A critical role for Type I IFN in arthritis development following Borrelia burgdorferi infection of mice

Bacterial culture and mouse infection

C3H/HeN mice were obtained from Charles River Breeding laboratories or NCI, depending on availability. C57BL/6 mice were purchased from NCI, and C3H/HeJ and C3H SCID mice were obtained from Jackson Laboratories. C3H TLR2−/− mice were originally generated by Deltagen Inc., obtained from Tularik, Inc. (now Amgen Inc., South San Francisco, CA) (26). C57BL/6 MyD88−/− mice were kindly provided by Dr. Shizuo Akira (27), and C57BL/6 TLR9−/− were obtained from the Mutant Mouse Resource at the University of California at Davis. C57BL/6 IFNAR1−/− mice were a kind gift from Dr. Murali-Krisna Kaja (University of Washington, Seattle, WA). All mice were housed at the University of Utah Animal Research Center and utilized with strict adherence to all institutional policies and guidelines established for the care and use of laboratory animals in biomedical research.

A clonal isolate of Borrelia burgdorferi strain N40 (cN40), kindly provided by Steve Barthold (University of California at Davis) (6), was propagated in BSKII media (28). For infection experiments, 2000 cN40 spirochetes were intradermally injected into the shaved back of 6 wk old, female mice in a volume of 20 μl. Uninfected, control mice received 20 μl of BSKII. Mice were sacrificed at 1,2, or 4 wks post-infection and ankle measurements taken with a metric caliper were compared with pre-infection measurements and recorded as the change in ankle measurement (29). For mock-infected control mice, as well as animals euthanized at 1 and 2 wks post-infection, all skin was carefully removed, and both rear ankle joints were excised and separately snap frozen in a dry ice-ethanol bath for subsequent RNA extraction and RT-PCR analysis of IFN-inducible genes.

At 4 wks post-infection, the most swollen ankle joint, as assessed by metric caliper measurement, was excised and placed in 10% neutral-buffered formalin (Sigma). Samples were then decalcified, paraffin-embedded, sectioned, and stained with hemotoxylin and eosin by AML Laboratories (Rosedale, MD). Arthritis severity was assessed in a blinded fashion, and overall scores for joints obtained from both mock-infected and B. burgdorferi-infected mice were assigned based on a scale of 0 (no arthritis) to 5 (most severe arthritis), according to the following parameters: overall lesion appearance, degree of neutrophil infiltration, amount of monocytic infiltration, tendon sheath thickness, and the presence of reactive and reparative responses. The least swollen joint was removed at 4 weeks post-infection for quantitative PCR assessment (qPCR) of B. burgdorferi numbers within the joint.

Antibodies and depletion protocols

The following antibodies were purchased from eBioscience and utilized for flow cytometry analyses: FITC-conjugated α-DX5 (CD49b) pan-NK cell antibody, PE-conjugated α-CD3ε, PE-conjugated α-Armenian hamster IgG, FITC-conjugated α-rat IgM, FITC-conjugated α-CD11c, PE-conjugated α-B220, and FITC-α-Hamster IgG. PE-α-Rat IgG2a was purchased from BD Pharmingen.

