Neurodegeneration Enhances the Development of Arthritis

The prevalence of neurodegenerative disorders increases rapidly with age in Western societies (1, 2). In addition, peripheral inflammatory diseases, such as arthritis, increase with age (3–5). Surprisingly, little is known about the interaction between neurodegeneration and peripheral inflammatory diseases. Over the last decades, the CNS has been considered as immunologically privileged (6); however, it is now widely acknowledged that tight interactions and feedback loops exist between the CNS and the peripheral immune system (7). For instance, the lymphatic system, a central feature of the immune system, is connected to the CNS (8). It is also well established that peripheral immune activation and inflammation induce CNS effects: this link, for instance, is reflected by the fever response in the CNS that is induced by IL-1 (9, 10). Furthermore, central pain sensitization and sickness behavior are triggered via the production of proinflammatory cytokines in the context of arthritis (11–13). In contrast, efferent pathways were described whereby the CNS influences the immune system (e.g., cholinergic neurons regulate activation and cytokine production of immune cells, such as macrophages, by activating the α7 subunit of nicotinic acetylcholine receptors) (14, 15).

Much less is known about the interactions between neurodegenerative disease and peripheral inflammation. In particular, it is unknown how neurodegeneration influences peripheral inflammatory processes. Neurodegenerative disorders, such as Alzheimer’s disease, are associated with immune activation, especially the activation of microglia around plaques, an increase in cytokine production, and activation of the complement pathway (16). Hence, peripheral immune activation may also affect neurodegeneration by influencing immune responses to plaques. Indeed, active immunization was attempted to resolve neurodegeneration (17, 18). Furthermore, resolution of inflammation in the context of neurodegeneration was shown to involve migration of peripheral immune cells into the brain (19, 20).

To examine whether neurodegeneration influences peripheral inflammation, we induced arthritis in tau-transgenic (tau-tg) PS19 mice. These mice express the human tau P301S mutant in the CNS under control of the murine prion protein promoter (21). Tau P301S is responsible for an inherited form of frontotemporal dementia and parkinsonism linked to chromosome 17 (FTDP17) (22). Tau-tg mice display neurodegeneration associated with tau protein deposition and early microglial activation in the CNS (21). We challenged these mice with collagen-induced arthritis (CIA), a standard model of autoimmune inflammatory arthritis that shares many pathological features with rheumatoid arthritis, including joint swelling and damage, mononuclear cell infiltration, and increased production of proinflammatory cytokines (23, 24). In addition, we tested whether induction of arthritis influences neurodegenerative disease, with special regard to the deposition of tau tangles in the CNS.

The transgenic TauP301S strain (21) [B6;C3-Tg(Prnp-MAPT*P301S)PS19Vle/J; Jackson Laboratory] (denoted as tau-tg in this article) was routinely backcrossed to C57BL/6 mice (Charles River). Mice were kept in a 12-h light–dark cycle and had free access to food and water. Animal experiments were conducted according to institutional and governmental guidelines (#55.2-2532.1-42/14 and #55.2-2532.1-59/14; Government of Unterfranken, Bavaria, Germany).

Six-month-old tau-tg mice and nontransgenic (non-tg) littermates were subjected to CIA via s.c. injection at the tail base with 100 μl of an 1:1 emulsion of chicken collagen type II (CII; 4 mg/ml) and CFA (both from Sigma-Aldrich) containing 5 mg/ml heat-inactivated Mycobacterium tuberculosis (H37Ra; Difco), followed by booster injections after 21 and 35 d. Every third day, body weight was recorded, and clinical arthritis was scored from 0 (no clinical signs) to 3 (severe swelling and erythema) for every paw. Mice were sacrificed on day 50; blood and serum were collected, followed by transcardial perfusion with ice-cold PBS prior to tissue harvesting. In total, we analyzed five independent cohorts including 90 tau-tg and 76 control mice. We used animals from at least two cohorts for all experimental analyses. Arthritic tau-tg animals and non-tg littermates with a score ≥ 3 (at least one swollen paw) were selected for flow cytometric and histological analyses.

