INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sepsis can result in a progressive dysfunction of multiple organ systems (multiple organ dysfunction syndrome, MODS) remote from the original site of infection, presumably because of the associated host-initiated generalized inflammatory response (1). The mechanisms of cell and organ injury underlying MODS are multiple and include microcirculatory hypoxia, formation of various inflammatory mediators, and toxin-induced cell and tissue injury. A number of circulating biologic markers have been proposed for the evaluation of the severity of sepsis, including cytokines (2) or the expression of adhesion molecules on the cell surfaces of immune effector cells (3). Although these markers describe population-specific patterns of response, they are highly variable among individuals, thus limiting their clinical use. Recently, interest has focused on the interaction of inflammation and apoptosis. Mitochondrial dysfunction has been closely linked to programmed cell death (4), and alterations in mitochondrial function have been described in muscle and liver mitochondria from septic rats and primates (8). Furthermore, mitochondrial dysfunction has been suggested as a potential mechanism explaining tissue hypoxia despite normal oxygen availability in sepsis (11, 12). More recently, T-lymphocyte mitochondrial alterations have been described in septic mice (13). We hypothesized that mitochondrial dysfunction might also be detected in peripheral blood monocytes in human sepsis, and may reflect the degree of systemic injury.

As such markers, we focused our interest on mitochondrial function, cell death by apoptosis or necrosis, and proteins regulating programmed cell death such as Fas ligand, heat shock protein 70 (Hsp70), and Bcl-2. Here we analyzed these functions in monocytes from septic patients and healthy control subjects. Our results indicate that mitochondrial membrane potential (m) alterations and subsequent cell death can be observed ex vivo in monocytes from septic patients. Alterations in m were more pronounced in non-survivor septic patients than in survivors, suggesting m to be a marker for the disease outcome.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection

This study protocol was approved by our Institutional Review Board for human experimentation (Medical ICU, Cochin Port-Royal Hospital). Informed consent was obtained from all patients' next of kin. Eighteen patients who fulfilled severe sepsis and septic shock criteria, as described by Bone and colleagues (14), were enrolled into this study. They were from the Medical Intensive Care Unit of Cochin Port-Royal Hospital, Paris, France. Inclusion criteria were evidence of both an infectious site and signs of inadequate organ perfusion (mental dysfunction, hypoxemia, elevated plasma lactate, or oliguria). Exclusion criteria included younger than 18 yr of age, preexistent neutropenia, chemotherapy during the previous 6 mo, active corticosteroid treatment or any other immunosuppressive therapy, and HIV+ patients. Both the Simplified Acute Physiology Score (SAPS II) (15) and the Logistic Organ Dysfunction (LOD) Score (16) were calculated for the first day in the Intensive Care Unit. The diagnosis of ARDS was assigned by the American-European consensus conference (17) criteria: acute onset, arterial hypoxemia with Pa_O2/FI_O2 ratio lower than 200 (regardless of PEEP level), bilateral infiltrates seen on chest radiography, pulmonary artery occlusive pressure lower than 18 mm Hg, or lack of clinical evidence of atrial hypertension. Seventeen healthy health care donors volunteered as a control group and were studied at the same time.

Reagents and Media

RPMI 1640 medium, fetal calf serum (FCS), HEPES buffer, glutamine, phosphate-buffered saline (PBS), and bovine serum albumin (BSA) were from Gibco Laboratories (Paisley, Scotland), Ficoll-Plaque from Pharmacia (Uppsala, Sweden), saponin and paraformaldehyde from Sigma Chemical Co (St. Louis, MO). Monoclonal antibody SPA-810, specific for the inducible form of Hsp70, was obtained from StressGen Biotechnologies (Victoria, BC, Canada), and fluoresceine isothiocyanate (FITC) conjugated antihuman Bcl-2 and the FITC conjugated antimouse IgG used as secondary antibody, from Dako Corp. (Carpinteria, CA). 5,5',6,6'-tetrachloro-1,1',3,3'-tetraethylbenzimidazolyl-carbocyanine iodide (JC-1) was purchased from Molecular Probes (Eugene, OR), the annexin-V-fluos staining kits from Boehringer Mannheim (Mannheim, Germany), and soluble Fas ligand from Medical & Biological Laboratories (Nagoya, Japan).

