All the animal experiments and procedures were approved by the Laboratory Animal Ethics Committee of Peking University First Hospital (approval number: J2022041, 20 May 2022). Male C57BL/6 J mice (age, 8–12 weeks) were purchased from SPF (Beijing, China) Biotechnology Corporation Ltd. [License No. SCXK(Jing) 2019-0010]. The mice were randomly assigned to the sham group (n = 3–5) and the septic group (n = 3–5 at each time point) for the detection of cytokine levels, organ damage scores, and splenic T-lymphocyte phenotype. They were housed in the specific pathogen-free environment of our laboratory animal facility (12-h day/night cycle at 25°C) with free access to food and water. Before use, the animals were allowed to acclimate to laboratory conditions for at least 1 week.

Sepsis was induced through cecal ligation and puncture (CLP), as described previously [17]. In brief, the mice were anesthetized with 0.5% pentobarbital sodium (50–60 mg/kg) intraperitoneally. Each mouse was placed in the supine position on an operating pad; then, its abdomen was shaved and sterilized with 75% alcohol. A 0.5–1.0-cm vertical midline incision was created through the abdominal wall, and the cecum was externalized. Next, 75% of the cecum was ligated (via distal ligation) with a 4-0 silk suture and penetrated completely with a 21-gauge needle. Two drops of feces were gently squeezed from penetration holes, and the cecum was re-placed into the peritoneal cavity. The abdominal incision was closed in two layers with 5-0 silk sutures and sterilized with iodine. Sham-operated mice underwent the aforementioned procedure but without cecum ligation or puncture. Finally, 1.0 mL of prewarmed saline was injected subcutaneously to resuscitate each mouse. The survival rate was assessed every 6–8 h, and all mice were observed until the end of post-CLP day 7.

A whole blood sample was collected via cardiac puncture and placed at room temperature for 4 h. Serum was separated from the blood through centrifugation at 4000 rpm at 4°C for 20 min (Centrifuge 5425 R, Eppendorf, Hamburg, Germany). Serum interleukin-6 (IL-6) and interleukin-10 (IL-10) levels were then determined using an enzyme-linked immunosorbent assay (ELISA) kits (Elabscience, E-EL-M0044c and E-MSEL-M0031, Wuhan, China), according to the manufacturer’s instructions.

Total RNA was extracted using TRIzol reagent (Invitrogen, 15596018, Waltham, MA, USA) and subjected to purity and concentration measurement using a Nanodrop2000 (Thermo Fisher Scientific, ND-1000 spectrophotometer, Wilmington, DE, USA). An A260/A280 ratio between 1.8–2.0 was considered qualified for purity. Complementary DNA was synthesized from 2 μg of total RNA using FastKing RT Enzyme (Tiangen Biotech, KR118, Beijing, China). Quantitative reverse transcription polymerase chain reaction (qRT-PCR) reagents were prepared using the PowerUp SYBR Green Master Mix (Thermo, A25742, Waltham, MA, USA). qRT-PCRs were performed on an ABI ViiA7 system (Applied Biosystems, CA, USA). The results were normalized to the glyceraldehyde-3-phosphate dehydrogenase (Gapdh). All qRT-PCR primer sequences are listed in Supplementary Table S1. Relative mRNA expression was analyzed using the 2^−ΔΔCt method on Excel 2019MS0 (version 2501Build 16.0.18429.20132, Microsoft, Washington, DC, USA).

Kidney, liver, heart, and lung tissues were fixed in 10% formaldehyde (Leagene, DF0118, Beijing, China) for at least 24 h. After dehydration in a series of ascending concentrations of ethanol (Mreda, M042753-500 mL, Beijing, China), samples were embedded in paraffin (Leica, 39601006, Wetzlar, Germany). The tissue blocks were then sectioned at 4-μm thickness using a microtome (Leica, RM2235), and the slices were stained with hematoxylin and eosin (H&E) (Solarbio, G1120, Beijing, China) for observation under a light microscope (Eclipse Ci-L, Nikon, Japan) in a high-power field (magnification, 400×). The degree of histopathological damage was scored by a pathologist blind to experimental groupings on a 0–4 scale (Supplementary Table S2) [18,19]. For each mouse, six randomly selected nonoverlapping fields were scored, and a mean score was calculated for each group.

