Treatment of Borrelia burgdorferi–Infected Mice with Apoptotic Cells Attenuates Lyme Arthritis via PPAR-γ

Lyme disease is an important vector-borne disease in the United States with an estimated 300,000 new cases each year (1). The spirochete Borrelia burgdorferi is the etiological agent of Lyme disease and is transmitted to mammalian hosts through the bite of infected Ixodes ticks (2). Humans infected with B. burgdorferi typically develop an expanding circular rash, erythema migrans, at the site of the tick bite, which can aid in clinical diagnosis (3). Prompt treatment of patients at this stage with antibiotics is critical, as individuals not treated during early infection can go on to develop debilitating, long-term complications, including chronic severe arthritis, carditis, and neurologic symptoms that can be difficult to treat.

A transient, recurring arthritis, typically of the knee, is the most common late manifestation of B. burgdorferi infection in untreated patients (4). The murine model of experimental Lyme arthritis recapitulates a portion of the disease found in humans and serves as an experimental model to investigate disease pathogenesis (5). Lyme arthritis development in the murine model is primarily the result of the infiltration of innate immune cells, mainly neutrophils, as mice without T and B cells are still susceptible (6). Development of arthritis correlates with the KC/CXCR2-mediated recruitment of neutrophils into the infected joint, followed closely by inflammatory monocytes (7, 8). Production of proinflammatory mediators drives pathogenesis, as their inhibition has been shown to modulate disease severity (9–13). In this model, arthritis undergoes spontaneous resolution following a peak of inflammation around 3–4 weeks postinfection (pi). Although specific Ab-mediated clearance of B. burgdorferi from infected joints is likely to play a role in arthritis resolution (14), the actual mechanisms mediating disease resolution are unclear. Inhibition of several pathways, including lipid mediator production (15, 16) and TLR signaling (17–19), resulted in poor arthritis resolution despite normal levels of anti-Borrelia Ab production.

Apoptosis, or programmed cell death, is an important mechanism contributing to the resolution of inflammation (20). Neutrophils are short-lived inflammatory cells programmed to undergo apoptosis, perhaps to limit their proinflammatory response. However, prolonged neutrophil and immune cell survival along with continued inflammatory cell recruitment can contribute to excess inflammation and host pathogenesis (20). The clearance of apoptotic cells (AC; efferocytosis) by macrophages and neutrophils is thought to contribute to the downregulation of proinflammatory immune responses (21). Resolution of inflammation is an important mechanism to repair host tissue damage caused by inflammatory cell infiltration and to replenish the population of resident macrophages depleted during the inflammatory response (22). AC have been shown to exhibit anti-inflammatory properties and perhaps contribute to the induction of inflammation resolution (23). Other models of inflammation, such as lung fibrosis, have shown a reduction in inflammation when treated with exogenous AC (21, 24). Thus, understanding how AC impact the resolution of inflammatory processes could lead to new insights in the treatment of chronic inflammatory diseases such as arthritis.

The pathway by which AC elicit an anti-inflammatory effect is still being determined, but peroxisome proliferator-activated receptor-γ (PPAR-γ) may be involved in this mechanism. PPAR-γ is a nuclear receptor that mediates transcriptional activation and suppresses inflammatory gene expression in macrophages and chemotaxis in neutrophils (25, 26). PPAR-γ activation has been shown to increase after AC administration in a lung fibrosis model, contributing to the downregulation of proinflammatory cytokines and promoting inflammation resolution (21). In this article, we show that a similar mechanism is active during experimental Lyme arthritis and, thus, may represent a general mechanism for limiting inflammatory processes.

C3H/HeJ mice, 4–6 wk of age, were used for all experiments. Both males and females were used for in vivo experiments and as a source of bone marrow cells. The mice were bred and housed in a specific pathogen-free facility and given continuous access to food and water. At specific time points following infection, mice were humanely euthanized via CO₂ inhalation as accepted by American Veterinary Medical Association guidelines. All studies were conducted in accordance with the guidelines of ACUC of the University of Missouri and performed under an approved protocol.