Cells

Twenty milliliters of blood were drawn on citrate and immediately processed at room temperature at three time points: (1) within 72 h after the beginning of severe sepsis (T1), (2) between the 7th and 10th days after the onset of severe sepsis (T2), and (3) at hospital discharge (T3). Human peripheral blood mononuclear cells (PBMC) were isolated by Ficoll gradient centrifugation. Cell death (by either necrosis or apoptosis) was studied on PBMC immediately after Ficoll gradient isolation to avoid any activation of programmed cell death in vitro. Alternatively, cells were cultured in RPMI-1640 supplemented with 10% fetal calf serum, 2 mM glutamine and 25 mM HEPES (complete medium) in a humidified atmosphere containing 95% air and 5% CO₂ at 37° C. Monocytes were isolated by adherence after a 45-min incubation at 37° C, followed by one PBS wash. m, Hsp70, and Bcl-2 expression were analyzed on adherent cells as purified after a 45-min incubation at 37° C, followed by one PBS wash. As positive controls for mitochondrial membrane depolarization, cells were exposed to hydrogen peroxide (H₂O₂) (4 mM for 4 h at 37° C)(5). Heat shock (44° C for 30 min with a 3-h recovery at 37° C) (5), and tobacco smoke extract (0.48 puff/ml, 3-h incubation at 37° C) (18) exposures were used as positive controls for Hsp70 and Bcl-2 overexpression, respectively.

Determination of Mitochondrial Membrane Potential (m)

m was measured by using the lipophilic cation JC-1, which selectively enters mitochondria. JC-1 exists in a monomeric form emitting at 527 nm after excitation at 490 nm. Depending on m, JC-1 forms J-aggregates that are associated with a large shift in emission (590 nm). Color dye changes reversibly from orange to green as mitochondrial membranes become depolarized. For staining, cell suspensions were adjusted to a density of 0.5 × 10⁶ cells/ml and incubated in complete medium with JC-1 (10 µg/ml) for 10 minutes at 37° C in the dark. Cells were then washed in PBS, resuspended in a total volume of 400 µl, and immediately analyzed by flow cytometry (EPICS Elite flow cytometer; Coulter, Miami, FL) equipped with a single 488-nm argon. A total of 5,000 cells were analyzed for green fluorescence with a 525-nm filter and for orange fluorescence with a 575-nm filter. All data were analyzed with Elite software version 4.02. Mitochondrial membrane disrupture, which goes along with a lower m, was always associated with an increase in percent cells with depolarized mitochondria (%m), i.e., the measurement determined for comparisons used in this study (19, 20).

Flow Cytometric Analysis of Cell Death by Apoptosis and Necrosis

Apoptosis was detected using annexin-V, which has high affinity for negatively charged phospholipids, such as phosphatidyl serine, that externalize towards the outer membrane upon programmed cell death. The simultaneous use of DNA staining by propidium iodide, which is excluded from both intact and apoptotic cells, allows for adequate detection of necrotic cells among the annexin-V positive cluster. Cells were washed and stained with annexin-V and propidium iodide in HEPES buffer as described by the manufacturer, and analyzed by flow cytometry. In all cases, a total of 5,000 cells/sample were analyzed in list mode for green fluorescence through a 525-nm filter and for red fluorescence through a 575-nm filter. Cells exclusively positive for annexin-V were considered as undergoing apoptosis, whereas cells positive for both propidium iodide and annexin-V were considered as necrotic (18, 21).

Soluble Fas Ligand Analysis

Soluble Fas ligand was measured using an enzyme-linked immunosorbent assay (ELISA) according to the manufacturer in plasma kept at 80° C before analysis.

