|Home | About | Journals | Submit | Contact Us | Français|
Neuroinflammation and increased production of tumor necrosis factor (TNF) in the central nervous system have been implicated in many neurological diseases including white matter disorders periventricular leukomalacia and multiple sclerosis. However, the exact role of TNF in these diseases and how it mediates oligodendrocyte injury remain unclear. Previously we demonstrated that lipopolysaccharide (LPS) selectively kills oligodendrocyte precursors (preOLs) in a non-cell autonomous fashion through the induction of TNF in mixed glial cultures. Here we report that activation of oligodendroglial, but not astroglial and microglial, TNFR1 is required for LPS toxicity, and that astrocytes promote TNF-mediated preOL death through a cell contact-dependent mechanism. Microglia were the sole source for TNF production in LPS-treated mixed glial cultures. Ablation of TNFR1 in mixed glia completely prevented LPS-induced death of preOLs. TNFR1-expressing preOLs were similarly susceptible to LPS treatment when seeded into wildtype and TNFR1−/− mixed glial cultures, demonstrating a requirement for oligodendroglial TNFR1 in the cell death. Although exogenous TNF failed to cause significant cell death in enriched preOL cultures, it became cytotoxic when preOLs were in contact with astrocytes. Collectively, our results demonstrate oligodendroglial TNFR1 in mediating inflammatory destruction of preOLs and suggest a previously unrecognized role for astrocytes in promoting TNF toxicity to preOLs.
Inflammatory responses associated with activation of microglia within the central nervous system (CNS) have been implicated in many pathological conditions including white matter disorder periventricular leukomalacia (PVL). PVL is the major form of cerebral white matter injury and the most important determinant of the neurologic morbidity in premature infants (Volpe 2003). Pathologically, it is characterized by focal necrotic lesions deep in the cerebral white matter that correlates with the development of cerebral palsy, and by diffuse white matter lesions leading to subsequent myelination abnormalities that likely contribute to cognitive and behavioral deficits frequently observed in survivors of prematurity (Volpe 2003). Diffuse PVL lesions are now recognized as the predominant form of brain injury of prematurity with hallmarks of prominently activated astrocytes and microglia and dying premyelinating oligodendrocytes (preOLs) (Haynes et al. 2003). Although the role of these reactive astrocytes and microglia remains unclear, they are likely involved in preOL injury by releasing highly reactive oxygen/nitrogen species and proinflammatory cytokines, such as tumor necrosis factor (TNF) (Haynes et al. 2005, Back 2006, Rezaie & Dean 2002). In fact, preOLs, the primary OL lineage cells in human cerebral white matter during the peak risk period for PVL (Back et al. 2001), are intrinsically highly sensitive to oxidative/nitrative (Back et al. 1998, Back et al. 2002) and excitotoxic insults (McDonald et al. 1998, Follett et al. 2000).
Hypoxia/ischemia and maternal/fetal infection are the two primary triggers for PVL (Volpe 2001). Accumulating experimental and clinical evidence suggests a strong link between the gram negative bacterial component lipopolysaccharide (LPS) and PVL. Many animal studies employing systemic, local, or intrauterine LPS administration and hypoxia/ischemia have demonstrated selective injury to developing cerebral white matter, leading to hypomyelination (reviewed by (Hagberg et al. 2002)). Proinflammatory cytokine TNF is a central player in cerebral ischemia and neuroinflammation, and may contribute to the pathogenesis of PVL. TNF is known to exert most of its biological functions by signaling through cognate receptors TNF receptor 1 (TNFR1, p55) and/or receptor 2 (TNFR2, p75). Increased TNF production and TNFR1/2 expression have been reported for human PVL (Deguchi et al. 1996, Kadhim et al. 2006). Furthermore, transgenic animal studies have demonstrated that overexpression of TNF in astrocytes results in OL apoptosis and demyelination (Akassoglou et al. 1998), implying a pathogenic effect of local TNF signaling in the CNS.
Using an in vitro model for inflammatory injury to preOLs, we demonstrated that LPS induces selective preOL death indirectly by activating microglia (Li et al. 2008, Li et al. 2005). We further identified that peroxynitrite underlies the potent direct killing capability of activated microglia to preOLs and that astrocytes can shift the killing mechanism to one dependent on TNF signaling (Li et al. 2008). However, neither the cellular source for TNF production nor the TNF receptor mediating preOL death was identified. Since primary glial cells all express TNF receptors (Dopp et al. 1997), they can all engage in TNF signaling. Here, we systematically dissected the cellular source for TNF production, determined the TNF receptor required for LPS toxicity, and further identified an essential role for oligodendroglial TNFR1 in TNF-mediated killing of preOLs in mixed glial cultures. Most importantly, we provided the first evidence demonstrating that astrocytes sensitize preOLs to TNF toxicity in a contact-dependent manner.
