The α2 Na+/K+-ATPase isoform mediates LPS-induced neuroinflammation

Na+/K+-ATPase is a transmembrane ion pump that is essential for the maintenance of ion gradients and regulation of multiple cellular functions. Na+/K+-ATPase has been associated with nuclear factor kappa B (NFκB) signalling, a signal associated with lipopolysaccharides (LPSs)-induced immune response in connection with activated Toll-like receptor 4 (TLR4) signalling. However, the contribution of Na+/K+-ATPase to regulating inflammatory responses remains elusive. We report that mice haploinsufficient for the astrocyte-enriched α2Na+/K+-ATPase isoform (α2+/G301R mice) have a reduced proinflammatory response to LPS, accompanied by a reduced hypothermic reaction compared to wild type litter mates. Following intraperitoneal injection of LPS, gene expressions of Tnf-α, Il-1β, and Il-6 was reduced in the hypothalamus and hippocampus from α2+/G301R mice compared to α2+/+ littermates. The α2+/G301R mice experienced increased expression of the gene encoding an antioxidant enzyme, NRF2, in hippocampal astrocytes. Our findings indicate that α2Na+/K+-ATPase haploinsufficiency negatively modulates LPS-induced immune responses, highlighting a rational pharmacological target for reducing LPS-induced inflammation.

www.nature.com/scientificreports/ functions, the α 2 isoform is functional primarily in astrocytes and developing neurons and the α 3 isoform is restricted to neurons 19,20 .
In the adult brain, the Na + /K + -ATPase α 2 isoform is enriched in astrocytes and important for extracellular K + clearance and function to support glutamate uptake from the synaptic cleft [21][22][23] and impair astrocytic K + clearance 24 . Accordingly, mutations in the ATP1A2 gene, which encodes the α 2 subunit isoform of the Na + / K + -ATPase, can cause familial hemiplegic migraine type 2 (FHM2) 25 , a subtype of migraine with aura 26 . Several Atp1a2 gene-modified mouse models have been made and extensively used to study the α 2 isoform in vivo 25 . While KO and KI homozygous mice dies immediately after birth [27][28][29][30] heterozygous α 2 knock-out (KO) and knock-in (KI) mice are viable.
FHM2 patients with the G301R mutation resulting from a gene variant in the ATP1A2 gene have migraine comorbidity with epilepsy, coma, motor symptoms and psychiatric disorders such as depression and obsessive-compulsive disorder (OCD) 31,32 . Heterozygous KI mice containing the G301R disease mutation (α 2 +/G301R mice) displayed FHM2-related phenotypes, including mood depression and obsessive-compulsive disorder (OCD)-like symptoms 33 , besides showed a greater susceptibility to epilepsy and disseminated cortical depression (CSD) 33,34 . Moreover, astrocyte-neuron primary in vitro cultures from α 2 (G301R/G301R) mice revealed impaired glutamate uptake 33 . Surprisingly, when submitted to spinal cord injury, the α 2 +/G301R mice display improved functional recovery and decreased lesion volume compared to littermate controls (α 2 +/+ ) 35 . These phenotypes were associated with changes in pro-and anti-inflammatory cytokines, with the cytokines TNF, IL-6, and IL-10 upregulated in the spinal cord of α 2 +/G301R and α 2 +/+ mice with a spinal cord injury 35 . Interestingly, the functional recovery of the α 2 +/G301R mice was improved compared to α 2 +/+ mice and correlated with a significantly reduced lesion size 35 . In line with this, astrocytes deficient of the mutant superoxide dismutase 1 (SOD1) that also was depleted from the α 2 Na + /K + -ATPase, were able to protect motor neurons from degeneration in co-cultured primary motor neurons 36 . In these SOD-deficient astrocytes, mitochondrial respiration and inflammatory gene expressions appeared induced.
The genetic mechanisms of the Atp1a2 pathophysiology thus appear to reply on the ion balance, however, the impact of this on the immune response remains unknown.
As an initial attempt to understand this, we challenged heterozygous α 2 +/G301R mice by lipopolysaccharides (LPS) administration. We hypothesised that the mice would be compromised compared to wild type litter mates as ion channels-related diseases contribute to similar symptoms, and can be used as potential targets to modulate immune response and to treat inflammatory disorders and cancer.
LPSs, known as lipoglycans and endotoxins, acts as a prototypical endotoxin, binding the CD14/TLR4/MD2 receptor complex in many cell types, which promotes the production and secretion of proinflammatory cytokines, nitric oxide, and eicosanoids. LPS can induce neuroinflammation in the brain, including NF-κB activation in rodents, which can lead to impaired cognitive performance 37,38 .
Interestingly, short interference (si)RNA-mediated knockdown of the gene encoding the α 2 isoform in primary astrocyte cultures prevented the LPS-mediated activation of ERK and NFκB 39 , but the precise mechanism through which the α 2 Na + /K + -ATPase pump regulates neuroinflammation remains elusive.
Here, we studied the α 2 Na + /K + -ATPase regulation of LPS-mediated neuroinflammation in the hypothalamus and hippocampus using a heterozygous mouse model (α 2 +/G301R ) that exhibits α 2 haploinsufficiency 33 . We found a reduction in the systemic production of the proinflammatory cytokines TNF-α, IL-1β, and IL-6 following LPS administration, as well as a reduced hypothermic response in α 2 +/G301R mice.