Analysis of Hsp70 and Bcl-2 Expression

Hsp70 expression was quantified by flow cytometry analysis as described (22). Briefly, a pellet of 10⁶ monocytes was fixed in 100 µl of 3% paraformaldehyde in PBS, kept for 10 min at room temperature, then washed by adding 1 ml of PBS containing 1% of BSA. For labeling, cells were permeabilized with 50 µl of 0.6% saponin and incubated with the antibody against the cytosolic inducible form of Hsp70 at a dilution of 1/100 in PBS/BSA for 10 min at room temperature. Unbound antibodies were removed by washing twice in PBS/BSA, and bound antibodies revealed using antimouse IgG-FITC conjugate diluted at 1/ 30 in PBS/BSA for 10 min at room temperature. Analysis was performed by flow cytometry with an EPICS Elite flow cytometer. A total of 5,000 cells were analyzed for green fluorescence through a 525-nm filter. Monocytes Bcl-2 expression was similarly quantified by flow cytometry in fixed and permeabilized cells after staining for 10 min at room temperature with an antihuman Bcl-2-FITC antibody (18).

Statistical Analysis

All values are expressed as means ± SEM. The data were analyzed using a nonparametric analysis of variance (Kruskal-Wallis) followed by a Mann-Whitney U test (Intercooled Stata 5.0, College Station, TX). p < 0.05 was the criterion for statistical significance. The relationships between two continuous variables were analyzed using Spearman's rank correlation tests.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Characteristics

Eighteen patients fulfilling criteria of severe sepsis were included in our study over a 6-mo period. There were eight women and 10 men with a mean age of 62.9 ± 4.3 yr, a SAPS II score of 50.9 ± 3.8, and a LOD score of 6.8 ± 0.6. The etiologies of sepsis were pneumonia (n = 6), peritonitis or digestive tract infection (n = 6), urinary tract infection (n = 2), mediastinitis (n = 1), central venous line infection (n = 1), osteomyelitis (n = 1), and cerebral malaria (n = 1) (Table 1). Seven patients died during their hospital stay, five after the first blood sampling and two after the second one. The SAPS II score was not significantly higher in the survivors than in the nonsurvivors (49.7 ± 4.1 versus 59.3 ± 6.7, respectively, NS), whereas the LOD score was significantly higher in the nonsurvivor group (5.9 ± 0.7 versus 8.7 ± 0.8, respectively, p = 0.03). Seventeen healthy volunteers from the health care donors (six male and nine female, 32 ± 2 yr) served as control group.

Mitochondrial Membrane Potential (m)

When measured at T1, septic patients exhibited an increased percentage of monocytes with depolarized mitochondria (%m) when compared with healthy control subjects (Figures 1 and 2A, p < 0.05). One representative experiment that illustrates the shift in fluorescence of JC-1 in a septic patient (A) relative to a control subject (B) is shown in Figure 1. The decrease m in septic patients leads to an increase in %m (i.e., percentages shown in right corners of low panels of Figure 1). The %m further increased at T2 (p < 0.01) (Figure 2A) returning to control values by hospital discharge (T3). Furthermore, among septic patients, %m was significantly higher in nonsurvivors than in survivors at T1 (26.4 ± 5.3% versus 10.1 ± 2.7%, respectively, p < 0.01) (Figure 2B). We found no significant correlation between the age of septic patients (ranging from 18 to 92 yr) and %m (r = 0.04, p = 0.8). Furthermore age of survivors of sepsis was similar to the entire septic group's age on entry (62.9 ± 4.3 versus 61.3 ± 4.1 yr, respectively), strongly suggesting that age is not an independent determinant of %m.

View larger version (18K): [in this window] [in a new window]   Figure 1.   Cytofluorometric analysis of mitochondrial membrane potential (m). One representative analysis of m in monocytes stained by JC-1 from a healthy volunteer (A) and a septic patient (B). Respective percentage of cells with depolarized mitochondria are indicated in the lower right corner of a control subject and a septic patient measured at the same time (4.8 versus 45%, respectively). Thus, a decrease in m corresponds to an increase in percent of monocytes with depolarized mitochondria (%m).

View larger version (23K): [in this window] [in a new window]   Figure 2.   Mitochondrial membrane potential (m). Cytofluorometric analysis of m in control subjects (n = 17), and in septic patients at different time points; T1 within 72 h after the beginning of severe sepsis (n = 18), T2 within 7 to 10 d after the onset of severe sepsis (n = 13), and T3 at discharge of the survivors (n = 11) (A). We observed an increase of percent cells with depolarized mitochondria (%m) in septic patients at both first time points (T1 and T2). However, values in septic patients returned to control values at hospital discharge (T3). Furthermore, there was a significant increase in %m within the first 72 h in nonsurvivors (n = 11) as compared with survivors (n = 7) (B). * p < 0.05, # p < 0.01.