Wildtype B6.129SF2/J and C57BL/6J mice, transgenic eGFP mice (003291, background C57BL/6J), and TNF (background B6.129SF2/J) and TNFR1 (background C57BL/6J) knockout mice were obtained from the Jackson Laboratory (Bar Harbor, ME). Recombinant TNF was obtained from R&D Systems (Minneapolis, MN). PDGF and basic FGF were from PeproTech (Rocky Hill, NJ). Recombinant GFP adenovirus was from Gene Transfer Vector Core, University of Iowa. Rabbit polyclonal antibodies against Iba1 were purchased from Wako Chemicals (Richmond, VA). Rat antimouse TNF (clone MP6-XT22) was obtained from eBioscience (San Diego, CA). Olig2 antibody was a generous gift from Dr. Richard Lu (University of Texas Southwestern Medical Center). Lactate dehydrogenase (LDH) cytotoxicity kit was from Roche Applied Science (Indianapolis, IN). Unless specified otherwise, all other reagents were from Sigma (St. Louis, MO).
Primary preOLs, microglia, astrocytes and mixed glial cultures were prepared from the forebrains of 1 to 2-d-old Sprage-Dowley rats or mice using a differential detachment method as detailed previously (Li et al. 2008). Briefly, mixed glial cultures were grown in poly-D-lysine coated culture plates or in flasks for individual cell type isolation. Microglia were isolated by shaking mixed glia-containing flasks for 1 h at 200 rpm. The purity of microglia at this stage was consistently >95%. To obtain highly purified microglial monocultures (>99%), cells were further plated into regular petri dishes for 1 h and then detached from the plates with ice cold EBSS. PreOLs isolated from mixed glial cultures were maintained in growth medium, i.e., the serum-free Basal Defined Medium (BDM) (DMEM, 0.1% bovine serum albumin, 50 µg/ml human apo-transferrin, 50 µg/ml insulin, 30 nM sodium selenite, 10 nM D-biotin and 10 nM hydrocortisone) supplemented with PDGF (10 ng/ml) and bFGF (10 ng/ml) for 5–9 days. The OL mono-cultures consisted of OL precursors [O4+, O1−] and were of at least 95% pure. Mature OLs were obtained by culturing preOLs in differentiation medium (BDM supplemented with 10 ng/ml CNTF and 15 nM T3) for 7–10 days.
Astrocytes were purified from the astrocyte layer in the flask that has been exposed to the specific microglia toxin L-leucine methyl ester (1 mM) for 1 h, followed by 1~2 cycles of subculture and repeated exposure to L-leucine methyl ester. The enriched astrocytes were consistently more than 95% pure with preOLs being the major contaminating cells. For astrocytes and preOLs co-cultures, astrocytes (2.4 × 105 cells per well) were plated into 24-well culture plates 1~2 days before seeding preOLs (6 × 104 cells per well), and the co-cultures were maintained in growth medium for 1–2 days before use. For bridged no cell contact co-cultures, preOLs grown on PO-coated coverslips with mini PDMS columns attached at four corners were transferred to 24-wells containing astrocytes. The distance between the astrocyte layer and the preOL layer was 1mm. Unless stated otherwise, all cell treatments were carried out in the growth medium. Mouse mixed glial cultures and various glia monocultures were prepared with the same methods as described above from wildtype (WT) and TNF and TNFR1 knockout (KO) mice. Mosaic mouse mixed cultures using cells from WT and KO mice were prepared by seeding isolated astrocytes (4 × 104/well) in poly-D-lysine coated 8 well chamber slide (Nunc, Rochester, NY), followed by microglia seeding (2 × 104/well) on the next day, and preOLs plating (0.8 × 104/well) plating the following day. Five to seven days after preOLs plating, these mixed glial cultures were subjected to vehicle or LPS treatment for 48 h and preOL survival was determined by counting the number of O4+ cells.
