Reactive microglia and astrocytes are present in lesions of white matter disorders, such as periventricular leukomalacia and multiple sclerosis. However, it is not clear whether they are actively involved in the pathogenesis of these disorders. Previous studies demonstrated that microglia, but not astrocytes, are required for lipopolysaccharide (LPS)-induced selective killing of developing oligodendrocytes (preOLs) and that the toxicity is mediated by microglia-derived peroxynitrite. Here we report that, when astrocytes are present, the LPS-induced, microglia-dependent toxicity to preOLs is no longer mediated by peroxynitrite but instead by a mechanism dependent on tumor necrosis factor-α (TNFα) signaling. Blocking peroxynitrite formation with nitric oxide synthase (NOS) inhibitors or a decomposition catalyst did not prevent LPS-induced loss of preOLs in mixed glial cultures. PreOLs were highly vulnerable to peroxynitrite; however, the presence of astrocytes prevented the toxicity. Whereas LPS failed to kill preOLs in cocultures of microglia and preOLs deficient in inducible NOS (iNOS) or gp91phox, the catalytic subunit of the superoxide-generating NADPH oxidase, LPS caused a similar degree of preOL death in mixed glial cultures of wild-type, iNOS−/−, and gp91phox−/− mice. TNFα neutralizing antibody inhibited LPS toxicity, and addition of TNFα induced selective preOL injury in mixed glial cultures. Furthermore, disrupting the genes encoding TNFα or its receptors TNFR1/2 completely abolished the deleterious effect of LPS. Our results reveal that TNFα signaling, rather than peroxynitrite, is essential in LPS-triggered preOL death in an environment containing all major glial cell types and underscore the importance of intercellular communication in determining the mechanism underlying inflammatory preOL death.
Glial cells are essential for the development and function of the CNS (Volterra and Meldolesi, 2005). Microglia are the resident macrophage-like cells in the CNS and extremely responsive to environmental stress and immunological challenges (Kreutzberg, 1996). Activation of microglia has been implicated in a number of neurological disorders, including the white matter disorders periventricular leukomalacia (PVL) (Haynes et al., 2005) and multiple sclerosis (Trapp et al., 1999). Reactive astrocytes are frequently present in various lesions of the nervous system. A localized activation of microglia and astrocytes may have both beneficial and detrimental effects on neighboring cells (John et al., 2003; Minghetti, 2005). Both cell types are prime immune regulators in the CNS, exhibit great plasticity toward injury, and are capable of producing many inflammatory mediators, including cytokines, such as tumor necrosis factor α (TNFα), and chemokines, as well as reactive oxygen/nitrogen species and trophic factors (Ridet et al., 1997; Dong and Benveniste, 2001; Hanisch, 2002; Pekny and Nilsson, 2005).
Brain injury in premature infants is a common perinatal disorder and a major cause of life-long neurological disability that accounts for enormous personal and societal burden. PVL is the major form of cerebral white matter injury that underlies most of the neurological sequelae, including cognitive deficits associated with prematurity (Volpe, 2001b; Haynes et al., 2005). Preoligodendrocytes (preOLs) are the major cell type selectively injured in PVL. Hypoxia/ischemia and immune responses in the CNS to maternal/fetal infection are two primary components of the pathogenesis of PVL (Volpe, 2001a; Riddle et al., 2006), which is characterized by both focal necrosis with loss of all cellular elements and diffuse white matter injury leading to subsequent myelination deficits (Volpe, 2003).
Considerable clinical, in vivo, and in vitro evidence points to a strong link between bacterial endotoxin lipopolysaccharide (LPS) and PVL (Gilles et al., 1976; Grether and Nelson, 1997; Lehnardt et al., 2002; Pang et al., 2003; Wang et al., 2006). Many studies have demonstrated selective white matter lesions in fetal and neonatal animals after local, systemic, or intrauterine administration of LPS (Hagberg et al., 2002). However, the mechanisms underlying this inflammatory injury to preOLs remain elusive. Microglia and astrocytes are profoundly activated in the diffuse white matter lesions of PVL (Haynes et al., 2003), suggesting a role in mediating preOL injury. In vitro, LPS toxicity to OLs is not cell autonomous and only occurs in cultures containing microglia through activation of the microglial Toll-like receptor 4 (TLR4) signaling pathway (Lehnardt et al., 2002). We recently demonstrated that peroxynitrite, a short-lived potent oxidant and the reaction product of nitric oxide (NO) and superoxide, is the toxic microglial factor responsible for LPS-induced death of preOLs (Li et al., 2005). Unexpectedly, we found that, although astrocytes do not contribute directly to LPS toxicity to preOLs, their presence completely changed LPS-induced cell death mechanisms. In this study, we demonstrate that, when astrocytes are present together with microglia and preOLs, LPS-induced, microglia-dependent toxicity to preOLs is mediated instead by a mechanism dependent on TNFα signaling.
