INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Systemic inflammatory response syndrome (SIRS) is frequently accompanied by a severe depression of the immune system that is most probably consecutive to the exacerbated antiinflammatory response (or compensatory antiinflammatory response syndrome, CARS) (1). Sepsis syndrome, surgery, trauma, hemorrhage, and thermal injury are associated with an exacerbated in vivo production of pro- and antiinflammatory cytokines as assessed by their increased levels in the bloodstream. Paradoxically, the capacity of circulating leukocytes from these patients to produce cytokines is reduced when compared with cells from healthy control subjects.

Monocytes from patients with sepsis show a reduced capacity to release tumor necrosis factor- (TNF-), interleukin-1 (IL-1 and IL-1), IL-6, IL-10, and IL-12 in response to lipopolysaccharide (LPS) stimulation (2). Reduced cytokine production has also been observed with other stimuli such as silica, staphylococcal enterotoxin B, and killed Streptococcus and Staphylococcus (5). Similarly, monocyte-derived cytokine production was significantly altered in patients undergoing abdominal aortic surgery (8), cardiac surgery associated with cardiopulmonary bypass (9), or trauma (10). The hyporeactivity has also been reported when studying the production of IL-1, soluble IL-1 receptor antagonist (IL-1ra), and IL-8 by LPS-activated neutrophils in patients with sepsis (11, 12). As well, the hyporeactivity of circulating leukocytes was demonstrated with peripheral blood lymphocytes. Indeed, in response to T cell mitogens, IL-2, IL-10, and interferon-gamma ex vivo productions were found to be altered in burns (13), trauma (10), cardiopulmonary bypass (14), surgery (15), and sepsis (13, 14).

This hyporeactivity resembles the phenomenon of endo- toxin tolerance. Nuclear factor-kappa B (NF-B) is the principal intracellular promoter of proinflammatory gene induction and changes in its nuclear and cytoplasmic levels (16) and its composition are thought to be responsible for the development of endotoxin tolerance. The NF-B family is composed of various members, p50 (NF-B1), p52 (NF-B2), p65 (RelA), RelB, and c-Rel, which can form homo- and heterodimers. Numerous studies have shown that the transactivator form of NF-B is the p65 unit whereas the p50 unit showed no or minimal activation capacities (19). These numerous and concordant reports in mammalian cells are contradicted by other reports showing transactivatory activities of p50p50 in yeast (23) or in cell-free in vitro transcription systems (24). Fujita and coworkers (24) found that p50p50 could behave as a gene activator when complexed to the Bcl-3 protein, but another report shows that Bcl-3 facilitates the NF-B transactivation by removing the inhibitory p50p50 from the B sites (25). Numerous stimuli, including LPS, induce the phosphorylation, ubiquitination, and subsequent degradation of the cytoplasmic inhibitor of NF-B, IB. NF-B can then translocate into the nucleus and induce the transcription of genes coding for proinflammatory proteins, such as TNF- and inducible nitric oxide synthase. Unlike naive cells, tolerized cells (those pretreated with LPS) have a predominance of the p50 homodimer of NF-B after an LPS challenge (26). Analysis of p50-deficient mice further established that the p50 subunit of NF-B plays a central role in endotoxin tolerance (27). Up-regulation, stabilization, or enhanced rate of synthesis of IB have also been suggested as part of the endotoxin tolerance phenomenon (28, 29). To understand the intracellular mechanisms involved in immunodepression of circulating leukocytes in patients with severe SIRS, we studied NF-B and IB expression in freshly isolated peripheral blood mononuclear cells (PBMC) of patients with severe sepsis and after in vitro LPS activation. We also analyzed the expression of these molecules in patients with major trauma, who present a similar inflammatory stress but without a primary infectious component to its etiology.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients with Severe Sepsis or Trauma and Healthy Control Subjects

Following approval of the study protocol by the institutional review board for human experimentation, a written informed consent was obtained for each patient before inclusion into the study. If patient consent was not possible, then a legal representative was asked. Nineteen consecutive patients with a severe sepsis, as defined by the Bone criteria (1), were included in this study. Inclusion criteria were a SIRS secondary to an infectious disease and associated with organ dysfunction, hypoperfusion, or hypotension. Patients were excluded if they were under 18 yr, had neutropenia, had received chemotherapy during the past 6 mo, were presently receiving corticosteroid therapy or any other immunosuppressive therapies, or were human immunodeficiency virus positive.

