IL-10 Deficiency Reveals a Role for TLR2-Dependent Bystander Activation of T Cells in Lyme Arthritis

Lyme disease, caused by the tick-borne spirochete Borrelia burgdorferi, is the most common vector-borne illness in North America with estimates of 300,000 cases annually (1). The most common late manifestation of Lyme disease is arthritis, with up to 60% of infected individuals reporting intermittent attacks of joint swelling and pain (2). This can usually be treated successfully with 1–2 mo of oral or i.v. antibiotics (1, 3). However, 10–20% of patients present with persistent arthritis for months or years after standard antibiotic regimen, referred to as antibiotic-refractory arthritis or posttreatment Lyme disease (4, 5). Patients are negative for the presence of B. burgdorferi DNA after standard antibiotic therapy, raising several questions regarding the etiology of this long-lasting illness (6–8). Persistent Lyme arthritis shares certain pathogenic themes with rheumatoid arthritis such as similar synovial lesions, a dominant Th1 CD4⁺ T cell response in synovial tissues, and high concentrations of proinflammatory chemoattractants for CD4⁺ and CD8⁺ T cells in synovial fluid (9–11).

Several hypotheses have been proposed to explain persistent symptoms following antibiotic treatment, including molecular mimicry, dysregulated inflammation, and persistence of bacterial Ags (12–18). Adoptive transfer of CD4⁺ T cells into Rag1-deficient mice, which lack B and T cells, exacerbated the severity of arthritis and suggested a role for dysregulated T cells in Lyme disease (19). More than 8% of the B. burgdorferi genome coding sequences are specific for Pam3Cys-modified lipoproteins, which possess the potent ability to stimulate host immune responses by interacting with TLR1/2 heterodimers (20–24). Experiments with TLR2 whole-body knockout mice have demonstrated a compromised host defense and T cell involvement (25–27). Additionally, TLR2 expression on T cells has been shown to act as a costimulatory signal for T cell activation (28–30), and it could potentially be playing a role in inappropriate T cell responses during B. burgdorferi infection.

There have been numerous reports implicating a variety of CD4⁺ T cells in the pathogenesis of human Lyme disease (9, 31–37). Studies have characterized patients with persistent symptoms as having Th1 CD4⁺ T cells and inflammatory cytokines and chemokines such as IFN-γ, IL-1β, IL-6, CXCL9, and CXCL10 present in the synovial fluid, sometimes even months following infection and treatment (31, 32). IL-10^−/− mice, which lack the anti-inflammatory mediator IL-10, exhibit transcriptional upregulation of these same inflammatory mediators found in patients, contain CD4⁺ T cells in the joints, and have elevated IFN-γ in the serum at 2 wk postinfection despite extremely low levels of B. burgdorferi in the joint tissue (38, 39). The absence of IL-10 exposes the pathogenic potential of T cells whose activation threshold has been lowered during B. burgdorferi infection. We have used the IL-10^−/− mouse as a model for persistent Lyme arthritis to further investigate the arthritis-promoting properties of CD4⁺ T cells and IFN-γ in response to B. burgdorferi infection.

Despite correlative evidence linking aberrant T cell responses to other inflammatory arthritis diseases such as rheumatoid arthritis (40), the role of dysregulated T cell responses in the development of persistent Lyme arthritis has yet to be elucidated. In the current study, we show that activation of either CD4⁺ or CD8⁺ T cells is required for the development of Lyme arthritis, which persists in the presence of extremely low levels of B. burgdorferi Ag. Using transgenic mouse models, we demonstrate that T cell activation occurs independent of Borrelia-specific TCR interaction. Furthermore, we show that B. burgdorferi infection increases TLR2 expression on T cells. Cell transfer experiments revealed TLR2 as a critical mediator of T cell activation following B. burgdorferi infection, which results in enhanced IFN-γ production and Lyme arthritis. These results identify a novel mechanism of Lyme arthritis development dependent on TLR2 bystander activation of CD4⁺ and CD8⁺ T cells.

C57BL/6, C57BL/6 IL-10^−/− (B6.129P2-Il10^tm1Cgn/J), C57BL/6 TCRα^−/− (B6.129S2-Tcra^tm1Mom/J), C57BL/6 Rag1^−/− (B6.129S7-Rag1^tm1Mom/J), and C57BL/6 TLR2^−/− (B6.129-Tlr2^tm1Kir/J) mice were obtained from The Jackson Laboratory. C57BL/6 Rag2^−/−/OT-I mice were purchased from Taconic. SMARTA TCR transgenic mice, specific for the immunodominant class II–restricted epitope of lymphocytic choriomeningitis virus (LCMV) gp_61–80 (41), were maintained on a C57BL/6 background and additionally crossed to IL-10^−/− mice (SMARTA/IL-10^−/− mice). SMARTA/IL-10^−/− mice were further crossed to TCRα^−/− mice. Rag2^−/−/OT-I mice were crossed to IL-10^−/− mice. Genotyping was performed according to protocol provided by The Jackson Laboratory for the IL-10 and TCRα mutation. Experiments were performed using mice 5–7 wk of age, with male and female mice equally distributed into experiment and control groups. To prevent colitis development, IL-10^−/−, SMARTA/TCRα^−/−/IL-10^−/−, and OT-I/Rag2^−/−/IL-10^−/− mice were kept on antibiotic water (trimethoprim and sulfamethoxazole) until 1 d prior to infection. Mouse colonies were housed in the pathogen-free facility at the University of Utah. All mice used in this study were cared for in accordance with the guidelines of the Institutional Animal Care and Use Committee of the National Institutes of Health, and all animal experiments were approved by the Institutional Animal Care and Use Committee.

