J. Neurochem. (2010) 112, 854–869. Abstract Sleep apnea (SA) can be effectively managed in humans but it is recognized that when left untreated, SA causes long‐lasting changes in neuronal circuitry in the brain. Recent neuroimaging studies gave suggested that these neuronal changes are also present even in patients successfully treated for the acute effects of SA. The cellular mechanisms that account for these changes are not certain but animal models of intermittent hypoxia (IH) during sleep have shown neuronal death and impairment in learning and memory. Reactive gliosis has a drastic effect on neuronal survival and circuitry and in this study we examined the neuro‐glial response in brain areas affected by SA. Glial and neuronal alterations were analyzed after 1, 3, 5 and 10 days of exposure to IH (8 h/day during the sleep phase, cycles of 6 min each, 10–21% O2) and observed significant astroglial hyperplasia and hypertrophy in parietal brain cortex and hippocampus by studying gliofibrillary acidic protein, Vimentin, S100B and proliferating cell nuclear antigen expression. In addition, altered morphology, reduced dendrite branching and caspase activation were observed in the CA‐1 hippocampal and cortical (layers IV–V) pyramidal neurons at short exposure times (1–3 days). Surprisingly, longer exposure to IH reduced the neuronal death rate and increased neuronal branching in the presence of persistent reactive gliosis. Up‐regulation of hypoxia inducible factor 1 alpha (HIF‐1α) and mdr‐1, a HIF‐1α target gene, were observed and increased expression of receptor for advanced end glycated products and its binding partner S100B were also noted. Our results show that a low number of hypoxic cycles induce reactive gliosis and neuronal death whereas continuous exposure to IH cycles reduced the rate of neuronal death and induced neuronal branching on surviving neurons. We hypothesize that HIF‐1α and S100B glial factor may improve neuronal survival under hypoxic conditions and propose that the death/survival/re‐growth process observed here may underlie brain circuitry changes in humans with SA.

S100B is a soluble protein secreted by astrocytes that exerts pro-survival or pro-apoptotic effects depending on the concentration reached in the extracellular millieu. The S100B receptor termed RAGE (for receptor for advanced end glycation products) is highly expressed in the developing brain but is undetectable in normal adult brain. In this study, we show that RAGE expression is induced in cortical neurons of the ischemic penumbra. Increased RAGE expression was also observed in primary cortical neurons exposed to excitotoxic glutamate (EG). S100B exerts effects on survival pathways and neurite extension when the cortical neurons have been previously exposed to EG and these S100B effects were prevented by anti-RAGE blocking antibodies. Furthermore, nuclear factor kappa B (NF-jB) is activated by S100B in a dose-and RAGE-dependent manner and neuronal death induced by NF-jB inhibition was prevented by S100B that restored NF-jB activation levels. Together, these findings suggest that excitotoxic damage can induce RAGE expression in neurons from ischemic penumbra and demonstrate that cortical neurons respond to S100B through engagement of RAGE followed by activation of NF-jB signaling. In addition, basal NF-jB activity in neurons is crucial to modulate the extent of pro-survival or pro-death S100B effects.

Extracellular S100B dramatically increases after brain injury. While low S100B levels are neuroprotective, micromolar S100B levels have shown in vitro to activate microglia and facilitate neuronal death. In astrocytes, S100B exposure activates nuclear factor kappa B (NF-jB) and induces proinflammatory mediators. On microglia and neurons S100B effects are essentially mediated by receptor for advanced glycation end products (RAGE)/NF-jB, but it is not clear if these intracellular cascades are activated by different S100B levels in astrocytes and whether increased extracellular S100B is sufficient to induce reactive gliosis. A better understanding of these pathways is essential for developing successful strategies to preserve the beneficial S100B effects after brain injury. Here, we show that microglia-depleted cultured astrocytes exposed to S100B mimicked several features of reactive gliosis by activating RAGE/Rac-1-Cdc42, RAGE/Erk-Akt or RAGE/NF-jB-dependent pathways. S100B effects include RAGE/Rac1-Cdc42-dependent astroglial hypertrophy and facilitation of migration as well as increased mitosis. S100B exposure improved the astrocytic survival to oxidative stress, an effect that requires Erk/Akt. S100B also activates NF-jB in a dose-dependent manner; increases RAGE proximal promoter transcriptional activity and augmented endogenous RAGE expression. S100B-exposed astrocytes showed a proinflammatory phenotype with expression of Toll-like receptor 2 (TLR 2), inducible nitric oxide synthase (iNOS) and interleukin 1-beta (IL-1b), and facilitated neuronal death induced by oxygen-glucose deprivation. In vivo, intracerebral infusion of S100B was enough to induce an astroglial reactive phenotype.Together, these findings demonstrate that extracellular S100B in the micromolar level activates different RAGE-dependent pathways that turn astrocytes into a pro-inflammatory and neurodegenerative phenotype.