Materials and Methods
LPS (Escherichia coli O111:B4) was obtained from Sigma (St. Louis, MO). Wild-type, mutant, or knock-out mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Various cytokines were obtained from R & D Systems (Minneapolis, MN). PDGF and basic FGF were from PeproTech (Rocky Hill, NJ). 3-Morpholino-sydnonimine (SIN-1), NG-monomethyl-L-arginine (l-NMMA), iron (III) tetrakis(1-methyl-4-pyridyl)porphyrin pentachloride (FeTMPyP), and peroxynitrite were purchased from Cayman Chemical (Ann Arbor, MI). Recombinant reporter adenovirus was from Gene Transfer Vector Core (University of Iowa, Iowa City, IA). Antibodies against CD68 and GFAP were from Millipore Bioscience Research Reagents (Temecula, CA), and inducible nitric oxide synthase (iNOS) was from BD Transduction Laboratory (San Jose, CA). Caspase inhibitors were from Axxora (San Diego, CA). Unless specified otherwise, all other reagents were from Sigma.
Primary cell cultures.
Primary preOLs, microglia, astrocytes, and mixed glial cultures were prepared from the forebrains of 1- to 2-d-old rats or mice using a differential detachment method (McCarthy and de Vellis, 1980; Li et al., 2005; Chen et al., 2007). Briefly, forebrains free of meninges were digested with HBSS containing 0.01% trypsin and 10 μg/ml DNase and triturated with DMEM containing 20% heat-inactivated fetal bovine serum and 1% penicillin–streptomycin. Dissociated cells were plated onto poly-d-lysine-coated 75 cm2 flasks or directly into 24-well plates for experiments using mixed glia and fed every other day for 7–10 d. Microglia were isolated by shaking the mixed glia-containing flasks for 1 h at 200 rpm. The purity of microglia was consistently >95%. After removing microglia, the flasks were subjected to shaking at 200 rpm overnight to separate preOLs from the astrocyte layer (Li et al., 2005). The suspension was plated onto uncoated Petri dishes for 1 h to further remove residual contaminating microglia/astrocytes. PreOLs were plated either by themselves or onto 24-well plates containing microglia or astrocytes for cocultures. PreOLs were maintained in a 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) containing 10 ng/ml PDGF and 10 ng/ml basic FGF for 5–9 d. The OL cultures were primarily progenitors and precursors (A2B5+, O4+, O1-negative (O1−), myelin basic protein−] and are therefore referred to as preOLs. Contamination by astrocytes and microglia was <2% in preOL monocultures. Astrocytes were purified from the astrocyte layer in the flask after being exposed to a specific microglia toxin l-leucine methyl ester (1 mm) for 1 h. The enriched astrocytes were consistently >95% pure, with preOLs being the major contaminating cells. For cocultures, fixed cell numbers of microglia (2–3 × 104 cells per well), astrocytes (1 × 104 cells per well), and preOLs (4–5 × 104 cells per well) were plated into 24-well culture plates and used within 2–4 d. Mouse mixed glial cultures were prepared with the same methods as described above from iNOS, gp91phox, TNFα, TNF receptors 1/2 (TNFR1/2), or interferon γ (IFNγ) knock-out mice.
Cell treatment and cell survival determination.
Cell death was induced by exposure to LPS (E. coli O111:B4; Sigma) or cytokines in the presence of various reagents as specified in the figure legends. Accumulation of nitrite in the medium was determined by the Griess reaction as described previously (Li et al., 2005). SIN-1 and peroxynitrite treatment was performed as described previously (Zhang et al., 2006). Cells were washed twice and then placed in Earle's balanced salt solution. Increasing concentrations of SIN-1 or authentic peroxynitrite were then added to cells. After 1 h incubation with SIN-1 or peroxynitrite, the cells were switched back to BDM medium, and cell survival was analyzed 20–48 h later. Survival of preOLs was determined by counting O4-positive (O4+) 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.