Blood samples were collected on the day of arrival in the intensive care unit (ICU) (Day 1) and when possible on Day 7. All nonsurvivors died before Day 7. Blood samples were also collected from four other patients on recovery at the time of their discharge from the ICU to see if their PBMC could translocate NF-B in response to LPS after complete recovery. As a group the 12 survivors were 57 ± 5 yr old (range 27-79), had a SAPS II score of 40 ± 4 (range 18-69), and 7 were male. As a group the 7 nonsurvivors were 47 ± 7 yr old (range 24-72), had a SAPS II score of 48 ± 6 (29-73), and 3 were male. The SAPS II scores of survivors and nonsurvivors were overlapping and were not found to be statistically different. The sepsis etiology was pneumonitis or acute lung injury (n = 7), peritonitis or digestive organ system infection (n = 5), urinary tract infection (n = 2), indwelling vascular catheter infection (n = 2), and meningitis (n = 1). Infections were due to Legionella pneumophilia (n = 1), Morganella catarrhalis (n = 1), Enteroccus fecalis (n = 2), Staphylococcus aureus (n = 1), Neisseria meningitidis (n = 1), Hafnia alvei (n = 1), Pseudomonas aeruginosa (n = 2), Escherichia coli (n = 5), or not identified (n = 5).

Thirteen consecutive patients with major trauma as defined by an Injury Severity Score (ISS)  25 were studied. Patients who had no measurable blood pressure or pulse on arrival and/or who were less than 17 yr old were excluded from the study. Blood samples were collected on the day of arrival in the ICU (Day 1). The mechanisms of traumatic injury were motor vehicle accidents, automobile-pedestrian accidents, and falls. As a group, the 13 trauma patients were 25 ± 9 yr old (range 17-46) and had mean ISS and SAPS II scores of 38 ± 9 (range 25-54) and 40 ± 12 (range 22-60), respectively, and all but one were male. Three patients died within their stay in the ICU. The mean age of these nonsurvivors was 33 ± 12 yr (range 22-46), with a mean ISS of 45 ± 4 (range 43-50) and a mean SAPS II of 47 ± 6 (range 44- 54). Their deaths were attributable to severe progressive brain injury. The patients were compared with 13 healthy control subjects (mean age 30.7 ± 8.9 yr [range 23-49] and 6 were male).

PBMC Isolation and Cytoplasmic and Nuclear Extracts Preparation

PBMC were isolated from blood freshly collected on sodium citrate by centrifugation on Ficoll-Hypaque (MSL, Eurobio, les Ulis, France). Before Ficoll, a fraction of the blood was centrifuged 5 min at 1,500 rpm and 1 ml of plasma was collected and put immediately at 20° C for further cytokine measurements. After isolation, cells were used immediately for preparation of nuclear and cytoplasmic extracts (ex vivo analysis) or were cultured first (in vitro analysis), for 1 h at 37° C in a 5% CO₂ incubator, in RPMI 1640 medium (Glutamax; Gibco Life Technologies, Paisley, UK) in the absence (LPS) or the presence (LPS⁺) of Escherichia coli (0111:B4 E. coli LPS at 1 µg/10⁶ cells/ml). Cellular extracts were prepared as previously described (30). PBMC were washed once with phosphate-buffered saline (PBS), adherent cells were harvested with a cell scraper, added to nonadherent cells, and suspended in buffer A (10 mM HEPES, pH 7.9, 1.5 mM MgCl₂, 10 mM KCl, 0.5 mM dithiothreitol [DTT], and 0.1% NP-40) supplemented with protease inhibitors. The protease inhibitors included 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 25 µg/ml aprotinin, 10 µg/ml chymostatin, 2 µg/ml antipain, 8 µg/ml pepstatin, 10 µg/ml leupeptin, 0.1 mg/ml ₁-antitrypsin, and 0.5 mM 3,4-dichloroisocoumarin (all from Sigma, St. Louis, MO). Cells were incubated 10 min at 4° C and then centrifuged for 2 min at 10,000 rpm. The supernatant corresponding to the cytoplasmic extract was frozen at 80° C. The pellet was suspended in buffer C (20 mM HEPES, pH 7.9, 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 25% glycerol, 0.5 mM DTT, and protease inhibitors) and incubated for 20 min at 4° C. Cells were then centrifuged for 10 min at 14,000 rpm, and the supernatant corresponding to the nuclear extract was harvested and kept at 80° C. Protein concentrations were determined according to the method of Bradford.