Cell death was induced by exposure to LPS (Escherichia coli O111:B4, Sigma) or TNF as specified in the figure legend. Survival of preOLs was determined by counting O4 positive cells with normal nuclei. Briefly, cells were treated in triplicate as specified for 24–48 h. After washing with PBS and fixation with 4% paraformaldehyde, cells were immunostained with O4 antibody (1:500). Total number of cells was revealed by staining all nuclei with Hoechst 33258. Five random, consecutive fields were counted in each coverslip under 200× magnification with a total of >1000 cells counted in the control conditions. Cell survival is expressed as mean ± SD. When monocultures of preOLs were used, cell viability was evaluated by Alamar blue, a tetrazolium dye that is reduced by living cells to a colored product, as described previously (Li et al. 2003). Cell toxicity was also assessed by lactate dehydrogenase (LDH) release assay according to the manufacturer’s protocol (Roche Diagnostics, Indianapolis, IN).
Purified rat preOLs were infected with adenovirus containing green fluorescent protein (GFP) cDNA (AdGFP) as described previously (Baud et al. 2004). Briefly, cells were exposed to 1 × 108 pfu/ml of AdGFP overnight in BDM followed by a complete medium change the next day. Cells were allowed to recover for 48 h. The infection rate was consistently more than 90% with minimum toxicity. Cells were then trypsinized off the culture dish and seeded at density of 1–2 × 104 per well into established mixed glial cultures. The next day, the cultures were treated with either vehicle or LPS for 24 h, and immunostained for Iba1 and GFP.
The levels of TNF in the culture media of cells treated as specified were measured using a commercially available ELISA kit according to the manufacturer’s instruction (eBioScience, San Diego, CA). Absorption at 450 nm was determined in a microplate reader (Fluostar Optima, BMG Labtech). The detection limit of the ELISA was 8 pg/ml.
Microglia and astrocyte monocultures were plated onto poly-D-lysine coated 6-well plates at a density of 1.0 × 106 cells per well in DMEM containing 10% fetal bovine serum. After 24 h, cells were rinsed twice with BDM media, and stimulated with LPS (0–1.0 µg/ml) or vehicle for 0–24 h. RNA was extracted using an RNeasy kit according to the manufacturers’ instructions (Qiagen, Valencia, CA). Residual DNA was removed by incubating RNA samples with DNase I for 15 min at room temperature followed by DNase inactivation at 65°C for 10 min according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). The samples were reverse transcribed to cDNA using the reverse transcription system kit (Promega, Madison, WI) and random primers in the presence or absence of reverse transcriptase for 10 min at 25°C followed by 15 min at 45°C, 5 min at 95°C, and 5 min at 4°C. RT-PCR was used to test for the presence of TNF, IL-1β, IL-10, TLR2, TLR3, TLR4, GFAP, Iba1 and β-actin cDNA using the following specific primers: TNF, forward-GCCCACGTCGTAGCAAAC, reverse-GCAGCCTTGTCCCTTGAA; IL-1β, forward-TGACCCATGTGAGCTGAAAG, reverse-AGGGATTTTGTCGTTGCTTG; IL-10, forward-GAGTGAAGACCAGCAAAGGC, reverse-TTGTCCAGCTGGTCCTTCTT; TLR2, forward-AGCTGGAGAACTCTGACCCA, reverse-CAAAGAGCCTGAAGTGGGAG; TLR3, forward-TGCGATTGGCAAGTTATTCG, reverse-GCGCAGGCTGTTGTAGGAAA; TLR4, forward-TGCTCAGACATGGCAGTTTC, reverse-TCAAGGCTTTTCCATCCAAC; GFAP, forward-CAGCTT CGAGCCAAGGAG, reverse-TGTCCCTCTCCACCTCCA; Iba1, forward-CTTTTGGACTGCTGAAAGCC, reverse-GTT TCTCCAGCATTCGCTTC; and β-actin, forward- AGACTTCGAGCAGGAGATGG, reverse- CCATCATGAAGTGTGACGTTG. Briefly, after an initial denature step at 95°C for 10 min, 100 ng cDNA per reaction was subjected to 30 cycles of PCR (95°C 15 sec, 56°C 15 sec, 72°C 30 sec) followed by a final elongation step at 72°C for 5 min. Products were electrophorezed on 2% agarose gels and visualized under UV light using a Bio-Rad Chemidoc XRS gel documentation system and Quantity-one software.
