SIRT3 Protects Against Cognitive Dysfunction Induced by Sepsis-Associated Encephalopathy via JNK/p66Shc-Regulated Mitochondrial Apoptosis in Mice

Sepsis-associated encephalopathy (SAE) is one of the severe central nervous system complications. Oxidative stress and synaptic dysfunction were involved in cognitive impairment induced by SAE. The mitochondrial nicotinamide adenine dinucleotide (NAD + ) dependent deacetylase, sirtuin3 (SIRT3), plays a critical role in regulating mitochondrial function. The aim of this study was to evaluate the effect of SIRT3 in cognitive dysfunction induced by SAE.


Results
SIRT3 expression was diminished in hippocampus of mice after LPS treatment. SIRT3-de ciency contributed to more severe contextual memory loss and synaptic dysfunction, decreased ratio of Bcl-2/Bax and increased Cyt C release to cytoplasm in hippocampus compared with wild-type controls. In HT22 cells, lysine acetylation levels of CypD were signi cantly increased after Mi-sup exposure and further enhanced with 3-TYP (SIRT3 deacetylation inhibitor) pretreatment, in association with the accumulation of ROS, declined MMP and increased MPTP opening, as well as the increased mitochondrial Cyt C release and phosphorylation levels of mitochondrial JNK and p66Shc-Ser36. SIRT3 overexpression restored CypD lysine acetylation levels and MPTP opening in HT22 cells after Mi-sup exposure and reduced mitochondrial JNK and p66Shc activation.

Conclusion
Taken together, our results showed that SIRT3-mediated CypD deacetylation was involved in LPS-induced hippocampal synaptic dysfunction, via ROS accumulation, declined MMP, increased MPTP opening, mitochondrial Cyt C release and mitochondrial apoptosis of hippocampal neuron via JNK/p66Shc pathway. Our results revealed that SIRT3 may be a promising therapeutic and diagnostic target for cognitive dysfunction induced by SAE.

Background
It is universally acknowledged that severe sepsis is related to multi-system damage including brain dysfunction, and sepsis-associated encephalopathy (SAE) is one of the severe central nervous system complications in patients in intensive care units for its cognitive dysfunction [1,2]. SAE is associated with increased morbidity and mortality, and it can in uence the short-term outcome and long-term recovery [3]. Oxidative stress and apoptosis in hippocampal neurons have been reported to involve in cognitive impairment in SAE [4,5].
One of the most important pathogenesis of SAE is mitochondrial dysfunction [6,7,8]. Mitochondrial is reported to be the major source of reactive oxygen species (ROS) and initiation of apoptosis [9,10,11]. Normal mitochondria produce ATP to support cellular metabolism, whereas damaged mitochondria release pro-apoptotic factors to initiate programmed cell death. Several studies found that in ammation reduces mitochondrial membrane potential and increases oxidative stress [12,13], however, the mediator of these in ammation-induced pathological alterations in the mitochondria has not been clearly identi ed.
Several studies showed that 66kDa Src homology 2 domain-containing protein (p66Shc) plays a speci c role in mitochondrial ROS generation and translates oxidative stress signals into cell apoptosis [24,25].
Modi ed p66Shc could translocate into the mitochondria and oxidizes cytochrome C (Cyt C) to yield H 2 O 2 , which in turn initiates cell apoptosis [26,27]. However, the effects of p66Shc on sepsis-induced mitochondrial dysfunction in central nervous system are largely unknown, and whether SIRT3 could interact with p66Shc in the process of mitochondrial dysfunction were still unexplored. In the present study, we investigated the effects of SIRT3-mediated cognitive impairment induced by SAE and the following mitochondrial mechanism.

Animals
Procedures involving animals were approved by the Animal Experimentation Ethics Committee of Ruijin Hospital, and were performed strictly according to the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. Male wild-type (WT) and SIRT3-knockout (SIRT3-/-) mice were used for the study. The mice were kept on a 12-hour light/dark cycle with free access to food and water. SAE mouse model SAE was induced in mice between 6 to 8-week-old by intraperitoneal (i.p.) administration of 10 mg/kg LPS (Escherichia coli, serotype 0111: B4, catalog L2630) from Sigma-Aldrich, similar to previous studies [28,29,30]. Animals were divided into 4 groups (n = 8 each): WT Control (WT Con) group, WT LPS group, SIRT3-/-Control (SIRT3-/-Con) group and SIRT3-/-LPS group. Mice were euthanized 24 hours after LPS administration. The hippocampal tissues were isolated and stored at -80 °C for further analysis.

