|Home | About | Journals | Submit | Contact Us | Français|
Oligodendrocytes are myelinating cells of the central nervous system. Multiple sclerosis (MS) is a demyelinating disease characterized by both myelin loss and neuronal degeneration. However, the molecular mechanisms underlying neuronal degeneration in demyelinating disorders are not fully understood. In the experimental autoimmune encephalomyelitis (EAE) demyelinating-mouse model of MS, inflammatory microglia produce cytokines, including interleukin-1β (IL-1β). Since microglia and noncanonical Wnt signaling components in neurons, such as the coreceptor Ror2, were observed in the spinal cords of mice with EAE (EAE mice), we postulated that the interplay between activated microglia and spinal neurons under EAE conditions is mediated through noncanonical Wnt signaling. EAE treatment upregulated in vivo expression of noncanonical Wnt signaling components in spinal neurons through microglial activation. In accordance with the neuronal degeneration detected in the EAE spinal cord in vivo, coculture of spinal neurons with microglia or the application of recombinant IL-1β upregulated noncanonical Wnt signaling and induced neuron death, which was suppressed by the inhibition of the Wnt-Ror2 pathway. Ectopic noncanonical Wnt signaling aggravated the demyelinating pathology in another MS mouse model due to Wnt5a-induced neurodegeneration. The linkage between activated microglia and neuronal Wnt-Ror2 signaling may provide a candidate target for therapeutic approaches to demyelinating disorders.
Oligodendrocytes are glial cells that myelinate neuronal axons in the central nervous system (CNS) (1). Myelin sheaths are disrupted during the early phase of oligodendrocyte disorders, such as multiple sclerosis (MS), which can be recovered by maturation and remyelination by preexisting oligodendrocyte precursor cells. However, during the late phase, demyelination is accompanied by neuronal degeneration, resulting in permanent and irreversible damage to the lesion. Thus, effective therapeutic approaches for demyelinating diseases are highly desired. The experimental autoimmune encephalomyelitis (EAE) mouse model, induced by injecting myelin oligodendrocyte glycoprotein (MOG) peptide, has been widely used as an MS-related model exhibiting CNS demyelination, inflammation, and motor abnormalities (2). The pathology of EAE is due to an immune reaction against target antigens. The induced inflammation causes neurological and pathological symptoms comparable to those in MS patients (3). Microglia are considered to be deeply involved in this process (4). Microglia exhibit either a typical activated M1 phenotype or an alternatively activated M2 phenotype (5). M1 microglia are thought to be proinflammatory and secrete various inflammatory cytokines, such as tumor necrosis factor alpha (TNF-α), interleukin-1β (IL-1β), and IL-18. M2 microglia are considered to be reparative cells that are involved in neuroprotection and recovery of damaged lesions through arginase activity and the production of neurotrophic factors (6). Thus, microglia can perform biphasic functions. Axonal degeneration occurs in both the acute and chronic phases of MS animal models. Thus, drug development for demyelinating disorders could also be applicable to protection from neurodegeneration. However, the molecular mechanisms underlying neuronal pathological changes in demyelinating disorders are largely unknown.
A growing number of reports indicate that canonical Wnt/β-catenin signaling plays a role in oligodendrocyte development and remyelination (7). For example, constitutive activation of β-catenin in mice using Olig2-Cre-mediated oligodendrocyte-specific recombination (which show normal adult myelination) delays remyelination after lysolecithin-induced demyelination in a cell-autonomous manner (8), suggesting the involvement of Wnt/β-catenin signaling in oligodendrocyte regeneration and remyelination.
Wnt signaling is classified as either canonical Wnt/β-catenin or noncanonical signaling; the latter activates various alternative signaling routes independent of β-catenin (9). Wnt4, Wnt5a, and Wnt11 are typical noncanonical Wnt ligands that have been reported to play a role in various pathologies, including cancer and inflammatory diseases (10, 11). Noncanonical Wnt proteins activate alternative signaling pathways, including Wnt/Ca2+ and Wnt/c-Jun N-terminal kinase (JNK) pathways (12). The coreceptor identified for noncanonical Wnt ligands is a receptor tyrosine kinase-like orphan receptor 2 (Ror2), which is essential for noncanonical Wnt ligands to properly perform their functions (13,–16). In support of this interaction, the embryonic phenotype of Ror2 homozygous knockout mice is closely related to the Wnt5a knockout phenotype (17, 18). However, the involvement of noncanonical Wnt signaling in the pathogenesis of demyelinating disorders remains unclear.
In this study, we found that inflammatory microglia and noncanonical Wnt signaling components in neurons were present in the gray matter of EAE demyelinating mice. Expression of noncanonical Wnt signaling components was increased in the EAE spinal cord through microglial activation. Coculture of spinal neurons with microglia, as well as the application of recombinant IL-1β, upregulated Ror2 expression and induced neuron death via the Wnt-Ror2 axis. Moreover, we show that Wnt5a overexpression in a plp4e/− mouse model of chronic demyelination (19) influences the demyelinating pathology through neurodegeneration mediated by the noncanonical Wnt pathway. These results suggest that microglia-induced noncanonical Wnt signaling in neurons plays a role in demyelinating pathogenesis.