To test if the apparently unaffected LPS-treated α 2 +/G301R mice experienced changes in body temperature, a characteristic effect of LPS treatment, body temperatures were measured at baseline and 4 h after LPS injection. In line with the observed phenotypes, the α 2 +/+ mice exhibited a severe drop in body temperature (− 3.85 ± 3 °C) after LPS administration, whereas the α 2 +/G301R mice exhibited a significantly smaller drop in body temperature (− 1.40 ± 3.3 °C) (Fig. 1b). To evaluate whether cytokine levels were altered in the mice treated with LPS and if these alterations were different in the α 2 +/G301R mice, blood samples were collected 4 h after LPS administration and cytokine levels were measured using an enzyme-linked immunosorbent assay (ELISA). As expected, the LPS-treated α 2 +/+ mice exhibited significantly increased serum tumour necrosis factor-α (TNF-α), interleukin-1 beta (IL-1β) and IL-6 levels compared with those exhibited by the PBS-treated α 2 +/+ mice ( Fig. 1c-e). The LPStreated α 2 +/G301R mice had significantly reduced serum levels of TNF-α (Fig. 1c), and IL-6 ( Fig. 1e) compared with those of the LPS-treated α 2 +/+ mice. The levels of IL-1β in the LPS-treated α 2 +/G301R mice were equal to those in the LPS-treated α 2 +/+ mice (Fig. 1d). Combined, these results show a reduced sickness behavior and hypothermic response observed in the α 2 +/ G301R mice compared to α 2 +/+ mice, and moreover, indicate that induction of TNF-α and IL-6 is compromised in response to LPS treatment in α 2 +/G301R mice. Our data show that LPS challenge promoted a serum increase of IL-1β that was not reversed in α 2 +/G301R mice, thus suggesting that α 2 activity does not interfere with the release of IL-1β in its mature form. +/G301R mice display decreased sickness behaviour. α 2 +/G301R mice experience hypothermia, and systemic proinflammatory cytokine levels compared to α 2 +/+ mice following LPS administration (500 µg/kg, intraperitoneally (IP) injected). (a) Example of sickness behaviour 4 h after LPS administration. As illustrated, the α 2 +/+ mice displayed sickness behaviour, less locomotion, and, briefly, hypothermia, in contrast to the α 2 +/G301R mice that displayed less sickness behaviour after LPS administration. (b) The differences in body temperature 4 h after LPS administration compared to baseline (t = 0) in the α 2 +/+ and α 2 +/G301R animals treated with PBS or LPS, with a significant different after LPS treatment. (c-e) The α 2 +/G301R mice and their WT littermates (α 2 +/+ ) were injected with saline or LPS (500 µg/kg), and the levels of TNF-α (c) (n = 12 for the α 2 +/+ mice and n = 14 for α 2 +/G301R mice), IL-1β (d) [α 2 +/+ (n = 6) and α 2 +/G301R mice (n = 7)] and IL-6 (e) [α 2 +/+ (n = 6) and α 2 +/G301R (n = 7)] were measured in the serum 4 h after saline/LPS treatment. The data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 (Kruskal-Wallis test followed by Dunn's post hoc test). www.nature.com/scientificreports/ α 2 na + /K + -ATPase is not required for LPS-induced cytokine production in macrophages. Macrophages are the main producers of proinflammatory cytokines, including IL-6 and TNF-α 40,41 . Therefore, next we studied whether the decreased systemic levels of these cytokines in α 2 +/G301R mice were due to a reduced macrophage response and, hence, whether the macrophage response to LPS involves the Na + /K + -ATPase α 2 isoform. To this end, we measured TNF-α production in response to LPS treatment in bone marrow-derived macrophages (BMDMs) isolated from α 2 +/G301R mice and wild-type (WT) littermates. Surprisingly, the levels of TNF-α in the BMDMs from both α 2 +/+ and α 2 +/G301R did not differ significantly at any of the time points analysed (Fig. 2a-d). However, it is noteworthy that although the levels of TNF-α did not differ between α 2 +/+ and α 2 +/G301R , in both cases they are significantly increased after 1, 2, 4 and 6 h of LPS treatment. This suggested that the Na + /K + -ATPase α 2 isoform may not be functionally required for LPS-induced TNF-α release in macrophages. Indeed, low levels of α 2 protein were detected by Western blotting in BMDM cells and the levels diminished during BMDM differentiation in vitro ( Fig. 2e and Supplementary Fig. 1). Furthermore, no significant differences in α 2 isoform expression in the α 2 +/+ and α 2 +/G301R BMDM cells were observed (Fig. 2f), suggesting a low basal level of α 2 protein not significantly affected by the α 2 haploinsufficiency. Under these conditions, this implies that the α 2 Na + /K + -ATPase might regulate the proinflammatory response to LPS in other cell types.
LPS-induced neuroinflammation is suppressed in the hippocampus and hypothalamus of heterozygous α 2 +/G301R mice. Because steroid ouabain (OUA), a Na + /K + -ATPases inhibitor, has an antiinflammatory effect in the rat hippocampus [42][43][44] , and because the hypothalamus endocrinologically regulates temperature control, we examined the α 2 protein level in the hippocampus and hypothalamus of α 2 +/G301R mice. No significant changes were observed in the levels of the α 2 isoform in the hypothalamus between the α 2 +/G301R and α 2 +/+ mice (Fig. 3a, Supplementary Fig. 2). Like the observations in macrophages, this may imply that the α 2 isoform may not be produced at levels that would permit a detectable difference. By contrast, the α 2 isoform levels were significantly reduced in the hippocampus in the α 2 +/G301R mice compared with the α 2 +/+ mice (Fig. 3b, Supplementary Fig. 3). As a control, we also examined the levels of the housekeeping α 1 isoform. No differences were detected in the α 1 isoform levels in the hypothalamus and hippocampus (Fig. 3c,d, Supplementary Fig. 4, 5), confirming that the housekeeping α 1 isoform is not affected by the G301R mutation in the α 2 isoform. Moreover, LPS injection did not influence the levels of the α 1 (Fig. 3c,d) or α 2 (Fig. 3a,b) proteins in the hypothalamus or hippocampus in either the α 2 +/+ or α 2 +/G301R animals compared to those in the PBS-treated littermates. Since a significant reduction of α 2 proteins levels was observed in hippocampus, but no difference in the hypothalamus, we next determined it this would influence the gene expression of TNF-α, IL-1β, and IL-6 genes.
Cytokine expression was measured specifically in the hypothalamus and hippocampus of the α 2 +/+ and α 2 +/ G301R mice after PBS or LPS treatment by reverse transcription quantitative polymerase chain reaction (RT-qPCR). As expected, LPS treatment significantly upregulated the transcription of Tnf-α, Il-1β, and Il-6 genes in the hypothalamus (Fig. 4a-c) and hippocampus ( Fig. 4d-f) of the α 2 +/+ mice compared with the PBS control group. By contrast, the transcription of these genes in the hypothalamus and hippocampus of the α 2 +/G301R mice was unaffected by LPS treatment compared with that in the PBS-treated littermates ( Fig. 4a-f).