Spirochetes from the N40 strain of B. burgdorferi were used for all infections (27). Frozen stocks stored at −80°C were added to 7 ml of complete Barbour-Stoenner-Kelly–H medium containing 6% rabbit serum (Millipore Sigma) and grown for 6 d to log phase at 32°C. Spirochetes were enumerated using dark-field microscopy and a Petroff-Hausser counting chamber (Hausser Scientific, Horsham, PA). Dilutions were made in sterile Barbour-Stoenner-Kelly–H medium so that each hind footpad was inoculated with 50 μl of medium containing 5 × 10⁴ B. burgdorferi. GFP-expressing B. burgdorferi [a kind gift from J. Carroll (28), National Institutes of Health] was used for the phagocytosis assays. B. burgdorferi was given at a multiplicity of infection of 10 for in vitro studies.

Jurkat cells were grown in complete RPMI 1640 media containing 10% FBS, and 2% penicillin–streptomycin at 37°C in 5% CO₂ until needed. Jurkat cells were pelleted at 300 × g at 4°C for 5 min, and the supernatant was discarded. Cells were resuspended in complete media and washed three times. Cells were resuspended in 1 ml of RPMI 1640 containing 10% FBS and CD95 Fas Ab (BioLegend) at 10 μl/ml. After 6 h, cells were counted and diluted to a concentration of 1 × 10⁶ AC/ml. For apoptotic neutrophils, bone marrow neutrophils (BMN) were harvested as described and left for 24 h at 37°C in 5% CO₂. AC were verified via flow cytometry by cleaved–caspase-3⁺ cells. Ten microliters was injected into the tibiotarsal joint synovial cavities of each mouse at indicated days for in vivo studies and given at a 2:1 (AC:cells) concentration for the in vitro studies.

Briefly, protein from joint tissues was extracted as previously described (7), and tissues were harvested and snap-frozen in liquid nitrogen. The frozen samples were wrapped in aluminum foil and pulverized with a hammer, then placed into 1 ml of cold homogenization buffer (HBSS; Life Technologies) containing 0.2% protease inhibitor mixture (Millipore Sigma). Samples were homogenized for 2 min on a bead beater and then sonicated on ice for 20 s at 60% amplitude. Centrifugation (2000 × g for 20 min, 4°C) and filtration through a 0.45-μm filter were then performed. Protein quantity was then assessed via a BCA protein quantification assay (Thermo Scientific). Samples were stored at −20°C until used for analysis.

Total protein was isolated as described above, and sample proteins (25 μg) were separated on a 10% SDS-PAGE gel and transferred electrophoretically to a Trans-Blot nitrocellulose membrane (Bio-Rad, Hercules, CA). Membranes were blocked with 5% milk in TBST, then incubated with a 1:1000 dilution of anti-mouse PPAR-γ E8 (Santa Cruz Biotechnology) or 1:1000 β-actin C4 (Santa Cruz Biotechnology) Ab. Secondary Ab used was a 1:2000 dilution of goat α-mouse (Santa Cruz Biotechnology). Blot was developed using Pierce ECL Western Blotting Substrate (Thermo Scientific) and imaged via x-ray film. Densitometry on bands was performed using ImageJ software.

Neutrophils were isolated as previously described (8). Briefly, bone marrow was harvested from C3H mouse tibias and femurs and separated on a 1-ml Histopaque-1119 and 5-ml Histopaque-1083 (Millipore Sigma) gradient at 700 × g for 30 min at room temperature (RT). ACK lysis buffer was added to the last 1.5 ml of the gradient, incubated for 3 min at RT, and spun down at 500 × g for 5 min. For CFSE-labeled neutrophils, the CellTrace Far Red Cell Proliferation Kit (Thermo Fisher) was used according to the manufacturer’s instructions.

Bone marrow was isolated from C3H mouse tibias and femurs and allowed to differentiate for 6 d in RPMI 1640 supplemented with 30% L929 cell-conditioned medium, 10% FBS, and 2% penicillin–streptomycin at 37°C in 5% CO₂. Adherent cells were scraped, washed, plated, and left overnight at 37°C in 5% CO₂.