Electrophoretic Mobility Shift Assay (EMSA)

A double-stranded oligonucleotide containing the NF-B consensus motif (Promega, Madison, WI) was end labeled with T4 kinase in the presence of [-³²P]ATP. Nuclear extracts (2 µg) were incubated in the binding buffer for 15 min at room temperature (4% Ficoll, 20 mM HEPES, pH 7, 35 mM NaCl, 60 mM KCl, 0.01% NP-40, 2 mM DTT, 0.1 mg/ml bovine serum albumin [BSA], and 1.5 µg/µl salmon sperm DNA). After 15 min, the radiolabeled nucleotide was added (150,000 cpm) and the mixture was again incubated for 15 min at room temperature. Electrophoretic mobility shift assay (EMSA) was performed in a 5% acrylamide gel in 0.5 × TBE. Gels were dried and subjected to autoradiography. The NF-B complexes were quantified using a PhosphorImager and the ImageQuant software (Molecular Dynamics). As all the samples could not be analyzed on the same gel, we used a positive control (PBMC from a healthy donor stimulated with LPS) that was the same for all the gels. All gels were exposed to the PhosphorImager screen for the same period of time (i.e., 24 h). Various amounts of the same nuclear extract were analyzed to ascertain the linearity of the signal measurement. The values obtained for this positive control allowed us to calibrate the EMSA to compare the cpm from one gel to another. The cpm obtained for this positive control on one gel was chosen as a reference and the values of all the other gels were corrected by a multiplying factor that took into account the values for this positive control. This calibration was not necessary when the p65p50/p50p50 was calculated. Specificity of binding was assessed by competition with excess of cold oligonucleotide and by supershift experiments using anti-p50 and anti-p65-specific polyclonal antibodies (Santa Cruz Biotechnology, Santa Cruz, CA).

Western Blot

Four micrograms of protein of cytoplasmic extracts were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS- PAGE) in a 12% gel and transferred onto nitrocellulose sheets (Hybond C; Amersham, Buckinghamshire, UK). The protein transfer was ascertained by Ponceau red coloration. The membranes were then washed with PBS and blocked with PBS containing 0.1% Tween 20 and 5% gelatin (PBS-T-G) for 1 h at room temperature. After five washes with PBS-T, the membranes were incubated with rabbit polyclonal IgG anti-I-B (C-21; Santa-Cruz) at 1/2000 or anti-p65 (sc-109X; Santa-Cruz) or anti-p50 (sc-114X; Santa-Cruz) both at 1/20,000 in PBS-T-G for 1 h at room temperature. After five washes, peroxidase-labeled goat anti-rabbit Ig polyclonal antibodies (Silenus, Hawthorn, Australia) were added at 1/2000 in PBS-T-G and incubated for 1 h at room temperature. After five washes, the blots were developed using ECL (Amersham). Densitometry analysis was performed on the Western blots using the NIH Image software.

IL-10 Measurement

IL-10 in the plasma of patients with severe sepsis and healthy control subjects was quantified using an enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems, Abingdon, UK).

Statistical Analysis

Data are given as mean ± SEM and were analyzed by Mann-Whitney U test using the Statview II software.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nuclear Expression of NF-B in PBMC from Septic Patients

The expression of NF-B was analyzed by EMSA. A representative example of one healthy control subject and one each of the survivors and nonsurvivors of severe sepsis is shown in Figure 1A. As various NF-B complexes were observed by EMSA, we characterized them by competition with cold oligonucleotides and p50 or p65-specific antibodies. As shown in Figure 1B, the addition of an excess of cold oligonucleotide, corresponding to the NF-B binding site, turned off the signal of the two upper bands, showing that the lowest band is nonspecific. An excess of an irrelevant oligonucleotide (corresponding to the AP-1 transcription factor) did not have any effect. The addition of an anti-p50 antibody caused the supershift of the two upper bands, whereas the anti-p65 antibody affected only the uppermost band. Thus, the upper complex corresponds to the p65p50 heterodimer and the intermediate band is the p50p50 homodimer. In the PBMC from healthy control subjects, the ex vivo expression of NF-B into the nucleus and that measured in vitro without stimulation were low, whereas after LPS stimulation, an increased nuclear translocation was observed. In contrast, survivors of severe sepsis had a lower ex vivo nuclear expression of NF-B at admission and this depressed level of expression persisted following LPS stimulation in vitro. For most of the nonsurvivors the results were different: the nuclear extracts contained detectable amounts of NF-B, even without stimulation of their PBMC. However, the NF-B was mostly composed of the p50p50 homodimer.