Hind paws were fixed for 6 h in 4% paraformaldehyde and decalcified in EDTA (Sigma-Aldrich). Paraffin sections (2 μm) were stained for tartrate-resistant acid phosphatase to analyze osteoclast numbers and bone erosions, as well as with H&E to determine the inflammation area. Histological analysis was performed using a Nikon microscope equipped with OsteoMeasure analysis software (OsteoMetrics). After perfusion, brains were removed, and one hemisphere was fixed overnight at 4°C in 4% paraformaldehyde. Brain paraffin sections (8 μm) were assessed for Gallyas silver staining and TUNEL assay, according to the manufacturer’s instructions (Roche). To detect microglia in brain, immunohistochemical staining was performed using rabbit anti–Iba-1 (dilution 1:1000; Wako) and secondary biotinylated goat anti-rabbit IgG (Vector Laboratories), followed by development via an ABC HRP Kit and a DAB Peroxidase Substrate Kit (both from Vector Laboratories). Tau tangles were visualized by staining with biotinylated mouse anti-hyperphosphorylated (hp) tau (clone AT8, dilution 1:100; Thermo Scientific) using a Mouse on Mouse (M.O.M.) Detection Kit (Vector Laboratories), according to the manufacturer’s instructions, and cyanine Cy5-conjugated streptavidin (dilution 1:300; Jackson ImmunoResearch). Splenic cryosections (6 μm) were stained with goat anti-choline acetyltransferase (ChAT; dilution 1:1000; Novus), as described previously (25), and detected with Alexa Fluor 488–labeled donkey anti-goat IgG (dilution 1:300; Invitrogen).

Quantification of Iba-1⁺, Gallyas⁺, and AT8⁺ cells was performed in a blinded manner using Adobe Photoshop CS5. Cells in a given area were counted manually. The area was converted from pixel to square millimeters according to calibration factors and image size. Data are represented as cell number per square millimeter of tissue area. To quantify AT8⁺ tau deposits in axons, degenerated axons, and extracellular space, we first marked the soma of AT8⁺ cells, which was then excluded from the counting procedure. Area of AT8⁺ tau deposits with a standardized defined threshold of fluorescence intensity was quantified automatically within a defined area by Adobe Photoshop and set in relation to total tissue area. To determine colocalization of AT8 and Iba-1, PFA-fixed paraffin sections were stained with biotinylated mouse anti-hp Tau (clone AT8, dilution 1:100; Thermo Scientific), followed by Cy5-conjugated streptavidin (dilution 1:300; Jackson ImmunoResearch) and rabbit anti–Iba-1 (dilution 1:1000; Wako), followed by Alexa Fluor 594–conjugated donkey anti-rabbit (1:300; Invitrogen). The number of Iba-1⁺ cells per section that were also AT8⁺ was enumerated by Adobe Photoshop CS5.

Serum concentrations of total IgM and IgG were measured by ELISA (Bethyl Laboratories). To determine collagen-specific IgG in sera, plates were coated with 10 μg/ml chicken CII (Sigma-Aldrich) overnight at 4°C. Sera (dilution 1:5000) were added, and anti-collagen Abs were detected using a rabbit anti-mouse HRP-conjugated Ab. To test the affinity of collagen-specific IgG, increasing concentrations (0–3.0 M) of potassium thiocyanate were used. Anti-Tau autoantibodies were measured as reported previously (26). Briefly, 96-well microplates were coated with 50 μl of recombinant human tau peptide (2.5 μg/ml, Tau-412 [1N4R]; rPeptide) in bicarbonate buffer (pH 9.6) overnight at 4°C. Wells were blocked with 200 μl of 1% BSA in 1× PBS (pH 7.1) for 1 h at room temperature. Plates were washed with 1× PBS and incubated for 2 h with 50 μl of sera (diluted 1:300 in blocking buffer) at 37°C. Plates were washed again and incubated with 50 μl of HRP-conjugated rabbit anti-mouse Ab (diluted 1:5000 in blocking buffer) for 90 min at 37°C. After washing, the reaction was visualized by adding 50 μl of tetramethylbenzidine (eBioscience) substrate solution for 30 min at room temperature in the dark. The reaction was stopped by adding 2 mol/l H₂SO₄ (eBioscience), and absorbance was read at 450 nm with a reference filter of 620 nm. Serum cortisol levels were quantified by ELISA, according to the manufacturer’s instructions (Innovative Research). Serum concentrations of cytokine/chemokine profiles were determined using multiplex bead technology (eBioscience).