MAR1-5A3, a mouse anti-mouse IgG1 monoclonal antibody directed against IFNAR1 was used to inhibit Type I IFN responses in C3H/HeN mice (30). The mouse IgG1 monoclonal antibody MOPC-21 (BioExpress Cell Culture Services, West Lebanon, N.H.) or the mouse anti-human IFNGR1 antibody GIR-208 (31) was utilized as an isotype control. Groups of 5 age-matched, female C3H/HeN mice received an intraperitoneal injection of PBS, or 2.5 mg of either MAR1-5A3, or an appropriate isotype control, one day prior to B. burgdorferi infection. IFNγ was neutralized using the hamster H22 monoclonal antibody (32). The hamster α-GST monoclonal antibody PIP was used as an isotype control. C3H/HeN mice were injected intraperitoneally with 0.5 mg of either H22 or PIP, or an equal volume of PBS, two days prior to B. burgdorferi infection. 0.25 mg of H22 or PIP, or PBS was administered on the day of infection (day 0), and booster injections were given every seven days thereafter for the duration of the study. To block both Type I IFN and IFNγ responses, intraperitoneal injections of either 2.5 mg MAR1-5A3 + 0.5 mg H22, 2.5mg MOPC-21 + 0.5 mg Hamster IgG, or an equal volume of PBS, were administered one day prior to B. burgdorferi infection. The rabbit polyclonal antibody α-asialo GM1 (Wako Chemicals, Richmond, VA) was utilized to deplete NK cells from C3H mice (33). Groups of 5 female, age-matched mice received 50 μl (1.8 mg) intraperitoneal injections of α-asialo GM1, rabbit IgG (Sigma), or an equal volume of PBS on the day of infection (day 0), and on day 4. α-asialo GM1 administration resulted in a 75% depletion of NK and NK T cells from the spleen, as assessed by flow cytometry utilizing the FITC-conjugated α-DX5 (CD49b) pan-NK cell antibody. Plasmacytoid dendritic cells (pDCs) were blocked with the α-PDCA-1 monoclonal antibody (Miltenyi Biotec). Groups of 10 female, age-matched mice were injected intraperitoneally with 500 μg of α-PDCA-1, rat IgG, or an equal volume of PBS one day prior to B. burgdorferi infection. Animals were then boosted with 250 μg of α-PDCA-1 or appropriate controls on the day of infection.

RNA extraction and RT-PCR analysis

Rear ankle joints excised from B. burgdorferi- and 16 mock-infected mice were homogenized in chilled guanidine thiocyanate (Roche), and RNA was extracted via cesium chloride ultracentrifugation (34). RNA was extracted from cultured cells with Trizol reagent (Invitrogen), according to the manufacturer’s instructions. RNA samples were subsequently reverse-transcribed into cDNA, as described elsewhere (15) and LightCycler-based PCR amplification was conducted utilizing the LightCycler FastStart DNA Master^plus SYBR green I Kit for the LightCycler 2.0 instrument (Roche), as previously described (35), or the LightCycler 480 (LC480) SYBR green I master mix for the LC480 machine (Roche). The number of transcript copies present in the starting template sample were calculated, and normalized to 1000 copies of mouse β-actin, as mentioned elsewhere (35). The following primer pairs span introns and were used to amplify cytokines and IFN-responsive transcripts:

• Oasl2F: TGGGAAGAAGAAAGGGATGGG,

• Oasl2R: GGGTTTGTCCAGAGTGTCCAATC;

• Cxcl9F: TTGGGCATCATCTTCCTGGAGCAG,

• Cxcl9R: GAGGTCTTTGAGGGATTTGTAGTGG;

• (LC only) Il-6dF: ACAACCACGGCCTTCCCTACTT (36),

• Il-6bR: TCTCTGAAGGACTCTGGCTTTGTC; (LC480 only)

• Il-6F: ACCGCTATGAAGTTCCTCTCTGC,

• Il-6dR: CCAGAAGACCAGAGGAAATTTTC;

• Il-12p40F: AGATGACATCACCTGGACCT,

• Il-12p40R: GCCATGAGCACGTGAACCGT. The primer sequences for β-actin, Igtp, Iigp1 and 16S rRNA have been reported previously (15, 37).

DNA extraction and quantitative PCR (qPCR) analysis

DNA was extracted from mouse tissues, as described elsewhere (38), employing sequential digestions with a 0.1% collagenase A solution and a 0.2 mg/ml proteinase K solution. Following a series of phenol/choloroform extraction and ethanol precipitation steps, the purified DNA was diluted to 5 μg/ml for use in qPCR assays. Continuous fluorescence monitoring PCR with SYBR green I as the detection dye (Molecular Probes) was conducted with the ntm17 primer pair and normalized to 1000 copies of mouse nidogen, as previously described (38).