The following antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA): anti-CD16/CD32 (sc-18867), anti-arginase 1 (sc-18351), and anti-phospho-JNK (sc-6254). The following antibodies were purchased from the specified vendors: anti-cleaved caspase 3 and anti-JNK (Cell Signaling Technology; 9661 and 9252, respectively), anti-Ror2 (GeneTex; GTX62277), anti-Iba1 (Wako, 019-19741), anti-neurofilament-M (hybridoma; 1C8), anti-Tau (bio-yeda; 1074-1), anti-green fluorescent protein (anti-GFP) (Nacalai Tesque; 04404-84), anti-choline acetyltransferase (anti-ChAT) (Chemicon; AB143), and anticalbindin (Swant; CB38). Recombinant rat IL-1β (PeproTech; 400-01B), lipopolysaccharide (LPS) (Sigma; L2654), uridine (Sigma; U3003), 5-fluoro-2′-deoxyuridine (FDU) (Sigma; F0503), mouse laminin (Asahi Glass; L0001), JNK inhibitor II (Calbiochem; 420119), and recombinant Frizzled-7/Fc chimera (R&D Systems; 6178-FZ) were purchased from the above-mentioned companies.
CaMK2-tTA mice (20) were used to express tTA in central nervous system neurons. CaMK2-tTA mice were mated with TetO-Wnt5a mice (21) to generate CaMK2-tTA; TetO-Wnt5a mice. tTA expression driven by the CaMK2 neuronal promoter leads to ectopic Wnt5a expression in neurons through tTA-mediated interaction with the TetO sequence. Genotyping was performed as reported previously (21). In the presence of doxycycline (DOX), tTA cannot bind to the TetO sequence, preventing additional transactivation of Wnt5a. In the absence of DOX, tTA binds to TetO and initiates transcription, resulting in Wnt5a overexpression in neurons. The mice were maintained in the absence of DOX to initiate Wnt5a transcription in CaMK2-positive neurons. Heterozygous (plp4e/–) transgenic mice overexpressing the proteolipid protein gene (plp) have been described previously (19). All experiments were performed with the permission of the Animal Research Committee of the National Institute for Physiological Sciences, Aichi, Japan.
Rat spinal neurons were cultured according to previously published methods with minor modifications (22). Briefly, embryonic day 15 (E15) rat spinal cords (n = 8) were carefully removed and cut into small pieces (approximately 1 mm3) using a scalpel. These pieces were trypsinized (0.05%; Gibco) for 40 min at 37°C, followed by centrifugation in a 15-ml centrifuge tube for 3 min at 1,000 rpm. The cells were resuspended in 2.5 ml Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS) and 5% horse serum and triturated with a pipette tip. The cells were seeded in a 90-mm noncoated dish and incubated in a 5% CO2 incubator for 30 min at 37°C to allow the tissue debris and nonneuronal cells to adhere to the bottom of the dish. Next, the cells were plated in 12-well plates precoated with poly-l-lysine–laminin at a concentration of 2.5 × 105 cells/well and incubated in a 5% CO2 incubator at 37°C. The next day, the medium was replaced with serum-free neurobasal medium (Gibco) with B27 supplement (Gibco) and 1× GlutaMax (Gibco). After 3 days in vitro (d.i.v.), 8.1 μM FDU (Sigma) and 4.1 μM uridine (Sigma) were added to the medium to inhibit nonneuronal-cell division. Half of the medium was removed and replaced with fresh supplemented neurobasal medium twice a week. The spinal cord neurons were maintained in culture for 7 days. The cells were fixed 8 h after IL-1β application and immunostained with anti-cleaved-caspase 3 antibody.
Rat neonatal cortices were dissociated, trypsinized, and cultured in poly-d-lysine (PDL)-coated flasks in DMEM with 10% FCS. By 10 days in vitro, the cultures consisted of microglia and oligodendrocyte precursor cells growing on an astrocyte monolayer. Microglia were purified from the mixed glial cultures by mechanical shaking at 180 rpm at 37°C for 30 min. Microglia were removed by differential adhesion from oligodendrocyte precursor cells on the surfaces of astrocytes. The purified microglia were seeded on a 90-mm noncoated dish and maintained in a 5% CO2 incubator. The purified microglia were stimulated with 0.5 μg/ml LPS for 3 h before passage and then seeded at a concentration of 1 × 104 cells/well in 12-well plates containing cultured spinal neurons. The cocultured cells were fixed 8 h after the application of microglia and processed for immunostaining with anti-cleaved-caspase 3 antibody.
MOG-EAE was induced in 8-week-old C57BL/6 wild-type mice by subcutaneous administration of 150 μg MOG peptide (amino acids 35 to 55; Peptide Institute Inc., Osaka, Japan) mixed with complete Freund's adjuvant (CFA) containing 4 mg/ml heat-killed Mycobacterium tuberculosis (Difco catalog number DF231141) on the day of immunization. Intraperitoneal administration of 400 ng of pertussis toxin was performed the day of immunization and 2 days postimmunization (dpi). The control mice for EAE induction were injected with CFA and pertussis toxin without the MOG peptide. Minocycline (Nichi-iko Pharmaceutical Co. Ltd., Toyama, Japan) was intraperitoneally injected into control mice and mice with EAE (EAE mice) at 33 mg/kg of body weight five times per week (23) from 6 dpi onward.
