Eight-week-old adult male wild-type (WT) C57BL/6J mice were supplied by Viton Lever (Beijing, China) Laboratory Animal Center. All mice were provided with quality inspection certificates.

We provided a single consolidated timeline figure that includes: cecal ligation and puncture (CLP) (time 0), timepoints for hippocampal sampling (6 h, 1–3 days, 7 days, 14 days), the necrosulfonamide (NSA)/C-178 dosing schedule (day of CLP + 3 days), open field test (OFT) timing, and barnes maze test (BMT)/testing days (Supplementary Fig. 1A).

In our experiment, the CLP model was used to induce sepsis. The 2% isoflurane (20 ml/L, 20240115, RWD Life Science, Shenzhen, Guangdong, China) and 1.5% isoflurane (15 mL/L) were respectively used for anesthesia induction and maintenance. The abdominal skin of the mice was thoroughly disinfected before surgery. Along the middle and lower abdomen, a straight line was drawn for the middle incision (incision of about 1.0 cm), and the skin and the peritoneum were cut open one by one. A 4-0 surgical suture was used to ligate the midpoint of the cecum, and a 22-G needle was applied to puncture the cecum, and squeezing the cecum to the puncture point of the outflow of the bowel content (the size of rice grains). Layer-by-layer suturing of the skin; disinfection of incisions. We administered lidocaine (0.1 mL, 0.2%, s.c., D32510281, Jinyao, Xiangyang, Hubei, China) for pain relief. After completion of the surgical procedure, prewarmed 0.9% normal saline (30 mL/kg) was administered s.c. to compensate for intraoperative fluid loss. Postoperatively, mice were placed in a 37 °C incubator until fully recovered, as indicated by free movement and foraging behavior. Imipenem/cilastatin (25 mg/kg, s.c., 250620728, Zhuhai Lianbang, Zhuhai, Guangdong, China) was administered in the perioperative period, as well as at 3 and 6 h postoperatively. Mice in the Sham group did not receive cecal ligation and puncture, but all other procedures were identical to those in the CLP group. Exclusion criteria: intraoperative death; incorrect ligation site or failed puncture; massive intra-abdominal hemorrhage; body weight deviating from the mean body weight of the same batch by $\pm$10%.

In our experiment, the OFT was used to detect the recovery of locomotor ability in mice after surgery (Supplementary Fig. 1B). Mice were placed in a behavioral chamber for 2 h to habituate. The Anymaze behavioral monitoring system (60000, Stoelting Company, Wood Dale, IL, USA) was connected to a custom-made open field test chamber (40 $\times$ 40 $\times$ 40 cm) constructed from polyvinyl chloride. A high-definition camera was installed inside to monitor mouse movement. Before testing, the floor and inner four walls of the behavioral chamber were wiped with alcohol (75%) and allowed to air-dry. Each mouse was allowed to move freely for 5 min in the behavioral chamber. The Anymaze behavioral monitoring system automatically detected and recorded the mouse’s behavior. The behavioral box was then sprayed with 75% alcohol to eliminate the mouse’s residual odor. After all tests were completed, the total distance traveled by the mice was extracted.

We used the BMT to detect the memory abilities of the mice (Supplementary Fig. 1C). The adaptation and training phases of the BMT were conducted 5 days prior to CLP. The a priori primary behavioral endpoint in the BMT was the latency of mice to first enter the target hole (escape from the noxious noise) on day 15 after CLP. The BMT was performed and analyzed by investigators who were blind to the group assignments of the mice. The BMT steps were as follows. Adaptation stage: The mouse was placed on the maze platform, and a transparent glass cup was covered. The buzzer (60170, Stoelting Company, Wood Dale, IL, USA) was turned on to provide a noxious auditory stimulus. After 1 min, the transparent glass cup was moved to guide the mouse to the target hole and into the target box, and the noise was terminated. The mouse was kept in the target box for 1 min and then returned to its cage. Alcohol (75%) was sprayed on the platform and the target box. Training phase: the duration was four days. The mouse was positioned on the Barnes maze platform, covered by an opaque box. The buzzer was turned on to make a noise, and the opaque box was taken away. The mouse was allowed to freely move on the platform for 2 min. If the mouse entered the target box, the buzzer was stopped immediately. This behavior was regarded as a successful escape. Conversely, the mouse was directed to the target box using a transparent glass cup, and the buzzer was turned off immediately. Day 1, each mouse received 3 training trials; Days 2–4, each mouse received 4 training trials. Training was conducted for a total of 15 trials over 4 days. Test phase: the mouse was positioned on the Barnes maze platform and covered by an opaque box. The buzzer was turned on, the opaque box was taken away, and the mouse explored freely on the platform for 2 min. The latency for the mouse to enter the target hole for the first time (escape latency) was noted. The time the mouse stayed in the target quadrant was also noted. Percentage of time spent in the target quadrant = time spent (target quadrant) / total time $\times$ 100%.