Collagenase release of ankle cells

To release cells from the ankle tissue, both rear ankles were excised from uninfected, 10, or 14 days post-B. burgdorferi-infected C3H or C57BL/6 mice following skin removal and incubated for 2 hours at 37°C in RPMI-1640 (Invitrogen) containing 10% fetal calf serum (FCS) and 1 mg/ml Collagenase A (Roche), as described previously (16). Two color flow cytometry analysis was conducted to assess whether the numbers of conventional (cDCs) (CD11c+B220−) or plasmacytoid (pDCs) (CD11c+ B220+) dendritic cells increase in the ankle joints following B. burgdorferi infection. The increase in cDC and pDC cell numbers was calculated relative to values obtained for uninfected mice.

Cell culture

Bone-marrow derived macrophages were isolated from the femurs and tibias of mice, as previously described (39), by culture in RPMI-1640 media (Invitrogen) supplemented with 30% L929 conditioned medium and 20% horse serum (Hyclone), demonstrated to be free of Mycoplasma with the MycoProbe Mycoplasma detection kit (R&D Systems). Harvested macrophages were replated in 6 well dishes at a density of 6 × 10⁵/ml in media lacking serum and containing 1% Nutridoma. After an overnight incubation, the media was removed and replaced with media alone or media containing 7.4 × 10⁶/ml live B. burgdorferi cN40 (washed twice in PBS to remove BSKII components). Macrophages were stimulated at 37°C, 5% CO₂ for 24 hours, after which the cells were harvested for RNA extraction.

To select for conventional dendritic cells utilizing a low-density protocol, bone marrow cells were cultured in RPMI-1640 media (Invitrogen) containing GM-CSF (kindly provided by Ray Daynes, University of Utah), as previously described (40). Harvested cells were magnetically labeled with α-CD11c microbeads, applied to MACS columns (Miltenyi Biotec), and the CD11c+ cells retained by the column were eluted. The purity of these cells was confirmed via flow cytometry staining using α-CD11c and α-B220 antibodies. On average, 90% of the cells recovered utilizing this protocol were CD11c+ B220−. 1.5 ml of mDCs (1.5 × 10⁶ cells) and 1.5 ml of media only, media plus 1000 U/ml rIFNβ (PBL Interferon Source), 7.4 × 10⁶/ml live B. burgdorferi (Bb), Bb + IFNβ, 10 μg/ml α-OspA (determined to be negative for Mycoplasma) (41), or Bb + α-OspA were added to 17 × 100 mm snap-cap tubes (Falcon) and incubated for 24 hours at 37°C, 5% CO₂. Cells were then collected by centrifugation, and RNA was extracted.

pDCs were isolated from the spleens of mice that had received daily intraperitoneal injections of rFlt3L (BioExpress Cell Culture Services, West Lebanon, N.H.) for 9 days (42). Spleens were removed and dissected into RPMI-1640 media containing 2 mg/ml Collagenase A and 20 μg/ml DNAse I. pDCs were magnetically labeled with α-PDCA1 microbeads (Miltenyi Biotec) and recovered via separation over a MACS column, followed by elution of PDCA1+ cells. Purity was assessed via flow cytometry analysis utilizing α-CD11c and α-B220 antibodies. On average, 95% of the cells recovered utilizing this protocol were CD11c+ B220+. pDCs were seeded into 24 well plates at 7 × 10⁵/ml in 1 ml of either complete media, or media containing the following stimuli: 100 U/ml rIFNα (PBL Interferon Source), 7.4 × 10⁶/ml live B. burgdorferi (Bb), Bb + IFNα, 10 μg/ml α-OspA (41), or Bb + α-OspA. Following a 24 hour incubation at 37°C, 5% CO₂, cells were harvested for RNA extraction.

Statistics

The non-parametric Mann-Whitney U-test was utilized for data values obtained from PCR, RT-PCR, and histopathological scoring of arthritis lesions. One-way ANOVA with the Bonferroni post-test was utilized to assess the significance of data obtained from ankle measurements and collagenase-digested ankle joints. All statistical analyses were conducted using GraphPad InStat3 version 3.0b for MacIntosh (GraphPad Software, San Diego, CA). For all tests, p < 0.05 was considered significant.