Mice were anesthetized and perfused transcardially with 4% paraformaldehyde (PFA), and cryosections were prepared with a cryostat. The samples were treated with 40 μg/ml proteinase K for 30 min and postfixed in 4% PFA for 15 min at room temperature. After the samples were washed with phosphate-buffered saline (PBS), they were treated with 0.1 M triethanolamine-HCl (pH 8.0), followed by the addition of acetic anhydride. RNA probe hybridization and posthybridization washing were performed as described previously (24). The samples were preincubated in blocking solution containing 10% heat-inactivated sheep serum in PBS–0.1% Triton X-100 (PBST) for 1 h and incubated with 1/2,000-diluted alkaline phosphatase (AP)-conjugated antidigoxigenin antibodies (Roche Diagnostics Corp.) in the blocking solution at 4°C overnight. After undergoing 3 washes with MABT buffer (0.1 M maleic acid, 0.15 M NaCl, and 0.1% Tween 20, pH 7.5), the samples were treated twice with NTMT buffer (0.1 M NaCl, 0.1 M Tris-HCl, 0.05 M MgCl2, and 0.1% Tween 20, pH 9.5). Nitroblue tetrazolium (NBT) and BCIP (5-bromo-4-chloro-3-indolylphosphate) (Roche) were used as the substrates for AP. The probe sequence for axin2 was digested from the FANTOM clone (RIKEN, Yokohama, Japan) of the cDNA library. The probes for Ror2 and Wnt11 were synthesized using PCR with the following primer sets (For, forward primer; Rev, reverse primer): Ror2, For, GTATGGAAAGTTCTCCATCG, and Rev, TCTGATCTGACCCTTCATGG; Wnt11, For, GGAAACGAAGTGTAAATGCC, and Rev, TGGTACTTGCAGTGACATCG.
The cultured cells were fixed with 4% PFA and used for immunostaining. Fixed cells or cryosections were blocked with 5% normal goat serum in PBST and then incubated with primary antibodies overnight at 4°C. After being rinsed with PBST, the sections were incubated with secondary antibodies. The secondary antibodies used were Alexa Fluor 488- or 564-conjugated goat anti-mouse, anti-rabbit, or anti-rat IgG (Molecular Probes). The Vectastain Elite ABC kit (Vector) was used for immunostaining with the horseradish peroxidase (HRP) substrate diaminobenzidine (DAB). Fluorescent, DAB, or AP signals in the tissue sections were visualized with an Olympus BX51 microscope. Fluorescent signals in the cultured cells were visualized using an IX71 fluorescence microscope (Olympus, Tokyo, Japan). For Nissl staining or anti-Ror2 antibody staining, specimens were embedded in paraffin and sectioned at 10 μm.
Total RNA was isolated using Sepasol (Nakarai, Kyoto, Japan), and first-strand cDNA was synthesized using ReverTra Ace (Toyobo, Osaka, Japan). Reverse transcription (RT)-PCR analysis was performed as described previously (25). Briefly, PCR products were visualized on 2% agarose gels containing ethidium bromide. The relative intensities of the target gene bands were quantified with ImageJ and normalized to that of β-actin. Real-time quantitative PCR (qPCR) was performed with 10 μl SYBR master mix reagent (TaKaRa, Otsu, Japan), 0.25 μM specific primer sets (the primer sequences for Ror2, Wnt5a, Wnt11, axin2, and β-actin are shown below), and 1 μl of the prepared cDNA. Real-time quantitative PCR analysis was performed using a StepOne analyzer (Life Technologies). The temperature profile consisted of 40 cycles of denaturation at 95°C for 15 s and annealing and elongation at 60°C for 1 min. To distinguish specific amplification from nonspecific amplification, the melting curve was analyzed after each PCR. To determine the amount of cDNA applied for the assay, purified PCR products were serially diluted and used as standards. The primer sets used in the study were as follows: Ror2, For, GTATGGAAAGTTCTCCATCG, and Rev, TCTGATCTGACCCTTCATGG; Wnt5a, For, CTTCCGCAAGGTGGGCGATGC, and Rev, TTGCACAGGCGTCCCTGCGTG; Wnt11, For, GTGGCTGCTGACCTCAAGACC, and Rev, TTCTTCATGCAGAAGTCAGGAG; axin2, For, AGTAGCGCCGTGTTAGTG, and Rev, ATGCCATCTCGTATGTAGGT; β-actin, For, TGTTACCAACTGGGACGACA, and Rev, GGGGTGTTGAAGGTCTCAAA.
Cells were lysed with SDS sample buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 6% 2-mercaptoethanol, and 10% glycerol) and subjected to SDS-PAGE, followed by immunoblotting using primary anti-phospho-JNK antibody, which was detected with HRP-conjugated goat anti-mouse IgG antibodies (MP Biomedicals, Solon, OH) using a chemiluminescence detection system (GE Healthcare Life Sciences, Uppsala, Sweden).
