toxicity has not been described. A study examining the safety of IL-10 administration in humans has also demonstrated no adverse symptoms or signs after IL-10 administration (61). However, IL-10 administration was associated with a decrease in cells expressing T cell markers, a reduction in T cell proliferation, and an inhibition of macrophage secretion of TNFα and IL-1 in response to an ex vivo challenge with endotoxin (61). The suppression of inflammation raises the specter of an increased propensity for the development of infections, and therefore the safety of IL-10 in humans still needs to be established.

The current study demonstrates that vIL-10 can interfere with critical steps involved in particle-induced inflammation, osteoclastogenesis, and bone loss in vitro and in vivo. The findings warrant further study of this important immune modulator as a candidate for inhibition of periprosthetic osteolysis. Since periprosthetic bone loss is a localized process, gene delivery can be confined to the intraarticular tissues surrounding the joint, thereby avoiding systemic effects. Further studies will be required to determine the optimal methods of delivery, efficacy, and safety and will include studies in larger animal models, which will possibly lead to effective therapeutic interventions in humans.

Acknowledgements

We would like to thank Drs. C. H. Evans and P. D. Robbins for providing us with the AdvIL-10 vector, and J. Harvey for critical assistance with the histologic assessments.

Ancillary