The adaptive immune response is not required for induction of IFN-profile genes in C3H mice

Previous studies demonstrated that a large number of IFN-responsive genes were significantly upregulated by 1 week post-B. burgdorferi infection within the rear ankle joints of arthritis-susceptible C3H mice, but not in mildly arthritic C57BL/6 mice, (15). Elevated transcription of IFN-responsive genes could be linked to the severe arthritis phenotype manifested by C3H, but not C57BL/6, mice at later timepoints. Alternatively, triggering of these genes might occur as a result of early infiltrating B and T lymphocytes, which have been reported to secrete IFNγ in response to B. burgdorferi (43–46). To distinguish between these two scenarios, C3H scid mice, which develop severe arthritis (9), were infected with B. burgdorferi, sacrificed at 1 week post-infection, and quantitative RT-PCR of selected IFN-responsive genes was conducted on RNA isolated from both rear ankle joints. Figure 1 depicts transcript levels obtained for four IFN-inducible genes: the IFNγ inducible GTPase, Igtp (A); the IFN-inducible GTPase 1, Iigp1 (B); the oligoadenylate synthase-like 2 gene, Oasl2 (C); and the T cell recruiting chemokine Cxcl9 (Mig) (D). While the majority of these genes have been previously annotated as IFNγ-inducible, increased transcription of these genes can also be triggered by Type I IFN (47–49). These four genes were selected as representatives of the IFN-responsive profile, as Affymetrix microarray analyses revealed these to be among the most highly upregulated genes at 1 week post-infection. For example, both Igtp and Iigp1 expression levels were elevated greater than 100 fold versus values recorded for uninfected mice (15). Each of the four examined IFN-responsive gene transcripts were elevated within the ankle joints of C3H scid mice to similar levels, relative to uninfected mice, as those obtained for wild-type C3H mice (Fig. 1). These data indicate that early recruitment/stimulation of adaptive lymphocytes is not responsible for the transcription of IFN-responsive genes within the ankle joints of C3H mice at 1 week post-infection. Instead, induction of these genes is mediated by cells of the innate immune response.

IFNAR1 blockage results in decreased arthritis severity and diminished IFN-responsive gene expression, but does not alter host defense

A previous report by Brown and colleagues demonstrated that C3H IFNγ−/− mice develop severe arthritis following B. burgdorferi infection, indicating IFNγ is not required for Lyme arthritis development (17). However, much controversy has remained in the field about the impact of IFNγ on arthritis development and resolution (10, 50, 51). When combined with our observation that IFN-responsive gene induction in C3H scid mice mimics that seen in wild-type mice (Fig. 1), it is highly probable that transcription of any IFN-inducible gene associated with arthritis development would be triggered by innate immune cells and could be a consequence of Type I IFN. In line with this hypothesis, the contribution of Type I IFN to Lyme arthritis severity in C3H mice at 4 weeks post-infection was assessed utilizing a mouse monoclonal antibody directed against murine IFNAR1, administered 1 day prior to infection. Overall, mice injected with the α-IFNAR1 monoclonal antibody exhibited reduced ankle swelling, compared with values obtained for mice treated with PBS or IgG1. Individually, a broad spectrum of arthritis phenotypes was noted for joints obtained from IFNAR1-blocked mice ranging from a drastic reduction in ankle swelling for two mice, to an intermediate phenotype for two mice, to moderately severe ankle swelling for one mouse (Fig. 2A). There was significant concordance between ankle swelling measurements and histopathology, as the two mice exhibiting the least amount of ankle swelling also received greatly reduced lesion scores. A significant reduction in lesion pathology was noted in α-IFNAR1-treated mice when compared with IgG1-treated mice (Fig. 2B). α-IFNAR1-injected mice displayed decreased edema, fewer neutrophils, and less thickening of the cranial tibial tendon sheath (Fig. 3D). These observations were in stark contrast to those noted for PBS- or IgG1 treated mice, both of which exhibited extensive edema, neutrophil infiltration, and thickening of the tendon sheath (Fig. 3B, 3C), features which are classic hallmarks of Lyme arthritis (5, 6). Similar results were also obtained in another experiment. To determine whether blockage of Type I IFN impacts immunological clearance of B. burgdorferi, qPCR was conducted on DNA extracted from ankle joints of α-IFNAR1-injected mice. There was no significant difference in spirochete numbers within the joints of IFNAR1-inhibited mice versus those detected in PBS- or IgG1-treated mice (Fig. 2C).