Cells were fixed with 4% paraformaldehyde in PBS for 10 min and blocked with TBST (50 mM Tris·HCl, pH 7.4/150 mM NaCl/0.1% Triton X-100) or TBS (for O4 immunostaining) containing 5% goat serum. The coverslips were then incubated with antibody O4 (1:500) or antibodies against Iba1 (1:1000) or GFAP (1:1000) overnight at 4°C. After washes, secondary antibody conjugated with either Alexa Fluor 488 or Alexa Fluor 594 (1:1000 dilution, Molecular Probes, Eugene, OR) was incubated with the coverslips for 1 h at room temperature. Following more washes, nuclei were stained with Hoechst 33258 at a final concentration of 2 µg/ml for 1 min. The coverslips were then washed 2–3 times and mounted onto glass slides with FluoroMount and kept in the dark at 4°C. Cell images were captured with a fluorescence microscope equipped with an Olympus DP70 digital camera (Olympus IX71).
For intracellular TNF staining, mixed glial cultures from C57BL/6 mice were washed twice with media and then treated with LPS at 0, 0.1 or 1.0 µg/ml for 8 or 24 h. During the last 5 h of stimulation, protein secretion was blocked using the protein transport inhibitor Brefeldin A (eBioscience, San Diego, CA) diluted 1:1000 in BDM alone or in BDM containing LPS. Cells were then fixed, permeabilized, blocked, and immunostained as described above.
Cell surface phosphatidylserine (PS) was detected with an annexin-V-FITC apoptosis detection kit (Sigma) according to the manufacturer’s instructions with slight modification. Cells were treated with LPS (1.0µg/ml) or media for 2, 6, 8, 12, 24 and 48h, at which time an aliquot of the supernatants was removed and stored at −80°C for subsequent TNF ELISA and LDH analyses. The cells were washed twice with sterile filtered TBS, and 500µl of binding buffer containing 5µl annexin-V-FITC was added to each well for exactly 10 min. The cells were then fixed with 2% PFA and stained for O4.
Cell lysates were subjected to SDS-polyacrylamide gel electrophoresis followed by electrotransferring of separated proteins to PVDF membrane. Nonspecific binding was blocked with TBS-T (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Tween-20) containing 5% non-fat milk for 1 h at room temperature. Anti TNFR1 (1: 500 dilution, Rabbit IgG, Santa Cruz Biotech, Santa Cruz, CA.), and β-actin (1:10,000 dilution) antibodies were diluted in TBS-T containing 5% non-fat milk and incubated overnight with the membrane at 4°C. After washing 3–5 times with TBS-T, the membrane was incubated with horseradish peroxidase conjugated secondary antibody (1:2000) for 1 h and was visualized by chemiluminescence using the SuperSignal detection kit (Thermo Scientific, Rockford, IL).
All cell culture treatments were performed in triplicate. Unless specified otherwise, data were presented as mean ± SD. Results were analyzed by one-way ANOVA followed by Bonferroni’s post-hoc test to determine statistical significance. Comparison between two experimental groups was based on two-tailed t test. P<0.05 was considered statistically significant.
As mentioned above, multiple lines of evidence have revealed a strong relationship between maternal/fetal infection, LPS, and pathogenesis of PVL. To elucidate the mechanism underlying endotoxin-induced white matter injury, an in vitro model was developed. Using co-cultures of preOLs with other cells such as microglia or astrocytes, we and others identified an indispensible role of microglia in mediating preOL cell death (Lehnardt et al. 2002, Li et al. 2005, Li et al. 2008). Consistent with previous findings, in mixed glial cultures prepared from neonatal rat forebrains, addition of LPS induced robust microglia activation and proliferation with concomitantly significant loss of preOLs (Fig. 1A–C). LPS activated microglia exhibit increased immunoreactivity to Iba1 and larger cell bodies and processes. A significant portion of activated microglia appeared to be amoeboid (Fig. 1C).
It is well accepted that apoptotic cells are rapidly cleared by phagocytes in vivo without intracellular content release (Lauber et al. 2004). We found that GFP+ preOLs, when seeded on top of established mixed glial cultures, were susceptible to LPS and their cell corpses were engulfed by microglia (Fig. 1D). Consistent with this finding, we found increased cell surface PS on preOLs in mixed glial cultures treated with LPS, as determined by Annexin V staining (Fig. 1E, H). Translocation of PS from the inner leaflet of plasma membrane to the outer leaflet represents a “eat-me” signal for phagocytes (Lauber et al. 2004). This LPS-induced PS translocation in preOLs correlated with TNF production and proceeded cell death (Fig. 1F, G). These results show that dying preOLs were efficiently removed by microglia via phagocytosis, and are in agreement with our observation that despite significant loss of preOLs in LPS-treated mixed glial cultures, there was minimal cell debris.
