Pro-Inflammatory Th1 and Th17 Cells Are Suppressed During Human Experimental Endotoxemia Whereas Anti-Inflammatory IL-10 Producing T-Cells Are Unaffected

Objective Sepsis is one of the leading causes of the deaths in hospitals. During sepsis, patients are exposed to endotoxemia, which may contribute to the dysregulation of the immune system frequently observed in sepsis. This dysregulation leads to impaired pro-inflammatory responses and may increase the risk for secondary infections in sepsis. The experimental human endotoxemia model is widely used as a model system to study the acute effects of endotoxemia. Under physiological circumstances, the immune system is tightly regulated. Effector T-cells exert pro-inflammatory function and are restrained by regulatory T-cells (Tregs), which modulate pro-inflammatory effector responses. Endotoxemia may induce inadequate Treg activity or render effector T-cells dysfunctional. It was the aim of the study to investigate effector T-cell and Treg responses in an experimental human endotoxemia model. Methods In a cross-over designed placebo-controlled study, 20 healthy male volunteers received an intravenous injection of either lipopolysaccharide (LPS) (0.8 ng/kg body weight) or a placebo (saline 0.9%). CD3+ T-cells, CD4+ T-cells, CD8+ T-cells, and intracellular cytokine profiles were measured with flow cytometry at baseline and at repeated points after LPS/placebo injection. Complete blood cell counts were obtained with an automated hematology analyzer and cytokines were quantified by ELISA. Results Circulating neutrophils were significantly increased 2 h after LPS injection (p < 0.001) while absolute number of CD3+ T-cells, CD4+ T-cells, and CD8+ T-cells decreased (p < 0.001). Effector T-helper-cells (THs) showed a significant—but transient—decrease of pro-inflammatory IFNγ, interleukin (IL)-2, TNFα, and IL-17A production after LPS injection (p < 0.001). In contrast, the frequency of Treg and the capacity to produce IL-10 were unchanged (p = 0.21). Conclusion Effector THs fail to produce pro-inflammatory Th1-/Th17-associated cytokines after LPS challenge. In contrast, IL-10 production by Treg is not affected. Thus, endotoxemia-induced suppression of pro-inflammatory THs might be considered as a contributing factor to immunoparalysis in sepsis.

Methods: In a cross-over designed placebo-controlled study, 20 healthy male volunteers received an intravenous injection of either lipopolysaccharide (LPS) (0.8 ng/kg body weight) or a placebo (saline 0.9%). CD3 + T-cells, CD4 + T-cells, CD8 + T-cells, and intracellular cytokine profiles were measured with flow cytometry at baseline and at repeated points after LPS/placebo injection. Complete blood cell counts were obtained with an automated hematology analyzer and cytokines were quantified by ELISA.
conclusion: Effector THs fail to produce pro-inflammatory Th1-/Th17-associated cytokines after LPS challenge. In contrast, IL-10 production by Treg is not affected. Thus, Suppression of T-Cells in Experimental Endotoxemia Frontiers in Immunology | www.frontiersin.org May 2018 | Volume 9 | Article 1133 inTrODUcTiOn Sepsis is one of the leading causes of deaths in hospitals. Thera peutic options for sepsis patients are limited and mortality rates remain high (1)(2)(3). This lifethreatening syndrome develops as a result of a dysregulated immune response to a pathogen (4). In this case, the clearance of the pathogen is inefficient and there is con tinuous activation of specific proinflammatory pathways (5). At the same time, the effector response of the immune system is disturbed; and the innate as well as the adaptive immune system are hyporesponsive (4,5). In addition, a postmortem study by Boomer et al. revealed that patients with fatal clinical course of sepsis showed signs of severe immunological dysfunction (6). Splenocytes had a reduced capacity to produce proinflammatory cytokines upon stimulation, and splenic Tcells were diminished in numbers (6). This socalled "immunoparalysis" is a severe immunosuppressive state that makes the host susceptible for secondary infections (4,5,7). Experimental human endotoxemia, in which lipopolysaccharide (LPS) is administered to healthy volunteers, has been established as a model to study the diverse effects of endotoxemia (8)(9)(10)(11)(12). In this model, features of dys functional immunity can be observed. Leukocytes from healthy volunteers with LPS exposure show reduced responsiveness to ex vivo stimulation with LPS and other tolllikereceptor agonists (7)(8)(9)(10)(11)13). Therefore, the human experimental endotoxemia model was also used in a doubleblind placebocontrolled pilot study to investigate agents, which may reverse immunoparalysis (7). Recent studies emphasize the growing importance of effector Tcells and regulatory Tcells (Tregs) in sepsis (14). Effector Tcell subsets have proinflammatory function, are classified according to signature cytokines, and have pivotal role in defense against pathogens-Th1 cells produce interferon gamma (IFNγ) and interleukin (IL)2 to support cellmediated immunity; Th17 cells produce IL17 (Th17) and have a crucial role in the inflammatory response against parasites, extracellular, and fungal pathogens (15,16). Treg subsets with antiinflammatory capacity limit pro inflammatory Tcell responses (17). Tregs balance Tcell homeo stasis, activation, and function via numerous different mechanisms including secretion of IL10 (17,18). IL10 has been suggested as an important regulator in sepsis (4,19). It has not been well studied which type of Tcell subsets are affected by endotoxemia, and the kinetics of the Tcell dysfunction are not exactly known. The aim of this study was to characterize Tcell responses during endotoxemia. Therefore, Tcell function of proinflammatory effector Tcells and antiinflammatory Tregs was analyzed in a human endotoxemia model.

