Online Laboratory Investigation
Septic Shock Sera Containing Circulating Histones Induce Dendritic Cell–Regulated Necrosis in Fatal Septic Shock Patients Loic Raffray, MD1,2; Isabelle Douchet, MS2; Jean-Francois Augusto, MD3; Jihad Youssef, MD1; Cecile Contin-Bordes, PhD2,4,5; Christophe Richez, MD, PhD2,5,6; Pierre Duffau, MD, PhD2,5,7; Marie-Elise Truchetet, MD, PhD2,5,6; Jean-Francois Moreau, MD, PhD2,4,5; Charles Cazanave, MD, PhD8; Lionel Leroux, MD, PhD9; Gaelle Mourrissoux, MD1; Fabrice Camou, MD1,5; Benjamin Clouzeau, MD, PhD10; Pascale Jeannin, MD, PhD3; Yves Delneste, PhD3; Claude Gabinski, MD1,5; Olivier Guisset, MD1,5; Estibaliz Lazaro, MD, PhD2,5,7; Patrick Blanco, MD, PhD1,2,4,5,7
Objectives: Innate immune system alterations, including dendritic cell loss, have been reproducibly observed in patients with septic shock and correlated to adverse outcomes or nosocomial infections. The goal of this study is to better understand the mecha-
Medical Intensive Care Unit, Saint Andre Hospital, University Hospital of Bordeaux, Bordeaux, France. 2 CNRS UMR 5164 CIRID, University of Bordeaux, Bordeaux, France. 3 INSERM U892-CRCNA, University of Angers, Angers, France. 4 Laboratory of Immunology and Immunogenetics, University Hospital of Bordeaux, Bordeaux, France. 5 GECMIA (Groupe Epidemiologie Clinique des Maladies Inflammatoires d’Aquitaine) Study Group, University Hospital of Bordeaux, Bordeaux, France. 6 Department of Rheumatology, University Hospital of Bordeaux, Bordeaux, France. 7 Department of Internal Medicine, University Hospital of Bordeaux, Bordeaux, France. 8 Department of Infectious and Tropical Diseases, University Hospital of Bordeaux, Bordeaux, France. 9 Department of Interventional Cardiology, University Hospital of Bordeaux, Bordeaux, France. 10 Medical Intensive Care Unit, Pellegrin Hospital, University Hospital of Bordeaux, Bordeaux, France. This work was performed at CNRS UMR 5164 CIRID, University of Bordeaux, Bordeaux, France. Drs. Raffray, Douchet, Lazaro, and Blanco contributed equally to this work. Supported, in part, by Centre National pour la Recherche Scientifique. Dr. Raffray received support for travel from Roche (expenses related to medical meeting in 2013) and Alexion (expenses related to medical meeting in 2012). Dr. Moreau disclosed government work. Dr. Camou’s institution received grant support from Novartis. The remaining authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved. DOI: 10.1097/CCM.0000000000000879 1
Critical Care Medicine
nisms behind this observation in order to better assess septic shock pathogenesis. Design: Prospective, controlled experimental study. Setting: Research laboratory at an academic medical center. Subjects: The study enrolled 71 patients, 49 with septic shock and 22 with cardiogenic shock. Seventeen healthy controls served as reference. In vitro monocyte-derived dendritic cells were generated from healthy volunteers. Interventions: Sera were assessed for their ability to promote in vitro dendritic cell death through flow cytometry detection in each group of patients. The percentage of apoptotic or necrotic dendritic cells was evaluated by annexin-V and propidium iodide staining. Measurements and Main Results: We observed that only patients with septic shock and not patients with pure cardiogenic shock were characterized by a rapid and profound loss of circulating dendritic cells. In vitro analysis revealed that sera from patients with septic shock induced higher dendritic cell death compared to normal sera or cardiogenic shock (p < 0.005). Sera from surviving patients induced dendritic cell death through a caspase-dependent apoptotic pathway, whereas sera from nonsurviving patients induced dendritic cell–regulated necrosis. Dendritic cell necrosis was not due to necroptosis but was dependent of the presence of circulating histone. The toxicity of histones toward dendritic cell could be prevented by recombinant human activated protein C. Finally, we observed a direct correlation between the levels of circulating histones in patients and the ability of the sera to promote dendritic cell–regulated necrosis. Conclusions: The study demonstrates a differential mechanism of dendritic cell death in patients with septic shock that is dependent on the severity of the disease. (Crit Care Med 2015; 43:e107–e116) Key Words: dendritic cell; histone; human; immunology; immunosuppression; septic shock
www.ccmjournal.org
e107
Raffray et al
S
epsis and septic shock (SS) lead to considerable morbidity and mortality in developed and developing countries (1). Despite advances in understanding the events that lead to SS, molecular therapies have failed to improve sepsis mortality (2, 3). Death of patients with SS usually does not result from the initial septic event but rather from subsequent nosocomial infections, and patients who survive severe sepsis often display severely compromised immune function, including innate immune function (4). Dendritic cells (DCs) are a key player in the innate arm of the immune system. DCs are specialized sentinel cells that bridge the innate and adaptive immune systems; they recognize pathogens using pattern recognition receptors, including Toll-like receptors (TLRs), and then migrate to T cell areas of lymphoid organs to present pathogen-derived antigens to antigen-specific T cells. Activated DCs up-regulate costimulatory molecules and produce cytokines that drive T cell priming and effector differentiation and activate various types of immune cells. DCs are a heterogeneous population of antigen-presenting cells (APCs) that comprised of two major classes: plasmacytoid DCs (pDCs) and conventional DCs (cDCs). The pDCs rapidly produce type 1 interferon following activation through nucleic acid–sensing TLRs, such as TLR7 and TLR9. cDCs are dedicated APCs that have a characteristic dendritic morphology and express high levels of major histocompatibility complex class II molecules. Altered functions of these cells have been implicated in the pathogenesis of cancer, autoimmune diseases, allergy, and infection (5–7). The involvement of DCs in SS pathogenesis has been assessed in both mice and humans. In mouse models of SS, several studies have shown that DC numbers are decreased in the blood as well as the tissue level (8–14). Apoptosis has been proposed to be the main mechanism responsible for these observations (15). Indeed, blocking apoptosis restores DC numbers and significantly improves survival in different mouse models of sepsis (16, 17) However, a recent comparative transcriptome analysis of a classical mouse model and human sepsis indicates very poor correlation between the pathophysiology of SS in mice and humans, emphasizing the need for human studies (18). In humans, several studies have described an early and profound loss of circulating cDCs and pDCs (10, 12, 14, 19). Interestingly, the decrease has been shown to be associated with decreased survival or nosocomial infections (10, 20). Although the mechanisms involved in this observation in humans are of great interest, they remain uncertain because only observational studies have been published to date. This study is aimed at identifying the mechanisms involved in DC loss in sepsis. We hypothesized that soluble factors present in patient sera could interfere with DC biology and/or survival.