The plasmid vector piLenti-siRNA-GFP carrying small interfering RNA (siRNA) for rat Ror2 or scrambled siRNA was purchased from Applied Biological Materials (British Columbia, Canada). Vesicular stomatitis virus G protein (VSV-G) and pseudotyped human immunodeficiency virus (HIV) vectors were used for virus preparation. Viral-vector production has been described previously (26, 27). Briefly, HEK293FT cells (ATCC, Manassas, VA) were cotransfected with a mixture of 4 plasmids using Lipofectamine 2000 transfection reagent. The 4-plasmid mixture was composed of pCAGGS-KGRIR, pCAGGS-RTR2, pCAGGS-VSVG, and piLenti-siRNA-GFP. The cells were cultured in DMEM supplemented with 10% fetal bovine serum at 37°C in a 5% CO2 atmosphere and plated on a 90-mm dish at a concentration of 5 × 106/10 ml medium 1 day before transfection. Five hours after transfection, the cells were washed and cultured in the above-mentioned maintenance medium. The medium containing virus particles was harvested 48 and 72 h after transfection. The medium was centrifuged at 3,000 rpm for 5 min, filtered through 0.45-μm-pore-size membranes, and ultracentrifuged at 50,000 × g for 2 h at 4°C. The viral pellet was suspended in 2 ml of Hanks balanced salt solution (HBSS; no calcium, no magnesium) medium and concentrated again by ultracentrifugation at 50,000 × g for 2 h at 4°C. The pellet was suspended in 100 μl neurobasal medium at 4°C. The virus aliquots were stored at −80°C.
To confirm the knockdown efficiency of siRNA for Ror2 in cultured spinal neurons, the cells were transduced with lentivirus harboring siRNA against Ror2 (piLenti-puro-siRor2-GFP). The next day, puromycin (final concentration, 1 μg/ml) was applied to the culture medium to remove untransduced cells, and the spinal neurons were maintained for an additional 2 days. Total RNA was isolated using Sepasol (Nakarai, Kyoto, Japan).
Results are expressed as means and standard errors of the mean (SEM). For comparison of two groups, Student's t test was used. P values of <0.05 were considered significant. Data for multiple comparisons were analyzed by one-way analysis of variance (ANOVA), followed by a Newman-Keuls post hoc test using the statistical program InStat (GraphPad Software, San Diego, CA). The level of significance was a P value of <0.05.
To investigate molecular events involved in demyelinating disorders, we utilized the MOG-EAE mouse model. By 14 dpi, MOG-EAE mice began to show deficits in locomotor activities. It has been reported that experimental activation of Wnt/β-catenin signaling delays remyelination (8). However, in contrast to canonical Wnt/β-catenin signaling, the roles of noncanonical Wnt signaling in demyelinating disorders remain unclear.
To examine whether noncanonical Wnt signaling components are present in the MOG-EAE demyelinating-mouse model in vivo, we performed RNA in situ hybridization. The expression of typical noncanonical Wnt ligand, Wnt11, and its coreceptor, Ror2, was increased in the gray matter of the EAE mouse spinal cord (Fig. 1A to toD).D). We also detected slight changes in the expression of axin2 (Fig. 1E and andF),F), encoded by a target gene of the canonical Wnt/β-catenin pathway (28, 29). To confirm the upregulation of Wnt signaling components, we performed real-time qPCR analysis on the EAE spinal cord (Fig. 1G). Wnt11 and Ror2 mRNA levels were significantly upregulated in the spinal cord under our EAE conditions, whereas the increases in Wnt5a and axin2 mRNA levels were not significant (Fig. 1G). These results indicated that noncanonical Wnt signaling in the spinal cords of EAE mice increased in vivo. To examine the cell types that express Ror2 and Wnt11 under our EAE conditions, we performed immunostaining with cell lineage-specific markers following Ror2 or Wnt11 in situ hybridization. The Ror2 signal was detected in Tau-positive spinal neurons of EAE mice (Fig. 1J). Ror2 signal was also present in a subset of motor neurons, which were detected with an anti-ChAT antibody (Fig. 1H). Similar to the Ror2 expression, Wnt11 expression was also detected in neurons, including ChAT-positive motor neurons in the EAE spinal cord (Fig. 1I). This suggests that spinal neurons are Wnt-producing cells in the MOG-EAE mice.
By 14 dpi in MOG-EAE mice, the number of Iba1-positive microglia had increased (Fig. 2A). Activated microglia express more M1 (classical proinflammatory) markers under pathological conditions, whereas they transfer to the M2 (alternative anti-inflammatory) phenotype in a resting state (5, 6). To examine the microglial phenotype induced by our MOG-EAE procedure, spinal cord sections were immunostained for the classic proinflammatory microglial marker (anti-CD16/CD32) or alternative anti-inflammatory microglia (anti-arginase 1) (Fig. 2B and andD).D). In the white matter of the EAE spinal cord, cells positive for either CD16/CD32 or arginase 1 were both observed at higher numbers, whereas CD16/CD32-positive microglia were predominant in the ventral gray matter at this stage (Fig. 2C and andE).E). These results suggest that proinflammatory microglia mainly influence gray matter in the EAE spinal cord. When the section was costained with Iba1 and Ror2, both Iba1-positive cells and Ror2-positive cells were observed in the gray matter of the EAE spinal cord (Fig. 2F).