Participants
The study is a singlecenter, placebocontrolled, randomized, and singleblinded trial in a crossover study design. Healthy men aged 18-40 years were recruited by public advertisement. The extensive screening and safety procedure consisted of per sonal interview, conducted by a physician, a physical examination including an assessment of blood and clinical chemistry param eters (complete blood cell count, Creactive protein, coagulation factors, lactatedehydrogenase, myoglobin, creatinkinase, liver enzymes, renal, and hormonal parameters). Laboratory screen ing was conducted before each study day (LPS vs. placebo) and up to 1 week after completing the study. Participants were exc luded with reported or current medical and psychological condi tions, body mass index (BMI) <19.0 or ≥ 27.0 kg/m 2 , current medication, smoking, regular and/or high alcohol consumption, severe allergies, or depression scores exceeding published cut offs of the Beck Dep ression Inventory (BDI, 14). Additional excluding factors were extensive sport exercises 24 h before and after the study days and vaccinations within the last 2 months. One participant did not complete the +72 h time point within the LPS condition due to a case of family related acute gastroen teritis. The study protocol was approved by the local ethics com mittee of the University Hospital Essen, Germany (permission sign: 156533BO). All volunteers provided written informed consent in accordance with the Declaration of Helsinki and received financial reimbursement for their participation in the study.

study Protocol
The study comprised a placebo and a LPS condition, i.e., parti cipants received either LPS or placebo on two otherwise iden tical study days. The participants either started with the placebo condition followed by the LPS condition or with the LPS con di tion followed by the placebo condition, with a minimum of 7 days between study conditions. The order of study days was ran domized and counterbalanced (www.randomizer.org was used for randomization). The study took place in a medically equip ped room under supervision of an internal physician. After the arrival of the participants, an intravenous catheter was inserted in a cubi tal vein for repeated blood drawing and endotoxin injection. After a rest of 30 min, vital signs including heart and breath ing rate, pulse oximetry (Kernmed Oled, Ettlingen, Germany), and blood pressure (Dinamap Compact T, Critikon, Nor derstedt, Germany) were measured. One hour after arrival, subjects received an intravenous injection of 0.8 ng LPS/kg of body weight (Eschericha coli LPS 200 ng/ml, LOT HOK354, USP The United States Phar macopeial Convention, Inc., Rockville, MD, USA as previously described) under continuous vital sign monitoring (20). LPS had been consigned to the German federal Agency for Sera and Vaccination (PaulEhrlich Institute, Langen, Germany) for a microbial safety testing and was stored in endotoxinfree borosilicate tubes (Pyroquant Diagnostik, MörfeldenWalldorf, Germany) at −20°C until use. At study days, blood samples for complete blood counts and cytokine analysis were collected in EDTAcoated tubes at defined time points: before (baseline), as Keywords: systemic inflammation, T-cells, endotoxemia-induced suppression of pro-inflammatory THs might be considered as a contributing factor to immunoparalysis in sepsis.