Infection of preOLs with recombinant adenovirus.
Purified preOLs were infected with adenovirus containing green fluorescent protein (GFP) cDNA (AdGFP) as we described previously (Baud et al., 2004a). Briefly, cells were exposed to 1 × 108 pfu/ml AdGFP overnight in regular culture medium followed by a complete medium change the next day. Cells were allowed to recover and express GFP for 48 h. The infection rate was consistently >90% with minimum toxicity. Cells were then trypsinized off the culture dish and seeded at density of 1–2 × 104 per well into regular mixed glial cultures. The next day, cell cultures were challenged with LPS or TNF for 24–48 h, and morphology and the extent of loss of GFP+ cells were analyzed.
Analysis of TNFα production.
The concentrations of TNFα in the culture media of cells treated as specified were measured using commercially available ELISA kits according to the instructions of the manufacturer (eBioScience, San Diego, CA). Absorption at 450 nm was determined in a microplate reader (Fluostar Optima; BMG Labtech, Offenburg, Germany). The detection limit of the ELISA was 8 pg/ml for TNFα.
Immunocytochemistry, terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling, and immunofluorescence microscopy.
After treatments, cells were fixed with 4% paraformaldehyde in PBS for 10 min, washed with PBS, and blocked with TBST (50 mm Tris-HCl, pH 7.4, 150 mm NaCl, and 0.1% Triton X-100) or TBS (for O4 immunostaining) containing 5% goat serum. The coverslips were incubated with antibody O4 (1:500) or antibodies against CD68 (1:100), GFAP (1:1000), or iNOS (1:1000) overnight at 4°C. After washes, secondary antibody conjugated with either Alexa Fluor 488 or Alexa Fluor 594 (1:1000 dilution; Invitrogen, Carlsbad, CA) was incubated with the coverslips for 1 h at room temperature. In the case of labeling fragmented DNA, terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) was performed using an In Situ Cell Death detection kit according to the protocol of the manufacturer (Roche, Indianapolis, IN). After more washes, nuclei were stained with Hoechst 33258 at a final concentration of 2 μg/ml for 1 min. The coverslips were then washed two to three times and mounted onto glass slides with FluoroMount and kept in the dark at 4°C. Cell images were captured with a fluorescence microscope (model IX71; Olympus, Tokyo, Japan) equipped with an Olympus DP70 digital camera.
All cell culture treatments were performed in triplicate. Results were analyzed by one-way ANOVA, followed by Bonferroni's post hoc t test to determine statistical significance. Comparison between two experimental groups was based on two-tailed t test. p < 0.05 was considered statistically significant.
LPS-induced killing of oligodendrocyte precursors is mediated through activated microglia
Previous studies showed that LPS causes significant toxicity to OLs in culture and in vivo (Merrill et al., 1993; Pang et al., 2000; Lehnardt et al., 2002; Li et al., 2005; Wang et al., 2006). Subsequent studies demonstrated that the effect of LPS on developing OLs is not cell autonomous but rather dependent on microglia activation through TLR4 (Lehnardt et al., 2002). Using mixed glial cultures containing all major CNS glial cell types (microglia, astrocytes, and OLs), we found that LPS selectively killed preOLs (O4+, O1−) (Fig. 1A). To determine which cell type was responsible for the LPS toxicity, we tested the effect of LPS on preOLs in various monocultures and cocultures. Consistent with previous findings (Lehnardt et al., 2002; Li et al., 2005), LPS had no effect on pure preOLs or preOLs cocultured with astrocytes but was highly toxic to preOLs when preOLs were cocultured with microglia (Fig. 1B). These results confirmed a pivotal role for activated microglia in LPS-mediated toxicity. Using this preOL plus microglia coculture model, we determined that peroxynitrite generated by LPS-activated microglia is the toxin responsible for LPS-induced death of preOLs (Li et al., 2005).
