Introduction
Sepsis is a major medical challenge with a high annual incidence rate. Despite improvements in critical care, however, the outcome from sepsis has improved little and mortality rates remain high [
1‐
3]. Earlier, the prevailing theory was that mortality from sepsis largely is a consequence of an overwhelming host inflammatory response [
4‐
6]. Failure of clinical trials targeting inflammatory mediators to improve the outcome from sepsis and recent insights prompted reconsideration of this concept [
4‐
8]. Today, it is recognized that the host's immune response during sepsis changes over time, resulting in both inflammation and profound immunosuppression in the later course of the disease. Many patients surviving the early phase of sepsis therefore often show signs of severe immunosuppression [
4‐
6,
9‐
16].
A number of immune dysfunctions have been reported in sepsis, including apoptosis of T lymphocytes and B lymphocytes, altered cellular cytokine production, increased levels of the anti-inflammatory IL-10, impaired phagocytosis, monocyte deactivation with diminished major histocompatibility class II molecule expression, and altered response to microbial products [
17‐
22]. The term immunoparalysis was proposed to describe the host's general inability to mount effective immune responses. We and other workers have demonstrated an association between low levels of the major histocompatibility complex class II molecule human leukocyte antigen (HLA)-DR on monocytes and the impairment of cellular immunity in sepsis, including decreased production of proinflammatory cytokines, impaired antigen presentation, and reduced
ex vivo lymphocyte response to recall antigens [
9,
20,
23,
24]. Importantly, prolonged downregulation of monocytic HLA-DR was associated with an adverse outcome from sepsis [
20,
24]. Consequently, a number of clinical pilot trials aiming to reverse immunoparalysis via immunomodulatory strategies were recently performed [
9,
25,
26].
In contrast to the extensively studied major population of classical CD14
bright monocytes, little is known about phenotypic and functional changes of CD16
positive (Fcγ receptor III) monocyte subsets in sepsis. In healthy individuals about 10 to 15% of circulating monocytes are CD16
positive cells, which express higher levels of HLA-DR and proinflammatory cytokines than CD16
negative monocytes after stimulation with microbial products. This CD16
positive subset has therefore been referred to as proinflammatory monocytes [
27‐
29]. Although expansion of CD16
positive monocytes was shown in sepsis [
30], it is currently unclear whether this subset undergoes functional deactivation similar to classical CD14
brightCD16
negative monocytes in sepsis.
Dendritic cells (DCs) are the most potent antigen-presenting cells (APCs) and play a key role in linking innate and adaptive host immune responses to microorganisms. Distinct subsets of circulating DCs can be identified in peripheral blood, including myeloid dendritic cells (MDCs) and plasmacytoid dendritic cells (PDCs) [
31]. Although arising from common precursor cells in the bone marrow, MDCs and PDCs are phenotypically and functionally different [
32]. For example, PDCs but not MDCs express the receptor for dsDNA (Toll-like receptor (TLR) 9), while TLR4, the receptor for bacterial lipopolysaccharide (LPS), is restricted to MDCs [
31]. Activation of MDCs by LPS via TLR4 results in the secretion of TNFα, IL-1β and IL-6, while PDCs secrete enormous amounts of IFNα after stimulation with the TLR9 ligand CpG oligonucleotides (ODN), and may play an important role in antiviral immunity [
31,
33].
Upon encountering microbial products, DCs undergo phenotypic and functional maturation and migrate to secondary lymphatic organs, where they induce adaptive T-cell responses. Compromised DC function was associated with increased disease severity and adverse outcome in animal models of sepsis [
34‐
36]. Increased apoptosis of DCs has been demonstrated in spleens from patients with sepsis, and an early decrease in circulating DCs was shown to correlate with increased disease severity and mortality [
37,
38]. Data on functional changes in DCs in sepsis patients, however, remain scarce.
The aim of the present study was to determine and compare phenotypic differences and functional changes in different monocyte and DC subsets over time in patients with sepsis and immunoparalysis.
Materials and methods
Study population and protocol
Sixteen consecutive patients (13 men, age 59 ± 9.7 years) with severe sepsis or septic shock and immunoparalysis hospitalized in the surgical intensive care unit of a tertiary care academic centre were included between January 2004 and January 2005. Sixteen healthy volunteers (13 men, age 46 ± 11.4 years) served as controls.
