Background: Abdominal sepsis due to intestinal leakage of endogenous gut bacteria is a life-threatening condition. In healthy individuals, T lymphocytes have essential functions in balancing the immune response to the commensal gut flora.
Aim: To determine how T lymphocytes shape the process of diffuse faecal peritonitis.
Methods: In colon ascendens stent peritonitis (CASP), a clinically relevant mouse model of diffuse peritonitis, the kinetics of systemic T cell activation were investigated by assessment of activation markers. CD4+ T cells were then depleted with monoclonal antibodies, and survival, bacterial dissemination and cytokine concentrations were measured. T cell receptor signalling was blocked with tacrolimus.
Results: In diffuse peritonitis, CD4+ T cells, both Foxp3− and Foxp3+, became systemically involved within hours and upregulated CTLA-4 and other activation markers. Depletion of the CD4+ T cells enhanced local bacterial clearance from the peritoneal cavity, reduced bacterial dissemination and improved survival. This was accompanied by increased immigration of granulocytes and macrophages into the peritoneum, indicating that CD4+ T cells inhibit the local innate immune response. Blockade of T cell receptor (TCR) signalling by tacrolimus did not influence the survival in this peritonitis model, showing that the inhibitory effects of the CD4+ T lymphocytes were independent of TCR-mediated antigen recognition.
Conclusion: In diffuse peritonitis caused by commensal gut bacteria the CD4+ T lymphocytes exert a net negative effect on the local anti-bacterial defence, and thereby contribute to bacterial dissemination and poor outcome.
Statistics from Altmetric.com
Invasion of gut bacteria into the peritoneal cavity due to intestinal leakage is a life-threatening complication of abdominal surgery.1 2 It carries a high risk of severe sepsis, which endangers the organism in two ways: first, exposure to large amounts of pathogen-associated molecular patterns (PAMPs) causes a systemic inflammatory response syndrome (SIRS), which can culminate in lethal septic shock; secondly, SIRS may be followed by a compensatory anti-inflammatory response syndrome (CARS), which compromises the antimicrobial defence.3 4 There is increasing evidence that T lymphocytes have an important role in shaping the balance between inflammation and anti-inflammation in bacterial infection. Their action can be beneficial or detrimental for the host.5
Apoptotic loss and functional impairment of T lymphocytes, which are frequently observed in sepsis, trauma and stroke, compromise the host defence against endogenous as well as exogenous bacteria.6–9 Accordingly, adoptive transfer of T cells from untreated animals prevented lethal infection with commensal gut bacteria, which otherwise was a regular consequence of cerebral ischaemia in mice.8 Furthermore, prevention of cell death improved the survival in a murine model of sepsis.10 11 This shows that effector T cells contribute to the control of systemic bacterial infection.
Conversely, the inflammatory response to pathogens or their PAMPs can be more dangerous than the direct effects of the invading micro-organisms. This is most clearly exemplified by lipopolysaccharide (LPS) bolus injections. LPS is harmless per se, as shown in different immunodeficient mutant mice (Toll-like receptor 4 (TLR4)-deficient, lipopolysaccharide-binding protein (LBP)-deficient, tumour necrosis factor (TNF)- or TNF receptor-deficient, interferon (IFN)γ-deficient), which are exquisitely susceptible to bacterial infection but highly resistant to purified LPS.12 In contrast, LPS causes lethal shock in immune-competent animals. This has been attributed to the massive secretion of inflammatory cytokines by cells of the innate immune system, in particular by monocytes and macrophages. However, it was recently shown that depletion of αβ-T lymphocytes strongly enhances the toxicity of LPS, suggesting that T cells modulate the inflammatory response of the innate immune system. This dampening effect has been attributed to the action of regulatory T cells (Tregs).13