Behavioral tests
Behavioral tests included the open eld test (OFT) and fear conditioning test (FCT). Spontaneous locomotor activity of the mice in OFT was evaluated as previously described [31]. The apparatus is made up of a white polyester resin chamber (40 cm x 40 cm x 40 cm). Mice were placed in the center of the arena and allowed to explore for 300 s, and the total distance moved and the time spent in the center were recorded. The chamber was cleaned with 75% ethanol after each trail.
The FCT was performed following the protocol as previously described [32,33]. On the day of training, mice were placed into an enclosed training chamber and allowed to explore for 180 s. Mice were then exposed to a tone (30 s, 70 dB, 3 kHz), followed by a 2 s foot shock (0.45 mA). Afterward, mice were left in the chamber for additional 30 s. The training period was repeated 5 times. 24 hours after the training session, mice were re-exposed in the same chamber for 300 s for the contextual fear conditioning test (a hippocampus-dependent task). 4 hours later, mice were placed into a novel chamber which was different in shape, color, and smell from the original chamber for the cued fear conditioning test (a hippocampusindependent task). After 180 s exploratory period, a training tone (30 s, 70 dB, 3 kHz) was applied for another 30 s and repeated 4 times at an interval of 40 s, and the time of freezing was recorded. The chamber was cleaned with 75% ethanol after each trial.
Nissl's staining Mice were anesthetized and transcardially perfused with pre-cooled heparinized physiological saline and 4% paraformaldehyde. After dehydration with gradient ethanol elution and cleared in xylene, cerebral tissues were sliced into 10 μm coronal sections using a cryostat (Leica CM1860 UV, Leica) and stored at -20 °C. Nissl staining (Beyotime Institute of Biotechnology, Shanghai, China) was conducted according to the manufacturer's instructions. In brief, the sections were soaked in 1% toluidine blue for 5 min at 50 °C, washed by double distilled water, and dehydrated with gradient ethanol elution. The slice sections were observed under a uorescent microscope.

TUNEL staining
The TdT-mediated dUTP Nick-End Labeling (TUNEL) staining was used to evaluate neuronal apoptosis in the hippocampus. In Situ Apoptosis Detection Kit was used with manufacturer's instruction (Beyotime Institute of Biotechnology, Shanghai, China). The nuclei were stained with DAPI, and the uorescent microscopy was used to evaluate neuronal apoptosis.
Anti-rabbit (catalog 147626) and anti-mouse (catalog 146552) secondary antibodies from Jackson ImmunoResearch were added and incubated for 2 h at 37 °C, and sections were then washed three times with PBST. After nal washing, sections were protected with coverslips, with the nucleus visualized with DAPI (catalog C1005, Beyotime Institute of Biotechnology, Shanghai, China). The brain tissues were observed and analyzed using uorescence microscope.
HT22 cells were xed in 4% paraformaldehyde in PBS for 15 min at room temperature. Cells were then permeabilized in 0.1% Triton X-100 in PBS for 15 min and blocked with 10% donkey serum in PBS for 1 h. The culture dishes were then incubated with anti-SHC phosphor S36 (catalog 54518, Abcam) and anti-TIMM44 (catalog HPA043052, Sigma) overnight at 4 °C, and for 1.5 h at room temperature for secondary antibody incubation. After nal washing, the nucleus visualized with DAPI (catalog C1005, Beyotime Institute of Biotechnology, Shanghai, China). The cells were observed using confocal microscopy.

NAD + and NADH measurements
Fresh mouse hippocampal brain tissue (20 mg) was collected and homogenized in 400 μL NAD + /NADH extract buffer on ice. Total NAD + and NADH were tested using an NAD + /NADH assay kit (Beyotime Institute of Biotechnology, Shanghai, China) according to the manufacturer's protocol [34]. NAD + /NADH ratio was calculated based on the valve of total NAD + and NADH (NAD + = total NAD + -NADH).