During the study interval, a total of 22 intensive care unit patients were screened for the presence of immunoparalysis, and all patients who fulfilled the inclusion criteria entered the analysis. The following inclusion criteria applied: age > 18 years, presence of severe sepsis or septic shock [
39], and presence of immunoparalysis (monocytic HLA-DR expression < 5,000 antibodies/cell). Hepatitis B or hepatitis C patients, HIV patients and patients receiving immunosuppressive drugs (for example, steroids) were excluded.
Disease severity was assessed daily using the Simplified Acute Physiology Score 2 and the Sequential Organ Failure Assessment score. Clinical data, microbiological data and 28-day mortality were recorded. Blood samples were collected on the day after enrolment (baseline) and at study day 28. Informed consent was achieved from the patient or respective representatives. The study was performed in adherence with the Declaration of Helsinki and was approved by the local ethics committee on human research.
For ex vivo cell culture, RPMI 1640 medium (PAA Laboratories, Pasching, Germany) was used. The medium was tested for low TNF-inducing capacity (TNFα release < 10 pg/ml) in heparinized whole blood samples from healthy controls. Endotoxin (LPS) from Escherichia coli (L-4516) was purchased from Sigma (Steinheim, Germany) and lipoteichoic acid (LTA) from Staphylococcus aureus (DSM 20233) was a kind gift from Dr S. Morath (Konstanz, Germany). Commercially available ODN CpG 2336 (ODN class A), CpG 2243 (control class A), CpG 2395 (ODN class C) and CpG 2137 (control class C) were used (Coley Pharmaceutical, Kanata, Canada).
Determination of cytokine secretion by monocytes and dendritic cells
Heparinized blood was diluted 1:5 in RPMI without supplements and was incubated (6 hours, 37°C, 5% CO
2) with 100 ng/ml LPS or 10 μg/ml LTA for cytokine measurement in the supernatants. For stimulation with ODN class A and ODN class C, peripheral blood mononuclear cells were isolated from heparinized venous blood samples by density gradient centrifugation using Ficoll-Paque (Pharmacia, Freiburg, Germany). Peripheral blood mononuclear cells were cultured at a concentration of 2 × 10
6 cells/ml in supplemented RPMI 1640 medium and were stimulated with 1 μg/ml ODN class A, ODN class C, ODN control class A or ODN control class C. After incubation (24 hours, 37°C, 5% CO
2) supernatants were separated from cells for cytokine measurement. Quantification of HLA-DR on circulating monocytes was performed using a standardized flow cytometric assay, as described elsewhere [
40].
For enumeration of DC subsets, 150 μl whole blood was stained with FITC-conjugated antibodies against lineage markers (lin1) (mixture of anti-CD3/CD14/CD16/CD19/CD20/CD56), anti-CD123-PE, anti-HLA-DR-PerCP and anti-CD33-APC (BD Biosciences, Heidelberg, Germany). PDCs were gated as lin1 -CD123+HLA-DR+ events, and MDCs as lin1 -CD33+HLA-DR+ events. After treatment with FACS Lysing Solution (BD Biosciences), at least 150 to 300 events per DC population were analyzed on a FACSCalibur using CellQuestPro (BD Biosciences) software. HLA-DR expression on DCs was measured as the mean fluorescence intensity. Absolute APC population frequencies were calculated as white blood cell counts multiplied by the ratio of the APC population over all leukocytes.
Intracellular cytokine staining by flow cytometry
For flow cytometric measurement of intracellular cytokines, heparinized blood samples were diluted 1:1 in RPMI without supplements and were stimulated with 100 ng/ml LPS and 10 μg/ml Brefeldin A (Sigma) for 6 hours (37°C, 5% CO2). After stimulation, cells were washed and stained with anti-CD14-FITC, anti-HLA-DR-PerCP and anti-CD33-APC (BD Biosciences). Leukocytes were fixed and permeabilized with FACS Lysing Solution and FACS Perm2 (BD Biosciences), and were stained with anti-TNFα-PE, anti-IL-1β-PE, anti-IL-6-PE, anti-IL-10-PE (BD Biosciences) or murine IgG1-PE (Immunotech, Marseille, France) as control.
Detection of cytokines, procalcitonin and C-reactive protein
Cytokine production was assayed in culture supernatants and plasma by ELISA. Commercial kits were used to determine IFNα (PBL Biomedical Laboratories, Piscataway, NJ, USA), TNFα, IL-1β, IL-6 and IL-10 in supernatants (R&D Systems, Minneapolis, MN, USA). IL-10 plasma levels were measured by ultrasensitive ELISA (lower detection limit, 0.78 pg/ml; Biosource, Nivelles, Belgium). Immunoluminometric assays (Lumi® PCT; Brahms, Hennigsdorf, Germany) were used to detect procalcitonin plasma levels. High-sensitivity C-reactive protein was measured immunoturbidometrically in a certified laboratory.