View larger version (43K): [in this window] [in a new window]   View larger version (46K): [in this window] [in a new window]   Figure 1.   (A) Nuclear expression of NF-B in PBMC from a healthy control subject, a survivor, and a nonsurvivor of severe sepsis analyzed by EMSA. The nuclear extracts were obtained from freshly isolated PBMC (ex vivo) or PBMC cultured in vitro for 1 h at 37° C in the absence (LPS) or the presence of E. coli LPS (+LPS). (B) The specificity of the bands was assessed by the incubation of the nuclear extracts with an excess of cold NF-B oligonucleotide or with an irrelevant cold oligonucleotide (corresponding to the transcription factor AP-1). The NF-B complexes were characterized by supershift experiments using antibodies specific to the p50 or the p65 subunits. (C) Mean ± SEM of the cpm for nuclear p50p50, (D) p65p50, and (E) p65p50 + p50p50 in PBMC from healthy control subjects, survivors (at admission of the patients in the intensive care unit [D1] and 7 d later [D7]), and nonsurvivors (D1) with severe sepsis. The cpm obtained for each patient were calibrated with an internal positive control (see METHODS). (F  ) The ratio ± SEM between nuclear p65p50 and p50p50 forms of NF-B is shown for healthy control subjects, survivors (on Days 1 and 7), and nonsurvivors with severe sepsis. p Values versus control subjects are indicated.

After EMSA, the p50p50 and p65p50 complexes were quantified using a PhosphorImager. The results corresponding to the NF-B expression on Day 1 for all patients with sepsis (n = 19) and survivors only for Day 7 (n = 12) are shown in Figures 1C to 1E. These data are compared with 13 healthy control subjects. PBMC from healthy control subjects expressed basal nuclear NF-B ex vivo both for p50p50 (Figure 1C) and p65p50 (Figure 1D). This expression was moderately down-regulated in vitro without LPS, whereas it was up-regulated by LPS stimulation with the heterodimer increasing its expression more than the homodimer. In contrast, the nuclear p65p50, the active form of NF-B, was reduced in all patients with severe sepsis as compared with control subjects (p = 0.05 ex vivo and 0.003 in vitro following LPS stimulation). The ex vivo expression of nuclear NF-B was significantly lower in PBMC from surviving patients with sepsis at Day 1 and Day 7 and at both time points LPS stimulation failed to induce the translocation of either p50p50 (Figure 1C) or p65p50 forms (Figure 1D). For the nonsurviving patients with sepsis, the in vitro nuclear expression of p65p50 after LPS stimulation was significantly lower than control subjects (Figure 1D), whereas the inactive form of NF-B (p50p50) was present in amounts comparable to those found for the control subjects (Figure 1C). Consequently, for survivors the total nuclear NF-B was significantly lower in PBMC both ex vivo and following LPS than control subjects, whereas this was not the case for nonsurvivors (Figure 1E).

Because the p65p50 dimer is a potent gene transactivator whereas the p50p50 dimer plays an inhibitory role, the ratio between these two forms as well as the absolute amount of either form is important in controlling the intracellular inflammatory response. Specifically, a ratio not unfavorable to the heterodimer is needed for gene activation (39, 41). As shown in Figure 1F, the p65p50/p50p50 ratio was around 1.5 ex vivo for healthy control subjects and reached 2.0 in vitro after LPS stimulation. Even on inclusion (Day 1) the ratio of p65p50 to p50p50 for survivors of severe sepsis was significantly lower than for control subjects both ex vivo and in vitro after LPS stimulation. By 7 d, however, the ex vivo p65p50/p50p50 ratio returned to values not dissimilar to those of healthy controls, but the in vitro response of the PBMC to LPS was still depressed. As described above, nonsurvivors expressed more NF-B than did survivors of severe sepsis. However, because most of their nuclear NF-B was present in the homodimer form, the p65p50/p50p50 ratio was lower than that of control subjects. Although this difference did not reach significance when assessed from PBMC ex vivo, after in vitro LPS stimulation, this ratio was significantly lower than that of control subjects. Furthermore, the p65p50/p50p50 ratio for nonsurvivors was also significantly lower than that of survivors (p = 0.02). In an attempt to ascertain the time course of recovery of NF-B responsiveness to LPS stimulation, we analyzed the nuclear expression of NF-B in PBMC from four patients with severe sepsis on their discharge from the ICU (on Days 11, 18, 36, and 49, respectively). Indeed, their NF-B EMSA profiles were similar to those of healthy control subjects demonstrating a nuclear translocation of p50p50 and higher amounts of p65p50 in vitro after LPS stimulation (data not shown).

