Initiation of the innate sterile inflammatory response that can develop in response to microparticle exposure is little understood. Here, we report that a potent type 2 immune response associated with the accumulation of neutrophils, eosinophils and alternatively activated (M2) macrophages was observed in response to sterile microparticles similar in size to wear debris associated with prosthetic implants. Although elevations in interleukin-33 (IL-33) and type 2 cytokines occurred independently of caspase-1 inflammasome signalling, the response was dependent on Bruton’s tyrosine kinase (BTK). IL-33 was produced by macrophages and BTK-dependent expression of IL-33 by macrophages was sufficient to initiate the type 2 response. Analysis of inflammation in patient periprosthetic tissue also revealed type 2 responses under aseptic conditions in patients undergoing revision surgery. These findings indicate that microparticle-induced sterile inflammation is initiated by macrophages activated to produce IL-33. They further suggest that both BTK and IL-33 may provide therapeutic targets for wear debris-induced periprosthetic inflammation.
Prosthetic implants for joint reconstruction are widely used; however, as many as 15% of joint replacements will fail, often requiring revision surgery1. The main cause is likely to be microparticles (MPs) released from prosthetic devices, commonly referred to as wear debris, that are thought to promote localized sterile inflammation leading to pain, osteolysis, loosening of the implant and ultimately failure of fixation. The immune signalling pathways through which these apparently inert solid MPs promote such harmful inflammation remain unclear and their characterization may reveal new targets for the development of effective therapies2,3,4. As macrophages are often associated with wear debris in periprosthetic tissue, these myeloid cells have been proposed to be essential players5. Also, in vitro cultures have indicated that wear debris can promote macrophage activation and inflammatory cytokine production, including TNF-α and IL-1β (refs. 2,6), with a number of reports concluding that MPs similar to wear debris may trigger type 1 immune responses. However, this remains controversial as the presence of endotoxin in some MP preparations or in infection-associated implants may skew the response7,8,9, obscuring sterile responses associated with the MPs themselves.
We have previously reported that endotoxin-free MPs, similar in size to wear debris particles, also induce a type 2 innate response with increases in M2 macrophages, neutrophils and eosinophils through Toll-like receptor 4 (TLR4)-independent pathways10. These findings are consistent with studies of other sterile particulate inert structures, including monosodium urate crystals11 and silica12, where type 2 immune responses were also observed. Here, we now interrogate the mechanisms that initiate innate sterile type 2 inflammation induced by metallic MPs of a similar size and composition to wear debris particles. Our studies reported herein indicate that BTK signalling specifically upregulates macrophage production of IL-33, which is required for the initiation of the MP-induced type 2 immune response. These studies thus indicate a critical role for IL-33-producing macrophages in driving the initiation of the sterile inflammatory response to inert particulates.
Results and discussion
MPs induce a caspase-1-independent type 2 innate response
We have reported that micrometre-sized titanium particles, similar in size to wear debris particles associated with prosthetic implants, induce a robust type 2 innate response independently of TLR4 by 48 h after inoculation10. To assess whether this response was characteristic of micrometre-sized metallic particles generally, C57BL/6 (BL/6) mice were inoculated intraperitoneally with either vehicle or similarly sized solid inert cobalt chrome alloy MPs (0.3–100 µm), which is a commonly used implant material2 and is used in all experiments described in this Article. At 2 days after MP inoculation, peritoneal exudate cells (PECs) were analysed for innate immune cell markers. Neutrophils (CD11b+, Ly6G+), eosinophils (c-Kit−, Siglec-F+) and M2 macrophages (F4/80+, CD206+) were increased (Supplementary Fig. 1A,B) following MP inoculation. Gene expression analysis showed that type 2 cytokines (Il-4, Il-5, Il-13), alternatively activated (M2) macrophage markers (Arg1, Retnla, Chil-3), cytokine alarmins (Il-25 and Il-33) and the type 2-associated Notch ligand, Jag1, were all upregulated in PECs exposed to MPs. In contrast, genes associated with innate type 1 or type 17 responses (Il-12, Inos, Delta4 and Il-17) were not increased, indicating polarization towards a type 2 immune response (Supplementary Fig. 1C) as early as day 2 after inoculation. Interestingly, modest increases in interferon-γ (IFN-γ) were also detected, probably derived from natural killer or other innate lymphoid cells.
