Male C57/BL6 mice aged 8 weeks (20–25 g) were purchased from Charles River Co., Ltd. (China, Beijing). For generation of conditional MD2 knockout (cKO) mouse lines with deletion of exon 1, male homozygous MD2^fl/fl mice on the C57BL/6J genetic background were crossed with female MD2^fl/+ mice carrying Cre. Heterozygous MD2^fl/+ mice were generated by crossing female MD2 cassette (MD2 cassette/+) mice carrying a targeted MD2 allele on the C57BL/6J genetic background with male protamine-Flp mice. CaMKII-Cre (#5359) mice with a C57BL/6J genetic background were obtained from The Jackson Laboratory. PCR genotyping of CaMKII cKO mice was performed using the following set of oligonucleotide primers: floxed allele, forward, 5′-TCT​CAG​TAC​TTC​GGA​GGC​AGG​ATG​A-3′, reverse, 5′-TAC​CCT​CCT​TTC​ACT​CCC​TCG​TTC​C-3'; Cre allele, forward, 5′-GGT​TCT​CCG​TTT​GCA​CTC​AGG​A-3′, reverse, 5′-CCTGTTGTTCA GCTTGCACCAG-3'. Mice were housed under standard conditions (22 ± 2°C, humidity 50%, and 12 h light-dark cycle from 8 a.m. to 8 p.m.) and had free access to food and water. All animal procedures were carried out in accordance with the guidelines for the Care and Use of Laboratory Animals of Fourth Military Medical University (and the number of this approval 20200515).

The following primary antibodies were used: rabbit anti-RIPK1/RIP1 (phospho Ser166) antibody (AGR66476, Arigo), rabbit anti-MLKL (phospho S345) antibody (ab196436, Abcam), rabbit anti-caspase-3 antibody (#9662, CST), rabbit anti-cleaved caspase-3 (Asp175) antibody (#9661, CST), rabbit anti-β-actin (13E5) antibody (#4970, CST), rabbit anti-MD2 antibody (No. GTX85517, GeneTex), and mouse anti-NeuN antibody [1B7] (ab104224, Abcam). The secondary fluorophore-bound IgGs (Alexa Fluor series) for immunofluorescence (IF) included donkey anti-rabbit IgG H&L (Alexa Fluor^® 594) (ab150064, Abcam) and donkey anti-mouse IgG H&L (Alexa Fluor^® 488) (ab150105, Abcam). Inhibitors, including Z-VAD-FMK [ab120487, Abcam, 5 μg/μl, intracerebroventricular (icv) administration], necrostatin-1 (Selleck, 3 μg/μl, icv), protease inhibitor cocktail (No. 16B140042, Biotool) and phosphatase inhibitor cocktail (No. B15002, Biotool) were used. Tat-fused peptides were custom-synthesized by the Peptide Synthesis and Purification Core Facility at GL Biochem. NeuroTrace™ Nissl 500/525 green fluorescent stains were purchased from Invitrogen.

The mice were fasted for 12 h but had free access to water before the surgical procedure. Then, anesthesia was induced by inhalation of 2% isoflurane through a face mask, and anesthesia was maintained with 1.5% isoflurane. Under sterile conditions, the cecum was ligated with 4–0 silk at the midpoint and punctured once with a 22-gauge needle. Then, the cecum was gently squeezed until a single droplet of fecal material came out of the puncture site. Next, the abdominal cavity was closed in two layers, followed by fluid resuscitation (preheated saline, 20 ml/kg), and the animal was returned to its cage. In the sham group, except for ligation and perforation, the rest of the steps were the same as those in the CLP model group. All animals received care with suitable analgesia according to institutional guidelines.

According to the high affinity between the cold-inducible RNA binding protein CIRP and MD2 (Qiang et al., 2013), we synthesized a blocking peptide to mimic the 106–125 domain of CIRP and link the trans-trans-activating (Tat) transmembrane functional domain (YGRKKRRQRRR) to form Tat-CIRP (YGRKKRRQRRR-GRGFSRGGGD RGYGG), which promotes penetration of the BBB. For other details, see our previous study (Fang et al., 2021).

Western blot analysis was carried out on the protein samples from the hippocampus. The mice were sacrificed, and the hippocampus was collected and stored at −80°C. Fresh hippocampal tissues were weighed and homogenized in lysis buffer containing protease and phosphatase inhibitors. After lysis on ice for 20 min, the homogenates were centrifuged at 12,000 rpm for 10 min at 4°C. Then, the supernatant was collected, and 10 µl was used for BCA protein quantification. The residual supernatant was denatured with 4×loading buffer at 100°C for 10 min and stored in a −20°C freezer for later use. Subsequently, 10 µl of the prepared sample was separated by SDS-PAGE and transferred to a PVDF membrane under a voltage of 100 V for 90 min. After the membrane was blocked with 50 g/l BSA or 5% nonfat milk for 2 h, it was incubated with primary antibody overnight in a refrigerator at 4°C followed by incubation with HRP-labeled IgG for 2 h at room temperature. After thorough washing with TBST, the Bio-Rad gel imaging system was used for chemical development.

