Surgery or trauma can cause systemic inflammation, and damage the brain-blood barrier, eventually induce neuroinflammation [52], which has been proved to be highly associated with postoperative cognitive dysfunction [8, 9, 53]. Intracerebroventricular injection of LPS can mimic neuroinflammation induced by surgery or trauma in animal studies [25, 54]. In the present study, we discovered that LPS-induced neuroinflammation increased apoptosis and neuronal loss in the dHPC, and ultimately lead to a selective spatial recognition memory impairment. This finding is consistent with previous studies which demonstrated that spatial memory is more vulnerable than non-spatial memory to LPS-induced damage [25, 54, 55]. In fact, the processing of spatial recognition memory is highly dependent on the dHPC [30, 56], while non-spatial recognition memory judgments for individual items or relative recency of items primarily depend on other brain regions, including mPFC and PRH [57, 58]. We further found that postconditioning with propofol or sevoflurane exerted a similar anti-neuroinflammatory effect via inhibiting microglia activation and the release of proinflammatory cytokines. However, postconditioning with sevoflurane but not propofol attenuated the apoptosis and neuronal loss and completely rescued the spatial cognitive impairment induced by neuroinflammation. Finally, we discovered that the persistent upregulation of extrasynaptic NR2B and the downstream CREB signaling pathway were involved in the neurotoxicity effect of postconditioning with propofol in the presence of pre-existing neuroinflammation.
An interesting finding of the present study is that only posttreatment with sevoflurane, but not propofol reversed the spatial recognition memory impairment after LPS injection. This is an intriguing observation because both inhalation and intravenous anesthetics are generally thought to have a neuroprotective effect in various experimental models including myocardial infarction [59], cerebral ischemia/reperfusion injury [1–5], and hemorrhage shock [60]. But recently, differential effects of sevoflurane and propofol on anxiety-like behavior in formalin-induced pain were reported in an animal study [61]. A clinical study also showed that sevoflurane-based anesthesia for on-pump cardiac surgery was associated with better short-term postoperative cognitive performance than propofol [62], however, the mechanism underlying this differential effects of sevoflurane and propofol on postoperative cognitive function has been unknown. Here, we found that postconditioning with sevoflurane persistently alleviated apoptosis and neuronal loss in the dHPC after LPS injection, whereas the anti-apoptotic effect of propofol postconditioning was only observed on one day after LPS injection, and increased apoptosis and neuronal loss still existed at a later time. One possible explanation for these results is that the anti-inflammatory effect of propofol postconditioning was more prominent than neurotoxicity in an early stage of LPS-induced acute neuroinflammation, while the neurotoxic effect was long-lasting and may account for the neuronal loss in the dHPC.
NR2B subunit is crucial to vital brain functions including synaptic plasticity, gene transcription, neuronal survival, and cognitive function [12, 63, 64]. After maturation, NR2B-containing NMDARs are mainly distributed at the extrasynaptic site, and excessive activation of extrasynaptic NR2B-containing receptors leads to neuronal death via activating downstream signaling pathways [20–24]. We found a selective increase of extrasynaptic NR2B-containing receptors after propofol postconditioning. In contrast, NR2B at the synaptic site was insensitive to propofol postconditioning. Moreover, other NMDAR subunits or AMPAR subunits were not modulated by propofol postconditioning. Meanwhile, we did not find LPS alone could alter NR2B expression, indicating the effect of propofol postconditioning on NR2B expression was exerted directly by itself, but not secondary to the anti-inflammation. Phosphorylation of ERK1/2 and CREB, as the downstream signaling molecules of synaptic NMDARs, play an important role in modulating the transcription of genes that promote spine formation, synaptic plasticity, neuronal survival, learning and memory [20]. In the current study, posttreatment with propofol reduced the levels of total ERK1/2 and nuclear CREB phosphorylation, which is associated with up-regulation of extrasynaptic NR2B-containing NMDARs. In addition, the increases in total and extrasynaptic NR2B-containing receptors expression and neuronal loss, deficits in downstream