PATIENTS AND METHODS Patients The study was conducted in the University hospital of Bordeaux, France, between June 2008 and December 2011. Seventy-one adult patients were enrolled included in this study and classified according to their diagnoses as follows: 49 SS and e108
www.ccmjournal.org
22 cardiogenic shock (CS). Seventeen healthy donors (HD) were recruited included in the study as a control group. The study was approved by the Local Ethics Committee for Scientific Research Conduct. As it was purely observational and did not require any additional blood sampling as compared to routine care, consent was waived for patients. SS was defined according to consensus definitions as sepsis associated with organ dysfunction and hypoperfusion or hypotension even after appropriate volume resuscitation (21). CS was characterized by a decreased cardiac output and evidence of tissue hypoxia in the presence of adequate intravascular volume. Severity was assessed by the Sequential Organ Failure Assessment (SOFA). Patients were investigated within 24 hours of admission and followed until remission or death. Mortality was defined as death occurring within 28 days after diagnosis. When possible, sera were collected at the beginning of the disease (day 0 in ICU, d0) and at the end of the disease (last blood drawn within the ICU, day final, df). Exclusion criteria were as follows: age less than 18 years, pregnancy, hematologic malignancy, HIV infection, and immunosuppressive treatment, including daily corticosteroids therapy greater than 0.5 mg/kg prednisone equivalent. Flow Cytometric Analysis of Circulating DCs The size of each DC subset was estimated by flow cytometry (BD Biosciences, Pont-de-Claix, France). Samples were analyzed on a four-color FACSCanto II apparatus and 106 WBCs were acquired. All lineage positive cells (CD3, CD8, CD19, CD14, CD16, and CD56) were gated out to allow for the gating of Lin– HLA-DR+ cells. Among these, CD11c+ CD123– (cDC) or CD11c– CD123+ (pDC) were selected. Absolute numbers of blood DC precursor subsets were expressed per milliliter of peripheral blood. Plasma Blood Mononuclear Cell Plasma blood mononuclear cells (PBMCs) used for culture experiments were obtained using Ficoll gradient centrifugation from apheresis blood samples. PBMCs were cultured in six-well plates (107 cells/well) overnight in RPMI1640 (Gibco, Life Technologies, Carlsbad, CA) supplemented with 8% fetal calf serum and with 10% sera from each patient or HD and with an increasing concentration of histones (Roche, Boulogne-Billancourt, France). Following culture, the culture medium was split into two glass tubes: one for detection of the percentage of lymphocyte death and the other for cDC and pDC counting. The bottom of the well was scraped to recover the monocyte population. After centrifugation, cells were labeled with anti CD3 and DR or anti CD11C CD123 and DR or with anti CD14 (Beckman Coulter, Villepinte, France) and cell death was determined by eosin dye exclusion, annexin-V fluorescein isothiocyanate, and propidium iodide (PI) staining (Apoptosis kit, BD Biosciences). Data were collected and analyzed using a BD FACSCanto II instrument with DIVA software (BD Biosciences). The percentage of death was determined by summing all populations except the double-negative population. April 2015 • Volume 43 • Number 4
Online Laboratory Investigation
Monocyte-Derived DC Generation and Culture Monocyte-derived DC (MDDC) preparation was performed as described previously (22). Briefly, CD14+ monocytes were purified from blood by depletion using monoclonal antibodies and Dynabeads (Life Technologies) and were cultured in six-well plates (1 × 106/well) for 4 days with recombinant human granulocyte-macrophage colony stimulating factor (GM-CSF) (100 ng/mL) and recombinant human interleukin (IL)-4 (20 ng/mL). These cells are refered thereafter as GM-IL4 monocytes. On day 4, cells were stained for flow cytometry. The purity and viability of DC preparations was assessed by flow cytometry and was above 95%. Recombinant human GM-CSF and recombinant human IL-4 were purchased from Miltenyi Biotec (Paris, France). Incubation With Sera After 4 days of culture, 200 μL of sera (10% vol/vol) was added to the MDDC and incubated overnight in different conditions:
Z-Val-Ala-Asp-fluoromethylketone (Coger, Paris, France) at 5 μM and 10 μM final concentration, human serum albumin (Vialebex, LFB biomedicament, Lille, France) at 200 ng/mL, recombinant form of human activated protein C (rh-aPC, drotrecogin alfa, Lilly, Indianapolis, IN) at 200 ng/mL, necrostatin at 10 μM, histones (Roche Diagnostics) at 200 ng/mL, and anti-histones (anti-H3 and anti-H4) at 1 μg/mL (Santacruz Biotechnologies, Dallas, TX). For each condition, GM-IL4 monocytes were cultured and analyzed in the same way and were used as our internal control. At the end of the time course, the culture medium containing MDDC was centrifuged at 1500 rpm for 5 minutes. MDDC were labeled with anti CD1a (Beckman Coulter) and cell death was determined by staining the cells with annexinV FITC and PI markers (Apoptosis kit, BD Biosciences). Data were collected and analyzed using a BD FACSCanto
Table 1. Clinical Characteristics at Baseline of Septic Shock Patients, Cardiogenic Shock Patients, and Healthy Controls at Admission Characteristics
No. of individuals Age (yr) Sex ratio male/female (% male) Primary injury (%)
p (CS vs SS)
Healthy Donors
CS
SS
17
22
49
30 (25.7–50)
55 (43–74)
65 (56.8–78)
0.045
6/11 (35)
13/9 (59)
28/21 (57)
NS
NA
Myocardial infarction (18)
Pneumonia (43)
Cardiac arrest (18)
Urinary tract sepsis (14)
Drug poisoning (22)
Abdominal sepsis (14)
Others (42)
Others (29)
Sequential Organ Failure Assessment admission
NA
14 (12–19)
13 (9.8–15.3)
NS
Length of ICU stay, median (d, range)
NA
4.5 (2–12)
8 (4–13.3)
0.04
Length of hospital stay (d, range)
NA
7 (2–19)
18 (5–28.3)
0.018
Death in ICU (%)
NA
11 (50)
19 (38.8)
NS
Death at day 28 (%)
NA
14 (64)
20 (40.4)
NS
Delay between admission and blood sampling (hr, range)
NA
4 (3–5)
2 (0–3)
NS
Vasopressor at admission (%)
NA
9 (41) including 5 epinephrine and 4 dobutamine
13 (26.5) with norepinephrine
NS
Vasopressor therapy at blood sampling (%)
NA
22 (100) including 16 epinephrine and 6 dobutamine
41 (83.6) including 39 with norepinephrine and 2 epinephrine
NS
Mechanical ventilation at blood sampling (%)
NA
5 (23)
17 (35)
NS
Mechanical ventilation during ICU stay (%)
NA
5 (23)
20 (41)
NS
Lactates (mmol/L)
NA
8.1 (2.2–10.3) (n = 18)
3 (1.9–5.5) (n = 49)
NS
C-reactive protein (mg/dL)
NA
10 (4.6–54)
177 (114.2–240.5)
< 0.0001
CS = cardiogenic shock, SS = septic shock, NS = not significant, NA = not applicable. Variables are expressed as median (interquartile range). Comparisons between SS and CS patients are done using nonparametric unpaired tests.
Critical Care Medicine
www.ccmjournal.org
e109
Raffray et al
II instrument with DIVA software (BD Biosciences). The percentage of death was determined by summing all populations except the double-negative population. The percentage of apoptosis was determined by the annexin-Vpositive population and the percentage of necrosis by the PI + Annexin– population. Caspase induction was evaluated through detection of caspase 3 activation and cleavage of PARP (poly ADP ribose polymerase) by using a commercially available flow cytometry assay (BD Biosciences). Nucleosome Quantification Nucleosome serum concentration was determined by enzymelinked immunosorbent assay (ELISA) using a commercial available kit (Cell death detection ELISAPLUS, Roche, Mannheim, Germany). In brief, the test is based on the “sandwich” ELISA principle using two mouse antibodies directed against histone and DNA. The results are expressed in arbitrary unit in reference to a positive control (100%). Nucleosome concentration in sera was determined in duplicate. Results were reproducible (< 5% variation) intra- and interassays. Statistical Analysis Data are expressed as numbers and percentages for qualitative variables. As quantitative variables do not follow a Gaussian distribution, medians and interquartile ranges are used.
Statistical significance of difference between groups are determined by nonparametric tests, correlation studies were determined with the Spearman correlation test, and statistical significance of differences of paired data were determined by paired t test. All statistical analyses were performed using GraphPad Prism 6.0 software (GraphPad Software, San Diego, CA).
RESULTS Early and Profound Circulating DC Loss Is Specific to Patients With SS Seventy-one patients were enrolled in this study. Patient characteristics are shown in Table 1. Patients with SS and CS were comparable in terms of sex ratio, SOFA score, and mortality. SS patients were older and had a longer length of ICU stay and overall hospital stay. Thirty-four patients died, 20 with SS and 14 with CS. As shown in Figure 1, SS patients had significantly lower cDC and pDC counts when compared with HD or patients with CS (median value of cDC, 3.4 × 103/mL vs 12.7 × 103/mL vs 10.7 × 103/mL, respectively; median values of pDC, 0.8 × 103/mL vs 7 × 103/mL vs 4.8 × 103/mL, respectively). Cell surface expression of HLA-DR on both DC subsets was lower in the SS population compared with patients with CS. Thus, we confirmed in this new cohort of patients that circulating DC numbers are significantly decreased in patients with SS.