Cell culture
Hippocampal neuronal cells (HT22) and microglia cells (BV2) were respectively cultured in Dulbecco's modi ed Eagle's medium (DMEM), 10 % fetal bovine serum (FBS) and 1 % streptomycin and penicillin in incubator containing 5 % CO 2 at 37 °C. To induce LPS-induced hippocampal neuron stress in vitro, BV2 cells were rst exposed to LPS (1 μg/ml) for 24 h, and the supernatants of LPS-stimulated microglia (Misup) were collected as previously described [35]. HT22 cells were then cultured in the Mi-sup for 24 h.
Co-immunoprecipitation (Co-IP) assays HT22 cell lysate was harvested, and Co-IP assay was performed using an Immunoprecipitation kit from Proteintech. Brie y, 2 μg of antibodies against the proteins of interest or a negative control IgG of the same species were added to the cell lysate. Then the IP buffer was mixed with protein A/G beads. The immunoprecipitate was separated by SDS-PAGE and analyzed by western blotting.

Plasmids and transfection
The pDC315-SIRT3 was designed and purchased from Shanghai Gene-Pharma Co. (Shanghai, China). The plasmid was transfected into HEK293 cells and the supernatant was collected. The viral supernatant was identi ed and ampli ed to obtain adenovirus-SIRT3. The plasmids were transfected into the HT22 cells to overexpress SIRT3 with Lipofectamine 2000 (Thermo Fisher Scienti c, Inc.) according to the manufacturer's protocol. Null vector transfection was used as the control group.

ROS staining
Total cellular ROS production was determined by detecting the uorescence intensity of dichloro uorescein diacetate (DCFH-DA) according to the manufacturer's protocol [36] (Reactive oxygen species assay kit, Beyotime, China). Brie y, HT22 cells were cultured in 12-well plates and incubated with freshly prepared DCFH-DA reagent in dark at 37 °C for 30 min. HT22 cells were washed three times with PBS and analyzed using uorescence microscope.
JC-1 staining JC-1 staining was used to evaluate MMP according to the manufacturer's protocol (Mitochondrial membrane potential assay kit with JC-1, Beyotime, China). Pretreated HT22 cells were washed three times with PBS, followed by the additional 1 ml JC-1 working solution per sample and incubated at 37 °C for 20 min. After incubation, HT22 cells were washed twice with JC-1 staining buffer and detected using uorescence microscope.
MPTP opening MPTP opening was measured by calcein-AM following the manufacturer's protocol (Mitochondrial permeability transition pore assay kit, Beyotime, China). HT22 cells were co-stained with 1 μM calcein-AM, 1mM CoCl 2 , and 200 nM Mitotracker for 20 min at 37 °C. After staining, HT22 cells were washed with PBS and captured by uorescent microscope. ImageJ software was used for quanti cation and measurement of uorescent intensity.

Western blot analysis
Frozen hippocampus sample were homogenized in ice-cold extraction buffer. Homogenates were centrifuged at 12,000 g for 15 min at 4 °C, the nal supernatants were obtained. In contrast to frozen hippocampus tissues, HT22 cells were washed twice with PBS and lysed in an extraction buffer. Mitochondrial extracts were obtained with a mitochondrial isolation kit (Beyotime Institute of Biotechnology, Shanghai, China). Protein concentrations were determined by BCA Protein Assay Kit (Beyotime Institute of Biotechnology, Shanghai, China). Equal amounts of protein were separated on a 10% SDS-PAGE and transferred to a PVDF membrane. After being blocked, the membranes were incubated with correspondent primary antibody at 4 °C for overnight. Antibodies against αtubulin (catalog 11244-1-AP), GAPDH (catalog 10494-1-AP), Bcl-2 (catalog 12789-1-AP), Bax (catalog 50599-2-Ig), Cytochrome C (catalog 66264-1-Ig), SHC (catalog 10054-1-AP) were obtained from Proteintech Group. Antibodies against VDAC (catalog D73D12), SIRT3 (catalog D22A3), PSD95 (catalog D27E11), BDNF (catalog 47808S) were from Cell Signaling Technology. Anti-Cyclophilin 40 (catalog 181983), anti-acetyl Lysine (catalog 80178) and anti-SHC phosphor S36 (catalog 54518) were from Abcam. Anti-JNK (catalog 7345) and anti-p-JNK (catalog 6254) were from Santa Cruz Biotechnology. After washing, the membrane was incubated with secondary antibodies (catalog SA00001-2, Proteintech Group) for 80 min at room temperature. Immunoreactive protein bands were detected using chemiluminescence reagents (Beyotime Institute of Biotechnology, Shanghai, China). In the present study, α-tubulin and GAPDH were used as loading control in hippocampus homogenates and whole-cell extracts, while VDAC was used as loading control in both hippocampus and cell mitochondrial extracts.