This indicates that part of the lack of a LPS immune response is associated with a significant reduced expression of the TNF-α, IL-1β, and IL-6 genes in both hippocampus and hypothalamus.
Tlr4 mRNA expression is reduced in astrocytes from α 2 +/G301R mice treated with LpS. It is well known that LPS acts as an agonist of TLR4 to induce neuroinflammation and the expression of proinflammatory cytokines. To determine whether the LPS-induced expression of proinflammatory cytokines is associated with a potential effect of α 2 haploinsufficiency in the TLR4 induction in hypothalamus and hippocampus, we investigated the expression levels of Tlr4 in the hypothalamus and hippocampus by RT-qPCR. No difference in Tlr4 gene transcription between the α 2 +/+ and α 2 +/G301R mice in either the PBS or LPS treatment group was observed in the hypothalamus or hippocampus ( Supplementary Fig. 6). However, as the α 2 isoform is specifically expressed in astrocytes, we speculated that α 2 haploinsufficiency would cause astrocytes from the α 2 +/G301R mice to have lower levels of α 2-Na + /K + -ATPase, and therefore, an effect on Trl4 mRNA expression may be noted only in astrocytes and not in samples containing all cells from the hypothalamus or hippocampus. We therefore measured Tlr4 expression in astrocytes isolated by bead sedimentation using an antibody against astrocyte cell surface antigen-2 (ACSA-2) 45 . In agreement with the established LPS-induced upregulation of TLR4, Tlr4 expression was significantly increased in astrocytes from both the hypothalamus (blue in Fig. 5a) and hippocampus (red in Fig. 5b) in the α 2 +/+ mice after LPS treatment compared with PBS treatment. Tlr4 expression in the astrocytes from the α 2 +/G301R mice was comparable with that in the astrocytes from the α 2 +/+ mice treated with PBS; however, no increase in Tlr4 expression was observed in the hypothalamus (Fig. 5a) or hippocampus (Fig. 5b) of the α 2 +/ G301R mice after LPS treatment, suggesting that the α 2 isoform is required for LPS-mediated effect on Tlr4 expression in astrocytes during the acute LPS-induced neuroinflammation.
Ligand-induced receptor dimerization is thought to initiate signal transduction through TLRs, which leads to the recruitment of signalling adaptor proteins and the regulation of TNF-α mRNA translation 46,47 . Therefore, we hypothesized that the reduced expression of Tlr4 in astrocytes from α 2 +/G301R mice could explain the reduced Tnf-α expression in hypothalamus and hippocampus of α 2 +/G301R mice. As the amount of astrocytes isolated form these brain areas are too low to measure TNF protein levels, we measured Tnf mRNA expression to investigate this, in astrocytes isolated from either the hypothalamus or the hippocampus of α 2 +/G301R mice compared to those isolated from α 2 +/+ (Fig. 5c,d), in line with the results of Tlr4 expression. In the hypothalamus there is a significant increase after LPS, though smaller than in α 2 +/+ astrocytes and in the hippocampus there is a significant reduction. www.nature.com/scientificreports/ Figure 2. The α 2 Na + /K + -ATPase isoform is not involved macrophage cytokine production in response to LPS. BMDMs derived from bone marrow cells from the femurs and tibias of α 2 +/+ and α 2 +/G301R mice were cultured in L929 cell supernatant for 6 days. After 6 days, the BMDMs were treated with LPS (100 ng/mL) or PBS (CTR). (a-d) The levels of TNF-α were measured in the supernatant 1 h (a), 2 h (b), 4 h (c), and 6 h (d) after LPS treatment. (e) α 2 Isoform levels were measured by Western blotting in bone marrow derived macrophages from α 2 +/+ and α 2 +/G301R mice at the indicated times during differentiation. β-Actin was used as a loading control. A protein extract from the central nervous system (CNS) was loaded as a positive control. (f) Densitometric analysis from day 0 (arbitrary units, A.U.) showing that no obvious differences existed in the α 2 protein levels in the BMDMs derived from the α 2 +/+ and α 2 +/G301R mice. The data are presented as the mean ± SEM from two individual experiments. The data are presented as the mean ± SEM. *p < 0.05 and **p < 0.01 (Kruskal-Wallis test followed by Dunn's post hoc test). www.nature.com/scientificreports/ Figure 3. The levels of α 2 Na + /K + -ATPase isoform are reduced in the hippocampus of α 2 +/G301R mice. α 1 and α 2 isoform protein levels assessed by Western blotting in the hypothalamus (blue bars) and hippocampus (red bars). (a,c) Representative digital images of Western blots and densitometric analysis (arbitrary units) of hypothalamic lysates from the α 2 +/+ and α 2 +/G301R mice 4 h after saline or LPS injection showed no differences in α 1 Na + /K + -ATPase (a) (n = 8-10 mice/group) and α 2 Na + /K + -ATPase (c) (n = 6 mice/group). (b,d) Representative digital images of Western blots and densitometric analysis of hippocampal lysates from the α 2 +/+ and α 2 +/G301R mice 4 h after saline or LPS injection showed no differences in α 2 Na + /K + -ATPase (b) (n = 10-11 mice/group) but showed a reduction in α 2 Na + /K + -ATPase irrespective of LPS treatment (d) (n = 7-9 mice/group). The results are expressed relative to the control (PBS) as the mean ± SEM from three individual experiments. *p < 0.05 (Kruskal-Wallis test followed by Dunn's post hoc test).

Scientific RepoRtS
Together, these results show that the lack of the TNF-α response in the α 2 +/G301R mice is partly due to the lack of α 2 Na + /K + -ATPase expressing astrocytes that can sense LPS through induced Tlr4 expression.
