Enhancement of NF-κB Activation in Lymphocytes Prevents T Cell Apoptosis and Improves Survival in Murine Sepsis¹

Sepsis continues to be a common and frequently fatal condition in the intensive care unit despite advances in critical care and antimicrobial therapy. At an annual incidence of 750,000 cases in the United States, its progression to severe sepsis and septic shock is associated with a case fatality rate of ∼30–50% (1, 2). Despite intensive and successful investigations and promising preclinical treatment trials, efforts to reduce the mortality rate from sepsis in patients have failed (3). Increasing costs of critical care treatment exert pressure to better understand the molecular mechanisms of this disease to generate new therapeutic strategies.

There is now general agreement that the inability to regulate the inflammatory response is a major pathogenic factor in the etiology of sepsis. Sepsis describes a complex clinical syndrome that results from a harmful and damaging host response to infection. Initially, appropriate bacterial-host interactions induce widespread activation of the innate immune response, which culminates in systemic inflammation largely borne by production of proinflammatory mediators such as cytokines or chemokines. Severe sepsis and septic shock develop when this response becomes increasingly amplified and dysregulated (4). An analysis of recent studies has demonstrated that the risk of death significantly correlates with the effectiveness of anti-inflammatory therapy (3). However, the profound proinflammatory response that initially occurs in sepsis is balanced by a series of anti-inflammatory molecules (e.g., soluble TNFRs, anti-inflammatory cytokines such as IL-10) that attempt to restore immunological homeostasis (4). This process and its functional consequences are viewed as a counterbalancing response to the initial proinflammatory state (4). Part of this anti-inflammatory reaction involves T cell hyporesponsiveness and anergy, which are well-known phenomena that develop after trauma or burns (5, 6). T cell hyporesponsiveness also occurs in most septic patients to some extent (7). However, this anti-inflammatory response can be excessive and thus can cause immunosuppression and probably promotes multiple organ failure (8).

The development of lymphocyte apoptosis is likely to be another consequence of inflammatory down-regulation in the course of sepsis. Moreover, the extensive apoptotic death of lymphocytes probably contributes to profound immunosuppression. Among other studies, Hotchkiss et al. (9, 10) have clearly demonstrated that extensive lymphocyte apoptosis in animal models of sepsis and in septic patients is an important pathogenetic factor. Hence, results from animal models demonstrate improved sepsis survival rates due to prevention of lymphocyte apoptosis (11).

In this context it is noteworthy that signal transduction via the NF-κB pathway is not only crucial for control of proinflammatory and anti-inflammatory immune responses (12, 13), but it also plays a role in control of programmed cell death (14). Mainly considered as a prosurvival transcription factor, NF-κB signaling targets a number of antiapoptotic genes (15). Moreover, prevention of TNF-α-induced programmed cell death is a hallmark of classical NF-κB signaling (16). It occurs in different cell types such as murine embryonic fibroblasts and Jurkat T cells (17) as well as in hepatocytes (18, 19).

As sepsis is characterized as systemic hyperinflammation due to cytokine overproduction, TNF-α-induced antiapoptotic NF-κB signaling may play a decisive role in the control of lymphocyte apoptosis during sepsis. In this work we examined whether lymphocyte apoptosis in a polymicrobial murine sepsis model is associated with reduced NF-κB activity. We also tested the hypothesis that enhanced lymphocyte NF-κB activation during sepsis prevents T cell death and investigated the effect of increased lymphocyte NF-κB activation on survival rates of septic mice.

Breeder mice of Rag1^−/− and Iκbα^+/− strains on the C57BL/6 background were purchased from The Jackson Laboratory. They were maintained on sterile standard laboratory chow and water ad libitum in individual ventilated cages under specific pathogen-free conditions in the animal facility of the University of Ulm (Ulm, Germany). All animal experiments were approved by the Animal Study Committee of the Regierungspräsidium (Tübingen, Germany).