To assess whether the caspase-1-dependent inflammasome might contribute to MP-induced inflammation, caspase-1-/- mice were inoculated with MPs. Increases in eosinophils and M2 macrophages were sustained; however, decreases in neutrophils were evident in caspase-1-/- mice when compared with inoculated wild-type controls (Supplementary Fig. 1A,B). Intriguingly, similar increases in type 2 cytokines were observed in caspase-1-/- mice (Supplementary Fig. 1C). Although caspase-1 generally promotes functional IL-1β activity13, increases in caspase-1-independent IL-1β have also been described in response to Mycobacterium tuberculosis infection14. However, our findings showed that IL-1β elevations were blocked in MP-inoculated caspase-1-/- mice, as shown in Supplementary Fig. 1D. In a separate experiment, using CC chemokine receptor 2 (CCR2) reporter mice, approximately 40% of the peritoneal macrophages were CCR2+ at 2 days after MP inoculation, indicating that they were blood monocyte-derived (Supplementary Fig. 1E). The MP-mediated innate immune response was sufficiently robust to function as an adjuvant that could prime in vivo Ag-specific DO11.10 T helper 2 (Th2) cell differentiation (Supplementary Fig. 2). Taken together, these findings indicated that blockade of the caspase-1-dependent inflammasome has little effect on the potent type 2 innate immune response to sterile inert MPs.
The MP-induced type 2 response is SYK dependent
In vitro studies have implicated the spleen tyrosine kinase (SYK) pathway in particulate-induced inflammation, triggered through receptor-independent cell membrane perturbations associated with frustrated phagocytosis15,16. Furthermore, inhibition of SYK signalling can block Ag-specific serum immunoglobulin-E (IgE) elevations following OVA-alum immunization12. To interrogate the role of SYK in MP-induced inflammation, we administered Bay 61–3606, a potent and selective SYK inhibitor17. Mice were orally gavaged with either Bay 61–3606 or vehicle twice per day starting 2 days before intraperitoneal MP inoculation. At day 2 after MP inoculation, peritoneal cells were analysed for immune cell phenotypes by fluorescence-activated cell sorting (FACS) analysis and cytokine gene expression by quantitative polymerase chain reaction (qPCR). MP-inoculated mice administered Bay 61–3606 showed marked reductions in neutrophils, eosinophils and M2 macrophages. Decreases in PEC type 2 cytokines, including Il-4, Il-5 and Il-13 mRNA, were also observed. In addition, MP-induced elevations in the cytokine alarmin, IL-33, known to trigger type 2 responses18, were also blocked, as measured by PEC Il-33 mRNA and peritoneal fluid (PF) protein levels (Supplementary Fig. 3A–D). PECs from MP-inoculated mice also showed elevations of activated pSYK (Y348), which was blocked by Bay 61–3606 (Supplementary Fig. 1E). As BTK signalling can be SYK dependent19, we also measured BTK phosphorylation. Enhanced PEC BTK phosphorylation was blocked by administration of Bay 61–3606 (Supplementary Fig. 3F). Thus, SYK inhibition generally inhibited immune cell infiltration and elevations in IL-33 and type 2 cytokines.
BTK drives type 2 inflammation independently of B cells
SYK signalling involves multiple downstream pathways, including BTK activation (Y551)20. To specifically examine the role of the BTK pathway, we used ibrutinib, which irreversibly inhibits BTK21,22,23. Neutrophils, eosinophils and M2 macrophages were reduced in MP-inoculated groups treated with ibrutinib. BTK blockade also inhibited M2 markers, type 2 cytokines, and Il-33 mRNA, PF IL-33 protein, and activated pBTK (Y551). However, ibrutinib administration did not alter MP-mediated activation of pSYK (Y348), consistent with its activation being downstream of SYK signalling (Supplementary Fig. 4). Thus, ibrutinib inhibited immune cell infiltration and elevations in type 2 cytokines and IL-33.
As BTK is associated with B cell activation and antibody-mediated signalling in innate immune cell populations19, we next examined whether B cells were required for MP-induced inflammation. BALB/c control mice or B-cell-deficient Jh-/- mice were inoculated with MPs and assessed at 48 h. Inflammation was similar in both strains (Supplementay Fig. 5A–C), indicating that MP-induced inflammation is independent of B cells. B cells were undetectable in Jh-/- mice (Supplementary Fig. 5D). As ibrutinib can also block T cell receptor (TCR) interleukin-2 inducible kinase (ITK) signalling24, we depleted T cells with an anti-CD4 (GK1.5) antibody, as described previously25, and inoculated mice intraperitoneally with MPs. CD4 T cell depletion had no effect on the MP-induced immune response (Supplementary Fig. 6A–C). Thus, neither B cells nor T cells are essential for this BTK-dependent MP-induced type 2 response.