Cell Death

Percent cell death (apoptosis or necrosis) was significantly increased in patients at T1 when compared with control subjects (Table 2). However, the increase was slight and only became statistically significant when both necrosis and apoptosis were pooled as cell death. In addition, no differences in percent cell death was observed at T2 or T3 as compared with control subjects. The percent of necrotic or apoptotic cells did not change over time between survivors (S) and nonsurvivors (NS) (Table 2). Finally, there was no significant correlation between %m and percent cell death (data not shown).

Soluble Fas Ligand

Soluble Fas ligand levels were below the detection threshold (i.e., < 50 pg/ml) in most of the septic patients and control subjects. Mean levels of Fas ligand in septic patients were 144 ± 24.6 pg/ml at T1 (n = 4), 87 pg/ml at T2 (n = 1), and 69 ± 6.8 pg/ml at T3 (n = 3) and 83 ± 6.5 pg/ml in control subjects (n = 5). There was no statistical difference in Fas ligand levels in the septic patients either between the different time points or between survivors and nonsurvivors.

Hsp 70 Expression

The expression of Hsp70 was significantly increased in septic patients at T1, but not at T2 or T3, as assessed by flow cytometry both as percent cells expressing Hsp70 or as mean fluorescence intensity (Figure 3A and B). However, in contrast to %m, no difference in Hsp70 expression was found between survivors and nonsurvivors at T1 (Figure 3C and D). Hsp70 expression did not correlate with m (data not shown).

View larger version (39K): [in this window] [in a new window]   Figure 3.   Hsp70 Expression. Cytofluorometric analysis of Hsp70 in control subjects (n = 17), and in septic patients at different time points; T1 within 72 h after the beginning of severe sepsis (n = 18), T2 within 7 to 10 d after the onset of severe sepsis (n = 13), and T3 at discharge of the survivors (n = 11). Basal Hsp70 expression of human monocytes obtained from healthy volunteers (Control ) were studied as well as after heat shock (44° C, 30 min, 3 h) as positive control (Control +). There was a significant increase of Hsp70 expression at the first time point that was not observed anymore at the second and third ones ( panels A and B). However, we did not find any difference between survivors (n = 11) and nonsurvivors (n = 7) ( panels C and D). Values are expressed as means ± SE either in percentage or in mean fluorescence intensity (MFI). * p < 0.05, # p < 0.01.

Bcl-2 Expression

There was no difference in Bcl-2 expression between septic patients and control subjects (Figure 4). However, survivors had a significantly higher expression of Bcl-2 than did nonsurvivors at T1. Because of the protective effects of Bcl-2 on mitochondria, we would expect that a decrease of Bcl-2 would be correlated with an increased %m. However, no significant correlation was observed between these variables (data not shown).

View larger version (36K): [in this window] [in a new window]   Figure 4.   Bcl-2 Expression. Cytofluorometric analysis of Bcl-2 in control subjects (n = 17), and in septic patients at different time points; T1 within 72 h after the beginning of severe sepsis (n = 18), T2 within 7 to 10 d after the onset of severe sepsis (n = 13), and T3 at discharge of the survivors (n = 11). Basal Bcl-2 expression of human monocytes obtained from healthy volunteers (control ) were studied as well as after exposure of tobacco smoke extract (0.48 puff/ml, 3-h incubation, 37° C) as positive control (Control +). There was no difference of Bcl-2 expression at any time point studied in septic patients as compared with healthy control subjects ( panels A and B). However, among septic patients, we found a significant decrease in Bcl-2 expression in nonsurvivors (n = 7) compared with survivors (n = 11) ( panels C and D). Values are expressed as means ± SE either in percentage or in mean fluorescence intensity (MFI). * p < 0.05, # p < 0.01.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our data show that in severe human sepsis, peripheral blood monocytes display significant alterations in mitochondrial function as attested by m decrease. Previous studies in severe sepsis have documented mitochondrial dysfunction in parenchymal cells, a dysfunction that has been suggested to be a factor in the development of multiple organ dysfunction syndrome. Our study supports and extends these findings to include the circulating immune effector cells. Indeed, we found a decreased m not only in monocytes from septic patients as compared with control subjects, but also in monocytes from nonsurvivor septic patients as compared with survivors. We also observed a slight increase in percent of cell death in the monocyte population from septic patients as compared with healthy control subjects. Although Hsp70 was increased in monocytes from septic patients, its expression was not different between survivors and nonsurvivors. In contrast, the anti-apoptotic molecule, Bcl-2, although not increased in septic patients, was significantly higher in nonsurvivors as compared with survivors.