Spleens were harvested from all mice after they were sacrificed. The spleens were then placed in a 10-mL culture dish containing 5 mL PBS and broken down using the rubber end of a sterile 3-mL syringe. Cells were passed through a 70-μm mesh filter (Biosharp, BS-70-CS, Anhui, China), and the single-cell suspensions were centrifuged and resuspended in PBS. After red cell lysis (Tiangen Biotech, RT122-01), each splenocyte suspension was adjusted to a concentration of 1 × 10⁷ cells/mL, and live cells were counted using trypan blue (Solarbio, C0040-50 mL) exclusion staining. The suspensions were preincubated with monoclonal antibodies (mAbs) against CD16/CD32 (1:200, Biolegend, 156603, San Diego, CA, USA) for 10 min on ice for Fc receptor blocking. This was followed by incubation for 20 min with the fluorochrome-conjugated mAbs against the following antigens: CD3 (AF700) (1:250, Biolegend, 100216), CD8 (BV421) (1:200, Biolegend, 100737), CD69 (APC) (1:100, Biolegend, 104513), and CD71 (PE) (1:250, Biolegend, 113807). Apoptosis was assessed using 7AAD (Percp) (1:20, Biolegend, 420403) and Apotracker Green (FITC) (1:200, Biolegend, 427401).

Next, we stained the splenocytes for KLF3. In brief, the splenocytes were fixed and permeabilized with a fixation/permeability working buffer (Invitrogen, 00-5523-00). They were incubated with anti-KLF3 goat primary antibody (10 μg/mL, Invitrogen, PA5-18030) for 45 min and then with Alexa Fluor 488-conjugated donkey anti-goat immunoglobulin G (1 μg/mL, Invitrogen, A-11055). Goat immunoglobulin G (Beyotime, A7007, Beijing, China) was used as the isotype control for the primary antibody. Flow cytometry was performed on a BD LSRFortessa X-20 (BD Biosciences, CA, USA). Finally, flow cytometric data were analyzed using FlowJo (version 10.8.1, Becton Dickinson, USA). Fig. 1 illustrates the gating strategy used for flow cytometry.

All data are presented as means ± standard deviations. Data from three or more groups were compared using one-way analysis of variance. Differences between CD4⁺ and CD8⁺ T-lymphocyte subpopulations were evaluated using Student’s t test. Correlation between independent data was assessed with Spearman’s correlation test. Survival rates in each group, presented using Kaplan-Meier curves, were evaluated using the log-rank test. GraphPad Prism (version 9.5; GraphPad, San Diego, CA, USA) was used for graphic and statistical analyses. A p-value of <0.05 was considered to indicate statistical significance.

CLP in rodents has become the most widely used model for experimental sepsis [17]. Here, we induced severe sepsis in our mice by ligating 75% of the cecum. To mimic the natural progression of sepsis, we did not administer any antimicrobials to the mice.

At around 12 h after CLP, all operated mice began demonstrating signs including malaise, chills, piloerection, generalized weakness, and reduced gross motor activity. Using our scoring system (Supplementary Table S3), we scored the whole-body status of septic mice at 18 h after CLP in terms of their mental state, autonomous motor activity, and myodynamia. As shown in Fig. 2A, the septic mice demonstrated a significantly lower whole-body status score than the sham-operated mice. Lethality was noted beginning at 24 h after CLP, and the 7-day mortality rate was 64.7% (Fig. 2B).

In the early stage after CLP, the septic mice demonstrated considerable increases in spleen Il10, Il6, Il1b, Ifng, and Tnfa mRNA levels (Fig. 3A) and serum IL-10 and IL-6 levels (Fig. 3B), which suggested that their systemic inflammatory responses had increased significantly. In contrast, spleen Klf3 mRNA levels decreased, and they were the lowest on post-CLP day 3 (Fig. 3C). Over the 7 days after CLP, the expression of the aforementioned proinflammatory and anti-inflammatory mediators in the spleen and serum decreased gradually, but serum IL-10 and IL-6 levels remained higher in the septic mice than in the sham-operated mice until the post-CLP day 7 (Fig. 3B). These results indicated that inflammatory response remained persistent in the septic mice. In the late stage after CLP, Klf3 mRNA expression increased gradually (Fig. 3C). Furthermore, significant negative correlations were observed in mRNA relative expression levels between Klf3 and cytokines Il10 (r = −0.7523, p = 0.0003) and Il6 (r = −0.7765, p < 0.0001) (Fig. 3D). Therefore, in the septic mice, changes in KLF3 expression demonstrated a trend opposite to changes in the systemic inflammatory response.