Isolated brain hemispheres and peripheral organs were snap-frozen and stored at −80°C. Protein was extracted by automated mechanical homogenization (Precellys homogenizer; Peqlab) in T-PER buffer–supplemented commercial proteinase inhibitor mixture without EDTA (Halt Protease Inhibitor Cocktail 100×; Thermo Scientific). Protein content was measured with a BCA Protein Assay Kit, according to the manufacturer’s instructions (Thermo Scientific). Brain tissue cytokine levels were determined using DuoSet ELISA kits, according to the manufacturer’s instructions (R&D Systems), and normalized to total brain protein. For immunoblotting, 20 μg of protein was separated by 10% SDS-PAGE, followed by blotting onto a nitrocellulose membrane. Blots were incubated with a polyclonal goat anti–pan-tau Ab (Santa Cruz Biotechnology) or anti-actin Ab (Sigma-Aldrich), followed by HRP-conjugated secondary Abs and development with ECL.

Brain hemispheres were snap-frozen and stored at −80°C. Total RNA of brain hemispheres was isolated via mechanical tissue homogenization (Peqlab) using a PARIS Kit (Thermo Scientific), according to the manufacturer’s instructions. RNA (1 μg) was subjected to cDNA transcription using oligo(dT) primers and MuLV Reverse Transcriptase (Roche). Real-time quantitative PCR was performed with SYBR Green I-dTTP (Eurogentec) and normalized to β-actin using the following primer pairs: hp-tau forward 5′-GGCATCTCAGCAATGTCTCC-3′ and reverse 5′-GGTATTAGCCTATGGGGGACAC-3′ and β-actin forward 5′-TGTCCACCTTCCAGCAGATGT-3′ and reverse 5′-AGCTCAGTAACAGTCCGCCTAGA-3′. Expression levels were calculated using the ΔΔ threshold cycle method.

Splenic single-cell suspensions were generated as described previously (27). Heparinized blood was subjected to erythrocyte lysis with hypotonic lysis buffer. Isolated spleen and blood cell pellets were taken up in RPMI 1640 (Life Technologies) supplemented with 5% FCS.

Brain cells were harvested after transcardial perfusion with ice-cold PBS, according to previous reports (28). Fc receptor blocking was performed for 30 min with anti-CD16/32 (BioLegend) for all samples. The following Abs against mouse Ags were used: CD45R (B220), CD45, CD138, GR-1, Ly6C (BioLegend), CD45, CD11b (eBioscience), and CCR2 (Novus). Cell staining was measured in a BD Gallios flow cytometer and analyzed using Kaluza software.

Evans Blue (Sigma-Aldrich) was dissolved in sterile 1× PBS and 200 μl of a 2% solution, injected i.p., and allowed to circulate for 24 h. Mice were sacrificed and perfused transcardially with 1× PBS, and brains were removed and transferred to reaction tubes containing 500 μl formamide (Carl Roth). After 24 h of incubation at 55°C, tubes were centrifuged, and absorbance of the supernatant was measured at 620 nm.

The following Abs were used for immunohistological analyses. Anti–Iba-1 polyclonal unconjugated (rabbit) (dilution 1:1000, cat. no. 019-19741, Wako Chemicals) and anti-rabbit IgG biotinylated (goat) (dilution 1:200, cat. no. BA-1000, Vector Laboratories) or Alexa Fluor 594–conjugated donkey anti-rabbit (1:300; Invitrogen) were used for microglia/macrophage detection in brain. Anti–PHF-tau Ser²⁰²/Thr²⁰⁵ AT8 biotinylated (mouse) (dilution 1:100, cat. no. MN1020B, Thermo Scientific) and Streptavidin Cyanine Cy5 (dilution 1:300, cat. no. 016-170-084, Jackson ImmunoResearch) were used to visualize phospho-tau in brain. Anti-ChAT polyclonal unconjugated (rabbit) (dilution 1:1000, cat. no. NB110-89724, Novus) and anti-goat IgG Alexa Fluor 488 (donkey) (dilution 1:300, cat. no. A-11055, Thermo Scientific) were used to quantify ChAT-expressing cells in the spleen. For anti-collagen IgG ELISA, anti-mouse Ig polyclonal HRP (rabbit) (dilution 1:5000, cat. no. p0260, Dako) was used as a secondary Ab. Immunoblotting for pan-tau in various organ lysates and brain was performed using anti-tau C-17 polyclonal unconjugated (goat) (dilution 1:1000, cat. no. sc-1995, Santa Cruz Biotechnology), secondary anti-goat IgG polyclonal HRP (donkey) (dilution 1:10,000, cat. no. 6420-05, SouthernBiotech), anti-Actin polyclonal (rabbit) (dilution 1:1,000, cat. no. A2066, Sigma-Aldrich), and secondary anti-rabbit Ig HRP (goat) (dilution 1:10,000, cat. no. 4010-05, SouthernBiotech).