Different subpopulations of Tregs can be distinguished. First, naturally occurring thymus-derived Tregs are CD4+CD25+ and express basal levels of CTLA-4 (CD152) as well as the transcription factor Foxp3. Secondly, antigen-induced Tregs are characterised by their ability to secrete interleukin (IL)10 (Tr1) or transforming growth factor (TGF)β (Th3).5 14 To date, Foxp3 is considered to be the most reliable marker for Tregs.15 In contrast, CTLA-4 is not exclusively expressed on Tregs, but is also induced on effector T cells after activation. The molecule performs an essential inhibitory function, counteracting the activating co-receptor CD28. Similarly to CTLA-4, the inducible co-stimulator (ICOS, CD278) is upregulated following T cell activation, but its ligation enhances the secretion of effector cytokines.16–18
Tregs are important for the prevention of excessive inflammation5 and the maintenance of immune tolerance. This is vital in the dynamic interactions between the commensal gut flora and the immune system,19 because PAMPs, such as LPS and CpG-DNA, are shared by pathogenic and commensal bacteria.19 20 It has recently been shown that even in germ-free animals a significant proportion of T cells are able to respond to intestinal bacteria in an antigen-specific manner with strong proliferation and secretion of inflammatory cytokines.21 22 In the absence of silencing signals, these enteroantigen-reactive T cells cause chronic inflammatory bowel disease.23 24 In an immune-balanced situation, the sources of silencing signals are Tregs.19 25 It has been shown that Tregs express TLRs, and that their suppressive activity can be elicited or increased by LPS.26
Given the fine-tuned balance between regulatory and effector T cells with reactivity against intestinal bacterial antigens, we were interested in the T cell reactions to systemic infection with gut bacteria. Therefore, we have studied their role in colon ascendens stent peritonitis (CASP).27 28 In this mouse model of diffuse peritonitis, the clinical situation of patients with colonic leakage following surgical procedures is mimicked by insertion of a stent with defined diameter into the ascending colon. CASP surgery leads to rapid and systemic infection and inflammation. Cytokines that are known to be involved in T cell–phagocyte interactions, such as IFNγ,28 IL1229 and IL10,30 were shown to have crucial functions in the CASP model. In the present study, we show that T cells very rapidly respond to the bacterial infection, and that they suppress the innate mechanisms of local bacterial clearance.
MATERIALS AND METHODS
Female C57BL/6 mice at the age of 8–12 weeks (Charles River, Sulzfeld, Germany) were housed in a conventional animal facility and adapted for at least 2 weeks. All experiments were performed according to the German animal safety regulations and approved by the local animal protection authority.
CASP surgery was performed as described before.27 28 Briefly, mice were anaesthetised with ketamine and a 16G stent was implanted into their colon ascendens. In sham operations, the stent was fixed outside the ascending colon without puncturing the colon wall. After 6–24 h the animals were anesthetised, and blood, peritoneal lavage fluid, liver, spleen and thymus were recovered.
Superantigen shock was induced by intraperitoneal injection of 20 μg of staphylococcal enterotoxin B (SEB) together with 20 mg of galactosamine.
For the depletion of CD4+ T cells, 150 µg of rat anti-mouse CD4 monoclonal antibody (mAb; clone GK1.5, ATTC) was injected intraperitoneally on days −3 and −1 before CASP surgery. This efficiently depleted CD4+ cells without affecting CD8+ T lymphocytes. Control mice received phosphate-buffered saline (PBS) (fig 1). To interfere with TCR signalling, animals received 50 μg of tacrolimus (FK506, Fujisawa, Munich, Germany) on days −3 and −1 before CASP surgery or induction of superantigen shock. Control animals received PBS.
For peritoneal lavage, 10 ml of PBS (ice-cold) were injected into the peritoneal cavity. The recovered cell suspension was washed twice in PBS and resuspended in cell culture medium (RPMI 1640, containing 10% fetal calf serum (FCS), glutamine and penicillin/streptomycin). Spleen and thymus cells were depleted of erythrothytes (BD lysis buffer) before washing and resuspension in cell culture medium.