The N40 isolate of B. burgdorferi (provided by S. Barthold, University of California, Davis, CA) was grown to late log phase in Barbour–Stoenner–Kelly-II medium supplemented with 6% rabbit serum (Sigma-Aldrich). Mice were infected with 2 × 10⁴ B. burgdorferi by intradermal injection into the skin of the back. Infection was confirmed in mice sacrificed at 1 and 2 wk by the culture of B. burgdorferi from the bladder as described (25). ELISA quantification of B. burgdorferi–specific IgM and IgG concentrations was used to confirm infection in mice sacrificed at and after 14 d of infection as previously published (25). To measure in vivo cytokine production, mice were injected i.v. with 0.25 mg of brefeldin A (Sigma-Aldrich) in PBS 6 h before sacrifice (42).

Ankle measurements were obtained using a metric caliper before and 4 wk after infection by an investigator blinded to the experimental group. Rear ankle joints were prepared for assessment of histopathologic lesions by removal of the skin, and tissue was fixed in 10% neutral buffered formalin. Joints were decalcified and embedded in paraffin, sectioned at 3 μm, and stained with H&E. Lesions were scored in a blinded fashion, with each slide receiving a score of 0–5 for the characteristics of the disease such as polymorphonuclear leukocyte and mononuclear cell (lymphocytes, monocytes, macrophages) infiltration into inflammatory processes, tendon sheath thickening (hypertrophy and hyperplasia of surface cells and/or underlying dense sheets of cells resembling immature fibroblasts, synoviocytes, and/or granulation tissue), reactive/reparative responses (periosteal hyperplasia and new bone formation and remodeling), and overall lesion (composite score based on all lesions observed in six to eight sections per joint), with 5 representing the most severe lesion, and 0 representing no lesion as previously described (43).

Abs used in neutralized or depletion studies were purchased from Bio X Cell and were aggregate and endotoxin free, and sterile. Cellular subsets were depleted by administering 0.2 mg of Ab into the i.p. cavity every 4 d beginning 1 d prior to infection as indicated: anti-CD4 (GK1.5) and anti-CD8 (YTS 169.4). Rat IgG2b (LTF-2) was used as the isotype control. IFN-γ was neutralized using 1 mg of anti–IFN-γ (XMG1.2) 1 d prior to infection, followed by additional doses of 0.5 mg of anti–IFN-γ every 4–5 d. Rat IgG1 (HRPN) was used as the isotype control for IFN-γ neutralization. Cellular depletions of CD4⁺ and CD8⁺ T cells were confirmed by flow cytometry (Supplemental Fig. 2).

CD3⁺CD4⁺ and CD3⁺CD8⁺ T cells were isolated from popliteal and inguinal lymph nodes for intracellular cytokine staining. For mice treated with brefeldin A, 1 × 10⁶ cells were stained with surface markers and incubated for 30 min at 4°C. Cells were fixed/permeabilized in 100 μl of Cytofix/Cytoperm (BD Biosciences). Cells were resuspended in 100 μl of Perm/Wash buffer with fluorochrome-conjugated IFN-γ, briefly vortexed, and incubated on ice in the dark for 45 min. Cells were resuspended in 200 μl of PBS plus 2% FBS for analysis by flow cytometry. Restimulation of IL-10^−/−, SMARTA/TCRα^−/−/IL-10^−/−, and OT-I/Rag2^−/−/IL-10^−/− cells was performed for 4 h with 20 ng/ml PMA and 1 μM ionomycin at 37°C in the presence of brefeldin A (1 μl/ml; GolgiPlug). For cell transfer experiments, CD3⁺CD4⁺ and CD3⁺CD8⁺ T cells were isolated from lymph nodes (popliteal and inguinal) and spleens of wild-type (WT) C57BL/6 and TLR2^−/− mice using the Pan T Cell Isolation Kit II (Miltenyi Biotec). T cell sorting purity was determined by flow cytometry.