Flow cytometric analyses were performed using anti-mouse CD16/32 unconjugated clone 93 (rat) (dilution 1:200, cat. no. 101302, BioLegend), anti-mouse/human CD45R/B220 PE/Cy5 clone RA3-6B2 (rat) (dilution 1:2000, cat. no. 103210, BioLegend), anti-mouse CD45 Pacific Blue clone 30-F11 (rat) (dilution 1:1000, cat. no. 103126, BioLegend), anti-mouse CD138 PE/Cy7 clone 281-2 (rat) (dilution 1:1000, cat. no. 142514, BioLegend), anti-mouse Ly-6G/Ly-6C (Gr-1) PE clone RB6-8C5 (rat) (dilution 1:2000, cat. no. 108408, BioLegend), anti-mouse Ly-6C PE/Cy7 clone HK1.4 (rat) (dilution 1:3000, cat. no. 128017, BioLegend), anti-mouse CD45 allophycocyanin clone 30-F11 (rat) (dilution 1:500, cat. no. 17-0451-82, eBioscience), anti-mouse CD11b PerCP-Cyanine5.5 clone M1/70 (cat. no. 45-0112-82, eBioscience), and anti-mouse CCR2 DyLight 488 polyclonal (rabbit) (cat. no. NBP1-48338G, Novus).

Statistical analyses were performed using the two-tailed Mann–Whitney U test, and data are presented as mean ± SEM, as well as medians. GraphPad Prism 5.03 software was used for statistical calculations and data presentation.

To determine whether progressive neurodegeneration can affect peripheral autoimmune-driven inflammation, we induced arthritis in 6-mo-old tau-tg mice and non-tg littermates. To induce arthritis, mice were immunized with chicken CII and CFA, followed by booster injections on days 21 and 35; disease development was monitored until day 50 (Fig. 1A, 1B). We observed a significantly higher incidence of CIA in tau-tg mice. In addition, the onset occurred earlier (day 21) in comparison with non-tg littermate controls (day 26) (Fig. 1C). Both findings were further associated with significantly higher clinical paw swelling scores in tau-tg mice (Fig. 1D) and significant changes in histomorphometric values with respect to tissue inflammation, bone erosion, and numbers of osteoclasts in paw sections (Fig. 1E). Thus, cerebral mutant human tau expression increases the susceptibility to arthritis.

To determine the reason for enhanced arthritis, we further characterized the inflammatory phenotype of tau-tg mice. Naive tau-tg mice displayed a sex-independent reduction in body weight (Fig. 1F), indicative of a cachectic phenotype often associated with inflammation. In fact, naive tau-tg mice revealed higher basal serum levels of typical proinflammatory cytokines and chemokines, such as TNF-α, IL-17A, and MCP-1, in comparison with wild-type (wt) littermates (Fig. 1G). During CIA, tau-tg mice exhibited significantly more pronounced increases in TNF-α, IL-17A, IFN-γ, and MCP-1 cytokine levels than did wt littermates (Fig. 1H). Of note, basal corticosteroid levels were not altered in tau-tg mice (Supplemental Fig. 1). Autoantibodies play a very important role in arthritis because they bind to collagen, which further results in complex formation that triggers a local inflammatory response in the joint (23). However, the specific humoral immune response against collagen, as assessed by Ab titer and Ab affinity, was not altered in tau-tg mice (Fig. 1I). Moreover, we could not detect differences in total IgG, total IgM, or spleen weight in tau-tg mice (Supplemental Fig. 2). Western blot analysis of brain and peripheral tissues was performed to exclude the possibility of altered immune cell activation due to transgenic human tau expression in peripheral tissues. We showed that transgenic human tau protein was only detectable in the brain (Fig. 1J). Thus, increased incidence of CIA in tau-tg mice is not based on ectopic tau expression. Taken together, overexpression of mutant human tau in the brain leads to enhanced cytokine responses and more severe arthritis.