Figure 1. Comparison of circulating conventional dendritic cells (cDCs), plasmacytoid dendritic cells (pDCs), T lymphocyte (TL), and monocyte counts in patients with septic shock (SS), cardiogenic shock (CS), and healthy donors (HD) (HD, circles, n = 17; CS, squares, n = 22; SS, rhombuses, n = 49). Comparisons are done using nonparametric unpaired tests between each group of patients (*p < 0.0001). Horizontal lines indicate medians. ns = not significant.
e110
www.ccmjournal.org
Sera From Patients With SS But Not With CS Induce DC Death To determine whether reduced viability accounted for the quantitative changes observed in the DC population, we first assessed cell death of normal PBMCs cultured in vitro overnight with 10% SS sera, CS sera, or HD sera. Cell death was evaluated by cell counting (eosin staining) as well as annexin-V and PI staining. Cell death was significantly increased in PBMCs cultured in the presence of SS sera (gray symbols), mainly on the BDCA2+ HLA-DR+ (pDCs), CD11c+HLADR+CD14– (cDCs), and CD11c+HLADR+CD14+ (monocytes) populations compared to cells cultured with HD sera (white symbols). By contrast, the rate of cell death was equivalent in the different lymphocyte subsets (Fig. 2A and data not shown), independently of the sera used. To confirm that serum from patients with SS was toxic to DCs, we cultured purified normal monocytes with GM-CSF and IL-4 for 4 days. The DCs were then exposed to different types of sera overnight, at different concentrations and assessed for cell death (Fig. 2B). We observed that SS sera (rhombuses) induced a higher DC death rate compared with HD sera (circles) in a dose-dependent manner. When compared with HD or CS sera, sera from patients with SS reproducibly induced a significantly higher DC death April 2015 • Volume 43 • Number 4
Online Laboratory Investigation
rate as assessed by annexin-V/PI staining (Fig. 2C) (median percentage of death 17.3, 15.5, and 33.8, respectively, with HD sera, CS sera, and SS sera). The eosin dye exclusion test confirmed the flow cytometry results by showing more eosinpositive cells in the MDDC cultured in the presence of SS sera (median, 80.4 × 103 cells/mL) compared to culture with HD (40 × 103 cells/ mL) or CS sera (50 × 103 cells/ mL) (Fig. 2D). Because several groups have observed that DC loss positively
correlates to a more severe outcome (10, 20), sera were split into two groups based on the clinical outcome of the patient: surviving (surv SS, n = 29) versus nonsurviving (non-surv SS, n = 20) patients. We observed that the percentages of serainduced DC death were equivalent in the two groups (Fig. 2E). Taken as a whole, those data suggest that sera from patients with SS are characterized by the presence of a soluble factor that induces monocyte and DC cell death.
Figure 2. Septic shock sera are toxic for circulating and monocyte-derived dendritic cells (MDDCs). A, Normal plasma blood mononuclear cells were cultured overnight with 10% normal serum (empty symbols, n = 6) or serum from septic shock (SS) patients (filled symbols, n = 9) drawn at any time point over the course of the disease. Cell death was evaluated by flow cytometry using annexin-V/propidium iodide (PI) staining. Results of individual cultures and the medians are shown in each cell population: T lymphocytes, plasmacytoid dendritic cells (pDCs), conventional dendritic cells (cDCs), and monocytes. B, MDDCs (see Patients and Methods section) were cultured overnight with different concentrations of serum from healthy donors (HD) (circles) or SS patients (rhombuses). Cell death was evaluated by flow cytometry by using annexin-V/PI staining. The results are reported as the mean of three independent experiments ± sd. C, MDDCs were cultured overnight with 10% serum from HD (circles, n = 8), cardiogenic shock (CS) patients (squares, n = 5), or SS patients (rhombuses, n = 56). Cell death was evaluated by flow cytometry using annexin-V/PI staining. D, MDDCs were cultured overnight with 10% serum from nine HD (circles), eight CS (squares) patients, or 24 SS (rhombuses) patients. Eosin dye was used to evaluate the number of dead cells. E, Sera from survival (surv SS, triangles, n = 36) or nonsurvival (non-surv SS, inverted triangles, n = 20) SS patients induce similar levels of cell death on MDDCs (ns = not significant, *p < 0.05, **p < 0.005, nonparametric unpaired tests; horizontal lines indicate medians).