White Blood cell count
Complete blood counts containing white blood cell (WBC), neutrophils, mono/lymphocytes, and platelets were obtained via automated hematology analyzer (KX21N, Sysmex Deutschland GmbH, Norderstedt, Germany) using EDTAanticoagulated peri pheral blood samples.

T-cell analysis
Surface phenotyping and intracellular cytokine analyses of Tcells were performed with freshly isolated PBMC. PBMCs were sta ined with antibodies for 30 min and analyzed immediately after washing with Dulbecco's phosphatebuffered saline (DPBS 1×; Gibco, Life Technologies) in case of surface phenotyping. Isotype controls were used to confirm specificity of staining and to dis criminate background staining.

Peripheral cytokine level
EDTA plasma concentrations of IL10, TNFα, and IP10 were measured by ELISA (Human Quantikine ELISA, R&D Systems, WiesbadenNordenstadt, Germany) at room temperature on a Fluostar Optima microplate reader (BMG Labtech, Offenburg, Germany). The sensitivity of the assays was 3.9 pg/ml for IL10, 0.7 pg/ml for TNFα, and 4.46 pg/ml for IP10.

statistical analysis
Mean values, their ±SEM and ranges, and normal distribution of data were calculated for each variable using the SPSS Software (SPSS 22.0, SPSS Inc., Chicago, IL, USA). Grubbs' test was used to identify outliers. The graphs were made using GraphPad Prism ® 6 (Version 6.01, GraphPad Software, Inc., La Jolla, CA, USA). Repeated measures (e.g., cytokine concentrations) were compared between LPS and placebo conditions using twoway repeated measure ANOVA with the repeated factors time and condition (i.e., LPS, placebo). To compare single measurement points separately between LPS and placebo, post hoc paired ttests were calculated with Bonferroni corrections for multiple testing, if not otherwise indicated. The pvalues <0.05 were considered to be statistically significant.

Demographic and clinical characteristics
Twenty healthy volunteers of Caucasian ethnicity with a mean age of 26.1 ± 0.9 years (range: 18-34) and a mean BMI of 24.2 ± 0.5 kg/m 2 (range: 19.3-26.9) were included in this cross over study. There was no difference with respect to the time to switch conditions comparing the two groups (LPS/placebo vs. placebo/LPS, 32 ± 11.4 days vs. 33 ± 8.4 days, p = 0.86). We did not observe differences between participants who received LPS before or after saline in any dependent variable (data not shown). Endotoxin injection induced a transient systemic inflamma tory response in all participants, which was characterized by a significant increase in WBC count. In addition, body temperature, heart rate, and respiratory rate increased after LPS administration (all p < 0.001, ANOVA interaction effects, Table 1).

cellular response to lPs injection
LPS application led to a significant increase in total neutrophil count peaking at 6 h postinjection (F = 92.51, p < 0.001; ANOVA interaction effect, Figure 1A), while monocytes (F = 14.16; p < 0.001; ANOVA interaction effect, data not shown) and lymphocytes (F = 108.8; p < 0.001; Figure 1A) significantly decreased. Maximum reduction of lymphocyte and monocyte counts occurred 4 or 2 h after LPS injection. Cell counts were nor malized after 24 h for lymphocytes and after 4 h for monocytes. The absolute numbers of circulating Tcells were diminished 3 h after LPS challenge and recovered at 24 h ( Figure 1B).
soluble systemic il-10 and iP-10 levels increase after lPs challenge Lowdose LPS injection induced significant changes in plasma concentrations of the proinflammatory cytokines interferon induced protein 10 and TNFα (IP10, F = 72.0, p < 0.001, Figure 2A; TNFα, F = 26.2, p < 0.001, both ANOVA interaction effects), as well as in the antiinflammatory cytokine IL10 (F = 21.5, p < 0.001; Figure 2B). The maximum IP10 concentration was reached 4 h after LPS administration, whereas the IL10 peak concentration occurred earlier at 2 h after LPS injection. IP10 and IL10 levels returned to baseline after 24 h (Figure 2).