Statistical analysis
Data were analyzed using GraphPad Prism 8 software and the data were presented as mean ± standard error. Statistical signi cance was evaluated by the Student's t-test for comparing two experimental groups. For datasets having multiple groups, statistical signi cance was evaluated using a one-way analysis of variance (ANOVA) followed by Tukey's test for the post hoc analysis. Results were considered as signi cant at P < 0.05.

SIRT3 expression is diminished in LPS-induced mice hippocampus
We rst determined whether SIRT3 expression levels were altered in the hippocampus of LPS-treated mice. Western blot analysis con rmed that SIRT3 was signi cantly reduced in both whole-hippocampal homogenates and mitochondrial fractions 24 h after 10 mg/kg LPS i.p. (Fig. 1A and 1B). This decreased level of SIRT3 in LPS-treated mice was also evident at the mRNA level in hippocampal homogenates (Fig.  1C). Consistent with the western blot analysis, the immuno uorescence staining con rmed that the expression of SIRT3 was signi cantly decreased (Fig. 1D).

SIRT3 de ciency exacerbates LPS-induced hippocampus-dependent cognitive impairments
The severity of cognitive impairments was evaluated in LPS-treated WT and SIRT3-/mice. Mice received OFC on day 5 and FCT on day 6 and 7 after LPS exposure ( Fig. 2A). Compared with WT Con mice, SIRT3 de ciency resulted in decreased survival rate within 7 d after LPS exposure (Fig. 2B). Compared with WT LPS mice, SIRT3 de ciency resulted in declined contextual freezing time after LPS exposure (Fig. 2C). Compared with WT Con mice, both WT LPS and SIRT3 de ciency mice contributed to the decreasing of post-tone freezing time in the auditor-cued fear conditioning test, however, there was no difference in the freezing time between two SIRT3 de ciency groups (Fig. 2D). The OFT was performed to determine whether the locomotor activity of mice was altered. There was no difference between four groups for the total distance (Fig. 2E) and the time spent in the center of arena (Fig. 2F). The results indicated that SIRT3 de ciency exacerbates LPS-induced hippocampus-dependent cognitive impairments.

SIRT3 de ciency exacerbates hippocampal neuronal apoptosis and synaptic dysfunction in LPSinduced SAE
We used Nissl's staining to assess whether SIRT3 de ciency lead to hippocampal neuron stress after LPS exposure. In dentate gyrus (DG) region of the hippocampus, WT Con group hardly had any neuronal damage, while the decreasing numbers of surviving neurons and irregular arrangement neurons were observed in the WT LPS group. In comparison with WT LPS group, remarkably increased numbers of Nissl-positive neurons and signi cantly decreased thickness of DG region granule cell layer were detected in SIRT3-/-LPS group, which indicating exacerbated hippocampal neuronal damage (Fig. 3A and 3C).
Then, TUNEL staining was utilized to explore the hippocampal apoptosis cells at 24 h post LPS induction. The numbers of TUNEL-positive cells in DG region were signi cantly increased in SIRT3-/-LPS group compared to WT LPS group. While both WT Con and SIRT3-/-Con group hardly observed TUNEL-positive cells ( Fig. 3B and 3D).
Synaptic dysfunction and degeneration contribute to the deterioration of memory performance [37,38].
Synapse-associated proteins (postsynaptic density protein 95, PSD95; brain derived neurotrophic factor precursor, pro-BDNF) were assessed by western blot in hippocampus (Fig. 3E). PSD95 levels were remarkably declined in WT LPS group in comparison with WT Con group, and showed further decline in SIRT3-/-LPS mice. Pro-BDNF levels were signi cantly increased in SIRT3 de ciency mice in comparison with WT mice, however, the levels within WT and SIRT3-/groups had no signi cant differences. All these results indicated that SIRT3 de ciency exacerbated hippocampal neuron apoptosis and synaptic dysfunction in LPS-induced SAE in mice.