We previously demonstrated that production of TNF is required for LPS toxicity to preOLs in mixed glial cultures (Li et al. 2008). However, the cellular source for TNF production was not identified. Since both microglia and astrocytes are the main responders in the CNS to inflammation and capable of producing various chemokines and cytokines in vivo and in vitro (Hanisch 2002, Farina et al. 2007), we examined which cell types or both of them produce TNF in response to LPS in cultures. Using enriched rat microglia and astrocyte monocultures, we found that LPS induced robust TNF production in microglia, but not astrocytes, in a concentration-dependent manner (Fig. 2A, B).
Previous studies suggest that astrocytes constitutively express low level of the LPS receptor, toll-like receptor 4 (TLR4) (Carpentier et al. 2005, Bowman et al. 2003). To further answer the question whether astrocytes produce TNF in response to LPS, we prepared highly purified monocultures (>99%), exposed them to LPS for 6 and 24 h, and examined various transcripts by RT-PCR (Fig. 2C). Whereas both astrocytes and microglia express TLR3, TLR4 was only detected in microglia, a finding consistent with a previous study (Lehnardt et al. 2002). LPS exposure resulted in a transient minor upregulation of TNF and IL-1β mRNA in astrocytes at 6 h, but not at 24 h (Fig. 2C). In contrast, LPS-stimulated microglia continuously expressed TNF and IL-1β. Additionally, the anti-inflammatory cytokine IL-10 was also detected in LPS-activated microglia, especially at the later time point, suggesting an autoregulatory mechanism for microglial activation, potentially mediated through phagocytosis of dying preOLs (Fig. 1).
To investigate whether astrocytes are capable of producing TNF in response to factors produced by LPS-activated microglia, we immunostained for TNF in mixed glia treated with LPS for 8 and 24h. Whereas no TNF immunoreactivity was detected in the absence of LPS stimulation, LPS at concentrations of 0.1–1 µg/ml induced robust TNF immunoreactivity in cells with activated microglial morphology (Fig. 2D). We did not detect a single TNF-positive cell that was also GFAP positive at both time points (>1000 TNF+ cells examined). Instead, TNF immunoreactivity was completely restricted to microglia at various activation stages, with the most active amoeboid microglia producing the strongest TNF signal (Fig. 2E, arrow vs. arrowhead). Taken together, our data demonstrate that microglia but not astrocytes produce TNF upon LPS stimulation.
Ablation of tnf in mixed glia completely prevented LPS toxicity (Li et al. 2008). If microglia are indeed responsible for TNF generation, then selective ablation of tnf in microglia, but not in astrocytes and OLs, should prevent LPS toxicity. To test this hypothesis, we isolated each individual cell type from WT and TNF KO mice, and reconstituted them in different combinations to create mosaic mixed glial cultures with a composition similar to that of regular mixed glial cultures. We then subjected these cultures to vehicle or LPS treatment for 48 h and examined preOL survival by counting the number of O4+ cells.
Consistent with our above finding in regular mixed glial cultures (Fig. 1A), LPS also induced significant loss of preOLs in reconstituted mixed cultures containing WT astrocytes, microglia and OLs (Fig. 3A). However, in mixed glia containing TNF KO instead of WT microglia, the LPS toxicity was abrogated (Fig. 3B). In contrast, TNF deficiency in astrocytes or OLs did not render any protection (Fig. 3C–E). These results clearly demonstrate an indispensible role for microglial TNF in mediating LPS toxicity to preOLs. Consistent with the survival data, analyses of secreted TNF in these mosaic mixed glial culture revealed that when LPS failed to cause preOL death in cultures containing TNF KO microglia (Fig. 3B), LPS-induced TNF production was also significantly diminished (Fig. 3G vs. F). It should be noted that LPS-induced TNF production was not completely abrogated in cultures containing TNF KO microglia, most likely due to contaminating WT microglia in WT astrocytes and OLs preparations.