IL-10 Concentration in the Plasma of Patients with Severe Sepsis and Correlation with the p65p50/p50p50 Ratio

IL-10 is a well-known immunosuppressive and antiinflammatory cytokine. We investigated whether the absence of nuclear translocation of NF-B found for patients with sepsis was linked to plasma IL-10 levels. On Day 1, IL-10 was present in the plasma of both survivors (80 ± 30 pg/ml) and nonsurvivors (255 ± 82 pg/ml), whereas it was under the limit of detection (20 pg/ml) for healthy control subjects (p = 0.002 and p = 0.001, respectively, versus controls). The IL-10 levels were higher in the plasma of the nonsurvivors as compared with survivors (p = 0.035). Importantly, we found a reverse correlation between plasma IL-10 levels in nonsurvivors and the NF-B p65p50/p50p50 ratio after in vitro LPS stimulation (r = 0.82, p = 0.046). This correlation was not found in ex vivo PBMC, nor in survivors either ex vivo or in response to LPS stimulation.

Nuclear Expression of NF-B in PBMC from Trauma Patients

The nuclear expression of NF-B was also studied in PBMC from patients with trauma who present a noninfectious inflammatory syndrome. Thirteen patients were studied. We found similar results for survivors and nonsurvivors and as the nonsurvivor group was small (three patients), despite high ISS scores of these patients, we show the values for all the patients with trauma together. As shown in Figure 2A, the ex vivo nuclear expression of NF-B on the first day of trauma was significantly lower both for the homodimer and heterodimer forms as compared with healthy control subjects. Furthermore, the nuclear NF-B was very low when compared with control subjects in vitro after LPS stimulation. The difference was not significant for p50p50, but was for p65p50. The p65p50/p50p50 ratio was significantly lower for patients with trauma versus control subjects only after LPS stimulation but not ex vivo (Figure 2B). These results show that an inflammatory insult without infection is sufficient to block NF-B nuclear translocation in response to an LPS stimulation. However, even if the number of nonsurviving patients with trauma in our study was low, we did not observe the high expression of p50p50 homodimer seen in nonsurvivors of severe sepsis.

View larger version (25K): [in this window] [in a new window]   Figure 2.   The nuclear expression of NF-B in PBMC from healthy control subjects and patients with trauma, at the day of their admission in the intensive care unit, was analyzed by EMSA. (A) cpm ± SEM corresponding to the p50p50 and p65p50 forms are indicated for freshly isolated PBMC (ex vivo) and after a 1 h in vitro incubation at 37° C in the presence of E. coli LPS (+LPS). (B) The ratio ± SEM between nuclear p65p50 and p50p50 forms of NF-B, corresponding to freshly isolated PBMC (ex vivo) and PBMC after a 1 h in vitro incubation at 37° C in the presence of E. coli LPS (+LPS) is shown for healthy control subjects and patients with trauma. p Values versus control subjects are indicated.

