Acute Lipopolysaccharide-Mediated Injury in Neonatal White Matter Glia: Role of TNF-, IL-1, and Calcium1
http://www.100md.com
免疫学杂志 2005年第13期
Abstract
Bacterial infection is implicated in the selective CNS white matter injury associated with cerebral palsy, a common birth disorder. Exposure to the bacterial endotoxin LPS produced death of white matter glial cells in isolated neonatal rat optic nerve (RON) (a model white matter tract), over a 180-min time course. A delayed intracellular Ca2+ concentration ([Ca2+]i) rise preceded cell death and both events were prevented by removing extracellular Ca2+. The cytokines TNF- or IL-1, but not IL-6, mimicked the cytotoxic effect of LPS, whereas blocking either TNF- with a neutralizing Ab or IL-1 with recombinant antagonist prevented LPS cytotoxicity. Ultrastructural examination showed wide-scale oligodendroglial cell death in LPS-treated rat optic nerves, with preservation of astrocytes and axons. Fluorescently conjugated LPS revealed LPS binding on microglia and astrocytes in neonatal white and gray matter. Astrocyte binding predominated, and was particularly intense around blood vessels. LPS can therefore bind directly to developing white matter astrocytes and microglia to evoke rapid cell death in neighboring oligodendroglia via a calcium- and cytokine-mediated pathway. In addition to direct toxicity, LPS increased the degree of acute cell death evoked by ischemia in a calcium-dependent manner.
Introduction
Cerebral palsy is a common human birth disorder manifest in the first years of life, which continues unabated throughout the lifetime of the patient. There is currently no effective remedy or prophylactic strategy (1, 2, 3). In the majority of cases, cerebral palsy involves a lesion centered upon the developing white matter adjacent to the ventricles (periventricular leukomalacia (PVL) 3). PVL arises in mid-gestation at a point when myelination is initiating (4). Injury of white matter glial cells is prevalent and may involve selective loss of oligodendroglia with subsequent hypomyelination (5, 6, 7).
An increasing number of studies have implicated intrauterine infection and chorioamnionitis in the genesis of PVL. Early reports demonstrated that i.v. injection of the bacterial endotoxin LPS can produce selective white matter injury in neonatal CNS (8, 9). Recent studies have shown that direct LPS injection into the developing brain can evoke white matter damage (10, 11), whereas induction of intrauterine infection can produce diffuse glial cell death and cavitation in fetal white matter (12, 13). The mechanism of LPS cytotoxicity in developing white matter appears to involve cell-cell interactions. For example, cultured oligodendroglia die when exposed to LPS provided that other glial types are present (14, 15, 16, 17), whereas media collected from LPS-treated microglia or astrocytes is toxic to oligodendroglia in culture (15, 18). A recent report suggests that microglia are the only neural cell type to express the LPS receptor TLR-4, although this has not been examined in vivo (17).
LPS can promote the release of cytokines from glia, whereas cytokine production is implicated in the association between intrauterine infection, preterm birth and neonatal brain damage (see Discussion). However, because i.v. injection of LPS produces systemic hypotension (19), whereas chorioamnionitis is associated with hypotension in the fetus (20), it is uncertain whether bacterial infection produces white matter injury as a direct effect, or subsequent to ischemia associated with hypotension. Experimental studies have repeatedly demonstrated that immature white matter is predisposed to ischemia during hypotension, and is highly sensitive to ischemic injury (1). The immature oligodendroglia that populate white matter at this age are exquisitely sensitive to ischemic injury (21, 22). The current study examines LPS-mediated cytotoxicity in glial cells in isolated white matter, to establish the potential for direct LPS-mediated white matter injury independent of hemodynamic factors. The mechanism and characteristics of direct LPS cytotoxicity are investigated and the pattern of glial cell binding of LPS examined in situ.
Materials and Methods
Optic nerves were dissected from P8-P12 (called "P10" throughout) Lister hooded rats and placed in artificial cerebrospinal fluid (aCSF) composed (in mM) 153 Na+, 3 K+, 2 Mg2+, 2 Ca2+, 131 Cl–, 26 HCO3–, 2 H2PO4–, 10 dextrose, bubbled with 95% O2/5% CO2. All procedures involving the use of animals were approved by local ethical review. At this age, the rat optic nerve (RON) is initiating the process of myelination, with the first wraps of myelin appearing (23, 24). This is the same developmental point at which periventricular white matter is subject to PVL in the fetus (4, 25).
Bacterial LPS (from Escherichia coli; Sigma-Aldrich) was dissolved in aCSF immediately before the experiment to give a final concentration of 1 μg/ml. Ischemia was induced by changing from aCSF to perfusion with zero-glucose aCSF that had been bubbled with 95% N2/5% CO2 for at least 60 min. The atmosphere in the recording chamber was switched simultaneously to 95% N2/5% CO2 (see Ref. 26 for further details). Ischemia was maintained for 60 min, at which point normal conditions were re-established for a further 15 min. Zero-calcium solutions were made up by excluding Ca2+ from the composition of aCSF and adding the calcium chelator EGTA at a concentration of 50 μM. IL-1, TNF-, IL-6, anti-TNF- (used at 10 μg/ml), and IL-1 receptor antagonist (IL-1ra; used at 5 μg/ml) were purchased from R&D Systems. Catalase (250 U/ml) and 4-hydroxy-2,2,6,6-tetramethylpiperidinyloxy (TEMPO; 500 μM) were purchased from Calbiochem. Unless otherwise stated, all other chemicals were purchased from Sigma-Aldrich. Data are presented as the mean ± SEM, with statistical significance determined by ANOVA with the Tukey posttest (PRISM; GraphPad).