Equal cytosolic levels of p65 in LPS-treated α 2 +/G301R hippocampus. LPS binding to TLR4 initiates an intracellular signaling cascade that results in NFκB activation. The activation of the transcription factor NFκB is initiated upon a nuclear translocation. NFκB is an important transcriptional regulator of neuroinflammation that is activated by LPS and/or proinflammatory cytokines, such as TNF-α, which trigger intracellular localization of the NFκB complex. Therefore, next to further dissect the mechanisms of α 2 isoform-mediated regulation of LPS response, we evaluated the nuclear translocation of NFκB in the hypothalamus and hippocampus of LPS-treated α 2 +/G301R mice. We examined nuclear protein levels via Western blotting using an antibody that recognize RelA (p65), a component of the NFκB complex. No significant changes were observed in cytoplasmic p65 levels in the hypothalamus (Fig. 6a, Supplementary Fig. 7), but there was a reduction in the hippocampus (Fig. 6b, Supplementary Fig. 8) of α 2 +/+ mice and the α 2 +/G301R mice treated with PBS or LPS. In the nucleus, active NFκB promotes the transcription of NFκB-responsive genes 48,49 , such as IL-1β and IL-6. Therefore, next we measured Il-6 and Il-1β mRNA levels in astrocytes isolated from both the α 2 +/+ and α 2 +/G301R mice subjected to PBS or LPS treatment. These results showed that Il-6 mRNA was upregulated in astrocytes isolated from the hypothalamus (Fig. 6c) and hippocampus (Fig. 6d) of α 2 +/+ and α 2 +/G301R mice after LPS treatment. Il-1β mRNA was significantly upregulated in astrocytes isolated from the hypothalamus (Fig. 6e) and the hippocampus of α 2 +/+ mice after LPS treatment (Fig. 6f). However, Il-1β mRNA was significantly reduced in hippocampal astrocytes of α 2 +/G301R mice after LPS treatment (Fig. 6f). Comparing the mRNA expression of Il-1b and Il-6 in whole hypothalamus (Fig. 4a-c) and Hippocampus (Fig. 4d-f), there is an overall agreement that the effect is very local to the astrocytes, as the specific cells that is subjected to α 2 haploinsufficiency. mice exhibited reduced Tlr4 and Tnf-α mRNA expression in astrocytes isolated from the hypothalamus (blue bars) and hippocampus (red bars) compared with that in the α 2 +/+ mice after PBS treatment. (a,b) qPCR analysis of Tlr4 relative to β-actin expression in the hypothalamus (a) and hippocampus (b) showed a significant reduction of Trl4 in these brain areas, notably after LPS treatment, between the α 2 +/+ (n = 3) and α 2 +/G301R (n = 3) animals. The qPCR analysis of Tnf-α relative to β-actin expression in the hypothalamus (c) and hippocampus (d) showed that a significant decrease in Tnf-α mRNA expression was observed in astrocytes from α 2 +/G301R mice (n = 3) after LPS treatment compared with α 2 +/+ astrocytes (n = 3). All data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 (Kruskal-Wallis test followed by Dunn's post hoc test). www.nature.com/scientificreports/ (n = 5) animals 4 h after LPS treatment. All data are presented as the mean ± SEM. *p < 0.05 (Kruskal-Wallis test followed by Dunn's post hoc test). RT-qPCR analysis of (c,d) Il- 6 and (e,f) Il-1β relative to β-actin expression in astrocytes isolated from the hypothalamus (c,e) and hippocampus (d,f) showed significant increases after LPS treatment for all except Il-1β in hippocampus of α 2 +/G301R mice (n = 3 for both groups). All data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 (Kruskal-Wallis test followed by Dunn's post hoc test). www.nature.com/scientificreports/ This suggest that despite no significant nuclear translocation of NFκB was observed after LPS treatment in the α 2 +/G301R mice, there is however, a significant induction of Il-1b and Il-6 mRNA, suggesting either that the nuclear translocation assay needs to be performed in astrocyte populations, or supplementary mechanisms mediate the expression of the measured cytokines.
In conclusion, these results show that Nrf2 mRNA transcription is activated in the α 2 +/G301R mice and of relevance to LPS-induced response.
To study if the observed increase in Nrf2 mRNA expression in LPS-treated hypothalamus and hippocampus of the α 2 +/G301R and α 2 +/+ mice resulted in subsequent expression of NRF2-responsive genes, we measured the expression of the NRF2-responsive genes Ho-1 and Nqo1 58 in astrocytes from the hypothalamus and hippocampus of α 2 +/+ and α 2 +/G301R mice. LPS treatment resulted in increased Ho-1 expression in astrocytes from the hypothalamus (Fig. 7c) and hippocampus (Fig. 7d) of α 2 +/+ mice compared to that in PBS-treated mice. However, LPS did not lead to an increase in Ho-1 expression in the hypothalamus or in the hippocampus of α 2 +/G301R mice (Fig. 7c,d). No differences in Nqo1 expression between the hypothalamic astrocytes from the α 2 +/ G301R and α 2 +/+ mice were observed after LPS treatment (Fig. 7e,f), while Nqo1 expression was increased in the hippocampal astrocytes from the of α 2 +/G301R mice 4 h after treatment with LPS (Fig. 7f). Combined this indicate that Nfr2 expression increases after LPS treatment in both hypothalamus or in the hippocampus of α 2 +/+ mice and α 2 +/G301R mice. Of NRF2-responsive genes, Ho-1 mRNA appears reduced in astrocytes from hypothalamus of α 2 +/G301R mice compared to α 2 +/+ mice, however, only a difference was observed for Nqp-1 mRNA in astrocytes from hippocampus of α 2 +/G301R mice compared to α 2 +/+ mice. α 2 haploinsufficiency correlate directly with the increased expression of Cox2 transcripts in hippocampal astrocytes. LPS-mediated inflammation triggers the expression level of COX genes, major inflammatory factors 59 . To determine the role of the α 2 isoform in LPS-induced Cox1 and Cox2 expression, we performed RT-qPCR analysis on astrocytes isolated from the hypothalamus and hippocampus of the PBS-and LPS-treated α 2 +/+ and α 2 +/G301R mice. We found that there was no difference in the expression of the Cox1 gene in the hypothalamic astrocytes of all groups evaluated (Fig. 8a). In α 2 +/G301R astrocytes from the hippocampus, we noted that Cox1 gene expression was significantly increased after LPS treatment, in contrast to PBS-treated α 2 +/G301R hippocampal astrocytes (Fig. 8b). The Cox2 mRNA expression in α 2 +/+ mice astrocytes was upregulated in hypothalamus after the challenge with LPS (Fig. 8c), but unchanged in hippocampus (Fig. 8d).