Since Iba1-positive cells and the expression of Ror2 were observed in the gray matter of the EAE spinal cord (Fig. 2F), we speculated that there was interplay between activated microglia and the noncanonical Wnt signaling pathway under EAE conditions. Minocycline has been used as a potent inhibitor of microglial activation (30). We therefore examined whether minocycline administration to demyelinating MOG-EAE mice suppressed Ror2 induction in the spinal neurons. Minocycline was intraperitoneally injected into EAE mice 5 times per week from 6 dpi onward. Intraperitoneal minocycline administration remarkably reduced the numbers of both CD16/CD32-positive microglia and arginase 1-positive microglia (Fig. 3A to toH).H). The upregulation of Ror2 in the spinal neurons that was observed in the MOG-EAE mice injected with saline was abolished by minocycline administration (Fig. 3I to toKK and andO).O). These results suggest that in vivo Ror2 upregulation is dependent on an increase in microglia positive for either CD16/CD32 or arginase 1 under MOG-EAE conditions. Axin2 expression remained unchanged by minocycline administration (Fig. 3L to toNN and andP).P). These results suggest that canonical Wnt/β-catenin signaling is independent of microglia under EAE conditions.
Although minocycline administration reduced the number of cells positive for activated microglial markers (Fig. 3), minocycline could have broad anti-inflammatory effects. To further demonstrate that microglia induce Ror2 expression in spinal neurons, we used cultured spinal neurons, which are useful for investigating cellular and molecular mechanisms. Microglia convert to activated states in response to pathological conditions, leading to transcriptional or functional remodeling and the acquisition of an immunocompetent phenotype (31). Activated microglia respond to EAE-induced inflammation and produce cytokines, such as IL-1β (3, 6). The cultured spinal neurons were thus treated with recombinant IL-1β for 3 h, which increased the expression of Ror2 mRNA (Fig. 4A). In addition, immunostaining with anti-Ror2 antibody in the IL-1β-treated cultured spinal neurons revealed that IL-1β also increased the expression of Ror2 protein (Fig. 4B and andC).C). TNF-α, another inflammatory cytokine, also increased Ror2 expression in cultured spinal neurons (see Fig. S1A in the supplemental material), suggesting that various inflammatory cytokines can perform similar functions in Ror2 induction. Here, we used IL-1β as one of the typical inflammatory cytokines. The properties of microglial cells and their activation have been studied in cultures stimulated with the bacterial endotoxin LPS, usually resulting in severe inflammatory responses (32). Ror2 protein expression was also induced when spinal neurons were cocultured with microglia pretreated for 3 h with LPS (Fig. 4D and andE).E). Several reports have demonstrated that IL-1β expression is increased in microglia 2 to 4 h after LPS treatment (33,–36). These results suggest that activated microglia upregulate Ror2 expression in spinal neurons through release of cytokines, including IL-1β.
Some reports have suggested that neuronal degeneration occurs during EAE development (37, 38). We thus analyzed nuclear morphology by Hoechst 33258 staining and fluorescence microscopy. Cells with uniformly stained nuclei were scored as healthy and viable cells, whereas those with nuclei exhibiting condensed and fragmented chromatin were classified as apoptotic cells. Increased numbers of condensed or fragmented cell nuclei could be observed in motor neurons that were positive for ChAT under EAE conditions (Fig. 5A to toC).C). In conjunction with the increased microglia under EAE conditions (Fig. 2), these results encouraged us to examine if the number of activated caspase 3-positive apoptotic neurons was increased by activated microglia under our culture conditions. When spinal neurons were cocultured with microglia, active-caspase 3-positive apoptotic neurons increased (Fig. 5D to toF),F), whereas a negligible number of cocultured microglia were apoptotic themselves (0.9% of the microglia). Furthermore, the application of IL-1β increased the number of active-caspase 3-positive apoptotic spinal neurons (Fig. 5G to toI).I). TNF-α also increased active-caspase 3+ cells in cultured spinal neurons (see Fig. S1B to D in the supplemental material), similar to IL-1β application. These results indicate that microglia