EPO is a major hematopoietic growth factor that is mainly produced in the kidney and fetal liver 
. It is also known to express in CNS tissue 
. EPO mRNA is constitutively expressed in the cortex and hippocampus of the brain 
. Various studies have focused on the function of EPO in CNS; for example, mice lacking EPO or EPOR exhibited increased apoptosis in the brain before they died from severe anemia in utero 
, and mice lacking EPOR in the brain suffered from reduced neurogenesis or impaired migration of neurons in a brain stroke model 
. Thus, EPO is considered to be a neuroprotective factor against hypoxic-ischemic and traumatic injuries and essential for neuronal development 
In the present study, we showed that the induction of EPO expression under hypoxic conditions was suppressed by the general anesthetic isoflurane in a concentration- and time-dependent manner in the mouse brain. Other anesthetics, including sevoflurane, halothane, N2
O, pentobarbital, ketamine, and propofol, showed a similar effect. As for the mechanism of this suppression, we found that the accumulation of HIF-2α, but not HIF-1α, protein under hypoxic conditions was suppressed with isoflurane in the mouse brain. This finding is consistent with a previous report indicating that EPO is a target gene for HIF-2α, rather than HIF-1α, in CNS 
. HIF-1α is expressed ubiquitously, but the expression of HIF-2α is tissue-specific 
. HIF-2α is expressed in astrocytes and endothelial cells in the CNS 
. Astrocytes are the major source of EPO in the CNS 
, and hypoxia-induced EPO upregulation is dramatically reduced in the astrocyte-specific HIF-2α knockout mouse 
. In the present study, various anesthetics, including isoflurane, pentobarbital, and ketamine, suppressed the accumulation of HIF-2α protein under hypoxic conditions in cultured astrocytes. Therefore, our results indicated that the hypoxia-induced activation of HIF-2 in astrocytes was inhibited by general anesthetics, which resulted in a significant suppression of EPO production.
Recently, various studies on astrocytes have been performed, and these cells are considered responsible for a wide variety of functions in the CNS, including synaptic transmission and information processing by neural circuit functions 
. In the present study, we showed that various anesthetics suppressed the accumulation of HIF-2α protein and EPO upregulation under hypoxic conditions in the mouse brain and cultured astrocytes. Considering the fact that the accumulation of HIFα proteins is induced by hypoxia, the suppression of oxygen consumption induced by general anesthetics may reduce the level of hypoxia and consequently decrease HIF-2α protein accumulation. Actually, most of all general anesthetics excluding ketamine and N2
O are known to decrease cerebral metabolic rate of oxygen 
. But the effect of anesthetics on metabolism of astrocytes is not well investigated. In the current study, pentobarbital, well known for suppressing the metabolism of the CNS 
, decreased oxygen consumption of astrocytes. On the other hand, ketamine and propofol did not change oxygen consumption. In addition, we previously reported that hypoxic brain EPO induction was preserved in hypothermic mice, although hypothermia is well known to reduce cerebral oxygen consumption 
. Therefore, these findings suggest that the suppressive effect of various anesthetics against HIF-2α protein accumulation and EPO induction cannot be explained only by the decrease in oxygen consumption.
The main target of general anesthetics differs with various anesthetics; for example, ketamine and N2
O act via N-methyl-D-aspartate (NMDA) receptors 
, whereas the volatile anesthetics, propofol and barbiturates act via γ-aminobenzoic acid-A (GABA-A) receptors 
. On the other hand, the effect of general anesthetics on glial cells is not well understood, except for the fact that volatile anesthetics inhibit the glutamate uptake of astrocytes 
. Our finding that various general anesthetics have an EPO-suppressive effect in in vitro
experiments suggests that general anesthetics have some common direct effects on astrocytes. This finding is quite surprising considering the diverse action mechanism of general anesthetics. Anesthetics modulate functions of macromolecules, which play an essential role in cellular signal transduction. For example, protein kinase C (PKC) 
, mitogen-activating protein kinases (MAPKs) 
, and reactive oxygen species (ROS) 
are modulated by anesthetics. PKC, MAPK, and ROS are also identified to affect HIF activity by modulating HIF-α protein translation rate, hydroxylation, and phosphorylation of HIF-α protein 
. Therefore, general anesthetics may affect astrocytes through modulation of such enzymes and mediators. But most of the studies considering the effect of anesthetics on HIF have focused on HIF-1 under normoxic conditions, and the effect on HIF-2 under hypoxic conditions is not well understood.