For Nissl staining, after deparaffinization and rehydration, the slices were immersed in 100% ethanol for approximately 5 min, 95% ethanol for 30 s, and 70% ethanol for 30 s. The samples were then washed in PBS three times, immersed in 1% Nissl for 8 min, and washed with PBS three times. Next, the samples were immersed in 70% ethanol for 2 min, 80% ethanol for 2 min, 95% ethanol for 2 min, 100% ethanol for 2 min (twice) and xylene for 2 min (twice). Finally, the slices were sealed by neutral gum. For TeuroTrace™ fluorescent Nissl staining, cryosections (10 μm) were prepared with standard protocols. Then, the samples were rehydrated in PBS for 40 min and washed twice for 10 min in PBS plus 0.1% Triton X-100. Next, the sections were covered with NeuroTrace™ stain (1:2,000) for at least 20 min. After thorough washing, DAPI-containing mounting tablets were used for mounting.

Mice were deeply anesthetized with 2% pentobarbital sodium (20 mg/kg, i.p.) and was rapidly perfused 50 ml of pre-cooled saline through the heart, followed by slow perfusion with 50 ml of 4% paraformaldehyde. The brain was taken out and fixed for 2 h in 4% paraformaldehyde. Then they were dehydrated with 20% and 30% sucrose solutions in sequence until the brain sank to the bottom. And they were made into frozen tissue section in 10 μm. The brain slices were washed 3 times with PBS for 10 min each time. After washing, the slices were blocked with donkey serum and 0.05% Triton X-100 for 2 h at room temperature, and were incubated with primary antibodies overnight at 4°C. After washing thoroughly, secondary antibody was added to the slices in a dark box for 2 h. After thorough washing, DAPI-containing mounting tablets were used to mount the slices. Then, they were placed in a confocal microscope.

Tat-CIRP or inhibitors were injected into lateral ventricles through stereotactic technology. Mice were anesthetized with 2% pentobarbital sodium (20 mg/kg, i.p.) and fixed in a stereotaxic device. Then, their eyes were covered by erythromycin ointment to prevent injury caused by long-term exposure. After a series of procedures, we identified the coordinates of the lateral ventricles (AP = −0.5 mm, ML = +1.0 mm, DV = −1.4 mm) and injected relevant drugs at a total volume of 1 μl.

The OFT is one of the most commonly used tests to assess anxiety-like behaviors in rodents and allows animals to freely explore open areas surrounded by walls. Under normal conditions, the animals show thigmotaxis (tend to avoid the central region and stay at the peripheral region), and increased or decreased center activity is associated with anxiolysis or anxiogenesis (Kuniishi et al., 2017). First, the mice were handled starting 14 days prior to the experiment to allow acclimatization to the investigator. Additionally, the mice were placed in the testing room 1 h before testing to acclimatize to the environment. The mice were placed in an apparatus consisting of a 40 cm × 40 cm × 40 cm polyvinyl chloride (PVC) box, and a camera was used to monitor their movements. All the mice were placed in the same starting position (the corner) of the PVC box at the start, and allowed to explore the test area for 5 min. During testing, the investigator left the room. The tests were all carried out between 6:30 p.m. and 10:30 p.m. After each test, the mice were returned to the cage, and the bottom and sidewalls of the box were cleaned with 75% alcohol. The next step did not start until the box was dry and there was no smell. Mouse movements and activity were recorded and analyzed by the Anymaze system.

An EPM is one of the traditional tests to evaluate anxiety-like behavior. In this test, four elevated arms radiate from a central platform in which two opposing arms are walled and the other two opposing arms are opened. A camera was placed on a stand above the arena. Before the test, the mice were allowed to acclimatize to the facility for 7 days. In addition, the mice were moved to the behavioral testing room 1 h before the test. During testing, the mice were placed at the intersection of the open and closed arms of the elevated plus maze and allowed to freely explore the maze for a total of 5 min. During testing, the investigator left the room. Mice tended to avoid open or elevated places, which was counterbalanced by their innate curiosity to explore areas that are new to them. More time in the open arms was associated with less anxiety, and more time in the closed arms was related to anxiety. Additionally, the tests were all carried out between 6:30 p.m. and 10:30 p.m. After each test, the mice were returned to the cage, and the arms were cleaned with 75% alcohol. The next step did not start until the device was dry and there was no smell. Mouse movements and activity were recorded and analyzed by the Anymaze system.