The approach of cecal ligation and puncture (CLP)⁴ is a well established and widely used clinically relevant model of polymicrobial peritonitis/sepsis (20). For this purpose female mice and mice 8 wk after adoptive transfer of fetal liver cells weighing 20–30 g were anesthetized, the cecum was isolated, ligated with 4–0 silk, and punctured twice with a 18-gauge needle. Sham-operated mice only had laparatomy and cecum manipulation. After the abdomen was closed in two layers the mice received fluid therapy with 1.0 ml of 0.9% sodium chloride solution s.c. and analgesia with 25 μg/kg buprenorphine i.m. At certain time points postsurgery, CLP and sham mice were killed and thymi and spleens were removed. For survival studies, CLP was performed by an investigator blinded to the identity of the mice. Besides the i.m. application of buprenorphine and 0.5 ml of 0.9% sodium chloride solution s.c. twice a day, no additional manipulations were performed. Survival was recorded for 6 days.

Fetal liver cells were harvested from E15 embryos and placed in 1 ml of IMDM, 2% FCS. For rapid genotyping, embryonic tissue was digested in 100 μl of 1× lysis buffer, containing 1 mg/ml proteinase K, and incubated for 1 h at 55°C. Genomic DNA was isolated and analyzed by PCR with IκBα-specific primers as described (21). Fetal liver single cell suspensions were prepared by repeated passage through a 21-gauge needle followed by one passage through a 26-gauge needle. Subsequently, cells were passed through a 40-μm cell strainer. Within 6 h, 1 × 10⁶ fetal liver cells were injected into the tail vein of an 8-wk-old Rag1^−/− female host. Before injection, host animals were sublethally irradiated with 600 rad of gamma irradiation. Two days before injection, host mice were maintained under sterile conditions using autoclaved cages, with water containing 25 mg/L neomycin sulfate and 13 mg/L polymyxin B sulfate. Before additional experiments, blood was taken and lymphopoetic reconstitution was verified via flow cytometric detection of peripheral CD4⁺ and CD8⁺ cells.

Lymphoid organs were fixed in 10% buffered formalin and embedded in paraffin. After routine processing, 5-μm sections were stained with H&E for histological analysis. TUNEL staining was performed using the In situ Cell Detection kit according to the manufacturer’s instructions (Boehringer Mannheim). Single cell populations from thymus and spleen were prepared as described (22) and were preincubated with mouse Fc block (BD Pharmingen) for 15 min on ice and washed with FACS buffer (0.1% BSA, 0.1% NaN₃ in PBS). Cells were then incubated with directly fluoresceinated mAbs for 1 h on ice and subsequently analyzed by means of flow cytometry. Conjugated anti-mouse CD3, CD4, CD8, CD69 (H1.2F3), and isotype control Abs were obtained from BD Pharmingen. Membrane integrity was routinely tested by exclusion of propidium iodide (PI), all preparations were >95% PI-negative. Cell death analysis was performed by means of DNA staining of permeabilized complete cells with PI. Briefly, 1 × 10⁶ thymocytes were washed twice in PBS, fixed and permeabilized in ice-cold 70% ethanol, washed in PBS, and resuspended in 0.5 ml of staining solution (0.2 mg PI, 2 μg of RNase in PBS). Flow cytometric analysis was performed within 1 h. Expanded hypodense population represents dead cells in general, whereas cells undergoing classical apoptosis normally accumulate in a distinct, narrow sub-G₁ peak.