or the inflammatory cytokines induce apoptosis in spinal neurons. It has been proposed that Ror2 mediates noncanonical Wnt signaling by activating the Wnt-JNK pathway (16) and that JNK functions in neuron death (39). Indeed, immunostaining with anti-phospho-JNK antibody revealed that application of IL-1β induced phosphorylation of JNK in cultured spinal neurons (Fig. 6A); JNK phosphorylation was confirmed by Western blotting using an anti-phospho-JNK antibody (Fig. 6B). Since Ror2 expression in spinal neurons was increased by coculture with microglia (Fig. 4), the application of IL-1β (Fig. 4), or EAE treatment in vivo (Fig. 1 and and3),3), we next assessed whether endogenous Ror2 contributed to IL-1β-dependent apoptosis. We performed a Ror2 knockdown analysis using RNA interference (siRNA). The expression of endogenous Ror2 in cultured spinal neurons was inhibited by a Ror2 siRNA-harboring lentiviral vector (Fig. 6C), which suppressed IL-1β-induced neuronal apoptosis compared to scrambled siRNA (Fig. 6D to toF),F), suggesting that endogenous Ror2 contributes to IL-1β-dependent neuronal apoptosis. To examine whether extracellular Wnt ligands have an effect on IL-1β-dependent neuron death, we investigated neuronal viability upon the addition of a potent soluble Wnt antagonist, Frizzled-7/Fc (40,–42), to cultured spinal neurons. One microgram per milliliter or 0.25 μg/ml of Frizzled-7/Fc suppressed the increase in active-caspase 3-positive cells induced by IL-1β in a dose-dependent manner (Fig. 6G to toJ).J). These results suggest that endogenous Wnt ligands contribute to IL-1β-dependent neuron death. JNK inhibitor II, which is a well-known compound that inhibits JNK (43), was applied to cultured spinal neurons. IL-1β increased the ratio of active-caspase 3-positive cells to the total number of cells, while JNK inhibitor II suppressed the IL-1β-induced apoptosis (see Fig. S2A to D in the supplemental material). To investigate the function of the JNK pathway in microglia-induced neuron death, we tested a coculture of microglia with or without JNK inhibitor II and measured whether it blocked neuronal apoptosis. Coculture of microglia with spinal neurons increased the ratio of active-caspase 3-positive cells versus the total number of cells, which was suppressed by JNK inhibitor (see Fig. S2E to G in the supplemental material). These results suggest that the JNK pathway contributes to neuron death induced by IL-1β application or coculture with microglia.
In the experiment described above, activated microglia upregulated components of the noncanonical Wnt pathway and induced neuron death via the Wnt-Ror2 axis in vitro. Loss-of-function effects of either Wnt5a or Wnt11 in vivo may be compensated by one another because Wnt5a and Wnt11 share common receptors and are both expressed in the spinal cord (Fig. 1). The physiological analysis of Wnt proteins in vivo has been hampered by functional redundancy (24, 44). Therefore, we utilized an inducible Wnt5a transgenic-mouse model (21), enabling in vivo examination of the relevance and activity of noncanonical Wnt signaling within particular cell lineages. Wnt4, Wnt5a, and Wnt11 are classified as typical noncanonical Wnt ligands (10). Since noncanonical Wnt proteins were increased in the EAE spinal neurons (Fig. 1), transgenic Wnt5a expression in the TetO-Wnt5a mice was driven by mating them with CaMK2-tTA mice (20), in which tTA expression was specifically induced in CNS excitatory neurons and cerebellar Purkinje cells and granule cells (CaMK2-tTA; TetO-Wnt5a mice) (20, 45). Previously, it has been reported that spinal neurons express both CaMK2α and CaMK2β in the ventral gray matter of the spinal cord (46). The level of Wnt5a mRNA was increased in the CaMK2-tTA; TetO-Wnt5a mice (Fig. 7A and andB).B). Heterozygous plp-transgenic (plp4e/–) mice overexpressing the proteolipid protein gene (plp) closely mimic the failure of remyelination observed in chronic demyelinated lesions of MS (47). At 2 months of age, plp4e/− mice have an abnormal paranodal structure (48). Although the number of myelinated axons is consistent with that of wild-type mice at this stage, the myelin in plp4e/− mice is thinner, showing a significant difference in the G ratio. Moreover, the conduction velocity (CV) is significantly decreased in 2-month-old plp4e/− mice (48, 49). Abnormalities in node-paranode complexes induce oxidative stress in oligodendrocytes or neurons (50, 51).