In astrocytes from the hypothalamus, Cox2 mRNA was significantly elevated in PBS-treated α 2 +/G301R mice, with no further induction was noted in astrocytes from the α 2 +/G301R mice upon LPS treatment (Fig. 8c). In astrocytes from the hippocampus, the Cox2 levels in the α 2 +/+ mice were not altered (Fig. 8d). In contrast, Cox2 mRNA expression was significant upregulated in LPS-treated hippocampal astrocytes from the α 2 +/G301R mice (Fig. 8d). Overall, this correlates very well with the tendency of increased nuclear levels of p65, which suggests increased NFκB activity in astrocytes from the hippocampus (Fig. 6b) and is responsible for high Cox2 expression 59 , as noted in the same cells (Fig. 8b).

LPS-induced memory impairment and anxiety-like behaviours are not significantly altered in α 2
+/G301R mice compared with WT littermates. Because long-lasting effects in memory and behavior have been observed following LPS administration (reviewed in 60 ), we wished to examine if the α 2 +/G301R mice were susceptible for behavior changes upon LPS administration.
To assess whether the α 2 isoform is involved in the cognitive and affective LPS-mediated effects, first, we examined the locomotor activity of the α 2 +/G301R mice versus the α 2 +/+ mice treated with saline or LPS using the open field test 4 h after PBS or LPS administration. LPS induced a reduction in locomotion and anxiety-like behaviours in both the α 2 +/+ and α 2 +/G301R mice (Fig. 9a,b). There was no difference in the total distance travelled (P = 0.5806) or the time spent in the middle zone of the arena (P = 0.7457) by LPS-treated α 2 +/+ and α 2 +/G301R mice. Similarly, no differences were observed between the two types of mice after PBS treatment.
Next, we investigated whether α 2 isoform haploinsufficiency can influence LPS-induced memory impairment. To this end, we assessed fear memory using the passive avoidance test that measures the latency to enter a dark environment in which an aversive stimulus (foot shock) has been previously experienced using a light-dark box paradigm. All four groups (PBS-and LPS-treated α 2 +/+ and α 2 +/G301R mice) showed similar baseline latencies to enter the dark chamber in the training stage ( Supplementary Fig. 9). We evaluated the latency for entry to the dark side of the box before the challenge with LPS and observed that the isoform α2-Na,K-ATPase mutation does not interfere with the time of entry to the chamber in the training stage ( Supplementary Fig. 9a) www.nature.com/scientificreports/ increased latency time for α 2 +/G301R mice, in the test, after treatment with LPS, it showed that α2 +/G301R mice are less cognitively affected than α 2 +/+ animals, after treatment with LPS (Fig. 9c), and we did not observe differences in latency between the mutated animals treated with PBS or LPS (Supplementary Fig. 9b). (a-f) RT-qPCR mRNA expression analysis for Nrf2 (a,b); Ho-1 (c,d); Nqo-1 (e,f) in the hypothalamus (blue bars) and hippocampus (red bars). All data are the mean ± SEM of three individual experiments. *p < 0.05, *p < 0.01, and ***p < 0.001 (two-way ANOVA followed by Tukey's post hoc test, n = 3-4 mice/group). www.nature.com/scientificreports/ The α 2 +/+ and α 2 +/G301R mice treated with PBS showed a significant increase in latencies in the probe stage, compared to LPS-treated α 2 +/+ mice (Fig. 9c). The increased time in latency was found significant for α 2 mice, in the test, after LPS treatment, suggesting that the α 2 +/G301R mice are less cognitive affected than their littermates, after LPS treatment.
Collectively, our results suggest that the reduction of the α 2 isoform does not interfere with the LPS-induced decrease in locomotor activity, increase in anxiety behaviours, or cognitive impairment. Activation of the immune system by LPS leads to production and release of proinflammatory cytokines such as TNF-α, IL-1β that act on the periphery and central nervous system leading to symptoms such as immobility and/or lethargy, piloerection, drowsiness, and ptosis, this symptoms are knowing as sickness behavior 61 , which may be accompanied by physiological changes such as hypothermia. The sickness behavior reduces with the process of resolving inflammation, but motivational deficits may last longer, such as anxiety. Given this, we can infer that the reduction in α 2 Na + / K + -ATPase activity interferes with sickness behavior, since we observed a less locomotion and hypothermia (Fig. 1a,b). However, there was no motivational change as observed by the open field test (Fig. 9a,b).

Discussion
Neuroinflammation is a critical factor in neurodegenerative diseases, including Parkinson's disease and Alzheimer's disease. We used a gene modified mouse with a knock-in mutation to investigate the role of the vital membrane ion pump Na + /K + -ATPase in mediating neuroinflammation. In recent years, there has been increasing evidence that the pump, the Na + /K + -ATPase, serve not only to maintain the electrochemical gradient across the cell's plasma membrane, but are highly involved in regulating intracellular signal cascades. www.nature.com/scientificreports/ We report that mice that are haploinsufficient for the astrocyte-enriched α 2 Na + /K + -ATPase isoform (α 2 +/G301R mice) have a reduced proinflammatory response to LPS, accompanied by a reduced hypothermic reaction. Following administration of LPS, the gene expression of TNF-α, IL-1β, and IL-6 was reduced in the hypothalamus and hippocampus from α 2 +/G301R mice compared to WT littermates (Fig. 10). The central nervous system regulates systemic inflammatory responses to endotoxin through humoral mechanisms via activation of the hypothalamic-pituitary-adrenal pathway, as well as through the activation of efferent www.nature.com/scientificreports/ vagal nerve pathways by reducing the production of pro-inflammatory cytokines produced by macrophages 62,63 . Therefore, we can assume that neuroinflammation reduction observed in the α 2 +/G301R animals could be due to a possible interference in the neuroimmune pathways, however studies must be carried out to understand this interference. Furthermore, activation of the immune system by LPS leads to production and release of proinflammatory cytokines such as TNF-α, IL-1β that act on the periphery and central nervous system leading to symptoms such as immobility and/or lethargy, piloerection, drowsiness, and ptosis, this symptoms are knowing as sickness behavior, which may be accompanied by physiological changes such as hypothermia. The sickness behavior reduces with the process of resolving inflammation, but motivational deficits may last longer, such as anxiety. Given this, we can infer that the reduction in α 2 Na + /K + -ATPase activity interferes with sickness behavior, since we observed a less locomotion and hypothermia (Fig. 1a,b). However, there was no motivational change as observed by the open field test (Fig. 9a,b). The α 2 +/G301R mice experienced increased expression of the gene encoding an antioxidant enzyme, Nrf2, in hippocampal astrocytes.