SIRT3 de ciency decreased the hippocampal NAD + /NADH ratio and accelerated mitochondrial apoptosis in LPS-induced SAE
Since SIRT3 is NAD + -dependent, we then tested the effect of LPS exposure on NAD + /NADH ratio in fresh hippocampus. After LPS treatment, the NAD + content was signi cantly declined in WT LPS mice compared with WT Con mice, and was further declined in SIRT3-/-LPS mice (Fig. 4A). Similar trends of NAD + /NADH ratio were also witnessed in four groups (Fig. 4B). These results indicated that hippocampal mitochondrial function was damaged after LPS treatment, and this damage was further exacerbated in SIRT3 de ciency mice.
Mitochondrial apoptosis-associated protein B-cell lymphoma 2 (Bcl-2), Bax and cytochrome C (Cyt C) in mitochondrial and cytosol extracts of hippocampus were accessed by western blot respectively. VDAC, the distinctive protein of mitochondrial, was used as normalization for mitochondrial extracts. Compared with WT LPS mice, the ratio of Bcl-2/Bax in mitochondrial extracts was sharply decreased in SIRT3-/-LPS mice, while the ratio in cytosol was increased. Similar trends were showed on the change of Cyt C in these two groups (Fig. 4C and 4D). These results showed that increased Bcl-2 and cytochrome C were released to cytosol from mitochondrial after LPS exposure in SIRT3-/mice, indicating accelerated mitochondrial apoptosis in LPS-induced neurological impairment.