Sera From Nonsurviving Patients Induce Primary Regulated Necrosis, Whereas Sera From Surviving Patients Induce Apoptosis Although the overall cell death rate was not different between MDDC incubated with sera from surv SS patients as compared to non-surv SS patients, we were intrigued by the remarkably different morphology of the cells when examined by optical microscopy. Cells cultured in the presence of sera from non-surv SS patients, but not those cultured in sera from surv SS patients, were characterized by the presence of cellular debris and clumping of dead cells (data not shown), suggesting that different mechanisms could be responsible for sera-induced DC death. Interestingly, flow cytometry analysis revealed that sera from surv SS patients induced early apoptosis (annexin-V+/ PI– population) and secondary necrosis (annexin-V+/PI+ population), whereas sera from non-surv SS patients induced mainly primary necrosis (annexin-V–/PI+ population) (Fig. 3A). This observation was confirmed in 20 sera from non-surv SS patients compared to 36 sera from surv SS patients (median percentage of necrosis, 11.7% vs 1.6%; p
Septic Shock Sera Containing Circulating Histones Induce Dendritic Cell–Regulated Necrosis in Fatal Septic Shock Patients Loic Raffray, MD1,2; Isabelle Douchet, MS2; Jean-Francois Augusto, MD3; Jihad Youssef, MD1; Cecile Contin-Bordes, PhD2,4,5; Christophe Richez, MD, PhD2,5,6; Pierre Duffau, MD, PhD2,5,7; Marie-Elise Truchetet, MD, PhD2,5,6; Jean-Francois Moreau, MD, PhD2,4,5; Charles Cazanave, MD, PhD8; Lionel Leroux, MD, PhD9; Gaelle Mourrissoux, MD1; Fabrice Camou, MD1,5; Benjamin Clouzeau, MD, PhD10; Pascale Jeannin, MD, PhD3; Yves Delneste, PhD3; Claude Gabinski, MD1,5; Olivier Guisset, MD1,5; Estibaliz Lazaro, MD, PhD2,5,7; Patrick Blanco, MD, PhD1,2,4,5,7
Objectives: Innate immune system alterations, including dendritic cell loss, have been reproducibly observed in patients with septic shock and correlated to adverse outcomes or nosocomial infections. The goal of this study is to better understand the mecha-
Medical Intensive Care Unit, Saint Andre Hospital, University Hospital of Bordeaux, Bordeaux, France. 2 CNRS UMR 5164 CIRID, University of Bordeaux, Bordeaux, France. 3 INSERM U892-CRCNA, University of Angers, Angers, France. 4 Laboratory of Immunology and Immunogenetics, University Hospital of Bordeaux, Bordeaux, France. 5 GECMIA (Groupe Epidemiologie Clinique des Maladies Inflammatoires d’Aquitaine) Study Group, University Hospital of Bordeaux, Bordeaux, France. 6 Department of Rheumatology, University Hospital of Bordeaux, Bordeaux, France. 7 Department of Internal Medicine, University Hospital of Bordeaux, Bordeaux, France. 8 Department of Infectious and Tropical Diseases, University Hospital of Bordeaux, Bordeaux, France. 9 Department of Interventional Cardiology, University Hospital of Bordeaux, Bordeaux, France. 10 Medical Intensive Care Unit, Pellegrin Hospital, University Hospital of Bordeaux, Bordeaux, France. This work was performed at CNRS UMR 5164 CIRID, University of Bordeaux, Bordeaux, France. Drs. Raffray, Douchet, Lazaro, and Blanco contributed equally to this work. Supported, in part, by Centre National pour la Recherche Scientifique. Dr. Raffray received support for travel from Roche (expenses related to medical meeting in 2013) and Alexion (expenses related to medical meeting in 2012). Dr. Moreau disclosed government work. Dr. Camou’s institution received grant support from Novartis. The remaining authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved. DOI: 10.1097/CCM.0000000000000879 1
Critical Care Medicine
nisms behind this observation in order to better assess septic shock pathogenesis. Design: Prospective, controlled experimental study. Setting: Research laboratory at an academic medical center. Subjects: The study enrolled 71 patients, 49 with septic shock and 22 with cardiogenic shock. Seventeen healthy controls served as reference. In vitro monocyte-derived dendritic cells were generated from healthy volunteers. Interventions: Sera were assessed for their ability to promote in vitro dendritic cell death through flow cytometry detection in each group of patients. The percentage of apoptotic or necrotic dendritic cells was evaluated by annexin-V and propidium iodide staining. Measurements and Main Results: We observed that only patients with septic shock and not patients with pure cardiogenic shock were characterized by a rapid and profound loss of circulating dendritic cells. In vitro analysis revealed that sera from patients with septic shock induced higher dendritic cell death compared to normal sera or cardiogenic shock (p < 0.005). Sera from surviving patients induced dendritic cell death through a caspase-dependent apoptotic pathway, whereas sera from nonsurviving patients induced dendritic cell–regulated necrosis. Dendritic cell necrosis was not due to necroptosis but was dependent of the presence of circulating histone. The toxicity of histones toward dendritic cell could be prevented by recombinant human activated protein C. Finally, we observed a direct correlation between the levels of circulating histones in patients and the ability of the sera to promote dendritic cell–regulated necrosis. Conclusions: The study demonstrates a differential mechanism of dendritic cell death in patients with septic shock that is dependent on the severity of the disease. (Crit Care Med 2015; 43:e107–e116) Key Words: dendritic cell; histone; human; immunology; immunosuppression; septic shock
www.ccmjournal.org
e107
Raffray et al
S
epsis and septic shock (SS) lead to considerable morbidity and mortality in developed and developing countries (1). Despite advances in understanding the events that lead to SS, molecular therapies have failed to improve sepsis mortality (2, 3). Death of patients with SS usually does not result from the initial septic event but rather from subsequent nosocomial infections, and patients who survive severe sepsis often display severely compromised immune function, including innate immune function (4). Dendritic cells (DCs) are a key player in the innate arm of the immune system. DCs are specialized sentinel cells that bridge the innate and adaptive immune systems; they recognize pathogens using pattern recognition receptors, including Toll-like receptors (TLRs), and then migrate to T cell areas of lymphoid organs to present pathogen-derived antigens to antigen-specific T cells. Activated DCs up-regulate costimulatory molecules and produce cytokines that drive T cell priming and effector differentiation and activate various types of immune cells. DCs are a heterogeneous population of antigen-presenting cells (APCs) that comprised of two major classes: plasmacytoid DCs (pDCs) and conventional DCs (cDCs). The pDCs rapidly produce type 1 interferon following activation through nucleic acid–sensing TLRs, such as TLR7 and TLR9. cDCs are dedicated APCs that have a characteristic dendritic morphology and express high levels of major histocompatibility complex class II molecules. Altered functions of these cells have been implicated in the pathogenesis of cancer, autoimmune diseases, allergy, and infection (5–7). The involvement of DCs in SS pathogenesis has been assessed in both mice and humans. In mouse models of SS, several studies have shown that DC numbers are decreased in the blood as well as the tissue level (8–14). Apoptosis has been proposed to be the main mechanism responsible for these observations (15). Indeed, blocking apoptosis restores DC numbers and significantly improves survival in different mouse models of sepsis (16, 17) However, a recent comparative transcriptome analysis of a classical mouse model and human sepsis indicates very poor correlation between the pathophysiology of SS in mice and humans, emphasizing the need for human studies (18). In humans, several studies have described an early and profound loss of circulating cDCs and pDCs (10, 12, 14, 19). Interestingly, the decrease has been shown to be associated with decreased survival or nosocomial infections (10, 20). Although the mechanisms involved in this observation in humans are of great interest, they remain uncertain because only observational studies have been published to date. This study is aimed at identifying the mechanisms involved in DC loss in sepsis. We hypothesized that soluble factors present in patient sera could interfere with DC biology and/or survival.