Detection of Cytoplasmic IB, p65, and p50

As p65 and p50 were barely detected in the nucleus of patients' PBMC (except for p50 for nonsurvivors of sepsis), we studied the expression of IB, to determine whether its up-regulation could explain the low expression of NF-B. IB belongs to the family of cytoplasmic inhibitors of NF-B and by binding to the dimer prevents its nuclear translocation. IB expression was analyzed by Western blot in cytoplasmic extracts of PBMC and the bands were quantified by densitometry. As expected, we saw IB in all the ex vivo cytoplasmic extracts of PBMC from healthy control subjects, with some individual variability of its expression, mean density = 40,609 ± 4,769 arbitrary units (Figure 3). In contrast, the ex vivo expression of IB was significantly lower for patients with severe sepsis (2,562 ± 805 arbitrary units) or major trauma (8,717 ± 2,627 arbitrary units). No difference was seen between survivors and nonsurvivors of severe sepsis. Upon LPS stimulation, no further reexpression of IB was observed in patients' PBMC (data not shown). Western blot analysis of p65 and p50 (Figure 3) revealed a low expression of these molecules within the cytoplasm of patients' PBMC. Indeed, the densitometric values found for patients were significantly lower than that of control subjects for both p65 and p50, for severe sepsis as well as major trauma. Thus, the low nuclear expression of NF-B was not due to its sequestration in the cytoplasm by IB but reflected a more generalized down-regulation of NF-B expression in these cells.

View larger version (21K): [in this window] [in a new window]   Figure 3.   Densitometric results (mean ± SEM) from Western blots for cytoplasmic IB, p65, and p50. The Western blot was performed on ex vivo cytoplasmic extracts of PBMC from healthy control subjects, surviving patients with sepsis on Days 1 and 7, and nonsurviving patients with sepsis and patients with trauma on Day 1.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Endotoxemia and severe immunodepression frequently accompany systemic inflammatory response syndrome (SIRS). Circulating leukocytes from patients with sepsis syndrome or following major surgery, trauma, hemorrhage, and thermal injury have a reduced capacity to produce cytokines in response to lipopolysaccharide (LPS) stimulation (2, 4, 8, 10). This phenomenon, known as "deactivation," "desensitization," "anergy," or "refractoriness," occurs in many inflammatory stresses and is very similar to another phenomenon described as "endotoxin tolerance" (31). Thus, we hypothesized that this immunodepression reflected intracellular changes similar to that reported in endotoxin tolerance and analyzed the nuclear expression of NF-B in PBMC from patients with SIRS of infectious (severe sepsis) or noninfectious origin (trauma).

To study the ex vivo and in vitro NF-B expression, we chose to analyze the whole mononuclear cell population rather than isolated monocytes to minimize cell manipulation that could interfere and induce activation signals altering NF-B expression. Numerous studies have shown that p65p50 is a potent transactivator whereas the p50p50 dimer is not (19). We found that all patients showed a reduced ex vivo nuclear expression of p65p50. For the p50 homodimer all the patients did not show the same profile. The survivors of severe sepsis and patients with trauma also had a very low p50p50 nuclear expression, whereas this form was present for nonsurvivors of sepsis in amounts comparable to control subjects. But more importantly, the PBMC of the patients were not able to perform NF-B translocation upon LPS activation, similar to endotoxin-tolerized cells. The survivors of severe sepsis and patients with trauma showed low expression of both active (p65p50) and inactive (p50p50) forms of NF-B after LPS stimulation, resembling what is found in some endotoxin tolerance experiments, where tolerance was associated with a depletion of both forms of NF-B (16), whereas nonsurvivors of severe sepsis showed a predominance of the inactive homodimer and a low p65p50/p50p50 ratio, similar to the tolerized cells described by Ziegler-Heitbrock (26). Besides general down-regulation of NF-B, the predominance of p50 homodimer over p65p50 is another inhibitory mechanism of gene activation by NF-B. Indeed, cotransfection experiments and reporter gene assays with plasmids carrying p50 and p65 have shown that p50 inhibits the activation induced by p65 in a dose-dependent manner (19). Franzoso and coworkers (21) have established that p50, cotransfected with similar amounts of p65, acts synergically for gene activation, since the p65p50 heterodimer has a greater affinity for the -site than the p65 homodimer. However, an increase of p50 plasmid amounts resulted in an inhibition of the reporter gene activity and EMSA showed that the p65p50 complex disappeared whereas the p50p50 homodimer was detected. The same group reported that an equivalent amount of p50 DNA, when cotransfected with p65 DNA, permitted transactivation, whereas a 3-fold excess of transfected p50 DNA over p65 markedly inhibited transactivation (25). Densitometric analysis of EMSA gels has shown that the p65p50/p50p50 ratio was of 1.4 ± 0.2 in LPS-stimulated cells, whereas it was of 0.8 ± 0.1 in tolerized cells after a second LPS challenge (26). Similarly, Goldring and coworkers (32) found that the p50p50 form was more abundant in tolerized cells and the ratio was in favor of this form.