Splenocytes from naive SMARTA/TCRα^−/−/IL-10^−/− and OT-I/Rag2^−/−/IL-10^−/− were isolated and 1 × 10⁶ cells were stimulated for 3 h at 37°C in the presence of 1 μM gp_61–80 peptide (GLKGPDIYKGVYQFKSVEFD), 1 μM OVA_257–264 peptide (SIINFEKL), or 5 μg/ml sonicated B. burgdorferi. Unstimulated cells incubated with media alone were used as a control. After stimulation, cells were fixed in 100 μl of 1.5% paraformaldehyde and incubated at 37°C for 10 min. After fixation, cells were stained with surface markers CD4, CD8, and Vα2 in PBS containing 2% FBS for 25 min on ice in the dark. Cells were permeabilized using 100 μl of cold 100% methanol for 10 min on ice. Cells were centrifuged and resuspended in 100 μl of PBS containing 2% FBS with pZAP-70/SYK (Y319/Y352) clone n3kobu5 (eBioscience) and incubated on ice in the dark for 45 min. Total ZAP-70 clone 1E7.2 (eBioscience) was used as a control. ZAP-70 phosphorylation was analyzed using flow cytometry.

RNA was purified from the tibiotarsal joints after the skin had been removed. Tissue was immediately immersed in RNA stabilization solution (Invitrogen) and stored at −80°C. Total RNA was recovered from homogenized tissue using the Direct-zol kit (Zymo Research). RNA recovered from tissue and cells was reverse transcribed, and transcripts were quantified using a Roche LC-480 quantitative RT-PCR (qRT-PCR) machine according to our previously described protocols (44). Primer sequences used in this study were as follows: CD8β (CD8), forward, 5′-CTCTGGCTGGTCTTCAGTATGA-3′, reverse, 5′-TCTTTGCCGTATGGTTGGTTT-3′; TLR2, forward, 5′-TTGAAGAACTCAGCCTGTAAG-3′, reverse, 5′-GCTTCCAGAGTCTCCAGTTTG-3′. Primer sequences for B. burgdorferi 16S rRNA (45), β-actin (44), CD4 (38), Ifng and Cxcl9 (46), and Cxcl10 (38) can be found in the indicated citations.

Flow cytometry data were collected on an LSRFortessa (BD Biosciences) flow cytometer and analyzed using FlowJo (v.10) software. FACS cell sorting was performed using FACSAria (BD Biosciences). Confirmation of cell sorting efficiency was performed using qRT-PCR. Cell surface stains were done in PBS containing 2% FBS. Intracellular stains for cytokines were done using a kit per the manufacturer’s instructions (BD Biosciences). Position of gates for sorting and analysis was based on analysis of appropriate isotype controls. Fluorochrome-conjugated Abs from BioLegend were CD3ε, CD4, CD8α, CD19, CD44, CD62L, IFN-γ, NK1.1, and Vα2. Fluorochrome-conjugated Abs from eBioscience were TLR2 and pZAP-70.

Blood was obtained from the tail vein of mice 2 wk postinfection and collected by submandibular puncture at the time of sacrifice. Serum was isolated and stored at −20°C prior to analysis. IFN-γ concentrations in serum samples were detected by sandwich ELISA using clone R4-6A2 as the capture Ab, and biotinylated Ab (XMG1.2) for detection. To determine B. burgdorferi–specific IgM and IgG concentrations, microtiter plates were coated with either sonicated B. burgdorferi or goat Ab to mouse IgG and IgM (Life Technologies). Serum dilutions were added to plates for 90 min at 37°C and bound murine Ig was detected by addition of HRP-conjugated Abs to murine IgG or IgM (Thermo Fisher Scientific). Ig content was estimated by comparing with standard curves using purified IgG or IgM as previously described (25).

The number of mice per group is annotated in corresponding figure legends. Statistical analysis was performed using Prism 6.0d software. Multiple-sample data sets were analyzed by one-way ANOVA followed by a Bonferroni post hoc test for pairwise comparisons, as appropriate and indicated in figure legends. Two-sample data sets were analyzed by a Student t test. Categorical data for histopathology were assessed by a Mann–Whitney U test. In all figures, data represent the mean ± SEM. A p value ≤ 0.05 was considered significant (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001).

The involvement of CD4⁺ T cells in the pathogenesis of human Lyme disease is seen even months following infection (9, 10), prompting investigation of a model that would allow assessment of the contribution of T cells. We have shown previously that the mild arthritis in C57BL/6 mice is increased in the absence of IL-10 despite greatly reduced numbers of B. burgdorferi in the joint tissue (38, 47). To determine whether C57BL/6 IL-10^−/− mice could be used as a model of the persistent Lyme arthritis seen in human patients, infected mice were followed for 18 wk. B. burgdorferi 16S rRNA was no longer detectable in the joint tissue of IL-10^−/− mice at this time (Table I). Additionally, IL-10^−/− mice displayed sustained upregulation of Ifng, Cxcl9, and Cxcl10 transcripts in the joint tissue (Table I), similar to observations in the synovial fluid of human patients months following antibiotic therapy (48, 49). In contrast, arthritis-resistant WT C57BL/6 mice continued to harbor detectable amounts of B. burgdorferi 18 wk postinfection; however, Ifng, Cxcl9, and Cxcl10 transcripts were not upregulated in the joint tissue (Table I). These results suggest that IL-10^−/− mice are a suitable model for the investigation of the arthritis-promoting properties of CD4⁺ T cells and IFN-γ in B. burgdorferi infection, as seen in human patients. This has not been possible in the widely studied C3H mouse, where neither T cells nor IFN-γ is required for arthritis development (46, 50, 51).