PATIENTS AND METHODS Patients The study was conducted in the University hospital of Bordeaux, France, between June 2008 and December 2011. Seventy-one adult patients were enrolled included in this study and classified according to their diagnoses as follows: 49 SS and e108
www.ccmjournal.org
22 cardiogenic shock (CS). Seventeen healthy donors (HD) were recruited included in the study as a control group. The study was approved by the Local Ethics Committee for Scientific Research Conduct. As it was purely observational and did not require any additional blood sampling as compared to routine care, consent was waived for patients. SS was defined according to consensus definitions as sepsis associated with organ dysfunction and hypoperfusion or hypotension even after appropriate volume resuscitation (21). CS was characterized by a decreased cardiac output and evidence of tissue hypoxia in the presence of adequate intravascular volume. Severity was assessed by the Sequential Organ Failure Assessment (SOFA). Patients were investigated within 24 hours of admission and followed until remission or death. Mortality was defined as death occurring within 28 days after diagnosis. When possible, sera were collected at the beginning of the disease (day 0 in ICU, d0) and at the end of the disease (last blood drawn within the ICU, day final, df). Exclusion criteria were as follows: age less than 18 years, pregnancy, hematologic malignancy, HIV infection, and immunosuppressive treatment, including daily corticosteroids therapy greater than 0.5 mg/kg prednisone equivalent. Flow Cytometric Analysis of Circulating DCs The size of each DC subset was estimated by flow cytometry (BD Biosciences, Pont-de-Claix, France). Samples were analyzed on a four-color FACSCanto II apparatus and 106 WBCs were acquired. All lineage positive cells (CD3, CD8, CD19, CD14, CD16, and CD56) were gated out to allow for the gating of Lin– HLA-DR+ cells. Among these, CD11c+ CD123– (cDC) or CD11c– CD123+ (pDC) were selected. Absolute numbers of blood DC precursor subsets were expressed per milliliter of peripheral blood. Plasma Blood Mononuclear Cell Plasma blood mononuclear cells (PBMCs) used for culture experiments were obtained using Ficoll gradient centrifugation from apheresis blood samples. PBMCs were cultured in six-well plates (107 cells/well) overnight in RPMI1640 (Gibco, Life Technologies, Carlsbad, CA) supplemented with 8% fetal calf serum and with 10% sera from each patient or HD and with an increasing concentration of histones (Roche, Boulogne-Billancourt, France). Following culture, the culture medium was split into two glass tubes: one for detection of the percentage of lymphocyte death and the other for cDC and pDC counting. The bottom of the well was scraped to recover the monocyte population. After centrifugation, cells were labeled with anti CD3 and DR or anti CD11C CD123 and DR or with anti CD14 (Beckman Coulter, Villepinte, France) and cell death was determined by eosin dye exclusion, annexin-V fluorescein isothiocyanate, and propidium iodide (PI) staining (Apoptosis kit, BD Biosciences). Data were collected and analyzed using a BD FACSCanto II instrument with DIVA software (BD Biosciences). The percentage of death was determined by summing all populations except the double-negative population. April 2015 • Volume 43 • Number 4
Online Laboratory Investigation
Monocyte-Derived DC Generation and Culture Monocyte-derived DC (MDDC) preparation was performed as described previously (22). Briefly, CD14+ monocytes were purified from blood by depletion using monoclonal antibodies and Dynabeads (Life Technologies) and were cultured in six-well plates (1 × 106/well) for 4 days with recombinant human granulocyte-macrophage colony stimulating factor (GM-CSF) (100 ng/mL) and recombinant human interleukin (IL)-4 (20 ng/mL). These cells are refered thereafter as GM-IL4 monocytes. On day 4, cells were stained for flow cytometry. The purity and viability of DC preparations was assessed by flow cytometry and was above 95%. Recombinant human GM-CSF and recombinant human IL-4 were purchased from Miltenyi Biotec (Paris, France). Incubation With Sera After 4 days of culture, 200 μL of sera (10% vol/vol) was added to the MDDC and incubated overnight in different conditions:
Z-Val-Ala-Asp-fluoromethylketone (Coger, Paris, France) at 5 μM and 10 μM final concentration, human serum albumin (Vialebex, LFB biomedicament, Lille, France) at 200 ng/mL, recombinant form of human activated protein C (rh-aPC, drotrecogin alfa, Lilly, Indianapolis, IN) at 200 ng/mL, necrostatin at 10 μM, histones (Roche Diagnostics) at 200 ng/mL, and anti-histones (anti-H3 and anti-H4) at 1 μg/mL (Santacruz Biotechnologies, Dallas, TX). For each condition, GM-IL4 monocytes were cultured and analyzed in the same way and were used as our internal control. At the end of the time course, the culture medium containing MDDC was centrifuged at 1500 rpm for 5 minutes. MDDC were labeled with anti CD1a (Beckman Coulter) and cell death was determined by staining the cells with annexinV FITC and PI markers (Apoptosis kit, BD Biosciences). Data were collected and analyzed using a BD FACSCanto
Table 1. Clinical Characteristics at Baseline of Septic Shock Patients, Cardiogenic Shock Patients, and Healthy Controls at Admission Characteristics
No. of individuals Age (yr) Sex ratio male/female (% male) Primary injury (%)
p (CS vs SS)
Healthy Donors
CS
SS
17
22
49
30 (25.7–50)
55 (43–74)
65 (56.8–78)
0.045
6/11 (35)
13/9 (59)
28/21 (57)
NS
NA
Myocardial infarction (18)
Pneumonia (43)
Cardiac arrest (18)
Urinary tract sepsis (14)
Drug poisoning (22)
Abdominal sepsis (14)
Others (42)
Others (29)
Sequential Organ Failure Assessment admission
NA
14 (12–19)
13 (9.8–15.3)
NS
Length of ICU stay, median (d, range)
NA
4.5 (2–12)
8 (4–13.3)
0.04
Length of hospital stay (d, range)
NA
7 (2–19)
18 (5–28.3)
0.018
Death in ICU (%)
NA
11 (50)
19 (38.8)
NS
Death at day 28 (%)
NA
14 (64)
20 (40.4)
NS
Delay between admission and blood sampling (hr, range)
NA
4 (3–5)
2 (0–3)
NS
Vasopressor at admission (%)
NA
9 (41) including 5 epinephrine and 4 dobutamine
13 (26.5) with norepinephrine
NS
Vasopressor therapy at blood sampling (%)
NA
22 (100) including 16 epinephrine and 6 dobutamine
41 (83.6) including 39 with norepinephrine and 2 epinephrine
NS
Mechanical ventilation at blood sampling (%)
NA
5 (23)
17 (35)
NS
Mechanical ventilation during ICU stay (%)
NA
5 (23)
20 (41)
NS
Lactates (mmol/L)
NA
8.1 (2.2–10.3) (n = 18)
3 (1.9–5.5) (n = 49)
NS
C-reactive protein (mg/dL)
NA
10 (4.6–54)
177 (114.2–240.5)
< 0.0001
CS = cardiogenic shock, SS = septic shock, NS = not significant, NA = not applicable. Variables are expressed as median (interquartile range). Comparisons between SS and CS patients are done using nonparametric unpaired tests.
Critical Care Medicine
www.ccmjournal.org
e109
Raffray et al
II instrument with DIVA software (BD Biosciences). The percentage of death was determined by summing all populations except the double-negative population. The percentage of apoptosis was determined by the annexin-Vpositive population and the percentage of necrosis by the PI + Annexin– population. Caspase induction was evaluated through detection of caspase 3 activation and cleavage of PARP (poly ADP ribose polymerase) by using a commercially available flow cytometry assay (BD Biosciences). Nucleosome Quantification Nucleosome serum concentration was determined by enzymelinked immunosorbent assay (ELISA) using a commercial available kit (Cell death detection ELISAPLUS, Roche, Mannheim, Germany). In brief, the test is based on the “sandwich” ELISA principle using two mouse antibodies directed against histone and DNA. The results are expressed in arbitrary unit in reference to a positive control (100%). Nucleosome concentration in sera was determined in duplicate. Results were reproducible (< 5% variation) intra- and interassays. Statistical Analysis Data are expressed as numbers and percentages for qualitative variables. As quantitative variables do not follow a Gaussian distribution, medians and interquartile ranges are used.
Statistical significance of difference between groups are determined by nonparametric tests, correlation studies were determined with the Spearman correlation test, and statistical significance of differences of paired data were determined by paired t test. All statistical analyses were performed using GraphPad Prism 6.0 software (GraphPad Software, San Diego, CA).
RESULTS Early and Profound Circulating DC Loss Is Specific to Patients With SS Seventy-one patients were enrolled in this study. Patient characteristics are shown in Table 1. Patients with SS and CS were comparable in terms of sex ratio, SOFA score, and mortality. SS patients were older and had a longer length of ICU stay and overall hospital stay. Thirty-four patients died, 20 with SS and 14 with CS. As shown in Figure 1, SS patients had significantly lower cDC and pDC counts when compared with HD or patients with CS (median value of cDC, 3.4 × 103/mL vs 12.7 × 103/mL vs 10.7 × 103/mL, respectively; median values of pDC, 0.8 × 103/mL vs 7 × 103/mL vs 4.8 × 103/mL, respectively). Cell surface expression of HLA-DR on both DC subsets was lower in the SS population compared with patients with CS. Thus, we confirmed in this new cohort of patients that circulating DC numbers are significantly decreased in patients with SS.