Intracellular Ca2+ concentration ([Ca2+]i) imaging
The low affinity, Ca2+-sensitive dye Fura-FF was used for Ca2+ imaging to limit the extent of Ca2+-buffering. 1 mM Fura-FF AM (Molecular Probes) stock solution was made in dry DMSO and 10% pluronic acid. RONs were incubated for 50 min at room temperature in aCSF containing 10 μM FURA-FF AM. RONs were maintained in hydrated 95% O2/5% CO2 atmosphere during the incubation period and were washed in aCSF before being mounted in the perfusion chamber. The ends of the optic nerves were fixed to a 22 x 40 mm glass coverslip with small amounts of cyanoacrylate glue, leaving the majority of the nerve completely free of glue (26). The coverslip was sealed onto a Plexiglas perfusion chamber (atmosphere chamber; Warner Instruments) with silicone grease. The aCSF was perfused through the chamber at a rate of 2–3 ml/min, with a fluid level of 2 mm completely covering the RON. 95% O2/5% CO2 was blown over the aCSF at a rate of 0.3 L/min. The chamber was mounted on the stage of an Nikon Eclipse TE200 inverted epifluorescence microscope. Chamber temperature was maintained closely at 37°C with a flow-through feedback tubing heater (Warner Instruments) positioned immediately before the aCSF entered the chamber, and a feedback objective heater (Bioptechs) that warmed the objective to 37°C. This combination of heating systems regulated the temperature of the bath and cover slip to 37°C, as established periodically with a temperature probe.
Cells within the RON were illuminated at 340, 360, and 380 nm by Monochromator (Optoscan; Cairn Research), and images were collected at 520 nm using an appropriate filter set (Chroma Technology). Images were taken with a cooled CCD camera (CoolSNAP HQ; Roper Scientific) every 60 s. This low recording frequency limited any damaging effect of illumination on the cells and reduced dye bleaching over the long recording times that were used. Changes in 340:380 ratio were taken to indicate changes in [Ca2+]i (27). The 360 intensity (isosbestic point) was monitored to assess the capacity of cells to retain dye. The sudden loss of 360 signal correlated with loss of cell membrane integrity and the release of dye into the extracellular space (see Ref. 26). Data were collected and stored with the image acquisition program Metafluor (Universal Imaging) running on Windows XP. Because of slow shifting of the preparation, cells were occasionally refocused during experiments. A small proportion of cells slowly drifted out of the focal plane and were discarded. Many of the experiments required long recording periods (190 min), which had a relatively low success rate due to dye fading and/or shifting of the preparation.
Dead cell counts
Ethidium bromide, a lipophobic nuclear dye that does not stain intact living cells, was used to stain nuclei of dead cells. RONs were dissected and placed in aCSF. The nerves were incubated for 180 min at 37°C in 1 ml of an experiment solution. RONs were maintained in hydrated 95% O2/5% CO2 atmosphere during the incubation period and were subsequently washed with fresh aCSF. RONs were then incubated for 10 min at room temperature with 0.5 nM ethidium bromide. RONs were then washed in aCSF and live-mounted on a Nikon LABOPHOT-2 epifluorescence microscope equipped with a 20x fluorescence objective and appropriate filter. Whole RON images (excluding the cut ends) were captured with a CCD camera, averaging 16 frames. Cell counting was performed blind using ImageJ (NIH). The mean number of dead cells found following exposure to any given condition was compared with the number seen in control RONs (180 min aCSF only), and are expressed as a percentage change relative to this level of control cell death.
LPS binding/immunohistochemistry
Whole P10 rat brains were dissected free, and RONs were dissected free and left on the optic chiasm, before fixation in 3% paraformaldehyde for 30 min at room temperature and mounting in Tissue Tec (Sakura). Sections (20 μm) were cut with a cryostat before mounting on slides. Fluorescently conjugated LPS, (LPS-Alexa Fluor 488, E. coli; Molecular Probes), was used to detect LPS binding (1 μg/ml in PBS). Sections were blocked for 60 min at room temperature with PBS containing 10% goat serum and 0.5% Triton X-100 before incubation in conjugated LPS for 120 min at room temperature. Sections were either simultaneously costained with monoclonal anti-glial fibrillary acidic protein (GFAP) conjugated to Cy3 (Sigma-Aldrich; 1:500 in PBS), or stained overnight at 5°C with isolectin IB4 conjugated to Alexa-Fluor 594 (Molecular Probes; 1:100 in PBS). Sections were washed briefly after incubation in LPS-conjugate before fixation with 4% formaldehyde for 30 min in PBS. Slices were then washed in PBS (2 x 15 min) and mounted on an Olympus IX70 inverted confocal microscope for image collection. Conjugated probes were used to eliminate cross-reactions and the specificity of LPS-conjugate binding was tested by displacement with unconjugated LPS (see Results). No bleed through of fluorescence was detected between channels.
Electron microscopy
RONs were exposed to 1 μg/ml LPS in aCSF or control aCSF for 180 min before washing in Sorenson’s buffer and postfixation in 3% glutaraldehyde/Sorenson’s for 90 min at room temperature. The RONs were subsequently postfixed with 2% osmium tetroxide and dehydrated in ethanol and propylene oxide. The nerves were infiltrated in epoxy resin and the ends cut back before taking ultrathin sections. Sections were counterstained with uranyl acetate and lead citrate and examined with a JEOL 100CX electron microscope. To avoid bias in the data, electron micrographs were collected blind (by R. Fern) to the experimental procedure used to produce each sample.
Results
The P10 RON is at a similar developmental point to the human mid-term periventricular white matter that is subject to PVL (24, 25, 28). Ultrastructural examination of control RONs showed the typical components previously reported at this age: premyelinated axons and some early myelinated axons; astrocytes; and cells of the oligodendroglial lineage. Immature oligodendroglial cells could be reliably identified by their characteristic nuclear morphology, narrow-bore endoplasmic reticulum and the presence of myelinated axons within the cell body or processes (29) (Fig. 1, A and B).