The mixed lineage kinase domain-like protein (MLKL) inhibitor NSA (ab143839, Abcam, Cambridge, UK) and the STING inhibitor C-178 (ab287033, Abcam) were injected into the lateral ventricle via cannula (Supplementary Fig. 1D). The anesthetic for the stereotaxic surgery was 2% pentobarbital sodium (20mg/kg, i.p., 250302, Mindong Lijiesun Pharmaceutical Co., Ningde, Fujian, China). The heads of the mice were sterilized before the operation. Centering on the Bregma point of mice, the trocar position points were: anteroposterior (AP): –0.50 mm; dorsoventral (DV): –1.50 mm; mediolateral (ML): –1.00 mm. Cranial drilling was performed in mice. Trocars were placed into the lateral ventricle, and cerebrospinal fluid was seen to overflow. The trocars were fixed to the cranial bones of the head. The mouse was then placed in an incubator until it was fully awake. CLP was performed one week later. Subsequently, the internal injection cannula, polyethylene (PE) tubing, locking nut, and injector were assembled in advance. The drug was drawn using the syringe pump, and the position of the drug solution was marked on the PE tubing to monitor whether the liquid level decreased during injection. The cannula cap was then removed, the internal injection cannula was slowly inserted into the guide cannula, and secured with the locking nut. The total injection time was 10 min. NSA was injected at a rate of 0.2 µL/min, and C-178 at 0.15 µL/min. After the injection was completed, the cannula was kept in place for approximately 10 min to allow sufficient drug diffusion, followed by slow withdrawal of the internal injection cannula. The cannula cap was reinserted and tightened. Cannula specifications: cannula length (C) 3.5 mm; outer diameter (D) 0.41 mm; gauge inner diameter 1 (G₁) 0.5 mm; gauge inner diameter 2 (G₂) 0.5 mm. Injector specifications: 0.5–100 µL. During the experiment, 2 mice were excluded due to cannula dislodgement. No mice were excluded because of cannula occlusion or surgical complications. To verify the correct placement of the cannula in the mouse lateral ventricle, needle-track histology was performed post-mortem (Supplementary Fig. 1E).

Euthanasia of experimental mice was carried out in strict accordance with the Guide for the Care and Use of Laboratory Animals issued by the National Institutes of Health (NIH). Mice were euthanized by intraperitoneal injection of an overdose (100 mg/kg) of pentobarbital sodium to achieve rapid and painless death. Complete cardiac and respiratory arrest was confirmed before tissue harvesting to verify the success of euthanasia.

We used WB to test the expression level of specific proteins. First, hippocampal-protein samples were prepared according to the standard protocol. Then, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels (10%, P0012A, Beyotime, Shanghai, China) of varying concentrations were selected according to the molecular weight of the target protein. For electrophoresis, the prepared samples were loaded, and electrophoresis was performed at 80 V until the 70 kDa band of the protein marker was visible, after which the voltage was adjusted to 100 V. Transfers were carried out under constant voltage at 100 V for 2 h. The membrane was blocked with 5% skimmed milk for 1.5 h, then washed three times with (1$\times$) tris-buffered saline with tween-20 (TBST, 1$\times$, T1085, Solarbio, Beijing, China), 5 min each time. Primary antibody incubation: p-STING (1:500, rabbit, AP1369, ABclonal, Wuhan, Hubei, China), STING (1:1000, rabbit, A21051, ABclonal), p-RIPK1 (1:1000, rabbit, ARG66476, Arigo, Hsinchu, Taiwan, China), RIPK1 (1:1000, rabbit, ab300617, Abcam), p-RIPK3 (1:1000, rabbit, ab195117, Abcam), RIPK3 (1:1000, rabbit, ab62344, Abcam), p-MLKL (1:1000, rabbit, ab196436, Abcam), MLKL (1:1000, rabbit, ab184718, Abcam), and GAPDH (1:1000, rabbit, ab181602, Abcam), overnight at 4 °C. Finally, secondary antibody incubation was performed for 2 h at 37 °C. We used Image Lab software (6.0, Bio-Rad, Hercules, CA, USA) for band quantification. Normalization was performed against the internal reference protein (GAPDH). We used calibration curves and avoided saturated exposure to ensure a linear range.