To further investigate the role of BTK signalling in this innate inflammatory response, we utilized mutant CBA/Nxid (BTK deficient) mice, which have a BTK loss of function mutation26,27. CBA/CaJ wild-type or BTK-deficient mice were inoculated intraperitoneally with either vehicle (PBS) or MPs. As shown in Fig. 1a,b, MP-induced increases in innate immune cell populations were blocked in BTK functional knockouts compared with inoculated wild-type mice. Gene expression of IL-33, type 2 cytokines and markers for M2 macrophages were also decreased (Fig. 1c). Although constitutive loss of BTK function triggers an intrinsic B-cell defect in BTK-deficient mice28, our findings that the response is intact in B-cell-deficient Jh-/- mice suggests B-cell-independent requirements for BTK in the MP-induced type 2 innate immune response.
IL-33-producing macrophages drive the type 2 MP response
The early BTK-dependent induction of IL-33 raised the possibility that this cytokine alarmin may promote the MP-induced type 2 immune response. Although not previously linked to sterile type 2 inflammation, IL-33 can drive type 2 responses in the context of helminth infection and allergic responses18,29. To determine the cellular sources of MP-induced IL-33 and associated type 2 cytokines, neutrophils (CD11b+, Ly6g+), eosinophils (Ly6g-, F4/80−, Siglec-F+) and macrophages (Ly6g−, Siglec-F-, F4/80+) were sort-purified from PECs for gene expression analysis at 2 days post MP inoculation. Macrophages were the main source of IL-33, while type 2 cytokines were differentially expressed by myeloid cells with neutrophils expressing elevated levels of IL-1β and IL-13 and eosinophils expressing higher levels of IL-4 and IL-5 (Fig. 2a,b). Non-myeloid cells, which included any innate lymphoid cells, did not express significant levels of type 2 cytokines.
IL-33 may exert its biological effects by binding to its cell surface receptor, ST2, or as a nuclear transcription factor regulating the transcription of diverse inflammatory genes18,30. To examine the role of IL-33 in driving the MP-induced type 2 response, we blocked IL-33 function by directly neutralizing IL-33 with a recombinant soluble Fc-ST2 chimeric molecule 5 h before and 24 h after MP inoculation31. As shown in Fig. 2c–e, neutralization of IL-33 signalling blocked the inflammatory response 48 h following MP inoculation. To confirm these results using a genetic approach, wild-type or IL-33-receptor-deficient (ST2-/-) mice were inoculated intraperitoneally with MPs and 2 days later PECs were collected and analysed using flow cytometry and qPCR. ST-2 deficiency blocked MP-mediated type 2 inflammation (Supplementary Fig. 7A,B). To assess whether the type 2 response can be rescued by exogenous IL-33 administration when BTK signalling is inhibited, we administered recombinant IL-33 (1 μg day–1) to ibrutinib-treated MP-inoculated mice and assessed inflammation at 48 h. As shown in Fig. 3a–c, IL-33 restored the innate type 2 responses. This rescue experiment thus indicates that BTK signalling is not required downstream of IL-33, consistent with a model where BTK triggers IL-33 that then initiates the type 2 innate response.
To investigate whether BTK signalling was also necessary for the helminth-induced type 2 immune response, BTK-deficient and control mice (wild-type mice) were orally inoculated with 200 L3 of the murine intestinal nematode parasite, Heligmosomoides polygyrus. Elevations in myeloid cells, type 2 cytokines and IL-4 bioactivity, as measured by mesenteric lymph node B cell major histocompatibility complex class II (MHC II) elevations32, were not impaired (Supplementary Fig. 8). Our findings thus indicate that this robust type 2 response to multicellular parasites utilizes other BTK-independent pathways for initiation of the response. This alternative pathway triggered during helminth infection may involve epithelial cells and also ILC2 cells. Indeed, recent studies indicate that IL-33, thymic stromal lymphopoietin (TSLP) and IL-25 are redundant in some type 2 response models, including the immune response to Schistosoma mansoni, necessitating blockade of all three alarmins to inhibit the type 2 response33. Apparently, initiation of MP-induced inflammation is instead dependent on BTK and associated IL-33 induction, raising the possibility that blockade of BTK or IL-33 may preferentially inhibit the response to MPs, leaving other immune responses, including those elicited by infectious agents, intact.