To visualize morphological features of dying preOLs and to validate the methodology used to quantify preOL survival in mixed glial cultures, we infected purified preOLs with adenovirus containing GFP cDNA and then seeded them into established mixed glial cultures that contained regular preOLs. Expression of GFP in infected preOLs allowed us to clearly visualize the morphology of the entire preOL, including their fine processes (Fig. 2A). Exposure of mixed glial cultures containing added GFP+ preOLs to LPS resulted in significant degeneration and loss of GFP+ cells when observed at 24 h after LPS treatment (Fig. 2B,C). It appeared that dying preOLs had multiple abnormal morphologies. Some appeared to have beaded processes and bulb formation (Fig. 2B, top and bottom left), whereas others appeared to form apoptotic bodies (Fig. 2B, bottom right). No degenerating GFP+ cells were observed in controls. The extent of preOL survival as determined by counting live GFP+ cells was similar to that determined by counting live O4+ preOLs. These data suggest that LPS-induced death of preOLs is not synchronous and may occur by multiple pathways. Our data also demonstrate that the method used to determine preOL survival by counting O4+ preOLs in this study is reliable and accurate.
Nitric oxide is not required for LPS toxicity in mixed glial cultures in which all CNS glial cell types are present
Peroxynitrite is a short-lived potent oxidant formed when NO reacts with superoxide anion, a reaction that occurs at a diffusion-limited rate (Pacher et al., 2007). We found that iNOS was rapidly upregulated in microglia in mixed glial cultures exposed to LPS (Fig. 3A,B). iNOS upregulation and thus NO production appeared to correlate with preOL process retraction and cell death (Fig. 3A). However, to our surprise and in striking contrast to the fact that blocking NO production effectively prevents LPS toxicity in cocultures of preOLs and microglia (Li et al., 2005), the NOS inhibitor l-NMMA had no effect on LPS toxicity in the mixed glial cultures despite the fact that it completely blocked NO production (Fig. 3C,D). The iNOS-specific inhibitor 1400W also did not abolish LPS toxicity when mixed glial cultures were used (Fig. 3E). Furthermore, the peroxynitrite decomposition catalyst and superoxide scavenger FeTMPyP (Misko et al., 1998), which blocks LPS toxicity in cocultures (Li et al., 2005), did not prevent preOL death in mixed glial cultures. These results indicate that, although astrocytes are not required for LPS toxicity (Fig. 1B), their presence changed the death mechanism from an NO-dependent mechanism to a mechanism independent of NO and peroxynitrite production.
To determine the mechanism underlying LPS-induced preOL death in mixed glial cultures, we next tested whether excitotoxicity and oxidative injury were involved because astrocytes are capable of releasing glutamate under inflammatory conditions and preOLs are known to be highly vulnerable to AMPA/kainate receptor-mediated excitotoxicity as well as oxidative stress (Oka et al., 1993; Back et al., 1998; Fern and Moller, 2000; Follett et al., 2000; Deng et al., 2003; Rosenberg et al., 2003; Baud et al., 2004b). Neither the AMPA/kainate receptor antagonist NBQX nor the antioxidants vitamin E and vitamin K2, shown previously to be effective in preventing oxidative injury to preOLs (Li et al., 2003), had any protective effect (Fig. 3E). To determine whether the cells died through a caspase-dependent apoptotic pathway, we next tested the effect of caspase inhibitors. The pan caspase inhibitor zVAD-fmk (N-benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl ketone) and the caspase-3 inhibitor z-DEVD-fmk (z-Asp-Glu-Val-fluoromethyl ketone) did not prevent the loss of preOLs (data not shown), indicating that a caspase-independent cell death pathway was operative. Consistent with this observation, there was minimal caspase-3 activation during LPS challenge in mixed glial cultures (data not shown). Because we have observed apparent apoptotic body formation in LPS-treated mixed glial cultures containing GFP+ preOLs (Fig. 2B), we asked whether dying preOLs contain fragmented chromatin, a common feature of apoptotic cell death. Unexpectedly, we did not observe TUNEL+ O4+ preOLs (data not shown). In contrast with this observation made in the mixed glial cultures, TUNEL+ preOLs were evident in cocultures of microglia and preOLs treated with LPS (Fig. 3F), in which we demonstrated previously that LPS-induced preOL toxicity is mediated by generation of peroxynitrite (Li et al., 2005). These results provide additional evidence for a different cell death mechanism induced by LPS when astrocytes are present. It should be noted that programmed cell death can take many forms, ranging from necrosis-like, apoptotic-like programmed cell death to classical apoptosis (Leist and Jäättelä, 2001).