Monocyte function plays a key role in sepsis. We previously established the in vitro effects of cellular stress on mitochondrial function, cell death, Hsp70, and Bcl-2 expression (5,18). Cell death may result from multiple insults and occurs along distinct pathways leading either to apoptosis or to necrosis (23). In septic animals, apoptosis has been observed in lymphocytes (24), parenchymal cells, including intestinal and lung epithelial cells, and vascular endothelial cells. Apoptosis has also been reported in autopsy liver specimens from septic patients (27).

Mitochondrial Function in Circulating Monocytes

Several investigators have documented uncoupled hepatic or muscle mitochondrial respiration in endotoxemic animals (8, 9, 12), although not confirmed by others (28, 29). More recently, Simonson and coworkers (10) reported an early inhibition of the intramitochondrial cytochrome a,a₃ along with morphologic mitochondrial changes in a baboon model of sepsis induced by infusion of Escherichia coli. Disruption of mitochondrial membrane has been reported to be an early and irreversible event leading to apoptosis (4), along with the opening of permeability transition pores in the mitochondrial inner membrane and the subsequent release of mitochondrial intermembrane proteins (cytochrome c, apoptosis-inducing factor) into the cytosol and activation of endogenous cysteine proteases (4). However, m alterations may also lead to ATP synthesis arrest, and maintenance of ATP levels appears as the main determinant leading cells to undergo either apoptosis or necrosis (6). Apoptosis does not occur in cells fully depleted in ATP, since ATP is needed to execute the apoptotic program, whereas a severe drop in ATP should lead to necrotic cell death. It is therefore likely that m alterations may be linked to either apoptosis or necrosis depending on the intensity of the effect. Our data support these hypotheses and extend them into the field of human sepsis. Septic patients have evidence of marked decrease of m at the time of presentation. Importantly, the m defect observed in septic patients was greater in nonsurvivors of sepsis than in survivors, although both groups were outwardly similar, as defined by their SAPS II score. That this energy failure does not persist in septic patients is suggested by the return of m to normal levels in survivors by the time they are discharged from the intensive care unit.

Cell Death: Necrosis and Apoptosis

Necrosis is accompanied by an intense inflammatory reaction, whereas during apoptosis, the potentially damaging inflammatory response is limited (23). Thus, apoptosis may contribute to downregulate inflammation in sepsis (30). Peripheral blood monocyte cell death by necrosis and apoptosis was slightly increased in our septic patient sample within the first 72 h, but not at later time points. This increase was only statistically significant when apoptotic and necrotic cells were grouped, suggesting that mitochondrial dysfunction in septic patients does not specifically lead to enough apoptotic cell death to be significant. However, apoptosis is not an all or nothing phenomenon, but one of progression to an irreversible point.

The overwhelming release of proinflammatory and anti-inflammatory cytokines in severe sepsis may also modulate programmed cell death. Alternatively, apoptosis might be much more important in tissue than in circulating cells in human sepsis with systemic organ dysfunction (25). The potential beneficial or adverse effects of apoptosis on circulating immune effector cells in sepsis are unknown and could represent a double-edged sword. Apoptosis could be beneficial by eliminating monocytes that produce proinflammatory cytokines; however, it could also be detrimental in sepsis by compromising host defense against microorganisms.