Determination of the bacterial load
The animals were sacrificed 20 h after CASP surgery. Blood, peritoneal lavage fluid and homogenised organ suspensions were incubated on Colombia blood agar (Becton Dickinson) for 22 h at 37°C, and bacterial colonies were enumerated.
Antibodies and staining reagents
The hamster anti-CTLA-4 mAb UC10-4F10 was a kind gift of Dr J Bluestone. Rat monoclonal antibodies directed against mouse CD3 (17A2), CD69 (H1.2F3), ICOS (7E.17G9) and Gr1.1 (RB6-8C5) coupled to either fluorescein isothiocyanate (FITC) or phycoerythrin (PE) as well as appropriate isotype controls were obtained from BD Biosciences (Heidelberg, Germany), anti-Mac-1 antibody (CI:A3-1) from Acris (Hiddenhausen, Germany), rat anti-Foxp3 unlabelled or coupled to biotin from NatuTec (Frankfurt/Main, Germany), and the rat anti-CD4-producing hybridoma GK1.5 was from ATCC. A biotinylated cocktail of mouse mAbs directed against Syrian and Armenian hamster immunoglobulin (Ig)G was from BD Biosciences, as were goat anti-rat IgG–FITC and streptavidin–TRITC (tetramethylrhodamine isothiocyanate). Mouse anti-FITC mAbs coupled to the fluorochrome Alexa488 and streptavidin–Alexa647 were obtained from Molecular Probes (Eugene, OR, USA), and the tyramide signal amplification kit (TSA, NEL700) from NEN (Boston, MA, USA). For intracellular staining, cells were permeabilised with the Fix and Perm permeabilisation kit (An der Grub Bio Research, Kaumberg, Germany).
Flow cytometry and fluorescence microscopy
For flow cytometric measurement of activation markers, T cells were stained with PE-labelled mAbs against CD69, CD25, ICOS or isotype controls in conjunction with FITC-labelled anti-CD3. For determination of CTLA-4 expression, cells were fixed, permeabilised and then incubated sequentially with an anti-CTLA-4 mAb, the biotinylated mouse anti-hamster mAb cocktail and streptavidin–Alexa647. Hamster IgG served as isotype control. Organ samples were snap-frozen in TissueTek embedding medium (Sakura, Zoeterwonde, The Netherlands) in liquid nitrogen. Cryostat sections (6 µm) were fixed in acetone (−20°C) and air dried. CTLA-4 was visualised on CD4+ T cells as follows: slides were incubated sequentially with 20% FCS in PBS, 0.1% H2O2 and a biotin-blocking buffer (DAKO, Glostrup, Denmark) to reduce non-specific staining, inactivate endogenous peroxidase and block endogenous biotin. They were then incubated with hamster anti-mouse CTLA-4 and rat anti-mouse CD4 in 20% FCS/PBS overnight. After 30 min incubation with a mixture of biotinylated anti-hamster mAb-cocktail and a FITC-coupled goat anti-rat IgG antiserum, the slides were postfixed with 4% paraformaldehyde in PBS for 15 min. A further peroxidase inactivation step (3% H2O2, 10 min) followed. Then TSA was used to enhance the CTLA-4 signal according to the manufacturer’s instructions. Finally, CTLA-4 was visualised with streptavidin–TRITC and CD4 with a mouse anti-FITC–Alexa488 mAb. Double staining for CTLA-4 and Foxp3 was performed similarly, using an anti-Foxp3 mAb followed by anti-rat IgG–FITC and anti-FITC–Alexa488. To stain Foxp3+ T cells on tissue sections, anti-CD3–FITC was used in conjunction with anti-FITC–Alexa488, and the binding of biotinylated anti-Foxp3 mAbs was visualised using TSA.