Neutrophil migration assays were performed as previously described (8). Briefly, BMN were harvested as described above and were incubated with AC at a 2:1 concentration for 2 h. Fifty nanograms per milliliter of leukotriene B₄ (LTB₄; Cayman Chemical) was added to the bottom of a 24-well transwell plate with a 3.0-μm polyester membrane (Greiner). Neutrophils were added to the top of the Transwell and allowed to migrate for 2 h. Cells were then collected from the bottom of the well, and the number of neutrophils migrating through the membrane was determined by flow cytometry and staining for CD45.2 (eBioscience) and Ly6G (BD Biosciences).

Neutrophil and macrophage phagocytosis of B. burgdorferi was determined by isolating bone marrow–derived macrophages (BMDM) and BMN as previously described and preincubating with AC at a 2:1 concentration for 1 h prior to the addition of GFP-expressing B. burgdorferi for the indicated times at a multiplicity of infection of 10. Cells were washed and stained for CD45.2 (eBioscience) for flow cytometry analysis. The percentage of cells phagocytosing B. burgdorferi was determined by the number of GFP⁺ cells over the total number of cells, expressed as a percentage.

RNA was isolated from cells using the RNeasy Mini Kit (QIAGEN) according to the manufacturer’s protocol. RNA was synthesized into cDNA using the SuperScript III First-Strand Synthesis system for RT-PCR (Invitrogen) according to the manufacturer’s instructions. Sample cDNA was then diluted to 50 ng/μl with real-time PCR performed via SYBR Green (Life Technologies). PPAR-γ cycle threshold (CT) values were normalized to GAPDH CT values, with ΔΔCT values shown as fold change compared with unstimulated macrophages.

Inflammatory cells were isolated from ankle joints as previously described (7). Briefly, ankles were removed and placed in 15-ml conical tubes containing 5 ml of 1× PBS with 4% FBS, 0.03 mg of DNase I (Millipore Sigma), and 30 μl of 100 mg/ml stock collagenase/dispase (Roche). Samples were placed on a rocker for 1 h at RT. Samples were then placed into sterile petri dishes with RPMI 1640 containing 10% FBS and flayed apart using sterile forceps. Cells were strained through a 70-μm filter (Greiner) and washed three times. Live cells were determined by counting using trypan blue exclusion. A total of 1 × 10⁶ cells were stained in a 96-well U-bottom plate (Greiner), with all samples treated with Fc Block (anti-CD16/CD32; eBioscience) for 15 min at 4°C. Cells were then stained for the following cell types: CD45.2 PE (eBioscience), F4/80 allophycocyanin (eBioscience), and Ly6G PE-Cy7 (BD Biosciences). Cells were permeabilized with saponin prior to staining with cleaved caspase-3 Alexa Fluor 488 (Cell Signaling Technology). Cells were washed and fixed in 4% paraformaldehyde for 15 min. Joint cellular infiltrate was analyzed using a Dako CyAn Flow Cytometer and Summit V5.0 software.

Ankles were harvested and snap-frozen in liquid nitrogen and homogenized in TRIzol reagent (Life Technologies) according to the manufacturer’s instructions for DNA extraction. Multiplex real-time PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems). Reactions for B. burgdorferi flagellin were normalized to copies of mouse nidogen within the same sample. Bacterial loads are expressed as copies of B. burgdorferi flagellin per 1000 copies of mouse nidogen (8).

Cytokine levels were measured from cell supernatants or ankle joint homogenates (7) using TNF-α (Applied Biosystems), IL-10 (BD Biosciences), and KC (R&D Biosystems) ELISA kits. Results were reported in picograms per milliliter or picograms per milligram protein.

The PPAR-γ agonist rosiglitazone and antagonist GW 9662 (Cayman Chemical) were used at a 10 μM concentration for all in vitro experiments. For in vivo experiments, stock rosiglitazone was diluted in sterile PBS to 1 mg/ml, and 10 μl was injected into the tibiotarsal joint synovial cavities of each mouse at 11 and 18 dpi.

In vivo experiments have groups of n = 4 and were performed in duplicate or triplicate, and in vitro experiments were performed in duplicate or triplicate with n = 3. Statistical significance for comparing multiple groups with a single control was assessed by ANOVA followed by a Dunnett test, whereas significance among multiple groups was assessed by ANOVA followed by a Bonferroni posttest. Additional data consisting of two groups used an unpaired, two-tailed Student t test. Statistical significance of differences in histology scores was determined by a Mann–Whitney U test. Significance levels are set at p < 0.05.