RONs subjected to 180 min of LPS treatment showed wide scale injury of immature oligodendrocytes (Fig. 2, C and D). Nuclear morphology appeared normal in these cells but mitochondria were swollen or ruptured and areas where the cell membrane had been lost could be identified in many cases. Similar changes were apparent in immature oligodendrocyte processes, although the degree of injury was less marked and cell membrane breakdown was not observed (Fig. 2E). No pathology was evident in axons, even those that were being ensheathed by immature oligodendrocytes that had died (Fig. 2C). Astrocytes were also largely unaffected in post-LPS treated RON (Fig. 2F).
The cytotoxic effect of LPS was confirmed by counting dead cells in P10 RONs following a standard 180 min exposure to 1 μg/ml LPS. RONs maintained in the presence of LPS had a 68.0 ± 10.2% increase in the number of dead cells compared with control (p < 0.001; Fig. 3). The cytotoxic effect of LPS was blocked by removing extracellular Ca2+ (+50 μM EGTA), or by addition of the reactive oxygen species (ROS) scavengers TEMPO + catalase (30) (Fig. 3). Neither zero Ca2+ nor ROS scavengers significantly affected the number of dead cells in the absence of LPS.
Cytokines are implicated as intermediaries in various bacterial infection models of white matter injury. Exposure for 180 min to the cytokines TNF- (1,000 U/ml) or IL-1 (1,000 U/ml), but not IL-6 (1,000 U/ml), mimicked the cytotoxic effect of LPS (1 μg/ml) (Fig. 4). Exposure to all three cytokines (with IL-1 increased to 25,000 U/ml) had no additive cytotoxic effect (Fig. 4). Removing the biological activity of TNF- by incubation in an inactivating Ab (10 μg/ml) abolished the cytotoxic action of LPS, as did incubation with an antagonist of IL-1, IL-1ra (5 μg/ml). Although 120 min exposure to IL-1 was toxic, 120 min of IL-1 was not toxic.
To probe the site of action of LPS, conjugated LPS binding was examined in P10 RON and brain. LPS binding was widespread throughout P10 RON, colocalizing with GFAP+ astrocyte somata and processes (Fig. 5, A–C). Conjugated LPS binding was displaced by coincubation with nonconjugated LPS (10 μg/ml) (Fig. 5, D and E), proving the selectivity of the binding. Microglial labeling with I-B4 revealed a sparse population of I-B4+ cells that bound conjugated LPS (Fig. 5, F–H). These cells were rare, and the great majority of LPS binding cells were I-B4–. Widespread binding of conjugated LPS was also found in P10 brain slices, again largely colocalized with GFAP+ astrocytes. Binding in brain slices was particularly intense on astrocytes surrounding blood vessels (Fig. 5, I–K).
Several reports suggest that bacterial and ischemic injury may be additive in developing white matter (31). Consistent with our previous results, ischemic conditions evoked [Ca2+]i rises associated with cell death in FURA-FF loaded P10 RON glia (Fig. 6, A and B), (29, 32). Qualitatively similar events were seen during ischemia in the presence of LPS (1 μg/ml) (Fig. 6, C and D), and the degree of cell death produced by ischemia was significantly higher in the presence of LPS (1 μg/ml) (Fig. 7).
We have recently reported that removing extracellular Ca2+ is protective against oligodendroglial death, and increases astrocyte death during acute ischemia in the P10 RON (29, 32). The overall level of glial death during ischemia was significantly reduced by removing extracellular Ca2+ (p < 0.01), and the increase in cell death during ischemia produced by LPS was absent in zero-Ca2+ (Fig. 7). Unlike for the LPS-mediated cell death produced by 180 min of exposure to LPS, the additive effect of LPS upon ischemic cell death was not reduced by the ROS scavengers TEMPO + catalase (Fig. 7).
Discussion
Induction of maternal bacterial infection in rabbits can reproduce a focal and diffuse white matter injury in fetal brain that has some similarities to PVL (12, 13, 33). Intracerebral or ventricular injection of LPS in neonatal rat can also produce white matter damage (10, 11). The current study demonstrates LPS toxicity in immature white matter oligodendrocytes in a whole mount preparation. LPS is therefore directly and acutely toxic to developing white matter glia, irrespective of vascular factors that may come into play during infection (20). The rapid death of immature oligodendrocytes found in this preparation was not reported following exposure to LPS in cell culture (15, 16), suggesting that interactions between cell types in situ underlies the acute toxicity.
LPS binds to microglia and astrocytes in situ
TLR-4 (+CD14) is the only known binding site for LPS (34). Cell culture studies report that microglia express TLR-4, whereas oligodendrocyte precursors do not (17), and astrocytes either do not (17) or express very low levels unless pre-exposed to LPS (35). LPS binding has not previously been examined in situ. We report extensive LPS binding in neonatal RON and brain. LPS binding was particularly marked around blood vessels. Microglial cells were shown to bind LPS, but the majority of LPS-bindings cells were GFAP+ astrocytes in both RON and brain. The acute cytotoxic effect of LPS upon immature oligodendrocytes is therefore mediated by other glial cell types, presumably mainly astrocytes.
LPS toxicity is delayed and is Ca2+ dependent
The acute cell death evoked by LPS reported in this study was preceded by a delayed sudden increase in [Ca2+]i and was prevented by removing extracellular Ca2+, demonstrating the Ca2+ dependence of the injury. The delayed nature of the [Ca2+]i rise is consistent with LPS having initial effects on astrocytes and/or microglial cells, which then mediate a delayed and lethal rise in oligodendrocyte [Ca2+]i. If this is correct, the initial effects that precede the oligodendrocyte [Ca2+]i rise take between 60 and 120 min and constitute the priming steps in an injury cascade.