Statistical analysis
For statistical analyses, SPSS for Windows software (version 12.0; SPSS, Inc., Chicago, IL, USA) was used. Data are presented as the mean ± standard deviation. The Mann – Whitney U test was used for comparison between patients and controls. Wilcoxon's test was used for comparison between baseline and day 28 in the patient group. P < 0.05 was considered significant.
Discussion
Altered monocyte function, including diminished HLA-DR expression and impaired proinflammatory cytokine response, was previously reported in patients with sepsis, severe trauma and major surgery. Such monocytic deactivation indicates a state of globally impaired immune functions and correlates with poor clinical outcome in critically ill patients. Nevertheless, whether this phenomenon is restricted to classical monocytes or is common to all monocyte and DC subsets is currently unclear. We demonstrate that sepsis-induced immune dysfunction affects all circulating myeloid APC subsets and that these functional alterations are long-lasting.
Today, it is well established that circulating monocytes represent a heterogeneous cell population. Among the antigenic markers, CD14 and CD16 (also known as Fcγ RIII) are commonly used to distinguish monocyte subsets (Table
2). In addition to the majority of monocytes that express high levels of CD14 but not CD16, a minor population of CD16
positive monocytes was identified. These cells have characteristic expression patterns distinct from classical monocytes, including high HLA-DR expression. CD16
positive monocytes may be subdivided into CD14
bright and CD14
dim cells. The latter subset has morphological and functional similarities to DCs, including a strong capacity to activate naïve T cells
in vitro [
42,
43]. Expansion of CD14
brightCD16
positive monocytes has been observed in patients with sepsis and other inflammatory conditions [
30,
42,
43]. Little is known, however, of the functional changes including both cytokine production and HLA-DR expression in CD16
negative and CD16
positive monocyte subsets during the course of sepsis.
Consistent with previous reports, we observed a significant increase in the circulating numbers of CD14
brightCD16
negative and CD14
brightCD16
positive monocytes in sepsis. Unlike these monocyte subsets, CD14
dimCD16
positive monocytes were significantly decreased in our patient population. This is in contrast with previous data demonstrating an increase in both CD14
bright and CD14
dimCD16
positive monocytes in neonates and children with sepsis [
44], and may reflect age-related differences in the differentiation and/or survival of CD16
positive monocytes.
Similar to classical monocytes, we observed that CD16
positive subsets show signs of profound functional deactivation in sepsis. Although HLA-DR levels differ between respective subsets, HLA-DR was diminished in all monocyte subsets in sepsis at baseline. Notably, CD14
dimCD16
positive monocytes showed only a slight reduction of HLA-DR expression at baseline while HLA-DR levels of both CD14
bright subsets remained significantly diminished in sepsis even at day 28. Moreover, consistent with previous data [
9,
45], we observed significantly reduced proinflammatory cytokine production (TNFα, IL-1β, IL-6) and increased anti-inflammatory cytokine levels (IL-10) after stimulation of whole blood with LPS and LTA in septic patients. Although we did not determine cytokine secretion in isolated monocyte subsets, we demonstrate reduced intracellular levels of TNFα and IL-6 in both CD16
negative and CD16
positive monocytes after LPS stimulation. Together with markedly diminished cytokine levels in the supernatants of LPS-stimulated whole blood cultures from septic patients (despite a significant increase in absolute numbers), this may indicate that both CD16
negative and CD16
positive monocytes undergo deactivation in sepsis.
Interestingly, in contrast to the markedly reduced cytokine levels in supernatants of stimulated whole blood cultures, differences in the percentage of cytokine-positive monocytes after LPS stimulation were less pronounced between patients and controls. Notably, the percentage of IL-1β-positive monocytes did not differ between septic patients and controls even at baseline, suggesting that proteolytic processing and/or secretion of IL-1β rather than synthesis and intracellular accumulation of inactive pro-IL-1β in monocytes are defective in sepsis. In fact, interference with the proteolytic cleavage of pro-IL-1β and secretion of mature IL-1β was proposed as a potential mechanism of the immunosuppressive effect of IL-10 [
46]. In addition to defects in cytokine transcription and translation, reduced monocytic cytokine production in sepsis may result from impaired post-translational processes involved in cytokine secretion.