To elucidate the mechanism that enhanced CIA in tau-tg mice, we analyzed proinflammatory immune cell populations. For this purpose, we performed flow cytometric analyses of spleen, blood, and brain tissue of untreated, arthritic tau-tg mice and wt controls at day 50. In general, CIA led to a significant increase in inflammatory monocytes (CD45⁺Ly6C^hiCD11b⁺), granulocytes (CD45⁺Ly6C^intCD11b⁺), and CCR2-expressing leukocytes (CD45⁺CCR2⁺) in the spleens of tau-tg and non-tg mice (Fig. 2A–C). We also observed increased plasmablasts (CD45⁺B220^lowCD138⁺) upon induction of CIA (Fig. 2A–C). Similar results were obtained for inflammatory monocytes in blood (Fig. 2D–F). Although the increase in proinflammatory cell populations in the spleen was expected during CIA (23, 29), it was very interesting to see that all described splenic immune cell subpopulations already were significantly enlarged under basal conditions in tau-tg mice (Fig. 2A–C). Further, naive tau-tg mice displayed significantly increased numbers of plasmablasts and inflammatory monocytes in the blood (Fig. 2D–F). These observations paralleled the elevated baseline amounts of proinflammatory cytokines in tau-tg mice (Fig. 1G). A specific effect was observed for the cell surface marker Ly6C. Together with CD11b, Ly6C is enriched in the myeloid lineage and, thus, is primarily expressed on monocytes and granulocytes. Cells with high expression levels of Ly6C are known to exhibit an inflammatory state (30–32). We found a mild upregulation of Ly6C on splenic inflammatory monocytes during arthritis but significantly higher Ly6C expression levels on granulocytes in arthritic tau-tg and wt mice. Interestingly, during CIA, only tau-tg animals exhibited a significant increase in Ly6C expression on blood inflammatory monocytes and granulocytes (Fig. 2A, 2D). Taken together, our data suggest that induction of neuroinflammation through transgenic expression of mutated tau alters peripheral innate immune cell subsets.

We next asked whether peripheral immune cellularity is in a balanced equilibrium with CNS immune cells. In fact, we detected significantly higher numbers of CD45^hi leukocytes in the brain of naive nonchallenged tau-tg mice (Fig. 2G, 2H), representing invading leukocytic cells (30, 33, 34). In accordance, these cells also expressed significantly higher levels of CCR2 (Fig. 2G), the chemokine receptor that controls the migration and activation of bone marrow–derived monocytes to sites of inflammation (35, 36). To assess whether this cell migration occurred actively or passively as a result of an opening of the blood–brain barrier during the course of neuroinflammation in tau-tg mice, we injected Evans blue dye into 10-mo-old tau-tg and wt mice. Because no bluish staining of brain tissue was detectable, indicating an intact blood–brain barrier, we concluded that cell migration in naive tau-tg mice occurred in an active way (Supplemental Fig. 3). Thus, transmigrating CD45^hi leukocytes could be the possible mediators of information exchange between the CNS and the peripheral immune system. Another possibility is the parasymphathetic vagus nerve, which is known to play a crucial role in the regulation of immune homeostasis via the release of acetylcholine. Acetylcholine dampens immune cell responses and, in this way, inflammatory processes as well (14, 37). To test this hypothesis, we investigated the splenic expression of ChAT; however, we could not detect differences in the numbers of ChAT-expressing cells in the spleens of naive tau-tg mice and non-tg littermates (Supplemental Fig. 4).

Previous studies demonstrated that systemic activation of the immune system in mice (e.g., induced by the administration of LPS) results in neuroinflammation and sickness behavior (38). Given that inflammation in the CNS of tau-tg mice elicits a proinflammatory state in the periphery, which, in turn, triggers a proinflammatory positive-feedback loop in the CNS, one would expect exaggerated inflammation in the brain of tau-tg mice with CIA. Therefore, we analyzed the abundance of the microglia activation marker Iba-1 on brain sections of naive and arthritic tau-tg and wt mice by immunohistochemistry. In accordance with the original publication, we observed increased numbers of Iba-1–expressing cells, indicative of microgliosis, in naive tau-tg mice in comparison with non-tg littermates, which, however, appeared milder than described (21) (Fig. 3A, 3B). Despite significantly increased Iba-1–expressing cells, arthritic tau-tg mice revealed no increase in cerebral cytokine production during the course of arthritis, which was in contrast to arthritic wt littermates (Fig. 3C). Although peripheral inflammation led to an upregulation of typical proinflammatory cytokines, such as IL-6, in the brains of wt mice, this central response to a peripheral stimulus was not detectable in tau-tg mice. Likewise, we did not observe an increase in the rate of cell apoptosis in brain sections of arthritic tau-tg mice, as determined via TUNEL assay (Fig. 3D). Hence, arthritis enhances cytokine expression in the CNS in wt, but not tau-tg, mice. In addition, we detected a boost in anti-tau Ab responses after induction of CIA, suggesting the induction of specific clearance mechanisms for tau-related pathology (Fig. 3E).