Adoptive transfer restores type 2 inflammation
Our findings indicated that after MP inoculation, IL-33 elevations and downstream type 2 immune responses were BTK-dependent and that the main source of IL-33 was macrophages. To assess whether BTK signalling in macrophages could restore IL-33 elevations and associated type 2 immunity in BTK-deficient mice, 1 × 106 sort-purified peritoneal macrophages, either from BTK-deficient or wild-type mice, were adoptively transferred into recipient BTK-deficient mice 24 h before MP inoculation. Analysis of the inflammatory response 48 h after MP inoculation showed significant increases in the immune cell infiltrate and cytokine responses in recipient BTK-deficient mice with transferred wild-type macrophages. In parallel experiments, transfer of macrophages from BTK-deficient mice into recipient BTK-deficient mice had little effect on the inflammatory response after MP inoculation (Fig. 4a–c). Thus, BTK signalling and associated IL-33 production in macrophages are sufficient to drive the development of sterile inflammation in response to MP inoculation. Epithelial cells are generally thought to be critical in triggering type 2 immunity through their release of IL-33 and other alarmins29,34. Our studies now reveal that BTK signalling in macrophages induces their production of IL-33 and consequent initiation of the type 2 immune response. Previous studies have shown that activated macrophages can produce IL-33 (refs. 35,36,37), although the significance of this observation in terms of immune response initiation has remained uncertain.
MPs directly promote IL-33 production by macrophages
To examine whether MPs could directly drive macrophage production of IL-33, highly purified peritoneal macrophages from untreated mice were cultured in vitro with MPs and PF from MP-inoculated mice. As shown in Supplementary Fig. 9, although MPs or PF alone had modest effects, the combination promoted marked increases in IL-33. To examine whether MP interactions with the macrophage, including phagocytosis, were involved, cytochalasin D, which inhibits the assembly of actin filaments and phagocytosis38,39, was added to the cultures. Interestingly, cell death was significantly increased following culture of macrophages with MPs, which was blocked by cytochalasin D. Physical interactions of the macrophage with the particle and IL-33 production were also markedly reduced. To assess whether cell death generally was important in triggering IL-33 production, inhibitors of necrosis (Necrox2)40,41, and apoptosis (ZVED-FMK)42, were added. These inhibitors blocked MP-induced increases in IL-33. Taken together, these results indicate that phagocytic and other physical interactions of macrophages with MPs trigger cell death pathways that promote IL-33 production.
MP-induced inflammation in articular tissues
Although the peritoneal model provides a readily studied approach to investigate inflammatory responses to MPs, the response in joint tissue may be different and more clinically relevant. To investigate MP-induced inflammation in localized joint tissue we developed an experimental model, based on previous studies43,44,45, where MPs are injected directly into the knee intra-articular region with 10 mg of MPs in a 20 μl volume. To examine the role of the SYK pathway, Bay 61–3606 was administered by oral gavage twice per day starting 2 days before bilateral intra-articular MP inoculation in the knee. As seen from haemotoxylin and eosin (H&E)-stained slides and FACS analysis, 2 days after MP administration a robust inflammatory infiltrate was observed, which was inhibited by Bay 61–3606 (Fig. 5a). We also examined effects of BTK signalling. Inhibition of BTK with ibrutinib markedly reduced synovial inflammation (Fig. 5b). To investigate whether MP-induced joint inflammation was also dependent on IL-33, we administered intra-articular injections of either FcST2 or control IgG at day 0 and 1 after MP inoculation. Treatment with Fc-ST2 significantly reduced MP-induced joint inflammation (Fig. 5c). Difficulty with isolation of viable immune cells from the murine joint tissue prevented reliable assessment of cytokine gene expression. Taken together, these data show that MP inoculation induces an inflammatory response in synovial tissues, similar to that seen in our peritoneal model and also dependent on SYK, BTK and IL-33 signalling. Recent studies indicate that different tissue microenvironments, which include tissue-resident macrophages and other immune and non-immune cells not present in peripheral blood, can support distinct responses46,47. Apparently, immune stimulation in these different in vivo tissue milieus were sufficiently similar to support the development of a common innate response.