The role of cytokines
The incidence of PVL in premature infants is correlated with elevated cytokine levels in umbilical cord blood, amniotic fluid and neonatal blood (33, 36, 37). We report that TNF- and IL-1 directly evoke acute cell death in RON glia, with the degree of cell death not significantly different from that produced by LPS. Although implicated in the inflammatory response to LPS (38), IL-6 was not found to cause acute glial cell death. Combined IL-1, IL-6, and TNF- exposure did not increase the degree of cell death, and it appears that maximal injury was achieved at the cytokine concentrations used in this study.
LPS-mediated cell death was reduced to the level seen in control experiments by the addition of a neutralizing Ab for TNF-, or of IL-1ra. Therefore, production of both cytokines is necessary for the acute cytotoxic effects of LPS. Maternal venous LPS injection elevates fetal brain TNF- and IL-1 mRNA within 60 min (39). In cell culture, LPS can induce TNF- release from microglia within 60–120 min and IL-1 within 120–180 min (the earliest time points measured) (40, 41, 42, 43). Cultured astrocytes are also capable of cytokine release (44), with LPS-induced TNF- release requiring the prior release of IL-1 (45, 46, 47). This is consistent with the current finding that 120 min of exposure to TNF- was toxic but IL-1 required 180 min of exposure for toxicity to become apparent. If astrocytes are the major source of the cytokine release induced by LPS, as the LPS binding data would suggest, this would explain the protection provided by blocking either TNF- or IL-1.
LPS-stimulated production of glial TNF- and IL-1 involves the activation of the transcription factor NF-B, which is evident within 60 min of LPS exposure in cell culture (48). Cytokines and ROS are also strong activators of NF-B in glia (48, 49), suggesting a feed-forward loop involving exposure to LPS, activation of NF-B, cytokine and ROS generation and further activation of NF-B (Fig. 8). This would be consistent with the protective effect of ROS scavengers seen in the results and with studies showing LPS evoked ROS generation in cultured microglia and astrocytes (48, 50, 51, 52, 53).
The toxic effect of TNF- upon oligodendroglia is well characterized (e.g., Refs. 18 and 54), and IL-1 toxicity of oligodendroglia has also been reported (55). It is unclear whether such toxicity is a direct pathway to the glial injury described in the current results, because TNF--mediated oligodendroglial cell death is thought to be slow. However, significant death of cultured rat oligodendroglia has been reported within 120 min of exposure to TNF- (18), and the generally slower oligodendrocyte death seen in other studies may be due to the low sensitivity to TNF--mediated injury in the mature cells used (54). However, a direct contribution of ROS to the oligodendroglial injury cannot be ruled out.
LPS and ischemia
Prior reports suggest that bacterial endotoxin- and ischemia-mediated immature white matter injury is additive (31). Bacterial endotoxins such as LPS evoke hypotension and can be predicted to induce ischemia in fetal white matter in vivo (19, 20). The current study shows additive injury by LPS and ischemia in isolated immature white matter, independent of any hemodynamic factors that may come into play in vivo. This additive effect was Ca2+ dependent; whereas both LPS- and ischemia-induced injury of RON oligodendrocytes is mediated by Ca2+ influx (29). However, the potentiation of ischemic injury by LPS occurred over a shorter time course (60 min) than was required for direct LPS toxicity, and was not affected by ROS scavengers that blocked direct toxicity. Therefore, it may be that interactions between LPS and ischemia involve different pathways than those operating during direct LPS toxicity.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by National Institute of Neurological Disorders and Stroke Grant NS44875 (to R.F.).
2 Address correspondence and reprint requests to Dr. Robert Fern, Department of Cell Physiology and Pharmacology, University of Leicester, P.O. Box 138, University Road, Leicester, U.K. LE1 9HN. E-mail address: RF34@le.ac.uk
3 Abbreviations used in this paper: PVL, periventricular leukomalacia; aCSF, artificial cerebrospinal fluid; [Ca2+]i, intracellular Ca2+ concentration; RON, rat optic nerve; IL-1ra, IL-1 receptor antagonist; ROS, reactive organ species.
Received for publication December 17, 2004. Accepted for publication April 20, 2005.
References
Volpe, J. J.. 1995. Neurology of the newborn W. B. Saunders, Philadelphia. .
Volpe, J. J.. 2001. Neurobiology of periventricular leukomalacia in the premature infant. Pediatr. Res. 50: 553-562.
Kinney, H. C., S. A. Back. 1998. Human oligodendroglial development: relationship to periventricular leukomalacia. Semin. Pediatr. Neurol. 5: 180-189.
Back, S. A., N. L. Luo, N. S. Borenstein, J. M. Levine, J. J. Volpe, H. C. Kinney. 2001. Late oligodendrocyte progenitors coincide with the developmental window of vulnerability for human perinatal white matter injury. J. Neurosci. 21: 1302-1312.
Flodmark, O., B. Lupton, D. Li, G. K. Stimac, E. H. Roland, A. Hill, M. F. Whitfield, M. G. Norman. 1989. MR imaging of periventricular leukomalacia in childhood. AJR Am. J. Roentgenol. 152: 583-590.
van de Bor, M., G. L. Guit, A. M. Schreuder, J. Wondergem, G. J. Vielvoye. 1989. Early detection of delayed myelination in preterm infants. Pediatrics 84: 407-411.
Back, S. A., B. H. Han, N. L. Luo, C. A. Chricton, S. Xanthoudakis, J. Tam, K. L. Arvin, D. M. Holtzman. 2002. Selective vulnerability of late oligodendrocyte progenitors to hypoxia-ischemia. J. Neurosci. 22: 455-463.