The severity of CLP-induced organ damage was also determined through histological analysis (Fig. 4). The H&E staining results demonstrated that severe kidney damage, including Bowman’s capsule expansion, tubular epithelial vacuolization, and brush border loss, occurred in the early stage after CLP surgery. The histological analysis also revealed that liver, heart, and lung injury became increasingly severe over the course of sepsis development.

Next, we employed flow cytometry to evaluate the phenotypes and proportions of splenic T lymphocytes. We identified CD3⁺ cells as T lymphocytes and then used CD8 to differentiate CD4⁺ and CD8⁺ T-lymphocyte subpopulations (Fig. 1).

As illustrated in Fig. 5A, in the early stage, CLP increased the proportion of CD69⁺ cells and the mean fluorescence intensity (MFI) of the transferrin receptor CD71 considerably, suggesting that CLP induced the activation of T lymphocytes. Subsequently, these indicators gradually decreased in the later stage. CD4⁺ and CD8⁺ cell subpopulations exhibited a temporal behavior similar to the overall T lymphocytes. Moreover, the changes in CD8⁺ cell activation were more pronounced than those in CD4⁺ cell activation, whereas cellular metabolism levels were higher in CD4⁺ cells than in CD8⁺ cells on post-CLP day 3 (Fig. 5B).

In immune responses, the activation and apoptosis of T lymphocytes are tightly regulated processes [20,21]. Sustained activation induces activation-induced cell death (AICD), thereby balancing effector function with immune homeostasis [22]. Moreover, stimulation via the T-cell receptor can activate apoptotic pathways [23]. In the current study, while CLP led to splenic T-lymphocyte activation (Fig. 5A), it significantly increased the proportion of apoptotic cells on post-CLP day 1 (Fig. 5C). Subsequently, apoptosis was downregulated in parallel to the significant decrease in T-lymphocyte activation on post-CLP days 3–5 (Fig. 5A). Finally, apoptosis increased again on post-CLP day 7, indicating a shift toward immunosuppression. These findings highlight the dynamic regulation of T-lymphocyte activation and apoptosis in response to sepsis induced by CLP, underscoring the complex interplay between these processes in modulating immune responses.

In T lymphocytes, apoptosis was accompanied by a gradual decrease in the proportion of splenic T lymphocytes to less than one-tenth of that in the sham-operated mice on post-CLP day 7 (Fig. 5C). Combined with the continual increases in serum inflammatory cytokine levels and sustained organ damage until the late stage after CLP (Figs. 3B and 4), a state of persistent inflammation and immunosuppression was supposed to be induced in the septic mice. In addition, over the 7-day period, the apoptosis proportions of CD4⁺ cells remained higher than those of CD8⁺ cells, resulting in an imbalance of the CD4⁺/CD8⁺ ratio (Fig. 5D)—another indicator of an acquired immune disorder associated with sepsis [16].

Supplementary Figs. S1 and S2 present all representative dot-plot data for the aforementioned analyses.

Flow cytometry was used to assess KLF3 expression in splenic T lymphocytes. Fig. 6A exhibits the histogram of isotype control in flow cytometry. As shown in Fig. 6B,C, in the sham group, the proportion of KLF3⁺ subset in T lymphocytes was 92.73 ± 1.27%. Concomitant with a considerable increase in systemic inflammatory response and T-lymphocyte activation on post-CLP day 1, the proportion of KLF3⁺ subset decreased to 78.03 ± 3.29%. When the CLP mice entered the persistent inflammation and immunosuppression stage, as illustrated in Figs. 3–5, the proportion of KLF3⁺ subset gradually increased to 96.27 ± 1.61% on post-CLP day 5. The relative decrease in KLF3 expression observed on post-CLP day 7 may be attributed to the excessive decrease in T lymphocyte count, however, the trend of increasing KLF3 expression on post-CLP days 3–5 was significant. Finally, while CD4⁺ and CD8⁺ lymphocytes exhibited differential expression of CD69 and CD71 (Fig. 5B), no significant differences in KLF3 expression levels were observed between these two subpopulations (Fig. 6D).

Notably, as detailed in Fig. 6E, an inverse correlation was observed in the proportions between the KLF3⁺ and CD69⁺ subsets of T lymphocytes (r = −0.7246, p = 0.0021) during the dynamic immune alterations. Moreover, KLF3 expression was negatively correlated with the proportion of apoptosis (r = −0.8683, p < 0.0001). Thus, KLF3 might be a suppressor of early T-lymphocyte hyperactivation and a prosurvival regulator in lymphocyte homeostasis.