Another important finding of the present study is the difference of behavior between HIF-1α and HIF-2α. Namely, HIF-1α was expressed even under normoxic conditions and 3-h hypoxic exposure did not affect HIF-1 protein accumulation distinctively in mice brains. In contrast, HIF-2α was barely expressed under normoxic conditions and clearly increased in response to hypoxic exposure. In in vitro
experiments, however, both HIF-1α and HIF-2α protein accumulation were observed under a 1% O2
condition, and various anesthetics significantly suppressed their induction. Although both HIF-1α and HIF-2α are considered to accumulate significantly under hypoxic conditions in in vivo
experiments using the mouse brain 
, some reports have shown that HIF-1α was expressed even under normoxic conditions 
. In most of these previous reports, HIF-1α protein increased in response to hypoxic exposure in the brain, but the extent varied 
. A possible explanation for the discrepancy is the difference of oxygen concentration. Previous report showed that HIF-1α protein accumulation was observed under 1% O2
condition but not 5% O2
condition in the neuronal cell line SK-N-BE cells 
. We exposed mice to 10% O2
in our studies, but 10% O2
might not be low enough to induce HIF-1α protein accumulation.
EPO has now been considered to be one of the promising agents for neuroprotection 
. Actually, in the clinical trials, erythropoietin showed neuroprotective effect against acute stroke 
, hypoxic-ischemic encephalopathy in newborns 
and delayed ischemic deficits following aneurysmal subarachnoid hemorrhage 
. In the current study, we showed the induction of EPO mRNA expression under hypoxic conditions was suppressed with isoflurane in a concentration- and time-dependent manner. Other anesthetics including sevoflurane, halothane, N2
O, pentobarbital, propofol and ketamine showed the same effect. Most of all anesthetics suppressed EPO mRNA induction with concentrations no more than clinically used, for example, isoflurane, sevoflurane and halothane showed this effect with 0.5%. Therefore, considering the neuroprotective effect of EPO, exposure to anesthetics beyond necessity should be avoided especially in cases hypoxia in brain may happen at greater risk like cardiovascular surgery.
According to the recent reports, anesthetic exposure in neonatal animals leads to neuronal death in certain circumstances 
. Such neurotoxicity has now been demonstrated for many anesthetics, including isoflurane, ketamine, midazolam, pentobarbital, N2
O, and propofol, and a positive correlation may exist between increased levels of anesthesia and increased severity of neuroapoptosis 
. The precise mechanisms by which injury is invoked are not clear, although an imbalance between excitatory and inhibitory input in the CNS during synaptogenesis may contribute to such an effect 
. On the other hand, EPOR is highly expressed in the developing mouse brain 
, and mice lacking EPO or EPOR experienced increased apoptosis in the brain before they died of severe anemia in the uterus 
. We did not investigate the effect of general anesthetics on brain EPO under normoxic conditions in neonatal animals. However, considering the pivotal role of EPO in brain development, general anesthetics may cause neuroapoptosis by suppressing EPO production in the brain. Further studies using neonatal animals should be performed.
In conclusion, we demonstrated that isoflurane inhibited hypoxia-induced EPO upregulation in the mouse brain and cultured astrocytes, most likely through suppression of HIF-2 activity. Other general anesthetics showed the same effect. Our findings suggest that general anesthetics have some direct effect on astrocytes and a major impact on the hypoxic response of the CNS.