Localized IFN-γ production has been observed in the synovial fluid of patients with persistent symptoms (52). Similarly, arthritis development in IL-10^−/− mice is dependent on both systemic and localized production of IFN-γ (38). Canonical sources of IFN-γ include Th1 CD4⁺ T cells, NK cells, and NKT cells, all of which are found in the joint tissue of B. burgdorferi–infected IL-10^−/− mice (38, 53). To define the cellular sources of IFN-γ in Lyme arthritis development, IL-10^−/− mice were infected with B. burgdorferi and cells were harvested from the draining popliteal and inguinal lymph nodes 4 wk postinfection, which is the peak of arthritis severity. Cells were restimulated in the presence of PMA and ionomycin for 4 h and then intracellularly stained for IFN-γ and analyzed by flow cytometry. This allowed for the identification of cells poised to produce IFN-γ. The frequency of both CD4⁺IFN-γ⁺ and CD8⁺IFN-γ⁺ T cells was increased compared with mock-infected mice (Fig. 1A). IFN-γ was not produced by NK or NKT cells in response to PMA/ionomycin stimulation (data not shown), thus suggesting that most IFN-γ–producing cells were T cells. The production of IFN-γ is specific to the draining lymph nodes, as T cells do not produce IFN-γ in other secondary lymphoid organs such as the spleen (data not shown). These proportional increases in IFN-γ⁺ lymphocyte populations were also reflected in total cell numbers (Fig. 1B). Furthermore, CD4⁺ and CD8⁺ T cells were producing more IFN-γ on a per cell basis compared with controls (Fig. 1B). Interestingly, although there were fewer IFN-γ⁺–producing CD4⁺ T cells than CD8⁺ T cells, CD4⁺ T cells were secreting more IFN-γ than CD8⁺ T cells as measured by mean fluorescence intensity. Taken together, these data suggest that both CD4⁺ and CD8⁺ T cells are poised to make IFN-γ following B. burgdorferi infection.

To capture the actual in vivo cytokine production during the course of B. burgdorferi infection, infected IL-10^−/− mice were injected i.v. with the Golgi blocking agent, brefeldin A, 6 h prior to sacrifice (42). CD4⁺ and CD8⁺ T cells were analyzed for activation and functional markers by flow cytometry. Intracellular staining revealed an increase in the frequency and number of CD4⁺ and CD8⁺ T cells producing IFN-γ throughout infection (Fig. 1C, 1D). Importantly, the numbers of both CD4⁺IFN-γ⁺ and CD8⁺IFN-γ⁺ T cell were equivalent by 4 wk postinfection, demonstrating that both are playing a critical role in IFN-γ production (Fig. 1D). At 4 wk postinfection, during peak arthritis severity, both CD4⁺ and CD8⁺ T cells had expanded ∼3-fold, suggesting that both CD4⁺ and CD8⁺ T cells are either trafficking to or proliferating in the draining lymph nodes during arthritis development (Supplemental Fig. 1). CD44 expression was assessed as an additional indicator of T cell activation status. Both the frequency and total number of CD4⁺CD44^hi and CD8⁺CD44^hi T cells increased throughout infection, demonstrating that B. burgdorferi induces activation of both CD4⁺ and CD8⁺ T cells (Fig. 1E, 1F). Taken together, these data indicate that CD4⁺ and CD8⁺ T cells are expanding in the draining lymph nodes upon B. burgdorferi infection and they are functionally active as demonstrated through increased IFN-γ production and CD44 expression.

Because CD4⁺ and CD8⁺ T cells are the primary cellular sources of IFN-γ in the draining lymph nodes, we hypothesized that they were both necessary and sufficient for the development of Lyme arthritis in IL-10^−/− mice following B. burgdorferi infection. To directly assess the contribution of CD4⁺ and CD8⁺ T cells, B. burgdorferi–infected IL-10^−/− mice were treated with depleting Abs directed toward CD4, CD8, or a combination of both throughout infection (see 2Materials and Methods). Depletion efficiency was measured throughout the length of the experiment by flow cytometric analysis of blood, and also analyzed at the time of sacrifice in the draining lymph nodes and the joints and determined to be >95% effective (Supplemental Fig. 2A–C). Arthritis severity in mice receiving anti-CD4 Ab was similar to mice receiving isotype control, whereas neutralization of CD8 partially reduced arthritis (Fig. 2A). The neutralization of both CD4 and CD8 reduced ankle swelling and histopathology scores (Fig. 2A, 2B). Mice that were treated with anti-CD4 displayed an increase in bacterial burden in joint tissue as compared with isotype control–treated animals (Fig. 2C), which is consistent with previous studies (54). Interestingly, CD4⁺ T cells play a role in host defense not shared by CD8⁺ T cells, as mice treated with anti-CD8 did not display an increase in bacterial burden in joint tissue (Fig. 2C). B. burgdorferi–infected IL-10^−/− mice expressed transcripts for Ifng, Cxcl9, and Cxcl10 in joint tissue that was significantly reduced only in mice treated with a combination of anti-CD4 and anti-CD8 (Fig. 2D–F). These data indicate that both CD4⁺ and CD8⁺ T cells contribute the expression of IFN-γ and IFN-γ–dependent genes in the joints of B. burgdorferi–infected mice. In the absence of CD4⁺ T cells, the number of CD8⁺IFN-γ⁺ T cells increased substantially, suggesting that CD8⁺ T cells are able to compensate for the loss of CD4⁺-derived IFN-γ (Supplemental Fig. 2D). The number of CD4⁺IFN-γ⁺ T cells also increased in the absence of CD8⁺ T cells. Taken together, these data demonstrate that CD4⁺ and CD8⁺ T cells act synergistically to drive Lyme arthritis. These findings additionally highlight the lack of correlation between the bacterial burden of B. burgdorferi and the development of arthritis.