Apoptotic cells were visualised with terminal deoxynucleotidyl transferase-mediated dUTP nick end labelling (TUNEL) using the In-Situ-Cell-Death-Detection Kit (Roche, Mannheim, Germany) according to the manufacturer’s instructions. Staining was analysed using the imaging program Metamorph (Visitron, Puchheim, Germany). Quantification was based on the sizes of the green- or red-stained areas, because the individual lymphocytes cannot be separated in the densely packed lymphoid organs. To compare, for example, Foxp3 expression on splenic T cells between experimental groups, the program determined the ratio between the size of the red (Foxp3) and green (CD3) areas (numbers of pixels) over the whole T cell area, which had been outlined manually. Results were expressed as percentages.
Statistical analysis was performed using GraphPad Prism for Windows software (GraphPad software, San Diego, CA, USA). Statistical differences in survival were assessed using Kaplan–Meier survival curves and log-rank test. Differences in bacterial loads, phenotype, cell invasion, cytokine expression and corticosterone between experimental groups were analysed with the two-tailed Mann–Whitney U-test. p-values <0.05 were considered to be significant.
Rapid involvement of T lymphocytes in polymicrobial sepsis
To assess the response of T cells in peritonitis, mice were subjected to CASP, and the activation markers CD69, ICOS and CTLA-4 were measured on splenic T cells. In untreated animals, only a small proportion of the CD3+ splenocytes was activated (6.6% (SD) (0.6%) expressed CD69, 2.0% (2.0%) ICOS and 2.1% (3.3%) CTLA-4). At 12 h after CASP, all markers were strongly upregulated: 22.8% (7.2%) of the T cells were now CD69+ (p<0.001), 7.4% (2.9%) expressed ICOS (p<0.01) and 14.8% (6.4%) were CTLA-4+ (p<0.001). Double-staining of cryosections revealed that CTLA-4 was expressed almost exclusively on CD4+ T cells. Surprisingly, increased CTLA-4 expression was observed as early as 3 h after CASP surgery (fig. 2). Sham-operated animals did not induce CTLA-4 on their splenic T cells (not shown). We then stained Foxp3 to label naturally occurring Tregs and see which T cell subpopulations expressed CTLA-4. While the density of Foxp3+CD3+ cells in the T cell areas had not changed significantly 24 h after CASP, they had strongly enhanced their expression of CTLA-4. In addition we observed strong de novo CTLA-4 expression on Foxp3-negative cells, which 24 h after CASP had increased from 30.6% (11.7%) to 55.2% (9.62%) of all CTLA-4-expressing T cells (n = 5, p<001, fig. 3). The findings demonstrate a very rapid systemic response of T lymphocytes, including Tregs, following an infection with commensal gut bacteria.
CASP was accompanied by very rapid onset of lymphocyte apoptosis. In the thymic cortex, this led to almost complete deletion of thymocytes within 18 h, whereas no increase or even a decrease of cell death was observed in the medulla (fig. 4). In the white pulp of the spleen, the fraction of apoptotic cells increased from 4.4% in untreated animals to 23.2% 12 h after CASP (p<0.001, not shown). A likely explanation is the strong increase of corticosterone concentrations in CASP (Supplementary fig 1).
CD4 depletion improved survival in sepsis
To address the role of CD4+ helper T cells in diffuse peritonitis, experimental mice were pretreated with an anti-CD4 mAb (GK1.5) on days −3 and −1 before CASP. Elimination of the CD4+ T cells improved survival: Whereas only 25% of control animals survived, the fraction increased to >75% after CD4+ T cell depletion (fig. 5, four independent experiments with similar outcome, p = 0.0004). Thus, unexpectedly, our results show a net negative effect of CD4+ T lymphocytes on the outcome of systemic infection with commensal gut bacteria. CD8+ T cells appeared to be largely unaffected as judged by their CD28 expression, which remained unchanged following depletion of CD4+ cells and was slightly upregulated in CASP in both animal groups (fig 6)
Bacterial dissemination and cytokine expression
The reduced mortality of CD4+ T cell-depleted animals was associated with an improved local bacterial clearance and reduced systemic bacterial dissemination. Significantly lower numbers of colony-forming units were detected 20 h after CASP surgery in the peritoneal cavity, the blood and in all investigated tissues (liver, spleen, kidney and lung; fig 7).