Gilles, F. H., A. Leviton, C. S. Kerr. 1976. Endotoxin leucoencephalopathy in the telencephalon of the newborn kitten. J. Neurol. Sci. 27: 183-191.(Catherine Sherwin and Rob)
Bacterial infection is implicated in the selective CNS white matter injury associated with cerebral palsy, a common birth disorder. Exposure to the bacterial endotoxin LPS produced death of white matter glial cells in isolated neonatal rat optic nerve (RON) (a model white matter tract), over a 180-min time course. A delayed intracellular Ca2+ concentration ([Ca2+]i) rise preceded cell death and both events were prevented by removing extracellular Ca2+. The cytokines TNF- or IL-1, but not IL-6, mimicked the cytotoxic effect of LPS, whereas blocking either TNF- with a neutralizing Ab or IL-1 with recombinant antagonist prevented LPS cytotoxicity. Ultrastructural examination showed wide-scale oligodendroglial cell death in LPS-treated rat optic nerves, with preservation of astrocytes and axons. Fluorescently conjugated LPS revealed LPS binding on microglia and astrocytes in neonatal white and gray matter. Astrocyte binding predominated, and was particularly intense around blood vessels. LPS can therefore bind directly to developing white matter astrocytes and microglia to evoke rapid cell death in neighboring oligodendroglia via a calcium- and cytokine-mediated pathway. In addition to direct toxicity, LPS increased the degree of acute cell death evoked by ischemia in a calcium-dependent manner.
Introduction
Cerebral palsy is a common human birth disorder manifest in the first years of life, which continues unabated throughout the lifetime of the patient. There is currently no effective remedy or prophylactic strategy (1, 2, 3). In the majority of cases, cerebral palsy involves a lesion centered upon the developing white matter adjacent to the ventricles (periventricular leukomalacia (PVL) 3). PVL arises in mid-gestation at a point when myelination is initiating (4). Injury of white matter glial cells is prevalent and may involve selective loss of oligodendroglia with subsequent hypomyelination (5, 6, 7).
An increasing number of studies have implicated intrauterine infection and chorioamnionitis in the genesis of PVL. Early reports demonstrated that i.v. injection of the bacterial endotoxin LPS can produce selective white matter injury in neonatal CNS (8, 9). Recent studies have shown that direct LPS injection into the developing brain can evoke white matter damage (10, 11), whereas induction of intrauterine infection can produce diffuse glial cell death and cavitation in fetal white matter (12, 13). The mechanism of LPS cytotoxicity in developing white matter appears to involve cell-cell interactions. For example, cultured oligodendroglia die when exposed to LPS provided that other glial types are present (14, 15, 16, 17), whereas media collected from LPS-treated microglia or astrocytes is toxic to oligodendroglia in culture (15, 18). A recent report suggests that microglia are the only neural cell type to express the LPS receptor TLR-4, although this has not been examined in vivo (17).
LPS can promote the release of cytokines from glia, whereas cytokine production is implicated in the association between intrauterine infection, preterm birth and neonatal brain damage (see Discussion). However, because i.v. injection of LPS produces systemic hypotension (19), whereas chorioamnionitis is associated with hypotension in the fetus (20), it is uncertain whether bacterial infection produces white matter injury as a direct effect, or subsequent to ischemia associated with hypotension. Experimental studies have repeatedly demonstrated that immature white matter is predisposed to ischemia during hypotension, and is highly sensitive to ischemic injury (1). The immature oligodendroglia that populate white matter at this age are exquisitely sensitive to ischemic injury (21, 22). The current study examines LPS-mediated cytotoxicity in glial cells in isolated white matter, to establish the potential for direct LPS-mediated white matter injury independent of hemodynamic factors. The mechanism and characteristics of direct LPS cytotoxicity are investigated and the pattern of glial cell binding of LPS examined in situ.
Materials and Methods
Optic nerves were dissected from P8-P12 (called "P10" throughout) Lister hooded rats and placed in artificial cerebrospinal fluid (aCSF) composed (in mM) 153 Na+, 3 K+, 2 Mg2+, 2 Ca2+, 131 Cl–, 26 HCO3–, 2 H2PO4–, 10 dextrose, bubbled with 95% O2/5% CO2. All procedures involving the use of animals were approved by local ethical review. At this age, the rat optic nerve (RON) is initiating the process of myelination, with the first wraps of myelin appearing (23, 24). This is the same developmental point at which periventricular white matter is subject to PVL in the fetus (4, 25).
Bacterial LPS (from Escherichia coli; Sigma-Aldrich) was dissolved in aCSF immediately before the experiment to give a final concentration of 1 μg/ml. Ischemia was induced by changing from aCSF to perfusion with zero-glucose aCSF that had been bubbled with 95% N2/5% CO2 for at least 60 min. The atmosphere in the recording chamber was switched simultaneously to 95% N2/5% CO2 (see Ref. 26 for further details). Ischemia was maintained for 60 min, at which point normal conditions were re-established for a further 15 min. Zero-calcium solutions were made up by excluding Ca2+ from the composition of aCSF and adding the calcium chelator EGTA at a concentration of 50 μM. IL-1, TNF-, IL-6, anti-TNF- (used at 10 μg/ml), and IL-1 receptor antagonist (IL-1ra; used at 5 μg/ml) were purchased from R&D Systems. Catalase (250 U/ml) and 4-hydroxy-2,2,6,6-tetramethylpiperidinyloxy (TEMPO; 500 μM) were purchased from Calbiochem. Unless otherwise stated, all other chemicals were purchased from Sigma-Aldrich. Data are presented as the mean ± SEM, with statistical significance determined by ANOVA with the Tukey posttest (PRISM; GraphPad).