Based on the observations of increased Iba-1–expressing cells and anti-tau Ab formation in arthritic tau-tg mice, we speculated that arthritis may influence neurodegenerative disease in these mice. One previous study indicated that peripheral inflammation may exacerbate pre-existing tau pathology (39). We performed Gallyas silver staining (40) on brain sections of naive and arthritic tau-tg mice to detect tau aggregates. Remarkably, we noted a highly significant reduction in the number of cells containing tau aggregates throughout the brain in tau-tg animals that developed CIA (Fig. 4A, 4B). The same observation was made for axonal and extracellular tau deposits using the PHF-specific anti-tau Ab AT8 (Fig. 4C, 4D). To determine whether this effect arose from the actual peripheral inflammation or from immune cell activation through immunization, we also performed Gallyas silver staining, as well as AT8 immunostaining, in tau-tg mice that were immunized and produced anti-collagen Abs (data not shown) but did not develop any clinical signs of arthritis during the 50 days of the experiment. Stunningly, these mice exhibited the same tau burden as did untreated control animals (Fig. 4A–D). These results demonstrate that peripheral inflammation, but not antigenic challenge alone, can resolve tau pathology in tau-tg mice. To exclude that the CIA-induced reduction in tau pathology is due to silencing of the transgene, we examined the actual transgene expression at the mRNA level in brain tissues of arthritic and naive tau-tg mice in the same experiment. Transgene expression was unaltered in tau-tg animals with persistent arthritis, indicating that the reduction occurred at the protein level (Fig. 4E). To determine whether the reduction in tau pathology in arthritic tau-tg mice is an active process, we compared the development of tau pathology from 6 to 8 mo in nonimmunized tau-tg mice (i.e., the time frame used to induce and monitor CIA). Tau pathology increased normally in the absence of CIA, and this increase was especially evident for axonal tau deposits (Fig. 4F). Hence, we propose that the reduced tau pathology in arthritic tau-tg mice is an active process.

To determine a potential clearance mechanism, we assessed the uptake of AT8⁺ neurons and/or AT8⁺ tau deposits by Iba-1⁺ cells via immunofluorescence. In accordance with our previous analyses (Fig. 4A, 4D), we observed increased uptake of AT8⁺ tau by Iba-1⁺ cells in the hippocampus, hypothalamus, and entorhinal cortex of arthritic tau-tg mice (Fig. 5), pointing to enhanced phagocytic activity of microglia under arthritic conditions.

In this article, we show that neurodegeneration induced by overexpression of mutant human tau protein significantly increases the susceptibility to arthritis. The same form of mutated tau protein was shown to provoke neurodegenerative disease in humans (22, 41). Our finding is likely due to a proinflammatory phenotype in tau-tg mice elicited by mutant tau protein expression in the CNS. Proinflammatory cytokines were already significantly higher in the periphery of tau-tg mice compared with wt mice under steady-state conditions. This proinflammatory signature elicited by CNS overexpression of mutant human tau consisted of cytokines such as TNF-α and IL-17A, which are important in the pathogenesis of arthritis (42) and, moreover, exert synergistic effects in triggering joint inflammation (43–45). Our data suggest that this proinflammatory environment in tau-tg mice facilitates the development of CIA. In agreement with these findings, the numbers of effector cells, such as CD11b⁺Ly6C^hi monocytes and plasmablasts, were increased in naive tau-tg mice and increased further after induction of CIA. Specifically, tau-tg mice, but not wt mice, exhibited a significant increase in Ly6C expression on blood inflammatory monocytes and granulocytes during arthritis. It is conceivable that these Ly6C⁺ inflammatory monocytes and granulocytes are the source of proinflammatory TNF-α and IL-17 (46).