Figure 1. Comparison of circulating conventional dendritic cells (cDCs), plasmacytoid dendritic cells (pDCs), T lymphocyte (TL), and monocyte counts in patients with septic shock (SS), cardiogenic shock (CS), and healthy donors (HD) (HD, circles, n = 17; CS, squares, n = 22; SS, rhombuses, n = 49). Comparisons are done using nonparametric unpaired tests between each group of patients (*p < 0.0001). Horizontal lines indicate medians. ns = not significant.
e110
www.ccmjournal.org
Sera From Patients With SS But Not With CS Induce DC Death To determine whether reduced viability accounted for the quantitative changes observed in the DC population, we first assessed cell death of normal PBMCs cultured in vitro overnight with 10% SS sera, CS sera, or HD sera. Cell death was evaluated by cell counting (eosin staining) as well as annexin-V and PI staining. Cell death was significantly increased in PBMCs cultured in the presence of SS sera (gray symbols), mainly on the BDCA2+ HLA-DR+ (pDCs), CD11c+HLADR+CD14– (cDCs), and CD11c+HLADR+CD14+ (monocytes) populations compared to cells cultured with HD sera (white symbols). By contrast, the rate of cell death was equivalent in the different lymphocyte subsets (Fig. 2A and data not shown), independently of the sera used. To confirm that serum from patients with SS was toxic to DCs, we cultured purified normal monocytes with GM-CSF and IL-4 for 4 days. The DCs were then exposed to different types of sera overnight, at different concentrations and assessed for cell death (Fig. 2B). We observed that SS sera (rhombuses) induced a higher DC death rate compared with HD sera (circles) in a dose-dependent manner. When compared with HD or CS sera, sera from patients with SS reproducibly induced a significantly higher DC death April 2015 • Volume 43 • Number 4
Online Laboratory Investigation
rate as assessed by annexin-V/PI staining (Fig. 2C) (median percentage of death 17.3, 15.5, and 33.8, respectively, with HD sera, CS sera, and SS sera). The eosin dye exclusion test confirmed the flow cytometry results by showing more eosinpositive cells in the MDDC cultured in the presence of SS sera (median, 80.4 × 103 cells/mL) compared to culture with HD (40 × 103 cells/ mL) or CS sera (50 × 103 cells/ mL) (Fig. 2D). Because several groups have observed that DC loss positively
correlates to a more severe outcome (10, 20), sera were split into two groups based on the clinical outcome of the patient: surviving (surv SS, n = 29) versus nonsurviving (non-surv SS, n = 20) patients. We observed that the percentages of serainduced DC death were equivalent in the two groups (Fig. 2E). Taken as a whole, those data suggest that sera from patients with SS are characterized by the presence of a soluble factor that induces monocyte and DC cell death.
Figure 2. Septic shock sera are toxic for circulating and monocyte-derived dendritic cells (MDDCs). A, Normal plasma blood mononuclear cells were cultured overnight with 10% normal serum (empty symbols, n = 6) or serum from septic shock (SS) patients (filled symbols, n = 9) drawn at any time point over the course of the disease. Cell death was evaluated by flow cytometry using annexin-V/propidium iodide (PI) staining. Results of individual cultures and the medians are shown in each cell population: T lymphocytes, plasmacytoid dendritic cells (pDCs), conventional dendritic cells (cDCs), and monocytes. B, MDDCs (see Patients and Methods section) were cultured overnight with different concentrations of serum from healthy donors (HD) (circles) or SS patients (rhombuses). Cell death was evaluated by flow cytometry by using annexin-V/PI staining. The results are reported as the mean of three independent experiments ± sd. C, MDDCs were cultured overnight with 10% serum from HD (circles, n = 8), cardiogenic shock (CS) patients (squares, n = 5), or SS patients (rhombuses, n = 56). Cell death was evaluated by flow cytometry using annexin-V/PI staining. D, MDDCs were cultured overnight with 10% serum from nine HD (circles), eight CS (squares) patients, or 24 SS (rhombuses) patients. Eosin dye was used to evaluate the number of dead cells. E, Sera from survival (surv SS, triangles, n = 36) or nonsurvival (non-surv SS, inverted triangles, n = 20) SS patients induce similar levels of cell death on MDDCs (ns = not significant, *p < 0.05, **p < 0.005, nonparametric unpaired tests; horizontal lines indicate medians).
Sera From Nonsurviving Patients Induce Primary Regulated Necrosis, Whereas Sera From Surviving Patients Induce Apoptosis Although the overall cell death rate was not different between MDDC incubated with sera from surv SS patients as compared to non-surv SS patients, we were intrigued by the remarkably different morphology of the cells when examined by optical microscopy. Cells cultured in the presence of sera from non-surv SS patients, but not those cultured in sera from surv SS patients, were characterized by the presence of cellular debris and clumping of dead cells (data not shown), suggesting that different mechanisms could be responsible for sera-induced DC death. Interestingly, flow cytometry analysis revealed that sera from surv SS patients induced early apoptosis (annexin-V+/ PI– population) and secondary necrosis (annexin-V+/PI+ population), whereas sera from non-surv SS patients induced mainly primary necrosis (annexin-V–/PI+ population) (Fig. 3A). This observation was confirmed in 20 sera from non-surv SS patients compared to 36 sera from surv SS patients (median percentage of necrosis, 11.7% vs 1.6%; p