IF was used to detect specific binding between antigen and antibody to detect target proteins in tissue samples. Brain tissue was harvested from CLP and Sham mice on postoperative day 1. Brain slices were blocked with the mixture (donkey serum (5%, ab7475, Abcam) and Triton (0.3%, ab286840, Abcam)) at 37 °C for 1 h. Incubation of primary antibodies: p-STING (1:100, rabbit, AP1369, ABclonal), p-MLKL (1:100, rabbit, ab196436, Abcam), neuronal antibody (1:200, mouse, ab104224, Abcam), ionized calcium-binding adapter molecule 1 (IBA1) antibody (1:200, goat, ab5076, Abcam), overnight at 4 °C.

Secondary antibody buffer containing Alexa Fluor 594 (1:300, S32356, Invitrogen, Carlsbad, CA, USA) or Alexa Fluor 488 (A-11001, Invitrogen, 1:300) was added dropwise. Incubate at 37 °C for 2 h in the dark. Subsequently, 4′,6-diamidino-2-phenylindole (DAPI) solution (1:200, ab104139, Abcam) was added and incubated for 10 min. After thorough washing and coverslipping, the sections were air-dried. Confocal microscopy (FV1200, Olympus, Tokyo, Japan) for photography. We defined the region of interest (ROI) as the region showing co-localization of the target protein and neurons in the hippocampal cornu ammonis 1 (CA1) area. The mean fluorescence intensity in the non-specific staining region adjacent to the ROI was measured as the background value. Corrected fluorescence intensity = mean intensity of ROI - mean intensity of local background. Six mice/group (gp) were used; three sections per mouse and three visual fields per section were analyzed for mean fluorescence intensity. Sections were coded by a third party, and quantification was performed by an investigator who was blind to group allocation.

Statistical analyses were performed using GraphPad Prism 9.0 software (GraphPad Software Co., San Diego, CA, USA). Experimental data were expressed as mean $\pm$ standard deviation. Two groups of data were tested using the t-test. The One-way or Two-way analysis of variance (ANOVA) (experimental group and drug group) was used to analyze more than two groups of experimental data. The Two-way ANOVA was performed using a mixed-model analysis. The BMT data were analyzed as the predefined test-day endpoint. For the Western blot time-course experiments, the sample size at each time point was n = 6. The multiple-comparison correction method was the Dunnett’s test. Statistical significance was set at p $\le$ 0.05.

Our experimental results indicated that the 14–day survival rate of CLP mice was 50.0% (7/14) (Supplementary Fig. 1F). The initial weight (%) of the CLP group was obviously lower than that of the Sham group (Supplementary Fig. 1G). The murine sepsis score (MSS) was obviously higher in the CLP group than in the Sham group within 10 days after surgery (Supplementary Fig. 1H). The surviving mice on day 14 after CLP were subjected to behavioral testing (Fig. 1A). First, the OFT data indicated that movement distance did not differ significantly between the two groups (Fig. 1B,C). Our results showed that the voluntary locomotion was not significantly damaged in mice 14 days after CLP, and subsequent behavioral studies at this time could exclude the effect of exercise. The BMT assessed the spatial memory ability of the mice on day 15 after CLP surgery (Fig. 1A). The results suggested that the primary latency time for CLP group mice escape the noxious stimulus was obviously longer than for Sham group mice (Fig. 1D,E), and there was no obviously significant difference in the time that the CLP group mice stayed at the target quadrant (Fig. 1F,G). This suggested that long-term spatial memory function was significantly impaired in mice after CLP.

Fig. 1.

Cognitive dysfunction occurred in septic mice. (A) Schematic timeline of BMT and OFT. (B) Trajectory diagram of OFT. (C) Statistical results of total distance (n = 7, NS: not significant). (D) Trajectory diagram of BMT and (E) Statistical results of primary latency (n = 7, **p 0.01 versus Sham). (F) Heat map of BMT and (G) Statistical results of time in the target quadrant (n = 7, NS: not significant). Statistical method: Student’s t-test. BMT, barnes maze test; OFT, open field test; CLP, cecal ligation and puncture.