Inflammatory cytokines are highly toxic to oligodendrocyte precursors in mixed glial cultures regardless of their ability to induce high levels of nitric oxide
Besides reactive nitrogen/oxygen species, activated microglia are also capable of producing a number of proinflammatory cytokines, including TNFα and interleukin-1β (IL-1β) (Hanisch, 2002). Because it has been shown previously that preOLs are highly sensitive to TNFα and interferon γ (IFNγ) (Merrill et al., 1993; Andrews et al., 1998) and combinations of these cytokines are powerful inducers of iNOS (Possel et al., 2000; Saha and Pahan, 2006), we asked whether these proinflammatory cytokines are toxic to preOLs in the mixed glial culture and whether blockade of inducible NO production abolishes cell death. Combinations of cytokines, such as TNFα, IFNγ, and IL-1β, induced profound preOL death as well as iNOS expression and NO production (Fig. 4A–C). However, although l-NMMA completely blocked NO production, it had no protective effect against this cytokine-induced toxicity, indicating that increased NO production in response to cytokines does not account for the toxicity. More importantly, this result also demonstrates that NO, even at high concentrations, was not toxic to preOLs in mixed glia cultures. To confirm our above findings, we prepared mixed glial cultures from mice deficient in iNOS or gp91phox, the catalytic subunit of the superoxide-generating NADPH oxidase, and subjected the cells to LPS or cytokine mixture treatment. We showed previously that microglia deficient in iNOS or gp91phox failed to kill preOLs during LPS activation. However, when astrocytes were present together with preOLs and microglia, LPS as well as TNFα plus IFNγ caused similar levels of preOL death in wild-type, iNOS−/−, and gp91phox−/− cells (Fig. 4D). Together, these data indicate that NO or peroxynitrite is not essential to LPS- or cytokine-induced preOL death in mixed glial cultures that include astrocytes.
TNFα production is necessary for LPS-induced death of oligodendrocyte precursors
Next, we investigated whether LPS-induced cytokine production accounts for LPS toxicity. Mixed glia exposed to LPS for 24 h produced significant amounts of extracellular TNFα (1299 ± 62 pg/ml; n = 3, mean ± SD) compared with controls (22.2 ± 27 pg/ml; n = 3, mean ± SD). Antibodies neutralizing soluble TNFα, but not control IgG, significantly prevented LPS-induced toxicity (Fig. 5A), indicating that LPS-induced TNFα production mediates, at least in part, preOL death in mixed glial cultures. In agreement with this observation, exogenous TNFα caused significant preOL death when applied to mixed glia (Fig. 5B). A similar level of toxicity was found when TNFα was added to mixed glial cultures containing GFP+ preOLs (data not shown). It should be noted that the level of toxicity induced by exogenous TNFα was often less than that induced by LPS, although TNFα was used at a concentration approximately two orders of magnitude higher than the level induced by LPS, indicating that endogenous TNFα is much more efficient in killing preOLs. Interestingly, the effect of TNFα appears to be not cell autonomous because TNFα by itself was not toxic to pure preOLs but was injurious to preOLs when other glial cells were simultaneously present (Fig. 5B), suggesting the importance of glial cell–cell communication in causing preOL death. Conditioned media collected from LPS-treated mixed glia were not toxic to pure preOLs (data not shown). These results suggest that cell–cell contact and/or other local factors are required for efficient LPS-triggered killing of preOLs.
TNFα signaling through TNF receptors mediates LPS-induced toxicity to preOLs
To verify that TNFα signaling accounts for LPS-induced preOL death in mixed glial cultures, we isolated mixed glia from wild-type mice and mice deficient in TNFα or TNFR1/2. Cells deficient in TNFα were unable to produce TNFα during LPS stimulation (Fig. 6A) and were resistant to LPS-induced killing of preOLs (Fig. 6B). Because TNFα signals through receptors TNFR1 and TNFR2, disruption of both receptors should silence TNFα-mediated signaling. Indeed, when mixed glia from TNFR1/2 knock-outs were subjected to LPS, TNFα was still produced, but preOL death was completely abolished (Fig. 6). In contrast, LPS exposure of wild-type cells resulted in marked killing of preOLs.