Similar to the other tests, the mice were given 7 days to acclimate to the testing facility and the investigator before the test. In addition, the mice were moved to the behavioral testing room 1 h before the test. During the test, mice were subjected to an 8-min swim session including 2 min in the adaptation stage and 5 min in the test stage in clear plexiglass cylinders (30 cm height × 20 cm diameter) filled with 17 cm of water (23 ± 2 °C). The water was changed every three mice. The test was performed between 6:30 p.m. and 10:30 p.m. under normal light conditions. The tests were recorded by a digital video camera, and the investigator left the room during the test. The immobility time, which was defined as passive floating with no additional activity other than that necessary to keep the animal’s head above the water was recorded. After each test, the mice were dried off and placed in an incubator until they recovered.

GraphPad Prism 8.0.2 was used to conduct the statistical analysis. All values except for the murine sepsis score (MSS) and the weights are presented as the mean ± SEM and were analyzed by using two-tailed Student’s t test (comparisons of two groups) or one-way analysis of variance (ANOVA) followed by Dunnett’s or Tukey’s post hoc test (comparisons of more than two groups). Survival curves and comparisons among curves were assessed through the Mantel-Cox log-rank test. A p value less than 0.05 was defined as significant.

First, the severity of sepsis in the mice was evaluated by MSS, following previous research (Shrum et al., 2014). In CLP mice, there was a positive correlation between the MSS and the level of serum IL-1β (Supplementary Figure S1C), indicating that the MSS reflected the severity of sepsis induced by CLP in these mice. The MSS significantly increased during the first 3 days and peaked on the third day. Then, the MSS gradually decreased in the following days (Supplementary Figure S1A), indicating the alleviation of sepsis. Similarly, the weight of the CLP-induced mice declined after the surgery and reached a minimum on the third day (Supplementary Figure S1B). Total distance (TD) as an indicator of OFT is often used to evaluate the general health of animals. Our results showed no difference in TD between the sham-treated and CLP-induced mice at 14 days after sepsis (Supplementary Figure S1F), which indicated that evaluating anxiety and depressive behaviors at this time point was reasonable. Then, the relationship between the MSS and outcome was analyzed. Through logistic regression, a significant relationship between scores on the first and the third day and the final mortality was found (Supplementary Table S1). Then, we established a group trajectory development model according to the MSS and final outcomes. The animals were divided into three groups according to their results. The first group showed a low probability of death with MSS<5 all times, and the third group presented high mortality in the first 3 days with MSS>10at the beginning and gradually increasing with time. The second group showed a balance between survival and death, with an inflection point at MSS≥5 on the third day relative to the first group (Supplementary Figure S1E; Supplementary Table S2). Therefore, we selected mice with MSS≥5 on the third day after CLP for subsequent behavioral experiments.

Next, a battery of behavioral tests was performed at 14 days after CLP to determine whether the septic mice present depression-like behaviors (Figure 1A). The OFT (Figure 1B), EPM (Figure 1E) and FST (Figure 1H) were performed. In the OFT, the number of entries to the center zone (Figure 1C) and the time spent in the center zone (Figure 1D) were decreased in the CLP group compared with the sham group. Similarly, the percentage of time spent in the open arm of the septic mice was significantly reduced in the EPM (Figure 1G). However, the animals presented no difference in the percentage of entry in the open arm (Figure 1F). The immobility time of the septic mice was increased in the FST (Figure 1I). The above results indicated that CLP induced long-term mental impairments in mice.

Then, we explored the possible mechanism of sustaining depression-like behaviors induced by sepsis. Pathological staining showed substantial neuronal death in the hippocampus of the septic mice (Figures 1J,K). These findings indicate the probable mechanism of hippocampal neuronal death underlying the long-lasting depression-like behaviors of the septic mice.

Apoptosis and necroptosis are two major PCDs that participate in the pathological processes of sepsis. To investigate their roles in SAE in septic mice, apoptosis-associated proteins and necroptosis-associated proteins in the hippocampus were detected. Our Western blot results indicated a marked increase in the amounts of p-Ripk1, p-MLKL (necroptosis markers) and cleaved caspase-3 (apoptosis markers) in the hippocampus during the first 3 days and a peak at 24 h after CLP surgery (Figures 2A–D; Supplementary Figure S2A–D). Moreover, the IF results demonstrated increased apoptosis and necroptosis in hippocampal neurons at 24 h after CLP (Figures 2E–H; Supplementary Figure S2E–H). These results revealed that apoptosis and necroptosis were increased in hippocampal neurons, which may mediate the pathological process of SAE.