The lithium-pilocarpine model of epilepsy reproduces in rodents several features of human temporal lobe epilepsy, by inducing an acute status epilepticus (SE) followed by a latency period. It has been proposed that the neuronal network reorganization that occurs during latency determines the subsequent appearance of spontaneous recurrent seizures. The aim of this study was to evaluate neuronal and glial responses during the latency period that follows SE. Given the potential role of astrocytes in the post-SE network reorganization, through the secretion of synaptogenic molecules such as thrombospondins, we also studied the effect of treatment with the α2δ1 thrombospondin receptor antagonist gabapentin. Adult male Wistar rats received 3 mEq/kg LiCl, and 20 h later 30 mg/kg pilocarpine. Once SE was achieved, seizures were stopped with 20 mg/kg diazepam. Animals then received 400 mg/kg/day gabapentin or saline for either 4 or 14 days. In vitro experiments were performed in dissociated mixed hippocampal cell culture exposed to glutamate, and subsequently treated with gabapentin or vehicle. During the latency period, the hippocampus and pyriform cortex of SE-animals presented a profuse reactive astrogliosis, with increased GFAP and nestin expression. Gliosis intensity was dependent on the Racine stage attained by the animals and peaked 15 days after SE. Microglia was also reactive after SE, and followed the same pattern. Neuronal degeneration was present in SE-animals, and also depended on the Racine stage and the SE duration. Polysialic-acid NCAM (PSA-NCAM) expression was increased in hippocampal CA-1 and dentate gyrus of SE-animals. Gabapentin treatment was able to reduce reactive gliosis, decrease neuronal loss and normalize PSA-NCAM staining in hippocampal CA-1. In vitro, gabapentin treatment partially prevented the dendritic loss and reactive gliosis caused by glutamate excitotoxicity. Our results show that gabapentin treatment during the latency period after SE protects neurons and normalizes PSA-NCAM probably by direct interaction with neurons and glia.

Sleep apnea (SA) causes long-lasting changes in neuronal circuitry, which persist even in patients successfully treated for the acute effects of the disease. Evidence obtained from the intermittent hypoxia (IH) experimental model of SA has shown neuronal death, impairment in learning and memory and reactive gliosis that may account for cognitive and structural alterations observed in human patients. However, little is known about the mechanism controlling these deleterious effects that may be useful as therapeutic targets in SA. The Receptor for Advanced Glycation End products (RAGE) and its downstream effector Nuclear Factor Kappa B (NF-κB) have been related to neuronal death and astroglial conversion to the pro-inflammatory neurodegenerative phenotype. RAGE expression and its ligand S100B were shown to be increased in experimental models of SA. We here used dissociated mixed hippocampal cell cultures and male Wistar rats exposed to IH cycles and observed that NF-κB is activated in glial cells and neurons after IH. To disclose the relative contribution of the S100B/RAGE/NF-κB pathway to neuronal damage and reactive gliosis after IH we performed sequential loss of function studies using RAGE or S100B neutralizing antibodies, a herpes simplex virus (HSV)-derived amplicon vector that induces the expression of RAGEΔcyto (dominant negative RAGE) and a chemical blocker of NF-κB. Our results show that NF-κB activation peaks 3 days after IH exposure, and that RAGE or NF-κB blockage during this critical period significantly improves neuronal survival and reduces reactive gliosis. Both in vitro and in vivo, S100B blockage altered reactive gliosis but did not have significant effects on neuronal survival. We conclude that both RAGE and downstream NF-κB signaling are centrally involved in the neuronal alterations found in SA models, and that blockage of these pathways is a tempting strategy for preventing neuronal degeneration and reactive gliosis in SA.