DiscUssiOn
This study reveals new insights into the differential effects of endotoxemia on anti and proinflammatory Tcell responses. Overall, absolute Tcell numbers were decreased sharply after LPS challenge within 3 h after injection. Proinflammatory effec tor THs lost the capacity to produce IL17A, IL2, TNFα, and IFNγ as early as 3 h after LPS injection and regained function post endotoxemia. The relative number of antiinflammatory Tregs and its capacity to produce IL10 remained stable and did not change during endotoxemia. In contrast to murine experimental endotoxemia, humans are highly sensitive to LPS (2, 4). Thus, healthy participants challenged with LPS respond with a rise of body temperature and fever, elevated heart, and breathing rate partially leading to tachycardia/ tachypnea, decreased blood pressure, and neutrophilia (11,(22)(23)(24)(25). Human experimental endotoxemia is a wellinvestigated model for systemic inflammation (7-13, 26, 27). This model is advanta geous over patient studies, as it is not biased by pretreatment or comorbidities. The experimental human endotoxemia model has been used in low and high dose settings to address different immunological questions regarding innate and cellular immune res ponses during systemic inflammation (0.4-4.0 ng/kg body weight) (7,11,14,22,(24)(25)(26)(27).
In our cohort, LPS challenge in the lower dosage range caused a clinical and an immunological response of the subjects, which was absent under placebo conditions indicating the efficacy of our model. There was a differential effect on Tcell populations in our model. The capacity of effector Tcells to produce proinflam matory cytokines such as IFNγ, TNFα, IL2, or IL17A was diminished 3 h after LPS injection. Impaired IFNγ production by Tcells has been described after administration of higher doses In the placebo condition, T-cell subset counts were increased 3 h after placebo injection as compared to baseline, following the circadian rhythm (nadir of naïve CD4 + T-cells/naïve CD8 + cytotoxic T-cells at 11:00 a.m., peak of naïve CD4 + T-cells/CD8 + T-cells around 2:00 a.m.) Data are given as mean (±SEM). Two-way ANOVA analysis with repeated measures was performed followed by post hoc Bonferroni-corrected paired t-tests. *p < 0.05, **p < 0.01, ***p < 0.001, results of post hoc Bonferroni-corrected t-test. For results of ANOVA see text. 5