Symptoms of demyelinating disorders are caused by both loss of myelin sheaths and neuronal degeneration (52); thus, their imbalance leads to phenotypes associated with demyelinating disorders. To evaluate the effects of Wnt5a overexpression on neurodegeneration, we analyzed 2.5-month-old plp4e/− mice, as they do not have severe demyelination at this stage. To overexpress Wnt5a in CaMK2-positive neurons in the demyelinating-mouse model, we mated CaMK2-tTA; TetO-Wnt5a mice with plp4e/− mice (CaMK2-tTA; TetO-Wnt5a; plp4e/−). We found that there were fewer motor neurons in the ventral horn in CaMK2-tTA; TetO-Wnt5a; plp4e/− spinal cords by 2.5 months of age (Fig. 7F to toJ).J). Nissl staining of spinal cord sections confirmed this defect (Fig. 7C to toE).E). Immunohistochemistry with an anti-Ror2 antibody in the gray matter of the ventral spinal cord revealed abundant expression of Ror2 in the CaMK2-tTA; TetO-Wnt5a; plp4e/− mice (Fig. 7K and andL).L). This result suggests that these cells are competent to respond to noncanonical Wnt signaling. Similar to the observation made in the EAE mice, IL-1β and TNF-α expression was upregulated in the CaMK2-tTA; TetO-Wnt5a; plp4e/− spinal cord (Fig. 7Q; see Fig. S1E to G in the supplemental material). CD16/CD32-positive cells, which are cytokine-producing cells, were observed in the ventral gray matter of CaMK2-tTA; TetO-Wnt5a; plp4e/− mice, whereas they were absent in wild-type mice at this stage (Fig. 7O and andP).P). In accordance with the JNK phosphorylation detected in vitro in cultured spinal neurons to which IL-1β had been applied (Fig. 6A and andB),B), phospho-JNK immunolabeling was detected in the CaMK2-tTA; TetO-Wnt5a; plp4e/− mouse spinal cord by immunostaining (Fig. 7M and andN),N), suggesting the JNK signaling pathway in the ventral horn of CaMK2-tTA; TetO-Wnt5a; plp4e/− mice might function in the pathway. Furthermore, immunohistochemistry using an anticalbindin antibody showed that Purkinje cells in the cerebellum were also decreased in CaMK2-tTA; TetO-Wnt5a; plp4e/− mice compared to the other genotypes (Fig. 7O to toS),S), probably owing to aggravation of cerebellar neuronal stress associated with early demyelination. These results suggest that upregulation of noncanonical Wnt exacerbates neurodegeneration in the spinal cord, as well as in the cerebellum, in this demyelinating-mouse model.
Previously, Sonomoto and coworkers demonstrated that IL-1β induces differentiation of human mesenchymal stem cells (MSCs) into osteoblasts and showed that Ror2 expression is significantly induced in human MSCs by IL-1β (53). To our knowledge, our finding is the first report that minocycline administration antagonizes the upregulation of Ror2 in the MOG-EAE spinal cord in vivo and that the application of recombinant IL-1β upregulates Ror2 expression in cultured spinal neurons. Since CD16/CD32+ microglia occupied a large proportion of activated microglia in the gray matter of the EAE spinal cord under these conditions (Fig. 2), proinflammatory microglia might provoke Ror2 expression in spinal neurons of the gray matter. Although it remains unclear how inflammatory cytokines elicit Ror2 expression in the spinal neurons, previous reports have demonstrated that IL-1β induces noncanonical Wnt5a expression via NF-κB in chondrocytes and that NF-κB can bind to the human Wnt5a promoter (54, 55). It is possible that inflammatory cytokines recruit transcription factors, such as NF-κB or AP-1, to induce Wnt11 or Ror2 expression in spinal neurons. Our results suggest that the progression of demyelinating disorders involves the interplay of microglia-derived cytokines and the noncanonical Wnt pathway through the upregulation of Ror2 in CNS neurons. Trapp et al. previously observed transected axons in multiple sclerosis lesions (52). In our study, condensed or fragmented cell nuclei were observed in motor neurons of EAE mice (Fig. 5), and Wnt5a overexpression in a chronic demyelinating-mouse model aggravated demyelination-associated neuronal loss (Fig. 7). However, few studies have characterized the molecular mechanisms underlying these pathological changes in neurons in demyelinating disorders. A previous paper reported that endogenous and exogenous galectin-1 (Gal1) plays a pivotal role in deactivating classically activated microglia, and Gal1 prevented microglial activation and promoted neuroprotection (37). Our study proposes another novel neurodegenerative effect of microglia in demyelinating disorders. Activated microglia secrete cytokines, such as IL-1β and TNF-α, and upregulate Ror2 expression in the spinal neurons. Wnt ligands activate downstream signaling through the Ror2 coreceptor and aggravate neuron death. It has been proposed that Ror2 mediates noncanonical Wnt signaling by activating the Wnt/JNK pathway (16). Noncanonical Wnt ligands have been reported to play a role in various pathologies, including cancer and inflammatory diseases (10, 11), and its downstream effector, JNK, functions in neuron death (39). It is thus possible that neuronal apoptosis induced by activated microglia involves Ror2 activation and subsequent upregulation of the JNK signaling pathway. In support of this, we observed JNK phosphorylation in both cultured spinal neurons to which IL-1β had been applied (Fig. 6A and andB)B) and the CaMK2-tTA; TetO-Wnt5a; plp4e/− mouse spinal cord (Fig. 7M and andN).N). We found that the expression of the typical noncanonical Wnt ligand, Wnt11, was increased in the gray matter of the EAE mouse spinal cord (Fig. 1). Double staining for ChAT showed that Wnt11 expression was present in a subset of motor neurons, suggesting that spinal neurons are Wnt-producing cells under our EAE conditions in vivo. Future experiments should be designed to address the precise contribution of noncanonical Wnt signaling to neuronal degeneration in a context-dependent and regionally specific manner, including microglial activation or neuronal degeneration, in view of the Wnt-Ror2 signaling. Our present study provides new insights into the molecular basis for neurodegeneration in demyelinating disorders.