DCs are key players in innate and adaptive immune responses. During infection, tissue-resident DCs recognize characteristic microbial patterns resulting in the uptake of pathogens, maturation and migration of DCs to lymphoid tissue, and activation of T-cell responses. In mice, previous studies demonstrated extensive depletion of DCs in secondary lymphatic organs after endotoxin challenge and polymicrobial sepsis [
34,
47,
48]. Markedly reduced numbers of DCs were also observed in the spleens of patients with sepsis [
38]. In a mouse model of polymicrobial sepsis induced by cecal ligation and puncture, increased numbers of apoptotic CD11c
+ DCs in mesenteric lymph nodes have been demonstrated as early as 24 hours after cecal ligation and puncture [
34]. Moreover, reduced numbers of circulating DCs in patients have been observed within 24 hours after onset of septic shock [
37]. These data indicate that DC apoptosis occurs early in sepsis, and prompted us to assess functional DC alterations in the course of sepsis. We observed a profound reduction in peripheral MDC and PDC counts in septic patients at baseline that remained significantly decreased on day 28 compared with controls. Whether this is due to ongoing DC apoptosis, due to increased migration of circulating precursor DCs into peripheral sites of inflammation or results from prolonged diminished re-population of DCs from the bone marrow, however, remains speculative [
49].
To the best of our knowledge, we are the first to demonstrate a marked and sustained functional impairment of circulating DC subsets in patients with sepsis. Similar to the phenotypic changes in monocytes resembling functional deactivation, peripheral blood DCs from septic patients showed a downregulation of surface HLA-DR expression and a reduced secretion of proinflammatory cytokines upon stimulation with microbial products. Relative to healthy controls, stimulation of MDCs from septic patients with LPS resulted in significantly reduced production of TNFα and IL-6, as indicated by intracellular cytokine staining. In addition, IFNα secretion by PDCs after stimulation with TLR9-activating CpG ODN was significantly decreased in sepsis, and this reduction exceeded (more than twofold) the decrease in PDC counts both at baseline and after 28 days.
Collectively, our data suggest a functional deactivation of both MDCs and PDCs during sepsis. Recent experimental findings provided experimental evidence for a crucial role of defective DC responses for the increased susceptibility to secondary infections during sepsis by demonstrating that increased mortality to an otherwise innocuous pulmonary
Aspergillus fumigatus or
Pseudomonas aeruginosa challenge in post-septic mice can be reversed by adoptive therapy using bone-marrow-derived DCs [
49,
50]. Moreover, significantly lower peripheral blood MDC and PDC counts have already been observed in nonsurvivors early after onset of septic shock [
37]. Whether the loss of DC correlated to the persistence of primary infections or to the occurrence of opportunistic infections, however, was not investigated. Further studies are needed to elucidate the specific consequences of the sustained loss and dysfunction of circulating DC subsets for impaired antimicrobial defenses in sepsis patients.
The mechanisms leading to altered cytokine responses and diminished major histocompatibility complex class II expression in DCs during sepsis are incompletely understood. Recent experimental studies
in vitro and
in vivo have demonstrated that DCs, similar to monocytes, become tolerant after exposure to microbial products – resulting in reduced production of proinflammatory cytokines upon repeated stimulation [
48,
51]. In addition to impaired cytokine responses, endotoxin-desensitized DCs were shown to be poor inducers of T-helper type 1 cell responses [
51]. Tolerance induction in DCs was also shown for other TLR ligands, including CpG ODN [
52]. The data presented here are consistent with a functional impairment of TLR4 and TLR9 agonist-induced cellular responses in MDCs and PDCs in patients with sepsis. In line with previous findings, we found increased IL-10 levels in septic patients at baseline and at day 28 [
17,
21]. IL-10 might contribute to the observed downregulation of HLA-DR on monocytes and DCs via enhanced re-endocytosis and sequestration of HLA-DR in monocytes [
17]. In line with recent findings demonstrating that IL-10 inhibits IFNα production in PDCs
in vitro, we observed in our study an inverse correlation of IL-10 levels with CpG-induced IFNα production by PDCs (
P < 0.001, Spearman rho = -0.83). Nonetheless, other factors including downregulation of TLR-receptor expression and altered TLR-induced signal transduction may play a role in monocyte and DC deactivation in sepsis [
48].
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
H-DV and CM were responsible for the study design. HP performed all experiments and recorded all clinical data. HP, JCS, and CM were responsible for data management and statistical analysis. HP, JCS and CM wrote the manuscript. HZ-B was responsible for patient recruitment and management, and participated together with H-DV in the interpretation of all data and revised the manuscript for important intellectual content.