Chonic and acute ethanol exposure inhibits a variety of immunological parameters and decrease resistance to infection (Bagby et al., 2006; Brown et al., 2006). Suppression of inflammatory responses by acute exposure to ethanol is particularly striking in both humans (Gluckman and MacGregor, 1978; Szabo et al., 1999) and animal models (Fitzgerald et al., 2007; Pruett et al., 2004b). The model used in the present study is intended to represent sepsis caused by microbial contamination of the peritoneal cavity (e.g., following penetrating abdominal trauma, ruptured appendix, cirrhosis of the liver, or other conditions that interfere with the normal barrier function of the mucosa). Increased rates of infection have been reported in patients with penetrating abdominal trauma following acute ethanol exposure (Gentilello et al., 1993). Similar results have been reported in burn patients (Germann et al., 1997). Sepsis caused by indigenous bacteria may begin as a polymicrobial infection, but in most cases. only one bacterium is isolated, and Escherichia coli is one of the most frequent (Huttunen et al., 2007). Decreased resistance to infection has also been reported in animal models (D'Souza et al., 1995; Pruett et al., 2004b), but this has not been investigated in as much detail as inhibition of production of particular inflammatory mediators. Studies in which inhibition of inflammatory responses has been implicated as a cause of decreased resistance to microbes have been relatively rare. However, such studies can potentially reveal much about the normal mechanisms by which inflammation promotes bacterial clearance and host survival because the effects of inhibiting a number of those mechanisms simultaneously are revealed by ethanol. In this instance, inhibiting production of several mediators at the same time is beneficial because it will provide an indication of a set of mediators that contribute to lethality. In the present study, evidence is presented indicating that mice treated with ethanol exhibit suppression of several mediators and processes of inflammation early, which is followed by an overgrowth of bacteria and possibly a lethal systemic inflammatory response.

Responses triggered though TLR4 and other Toll-like receptors are inhibited by acute ethanol exposure (Goral and Kovacs, 2005; Pruett et al., 2004b), and many investigators (including the present authors) have suggested that this contributes to decreased resistance to infection. To test this assumption, effects of ethanol in wild-type mice were compared with its effects in mice lacking MyD88, myeloperoxidase (MPO), or with a mutant hyporesponsive TLR4. The results unexpectedly demonstrated that inhibition of TLR4 signaling by ethanol is probably not a major cause of decreased resistance to sepsis in the experimental system used here.

Results shown in Figure 2 indicate the number of viable E. coli isolated at various times after challenge from the peritoneal cavity of control and ethanol-treated mice. The results demonstrate that after a brief increase in number, bacteria are progressively cleared over time in control mice. In contrast, the bacteria continue to grow though 21 h in ethanol-treated mice. The SEM increases at the later time points in ethanol-treated mice. This very likely reflects the fact that some of the mice in the ethanol-treated group will survive (as in Fig. 1). Survivors may have begun clearing bacteria by 21 h, whereas nonsurvivors would not be expected to do this.

As we reported previously, most of the cells in the peritoneal cavity of normal B6C3F1 female mice are macrophages (Fig. 3A). In contrast, 18 h after challenge with E. coli, the predominant cell type was the neutrophil in control mice and very few bacteria remained (Fig. 3B). In contrast, bacteria were remarkably abundant in mice treated with ethanol before E. coli challenge (Fig. 3C) due to either increased phagocytosis or decreased clearance of the bacteria. There was also a lower percentage of neutrophils than in the control group (shown in detail in Fig. 4), indicating impaired chemotaxis or migration of neutrophils to the initial site of infection or increased death of neutrophils.

Although the numbers of bacteria phagocytosed by each macrophage or neutrophil could not be determined precisely by the type of microscopic analysis done here, quantifying the overall process of phagocytosis and clearance was accomplished by discriminating between neutrophils/macrophages with associated bacteria and the ones without bacteria. Because it was possible that some of the bacteria observed were associated with phagocytes on the cell surface and not phagocytosed, a criterion of three bacteria per cell was used to classify a cell as positive for bacteria. Cells with less than three associated bacteria were classified as negative. This is not intended as precise quantitation of phagocytosis but was done just to give a general impression of how the phagocytosis of bacteria by macrophages/neutrophils after E. coli challenge is affected by ethanol. Results of such an analysis are shown in Figure 4. The percentage of macrophages with bacteria increased from 1 to 4 h in control mice, probably reflecting continued phagocytosis. After 4 h, the percentage of macrophages with bacteria decreased in control mice as bacteria were killed and digested by the phagocytic cells. Ethanol increased the percentage of macrophages with bacteria throughout the time course. At the early time points (1–4 h), this could be due to increased phagocytosis or decreased clearance because the number of cells with bacteria depends on the rate of phagocytosis and the rate of clearance. However, at later times, it seems unlikely that enhanced phagocytosis could explain the increased number of bacteria because bacteria had mostly been phagocytosed by then in the ethanol-only group. Thus, it seems more likely that the increased number of cells with bacteria (which corresponds with the increased bacterial number overall shown in Fig. 2) is due to macrophage and/or neutrophil dysfunction leading to inefficient bacterial killing. In control mice, the percentage of neutrophils in the peritoneal cavity increases over time, but this increase was inhibited by ethanol, which is consistent with inhibition of some of the chemokines that attract neutrophils (see next paragraph).