We next explored the possibility that T cells were being activated in a manner independent of classical TCR-mediated recognition. To address this possibility, the relative distribution of 14 different TCR β-chain V region (Vβ) subsets were examined in T cells from draining lymph nodes of B. burgdorferi–infected IL-10^−/− mice using a TCR Vβ repertoire panel (BD Biosciences). TCR Vβ profiling is a standard assay routinely used in our laboratory (55, 56), and Ag-specific T cell activation results in population skewing toward a few selected TCR Vβ families. Stimulation of WT C57BL/6 lymph node cells with the immunodominant glycoprotein, gp_61–80, from LCMV results in expansion of TCR Vβ 8.3 (data not shown). In contrast, bystander T cell responses involve a wide diversity of TCRs, suggesting Ag-independent activation (57). IL-10^−/− mice displayed increased numbers of CD4⁺ and CD8⁺ T cells in the draining lymph nodes during B. burgdorferi infection (Fig. 3A, Supplemental Fig. 1). However, there was no selective expansion of any individual Vβ subset on either CD4⁺ or CD8⁺ T cells during B. burgdorferi infection, which is indicative of a polyclonal or nonspecific T cell response. (Fig. 3B, 3C). Similarly, we did not observe any change in the frequency of IFN-γ–producing cells among TCR Vβ subsets following infection (Fig. 3C). These results support the hypothesis that T cell expansion in the draining lymph nodes during B. burgdorferi infection is independent of Ag specificity.

To directly confirm Ag-independent activation of CD4⁺ T cells, TCR transgenic SMARTA mice were used. SMARTA mice possess CD4⁺ T cells with monoclonal specificity for the MHC class II–restricted epitope of LCMV gp_61–80. This epitope is not found in B. burgdorferi, and thus SMARTA CD4⁺ T cells cannot not become activated through TCR engagement during B. burgdorferi infection. To determine whether Ag-independent activation was occurring during B. burgdorferi infection, SMARTA mice were crossed to IL-10^−/− mice to generate TCR transgenic SMARTA/IL-10^−/− mice (see 2Materials and Methods). To further eliminate the possibility that residual non-SMARTA TCRs (present on ∼5% of T cells) or dual TCRs on SMARTA T cells could recognize B. burgdorferi Ags, the SMARTA/IL-10^−/− mice were further crossed to a TCRα^−/− background (SMARTA/TCRα^−/−/IL-10^−/−).

SMARTA/TCRα^−/−/IL-10^−/−mice and IL-10^−/− were infected with B. burgdorferi and sacrificed 4 wk postinfection. SMARTA/TCRα^−/−/IL-10^−/−mice developed severe arthritis and histopathologic lesions, suggesting that B. burgdorferi–specific T cells are not responsible for arthritis development (Fig. 4A, 4B). Further histopathologic analysis of infected SMARTA/TCRα^−/−/IL-10^−/−mice revealed that the tendon sheath is reactive and thickened. Synoviocytes (arrowheads) covering the tendon and lining the tendon sheath are reactive and increased in number and size. Adjacent supporting stroma (extracellular matrix) under the tendon sheath is also thickened and contains neutrophils (subacute inflammation) admixed with lymphocytes and occasional macrophages (i.e., mononuclear inflammatory cells) as well as immature fibroblasts and collagen fibers (Fig. 4B). Examination of the draining lymph nodes from B. burgdorferi–infected SMARTA/TCRα^−/−/IL-10^−/− mice revealed an increase in CD4⁺ T cells compared with mock-infected controls, indicating expansion of TCR transgenic T cells (Fig. 4D). Additionally, the number of CD4⁺IFN-γ⁺ T cells increased in B. burgdorferi–infected SMARTA/TCRα^−/−/IL-10^−/− mice, indicating that classical TCR recognition of B. burgdorferi Ag is not required for IFN-γ secretion by CD4⁺ T cells (Fig. 4E). Concentrations of B. burgdorferi 16S rRNA were observed in the joint tissue of SMARTA/TCRα^−/−/IL-10^−/− mice (Fig. 4F). Joint tissue was also analyzed for the presence of Ifng, Cxcl9, and Cxcl10 transcripts, all of which were elevated in infected SMARTA/TCRα^−/−/IL-10^−/−mice compared with mock-infected controls (Fig. 4G–I). We observed an increase in B. burgdorferi–specific IgM and a decrease in B. burgdorferi–specific IgG in infected SMARTA/TCRα^−/−/IL-10^−/−mice compared with IL-10^−/− mice (Fig. 4J, 4K), suggesting that there is a lack of T cell help for class switching. To confirm that there were no cross-reactive Ags present in B. burgdorferi, phosphorylation of the direct substrate of TCR activation, ZAP-70, was assayed. SMARTA/TCRα^−/−/IL-10^−/−CD4⁺ T cells displayed increased phosphorylation of ZAP-70 when stimulated with their cognate Ag, gp_61–80. However, when stimulated with B. burgdorferi, there was no phosphorylation of ZAP-70, indicating that TCR signaling is not occurring (Fig. 4L). It has been previously shown that neutralization of IFN-γ in infected IL-10^−/− mice significantly reduced ankle swelling and histopathology scores (38). IFN-γ neutralization in infected SMARTA/TCRα^−/−/IL-10^−/−mice also resulted in a significant decrease in arthritis severity and histopathology scores (Supplemental Fig. 3A, 3B). However, arthritis was still elevated above mock-infected animals, indicating that other components are also contributing to arthritis. Interestingly, IFN-γ neutralization did not affect expansion of SMARTA/TCRα^−/−/IL-10^−/− CD4⁺ T cells in the draining lymph nodes (Supplemental Fig. 3C) but resulted in decreased expression of Ifng, Cxcl9, and Cxcl10 in joint tissue, providing further support that IFN-γ influences arthritis development upon B. burgdorferi infection in CD4⁺ transgenic mice (Supplemental Fig. 3E–G). Overall, these data demonstrate that B. burgdorferi infection results in Ag-independent expansion and activation of CD4⁺ T cells in the draining lymph node, resulting in the development of arthritis.