Intracellular Ca2+ concentration ([Ca2+]i) imaging
The low affinity, Ca2+-sensitive dye Fura-FF was used for Ca2+ imaging to limit the extent of Ca2+-buffering. 1 mM Fura-FF AM (Molecular Probes) stock solution was made in dry DMSO and 10% pluronic acid. RONs were incubated for 50 min at room temperature in aCSF containing 10 μM FURA-FF AM. RONs were maintained in hydrated 95% O2/5% CO2 atmosphere during the incubation period and were washed in aCSF before being mounted in the perfusion chamber. The ends of the optic nerves were fixed to a 22 x 40 mm glass coverslip with small amounts of cyanoacrylate glue, leaving the majority of the nerve completely free of glue (26). The coverslip was sealed onto a Plexiglas perfusion chamber (atmosphere chamber; Warner Instruments) with silicone grease. The aCSF was perfused through the chamber at a rate of 2–3 ml/min, with a fluid level of 2 mm completely covering the RON. 95% O2/5% CO2 was blown over the aCSF at a rate of 0.3 L/min. The chamber was mounted on the stage of an Nikon Eclipse TE200 inverted epifluorescence microscope. Chamber temperature was maintained closely at 37°C with a flow-through feedback tubing heater (Warner Instruments) positioned immediately before the aCSF entered the chamber, and a feedback objective heater (Bioptechs) that warmed the objective to 37°C. This combination of heating systems regulated the temperature of the bath and cover slip to 37°C, as established periodically with a temperature probe.
Cells within the RON were illuminated at 340, 360, and 380 nm by Monochromator (Optoscan; Cairn Research), and images were collected at 520 nm using an appropriate filter set (Chroma Technology). Images were taken with a cooled CCD camera (CoolSNAP HQ; Roper Scientific) every 60 s. This low recording frequency limited any damaging effect of illumination on the cells and reduced dye bleaching over the long recording times that were used. Changes in 340:380 ratio were taken to indicate changes in [Ca2+]i (27). The 360 intensity (isosbestic point) was monitored to assess the capacity of cells to retain dye. The sudden loss of 360 signal correlated with loss of cell membrane integrity and the release of dye into the extracellular space (see Ref. 26). Data were collected and stored with the image acquisition program Metafluor (Universal Imaging) running on Windows XP. Because of slow shifting of the preparation, cells were occasionally refocused during experiments. A small proportion of cells slowly drifted out of the focal plane and were discarded. Many of the experiments required long recording periods (190 min), which had a relatively low success rate due to dye fading and/or shifting of the preparation.
Dead cell counts
Ethidium bromide, a lipophobic nuclear dye that does not stain intact living cells, was used to stain nuclei of dead cells. RONs were dissected and placed in aCSF. The nerves were incubated for 180 min at 37°C in 1 ml of an experiment solution. RONs were maintained in hydrated 95% O2/5% CO2 atmosphere during the incubation period and were subsequently washed with fresh aCSF. RONs were then incubated for 10 min at room temperature with 0.5 nM ethidium bromide. RONs were then washed in aCSF and live-mounted on a Nikon LABOPHOT-2 epifluorescence microscope equipped with a 20x fluorescence objective and appropriate filter. Whole RON images (excluding the cut ends) were captured with a CCD camera, averaging 16 frames. Cell counting was performed blind using ImageJ (NIH). The mean number of dead cells found following exposure to any given condition was compared with the number seen in control RONs (180 min aCSF only), and are expressed as a percentage change relative to this level of control cell death.
LPS binding/immunohistochemistry
Whole P10 rat brains were dissected free, and RONs were dissected free and left on the optic chiasm, before fixation in 3% paraformaldehyde for 30 min at room temperature and mounting in Tissue Tec (Sakura). Sections (20 μm) were cut with a cryostat before mounting on slides. Fluorescently conjugated LPS, (LPS-Alexa Fluor 488, E. coli; Molecular Probes), was used to detect LPS binding (1 μg/ml in PBS). Sections were blocked for 60 min at room temperature with PBS containing 10% goat serum and 0.5% Triton X-100 before incubation in conjugated LPS for 120 min at room temperature. Sections were either simultaneously costained with monoclonal anti-glial fibrillary acidic protein (GFAP) conjugated to Cy3 (Sigma-Aldrich; 1:500 in PBS), or stained overnight at 5°C with isolectin IB4 conjugated to Alexa-Fluor 594 (Molecular Probes; 1:100 in PBS). Sections were washed briefly after incubation in LPS-conjugate before fixation with 4% formaldehyde for 30 min in PBS. Slices were then washed in PBS (2 x 15 min) and mounted on an Olympus IX70 inverted confocal microscope for image collection. Conjugated probes were used to eliminate cross-reactions and the specificity of LPS-conjugate binding was tested by displacement with unconjugated LPS (see Results). No bleed through of fluorescence was detected between channels.
Electron microscopy
RONs were exposed to 1 μg/ml LPS in aCSF or control aCSF for 180 min before washing in Sorenson’s buffer and postfixation in 3% glutaraldehyde/Sorenson’s for 90 min at room temperature. The RONs were subsequently postfixed with 2% osmium tetroxide and dehydrated in ethanol and propylene oxide. The nerves were infiltrated in epoxy resin and the ends cut back before taking ultrathin sections. Sections were counterstained with uranyl acetate and lead citrate and examined with a JEOL 100CX electron microscope. To avoid bias in the data, electron micrographs were collected blind (by R. Fern) to the experimental procedure used to produce each sample.
Results
The P10 RON is at a similar developmental point to the human mid-term periventricular white matter that is subject to PVL (24, 25, 28). Ultrastructural examination of control RONs showed the typical components previously reported at this age: premyelinated axons and some early myelinated axons; astrocytes; and cells of the oligodendroglial lineage. Immature oligodendroglial cells could be reliably identified by their characteristic nuclear morphology, narrow-bore endoplasmic reticulum and the presence of myelinated axons within the cell body or processes (29) (Fig. 1, A and B).