A study has reported that neuronal necroptosis was closely related to neurodegenerative diseases [7]. If so, then how does necroptosis change in hippocampal neurons of septic mice? The WB was performed to ascertain the expression level of necroptotic proteins in hippocampal tissues of septic mice at 6 h, 1 day, 2 days, 3 days, 7 days, and 14 days. Our data showed that the expression of ^S166p-RIPK1/RIPK1, ^S232p-RIPK3/RIPK3, and ^S345p-MLKL/MLKL proteins in the hippocampal tissues of septic mice was significantly elevated at 1, 2, and 3 days (Fig. 2A–D; The original Western blot images can be found in the Supplementary Materials-Original WB). These results demonstrated that necroptosis occurred in the hippocampal tissues of septic mice. The ^S345p-MLKL protein was immunofluorescently stained with neurons (NeuN). The IF intensity of ^S345p-MLKL in the hippocampal CA1 region co-localized with mouse hippocampal neurons on day 1 after CLP surgery was significantly different (Fig. 2E,F). In addition, our IF results demonstrated significant neuronal loss in the hippocampal CA1 region of mice on the third day after CLP surgery (Supplementary Fig. 1I). This suggested that the necroptosis protein ^S345p-MLKL was expressed in neurons in the hippocampal CA1 region of SAE mice.

Fig. 2.

Necroptosis arose in hippocampal neurons of SAE mice. (A) WB and (B–D) statistical results of necroptotic proteins (n = 6, *p 0.05, **p 0.01). An internal control was GAPDH. (E) IF staining of ^S345p-MLKL (scale bar = 40 µm), and (F) Statistical results of fluorescence intensity (n = 6, **p 0.01). Statistical method: Student’s t-test. WB, Western blot; NeuN, neurons; IF, immunofluorescence; RIPK, receptor-interacting protein kinase; MLKL, mixed lineage kinase domain-like protein; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; DAPI, 4^′,6-diamidino-2-phenylindole.

The effect of necroptosis on cognitive dysfunction in SAE mice was further verified by inhibiting ^S345p-MLKL protein expression. In this experiment, a cannula was embedded in the lateral ventricle of 8-wk-old WT mice. MLKL inhibitor NSA (2 µL; 1 mg/mL) was injected through the cannula on the day of CLP, and the drug was administered for 3 consecutive days (Fig. 3A). First, WB was used to verify the effectiveness of the NSA. The results showed that ^S345p-MLKL/MLKL protein expression in the hippocampal tissue of the Sham+NSA group was obviously lower than that in the Sham+ dimethyl sulfoxide (DMSO) group, and ^S345p-MLKL/MLKL protein expression in the hippocampal tissue of the CLP+NSA group was obviously lower than that of the CLP+DMSO group (Fig. 3B,C; The original Western blot images can be found in the Supplementary Materials-Original WB). Further, the OFT results demonstrated that movement distance did not differ significantly among all groups (Fig. 3D,E). The BMT results demonstrated that the CLP+NSA group had an obviously shorter primary latency time to find the target hole (escape) than did the CLP+DMSO group (Fig. 3F,G), whereas the difference in the time spent in the target quadrant was not significant (Fig. 3H,I).

Fig. 3.

Necroptosis caused cognitive dysfunction in SAE mice. (A) Schematic timeline of BMT, NSA injection, and OFT. (B) WB and (C) Statistical results of ^S345p-MLKL/MLKL (n = 6, *p 0.05 versus Sham+DMSO, ^#p 0.05 versus CLP+DMSO). An internal control was GAPDH. (D) Trajectory diagram of OFT and (E) Statistical results of total distance (n = 7, NS: not significant). (F) Trajectory diagram of BMT and (G) Statistical results of primary latency (n = 7, ^#p 0.05 versus CLP+DMSO). (H) Heat map of BMT and (I) Statistical results of time in the target quadrant (n = 7, NS: not significant). Statistical analysis: Two-way ANOVA with Sidak’s multiple comparisons test. SAE, sepsis-associated encephalopathy; NSA, necrosulfonamide; DMSO, dimethyl sulfoxide; ANOVA, analysis of variance.