Infection of C3H mice with B. burgdorferi induces a transient arthritis that peaks around 3 wk pi and then undergoes spontaneous resolution. Macrophages and neutrophils make up the majority of the inflammatory infiltrate in the joint during murine Lyme arthritis (29). These phagocytic cells are efficient in their uptake and killing of B. burgdorferi in vitro (30) and in vivo (31). During in vitro cell cultures, phagocytosis of B. burgdorferi has been reported to induce apoptosis in murine BMDM (32) and in human peripheral blood monocytes (33). Quantification of AC in the joints of B. burgdorferi–infected mice has not been reported; however, clearance of these cells is likely to play an important role in mediating arthritis resolution and the return of the tissue to homeostasis. To determine how the number of AC within the infected joint changes over the course of infection, C3H mice were infected with 1 × 10⁵ B. burgdorferi and were sacrificed at various times pi. Ankle joints were harvested for flow cytometric analysis of apoptotic neutrophils and macrophages. The activation of caspase-3 into cleaved caspase-3 was used to identify AC, taking advantage of the fact that cleavage of caspase-3 is downstream of both intrinsic and extrinsic apoptotic pathways (34). Apoptotic neutrophils were identified by CD45.2⁺, Ly6G⁺, and cleaved–caspase-3⁺ cells, and apoptotic macrophages were identified by CD45.2⁺, F4/80⁺ and cleaved–caspase-3⁺ cells. The number and percentage of apoptotic neutrophils within the infected ankle joints significantly increased over the course of infection (Fig. 1A, 1C), as did the number and percentage of apoptotic macrophages (Fig. 1B, 1D). Both apoptotic neutrophils and macrophages increased significantly around day 21 pi and peaked at day 28. This correlates well with the peak of inflammation, which usually occurs around day 21 in this model. By day 35, most of the AC had been cleared, which correlates well with arthritis resolution.

Ankle swelling is a convenient way to monitor inflammatory changes within the B. burgdorferi–infected joint and generally correlates well with arthritis severity (35). The uptake and clearance of AC is a critical mechanism for resolution of inflammatory processes (36) and has spurred the development of AC as a potential therapeutic because of their anti-inflammatory effects (37, 38). To determine if AC administration would alter Lyme arthritis progression, C3H mice were infected with 1 × 10⁵ B. burgdorferi via footpad injection and administered 1 × 10⁴ AC via intra-articular ankle injection on 11 and 18 dpi. These time points were chosen to avoid downregulation of neutrophil and macrophage effector functions during times of B. burgdorferi clearance within the ankle joints (16) and to be given earlier than when the influx of apoptotic neutrophils and macrophages occurs during resolution (Fig. 1). The AC used were Jurkat cells previously treated with anti-Fas Ab as described (39). Ankle swelling, B. burgdorferi loads within the ankle, and inflammatory infiltrates and cytokines within the ankle joint were assessed at various time points throughout the infection. As shown in Fig. 2, AC or live Jurkat cell administration had little or no effect on ankle swelling in uninfected mice (Fig. 2A), whereas the addition of AC to infected mice significantly reduced ankle swelling at 21, 28, and 35 dpi compared with infected controls (Fig. 2B). The addition of live Jurkat cells had no impact on ankle swelling in infected mice throughout the infection time course (Fig. 2B). Similarly, the addition of exogenous AC did not impact the host immune response to B. burgdorferi infection, as clearance of B. burgdorferi from joint tissues within the ankle joints throughout infection were not altered (Fig. 2C). However, analysis of the joint inflammatory infiltrate on 12 and 19 dpi showed significant decreases in neutrophils in the joints of AC-treated mice compared with control-treated animals, whereas macrophage numbers remained unchanged (Fig. 2D, 2E), suggesting that the presence of increased numbers of AC can limit inflammatory processes or promote early resolution. To gain more insight into the mechanism, we attempted to measure cytokine production within the infected joint (7). TNF-α is typically produced by activated macrophages, and its expression is downregulated following uptake of AC (40). We were unable to measure TNF-α from the infected mouse joints (data not shown), similar to our previous report (7). We have similarly shown that the production of KC is critical for neutrophil recruitment and the development of Lyme arthritis (7, 8). However, we did not find an effect of AC treatment on KC levels in the joint (Fig. 2F). Efferocytosis is known to increase macrophage production of IL-10 (37) and promote resolution of inflammation. We found a trend toward higher IL-10 levels in the joints of AC-treated mice, but these did not reach statistical significance (Fig. 2G). Altogether, these data show that treatment of B. burgdorferi–infected mice with AC attenuated Lyme arthritis development and promoted disease resolution without a detrimental effect on host response.