To determine whether CD8⁺ T cells were contributing to the development of Lyme arthritis in an Ag-independent mechanism, TCR transgenic Rag2^−/−/OT-I mice were used. Rag2^−/−/OT-I mice possess CD8⁺ T cells with monoclonal specificity to chicken OVA-derived MHC class I–restricted epitope OVA_257–264. Rag2^−/−/OT-I mice were crossed to IL-10^−/− mice to generate OT-I/Rag2^−/−/IL-10^−/− mice. OT-I/Rag2^−/−/IL-10^−/− and IL-10^−/− mice were infected with B. burgdorferi and sacrificed 4 wk postinfection. OT-I/Rag2^−/−/IL-10^−/− mice developed severe arthritis and histopathologic lesions similar to IL-10^−/− mice (Fig. 5A, 5B). Examination of the draining lymph nodes revealed a trending increase in CD8⁺ T cells from OT-I/Rag2^−/−/IL-10^−/− mice compared with mock-infected controls (Fig. 5C), similar to CD8⁺ T cells in the IL-10^−/− experimental group included for comparison (Fig. 5C). Additionally, there was a trending increase in CD8⁺IFN-γ⁺ T cells in the lymph nodes of OT-I/Rag2^−/−/IL-10^−/− compared with mock-infected controls, indicating that CD8⁺ T cells are capable of producing IFN-γ following Borrelia-induced bystander activation (Fig. 5D). We observed concentrations of B. burgdorferi 16S rRNA in the joints of OT-I/Rag2^−/−/IL-10^−/− mice, likely due to the absence of B cells (Fig. 5E). There was also a similar type II IFN profile observed in joint tissues, as Ifng, Cxcl9, and Cxcl10 transcripts were all elevated in B. burgdorferi–infected OT-I/Rag2^−/−/IL-10^−/− mice as compared with mock-infected controls (Fig. 5F–H). The use of a second TCR transgenic mouse that consists of CD8⁺ T cells recognizing a different Ag (OVA) further addresses the unlikely possibility of a cross-reactive Borrelia epitope. Phosphorylation of ZAP-70 did not occur when OT-I CD8⁺ T cells were stimulated with B. burgdorferi, indicating that T cell activation was not occurring through the TCR (Fig. 5I). Similar to the observations made for CD4⁺ T cells, these results indicate that CD8⁺ T cell activation and IFN-γ production occur via bystander activation during B. burgdorferi–induced arthritis. Overall, our results indicate a critical role for bystander T cell activation in mediating IFN-γ–dependent Lyme arthritis.