For cell extract preparation cells (2 × 10⁷) were resuspended in lysing buffer (50 mM Tris-HCl (pH 7.6), 250 mM NaCl, 3 mM EDTA, 3 mM EGTA, 1% Triton X-100, 0.5% Nonidet P-40, 10% glycerol) and lysed on ice for 30 min followed by centrifugation at 14,000 rpm for 30 min. The supernatant (protein extract) was collected and stored at −80°C. All steps were conducted at 4°C. All buffers were supplemented with 1 mM β-glycerolphosphate, 2 mM DTT, 1 mM PMSF, 10 μM leupeptin, 2 mM p-nitrophenyl phosphate, and 0.1 mM orthovanadate. Immunoblots were performed using Bio-Rad Protean II Minigels with 10 μg of protein extracts per lane. Proteins were blotted on nitrocellulose, and membranes were blocked with 7.5 or 10% nonfat dried milk in PBS, with 0.3% Tween 20. Abs used were IκBα (C-21; Santa Cruz Biotechnology), receptor interacting protein (RIP, clone 38; BD Biosciences), actin (C-11; Santa Cruz Biotechnology), p100/p52 (no. 4882; Cell Signaling Technology), p65 (F-6; Santa Cruz Biotechnology), p65 (for immunoprecipitation) (C-20; Santa Cruz Biotechnology), phospho-p65 (Ser⁵³⁶; Cell Signaling Technology), acetylated lysine (Ac-K103; Cell Signaling Technology), Bcl-x_L (2H12; BD Biosciences), c-IAP2 (cellular inhibitor of apoptosis protein, H-85; Santa Cruz Biotechnology), Bcl-2 (no. 554279; BD Biosciences), and caspase-3 (8G10; Cell Signaling Technology). For band shift assays, cell extracts (10 μg) were incubated in a 10-μl reaction for 30 min with 0.1 μg/μl polyinosinic-polycytidylic acid (Pharmacia) and 20,000 cpm [α-³²P]dATP-labeled oligonucleotide in 1 mM DTT, 10 mM HEPES (pH 7.6), 50 mM KCl, 6 mM MgCl₂, 1.2 mM CaCl₂, 1 mM DTT, 5% glycerin. Complexes were separated in native 4% polyacrylamide gels for 3 h. Gels were dried and exposed to x-ray films. The following oligonucleotides were annealed to form double-stranded gelshift oligonucleotides: NF-κB (HIVκB-site) sense 5′-GGATCCTCAACAGAGGGGACTTTCCGAGGCCA-3′, reverse 5′-GGATCCTGGCCTCGGAAAGTCCCCTCTGTTGA-3′; and NF-1 sense 5′-TTTTGGATTGAAGCCAATATGATAA-3′, reverse 5′-TTATCATATTGGCTTCAATCCA-3′.

For statistical analysis differences in group survival were determined using Fisher’s exact p test. Other data were analyzed using the Mann-Whitney U test. Statistical significance was set at p < 0.05. All data were expressed as mean ± SE.

Thymi were harvested 36 h after sham or CLP surgery to investigate the onset of programmed cell death. As expected, thymocyte apoptosis as well as structural disintegration was clearly more pronounced in the course of polymicrobial sepsis compared with sham mice as demonstrated by TUNEL and H&E staining (Fig. 1,A). Apoptotic effector mechanisms were further verified via detection of increased procaspase-3 cleavage in septic mice compared with sham mice (Fig. 1,B). Cellular analysis of septic thymi revealed a marked decrease in cellularity compared with sham-operated mice (data not shown). As demonstrated by flow cytometry, low thymocyte numbers particularly derive from a decline of premature thymocytes, namely CD4⁺CD8⁺ double positive (DP) cells (Fig. 1,C). Thus, the pattern of thymocyte apoptosis seen in this sepsis model is consistent with previous data (23, 24). Interestingly, NF-κB DNA binding activity significantly decreased over time in septic thymocytes. Compared with sham mice, an initial increase of activation after 3 and 6 h upon CLP surgery is followed by a dramatic inactivation or suppression of NF-κB binding activity in septic thymi (Fig. 1,D). This response is associated with clearly reduced expression of NF-κB-dependent antiapoptotic proteins such as Bcl-x_L and c-IAP2, but not Bcl-2 (Fig. 1 E).

To more precisely evaluate potential mechanisms that limit the duration and magnitude of NF-κB signaling in septic thymocytes, distinct negative regulators were analyzed. So far, the mechanisms that limit the amount of NF-κB activation are incompletely understood but are known to operate at several levels. De novo NF-κB-induced IκBα gene expression plays a key inhibitory role in this pathway (25). IκBα was expressed in similar levels in both sham and CLP thymi (Fig. 2,A), suggesting other counter-regulatory mechanisms in this model. The alternative pathway of NF-κB activation via IκB kinase α that promotes p100 processing to p52 and transcription by p52-RelB heterodimers (26, 27) did not differ between control and septic thymocytes (Fig. 2,B). Hence, this pathway seems not to play a major role in down-regulating NF-κB in septic thymocytes. Another negative regulatory mechanism of NF-κB depends on A20, an E3 ubiquitin ligase that promotes ubiquitination of RIP, an essential mediator of the proximal TNFR1 signaling complex (28). A20 down-regulates TNF-α-induced NF-κB signaling by catalyzing RIP ubiquitination, thereby promoting its subsequent proteasomal degradation (29). Interestingly, RIP protects thymocytes from TNFR2-induced cell death (30). To investigate whether A20-induced RIP degradation is involved in NF-κB down-regulation RIP expression was monitored in septic thymocytes. However, RIP levels were stable over 36 h in both sham and septic thymocytes (Fig. 2 A), suggesting that this counter-regulatory mechanism is of minor importance as well.