SIRT3-mediated deacetylation of CypD contributed to increased ROS generation, declined MMP and increased MPTP opening in HT22 cells with LPS-induced microglia supernatants
Subsequently, we determined how SIRT3 de ciency promoted mitochondrial dysfunction in hippocampal neurons. Immuno uorescent staining revealed co-localization of SIRT3 and NeuN in hippocampus (Fig.  5A). HT22 cells (a hippocampal neuronal cell line) were then taken for experiments.
To mimic LPS-induced hippocampal neuron stess in vitro, BV2 cells (a microglia-derived cell line) were rst exposed to LPS (1 μg/ml) for 24 h, and the supernatants of LPS-stimulated microglia (Mi-sup) were collected. HT22 cells were then cultured in the Mi-sup for 24 h. In HT22 cells, SIRT3 levels were decreased with Mi-sup rather than LPS (1 μg/ml) treatment. Mi-sup treatment downregulated PSD95 and upregulated pro-BDNF levels (Fig. 5B). Therefore, LPS did not directly alter PSD95 and pro-BDNF levels, but the Mi-sup attributed to synaptic dysfunction in HT22 cells. Mi-sup were then used in the following experiments.
Cyclophilin D (CypD) is the key component of mitochondrial MPTP and lysine acetylation (Ace-K) levels of CypD leads to an increased susceptibility to MPTP opening [39,40]. Compared with control group, CypD Ace-K levels were signi cantly increased after Mi-sup exposure (Fig. 5C). Next, CypD Ace-K levels were evaluated in HT22 cells with different SIRT3 deacetylation activity, by means of pretreated with 50 μM 3-TYP (SIRT3 deacetylation activity inhibitor) or transfected with SIRT3 overexpression plasmid (SIRT3-oe). CypD Ace-K levels were increased pretreated with 3-TYP after Mi-sup exposure, while decreased with SIRT3 overexpression, indicating that CypD Ace-K levels were signi cantly altered with different SIRT3 deacetylation activity in HT22 after Mi-sup treatment ( Fig. 5D and 5E).
Oxidative stress of HT22 cells was evaluated by ROS staining. Compared with control group, ROS generation were increased with Mi-sup treatment, the effects of which were further increased with 3-TYP pretreatment. After SIRT3 overexpression, the effects of ROS generation were signi cantly inhibited ( Figure 5F). MPTP opening and MMP were evaluated by observing the relative uorescence intensity of calcein AM, Co 2+ and JC-1 in four groups (Fig. 5G and 5H). MPTP opening was increased in HT22 cells after Mi-sup treatment, while it further increased in Mi-sup+3-TYP group, and the effects were attenuated in Mi-sup+SIRT3-oe group. These results indicated that SIRT3-mediated deacetylation of CypD contributed to increased ROS generation, declined MMP and MPTP opening in HT22 cells with LPSinduced microglia supernatants.
6. SIRT3 deacetylation activity contributed to synaptic dysfunction and mitochondrial apoptosis in HT22 cells with LPS-induced microglia supernatant via JNK/p66Shc pathway Several pathways capable of regulating ROS production and mitochondrial apoptosis, among which the phosphorylation of serine (Ser) 36 in p66Shc has been indicated as a key regulatory step. P66Shc translocate to the mitochondria and oxidizes Cyt C to yield H 2 O 2 , which in turn initiates cell apoptosis.
Studies indicated that p-p66Shc-Ser36 was regulated by JNK [41]. Meanwhile, JNK1/2 can translocate to the mitochondrial under stress resulting in accelerated p66Shc phosphorylation, which in turn increased ROS production and cell death [42,43]. But it remains unclear whether SIRT3 deacetylation activity was involved in this key process.
The levels of p-p66Shc-Ser36 and p-JNK in hippocampal mitochondrial extracts were remarkably increased in WT LPS mice in comparison with WT Con mice, and both levels were further increased in SIRT3-/-LPS mice, indicating that SIRT3 de ciency exacerbates LPS-induced synaptic dysfunction via the activation of mitochondrial JNK and p66Shc (Fig. 6A).
In HT22 cells, Mi-sup treatment signi cantly downregulated PSD95 and upregulated pro-BDNF levels, and these changes were partly abolished by SIRT3 overexpression. The ratio of Bcl-2/Bax and Cyt C levels were decreased after Mi-sup treatment, and these phenomena were partly reversed by SIRT3 overexpression (Fig. 6B). The ratio of p-p66Shc-Ser36/p66Shc and p-JNK/JNK in HT22 mitochondrial extracts were increased after Mi-sup treatment. After SIRT3 overexpression, the mitochondrial p-p66Shc-Ser36 and p-JNK levels were sharply decreased after Mi-sup treatment. These results indicated that SIRT3 can protect HT22 cells from mitochondrial apoptosis after Mi-sup treatment by regulating mitochondrial JNK and p66Shc activation (Fig. 6C).
In HT22 cells, the changes of PSD95 and pro-BDNF levels after Mi-sup treatment were further exacerbated in Mi-sup+3-TYP group, and the levels could be reversed by 30 μM SP600125 (JNK inhibitor) pretreatment. The ratio of Bcl-2/Bax and Cyt C levels were further decreased in Mi-sup+3-TYP group, which were reversed by SP600125 pretreatment (Fig. 6D). The increasing of p-p66Shc-Ser36/p66Shc and p-JNK/JNK in HT22 mitochondrial extracts were further exacerbated in Mi-sup+3-TYP group. Then, we witnessed the level of p-p66Shc-Ser36 was inhibited by SP600125 pretreatment, which indicated that p66Shc-Ser36 phosphorylation was regulated by JNK activation after Mi-sup exposure (Fig. 6E). To corroborate these results, mitochondrial p-p66Shc-Ser36 accumulation was determined by immuno uorescence (Fig. 6F). In agreement with western blotting results, p-p66Shc-Ser36 co-localized with TIMM44 in HT22 cells was witnessed after Mi-sup treatment, and the phenomenon of co-localization of p-p66Shc-Ser36 with mitochondrial was further increased in Mi-sup+3-TYP group, which was decreased after SIRT3 overexpression. These data indicated that JNK/p66Shc pathway was involved in SIRT3 deacetylation activity mediated mitochondrial apoptosis in HT22 cells with LPS-induced microglia supernatants.