We next questioned whether microglia-derived TNF exerts its toxic effects on preOLs in an autocrine or a paracrine fashion. Since ligation of TNFR1 has been shown to be primarily related to cell death and activation of TNFR2 to cell proliferation and protection (Arnett et al. 2001, Probert et al. 2000), we tested whether TNFR1 is responsible for the LPS toxicity. Indeed, ablation of TNFR1 in all glial cells completely prevented LPS-induced LDH release and loss of preOLs, even though TNFR1 KO cultures produced similar levels of TNF as the WT (Fig. 4A–D). These data demonstrate that TNFR1 is necessary for LPS-induced killing of preOLs.
PreOLs, microglia and astrocytes all express TNFR1 (Dopp et al. 1997). As activation of the TNFR1 pathway in either microglia or astrocytes could potentially induce OL death via secreted factors, it was important to determine which cell was responsible for the TNFR1-mediated effect. Therefore, we isolated GFP-expressing preOLs, which also expressed TNFR1, and seeded them onto either WT or TNFR1 KO mixed glial cultures (Fig. 4E). If oligodendroglial TNFR1, but not microglial and astroglial TNFR1, is indeed essential for LPS-induced preOL death, these exogenously added GFP+/TNFR1+ preOLs should demonstrate similar vulnerability toward LPS regardless being in WT or TNFR1 KO mixed glia. On the other hand, if activation of TNFR1 on microglia or astrocytes is responsible for LPS-induced preOL death, then these GFP+/TNFR1+ preOLs should be less vulnerable in TNFR1 KO mixed glial cultures. We found that upon LPS treatment, similar number of GFP+/TNFR1+ preOLs underwent degeneration regardless the TNFR1 status of the recipient mixed glial cultures (Fig. 4E, F). As an internal control, the endogenous TNFR1−/− preOLs in the TNFR1−/− mixed glial cultures were completely resistant to LPS (Fig. 4B). These data strongly suggest that ligation of oligodendroglial TNFR1 by TNF generated from microglia accounts for LPS-induced death of preOLs in an environment where all types of glia co-exist.
Our results demonstrated that activation of oligodendroglial TNFR1 by LPS-induced production of TNF is required for selective killing of preOLs in mixed glial cultures. Interestingly, when preOLs were exposed to increasing concentrations of recombinant TNF, no significant cell loss was observed (Fig. 5A). Similarly, TNF also had limited effect on mature OLs (Fig. S1A). Conditioned media from mixed glia previously treated with LPS were not toxic when added to pure preOLs (Fig. 5B, C). This is in sharp contrast to selective preOL death in LPS-treated mixed glial cultures. Thus, the above data indicate that a non-cell autonomous pathway is operative together with oligodendroglial TNFR1 for TNF-mediated toxicity in mixed glial cultures.
We previously showed that LPS-activated microglia kill preOLs via peroxynitrite in co-cultures, but via TNF in mixed glial cultures (Li et al. 2005, Li et al. 2008). Since TNF is not toxic to pure preOLs as demonstrated above, we hypothesized that astrocytes directly contribute to TNF-mediated toxicity. Whereas preOLs themselves were not sensitive to TNF, they became so when cultured together with astrocytes (Fig. 6A), demonstrating an essential role for astrocytes in TNF-induced loss of preOLs. In contrast, mature MBP-expressing OLs were not significantly affected by TNF (Fig. S1B). This observation is in agreement with earlier studies demonstrating maturation-dependent vulnerability of preOLs to oxidative stress, excitotoxicity and cytokines (Haynes et al. 2005, Butts et al. 2008, Andrews et al. 1998, Horiuchi et al. 2006). Interestingly, when preOLs were co-cultured with astrocytes in the absence of cell-cell contact, as illustrated in Fig. 6B, they were no longer vulnerable to TNF (Fig. 6C). Similarly, conditioned medium from TNF-treated astrocytes was not toxic to preOLs. These results suggest that direct cell contact between preOLs and astrocytes is a prerequisite for TNF toxicity.
One possibility is that astrocytes promote TNF-mediated preOL death by upregulating TNFR1 in preOLs since oligodendroglial TNFR1 is required for the toxicity (Fig. 4). However, LPS neither increased the level of total TNFR1 of mixed glia (Fig 7A, top panel), nor did it induce TNFR1 specifically in preOLs that were plated into TNFR1−/− mixed glial cultures (Fig. 7A, lower panel), a condition where these exogenously added preOLs were susceptible to LPS treatment. Moreover, although astrocytes are required for TNF-induced toxicity to preOLs, activation of astrocytic TNFR1 is not (Fig. 7B). This is in line with our above finding (Fig. 4) and demonstrates that TNFR1 on preOLs is necessary but not sufficient for TNF-induced cell death.