Here, we demonstrated by complementary in vitro and in vivo experimentation that the production of proinflammatory cytokines upon LPS stimuli is highly dependent on the astrocyte-specific α 2 isoform of Na + / K + -ATPase. α 2 Haploinsufficiency differentially altered the production of TNF-α, IL-6, and IL-1 in LPS-treated astrocytes. This appeared to be associated with a reduced Tlr4 expression.
The LPS-induced immune response in hypothalamus and hippocampus is different and may very well associated differential pathway activation promoted by TLR4, such as NFkB and MAPK, but may also involve other pathways 64 , yet to be elucidated. Interestingly, this caused a decrease in the hypothermic response in α 2 +/G301R mice compared with their α 2 +/+ littermates, which might be associated with a reduced expression of Nrf2, a mediator of inflammatory pathways 50,51 , in the hypothalamus in response to LPS. The LPS induced impairment of fear memory and anxiety-like behaviours observed in α 2 +/+ mice was not affected by α 2 isoform haploinsufficiency. Hypothermia is a thermoregulatory response to systemic inflammation that can be induced in rodents in response to systemic LPS challenge 65,66 . Although the molecular mechanisms of LPS-induced hypothermia are Figure 10. Schematic drawing of the proposed action upon LPS treatment in astrocytes expressing the α 2 Na +/ K + -ATPase (noted α 2 NKA in the figure). (a) In astrocytes with normal α 2 Na +/ K + -ATPase function, the intracellular responses to LPS will lead to an increasement of several inflammatory genes and cause LPS-related phenotypes in the mouse, including hypothermia, reduced locomotion and impaired spatial memory. (b) In astrocytes with α 2 Na +/ K + -ATPase haploinsufficiency, the intracellular responses to LPS will lead to a significantly reduced expression of several inflammatory genes and compromise LPS-related phenotypes in the mouse, leading to less hypothermia, reduced locomotion and reduced impaired spatial memory. www.nature.com/scientificreports/ still poorly understood, the participation of cytokines has been observed in the development of hypothermia 67 . The reduced cytokine levels we found in α 2 +/G301R mice could explain the lower hypothermic response in these animals following LPS treatment.
Previous findings have shown that ouabain, which acts as a Na + /K + -ATPase inhibitor, can protect motor neurons from mutant SOD1-induced astrocyte degeneration and reduce the neuroinflammation resulting from LPS 36,44 . In addition, haploinsufficiency-causing Na + /K + -ATPase α 2 isoform G301R mutation decreases lesion volume and improves functional outcomes after acute spinal cord injury in mice 35 . Furthermore, silencing of α 2-Na + /K + -ATPase reduces LPS-induced inflammation in glial cells 39 .
We previously studied TNF, IL-6 and IL-10 production in the spinal cord of α 2 +/G301R and α 2 +/+ mice with spinal cord injury 35 . We found upregulation of these cytokines in mice with spinal cord injury compared to uninjured mice but we observed no significant differences between the two genotypes; only IL-1β appeared to display a tendency to be upregulated 35 , but it is important to consider that IL-1β is produced as precursor molecules, pro-IL-1β, which is cleaved in its active form by a caspase family cysteine protease, the IL-1β converting enzyme (ICE) 68 . Although there are different mechanisms proposed for post-translational processing of IL-1β, there is still no definite mechanism. Our data show that LPS challenge promoted a serum increase of IL-1β that was not reversed in a 2 gene modified animals, thus suggesting that deficiency in a 2 activity does not interfere with the release of IL-1β in its mature form. Combined, these previous and our new findings support a role for the α 2 isoform of Na + /K + -ATPase in LPS-induced neuroinflammation, by means of mediating cytokine expression.
LPS induces an innate immune response and the production of various proinflammatory cytokines via TLR4 activation. In view of this, we hypothesized and confirmed that α 2 Na + /K + -ATPase haploinsufficiency prevents an increase in Tlr4 expression after LPS exposure in astrocytes of the hypothalamus and hippocampus.
Recently, a study showed that neuronal activity regulates the astrocytic signalling of the nuclear master transcription factor NRF2 through the secretion of glutamate and other soluble factors 69 . Thus, we hypothesized that α 2 +/G301R mice constitutively exhibit an increase in NRF2 activity that protects these animals from LPS-induced neuroinflammation, as NRF2 is capable of negatively regulating NFkB activity and the consequent reduced expression of pro-inflammatory cytokines. Confirming this hypothesis, we found that Nrf2 expression was upregulated in hippocampus, and Nrf2 expression was further enhanced upon LPS treatment, suggesting this as one mechanism that reduce NFkB activity in cells haploinsufficient for the α 2 isoform.
Moreover, the NRF2-responsive antioxidant factor Nqo-1 mRNA was upregulated in the hippocampus in α 2 +/G301R mice in the absence of inflammatory stimulation, and only marginally increased upon LPS treatment, suggesting the involvement of NRF2 activation in the reduced neuroinflammatory response to LPS exposure in α 2 +/G301R mice. This is in line with the fact that NQO-1 exhibits anti-inflammatory activity by inhibiting the induction of TNF and IL-1β expression by LPS in human monocytes 51,70 .
Peripheral LPS administration promotes NFκB activation in various regions of the central nervous system, leading to the stimulation of proinflammatory cytokines 71 . Our results confirmed that LPS-induced activation of NFĸB occurred by measuring reduced levels of cytoplasmic p65, which increased proinflammatory genes within 4 h in the hippocampus of α 2 +/+ mice. IL-1β has been associated with cognitive impairment during the inflammatory process, and the intrahippocampal administration of IL-1β induces impaired memory consolidation and reconsolidation in rats 72 . Although our results showed a reduction in IL-1β expression in the hippocampus 4 h after LPS administration in α 2 +/G301R animals, we observed that that α 2 haploinsufficiency does not affect the LPS-induced effect on memory impairment and anxiety.