Posttranslational modifications by phosphorylation or acetylation, e.g., of p65 efficiently modulate DNA-binding and/or target gene transactivation of NF-κB, thereby controlling the strength and duration of the nuclear NF-κB response. Thus, hyperphosphorylation appears to be critical for full-target gene transactivation by the NF-κB complex (31), whereas p65 deacetylation of activated NF-κB terminates its transcriptional response (32). In this experiment, slightly reduced phosphorylation status of p65 but clearly decreased acetylation was detected in thymocytes 36 h upon sepsis induction compared with sham mice (Fig. 2, C and D). Deacetylated p65-containing NF-κB complexes bind IκBα more efficiently resulting in increased export to the cytoplasm (32), a process that reasonably explains NF-κB inactivation in thymocytes during murine sepsis.

We postulated that augmentation of thymocyte NF-κB activation during sepsis may reduce programmed cell death. To investigate this hypothesis, we performed adoptive transfer experiments. IκBα-deficient and wild-type (wt) embryonic fetal liver stem cells were transferred into lymphopenic Rag1^−/− host mice. Using a similar approach normal development of IκBα-deficient lymphocytes in vivo was demonstrated earlier (33). It is well known that Iκbα^−/− cells reveal increased inducible and sustained NF-κB DNA binding activity upon stimulation (25). Basically, we confirmed these data. Successful lymphopoetic reconstitution was verified after 6 wk and CLP surgery was performed thereafter. Upon CLP, IκBα-deficient thymocytes demonstrate increased NF-κB activation throughout the observation period compared with wt-reconstituted host thymi (Fig. 3, A and B). Also, apoptotic rates of IκBα^−/− thymocytes were clearly lower compared with wt thymocytes as demonstrated by TUNEL (Fig. 3,C) and PI staining (Fig. 3,D). As demonstrated earlier by Chen et al. (33), cellularity of IκBα-deficient thymi was reduced compared with wt-reconstituted thymi before CLP, which is not due to apoptosis (data not shown). However, thymus cellularity was similar upon sepsis induction in both types of host mice, which suggests increased cell death in wt hosts (data not shown). Flow cytometric detection of CD4- and CD8-expressing cells revealed an alteration in the cell populations during sepsis in IκBα^−/− compared with wt-reconstituted thymi. Hence, the proportion of single positive cells (CD4⁺ or CD8⁺) was increased in wt radiation chimeras compared with IκBα^−/− radiation chimeras, and CD4⁺CD8⁺ DP cells decreased. This pattern was mostly pronounced 36 h upon sepsis induction (Fig. 3 E). Notably, the decrease of CD4⁺CD8⁺ DP T cells is a hallmark of murine sepsis (23). Thus, under septic conditions IκBα-deficient thymi demonstrate reduced apoptotic rates and increased maturation of thymocytes.

Are septic IκBα^−/− radiation chimeras demonstrating a rise of peripheral T cells that enhances the immune status during sepsis? Most interestingly, numbers of CD3-positive T cells in peripheral lymph organs, such as the spleen, are increased in septic IκBα^−/− host mice compared with wt controls (Fig. 4,A). The functional status of IκBα^−/− T cells was determined by staining with anti-CD69 Abs, a surface marker indicating transient activation of T cells. Thus, the fraction of cells positively stained for CD69 during sepsis was moderately but clearly higher in the population of IκBα-deficient CD3⁺ splenocytes compared with that of wt CD3⁺ splenocytes (Fig. 4,B). In other words, IκBα^−/− T cells are more readily activated during sepsis. Moreover, survival curves of radiation chimeras during sepsis clearly demonstrated that increased NF-κB activation of IκBα^−/− T cells improves the outcome. Upon CLP surgery the 6-day mortality rate of 100% in normal C57/BL6 mice and IκBα^+/+ reconstituted Rag1^−/− host mice was reduced to ∼60% in IκBα^−/− radiation chimeras (Fig. 4 C). In conclusion, our data suggest that augmented lymphocyte NF-κB activation improves the adaptive immune system during murine polymicrobial sepsis.