Clearance of AC by unactivated macrophages is a normal housekeeping function and does not alter macrophage activity (41). In contrast, uptake of AC by activated macrophages decreases their production of proinflammatory mediators and increases their production of anti-inflammatory mediators, such as IL-10 (41). We investigated the effect of AC clearance on unactivated BMDM or those activated by coculture with live B. burgdorferi. Unactivated BMDM made no detectable levels of TNF-α or KC and only low levels of IL-10, and these were not influenced by the addition of AC to the culture (Fig. 3A–C). BMDM activated by coculture with B. burgdorferi, however, made high levels of TNF-α and KC and moderate levels of IL-10. Addition of AC to these cultures significantly decreased TNF-α and KC levels and increased IL-10 levels (Fig. 3A–C). Nearly identical results were seen when apoptotic neutrophils were used rather than apoptotic Jurkat cells (Supplemental Fig. 1), similar to a previous report using bleomycin-activated cells (21). Thus, even when B. burgdorferi are present, uptake of AC can significantly decrease macrophage inflammatory responses.

Next, we wanted to determine how AC impacted macrophage and neutrophil function, including phagocytosis of B. burgdorferi and chemotaxis. As shown in Fig. 4A, the presence of AC decreased the percentage of macrophages phagocytosing B. burgdorferi. This was true for neutrophils, as well (Fig. 4B), with a significant decrease in the percentage of neutrophils phagocytosing B. burgdorferi compared with controls. Live Jurkat cells had no effect on macrophage or neutrophil phagocytosis of B. burgdorferi (Fig. 4A, 4B). Additionally, we wanted to see if AC would impact neutrophil migration to the neutrophil chemoattractant LTB₄, known to play an important role in neutrophil recruitment during Lyme arthritis (16). Neutrophil incubation with AC significantly decreased the number of neutrophils migrating to LTB₄ compared with controls (Fig. 4C). These data show that AC can also alter macrophage and neutrophil function in vitro in the presence of B. burgdorferi.

The transcription factor PPAR-γ is a known negative regulator of macrophage activation and plays an important role in AC-mediated anti-inflammatory processes (21, 25). We first measured PPAR-γ mRNA levels in BMDM cultured alone or in the presence of AC (Fig. 5A). Addition of AC alone had no effect on macrophage PPAR-γ transcription levels. Activation of BMDM by coculture with B. burgdorferi alone increased PPAR-γ mRNA levels by 2-fold, but this was not statistically significant compared with BMDM alone. Addition of both B. burgdorferi and AC, however, significantly increased macrophage PPAR-γ transcription levels in vitro by almost 5-fold (Fig. 5A). We then wanted to determine if PPAR-γ expression within the ankle joint increased in vivo during B. burgdorferi infection following the administration of AC. C3H mice were infected with 1 × 10⁵ B. burgdorferi and administered AC directly into the tibiotarsal joint on 13 dpi, and PPAR-γ expression was measured 24 h later via Western blot. PPAR-γ levels significantly increased in ankles administered AC versus mock-treated controls (Fig. 5B, 5C). These data indicate that PPAR-γ is upregulated when AC are cleared during B. burgdorferi infection and could possibly explain how AC contribute to inflammation resolution.