It is known that IFNγ potently activates astrocytes and that transgenic overexpression of IFNγ in mature OLs results in demyelination (Corbin et al., 1996; Horwitz et al., 1997). IFNγ was found to be upregulated in reactive astrocytes in the diffuse white matter lesions of PVL (Folkerth et al., 2004). Because TNFα and IFNγ act synergistically in causing injury to preOLs and their toxicity is also independent of NO production (Fig. 4), we asked whether IFNγ may act together with TNFα in causing preOL death in mixed glial cultures treated with LPS, a possibility that also explains why exogenous TNFα appears less toxic than LPS. We found that, in IFNγ−/− mixed glial cultures, LPS was as effective in killing preOLs as in wild-type cultures (Fig. 6B). Therefore, we conclude that TNFα-mediated signaling is essential for LPS-induced toxicity to preOLs in mixed glial cultures and that this toxicity is independent of IFNγ.
Astrocytes prevent peroxynitrite-induced death of oligodendrocyte precursors
As we showed above and previously (Li et al., 2005), LPS is toxic to preOLs but only in the presence of microglia (Fig. 1B). We identified peroxynitrite as the microglial toxin responsible for LPS-induced death in cocultures of preOLs and microglia. In contrast, LPS-treated astrocytes had minimal effect on preOL viability, consistent with previous findings that microglia but not astrocytes and OLs express functional TLR4 in vitro (Lehnardt et al., 2002). However, in the present study, we found that, in an environment in which all three CNS glial cell types are present, the mechanism underlying LPS-induced toxicity is no longer mediated through peroxynitrite but rather by a different mechanism involving TNFα signaling. This apparent dilemma could be resolved if astrocytes protect against peroxynitrite-induced preOL death, thereby allowing other mechanisms such as those mediated by TNFα to become dominant. This hypothesis is in agreement with evidence that astrocytes have high antioxidative capacities (Peuchen et al., 1997).
To test this hypothesis, we first treated preOLs with SIN-1, a widely used peroxynitrite generator (Zhang et al., 2006), and found that preOLs are indeed highly vulnerable to peroxynitrite (Fig. 7). Even lower concentrations of SIN-1 (200 μm) caused nearly complete cell death when preOLs were evaluated 24 h later. We then tested the effect of SIN-1 on preOL viability in mixed glial cultures in which OLs, microglia, and astrocytes coexist. To our surprise, SIN-1 did not kill preOLs in these cultures, even at concentrations of 1 mm and when evaluated 48 h later. To determine the major cell type responsible for this protection of preOLs against the toxic effect of SIN-1, we examined the effect of SIN-1 on preOL viability when preOLs were cocultured with microglia or astrocytes. As predicted, preOLs were still sensitive to SIN-1 when cocultured with microglia, in agreement with our previous identification of peroxynitrite as the microglial toxin that mediates LPS toxicity. In contrast, SIN-1 had minimal effect when preOLs were cocultured with astrocytes (Fig. 7). Similarly, authentic peroxynitrite was also highly toxic to preOLs. At concentrations of 100 μm, peroxynitrite, but not vehicle controls, caused massive preOL death (Fig. 8). However, the presence of astrocytes significantly prevented the peroxynitrite-induced preOL death (Fig. 8). High concentrations of peroxynitrite were not specifically toxic to preOLs but were toxic to astrocytes as well. In summary, our data demonstrate that, although astrocytes prevent LPS-activated, microglial peroxynitrite-mediated toxicity, they do not change preOL cell fate but rather shift the death mechanism toward a TNFα-dependent mechanism.
This study has investigated the cellular mechanism by which LPS induces injury to oligodendrocyte precursors, the cell type predominantly damaged in the diffuse white matter lesion of PVL. To investigate the mechanism underlying LPS-induced selective loss of preOLs in primary mixed glial cultures, various single and cocultures were prepared. Consistent with previous reports (Lehnardt et al., 2002; Li et al., 2005), we found that activation of microglia, but not astrocytes, is absolutely required for LPS toxicity. In preOL-microglia cocultures, a diffusible potent oxidant, peroxynitrite, was identified as the primary underlying toxic factor killing preOLs (Li et al., 2005). Blocking peroxynitrite formation by preventing either NO production or superoxide production or enhancing peroxynitrite decomposition abolished preOL death in these cocultures. However, when the mechanism of LPS-induced toxicity to preOLs was reexamined in mixed glial cultures in which astrocytes were also present, paradoxically, LPS-induced toxicity was independent of peroxynitrite but instead relied on TNFα signaling, although copious amount of NO was produced. This apparent paradox was reconciled by the fact that astrocytes block peroxynitrite-mediated toxicity to preOLs.