CREB signaling pathway, along with associated spatial recognition memory impairment induced by propofol postconditioning were reversed by systematical treatment with low dose memantine, previously shown to selectively block extrasynaptic NMDAR current while preserving synaptic NMDAR current [18]. Treatment with a low dose of memantine alone did not affect behavioral performance in distinct memory tasks [65], however, memantine at a 30-fold higher dose, which should block both synaptic and extrasynaptic NMDAR current, worsened the neurodegeneration and impaired cognitive function [18, 65]. Meanwhile, another NR2B antagonist Ro25-6981 was used to confirm the effect of persistent upregulation of extrasynaptic NR2B on neurotoxicity induced by propofol postconditioning after LPS injection. Ro25-6981 is commonly applied by local brain region injection or intraperitoneal injection in animal studies [66, 67]. And several studies showed that antagonism of NR2B by Ro25-6981 could inhibit its up-regulation [68, 69]. Here, consecutive administrations of Ro25-6981 at a very low dose (0.375µg) inhibited up-regulation of total and extrasynaptic NR2B subunit in the dHPC, reversed down-regulation of ERK1/2 and CREB phosphorylation, excessive apoptosis, neuronal loss, and improved the cognitive deficit after propofol postconditioning. We did not examine the effects of memantine and Ro25-6981 on normal object-place recognition memory, but it is known that memantine or Ro25-6981 produces its pharmacological effect in a dose-dependent manner [66, 70], and it was confirmed that Ro25-6981 with 0.375µg did not impair the normal object recognition memory, but only eliminated the changes of recognition memory induced by NR2B subunit alteration, while a high dose of Ro25-6981 (3.75µg) could disrupt the normal recognition memory [26].
Many conflicting observations regarding the effects of general anesthetics on NR2B expression and cognitive function were reported previously. For instance, isoflurane up-regulated extrasynaptic NR2B in the hippocampus in elderly rats and impaired cognitive function [24]; and isoflurane or sevoflurane up-regulated NR2B in the hippocampus, facilitated LTP in CA1 area, and improved cognition in adult mice [14, 16]. Sevoflurane induced NR1 and NR2B up-regulation, NR2A down-regulation in the PFC, and caused cognitive deficits in adolescent rats [71], while propofol down-regulated NR1, NR2B, NR2A, and PSD-95 in the PFC of developing rat brain, and caused long-term cognitive dysfunction [17,72]. The possible reasons for this discrepancy might be different ages, species of animals, and strategies of drug treatment. Moreover, the above studies were more concerned with the effects of general anesthetics exposure alone on NR2B expression and cognitive function. However, when LPS-induced neuroinflammation has already existed, general anesthetics may exert different effects on NR2B expression and cognitive function. It is still unknown why postconditioning with sevoflurane or propofol has different modulation on NR2B expression in the present study. One possibility is that although sevoflurane and propofol exert their anesthetic effect through GABA receptor and NMDA receptor, NR2B subunit is more sensitive to propofol than sevoflurane [73,74]. Besides, many posttranslational mechanisms including protein-protein interaction, phosphorylation, palmitoylation, ubiquitination, and proteolysis contribute to regulating the specific synaptic and extrasynaptic NMDARs distribution [75], but to our knowledge, only a few studies demonstrated that general anesthetics ur-regulated the expression of serine/threonine cyclin-dependent kinase 5 (Cdk5) [76,77], which is implicated in regulating the surface expression of NMDARs by NR2B phosphorylation at tyrosine1472 [78]. Therefore, further study should be conducted to elucidate the exact mechanism of the selective increase of extrasynaptic NR2B-containing NMDARs induced by propofol postconditioning after LPS injection.
In summary, under the pre-existing neuroinflammation, postconditioning with sevoflurane can provide reliable neuroprotection through attenuating LPS-induced neuroinflammation, apoptosis, neuronal loss, and improving spatial recognition deficit. However, posttreatment with propofol leads to a selective long-lasting upregulation of extrasynaptic NR2B-containing NMDARs in the dHPC, which down-regulates the CREB signaling pathway and impair the spatial recognition memory. The differential effects of postconditioning with sevoflurane or propofol suggest that sevoflurane seems to be more appropriate than propofol in preserving cognitive function.