To determine whether the decrease in arthritis severity observed as a result of IFNAR1-blockage was linked to a suppression of IFN-inducible genes, the effect of α-IFNAR1 pre-treatment in triggering of IFN-responsive genes at 1 week post-infection was also assessed. Transcript levels for all four examined IFN-responsive genes were significantly suppressed within the ankle joints of C3H mice that received the α-IFNAR1 blocking antibody, when compared with message levels obtained for infected mice treated with either PBS or IgG1. The diminished induction of transcription observed for these genes in the presence of α-IFNAR1 was striking, as message levels were reduced by 68% for Cxcl9, 70% for Iigp1, and 81% for Igtp. Intriguingly, induction of Oasl2 transcription was completely suppressed, as message levels in α-IFNAR1-treated mice were reduced to the baseline level obtained for uninfected C3H mice (Table I). Together, these data provide strong evidence that Type I IFN is involved in induction of the IFN-responsive profile at 1 week post-infection. The above results also strongly link Type I IFN with arthritis development and indicate that Type I IFN may partially exert its pro-arthritic effects through enhanced transcription of IFN-responsive genes by innate immune cells residing within ankle tissue. The kinetics of B. burgdorferi dissemination within α-IFNAR1-blocked mice was also monitored at 1 week post-infection to assess whether the spirochete burden within the rear ankle joints differed between these mice and those receiving IgG1 or PBS prior to infection. Quanititative RT-PCR utilizing primers specific for 16S rRNA revealed low but overlapping B. burgdorferi numbers within the joints of mice from all treatment groups at 1 week post-infection (data not shown), indicating that administration of α-IFNAR1 did not alter bacterial numbers in the joints.

IFNγ blockage has no effect on arthritis development or host defense to B. burgdorferi but does suppress induction of IFN-responsive gene transcripts at 1 and 2 weeks post-infection

As an additional test of the selective effect of Type I IFN on arthritis development, we decided to inhibit Type II IFN. Mice were administered α-IFNγ blocking antibody, PBS, or hamster IgG (hIgG) at various intervals both prior to, and following B. burgdorferi infection. Blockage of IFNγ did not impact arthritis development in mice examined at 4 weeks post-infection (data not shown), consistent with a previous report that IFNγ−/− C3H mice develop severe Lyme arthritis similar to that noted in wild-type mice (17). The effect of IFNγ inhibition on B. burgdorferi numbers within ankle tissues at 4 weeks post-infection was assessed by qPCR and there were no significant differences in the number of spirochetes detected within the ankle joints of α-IFNγ treated mice, when compared with values obtained for control treatment groups (data not shown). These data support Brown and colleagues’ previous conclusion that control of spirochete numbers within the ankle joints of infected C3H mice is IFNγ-independent (17).

Although arthritis proceeded uninhibited in IFNγ-blocked mice, α-IFNγ-treated mice exhibited a striking 100% reduction in the upregulated expression of the IFN-responsive genes Igtp and Cxcl9, when compared with mice receiving PBS or hIgG at 1 week post-infection. Iigp1 transcript levels were 97% suppressed in α-IFNγ-blocked mice (Table II). Oasl2 gene induction can be mediated by either Type I IFN or IFNγ (47, 48). Interestingly, IFNγ inhibition caused a 73% suppression of Oasl2 (Table II). In light of the 100% reduction in Oasl2 message levels obtained with the α-IFNAR1 blocking antibody (compare Table I and Table II), these data suggest that Oasl2 induced expression is more effectively reduced by Type I IFN blockage. Our previous microarray study found that while expression of IFN-responsive genes declined by 2 weeks post-B. burgdorferi infection, message levels for many transcripts were still several-fold higher than those obtained for uninfected mice(15). For this reason, the effect of IFNγ blockage was also assessed at 2 weeks post-infection. Markedly reduced transcript levels (90–100% suppression) were obtained for Igtp, Iigp1, and Cxcl9, whereas α-IFNγ treatment had little effect on Oasl2, suppressing its expression by only 43%, when compared with PBS- and hIgG- treated mice (Table II). Analyses of 16S rRNA transcript levels indicated that similar, low levels of B. burgdorferi were present within the joints of mice from all treatment groups at both 1 and 2 weeks post-infection (data not shown), indicating that IFNγ blockage did not alter the kinetics of bacterial dissemination. Finally, we assessed the effect of simultaneous blockage of both Type I IFN and IFNγ on induction of IFN-responsive genes at 1 week post-infection. Induced expression of all four examined IFN-responsive transcripts was 100% abolished within the joints of C3H mice injected with both α-IFNAR1 and α-IFNγ, as message levels were suppressed below the baseline values obtained for untreated, uninfected mice (Table III). Together these results support the unique role of Type I IFN in arthritis development, but also provide compelling evidence that induction of IFN-responsive genes within the joints of C3H mice at 1 week post-infection reflects the contribution of both Type I and II IFNs.