In this study we demonstrated that oligodendroglial TNFR1 is indispensible for LPS-triggered death of preOLs, and revealed a non-cell autonomous death pathway that involves astrocytes and production of TNF from activated microglia. We further demonstrated that microglia not only are responsible for LPS toxicity to preOLs but also actively phagocytose dying preOLs in culture. We showed that conditioned media derived from LPS-challenged mixed glia is not toxic to preOLs, and that recombinant TNF does not cause significant cell loss in pure preOLs, suggesting that activation of TNFR1 alone is not sufficient to cause cell death. In contrast, astrocytes render preOLs vulnerable to TNF in a contact-dependent manner (Fig. 8). Our results suggest, for the first time, that astrocytes contribute positively to TNF-mediated toxicity to preOLs, and may reconcile previous contradictory observations from different laboratories on the effect of TNF on OLs. Our data clearly indicate that intercellular interactions among various glial cells are important and necessary for preOL destruction under inflammatory conditions. Complete understanding of these communications and death of oligodendroglial cells would enable development of strategies that selectively target deleterious aspects of proinflammatory cytokines such as TNF while preserve their beneficial effects.
Multiple lines of evidence have indicated a deleterious role for TNF in OL injury in vivo (Probert et al. 2000, Akassoglou et al. 1998, Probert et al. 1995, Stalder et al. 1998), in organotypic cultures (Selmaj & Raine 1988), and in vitro (Jurewicz et al. 2005, Andrews et al. 1998, Li et al. 2008). Production of TNF and expression of TNF receptor R1 and R2 are found to be locally up-regulated in human PVL (Kadhim et al. 2001, Deguchi et al. 1996) as well as in multiple sclerosis (MS) lesions (Selmaj et al. 1991a, Hofman et al. 1989, Raine et al. 1998). Blockade of TNF prevents initiation of pathology and ameliorates progression of demyelination in animal models of MS (Selmaj et al. 1991b, Korner et al. 1997, Baker et al. 1994, Ruddle et al. 1990). Moreover, localized over-expression of TNF in the CNS selectively induces OL death and demyelination in the context of an intact blood brain barrier and absence of immune cell infiltration into the CNS parenchyma (Akassoglou et al. 1998, Kassiotis & Kollias 2001, Probert et al. 1995). This TNF pathology appears to be mediated through TNF/TNFR1 since both cytotoxic and inflammatory effects of TNF are abrogated in mice deficient in TNFR1 (Akassoglou et al. 1998). Other animal studies also revealed critical roles for TNF/TNFR signaling in triggering CNS inflammation and demyelination (Probert et al. 2000, Akassoglou et al. 1997, Stalder et al. 1998), and demonstrated that macrophage/microglial cells, but not lymphocytes play a central role in mediating injury (Stalder et al. 1998). Unexpectedly, anti-TNF therapies in human MS was of little efficacy or even resulted in worsened symptoms in some patients (TNF neutralization in MS 1999), clearly indicating the complexity of TNF signaling and call upon further understanding and investigation into these pathways. On the other hand, several studies have demonstrated a protective role for TNF. Using genetically modified mice, TNF was found to be required for preOLs proliferation and remyelination in a toxin-induced demyelination model, and this beneficial effect of TNF appears to be mediated by TNFR2 (Arnett et al. 2001). Although TNF is elevated in PVL and MS as well as in many other neurological disorders including ischemia, neurodegenerative diseases and traumatic brain injury, whether TNF signaling actively contributes to pathogenesis, prevents cell injury, and/or participates in repair process in these disorders has yet to be established due to the complex actions of TNF on neurons and glia (reviewed by (McCoy & Tansey 2008, Chadwick et al. 2008). The contrasting protective and deleterious effects of TNF in models of CNS demyelination suggest that selectively targeted, rather than global blockade of TNF signaling events, is required for achieving any meaningful therapeutic outcome. Our current study indicates that selective inhibition of oligodendroglial TNFR1 may offer therapeutic advantages and avoid the adverse effect of TNF-targeted therapies in the CNS. Future animal studies employing conditional knockout of TNFR1 in OL lineage cells is thus needed to explore this revenue.