TNF, IL6, monocyte chemoattractant protein-1 (MCP-1) and IL10 were strongly upregulated in peritoneum and serum 12 h after CASP. Differences between CD4-depleted and control animals were generally small. However, in the T cell-depleted animals, cytokine levels in the peritoneum tended to be increased, which contrasted with consistently lower serum cytokine concentrations (Supplementary table 1). This suggests that the elimination of CD4+ T cells relieved a restraint from the local antibacterial defence mechanisms in the peritoneum, so that fewer bacteria reached the general circulation. Consequently, the severity of the systemic inflammatory response was mitigated and the outcome improved. While CD4 depletion per se did not lead to invasion of inflammatory cells into the peritoneum, in CASP the influx of immune cells, mainly granulocytes and monocytes, was strongly increased in CD4+ T cell-deficient animals (fig 8), indicating a negative effect of T cells on granulocyte and monocyte extravasation. However, no T cells were observed in the peritoneum at any time, suggesting that their effect in immune cell migration patterns was indirect.
Effects of tacrolimus
How the T cells were activated after the onset of polymicrobial sepsis following CASP surgery is unclear. Usually, T cells become activated by cross-linkage of the TCR with a T cell-specific peptide (or superantigen) bound to major histocompatibilty complex (MHC) molecules on antigen-presenting cells (APCs). Tregs have been reported to express TLRs and to respond directly to molecular patterns such as LPS.26 To discriminate between TCR-dependent and independent activation pathways, TCR signalling was blocked by tacrolimus (FK506). Superantigen shock (SEB), well known to be TCR-mediated, was used as a control. It was completely prevented by pretreatment with tacrolimus (fig 9B, p<0.05). In contrast, inhibition of TCR signalling did not influence the survival of CASP (fig 9A), suggesting TCR-independent activation of T cells in diffuse peritonitis.
The results show that in diffuse polymicrobial peritonitis, CD4+ T cells, including Tregs, became systemically activated within hours. Their action inhibited the local antibacterial defence, thereby contributing to bacterial dissemination and poor outcome.
Similar to the situation in human sepsis and other mouse models of bacterial infection,4 10 31 we have observed in CASP a rapid loss of immune cells due to apoptosis, most pronounced in the thymic cortex. This was probably caused by glucocorticoids,32 33 which were released at very high concentrations. Death receptors and their ligands, such as TNF and Fas-ligand, could also contribute to cell death induction in CASP. In addition, there was increased expression of CTLA-4 on the surviving cells. Upregulation of CTLA-4 has also been observed in human SIRS and sepsis, where it was attributed to the selective loss of CTLA-4-negative T cells.34 35 CTLA-4+ T cells and Tregs appear to be relatively resistant to apoptosis induced by different agents including corticosteroids.36 37 However, in the observed time frame, cell death in the spleen was not dramatic and cannot be the only explanation for the increase of CTLA-4-expressing T cells in CASP. Furthermore, there was no significant increase in the fraction of Foxp3+ T cells. In addition, we observed a strong enhancement of the CTLA-4 staining intensity, both on Foxp3+ and on Foxp3− T cells. This can only be explained by an induction of CTLA-4 protein expression on Tregs and effector T cells. We have previously found similar increases in CTLA-4 expression in human malaria as well as in SIRS and sepsis (38 and unpublished observations). CTLA-4 induction regularly occurs after T cell activation and is thought to represent a feedback mechanism, which limits their proliferation and cytokine secretion.16
In addition to passive impairment of immune effector functions, CD4+ T cells actively suppressed the local innate immune response in CASP. This is illustrated by the reduced bacterial dissemination and the improved disease outcome after depletion of CD4+ T cells. Elimination of the CD4+ T cell subpopulation resulted in an increased influx of innate inflammatory cells into the peritoneal cavity, with improved local bacterial clearance, resulting in a reduction of systemic inflammation. We have shown previously that the very early influx of granulocytes into the peritoneal cavity is decisive for the outcome of murine peritoneal salmonella infection.39 Enhancement of an antibacterial immune response following depletion of CD4+ T cells has been reported before for Listeria monocytogenes, where the suppressive effect was attributed to CD4+CD25+ natural Tregs.40 These cells also suppress the inflammatory responses against Helicobacter spp. as well as against a number of parasites and viruses.5