Discussion
Our study showed that SIRT3 de ciency exacerbates sepsis-induced hippocampus-dependent cognitive impairments via enhancing oxidative stress and synaptic dysfunction in mice. Our study showed that SIRT3-mediated CypD deacetylation was involved in LPS-induced hippocampal synaptic dysfunction, via ROS accumulation, declined MMP, increased MPTP opening, mitochondrial Cyt C release and mitochondrial apoptosis of hippocampal neuron via JNK/p66Shc pathway. Our results revealed that SIRT3 may be a promising therapeutic and diagnostic target for cognitive dysfunction induced by SAE.
WT mice treated with 10 mg/kg LPS in our study showed declined freezing time in both contextual and cue memory, suggesting that our SAE model was successful. Compared with SIRT3-/-Con group, the mice in SIRT3-/-LPS group showed signi cantly declined hippocampal-dependent cognitive dysfunction, indicating that hippocampus was more susceptible to be affected by LPS exposure.
SIRT3 has been reported to regulate almost every aspect of mitochondrial function like mitochondrial biogenesis and dynamics [44,45,46,47], ROS metabolism, ATP production and maintenance of mitochondrial integrity [48,49]. It was expressed maximally in metabolically active tissues like brain, heart, kidney, liver, brown adipose tissue and skeletal muscles. Altered mitochondrial function in the brain, like declined ATP production and increased oxidative damage from ROS, was witnessed in different SIRT3 deacetylation activity under different stress [50,51]. SIRT3 de ciency mice showed mild hippocampal-dependent cognitive dysfunction and morphologic change of neurons in DG region, but signi cantly exacerbated after LPS treatment in our study. This phenomenon indicated that SIRT3 knockout mice exhibited almost normal physiology under normal condition, but tended to develop deteriorated brain dysfunction and poor remote memory under stress [52,53].
At the cellular level, previous studies found that HT22 cells responded to Mi-sup rather than LPS in cell viability and neuron apoptosis. LPS did not increase Ca 2+ and neither promoted neuronal apoptosis in cultured primary hippocampal neurons expressing low levels of TLR4 [54]. LPS may not directly impair neurons but induce neuroin ammation by rstly activating microglia. Therefore, LPS-stimulated microglia supernatant was used for the condition medium in our study. We witnessed increased CypD Ace-K levels, increased ROS generation, decreased MMP and increased MPTP opening and declined protective synaptic protein levels in HT22 cells after Mi-sup exposure, which further deteriorated with 3-TYP pretreatment. ROS production and low MMP are two inducers to MPTP opening. A long-lasting MPTP opening can lead to irreversible consequesces, including matrix swelling, dissipation of mitochondrial potential, ATP hydrolysis into ADP, and uncoupling of oxidative phosphorylation, causing cell death [55]. These results indicated that hippocampal neurons with decreased SIRT3 deacetylation activity were prone to suffer from mitochondrial apoptosis and synaptic dysfunction after Mi-sup exposure. However, the extracellular microenvironment and the speci c interaction between hippocampal neuron and microglia remain to be explored.
Our study found that LPS exposure activated JNK and induced mitochondrial p66Shc-Ser36 phosphorylation. P66Shc plays an important role in the generation of mitochondrial ROS and serves as antioxidants in many pathological conditions and disease [56,57]. PKCβ has been proposed as p66Shc-Ser36 kinase but JNKs may also phosphorylate this residue. Studies showed application of speci c inhibitors caused a pronounced decrease in p66Shc-Ser36 phosphorylation only in the case of JNK1/2, and p-p66Shc-Ser36-dependent ROS production and mitochondrial apoptosis was sharply diminished in JNK1/2 de cient MEFs [58], indicating JNK was critical for p66Shc activation. In our study, we showed that phosphorylation of p66Shc-Ser36 and JNK were regulated by SIRT3 deacetylation activity, and phosphorylation of p66Shc-Ser36 was regulated by JNK activation. These data indicated that SIRT3mediated JNK/p66Shc pathway was involved in mitochondrial apoptosis and synaptic dysfunction in HT22 cells with LPS-induced microglia supernatants.

Conclusions
In summary, the present study suggested that SIRT3-mediated CypD deacetylation was involved in LPSinduced hippocampal synaptic dysfunction, via ROS accumulation, declined MMP, increased MPTP opening, mitochondrial Cyt C release and mitochondrial apoptosis of hippocampal neuron via JNK/p66Shc pathway. Our results revealed that SIRT3 may be a promising therapeutic and diagnostic target for cognitive dysfunction induced by SAE. The data and materials of this study are available from the corresponding authors for reasonable requests.

Competing interests
The authors declare that they have no competing interests.