B. burgdorferi possess lipoproteins with a potent ability to stimulate host pathogen recognition receptors such as TLRs. TLR2 has been implicated as an important player in host defense, as TLR2^−/− mice experience deficiencies in clearing B. burgdorferi and harbor more bacteria in the joints compared with WT mice (25, 26). It has recently been appreciated that both CD4⁺ and CD8⁺ T cells are capable of expressing TLR2 (29, 58). To evaluate the expression of TLR2 on T cells during B. burgdorferi infection, flow cytometry and qRT-PCR were performed on cells from the draining lymph nodes of infected IL-10^−/− mice. Draining lymph nodes were collected 2 wk postinfection and CD4⁺, CD8⁺, and CD19⁺ cells were isolated by flow cytometry. Tlr2 transcripts were significantly elevated in both T and B cells following B. burgdorferi infection (Fig. 6A). The increase of Tlr2 transcript was specific, as Tlr1, Tlr6, Tlr7, and Tlr9 transcripts remained unchanged (Supplemental Fig. 4). At peak arthritis severity, the total number of CD4⁺ and CD8⁺ T cells expressing TLR2 protein had increased ∼10-fold compared with mock-infected mice, indicating the potent ability of B. burgdorferi to upregulate TLR2 on T cells (Fig. 6B). The expression of TLR2 on both CD4⁺ and CD8⁺ T cells increased and peaked at 3 wk postinfection (Fig. 6C, 6D). This finding also coincided with the highest number of CD4⁺IFN-γ⁺ and CD8⁺IFN-γ⁺ T cells in the draining lymph node (Fig. 1D). To determine whether this occurred in the absence of a B. burgdorferi–specific TCR, we analyzed the expression of TLR2 in both SMARTA/TCRα^−/−/IL-10^−/− and OT-I/Rag2^−/−/IL-10^−/− mice. TLR2 expression was increased on both CD4⁺ and CD8⁺ TCR transgenic T cells upon B. burgdorferi infection as evidenced by the increase in frequency and total number of TLR2⁺ T cells (Fig. 6E–H). These results indicate that B. burgdorferi infection leads to increased transcription and expression of TLR2 on both CD4⁺ and CD8⁺ T cells, and could play a role in T cell activation.

The observation of B. burgdorferi–induced TLR2 expression on T cells prompted us to directly assess its role in arthritis development. C57BL/6 Rag1^−/− mice were infected with B. burgdorferi 2 wk prior to an i.v. injection of T cells isolated from the lymph nodes and spleens of healthy naive WT or TLR2^−/− C57BL/6 mice in a method described previously (19). In this experiment donor and recipient mice expressed IL-10. Infected Rag1^−/− mice that did not receive T cells served as controls and had mild arthritis at 5 wk postinfection (Fig. 7A, 7B). Rag1^−/− mice reconstituted with WT T cells developed severe arthritis compared with controls. Strikingly, mice that had received TLR2^−/− T cells were protected from severe arthritis as measured by ankle swelling and histopathology scores (Fig. 7B, 7C). Mice that had received WT T cells had a significant increase in serum IFN-γ compared with mice receiving TLR2^−/− T cells. There was no IFN-γ in the serum of mice that were not reconstituted with T cells (Fig. 7D). Furthermore, mice that had received WT T cells displayed an increase in Ifng, Cxcl9, and Cxcl10 transcripts in the joint compared with mice reconstituted with TLR2^−/− T cells (Fig. 7E–G). IFN-γ was still present in the serum and joints of mice receiving TLR2^−/− T cells, indicating that other receptors for microbial patterns could contribute to the IFN-γ profile and the development of arthritis. It is also possible that other cytokines could also be contributing to bystander activation of T cells. Interestingly, the presence of either WT or TLR2^−/− T cells drastically reduced the bacterial burden in the joint (Fig. 7F), suggesting that T cells play a role in enhancing host defense independent of TLR2. Taken together, these data reveal TLR2 as a critical mediator of T cell activation following B. burgdorferi infection, which results in enhanced IFN-γ production and Lyme arthritis.

We conclude that persistent Lyme arthritis can be initiated and sustained through bystander activation of both CD4⁺ and CD8⁺ T cells. This is consistent with the absence of selective expansion of any individual Vβ subset on either CD4⁺ or CD8⁺ T cells during B. burgdorferi infection. Furthermore, Ag-independent expansion and activation of TCR transgenic CD4⁺ or CD8⁺ T cells was sufficient to induce arthritis. B. burgdorferi sequentially expresses an abundance of lipoproteins that are critical in the activation of TLR2 expressed by innate immune cells. We now demonstrate that B. burgdorferi induces TLR2 expression on T cells, which results in increased T cell activation and arthritis development. To our knowledge, this is the first example of CD4⁺ T cells or CD8⁺ T cells promoting arthritis pathogenesis in a TCR-independent mechanism.

Dysregulated immune responses are thought to play a key role in the pathogenesis of arthritis development. For example, Lyme disease patients with persisting symptoms after completing the standard dose of antibiotics often have lower frequencies of regulatory T cells and higher expression of coactivation receptors in their synovial fluid (30). This suggests that an inappropriately amplified immune response is contributing to arthritis development. Other studies have shown that chronic states of inflammation involving both systemic and local production of proinflammatory cytokines such as TNF and IL-6 are required for the pathogenesis of rheumatoid arthritis (59, 60). These results suggest that future CD4⁺ and CD8⁺ T cell studies should focus on costimulatory signals known to influence T cell activation and their role in Lyme and other arthralgias.