The determination of a peroxynitrite-independent cell death pathway when astrocytes are present is based on the following evidence: (1) blocking NO production with the NOS inhibitors l-NMMA or 1400W had minimal effect on LPS toxicity in mixed glial cultures; (2) a peroxynitrite decomposition catalyst and superoxide scavenger, FeTMPyP, as well as other antioxidants did not protect preOLs; (3) disruption of genes encoding iNOS or the gp91phox NADPH oxidase, two enzymes that we identified previously to be indispensable for LPS-induced microglial killing of preOLs in cocultures with microglia (Li et al., 2005), did not prevent LPS-induced toxicity in mixed glial cultures; and (4) astrocytes blocked peroxynitrite toxicity to preOLs. These data suggest that astrocytes respond to activated microglia and unmask a cell death mechanism mediated by TNFα signaling.
Microglia activated by LPS have been shown to release proinflammatory cytokines such as TNFα and IL-1β (Hanisch, 2002). As a potent source of immunologically relevant cytokines, including TNFα, astrocytes also play a pivotal role in the type and extent of CNS immune and inflammatory responses. A previous study showed that induction of iNOS in astrocytes by IFNγ and IL-1β potentiates NMDA-receptor mediated excitotoxicity (Hewett et al., 1994). Activated microglia also enhance TNFα production and glutamate release from astrocytes, resulting in amplified neurotoxicity (Bezzi et al., 2001). Neuropathological studies have revealed that, in human PVL cases, but not age-matched controls, abundant hypertrophic reactive astrocytes and activated microglia populate diffuse white matter lesions (Haynes et al., 2003). Therefore, it is most likely that intercellular communication among these activated glia may play an important role in the pathogenesis of PVL. Our data demonstrate that, with the peroxynitrite cell death pathway inhibited by astrocytes, a cell death pathway orchestrated by TNFα/TNFR signaling becomes dominant for LPS-triggered injury to preOLs in culture.
Multiple lines of evidence suggest that several proinflammatory cytokines, including IFNγ (Folkerth et al., 2004), TNFα (Deguchi et al., 1996; Kadhim et al., 2001), and IL-6 (Yoon et al., 1997) and IL-2 (Kadhim et al., 2002) are elevated in PVL and may play pivotal roles in perinatal white matter injury (Dammann and Leviton, 1997; Rezaie and Dean, 2002; Pang et al., 2006; Smith et al., 2007). Proinflammatory cytokines such as TNFα are also potent regulators of glial activation and iNOS induction (John et al., 2003). TNFα is a prototypical proinflammatory cytokine that plays a central role in initiating inflammatory reactions of the innate immune system, in part through the induction of expression and release of other cytokines (Wajant et al., 2003). TNFα exerts its biologic functions by binding to and signaling through TNFR1 and TNFR2 in an autocrine and/or paracrine manner. In situ immunohistochemical studies revealed locally increased TNFR1/2 expression and TNFα production in both reactive astrocytes and microglia in human PVL lesions (Deguchi et al., 1996; Yoon et al., 1997; Kadhim et al., 2001). TNFα/TNFR1 was also responsible for optical nerve OL death and subsequent retinal ganglion cell loss in a mouse model of glaucoma (Nakazawa et al., 2006). Transgenic overexpression of TNFα in astrocytes, but not neurons, results in demyelination and selective OL apoptosis (Akassoglou et al., 1998). It is not clear whether preOLs are damaged in these transgenic mice during early development. Other in vivo studies also revealed critical roles of the TNF/TNFR signaling pathway in CNS inflammation and white matter degeneration (Akassoglou et al., 1997; Probert et al., 2000). Our data demonstrate that TNFα/TNFR signaling is necessary for LPS-initiated death of preOLs in mixed glial cultures. The cellular source for LPS-induced TNFα production in mixed glia was not defined in the current study, but activated microglia are most likely responsible for the LPS-induced initial TNFα production given our observation that microglia respond robustly to LPS and produce NO and TNFα in culture, whereas astrocytes respond poorly (J. Li and J. Peng, unpublished observation). However, astrocytes, microglia, and preOLs all express TNFα receptors (Dopp et al., 1997) and thus all are capable of engaging in TNFα/TNFR signaling. Therefore, TNFα released from activated microglia may activate astroglial TNFα receptors, resulting in additional production of this cytokine and/or other toxic factors and amplification of microglial responses to LPS. Interestingly, TNFα itself had minimal effect on preOL viability in preOL monocultures, suggesting a non-cell-autonomous cell death pathway. The actual mediator(s) regulated by TNFα signaling and responsible for preOL death in mixed glial cultures remains to be identified.