Next, we tested whether suppression of apoptosis or necroptosis in the hippocampus can alleviate the behavioral changes induced by sepsis. Necrostatin-1 (Nec-1) [a specific inhibitor of necroptosis (Cao and Mu, 2021)] and benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl ketone (Z-VAD-FMK) [a specific inhibitor of apoptosis (Chang et al., 2020)] were administered intracerebroventricularly to mice immediately after CLP surgery. Then, their behaviors were observed 14 days after CLP (Supplementary Figure S3A, B). Surprisingly, there were no differences among the groups in survival rate or mental disorder-associated behaviors (Supplementary Figure S3C–H). Moreover, there was no significant improvement in the MSS or weight or reduction in neuronal loss under inhibitor treatment (Supplementary Figure S4, S5). These results indicated that inhibiting apoptosis or necroptosis alone could not rescue the depression of the septic mice.

Then, we explored the reasons why inhibition of apoptosis or necroptosis failed to alleviate depression in the septic mice. Our results showed that Nec-1 treatment facilitated the expression of cleaved caspase-3 and that the administration of Z-VAD-FMK accelerated the phosphorylation of Ripk1 (Figures 3A–D), which indicated a relationship according to Ramsay’s rule between apoptosis and necroptosis. However, synergy using Nec-1 and Z-VAD-FMK increased the death of the septic mice (Supplementary Figure S3C), which may be attributed to the superimposed toxicity of inhibitors and solvent.

Given the current unsatisfactory treatments, we attempted to identify a target that affects both apoptosis and necroptosis. MD2 is an essential cofactor protein of TLR4 that participates in the inflammatory response and is a key mediator of apoptosis and necroptosis in stroke models (Fang et al., 2021). Here, we explored the expression of MD2 in neurons in the hippocampus of the septic mice. A prominent increase in MD2 expression was detected at 24 h after CLP surgery (Figures 4A–D). To further confirm the crucial role of MD2 in the pathology of SAE, we constructed transgenic mice that with specific MD2 knockout in excitatory neurons (see our previous research for details (Fang et al., 2021)). The expression of apoptosis- and necroptosis-associated proteins was reduced in the hippocampus of the CaMKII-MD2^fl/fl mice compared with their littermates (Figures 5A–D). Moreover, the amount of HMGB1, a typical damage-associated molecular pattern (DAMP) molecule, was decreased (Figures 5E,F). Accordingly, less cell death was observed in the hippocampus of the CaMKII-MD2^fl/fl mice than the control mice (Figures 5G,H). Then, we tested proinflammatory factors in the hippocampus of the CaMKII-MD2^fl/fl mice. The expression of IL-6 was decreased in the CaMKII-MD2^fl/fl mice (Figures 6G–I).

Then, a series of behavioral experiments were performed (Supplementary Figure S6A), revealing that the CaMKII-MD2^fl/fl mice exhibited fewer behavioral changes 14 days after CLP (Figures 6A–F). Moreover, the MSS and weight loss of the CaMKII-MD2^fl/fl mice with sepsis were mitigated compared with those of their littermates (Supplementary Figure S6C, D).

After demonstrating the critical role of MD2 in SAE, we tried to find a potential clinical treatment for SAE. In our previous study, we designed a short peptide to disturb the function of MD2 (Fang et al., 2021). To confirm its utility in the sepsis model, TC was administered via icv injection into the mice during the first 3 days after CLP, and the behaviors were tested on the fourteenth day after CLP (Figures 7A,B). The TC administration group showed fewer behavioral changes than the CLP group (Figures 7D–J). The MSS of the TC group was lower than that of the CLP group without any other treatment (Supplementary Figure S7A). The weight of the TC group declined quickly during the period of drug delivery; however, the speed and degree of weight recovery were better than those of the saline groups (Supplementary Figure S7B). There was only a trend of improved survival in the TC-treated group, which was treated with icv administration (Figure 7C). Neuronal loss in the septic mice was alleviated with TC (Figures 8A,B). These results show that TC has good application prospects for treating SAE. Then, the mice were systematically administered TC within 3 days after CLP with a dosage of 50 mg/kg. The survival rate, MSS, and weight of mice were recorded for 14 days after CLP. Behavioral changes were tested on the fourteenth day after CLP. The survival rate of CLP + TC group was increased compared with the CLP group (Supplementary Figure S8A). In the CLP + TC group, improvements in the MSS, and weight of mice were also accelerated (Supplementary Figure S8B–C). Furthermore, the percentage of open arm entries and durations in the open arms were increased in the CLP + TC group compared with the CLP group (Supplementary Figure S8D). These results indicate that TC facilitates recovery from sepsis.