The data indicated that the expression of ^S366p-STING/STING proteins in the hippocampal tissue of the SAE mice was significantly higher 1 day and 2 days after modeling than in the Sham group (Fig. 4A,B; The original Western blot images can be found in the Supplementary Materials-Original WB). The IF staining of ^S366p-STING protein with neurons and microglia was performed. The data demonstrated that ^S366p-STING in the hippocampal CA1 region was mainly co-localized with hippocampal neurons in CLP mice, and there was a statistically significant difference in their IF intensities (Fig. 4C,D). There was a small amount of co-localization with microglia in the hippocampal CA1 region (Fig. 4E,F). This suggested that p-STING had elevated expression in neurons of the hippocampal CA1 region in SAE mice.

Fig. 4.

Spatiotemporal expression of STING protein increased in the hippocampus of SAE mice. (A) WB and (B) Statistical results of ^S366p-STING/STING protein expression levels (n = 6, *p 0.05; **p 0.01 versus Sham). Statistical method: Student’s t-test. An internal control was GAPDH. (C,E) IF staining of ^S366p-STING (scale bar = 40 µm), (D) Statistical results of ^S366p-STING fluorescence intensity in hippocampal neurons, and (F) Statistical results of ^S366p-STING IF-positive cells in hippocampal microglia (n = 6, *p 0.05, **p 0.01). Statistical method: Student’s t-test. STING, stimulator of interferon gene; IBA1, ionized calcium-binding adapter molecule 1.

In the present experiment, 1.5 µL of STING inhibitor C-178 (1 mg/mL) was injected on the day of CLP, and the drug was administered for 3 consecutive days (Fig. 5A). The data indicated that the expression level of ^S366p-STING/STING, ^S232p-RIPK3/RIPK3 and ^S345p-MLKL/MLKL protein in the hippocampal tissue of CLP+C-178 group mice was obviously lower than that in the CLP+DMSO group (Fig. 5B,C,E,F; The original Western blot images can be found in the Supplementary Materials-Original WB), however, there was no significant decrease in ^S166p-RIPK1/RIPK1 protein expression (Fig. 5B,D). The results indicated that inhibition of STING down-regulated the level of ^S232p-RIPK3/RIPK3 and ^S345p-MLKL/MLKL proteins, however, it could not down-regulate the expression of ^S166p-RIPK1/RIPK1 protein. In addition, OFT data indicated that there was no significant difference in the locomotor ability of mice among the groups (Fig. 5G,H). BMT results demonstrated that the primary latency time to find the target hole (escape) was obviously shorter in the CLP+C-178 group mice than in mice of the CLP+DMSO group (Fig. 5I,J), but the difference in the time spent in the target quadrant was not statistically significant (Fig. 5K,L). Those results indicated that inhibition of STING partially improved BMT performance in SAE mice.

Fig. 5.

Inhibition of STING reduced the expression of associated necroptosis proteins and improved cognitive function in SAE mice. (A) Schematic timeline of BMT, C-178 injection, and OFT. (B) WB. (C–F) Statistical results of ^S166p-RIPK1/RIPK1, ^S232p-RIPK3/RIPK3 and ^S345p-MLKL/ MLKL protein expression levels after C-178 inhibitor (n = 6, *p 0.05, **p 0.01 versus Sham+DMSO, ^#p 0.05, ^#⁢#p 0.01 versus CLP+DMSO, NS: not significant). An internal control was GAPDH. (G) Trajectory diagram of OFT, (H) Statistical results of total distance (n = 7, NS: not significant). (I) Trajectory diagram of BMT and (J) Statistical results of primary latency time (n = 7, ^#p 0.05 versus CLP+DMSO). (K) Heat map of BMT and (L) Statistical results of time in the target quadrant (n = 7, NS: not significant). Statistical analysis: Two-way ANOVA with Sidak’s multiple comparisons test.

In this experiment, immunofluorescence and Co-immunoprecipitation (Co-IP) tests were used to detect the interaction between STING and RIPK3 protein. The results of IF showed that STING and RIPK3 proteins co-localized abundantly in neuronal cells in the hippocampal CA region of SAE mice (Fig. 6A). Co-IP results showed that STING interacted with RIPK3 protein (Fig. 6B; The original Western blot images can be found in the Supplementary Materials-Original WB) in neurons.

Fig. 6.

STING interacted with the RIPK3 protein. (A) IF of STING co-localized with RIPK3 (scale bar = 40 µm, 240 µm). (B) Co-IP of STING and RIPK3 proteins in lysates of hippocampal tissues. Co-IP, Co-immunoprecipitation; WCL, whole cell lysates; IB, immunoblotting; IP, immunoprecipitation.