To further confirm the role of PPAR-γ in AC alteration of immune cells, we next wanted to determine if using the PPAR-γ agonist rosiglitazone would cause a similar alteration in cytokine production of B. burgdorferi–cocultured macrophages as AC. B. burgdorferi–cocultured macrophages treated with rosiglitazone had significantly reduced levels of TNF-α (Fig. 6A) and KC (Fig. 6B) and significantly increased levels of IL-10 (Fig. 6C), which were similar to trends shown with AC (Fig. 3). In contrast, treatment of BMDM with the PPAR-γ inhibitor GW 9662 partially but significantly counteracted the effect of AC on macrophage TNF-α and KC production (Fig. 6D, 6E) and completely inhibited the AC induction of IL-10 (Fig. 6F). We next investigated the effect of PPAR-γ on neutrophil migration and macrophage and neutrophil phagocytosis of B. burgdorferi. In contrast to the inhibitory effects of AC on macrophage and neutrophil phagocytosis of B. burgdorferi, treatment of BMDM and BMN with rosiglitazone or GW 9662 had no effect on their ability to phagocytose B. burgdorferi (Fig. 7A, 7B, 7E, 7F). These data suggest that AC induce other signals besides PPAR-γ activation, which impacts bacterial phagocytosis. Similar to the effect of AC, treatment of neutrophils with rosiglitazone inhibited their migration toward LTB₄ (Fig. 7C), and treatment with GW 9662 modulated the effects of AC on neutrophil migration (Fig. 7G), suggesting neutrophil recruitment is primarily impacted by PPAR-γ–mediated mechanisms. Finally, PPAR-γ activation appears to have little effect on efferocytosis, as treatment of BMDM with either rosiglitazone or GW 9662 did not alter macrophage clearance of AC in vitro (Fig. 7D, 7H).

Finally, we wanted to determine if administration of rosiglitazone in vivo during B. burgdorferi infection would impact ankle swelling and immune cell recruitment to the ankle joints. To determine if administration of rosiglitazone would alter Lyme arthritis progression, C3H mice were infected with 1 × 10⁵ B. burgdorferi via footpad injection and administered rosiglitazone via intra-articular ankle injection on 11 and 18 dpi. Ankle swelling was significantly reduced at 21 dpi in the rosiglitazone-treated mice compared with controls (Fig. 8A), similar to treatments with AC. Additionally, both neutrophil (Fig. 8B, 8E) and macrophage (Fig. 8C, 8F) numbers were significantly reduced in the rosiglitazone-treated mice on both 28 and 35 dpi. These data show that rosiglitazone treatment decreases ankle swelling and inflammatory cell recruitment to the ankle joints during B. burgdorferi infection. Unexpectedly, rosiglitazone treatment decreased B. burgdorferi loads in the ankle joints on day 28 pi and significantly lowered spirochete loads on day 35 pi. This effect was not seen during AC treatment of B. burgdorferi–infected mice and suggests that rosiglitazone may induce bacterial clearance by other mechanisms besides PPAR-γ activation (42).

Infection of susceptible C3H mice with B. burgdorferi leads to the development of a transient inflammatory arthritis that peaks in severity after several weeks and then undergoes spontaneous resolution. As such, experimental Lyme arthritis is an excellent model to investigate mechanisms of inflammatory disease induction and resolution. Although the mechanisms that drive arthritis development in this model have been investigated for quite some time, mechanisms of arthritis resolution have received little attention. Previously, we have reported that B. burgdorferi infection of C3H mice deficient in the eicosanoid metabolic pathway enzyme 5-lipoxygenase resulted in exacerbated arthritis and a failure to resolve joint inflammation in a timely manner (16). In addition, macrophages from these mice were defective in their clearance of apoptotic neutrophils in vitro, suggesting that defective efferocytosis might be driving the failure of arthritis resolution. In the current study, we investigated the effects of AC on macrophage and neutrophil responses to B. burgdorferi in vitro and the effect of exogenous AC treatment of B. burgdorferi–infected mice on Lyme arthritis. Our results suggest that clearance of AC likely plays an important role in mediating arthritis resolution by decreasing proinflammatory cytokine production, increasing the production of IL-10, and inhibiting the recruitment of neutrophils, all via the activation of PPAR-γ.