MPs promotes chronic inflammation and tissue damage
To more closely model the clinical case of articular tissues being repeatedly exposed to wear debris, we developed a repeated-injection model of MP-induced inflammation. In this model, bilateral intra-articular knee injections of MPs were given to C57BL/6 mice once per week over the course of 14 days. With this model, we examined both MP-induced inflammation and potential pathology in peri-articular tissues. As shown in Supplementary Fig. 10, repeated MP inoculations result in robust inflammation, including persistent presence of neutrophils and M2 macrophages and associated fibrosis. In addition, the sustained chronic inflammation and fibrosis was associated with loss of both cartilaginous and osseous tissues. Taken together, these data suggest that a localized chronic type 2 innate immune response occurs in tissues repeatedly exposed to ‘wear debris’ MPs.
Failed human joint arthroplasty reveals type 2 inflammation
Immunohistology was performed on periprosthetic and peri-articular tissue samples from patients undergoing revision total joint replacement or primary total joint arthroplasty. Multiple criteria were used to exclude any patients with septic tissues. H&E staining (Fig. 6a,b) showed monocytic immune cell infiltration surrounding wear debris in periprosthetic tissue. Dual immunofluorescent staining for CD68 (macrophages) and CD206 (mannose receptor (MRC1)) (Fig. 6c,d), or CD163 (Fig. 6e,f) (both M2 markers), or iNOS (Fig. 6g,h) (M1 marker) revealed increases in M2 macrophages in tissues from revision arthroplasty with increases in M1 macrophages predominantly in patients undergoing primary joint replacement.
Inflammatory response markers were assessed by qPCR. As shown in Fig. 6i, M2 macrophage markers including insulin-like growth factor 1 (IGF-1), fibronectin 2 (FN2), Arg1, CCL18, MRC1 and chitinase-1 (CHIT1) were increased in patients undergoing revision surgery. Elevations in the type 2 cytokine, IL-13, and IL-33 were also detected in this group. These factors are characteristically activated in type 2 responses and human M2 macrophages48. In contrast, IL-12, CXCR4, or iNOS expression, characteristic of M1 macrophages, were not elevated. Rigorous exclusion of patients with associated infections thus revealed the underlying type 2 inflammation triggered by implants.
In summary, our findings show that MPs induce cell death pathways in macrophages, which then express IL-33, thereby acting as essential players in the initiation of the type 2 sterile inflammatory response (Supplementary Fig. 11). Blockade of conventional signalling pathways associated with the type 1 inflammatory response, including caspase-1 signalling, may have little effect on these responses. Instead our findings indicate that BTK signalling in macrophages initiates the sterile inflammatory response through its upregulation of the cytokine alarmin IL-33. Furthermore, these results suggest that BTK and IL-33 are potential therapeutic targets for controlling inflammatory responses triggered by MPs, including wear debris particles contributing to failure of fixation in orthopaedic implants.
BALB/c, C57BL/6 J, CBA/CaHN-Btkxid/J and CBA/CaJ mice were obtained from the Jackson Laboratory (Bar Harbor, ME USA). Breeding pairs of DO11.10 TCR transgenic BALB/c mice were obtained from The Jackson Laboratory and housed in a specific pathogen-free facility during the experiments at Rutgers – New Jersey Medical School research animal facility, Newark, NJ. Caspase-1 gene knockout mice were obtained from T.-D.K. (St. Jude Hospital, Memphis, TN, USA) and B-cell-deficient Jh–/– mice on BALB/c background were obtained from TACONIC Laboratory (Hudson, NY, USA). All mice for our experiments at Rutgers – New Jersey Medical School research animal facility were housed in a specific pathogen-free, barrier facility excluding common adventitious pathogens. The studies have been reviewed and approved by the Institutional Animal Care and Use Committee at Rutgers—the State University of New Jersey. Experiments with ST2-/- mice were performed at the Children’s Hospital of Philadelphia, USA with the approval of The Children’s Hospital of Philadelphia Institutional Animal Care and Use Committee. The experiments herein were conducted according to the principles set forth in the Guide for the Care and Use of Laboratory Animals, Institute of Animal Resources, National Research Council, Department of Health, Education and Welfare (US National Institutes of Health).