RONs subjected to 180 min of LPS treatment showed wide scale injury of immature oligodendrocytes (Fig. 2, C and D). Nuclear morphology appeared normal in these cells but mitochondria were swollen or ruptured and areas where the cell membrane had been lost could be identified in many cases. Similar changes were apparent in immature oligodendrocyte processes, although the degree of injury was less marked and cell membrane breakdown was not observed (Fig. 2E). No pathology was evident in axons, even those that were being ensheathed by immature oligodendrocytes that had died (Fig. 2C). Astrocytes were also largely unaffected in post-LPS treated RON (Fig. 2F).
The cytotoxic effect of LPS was confirmed by counting dead cells in P10 RONs following a standard 180 min exposure to 1 μg/ml LPS. RONs maintained in the presence of LPS had a 68.0 ± 10.2% increase in the number of dead cells compared with control (p < 0.001; Fig. 3). The cytotoxic effect of LPS was blocked by removing extracellular Ca2+ (+50 μM EGTA), or by addition of the reactive oxygen species (ROS) scavengers TEMPO + catalase (30) (Fig. 3). Neither zero Ca2+ nor ROS scavengers significantly affected the number of dead cells in the absence of LPS.
Cytokines are implicated as intermediaries in various bacterial infection models of white matter injury. Exposure for 180 min to the cytokines TNF- (1,000 U/ml) or IL-1 (1,000 U/ml), but not IL-6 (1,000 U/ml), mimicked the cytotoxic effect of LPS (1 μg/ml) (Fig. 4). Exposure to all three cytokines (with IL-1 increased to 25,000 U/ml) had no additive cytotoxic effect (Fig. 4). Removing the biological activity of TNF- by incubation in an inactivating Ab (10 μg/ml) abolished the cytotoxic action of LPS, as did incubation with an antagonist of IL-1, IL-1ra (5 μg/ml). Although 120 min exposure to IL-1 was toxic, 120 min of IL-1 was not toxic.
To probe the site of action of LPS, conjugated LPS binding was examined in P10 RON and brain. LPS binding was widespread throughout P10 RON, colocalizing with GFAP+ astrocyte somata and processes (Fig. 5, A–C). Conjugated LPS binding was displaced by coincubation with nonconjugated LPS (10 μg/ml) (Fig. 5, D and E), proving the selectivity of the binding. Microglial labeling with I-B4 revealed a sparse population of I-B4+ cells that bound conjugated LPS (Fig. 5, F–H). These cells were rare, and the great majority of LPS binding cells were I-B4–. Widespread binding of conjugated LPS was also found in P10 brain slices, again largely colocalized with GFAP+ astrocytes. Binding in brain slices was particularly intense on astrocytes surrounding blood vessels (Fig. 5, I–K).
Several reports suggest that bacterial and ischemic injury may be additive in developing white matter (31). Consistent with our previous results, ischemic conditions evoked [Ca2+]i rises associated with cell death in FURA-FF loaded P10 RON glia (Fig. 6, A and B), (29, 32). Qualitatively similar events were seen during ischemia in the presence of LPS (1 μg/ml) (Fig. 6, C and D), and the degree of cell death produced by ischemia was significantly higher in the presence of LPS (1 μg/ml) (Fig. 7).
We have recently reported that removing extracellular Ca2+ is protective against oligodendroglial death, and increases astrocyte death during acute ischemia in the P10 RON (29, 32). The overall level of glial death during ischemia was significantly reduced by removing extracellular Ca2+ (p < 0.01), and the increase in cell death during ischemia produced by LPS was absent in zero-Ca2+ (Fig. 7). Unlike for the LPS-mediated cell death produced by 180 min of exposure to LPS, the additive effect of LPS upon ischemic cell death was not reduced by the ROS scavengers TEMPO + catalase (Fig. 7).
Discussion
Induction of maternal bacterial infection in rabbits can reproduce a focal and diffuse white matter injury in fetal brain that has some similarities to PVL (12, 13, 33). Intracerebral or ventricular injection of LPS in neonatal rat can also produce white matter damage (10, 11). The current study demonstrates LPS toxicity in immature white matter oligodendrocytes in a whole mount preparation. LPS is therefore directly and acutely toxic to developing white matter glia, irrespective of vascular factors that may come into play during infection (20). The rapid death of immature oligodendrocytes found in this preparation was not reported following exposure to LPS in cell culture (15, 16), suggesting that interactions between cell types in situ underlies the acute toxicity.
LPS binds to microglia and astrocytes in situ
TLR-4 (+CD14) is the only known binding site for LPS (34). Cell culture studies report that microglia express TLR-4, whereas oligodendrocyte precursors do not (17), and astrocytes either do not (17) or express very low levels unless pre-exposed to LPS (35). LPS binding has not previously been examined in situ. We report extensive LPS binding in neonatal RON and brain. LPS binding was particularly marked around blood vessels. Microglial cells were shown to bind LPS, but the majority of LPS-bindings cells were GFAP+ astrocytes in both RON and brain. The acute cytotoxic effect of LPS upon immature oligodendrocytes is therefore mediated by other glial cell types, presumably mainly astrocytes.
LPS toxicity is delayed and is Ca2+ dependent
The acute cell death evoked by LPS reported in this study was preceded by a delayed sudden increase in [Ca2+]i and was prevented by removing extracellular Ca2+, demonstrating the Ca2+ dependence of the injury. The delayed nature of the [Ca2+]i rise is consistent with LPS having initial effects on astrocytes and/or microglial cells, which then mediate a delayed and lethal rise in oligodendrocyte [Ca2+]i. If this is correct, the initial effects that precede the oligodendrocyte [Ca2+]i rise take between 60 and 120 min and constitute the priming steps in an injury cascade.