While the present study addressed the regulation of cytokines in relation to the Na + /K + -ATPase in astrocytes from hypothalamus and hippocampus, future studies must delineate the mechanisms in other brain structures as well as the α 2 isoform is expression in astrocytes throughout the brain. Of immediate interest in this context is the synergy between signaling pathways that mediate the α 2 -dependent LPS-induced neuroinflammation. Studies have demonstrated the existence of an important interaction between adipocytes and macrophages for the amplification of the inflammatory response induced by LPS 73 , and knowing that adipocytes express the α 2 Na + / K + -ATPase isoform 74,75 , it is likely that the a2 isoform in adipocytes contributes to the reduction of circulating levels of cytokines, however, this remains to be explored.
In summary, this study provides an undescribed link between the α 2 Na + /K + -ATPase and inflammation signaling in vivo. Overall, regulation of the astrocyte α 2 isoform represents a significant regulator of inflammatory responses in the brain, with makes the α 2 isoform a likely candidate for depressing neuroinflammation, and perhaps also neurodegenerative conditions.

Methods
Experimental animals. Mice were cared for in accordance with the protocols and guidelines that The Danish Animal Inspectorate approved under the Ministry of Food and Agriculture, Denmark (J. No. 2013-15-2934-00815 to KLH). We confirm that all experimental protocols were approved by the Animal Facility and the veterinary surgeon at the Department of Biomedicine, Aarhus University, Denmark.
Our mice were bred on a C57/BL6JRj (Janvier) background. All animals were housed under a reverse light/ dark cycle to prevent daytime experiments from interfering with their normal sleep cycles.
The animals were housed in ventilated cages with 1-3 cage mates under a 12-h light/dark cycle and controlled temperature and humidity and with free access to food and water. The animals were treated with 500 µg/kg LPS (#L2630, O11:B4) (Sigma-Aldrich, St. Louis, MO) or PBS. Four hours after LPS administration, the mice were anaesthetized by isoflurane inhalation and euthanized by decapitation, and the brains were immediately removed Genotyping. Heterozygous α 2 +/G301R mice and α 2 +/+ mice were genotyped 33  . On day 2, 5 mL of supplemented RPMI 1640 medium and 40% L929 conditioned medium was added. On day 4, the non-adherent cells were removed from the flask, the media was replaced, and the remaining adherent cells were maintained in culture for 6 days in 20% L929 conditioned medium. On day 7, the cells were transferred to 24-well plates (4 × 10 5 cells per well) and cultured for 4 h before use. The cells were treated with PBS or LPS (100 ng/mL) for 1, 2, 4, and 6 h. Supernatant samples were collected for the analysis of TNF-α via ELISA.
Body temperature. The rectal body temperature was measured using a rectal probe (TFN 530 SMP Thermometer, Ebro) at the time of the first injection and 4 h after the injection of LPS or PBS.
Measurement of cytokine levels. Four hours after LPS or PBS injection, blood was collected in 15-mL conical tubes and centrifuged at 3,000 rpm for 10 min to obtain the serum. The concentrations of TNF-α (#88-7324), IL-1β (#MLB00C), IL-6 (#M6000B), and IFN-γ (#MIF00) were measured by mouse-specific sandwich ELISA according to the manufacturer's instructions (eBioScience, Santa Clara, California, USA and R&D Systems, Minneapolis, USA) 78 . Briefly, samples were added to coated microwells with antibodies against TNF-α, IL-1β, IL-6, and IFN-γ along with a biotin-conjugated antibody (horseradish peroxidases; Mouse IL-1β Conjugate # 893830), polyclonal antibody specific for mouse IL-1β conjugated to horseradish peroxidase with preservatives (R&D Systems), Mouse IL-6 Conjugate # 892665, polyclonal antibody against mouse IL-6 conjugated to horseradish peroxidase with preservatives (R&D Systems), Mouse IFN-γ Conjugate (# 892666, polyclonal antibody specific for mouse IFN-γ conjugated to horseradish peroxidase with preservatives (R&D Systems). The Mouse TNF-α Conjugate (#88-7324), (eBioScience). The plate was incubated for 2 h at room temperature. The wells were washed, streptavidin-HRP was added to the entire plate, and the plate was incubated for 1 h at room temperature. Subsequently, the wells were washed and then incubated with TMB substrate (Thermo Fisher Scientific, Roskilde, Denmark) solution for 30 min while being protected from light. The reaction was stopped with a stop solution, and the absorbance was measured using a spectrophotometer at 450 nm. The concentrations of the cytokines were measured based on the standard curve.
Protein extraction and immunoblot analysis. The hippocampus and hypothalamus were isolated according to the protocol of the CelLytic NuCLEAR Extraction Kit (Sigma-Aldrich, St. Louis, MO). In brief, the tissues were homogenized in lysis buffer containing 10 mM HEPES (pH 7.9) with 1.5 mM MgCl 2 , 10 mM KCl, 0.1 M dithiothreitol (DTT) solution and a protease inhibitor cocktail, and they were centrifuged at 10,000×g for 20 min. The supernatant representing the cytosolic fraction was transferred to a new tube. The pellet was resuspended in an extraction buffer containing 1.5 μL 0.1 M DTT and 1.5 μL protease inhibitor cocktail. The solution was allowed to stand on ice for 30 min with shaking at brief intervals followed by centrifugation at 20,000×g for 5 min. The supernatant, which contained the nuclear protein fraction, was transferred to a clean chilled tube. The proteins from the cytosolic fractions of the hippocampus and hypothalamus (20 μg) were separated by size by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis. The proteins were blotted onto a nitrocellulose membrane (Pharmacia-Amersham, Amersham, UK) and incubated with anti-α1 (1:500) (a6f.-c, Developmental Studies Hybridoma Bank, USA), anti-α2 (1:5,000) (07674, EMD Millipore, USA), anti-p65 (1:1,000) (CST-8242T, Cell Signalling), and actin (1:1,000) (A2066, Sigma-Aldrich,) primary antibodies overnight at 4 °C 33 . The secondary antibodies included horseradish peroxidase-conjugated pig anti-rabbit and pig anti-mouse (1:2,000) (Dako, Glostrup, Denmark) antibodies. The proteins recognized by the antibodies were revealed via an Amersham ECL Western Blotting Detection Kit, following the instructions of the manufacturer (GE Healthcare, Buckinghamshire, UK). To standardize and quantify the immunoblots, we used the photo documentation system of the LAS 3000 imager (Fujifilm, Tokyo, Japan) and ImageJ software (US National Insti- Then, the tubes were connected to the gentleMACS Octo Dissociator with Heaters, and the gentleMACS 37C_ NTDK_1 programme was used for tissue dissociation. Then, the samples were centrifuged briefly, and the cells were resuspended in D-PBS. The cell suspension was filtered with a 70-μm cell strainer (MACS SmartStrainer), which was washed with 10 mL of D-PBS supplemented with BSA (0.5%). Thereafter, the cell suspension was centrifuged at 300×g for 10 min at room temperature, and the cells were suspended in D-PBS supplemented with BSA (0.5%).