Our data support a prominent role for activated microglia in inflammatory OL injury. Primary microglia express an array of TLRs including TLR4 that transduces LPS signaling, respond robustly to LPS stimulation, produce proinflammatory cytokines such as TNF and IL-1β, and cause selective preOL death via a mechanism involving oligodendroglial TNFR1. Interestingly, we also observed efficient engulfment of dying/dead preOLs by reactive microglia. It is not clear whether these phagocytozing microglia are of different populations or activation stages from microglia that are producing TNF and kill preOLs. When removing dying cells or myelin debris, phagocytozing microglia were shown to release anti-inflammatory factors (Magnus et al. 2001, Liu et al. 2006), consistent with our observation on microglial production of the anti-inflammatory cytokine IL-10 over time (Fig. 2C). Recent evidence suggests that when challenged, microglia undergo a sequence of activated states with different phenotypic responses and functional diversity, being either protective or toxic, but how these are regulated remain unclear (Hanisch & Kettenmann 2007). Indeed, LPS activated microglia in mixed glial cultures exhibit an array of morphology with amoeboid microglia, the likely fully activated state, producing the highest amount and other activation stages producing lower amount of TNF (Fig. 2E). It is possible that it is these amoeboid microglia that are cytotoxic to preOLs while some less activated microglia may be beneficial either by removing dying cells or by secreting trophic factors. Further studies will be required to investigate the underpinning mechanisms. On the other hand, this microglial functional diversity also highlights the need for specific blockade of only deleterious signaling events.
To date, the mechanism by which TNF causes a toxic effect to OL lineage cells remains unresolved. Conflicting data on the effect of TNF on OL lineage cells have been reported from various laboratories. While some found toxicity of TNF on cultured OLs, others did not; and some observed developmental-dependent toxicity (Merrill & Scolding 1999, Andrews et al. 1998, Horiuchi et al. 2006, Scurlock & Dawson 1999, Cammer 2000, Jurewicz et al. 2005, Pang et al. 2005, Taylor et al. 2010). In light of our finding of an essential role for astrocytes in TNF-mediated toxicity to preOLs, some discrepancies may in part due to different culture compositions and the presence of astrocytes. How exactly astrocytes promote TNF toxicity to preOLs is currently unknown, but direct cell-cell contact appears to be essential. TNF is a pleiotropic cytokine that exerts multiple cellular responses including apoptosis, cell survival and proliferation in a receptor- and cell context-dependent manner. Both OLs and astrocytes express receptors for TNF, therefore, it is possible that astrocytes respond to TNF (through receptor other than TNFR1, Fig. 7B) and express factors that, through direct interaction with preOLs, enhance TNF/TNFR1 death signaling in preOLs or suppress survival signals. One possibility for contact-dependent cell killing is through gap junctions since gap junctions are known to couple astrocytes and OLs (Nagy & Rash 2000, Orthmann-Murphy et al. 2008) and have been previously suggested in propagation of cell injury (Lin et al. 1998, Farahani et al. 2005, Froger et al. 2010). Further study is needed to dissect, at molecular level, the interplay between OLs and astrocytes under inflammatory conditions. It should be noted that despite astrocytes are generally considered protective by releasing trophic factors such as CNTF, LIF and IL-11, a pathological role for reactive astrocytes has also been implicated in white matter diseases (Williams et al. 2007, Haynes et al. 2005). It is also noteworthy that neurons or axons could directly or indirectly affect preOL injury, or vise versa, in the cerebral white matter, as axonal injury was recently reported in PVL diffuse lesions (Haynes et al. 2008). Since reciprocal axon-glial interactions are essential for myelinogenesis of the developing brain (Barres & Raff 1999, Simons & Trajkovic 2006), it would be interesting to examine whether activated microglia or TNF differentially affect various cell population in the white matter.
In summary, our study reveals a detrimental role for engaging oligodendroglial TNFR1 in LPS-triggered death of preOLs in an environment where all glial cells coexist, identifies an essential role for astrocytes in TNF toxicity to preOLs, and offers new insights into the mechanisms of TNF-mediated injury to preOLs. This study also suggests that selected inhibition of TNFR1 signaling in preOLs may be beneficial in white matter diseases.
This work was supported by NIH grant R01NS060017, by the research grant RG3975 from the National Multiple Sclerosis Society, and by the start-up funds from Texas A&M University. We thank Jie Peng and Sarah Dobrovolny for technical assistance.
The authors have no conflict of interest to declare.