The role of cytokines
The incidence of PVL in premature infants is correlated with elevated cytokine levels in umbilical cord blood, amniotic fluid and neonatal blood (33, 36, 37). We report that TNF- and IL-1 directly evoke acute cell death in RON glia, with the degree of cell death not significantly different from that produced by LPS. Although implicated in the inflammatory response to LPS (38), IL-6 was not found to cause acute glial cell death. Combined IL-1, IL-6, and TNF- exposure did not increase the degree of cell death, and it appears that maximal injury was achieved at the cytokine concentrations used in this study.
LPS-mediated cell death was reduced to the level seen in control experiments by the addition of a neutralizing Ab for TNF-, or of IL-1ra. Therefore, production of both cytokines is necessary for the acute cytotoxic effects of LPS. Maternal venous LPS injection elevates fetal brain TNF- and IL-1 mRNA within 60 min (39). In cell culture, LPS can induce TNF- release from microglia within 60–120 min and IL-1 within 120–180 min (the earliest time points measured) (40, 41, 42, 43). Cultured astrocytes are also capable of cytokine release (44), with LPS-induced TNF- release requiring the prior release of IL-1 (45, 46, 47). This is consistent with the current finding that 120 min of exposure to TNF- was toxic but IL-1 required 180 min of exposure for toxicity to become apparent. If astrocytes are the major source of the cytokine release induced by LPS, as the LPS binding data would suggest, this would explain the protection provided by blocking either TNF- or IL-1.
LPS-stimulated production of glial TNF- and IL-1 involves the activation of the transcription factor NF-B, which is evident within 60 min of LPS exposure in cell culture (48). Cytokines and ROS are also strong activators of NF-B in glia (48, 49), suggesting a feed-forward loop involving exposure to LPS, activation of NF-B, cytokine and ROS generation and further activation of NF-B (Fig. 8). This would be consistent with the protective effect of ROS scavengers seen in the results and with studies showing LPS evoked ROS generation in cultured microglia and astrocytes (48, 50, 51, 52, 53).
The toxic effect of TNF- upon oligodendroglia is well characterized (e.g., Refs. 18 and 54), and IL-1 toxicity of oligodendroglia has also been reported (55). It is unclear whether such toxicity is a direct pathway to the glial injury described in the current results, because TNF--mediated oligodendroglial cell death is thought to be slow. However, significant death of cultured rat oligodendroglia has been reported within 120 min of exposure to TNF- (18), and the generally slower oligodendrocyte death seen in other studies may be due to the low sensitivity to TNF--mediated injury in the mature cells used (54). However, a direct contribution of ROS to the oligodendroglial injury cannot be ruled out.
LPS and ischemia
Prior reports suggest that bacterial endotoxin- and ischemia-mediated immature white matter injury is additive (31). Bacterial endotoxins such as LPS evoke hypotension and can be predicted to induce ischemia in fetal white matter in vivo (19, 20). The current study shows additive injury by LPS and ischemia in isolated immature white matter, independent of any hemodynamic factors that may come into play in vivo. This additive effect was Ca2+ dependent; whereas both LPS- and ischemia-induced injury of RON oligodendrocytes is mediated by Ca2+ influx (29). However, the potentiation of ischemic injury by LPS occurred over a shorter time course (60 min) than was required for direct LPS toxicity, and was not affected by ROS scavengers that blocked direct toxicity. Therefore, it may be that interactions between LPS and ischemia involve different pathways than those operating during direct LPS toxicity.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by National Institute of Neurological Disorders and Stroke Grant NS44875 (to R.F.).
2 Address correspondence and reprint requests to Dr. Robert Fern, Department of Cell Physiology and Pharmacology, University of Leicester, P.O. Box 138, University Road, Leicester, U.K. LE1 9HN. E-mail address: RF34@le.ac.uk
3 Abbreviations used in this paper: PVL, periventricular leukomalacia; aCSF, artificial cerebrospinal fluid; [Ca2+]i, intracellular Ca2+ concentration; RON, rat optic nerve; IL-1ra, IL-1 receptor antagonist; ROS, reactive organ species.
Received for publication December 17, 2004. Accepted for publication April 20, 2005.
References
Volpe, J. J.. 1995. Neurology of the newborn W. B. Saunders, Philadelphia. .
Volpe, J. J.. 2001. Neurobiology of periventricular leukomalacia in the premature infant. Pediatr. Res. 50: 553-562.
Kinney, H. C., S. A. Back. 1998. Human oligodendroglial development: relationship to periventricular leukomalacia. Semin. Pediatr. Neurol. 5: 180-189.
Back, S. A., N. L. Luo, N. S. Borenstein, J. M. Levine, J. J. Volpe, H. C. Kinney. 2001. Late oligodendrocyte progenitors coincide with the developmental window of vulnerability for human perinatal white matter injury. J. Neurosci. 21: 1302-1312.
Flodmark, O., B. Lupton, D. Li, G. K. Stimac, E. H. Roland, A. Hill, M. F. Whitfield, M. G. Norman. 1989. MR imaging of periventricular leukomalacia in childhood. AJR Am. J. Roentgenol. 152: 583-590.
van de Bor, M., G. L. Guit, A. M. Schreuder, J. Wondergem, G. J. Vielvoye. 1989. Early detection of delayed myelination in preterm infants. Pediatrics 84: 407-411.
Back, S. A., B. H. Han, N. L. Luo, C. A. Chricton, S. Xanthoudakis, J. Tam, K. L. Arvin, D. M. Holtzman. 2002. Selective vulnerability of late oligodendrocyte progenitors to hypoxia-ischemia. J. Neurosci. 22: 455-463.
Gilles, F. H., A. Leviton, C. S. Kerr. 1976. Endotoxin leucoencephalopathy in the telencephalon of the newborn kitten. J. Neurol. Sci. 27: 183-191.(Catherine Sherwin and Rob)