Myelin debris removal.
Prior to the isolation of astrocytes, we performed a debris-removal step using protocols from Miltenyi Biotec's Myelin Removal Kit (catalogue number 130096733). Following dissociation, the cells were incubated for 15 min at 4 °C with Myelin Removal Beads II, following the ratio of 500 mg of brain tissue in 1,800 μL of buffer + 200 μL of myelin removal beads) 45 . The cells were washed with 0.5% BSA in PBS and centrifuged at 300×g for 5 min to remove any unbound spheres from the pellet. After that, the pellet was resuspended in 500 μL of buffer, the suspension was added to a LS column prepared in the MACSMidi magnetic cell separator, and the flow-through was collected. The column was further washed three times with 3 mL of buffer to ensure the removal of the unlabelled cells. The cells retained on the column were eluted in 5 mL of buffer. The flow-through was used in subsequent steps for astrocyte isolation.
isolation of astrocyte cells. The astrocytes were positively selected using the protocol of the Miltenyi Biotec Anti-ACSA-2 Kit (catalogue number 130097678). Up to 1 × 10 7 dissociated cells were suspended in 80 μL of buffer (0.5% BSA in PBS) and incubated with 10 μL of FcR blocking buffer for 10 min at 4 °C, followed by incubation with 10 μL of ACSA-2 MicroBeads for 10 min at 4 °C. Then, the cells were washed with 1 mL of buffer (0.5% BSA in PBS) and centrifuged at 300×g for 10 min to remove the excess beads from the solution. After the removal of the lavage solution, the pellet was resuspended in 500 μL of buffer, and the suspension was added to a prepared LS column installed in the MACSMidi magnetic cell separator. The column was washed with 3 mL of buffer three successive times to remove the unlabelled cells. After the column was removed from the magnetic separator, the astrocytes were eluted in 5 mL of buffer. The number of cells was then determined, and total RNA was extracted.

Reverse transcription quantitative PCR (RT-qPCR).
Total RNA was isolated from astrocytes isolated from astrocytes from the hippocampus and hypothalamus with the RNeasy Plus Mini Kit (Qiagen), according to the manufacturer's instructions. Complementary deoxyribonucleic acid (cDNA) was generated from 500 ng total RNA using the PrimeScript RT Reagent Kit (Takara BIO INC). The Tnfα, Il-1β, Il-6, Tlr4, Nrf2, Ho-1, and Nqo-1β, Cox1/2, Ptgds and β-actin (our reference) gene expression levels were measured via quantitative PCR (qPCR) using the TaqMan gene expression assays (Thermo Fisher Scientific) noted below:

β-Actin Mm02619580_g1
Real-time PCR analysis was performed in triplicate, with each reaction including 20 ng of cDNA, 5 μL of 2 × LightCycler 480 Probed Master Mix. (Roche, Basel, Switzerland), 0.5 μL of TaqMan Gene Expression Assay, and nuclease-free water up to a final volume of 10 μL. The PCRs were run in the Roche LightCycler 96-well system with the following protocol: 2 s at 50 °C for uracil-DNA glycosylase enzyme activation, 10 min at 95 °C for DNA Polymerase activation, 45 cycles of 15 s at 95 °C for denaturation, and 1 min at 60 °C for annealing and extension followed by a final denaturation step of 95 °C for 5 min 79 www.nature.com/scientificreports/ were set relative to the expression of the reference gene via the delta-delta-Ct method 81 . As a negative control, RT-PCRs with no template were used.

Behavioural analysis.
Open field test. The mice were placed in the centre of an open-field apparatus (50 × 50 cm) (Stoelting Europe; Dublin, Ireland) and monitored for 15 min using ANY-maze software V4.99 (Stoelting, USA) 82 . The system automatically recorded the total distance travelled (m) and the time (s) spent in the centre zone. Each mouse (n = 10-12 for each group) was tested once 4 h after PBS or LPS injection, and the open-field setup was cleaned with 70% ethanol and wiped with paper towels between each trial.
Passive avoidance test. The training was initiated on the acquisition day, 24 h after PBS or LPS injection. Each mouse was placed in a brightly lit compartment with an electronically controlled door leading into a dark compartment 82 . The latency (s) for the mouse to enter the dark compartment was recorded. Once in the dark compartment, the door closed, and the mouse received an electric shock (0.42 mA for 1 s). The test was performed 72 h after PBS or LPS injection, the mouse was reintroduced to the same brightly lit compartment, and the latency to enter the dark compartment was recorded as an indicator of memory for the shock.
Statistical analysis. The qPCR results were analysed via the delta-delta-Ct method, according to Schmittgen and Livak (2008) and calculated using REST 2009 Software (Qiagen, Duesseldorf, Ger). Normality was assessed through the D' Agostino and Pearson omnibus normality test, and, for parametric analyses. Parametric analyses were conducted through two-way ANOVA followed by Tukey's post-test. Non-parametric analyses were conducted through Kruskal-Wallis test followed by Dunn's post hoc test. Differences were considered to be significant at p < 0.05, and all results are expressed as the mean ± standard error of the mean (SEM) of the indicated number of experiments. All analyses were performed using the Prism 6 software package (GraphPad Software, San Diego, CA, USA).