Dose and administration of MPs, inhibitors and blocking antibody
Cobalt chrome MPs were obtained from Sandvik Osprey Ltd (Neath, Wales, UK). Sterile particles were prepared and redissolved in sterile PBS, as previously described10. These particles were separated using graded ethanol solution for sedimentation and analysed by low-angle laser light scattering with a Microtrac X-100 (Bioengineering Solutions, Chicago, IL, USA). MPs ranging from 0.3 to 100 µm were used and a dose of 50 mg per mouse was given intraperitoneally for in vivo studies in mice. Bay 61–3606 was obtained from Sigma-Aldrich (St. Louis, MO, USA) and dissolved in sterile PBS (vehicle) and was administered to C57BL/6 female mice (5–6 weeks old) by oral gavage twice a day at 10 mg kg–1 body weight starting 2 days before inoculation of MPs. Ibrutinib was commercially obtained from Selleck Chemicals (Houston, TX, USA) and was dissolved in a vehicle consisting of 30% polyethylene glycol + 1% Tween80 + 1% DMSO and filtered through a 0.22 pore size filter (Durapore) obtained from Millipore (Billerica, MA, USA). Ibrutinib at a dose of 10 mg kg–1 body weight was delivered by oral gavage with a gavage-feeding needle once a day starting 1 day before MP inoculation. Recombinant IL-33/vehicle (1 μg day–1) (R&D systems, Minneapolis, MN, USA) was administered (intraperitoneally) to ibrutinib-treated mice. To abrogate in vivo CD4+ T-cell function, 1 mg of anti-CD4 mAb (monoclonal antibody) (GK1.5) (BioXcell, West Lebabon, NH, USA) was given intraperitoneally 1 day before MP inoculation. CD4+ T-cell depletion was confirmed by flow cytometry of lymph node cell suspensions. To block the IL-33 signalling pathways, we administered MP-inoculated mice with 7.5 μg of either recombinant human IgG1FC (isotype Fc) or recombinant mouse ST2/IL-33RFc (R&D systems, Minneapolis, MN, USA) intraperitoneally, 5 h before MP inoculation and again another dose on day 1 after MP inoculation. IL-33 signalling in the knee synovium was blocked by combined intra-articular (5 μg) and peritoneal (2.5 μg) administration of either Fc-ST2 or control IgG at day 0 and day 1 after MP inoculation.
PEC suspensions were blocked with Fc Block (BD Biosciences, San Jose, CA, USA) and subsequently stained with specific antibodies including: anti-Ly6G FITC, anti-CD11cPE, anti-CD11b PerCP/CY5.5, Siglec-F PE, anti-c-kit APC, F4/80 APC, anti-CD4APC, anti-CD19 PerCP/Cy5.5, CD3PE (BD Biosciences, San Jose, CA, USA) and anti-CD206PE or anti-CD206 Alexa Fluor 488 (Biolegend, USA). Cells were collected 4 h after MP inoculation for analysis of phosphorylation of BTK (Y-551) or SYK (Y-348). Intracellular staining was performed using anti-mouse phospho-BTK/ITK (Y551/Y511) PE (eBiosciences, San Diego, CA, USA), anti-mouse phospho-SYK (Y-348) PE antibody (BD Biosciences, San Jose, CA, USA) and FOXP3/Transcription Factor Staining Buffer Set, which was used as per the manufacturer’s instructions (eBiosciences, San Diego, CA, USA). For carboxyfluorescein succinimidyl ester (CFSE)-labelled cells, anti-CD4 PerCP (BD Biosciences, San Jose, CA, USA) and KJ1–26 PE (BD Biosciences, San Jose, CA, USA) were used to phenotype DO11.10 T cells and cell-cycle progression was examined as previously discussed10. For FACS analysis of whole synovial tissues, samples were digested at 37 °C for 1 h with 0.1% collagenase in RPMI supplemented with 10% FBS, 1,000 U ml–1 penicillin, 1,000 U ml–1 streptomycin and 2 mM l-glutamine. Single-cell suspensions were then prepared and blocked and stained as described above.
Sorting and adoptive transfer of peritoneal innate cell populations
For gene expression analysis of peritoneal innate cell populations, PECs were harvested from C57BL/6 wild-type mice and were separated through electronic sorting employing FACSAria II (BD Biosciences, San Jose, USA) (see Supplementary Fig. 12 for gating strategy). For adoptive transfer experiments, macrophages were purified from wild-type (CBA/J) or BTK mutant mice (CBA/NXID) with biotin anti-mouse F4/80 antibody (Biolegend, USA) and anti-biotin beads obtained from Miltenyi Biotec, USA. Twenty-four hours before MP inoculation, 1 × 106 peritoneal macrophages from either XID or wild-type mice were transferred to XID recipient mice.