The most intriguing result of this study is the down-regulation of thymocyte NF-κB activity in the course of sepsis. This result is in clear contrast to its well-known key function during systemic hyperinflammation in systemic inflammatory response syndrome, sepsis, and septic shock (34). In this experiment, the suppression of NF-κB in septic lymphocytes is occurring late in sepsis and most likely parallels a phase that is known as anergy or sometimes described as CARS (compensatory anti-inflammatory response syndrome) (35). The mechanisms that cause this anti-inflammatory response are poorly understood. We provide data that suppression of NF-κB in T cells during sepsis might contribute to this phenomenon. However, with respect to the exact knowledge of NF-κB activation, its inactivation is poorly understood. In our sepsis model, well-known mechanisms that negatively regulate NF-κB signaling, such as production of the inhibitor protein IκBα and subsequent suppression of classical NF-κB signaling or induction of the alternative pathway that may lead to IκB kinase α-dependent accelerated turnover of RelA and c-Rel (36), are at best partially responsible for NF-κB down-regulation. Also, negative NF-κB regulation via A20-dependent degradation of RIP (29), an essential mediator of TNF-α-induced signaling (28), is likely to be of minor importance. Posttranslational modifications such as phosphorylation and acetylation of RelA-containing complexes, however, are known to positively correlate with the duration of the NF-κB response (32). Especially acetylation of transcription factors leads to alterations in various nuclear functions such as DNA binding, transcriptional activity, and protein-protein interactions (37). In addition, chromatin acetylation is known to facilitate DNA accessibility (38). Hence, deacetylation of RelA and histone proteins via specific histone deacetylase (HDAC) correlates with reduced gene transcription or gene silencing (32, 38). Interestingly, recent data demonstrate that bacterial infections of epithelial cells affect the histone acetylation-deacetylation balance leading to NF-κB activation and expression of proinflammatory genes (39). Though it was shown that NF-κB may associate with HDACs, which leads to repression of NF-κB-regulated genes (40), it is not fully understood whether deacetylation of p65 is a specific task or a by-product of HDAC activity. Our results, however, suggest that in the course of sepsis, negative regulation of NF-κB in thymocytes primarily occurs via deacetylation of p65. This response might be the result of common enhanced HDAC activity in the context of a compensatory anti-inflammatory response syndrome and immunosuppression.

T cell development in the thymus starts with CD4⁻CD8⁻ double negative (DN) thymocytes, which progress to CD4⁺CD8⁺ DP thymocytes and finally to CD4⁺ or CD8⁺ single positive thymocytes, which exit the thymus and enter the circulation (41). All thymocytes have some constitutive NF-κB activity, but activity is particularly high in later DN thymocytes and essentially promotes the progression to late DN stages via pre-TCR signaling (42). Notably, ex vivo NF-κB inhibition in isolated late DN cells triggers apoptosis (42), suggesting that NF-κB provides a surviving signal. As in our study, it is known that murine sepsis particularly causes a loss of DP thymocytes, suggesting a block in transition of DN to DP cells (23). Most interestingly, forced activation of NF-κB by expression of a constitutively active Ikkβ transgene in RAG1-deficient (MLR-1) thymocytes, which cannot assemble the pre-TCR, allows partial progression of thymocytes to the DP stage (42). This result is consistent with our own demonstrating that enhancement of NF-κB activation improves the number of DP thymocytes during sepsis. Taken together, these data support our observations of thymocytic NF-κB suppression in the course of sepsis leading to increased apoptosis of late DN or DP thymocytes.