Our data do not support a role for iNOS in LPS-induced preOL death in mixed glial cultures. However, this does not necessarily mean that reactive nitrogen species, in particular peroxynitrite, play no role in white matter injury in vivo. Immunoreactive nitrotyrosine, a footprint of peroxynitrite formation, is found in astrocytes and preOLs in the diffuse component of human PVL lesions (Haynes et al., 2003). Interestingly, morphologically identified microglia/macrophages within the subacute necrotic foci, but not in the diffuse lesions, are immunostained positively for nitrotyrosine (Haynes et al., 2003), indicating that a robust and significant burst of oxidative/nitrative stress may be present in the necrotic foci. In culture, peroxynitrite is highly toxic to preOLs and is indeed the molecule directly responsible for the death of preOLs triggered by acutely activated microglia (Li et al., 2005). However, the presence of astrocytes switches the death mechanism from a peroxynitrite-dependent to a TNFα-dependent pathway. These results have several important implications. First, peroxynitrite generated by activated microglia/macrophages within the necrotic foci in PVL may play a role there in direct killing of preOLs. Conversely, in the diffuse white matter lesion, which is populated by abundant reactive astrocytes as well as microglia, a different mechanism of toxicity, such as that mediated by TNFα signaling, may be more important. Second, understanding how astrocytes communicate with activated microglia and influence the mechanism of toxicity and identifying TNFα-regulated mediators of preOL injury should provide us with mechanistic insights that can be exploited to augment protective signals while suppressing deleterious signals. It should be noted that the role of reactive nitrogen species in neonatal white matter injury has yet to be established in animal models. Blocking iNOS may have only a limited beneficial effect. In fact, in inflammatory demyelination models of multiple sclerosis, ablation of the iNOS gene actually exacerbates OL injury and clinical symptoms (Fenyk-Melody et al., 1998; Sahrbacher et al., 1998). The role of iNOS and its product NO in neonatal white matter injury in vivo therefore remains unknown. Multiple pathogenic mechanisms of preOL destruction are likely to exist. Combinatorial approaches such as blocking TNFα signaling and NO production may prove beneficial in preventing white matter injury.
In summary, this study highlights the importance of both astrocytes and microglia in mediating and regulating injury to oligodendrocyte precursors and identifies a distinct cellular mechanism by which endotoxin-activated microglia kill preOLs in an environment in which all three types of CNS glial cells interact. Although peroxynitrite is upregulated, its toxicity to preOL is blunted by astrocytes. Instead, activation of TNFα/TNFR signaling results in preOL death when astrocytes are present. Our study provides new mechanistic insights into inflammatory injury to preOLs and underscores the necessity to consider cell–cell interactions when developing new strategies for the prevention and treatment of white matter injury.
This work was supported in part by National Institutes of Health Grants P01NS38475 (J.J.V., P.A.R.) and NS060017 (J.L.); by grants from the National Multiple Sclerosis Society, the Hearst Foundation, the United Cerebral Palsy Foundation, and the Priscilla and Richard Hunt Fellowship (J.L); by the startup fund from Texas A & M University (J.L.); and by a National Institutes of Health Developmental Disability Research Center Grant HD18655 (to Children's Hospital). We thank Ling Dong and Leon Massillon for technical assistance.
- Correspondence should be addressed to either of the following: Dr. Paul A. Rosenberg, Department of Neurology, Children's Hospital, 300 Longwood Avenue, Boston, MA 02115, ; or Dr. Jianrong Li at her present address, Department of Veterinary Integrative Biosciences, Texas A & M University, Mail Stop 4458, College Station, TX 77843,