In vitro MP culture with sorted macrophages
PECs were collected from C57BL/6 wild-type mice and macrophages were separated (>95% pure) using biotinylated anti-mouse F4/80 antibody (Biolegend, USA) followed by anti-biotin beads (Miltenyi Biotec, USA). 0.25 million purified macrophages were cultured with 0.25 mg MPs in media containing 2% PF harvested from MP-inoculated mice. In some cultures, inhibitors of necrosis (30 μM Necrox2 (Enzo Life Sciences, Inc., USA)), apoptosis (100 μM ZVED-FMK (R&D Systems, Inc., USA)) and phagocytosis (2 μM cytochalasin D (Enzo Life Sciences, Inc., USA)) were added. After 48 h, supernatants were assayed for IL-33 as described above (IL-33 bioassay). The percentage of necrosis was determined by staining cells with acridine orange and propidium iodide (AOPIstain, Nexcelom Bioscience, USA) and cells were assayed for viability by using Cellometer Auto2000 (Nexcelom Bioscience, USA). To assess MP–macrophage interactions following cytochalasin D administration, the per cent of macrophages with associated MPs was determined using images (40×) obtained with the EVOS XL Core microscope (Life Technology, USA).
Adoptive transfers of D011.10 TCR CD4+ T cells
Spleen and lymph nodes were collected from wild-type DO11.10 TCR-transgenic mice. DO11.10 OVA-specific TCR transgenic CD4+ cells were isolated, CFSE-labelled, and adoptively transferred to recipient BALB/c mice intravenously, as previously described10. Thirty micrograms OVA peptide alone or with either 50 mg MPs or with 4 mg alum were injected intraperitoneally into the recipient mice 2 days after transfer. HPLC−purified OVA323–339 (OVA peptide) was synthesized at the Rutgers – New Jersey Medical School, Molecular Resources Facility (MRF).
Intra-articular immunization with MPs
Mice were given bilateral intra-articular knee injections with either vehicle or 10 mg of MPs in a 20 μl volume. Knee tissues were collected after 2 days, fixed in 10% buffered formalin, and decalcified by EDTA. The slides were examined using a Zeiss Axioscope2 and the images were captured and analysed by Axiovision (Thornwood, NY, USA) software in 100× magnification. The area of immune cell infiltration and the total synovial tissue area were measured using Axiovision software. We also developed a repeated-injection model of MP-induced chronic inflammation. In this model, intra-articular MP inoculations of the knee were given once per week over the course of 14 days. For pathologic analysis, formalin-fixed tissue samples were decalcified with EDTA, sections were picrosirius red stained and digitally imaged to assess the level of fibrosis. For cartilage analysis, sections were Safranin-O stained. For analysis of changes to osseous tissue and inflammation, H&E stained slides were used.
Parasite inoculation and characterization of cells
Two hundred infective Hp L3 were inoculated in mice using a rounded gavage tube, and at day 7, peritoneal cells and mesenteric lymph nodes were harvested. Flow cytometric analysis of PECs was performed as described above for MPs. For B cells, expression of MHC II was monitored by flow cytometery using B220, CD3 and MHC II from BD Biosciences, and used as per the manufacturer’s instructions. Mesenteric lymph nodes were also analysed for gene expression as described above.
Human tissue samples from primary and revision arthroplasty were fresh frozen and sectioned at 5 µm thickness with an HM505E cryostat. Samples were fixed in acetone at –20 °C and blocked using Protein Block (Dako, Carpinteria, CA, USA), FcX (Biolegend, San Jose, CA, USA) and 10% normal goat serum (Jackson Immunoresearch, West Grove, PA, USA). The primary antibodies used were mouse anti-human CD68, biotin anti-CD206, biotin anti-CD163 (Biolegend, San Jose, CA, USA) and rabbit anti-human iNOS (Abcam, Cambridge, MA, USA). The secondary antibodies used were Alexa Fluor (AF)488-conjugated goat anti-mouse, AF555-conjugated goat anti-rabbit and streptavidin Rhodamine Red-X (Invitrogen, Eugene, Oregon, USA). Nuclear staining was performed using Vectashield with DAPI (Vector Laboratories, Burlingame, CA, USA). Digital images were obtained with a Leica DM6000B and a Hamamatsu Orca-Flash 4.0 camera and tiled using Leica Application Suites Advanced Fluorescence (Leica Microsystems, Buffalo Grove, IL, USA). Fluorescence intensities and exposure times were normalized to appropriate control images.