It is worth noting that the up-regulation of p50p50 appeared instrumental for the reduced TNF- and inducible nitric oxide synthase genes expression observed in tolerized cells (32). The inability of macrophages derived from p50-deficient mice to develop endotoxin tolerance further reinforces the role played by the p50 subunit of NF-B in conferring endotoxin tolerance (27). Our study shows that endotoxin tolerance and NF-B modulation also occur in vivo in patients with SIRS, independent of the presence of an infectious insult (i.e., it was observed for both patients with sepsis and trauma), in agreement with the nonspecific nature of the so-called "endotoxin tolerance" (31). However, the mechanism of tolerance seems to be different for the nonsurvivors of severe sepsis, when compared with the survivors and with patients with trauma.

A previous study that addresses NF-B expression during sepsis reported data on only the ex vivo expression of this nuclear factor: Böhrer and coworkers (33) reported a higher ex vivo nuclear expression of NF-B in PBMC in nonsurvivors. This work, however, did not include a comparison with healthy control subjects and did not quantify p65p50 and p50p50 expression. Our data agree with this previous study in confirming that total NF-B content is higher in the nucleus of PBMC from nonsurvivors as compared with survivors, although this difference did not reach significance in our study (Figure 1E, p = 0.07). Furthermore, we found that the nuclear p65p50, the active form of NF-B, was significantly reduced in all patients with severe sepsis as compared with control subjects. More importantly, we demonstrated that in the nonsurvivors, NF-B was mostly composed of the inactive form and that the p65p50/p50p50 ratio was significantly lower than in survivors. A similar observation obtained with patients with trauma suggests that infection per se is not a prerequisite for the disturbance of NF-B expression. Furthermore, in contrast to patients with sepsis, in patients with trauma the analysis occurs at the onset of the inflammatory process. Thus, our data suggest that the dysregulation of NF-B expression is induced very early after the inflammatory insult.

It has been shown that the immunosuppressive and antiinflammatory cytokine IL-10 could alter NF-B expression and translocation and contribute to cell desensitization (34). Furthermore, we recently reported that in vitro IL-10 altered the expression of both p65p50 and p50p50 (30). Thus, we investigated whether the absence of nuclear translocation of NF-B found for patients with sepsis was linked to plasma IL-10 levels. In agreement with previous reports (35), the IL-10 levels were higher in the plasma of the nonsurvivors as compared with survivors. Importantly, we found a reverse correlation between plasma IL-10 levels in nonsurvivors and the NF-B p65p50/p50p50 ratio after in vitro LPS stimulation. This correlation was not found in ex vivo PBMC, nor in survivors either ex vivo or in response to LPS stimulation. This observation suggests that IL-10 may well be an actor in the cell desensitization and alteration of the NF-B cascade in patients with life-taking sepsis.

Several investigators have analyzed IB expression in in vitro endotoxin-tolerized cells. Some found an increased IB level in tolerized cells (28, 29). In contrast, in another study (36), after the first exposure to LPS and prior to the second challenge, the cells did not express cytoplasmic IB, similar to what we found within PBMC derived from patients with severe sepsis or trauma. Indeed, the ex vivo expression of IB was quite low in the case of many of them, whereas it was detected in more important amounts in PBMC from healthy control subjects. Thus, the absence of nuclear NF-B was not the consequence of its cytoplasmic sequestration by IB. Our results suggest that there is a general down-regulation of NF-B expression in these cells as assessed by the low levels of NF-B measured within the cytoplasm. This observation is in accordance with the work of Blackwell and coworkers (17) who studied tolerized cell lines.

Our observation on a dysregulation of NF-B translocation in PBMC of patients with severe sepsis and the low presence of cytoplasmic IB suggests that although the successful use of drugs in animal models to improve sepsis has been shown, either by inhibiting NF-B activation (37) or by inhibiting proteolysis of IB (38), this approach may not be appropriate to cure patients with sepsis, unless they can be delivered at the onset of inflammation or within defined compartments. Indeed, in contrast to the hyporeactivity of circulating cells, an exacerbated production of cytokines by these cells has often been demonstrated in nonhematopoietic tissues (39, 40). Then it would be helpful to further analyze the expression of NF-B and IB in leukocytes recruited within the inflamed tissues.