Brinkhoff et al. Suppression of T-Cells in Experimental Endotoxemia
Frontiers in Immunology | www.frontiersin.org May 2018 | Volume 9 | Article 1133 of LPS (2.0 and 4.0 ng/kg bodyweight); in line with our study, IFNγ production normalized after 24 h (13,22). Whereas in these studies, cytokine concentrations in supernatants were deter mined, we clearly demonstrated a lack of function and impair ment of the circulating CD4 + TH compartment to produce IFNγ on singlecell level (13,22). Van de Veerdonk et al. reported deficient Candidaspecific IL17A + Tcell responses in healthy subjects receiving a relatively high dose of LPS (2 ng/kg body weight) and in patients with gram negative sepsis (28). We found that even a lower dose of LPS led to a reduced capacity of effector Tcells to produce IL17A; however, this effect was transient and effector Tcells recovered 24 h after LPS challenge. We further studied potential causes for the dysfunction of the effector TH compartment. Increased numbers of or enhanced activity of antiinflammatory Treg may lead to potent inhibition of effector Tcells (18,19). LPS exposure may have a direct effect on Treg and is reported to promote Treg function (29,30). Interes tingly, neither the relative fraction nor the functional capacity of Treg to produce IL10 was altered after LPS administration in our study. Ronit et al. observed a relative increase of Treg, which returned to baseline 24 h after LPS injection (14). In addition, the authors reported a marked suppression of a broad range of cytokines, including IL10, when whole blood was stimulated with the Tcell mitogen phytohemagglutinin. These results conflict with our findings. We found IL10 production to be unaffected on single Tcell level and demonstrated that the relative fraction of Treg remained stable after LPS challenge. However, there are several methodological differences between the study of Ronit et al. and ours, which may explain the observed discrepancies. First, the dosage of LPS administered to healthy volunteers was much lower in our study. Second, we chose, in contrast to Ronit et al., a placebocontrolled crossover study design and followed a    (31,32). Alternatively, it has to be considered that proinflammatory Tcells were not rendered anergic but were simply depleted from the circulation by increased apoptosis, pooling in the spleen, or increased adhesion to vessel walls (4,33,34). In addition, Tcells may migrate to peripheral tissues such as lung or liver during systemic inflammation. Depletion or migration of Tcells would have prevented detection of proinflammatory Tcell subsets in our ex vivo assay.
The limitation of our study is that a single endotoxin chal lenge was administered; thus, we investigated in our model the immune response to an acute challenge of LPS which does not resemble sepsis. Accordingly, immunosuppression in our experi mental human endotoxemia model was transient and shortlived in contrast to persistent, longlasting immunoparalysis in sepsis (4,7,11,14). Kox et al. studied earlier the effect of repeated LPS challenges in healthy volunteers (35). A diminished in vivo response to LPS in terms of attenuated TNFα, IL6, and IL10 serum levels was observed upon rechallenge. The ex vivo Tcell response has not been assessed in these kinds of studies, and it will be of interest to know whether repeated LPS challenges have similar impact on Tcell function. Furthermore, immunoparalysis is usually observed days after onset of sepsis whereas transient immunosuppression in the human endotoxemia model is already detectable after hours. Therefore, the immunological response to a single endotoxin challenge in the human experimental endo toxemia model reflects a physiological mechanism to limit the immune response after an initial proinflammatory phase rather than a dysregulation of the immune system. Notwithstanding, the mechanisms inducing the immunosuppressive state are assumed to be similar and thus, this model is also used in interven tional trials to assess pharmacological interventions to abrogate immunoparalysis (7). Our findings that Th1 and Th17 cells are profoundly suppressed by endotoxemia whereas IL10 pro duction by Tcells remains unaffected, may add to the understanding dys regulated immunity sepsis. Indeed, a postmortem study revealed that production of Th1associated cytokines by splenic Tcells from septic patients was significantly reduced (6). In addition, diminished Th17 cell responses have been reported in patients with sepsis (28). Another study assessed Tcellderived IL10 production and found no difference between septic patients and controls, which is in line with our data (36). Altogether, this may suggest that restoration of proinflammatory immunity is more important than abrogation of antiinflammatory immunity as future therapeutic strategy to break immunoparalysis.
In conclusion, we could demonstrate that even a low dose of LPS induces potent suppression of proinflammatory Tcell subsets and does not affect the capacity of antiinflammatory Treg to produce IL10.

eThics sTaTeMenT
This study was carried out in accordance with the recommenda tions of the local ethics committee of the University Hospital Essen, Germany. The protocol was approved by the local ethics committee of the University Hospital Essen, Germany (156533 BO). All subjects gave written informed consent in accordance with the Declaration of Helsinki.
aUThOr cOnTriBUTiOns AB designed the study, performed the experiments, performed data analysis, and wrote the manuscript. AS and JK performed the experiments, performed data analysis, and wrote the manuscript. HE, SD, and SB performed data analysis and wrote the manu script. MS and AK designed the study and wrote the manuscript. OW and BW designed the study, performed data analysis, and wrote the manuscript.

acKnOWleDgMenTs
The authors would like to thank Alexandra Kornowski for techni cal assistance, Larissa Lueg and Eva Stemmler for data collection, and the staff of "Microbial safety" of the PaulEhrlichInstitute (Langen, Germany) for endotoxin and sterility testing. This study was funded by an internal research grant of the Medical Faculty of the University of DuisburgEssen (IFORES) program and a research grant of the Dr. Werner JackstädtStiftung (to BW). The funding organizations were not involved in study design, neither in collection, analyses, statistics and manuscript writing process, in the decision to submit the article, nor in the choice of journal for publication.

FUnDing
This study was funded by an internal research grant of the Medical Faculty of the University of DuisburgEssen (IFORES) program (to AB) and a research grant of the Dr. Werner JackstädtStiftung (to BW).