During B. burgdorferi infection of mice, neutrophils and monocytes are recruited to the infected joint, where they help clear infecting spirochetes and then undergo apoptosis and are cleared by macrophages (33). We measured levels of AC in the joints of B. burgdorferi–infected mice and showed that AC numbers roughly coincided with arthritis development and resolution, as measured by ankle swelling. This result suggests that AC clearance may play a role in arthritis resolution, as defects in timely clearance of AC appear to delay disease resolution in experimental Lyme disease (16), K/BxN serum-transfer arthritis (43), and gout (44). In support of this, we treated B. burgdorferi–infected mice with AC during the development and near the peak of disease. This treatment significantly reduced joint swelling and led to earlier resolution in some animals. In addition, AC treatment led to a significant decrease in neutrophil infiltration into the infected joint, suggesting a decrease in neutrophil recruitment, an increase in neutrophil clearance, or both leading to decreased arthritis severity and enhanced arthritis resolution. Although AC treatment of macrophages resulted in significant alterations in cytokine production in vitro, we could not demonstrate this in vivo, although there was a trend in higher IL-10 levels in the joints of AC-treated animals. This may be partially due to the limited number of AC that could be injected directly into the tibiotarsal joint cavity (1 × 10⁴ AC). Others have reported greater effects of AC on arthritis severity using autoimmune arthritis models (collagen-induced arthritis or methylated BSA) using 10⁷ or 10⁸ AC delivered via i.v. or i.p. injection prophylactically (45, 46) or therapeutically (47). Rather than looking for a way to deliver greater numbers of AC, we are working to develop a cell-free mechanism to promote arthritis resolution.

AC administration has been shown to be beneficial in other noninfectious models of disease, such as lung fibrosis (21) or carrageen mouse paw edema (24). However, in other models of infectious disease, AC efferocytosis can be beneficial or harmful to the host depending on its effect on bacterial clearance. Phagocytosis of Escherichia coli or Staphylococcus aureus promote apoptosis of the phagocytosing cells and their efficient clearance to aid in inflammation resolution (48). However, efferocytosis of apoptotic neutrophils containing Chlamydia pneumoniae allows for the bacterium to infect macrophages via a “Trojan horse” mechanism (49). Phagocytosis of B. burgdorferi by mouse BMDM or human monocytes has been shown to induce apoptosis in those cells (32, 33). In this study, we show that coculture of macrophages or neutrophils with AC led to a decrease in their uptake of B. burgdorferi. However, in vivo treatment of mice with AC had no significant effect on spirochete numbers within the joint. This could be due to the low numbers of exogenous AC injected, or perhaps the differences in kinetics between recruited neutrophils and AC ensures that most of the spirochetes are cleared prior to the induction of neutrophil apoptosis.

To investigate the mechanism of AC induction of arthritis resolution, we added AC to macrophage cultures and measured cytokine production. Unstimulated BMDM made no KC or TNF-α and only very low levels of IL-10, and addition of AC had no effect on these levels. Activation of BMDM with B. burgdorferi led to high levels of TNF-α and KC production and modest levels of IL-10, similar to previous reports using murine (50) and human cells (33). Addition of AC caused a decrease in TNF-α and KC levels and an increase in IL-10, as previously described for human monocyte-derived macrophages (40). We also showed that macrophage and neutrophil phagocytosis of B. burgdorferi and neutrophil chemotaxis in response to LTB₄ were decreased by coculture with AC. Previous studies have linked AC effects on inflammation resolution to increased activation of PPAR-γ (21, 24, 51). PPAR-γ has a variety of functions, including glucose metabolism, promoting adipogenesis, and regulating fatty acid synthesis and storage, but more recently it has been linked to inhibition of inflammatory signaling through NF-κB (52). We show that AC treatment induced PPAR-γ expression both in vitro and in vivo and that treatment of cells or mice with the PPAR-γ agonist rosiglitazone mimicked the effects of AC treatment, and this could be blocked by treatment with the PPAR-γ antagonist GW 9662.

Together, our results show that AC are in abundant number during the time of inflammation resolution during Lyme arthritis, and these cells are able to alter macrophage and neutrophil phenotype from proinflammatory to proresolution. Additionally, the administration of the PPAR-γ agonist rosiglitazone during Lyme arthritis elicits similar effects as AC by decreasing immune cell recruitment. These results suggest PPAR-γ agonists, such as rosiglitazone, may be useful to ameliorate Lyme arthritis and induce resolution in individuals with posttreatment Lyme disease syndrome.

The authors have no financial conflicts of interest.