Plasmacytoid (pDC) and conventional DC (cDC) traffic to the joints following infection, but are not the initial source of B. burgdorferi-induced Type I IFN

In an attempt to identify the cell type(s) responding to Type I IFN to trigger arthritis development, we next questioned whether dendritic cells, either resident in, or trafficking to the joint following infection, are involved in the production of Type I IFN and/or in Lyme arthritis development. pDCs have long been recognized as the major producers of, and responders to Type I IFN, but cDCs also possess the ability to both synthesize and react to Type I IFN (52, 53). To this end, we assessed whether pDCs and/or cDCs traffic to the joint following B. burgdorferi infection. Cells released from collagenase-digested ankle joints excised from C3H mice at 7 and 14 days post-infection were analyzed for CD11c and B220 surface expression, and compared with values obtained from uninfected ankle joints. By 7 days post-infection an increase in the numbers of both CD11c+B220- (myeloid or cDCs) and CD11c+B220+ (pDCs) cells were observed in C3H ankle joints, with even greater numbers observed at 14 days post-infection (Fig. 4). However, blockage of pDCs in C3H mice utilizing α-PDCA-1 antibody had no impact on arthritis severity (data not shown). Taken together, these results demonstrate that infiltration of both cDCs and pDCs by 1 week post-infection may occur as a response to infection and inflammation, but that pDCs do not mediate arthritis development in C3H mice.

However, since increased numbers of DCs were detected in infected joint tissue as early as 1 week post-infection, a time-frame coincident with IFN-responsive gene induction, we also investigated the ability of both splenic pDCs and bone marrow-derived cDCs to induce expression of the IFN-profile in response to in vitro B. burgdorferi-stimulation. Although both cDCs and pDCs exhibited elevated IL-6 and/or IL-12 transcripts and protein (data not shown) in the presence of live B. burgdorferi, IFN-responsive transcripts were not induced in response to bacterial stimulation alone. In addition, pDCs stimulated in the presence of IFNα and B. burgdorferi also failed to upregulate IFN-inducible genes. However, a synergistic increase in expression was detected in cDCs administered IFNβ and live spirochetes, suggesting that priming may be required for optimal cDC contribution to the IFN response (Table IV). It has been reported that pDCs exhibit a diminished capacity for phagocytosis of antigens, relative to other APCs, such as cDCs and macrophages (52). To enhance the ability of pDCs to internalize live bacteria, antibody to the abundantly expressed outer surface protein A (OspA) was added to pDCs or cDCs, prior to the addition of B. burgdorferi. Unexpectedly, incubation of cDCs with α-OspA antibody resulted in induction of all four examined IFN-responsive genes, but not in IL-12 or IL-6 transcripts. In the presence of opsonizing antibody, neither Borrelia-stimulated pDCs nor cDCs exhibited further upregulation of IFN-responsive genes (Table IV). Taken together, the above data strongly imply that while DCs may contribute to either amplification of IFN-profile genes, or resolution of this response in vivo, they are unlikely to be the first cell type driving the transcription of Type I IFN-responsive genes in response to B. burgdorferi infection.