Single-cell lymph node suspensions were cultured with 10 µg ml–1 OVA peptide for 3 days on anti-IL-4 (Clone BVD4-1D11.2, BD Biosciences, San Jose, CA, USA) or anti-IFN-γ (BD Biosciences, San Jose, CA, USA) coated plates, and after 72 h plates were washed, incubated overnight with secondary biotinylated anti-IL-4 or anti-IFN-γ antibodies, and then developed with streptavidin–alkaline phosphatse (Jackon Immuno Research Laboratories, West Grove, PA, USA) as previously described10. Results were quantitated by counting the number of positive colonies under a microscope.
Bioassays of IL-1β and IL-33 in PF
Four hours after MP inoculation, the peritoneum was flushed with 1 ml of sterile PBS and the cell suspension was centrifuged and the supernatant was collected and then stored at –80 °C. Enzyme-linked immunosorbent assay (ELISA) was performed using either IL-1β or IL-33 Ready-Set-Go! (eBioscience, San Diego, CA, USA) as per the manufacturer’s instructions.
CD4+ and DO11.10 T-cell sorting
Single-cell suspensions from lymph nodes of DO11.10 mice were incubated with anti-CD4 microbeads (Miltenyi Biotec, San Diego, CA, USA) and CD4+ T cells were sort purified. The KJ1-26+ population was then positively selected as previously described10. Sorted populations were 85–90% pure.
Quantitation of cytokine gene expression
Total RNA was extracted from PECs with Trizol (Invitrogen, Eugene, Oregon, USA). RNA extraction from human and murine synovial tissues was performed using RNA-Bee (Amsbio, Milton, Abingdon, UK). Isolated RNA was reverse transcribed using Super Script II reverse transcriptase (Invitrogen, Eugene, Oregon, USA). The gene specific Taqman assay (Applied Biosystems, Waltham, MA, USA) was used for amplification and detection of different genes with Applied Biosystems 7500 sequence detector. Gene expression fold changes of different mRNA of treated samples were quantified relative to the untreated samples after normalizing to 18s RNA, as previously described10.
Collection and analysis of clinical tissue samples
Prospective clinical studies were approved by an Institutional Review Board (IRB) protocol. Periprosthetic and peri-articular tissue samples were collected from human subjects undergoing revision total joint replacements or primary total joint arthroplasty, respectively. Primary joint arthroplasty samples served as controls for all experiments. Multiple criteria were used to exclude septic loosening, using previously described methods49, including analysis of erythrocyte sedimentation rates, C-reactive protein (CRP) levels, and culturing of tissue and joint aspirates. Retrieved tissues were frozen for subsequent cryosectioning or homogenized for RNA purification.
The mean ± standard error of the mean (s.e.m.) is indicated in relevant figures. Statistical analyses included ANOVA followed by individual two-way comparisons with Tukey test or paired t-test using Graphpad Prism version 6.0 (Graph Pad Software Inc., La Jolla, CA, USA). Statistical differences at a level of P < 0.05 between groups was considered significant.
Institutional Review Board (IRB) study approval
Proposed human studies for this investigation were approved by the Rutgers – New Jersey Medical School IRB. All investigations were conducted in conformity with ethical principles of research, and informed consent for participation in the study was obtained. All investigations involving human subjects were performed at Rutgers – New Jersey Medical School, Newark, NJ.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by National Institutes of Health grants: R01DK11379002, R01AI13163401A1, R01AI121250, R01AI101935, RO1AI124346, RO1AR060782 and R01AI114647-01A1. We would like to acknowledge Sandvik Osprey Powders, UK for generously providing the micrometre-sized CoCr material used in this paper. In addition we would like to thank E. M. Adler, M. C. Riley, F. R. Patterson and J. S. Hwang (Department of Orthopaedic Surgery, Rutgers – New Jersey Medical School) and S. R. Peters (Department of Pathology and Laboratory Medicine, Rutgers – New Jersey Medical School) for providing clinical samples and assistance in tissue preparation.