Our finding of a tau-induced proinflammatory state already in the early phases of neurodegenerative disease is remarkable, because we revealed that tau expression is restricted to the CNS and not found at peripheral sites. This proinflammatory state requires sensing of CNS tau pathology by the peripheral immune system, with subsequent migration of immune cells to the brain. The increased levels of circulating cytokines, peripheral effector cell activation, and the development of anti-tau Abs in tau-tg mice, but not in wt controls, clearly support this concept. Thus, neurodegeneration increases the susceptibility to inflammatory arthritis, which may also explain the increased prevalence of arthritis in aged individuals, in whom neurodegeneration is more prevalent as well (47). Interestingly, aged people frequently develop rheumatoid arthritis with a more inflammatory phenotype (48), which might also be linked to prevalent subclinical neurodegenerative processes occurring in normally ageing individuals (49, 50).

An unexpected finding in our study was that CIA elicited the resolution of neurodegeneration, indicating the existence of a negative-feedback loop between neurodegeneration and peripheral inflammation. In fact, induction of arthritis in tau-tg mice reduced tau tangle burden in several brain regions, such as the entorhinal cortex, the hippocampus, the hypothalamus, and the ventral striatum. Interestingly, previous studies showed that immunization can reduce amyloid β (Aβ) plaque burden (17, 18, 51). Furthermore, induction of arthritis also reduces Aβ plaque burden by local phagocyte activation (52). In support of this study, we show that Iba-1–expressing cells engulfed tau under CIA conditions, which went along with decreased tau deposits. This finding points to an active clearance process based on phagocytosis of tau by activated microglia. The above-mentioned clearance process appears to operate in a region-specific manner, because the cortex was somehow excluded from this process.

In our study, we found that anti-tau Ab responses were significantly enhanced in the context of arthritis. Immune responses against tau, which are also observed in tau-tg mice under steady-state conditions, may foster the clearance of tau protein deposits in the CNS. Indeed, phagocytosis of tau protein is increased by administration of anti-tau Abs (53). Although we excluded a general opening of the blood–brain barrier in our experiments, it is possible that Abs enter the CNS by the opening of tight junctions in the choroid plexus during inflammation, which accounts for enhanced IgG levels in the CNS in models of experimental encephalomyelitis (54). Support for such a concept comes from our observation that resolution of tau tangles in the CNS specifically occurred in mice that developed the inflammatory phase of arthritis and not only an anti-collagen immune response, suggesting that inflammation may be required to allow CNS activity of peripheral immunological constituents, such as monocytes or Abs.

Resolution of tau tangles is considered an active process, associated with limitation of inflammation resulting from tau deposition (55, 56). Importantly, transgenic tau expression itself was not inhibited by the induction of arthritis, suggesting a clearance mechanism coming into place. Indeed, we saw a significant increase in the number of Iba-1–expressing cells in tau-tg mice that developed arthritis. Microglia activation can be shown by Iba-1 expression and is considered an important regulatory response to Aβ plaque formation, allowing phagocytosis of these structures and, thereby, preventing neuroinflammation and further damage (57). We cannot exclude that the increment in Iba-1–expressing cells during CIA in tau-tg mice is also due to immigrating phagocytic cells from the periphery, because activated macrophages can also express Iba-1 (58). Although neuroinflammation was shown to promote early tau tangle formation in the mouse model that we used in this study (21), we found that the levels of proinflammatory cytokines and chemokines, such as TNF-α, IL-6, and MCP-1, are lower in the CNS of arthritic tau-tg mice compared with arthritic wt mice. In addition, migration of peripheral CCR2⁺ monocytes into the brain was less pronounced in tau-tg mice with induced arthritis, indicating an overall decrease in inflammatory activity in the CNS of these mice.

In summary, we show that the presence of tau-induced neurodegeneration elicits an inflammatory state with increased proinflammatory cytokine expression, augmentation of proinflammatory monocytes, and increased susceptibility to arthritis. In turn, the same inflammatory response, particularly if further induced by arthritis, limits tau-mediated neuropathology, presumably by inducing microglia or macrophage-mediated clearance of tau deposits.

We thank Barbara Happich, Hedwig Symowski, Isabell Schmidt, and Silke Winkler for technical assistance. This work was done in partial fulfillment of the Ph.D. thesis of S.C.L.

The authors have no financial conflicts of interest.