With the progression of severe sepsis counter-regulatory suppression of proinflammatory NF-κB signaling may reflect a compensatory anti-inflammatory response syndrome on a molecular basis. As this pathway regulates both innate and adaptive immune responses (43), its inactivation in lymphocytes is likely to have dramatic consequences on host defense mechanisms. Thus, negative regulation of NF-κB in T cells may explain, at least in part, the immunosuppression that is observed during late sepsis. Moreover, immunosuppression in murine and human sepsis is closely related to lymphocyte apoptosis (10, 11). In our study, NF-κB down-regulation during late sepsis is associated with reduced expression of antiapoptotic proteins and increased thymocyte apoptosis. Reduced peripheral T cell numbers are likely due to increased thymocyte apoptosis under septic conditions. This finding is consistent with the observation that NF-κB-defective lymphocytes from transgenic mice that specifically express a transdominant form of IκBα (IκB super-repressor) reveal impaired ex vivo proliferation and increased apoptotic rates upon mitogenic stimulation, whereas basal thymopoiesis is normal (44). In this context it is most interesting that signaling via the NF-κB pathway is essentially required to protect T cells from TNF-α-induced apoptosis (22), a cytokine that plays a significant pathophysiological role in sepsis (4). Consequently, we were able to demonstrate that augmentation of lymphocytic NF-κB constitutive and inducible activity via adoptive transfer of IκBα-deficient fetal liver stem cells into lymphopenic Rag1^−/− host mice and subsequent sepsis induction reduces thymocyte apoptosis and improves T cell numbers, T cell activation, and survival. These data are in line with results from Hotchkiss et al. (45) who showed that prevention of lymphocyte apoptosis through the use of caspase inhibitors or Bcl-2 overexpression (46) reduces lymphocyte apoptotic rates and improves survival of septic mice.

It is generally believed that septic challenge provokes an intense hyperinflammatory response to eliminate the underlying pathogen. The amount of inflammation, however, determines the septic disease course. Proinflammatory NF-κB signaling, however, is likely to play an important role in this scenario. NF-κB inhibitors such as epigallocatechin-3-gallate (47), ethyl pyruvate (48), and parthenolide (49) have been shown to improve survival in rodent CLP-type models of sepsis. Similarly, it was shown that fatal outcome of sepsis is associated with highly increased NF-κB activity in mononuclear cells of septic patients (50, 51). This outcome is not contradictory to our results. These studies focus on early sepsis and interventions were started not later than 1 h (47, 49) upon CLP surgery with 18-gauge puncture or within 24 h in a milder than our CLP model (48). Moreover, relative lymphopenia is a common observation during sepsis. As differential blood counts are not included in the septic patient studies, the NF-κB measurements are likely to reflect primarily the activation of peripheral monocytes in the context of an initial hyperinflammatory response. Taken together, the aforementioned studies investigate an early, rather proinflammatory role of NF-κB signaling in sepsis. By the time the host increasingly succeeds in overcoming the infection, mechanisms are induced to counter-regulate and control this initial response to finally resolve the inflammation. Considering the results of our study, it is conceivable that fulminant initial NF-κB-driven inflammation by innate immune cells may be associated with or may even cause subsequent suppression of NF-κB signaling in the adaptive immune system, particularly during later time points. Excessive negative regulation of NF-κB, however, may result in deterioration of immunological functions and failure to ward off the infection, a situation often seen in clinical settings. Our data support the idea that antiapoptotic NF-κB signaling in lymphocytes is important to mount proper adaptive immune responses in the course of sepsis to fight the pathogenic microbe. Possibly, preterm or excessive down-regulation of NF-κB signaling may promote immunosuppression and anergy resulting in multiple organ failure and death. This study provides evidence for potential new therapeutic interventions in late stages of sepsis by augmentation of NF-κB activity, e.g., via the inhibition of HDACs or via IκBα RNA interference. Moreover, it points at potential adverse effects such as lymphopenia that may be caused by inadequate NF-κB inhibition.

We thank F. Weih and S. Schlottmann for experimental support and B. Stahl, S. Heydrich, and R. Mayer for technical assistance. We are particularly grateful to M. Karin for fruitful discussions and for support in preparing this manuscript.

The authors have no financial conflict of interest.