Bone marrow-derived macrophage induction of IFN-responsive genes in response to in vitro B. burgdorferi-stimulation is Type I IFN-dependent and occurs via a MyD88-independent signaling pathway

Another innate immune candidate cell, the bone marrow-derived macrophage (BMMφ) was then evaluated for its ability to upregulate IFN-inducible genes in response to B. burgdorferi. Twenty-four hours after the addition of B. burgdorferi, transcripts for Igtp (Fig. 5A), Iigp1 (Fig. 5B), Oasl2 (Fig. 5C), and Cxcl9 (Fig. 5D) were upregulated in BMMφ isolated from both C57BL/6 and C3H/HeN mice. BMMφ from C3H/HeJ mice, which lack a functional Tlr4 gene (54), also displayed upregulation of IFN-responsive transcripts, indicating that transcription of these genes was not due to low-level LPS contamination of our B. burgdorferi grown in complex media (Fig. 5). To assess whether borrelial lipoproteins or unmethylated CpG DNA were being sensed by TLRs 2 and 9, respectively, to enact transcription of IFN-inducible genes, BMMφ isolated from TLR2−/− and TLR9−/− mice were stimulated with B. burgdorferi. Transcription of IFN-inducible genes proceeded in the absence of TLR 2 or 9, indicating that these TLRs are not utilized by B. burgdorferi to upregulate IFN-responsive transcripts. Induction of IFN-responsive transcripts is also MyD88-independent, as MyD88−/− BMMφ stimulated with B. burgdorferi upregulate IFN profile genes. Intriguingly, BMMφ isolated from IFNAR1−/− mice failed to induce IFN-responsive genes as a result of B. burgdorferi stimulation. This observation provides strong evidence that Type I IFN modulates transcription of IFN-responsive genes in macrophages, and underscores the utility of the macrophage as a model cell to examine the mechanisms of IFN-responsive gene induction following B. burgdorferi infection.

NK cell depletion has no effect on arthritis development or host defense to B. burgdorferi at 4 weeks post-infection, but NK cells are major IFNγ contributors to induction of the IFN-responsive profile at 1 week post-infection

Our above results demonstrating that IFNγ blockage suppressed IFN-responsive gene transcription without impacting arthritis severity was intriguing (Table I), especially in light of conflicting data in the field concerning a role for IFNγ in Lyme arthritis development (17, 50, 51). Our finding that IFN-responsive genes are still induced in C3H scid mice (Fig. 1) indicated that cells other than T lymphocytes were responsible for both the Type I- and IFNγ-inducible portions of the profile. NK cells are potent producers of IFNγ, and have been previously shown to secrete this cytokine in response to B. burgdorferi (55) (56). NK cells were depleted from C3H mice using α-asialo GM1, to determine whether these cells impact arthritis development or host defense to B. burgdorferi. α-asialo GM1 was employed to neutralize NK cells because C3H mice produce the NK1.2 allele, to which there are no commercially available antibodies. In agreement with data previously obtained with the footpad-injection model (56), ankle swelling measurements and histopathological lesion scores obtained for α-asialo GM1-injected mice did not differ significantly from those recorded for PBS- or vehicle-treated animals (data not shown). Similar numbers of spirochetes were present in both the joints of α-asialo GM1-treated mice and in control mice at 4 weeks post-infection (Table V), indicating that NK cell depletion does not affect the ability of the host immune system to respond to B. burgdorferi challenge. We also examined whether NK cells are a major source of IFNγ driving the induction of the IFN-profile at 1 week post-B. burgdorferi-infection. Strikingly, mice treated with α-asialo GM1 exhibited a significant reduction (77–98%) in the expression levels of Igtp, Iigp1, and Cxcl9, when compared with message levels obtained for PBS- and rabbit IgG-treated mice. In contrast, Oasl2 message levels were only decreased by 54% within the joints of α-asialo GM1 treated mice (Table V), an observation that supports other data presented in this report indicating that the magnitude of Oasl2 expression shows greater dependence on Type I IFN. These data clearly demonstrate that NK cells are a major IFNγ source contributing to induction of IFN-responsive profile genes at 1 week post-infection. However, as we have reported here, in confirmation of previously published reports (17, 56), NK cells and IFNγ production do not contribute to arthritis development, reinforcing the idea that the entire IFN-profile is not required for arthritis development.