In the case of commensal gut bacteria, Tregs have a particularly important part to play, because life-long tolerance has to be maintained.19 41 42 This may explain why suppressive activity dominates the T helper cell response against enteric bacteria in CASP. Natural Tregs, which control the immune response to the intestinal microflora, constitutively express CTLA-4 and rapidly upregulate this molecule after activation as well as in response to corticosteroids.22 43
There are indications that the T cell-mediated suppression of antibacterial inflammation may even be “hard wired” in the immune system: CD4+CD25+ Tregs which control antibacterial responses have been found in naïve and even in germ-free animals. Also, in contrast to CD4+CD25– effector T cells or CD8+ T cells, which recognise bacterial antigens after processing and presentation in the context of the MHC, “conventional” antigen recognition has not been shown for the Tregs, which suppress them.40 Our data agree with these observations. In contrast to superantigen-induced T cell activation, which is TCR dependent,44 the net suppressive effect of CD4+ T cells in CASP was not abolished by tacrolimus (FK506), a drug which blocks TCR signalling. This indicates that the T cell involvement in CASP must be at least partially antigen independent and mediated by other T cell surface receptors. It has been shown that “natural” Tregs express TLRs, which may enable them to recognise conserved microbial structures similar to innate immune cells.26 There is also increasing evidence that, besides controlling adaptive immunity, CD4+ Tregs feed back on innate immunity and downregulate innate inflammatory responses.13 45
Our results identify CD4+ T cells as important players in acute bacterial infection. They show that, in the case of systemic dissemination of commensal bacteria, T cells may dampen the defence mechanisms to a degree which is life threatening. The findings may open a door for the development of new strategies to treat these dangerous conditions, but this will not be an easy task. In the attempt to enhance the bacterial clearance but avoid general immune suppression, which is associated with CD4 depletion, we have blocked CTLA-4 in CASP to interfere selectively with the dampening signals. As expected, this treatment reduced the bacterial load, but it did not improve survival, presumably because the animals succumbed to hyperinflammation (not shown). CD4+ T cells are composed of numerous subpopulations—besides Tregs there are Th1, Th2 and Th17 cells—each of which probably has unique regulatory functions in systemic bacterial infection. Further work is needed to define their precise roles. Activated Th1 cells, for example, secrete large amounts of IFNγ, a cytokine which is essential for survival of CASP.28 It was therefore unexpected that depleting CD4+ T cells conferred a survival advantage. Our results imply that the IFNγ must be derived from a different source. Th17 cells have been implicated in defence against extracellular bacteria, because they regulate the maturation and activation of granulocytes.46 However, in CASP they were not necessary for survival. The situation may be different when the infecting bacteria are more virulent than the commensal gut bacteria, which cause sepsis in CASP.
Since CD4+ T cells fulfil multiple functions in bacterial infection, their complete elimination will not be an option in the treatment of sepsis. The challenge will be to identify checkpoints in the T cell system, which are amenable to therapeutic manipulation, yet do not result in unbalancing the immune system.
The authors are grateful to J Bluestone for the kind gift of the anti-mouse CTLA-4 mAb, to A Müller for expert technical assistance, and to S Holtfreter for helpful comments.
Funding: This work was financially supported by the Bundesministerium für Forschung und Technologie (BMBF-NBL3, FKZ 01ZZ0403, Modul E5) and by the Deutsche Forschungsgemeinschaft (GRK-840).
Competing interests: None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.