Previous studies have reported that Wnt signaling plays a critical role in an animal model of demyelination and remyelination (8, 56). Fancy et al. focused on canonical Wnt/β-catenin signaling components, such as β-catenin, axin2, and adenomatous polyposis coli (APC), in oligodendrocytes. Constitutive in vivo activation of β-catenin or axin2 deficiency showed delayed oligodendroglial differentiation and remyelination after lysolecithin lesioning in an oligodendroglial-cell-autonomous manner (8, 56). Canonical Wnt signaling activates the GSK3β/β-catenin pathway, while noncanonical Wnt signaling activates alternative signaling pathways, including Wnt/Ca2+ and Wnt/JNK pathways (57,–60). Thus, Wnt ligands fractionate downstream pathways, depending upon the context. In fact, although MOG-EAE application induced a slight upregulation of axin2 expression, reduction in microglial numbers by minocycline administration did not affect axin2 expression, an indicator of canonical Wnt/β-catenin signaling activity. This is in contrast to the reduced expression of Ror2 in spinal neurons caused by minocycline administration. Thus, upregulation of noncanonical Wnt signaling in our MOG-EAE mice might be dependent on the microglial number, whereas the activity of Wnt/β-catenin signaling is independent of the microglia and cytokine axis. In addition, although previous papers reported that the involvement of Wnt/β-catenin signaling in oligodendrocyte regeneration and remyelination occurs in an oligodendroglial-cell-autonomous manner (8, 56), we found that the noncanonical Wnt signaling induced in neurons plays a role in demyelination-associated neurodegeneration. The delicate balance between oligodendroglial Wnt/β-catenin signaling and neuronal noncanonical Wnt signaling has a gross manifestation in demyelinating disorders.
In this study, we used two mouse models for demyelinating disorders. EAE is a model of acute demyelinating diseases. The general form of MS, termed relapsing-remitting MS (RRMS), is related with acute inflammatory events leading to neurological dysfunctions. The EAE mouse model shows functional recovery during relapses. Although condensed or fragmented cell nuclei were observed in EAE-conditioned motor neurons in our study (Fig. 5), ChAT-positive immunolabeling had been retained in the gray matter of MOG-EAE mice under our experimental conditions. Moreover, some previous studies have shown that the effects of gene-targeting or transgenic approaches on neuronal degeneration were less obvious than those pertaining to demyelination in EAE mice. For example, Mills Ko et al. (61) compared the extents of spinal cord demyelination and spinal cord axon loss induced by EAE treatment in astroglial CXCL10 knockout and control mice. Acute spinal cord demyelination was less severe in the absence of astroglial CXCL10, whereas spinal cord axon loss was indistinguishable between the two groups of mice under EAE conditions. For this reason, we used a chronic demyelinating-mouse model (plp4e/−) for the Wnt5a overexpression experiment. The ventral horn of the CaMK2-tTA; TetO-Wnt5a; plp4e/− mouse spinal cord showed ChAT+ motor neuron loss by 2.5 months of age (Fig. 7). CaMK2-tTA; TetO-Wnt5a; plp4e/− mice also exhibited a decrease in cerebellar Purkinje cells (Fig. 7). plp4e/− mice at 2.5 months of age have been reported to show abnormal paranodal disruption leading to decreased neuronal CV (48, 49). plp4e/− mice can produce a compacted myelin sheath, although it is thinner than that of wild-type mice (49). Abnormal node-paranode complexes cause oxidative stress in oligodendrocytes and neurons (50, 51, 62,–64). Interestingly, abnormal paranodal structure alters gene expression in CNS neurons (unpublished data). Thus, these results suggest that the 2.5-month-old plp4e/− demyelinating-mouse model exhibits neuronal stress leading to altered gene expression in CNS neurons, which may be modulated by the Wnt-Ror2 pathway.
Our present study provides new insights into the molecular basis of demyelination-related neurodegeneration, in view of the Wnt-Ror2 signaling induced by microglia. Microglia are also activated in various inflammatory conditions, such as ischemia or traumatic injury. Activated microglia might also induce noncanonical Wnt signaling, but not Wnt/β-catenin signaling, in the surrounding neurons in those inflammatory environments. Our discovery may therefore open a new path toward the development of therapeutic approaches in diverse inflammatory conditions. This study sheds light on noncanonical Wnt components and/or antagonists that might be applicable to drug development for neuroprotection in demyelinating disorders. Stabilization of axin2 by administration of XAV939 has been reported to improve spinal cord remyelination (56). The combinatorial modulation of both Wnt/β-catenin signaling in oligodendrocytes and the noncanonical Wnt pathway in neurons might be highly efficient in drug therapy for demyelinating disorders.
We thank Tetsushi Kagawa (Department of Stem Cell Regulation, Medical Research Institute, Tokyo Medical and Dental University [TMDU], Tokyo, Japan) for the gift of the PCR primers. We also thank Shinji Takada (Okazaki Institute for Integrative Bioscience and National Institute for Basic Biology, National Institutes of Natural Sciences, Okazaki, Japan) and Seiji Hitoshi (Department of Integrative Physiology, Shiga University of Medical Science, Otsu, Japan) for helpful discussions. We thank Rie Taguchi for her technical assistance.
This work was supported by a Grant-in-Aid for Scientific Research (Kakenhi Projects 25117001 “glial assembly”).
We declare that we have no conflict of interest.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/MCB.00139-16.