The involvement of T cells in Lyme arthritis development and resolution has been debated. Several reports suggest that the adaptive immune response is not required for acute Lyme arthritis development in genetically susceptible mouse strains (C3H), as scid and rag mice, which lack both functional B and T cells, develop arthritis upon B. burgdorferi infection (51, 61). However, adoptive transfer of CD4⁺ T cells into rag mice was shown to exacerbate the severity of Lyme arthritis, implicating a role for dysregulated T cells in Lyme disease (19). Efforts have been made to understand the protective and disease-promoting roles of the Ag-specific T cell response to B. burgdorferi (19, 33, 54). Initial studies suggested that T cell responses to a spirochetal epitope were cross-reacting with a similar epitope of a self-protein, resulting in molecular mimicry and autoimmune-like pathology (62). In support of this autoimmunity hypothesis, specific HLA-DR alleles such as DRB1*0101 and DRB1*01 have been linked as genetic risk factors for the development of antibiotic refractory arthritis (63). Recent studies have also identified autoantigens that are capable of acting as T cell targets in patients with Lyme disease (13, 64–66). Bystander activation of T cells may provide an environment for autoimmune-mediated disease development. Previous studies have identified bystander activation of γδ T cells in synovial fluid of patients with persistent symptoms, which is dependent on TLR activation of myeloid cells (35, 36). The mechanism by which these T cells are activated has yet to be determined, and future studies are needed to identify the primary signals that promote bystander activation. Our studies clearly provide a potential scenario in which localized activation of bystander T cells could result in the expansion of clones recognizing self-antigens.

Ag-independent or bystander T cell activation was first demonstrated in virally infected mice, which resulted in expansion and proliferation of heterologous polyclonal T cells (67). More recent studies have begun to illuminate the mechanisms driving this TCR-independent expansion of T cells and have implicated certain proinflammatory cytokines such as IFN-γ, IL-12, IL-15, and IL-18 as capable of causing bystander activation of mucosal-associated variant T cells and memory CD8⁺ T cells (68–72). However, less is known about the signals that initiate CD4⁺ T cell bystander activation. The ability of both CD4⁺ and CD8⁺ T cell bystander activation to exacerbate Lyme arthritis is unique. The presence of CD8⁺ T cells was sufficient to fully promote Lyme arthritis in IL-10^−/− mice, despite the absence of CD4⁺ T cells and B cells. CD4⁺ T cells have a dual role in host pathogenesis and host defense in this model system, as bacterial burden in the joints was higher in the absence of CD4⁺ T cells. Differences in bacterial load between IL-10^−/− and CD4 TCR transgenic mice indicate that although a Borrelia-specific CD4⁺ T cell response is required for clearance of B. burgdorferi, it is not required to promote arthritis severity. This finding is consistent with previous publications using other mouse strains that show that Borrelia load does not correlate with arthritis severity (73, 74). It was also revealed that CD4⁺ T cells play a role in the generation of B. burgdorferi–specific Abs, as their absence during infection resulted in decreased B. burgdorferi–specific IgG. This unique host defense feature of CD4⁺ T cells is not shared by CD8⁺ T cells, and it implicates a role for CD4⁺ T cell Ag specificity in promoting bacterial clearance. Our findings suggest that CD8⁺ T cells could even provide a therapeutic target for control of inflammation without compromising the ability to clear infection.

Several studies have reported a costimulatory role for TLRs on CD8⁺ T cells. TLR2 expression on CD8⁺ T cells leads to a decreased activation threshold for costimulatory signals received by APCs (75). Our data support these results, as TLR2 expression increased on both CD4⁺ and CD8⁺ T cells during B. burgdorferi infection. The expression of TLR2 on T cells also coincided with the increase in IFN-γ⁺ T cells, further suggesting that TLR2 is playing a direct role in IFN-γ secretion. Strikingly, arthritis severity was greatly reduced when TLR2 was selectively missing from T cells, revealing TLR2 as a critical component of bystander T cell activation driving arthritis. Accumulating evidence supports significant alterations in T cell TLR expression in patients with infectious diseases. For example, T cells from patients with chronic hepatitis C express higher transcript levels of several TLRs in CD8⁺ T cells compared with healthy controls (76). More studies are needed to assess the effect of altered TLR expression on T cells and its contribution to pathogenesis and immune defense.

This study highlights the Ag-independent role of both CD4⁺ and CD8⁺ T cells in the development of Lyme arthritis. Importantly, several observations in the IL-10^−/− mouse model are recapitulated in posttreatment Lyme disease patients (31, 32). Synovial tissue from patients contains elevated numbers of CD4⁺ T cells, as well as elevated concentrations of IFN-γ, CXCL9, and CXCL10, compared with patients whose arthritis resolves after antibiotics. This is unique, as other mouse models have not allowed for the assessment of T cell contribution. Therefore, our study reveals a role for T cell–driven arthritis, providing insight into the pathogenic potential of inflammatory dysregulation in Lyme disease.

We acknowledge Stephanie Meek, Peter Pioli, Robert Lochhead, Tyson Chiaro, Dylan Beyer, and Garrison Asper for technical assistance, training, advice and manuscript review, as well as James Marvin and Tessa Galland from the University of Utah Flow Cytometry Core Facility for expertise and assistance.

The authors have no financial conflicts of interest.