Several methodologic limitations in this study may make the connections between cell injury and cell death less clear than may be the case. Potentially, many of the cells assayed with impaired m would have displayed evidence of early apoptosis if other intracellular markers of apoptosis were assayed. Our measurements of apoptotic cell death might also be underestimated late apoptosis since monocytes exhibiting both positive PI and annexin labeling were considered as necrotic, but may correspond to "late" apoptotic cells in which plasma membrane integrity is compromised. Finally, we cannot exclude the idea that Ficoll gradient preparation, used to separate mononuclear cells from red blood and polymorphonuclear cells, might have eliminated necrotic cells. One potential explanation for our identifying only a small increase in cell death is that necrotic and apoptotic cells are actively and rapidly removed by the host reticuloendothelial system (23). If this is true, then only early markers of the apoptotic cascade such as the loss of m (4), occurring before other characteristic features of apoptosis, would be observed in circulating cells from patients with sepsis. This may indeed explain why in our study the decrease in m was more pronounced than the increase in cell death in circulating monocytes.

Soluble Fas Ligand

Fas ligand (FasL or CD95L, a member of the tumor necrosis factor superfamily) interacts with its receptor, Fas (APO-1 or CD95, a member of the tumor necrosis factor receptor superfamily), to regulate immune cell activation, differentiation, proliferation, and survival (31). This pathway has also been proposed to be involved in sepsis-induced apoptosis (32). Apoptosis induced by sepsis (cecal ligation and puncture) has been described to be significantly lower in Fas ligand-deficient mice and was associated with a lower mortality rate (32). However, this remains controversial since other investigators have not observed this beneficial effect (33). In our study we did not find any difference in plasma Fas ligand levels at any time points or between survivors and nonsurvivors. These results are in agreement with a preliminary report confirming the lack of increase of sFas ligand (34), though another study has shown an increase of soluble Fas (35). Further studies are required to evaluate the role of Fas in severe human sepsis.

Hsp70 and Bcl-2 Modulation in Human Sepsis

Hsp70 plays a major role in the regulation of apoptosis (36- 38). Overexpression of Hsp70 protects cells and whole animals against oxidative injury through a mechanism involving mitochondria (5, 38). Heat shock-induced Hsp70 overexpression attenuates lipopolysaccharide (LPS)-mediated apoptosis in cultured sheep pulmonary artery endothelial cells (37). Among the antiapoptotic molecules, both Hsp70 and Bcl-2 appear of particular interest in the context of sepsis (13, 38). Although we found an increase in Hsp70 expression at T1 in our septic cohort, this increase was not sustained (7 to 10 d) despite the fact that most of the patients still exhibited sepsis symptoms. As already described by others (39), we did not find any difference in the expression of Hsp70 between survivors and nonsurvivors. Hotchkiss and colleagues (40) and Ryan and coworkers (41) provided the first demonstrations of heat shock response-mediated protection against sepsis. Subsequent studies showed that prior induction of the heat shock response, by either whole body hyperthermia (heat shock) or intravenous sodium arsenite, protected rats against the lethal effects of sepsis (42, 43). Although not defining the mechanism of protection, these studies raise the intriguing possibility that the heat shock response may be an important endogenous defense mechanism against sepsis-induced injury (38). Hsp70, as well as other molecular chaperones, participates in protein integrity and in cytoprotection.

Bcl-2 represents another mechanism involved in the regulation of apoptosis in sepsis. Bcl-2 protects cells from apoptosis induced by different stimuli such as anti-Fas antibody, both in vitro and in vivo (44). Bcl-2-related proteins prevent mitochondrial dysfunction and apoptosis (45, 46). In pigs, the endotoxic shock-mediated liver and spleen cell apoptosis was associated with a decrease in Bcl-2 expression (47). Finally, Hotchkiss and colleagues (13) showed that overexpression of Bcl-2 in transgenic mice decreases apoptosis and improves survival in a sepsis model. It is also known that Bcl-2 prevents apoptotic mitochondrial dysfunction by regulating proton flux (48) or mitochondrial transition pore permeability (46). Our observation that nonsurvivors had lower Bcl-2 expression which did not correlate with %m cannot be explained. Potentially, this may represent a statistical power issue, as the trend was there, but may require a larger number of subjects to reach significance.

We conclude that mitochondrial dysfunction commonly occurs in circulating monocytes of septic patients, and is associated with both increased cell death and Hsp70 expression. However, in the cohort of septic patients, overexpression of Bcl-2 is associated with a better outcome. Therapeutic interventions targeting cell death or using direct measures of mitochondrial membrane function may open new possibilities to the diagnosis and treatment of severe human sepsis (49).