The effects of ethanol on cytokine and chemokine production at various times after E. coli challenge are indicated in Figure 5. Most of the cytokines and chemokines tested here were significantly decreased (all but interleukin [IL]-10 and IL-1β) at one or more time points by one dose of ethanol 30 min before challenge with E. coli. It is particularly interesting that for IL-1β and IL-6, two cytokines strongly associated with inflammation and acute phase responses, the concentration was increased at 21 h, even if it was significantly decreased at earlier times. This is consistent with the much greater bacterial load in the ethanol-treated mice at 21 h than at any other time for any other group (Fig. 2). It is not clear why this was not the case for other cytokines, but there may be a regulatory mechanism that prevents excessive production of these cytokines. The results are also consistent with the known sequence of cytokine production with peak concentrations of tumor necrosis factor-α produced early, interferon (IFN)-γ considerably later, and most others at an intermediate time.

To examine the role of LPS acting though TLR4 and the role of other TLRs in this system, mutant and transgenic mice were used. The C3H/HeJ mouse strain has a mutation in TLR4 that renders it hyporesponsive to LPS (Cross et al., 1995). Therefore, this strain was used to evaluate the role of TLR4 in this experimental system. The adapter molecule MyD88 can initiate TLR signaling for all TLRs except TLR3 (Johnson et al., 2008). Thus, MyD88-knockout mice were used to assess the role of TLRs in general (except TLR3) in this experimental system. The results shown in Figure 6 demonstrate that TLR4 is required for lethality but apparently is dispensable for survival (and presumably for bacterial clearance). Ethanol decreased survival in both TLR4-mutant C3H/HeJ mice and the corresponding wild-type strain. This suggests that the inhibition of TLR4 signaling by ethanol is probably not a major cause of increased mortality because C3H/HeJ mice, which have even less TLR signaling though LPS than ethanol-treated C3H/HeOuJ mice, survive better than C3H/HeOuJ mice with or without ethanol treatment. Survival of C3H/HeOuJ mice was less than noted in Figure 1 for B6C3F1 mice. This is probably due to genetic differences between resistance-related genes in the C3H and C57Bl/6 (the wild-type mice for MyD88 knockouts and MPO knockouts were C57Bl/6). This suggests that complementary genes enhance survival in the B6C3F1 hybrid and that resistance is less effective in C57Bl/6 mice. Absence of MPO did not significantly decrease survival but caused a nonsignificant increase in survival in control groups (Fig. 6). Ethanol significantly and similarly decreased survival in both wild-type and MPO-knockout mice, suggesting that MPO is not critical for survival in this model and that MPO does not have a major role in the inhibition of survival caused by ethanol.

The results shown in Figure 7 indicate that LPS is a major contributor to the induction of cytokines and chemokines by E. coli. The production of cytokines and chemokines was ~2- to 15-fold greater in C3H/HeOuJ mice than in the TLR4-mutant C3H/HeJ mice. Thus, the poor response of C3H/HeJ mice to LPS substantially decreases the response of these mice with regard to all tested cytokines and chemokines, even though gram-negative bacteria have components that can activate many receptors in addition to TLR4. The results shown in Figure 7 also indicate that ethanol alone (at 6 g/kg) does not induce these cytokines or chemokines to concentrations significantly greater than found in control mice and that ethanol at 4 g/kg significantly modulates production of most cytokines and chemokines. This is of interest because, in this mouse model for binge drinking, a dosage of 4 g/kg produces a peak blood ethanol concentration of ~65mM (Carson and Pruett, 1996), a concentration not uncommon in humans (Jones and Holmgren, 2009; Urso et al., 1981).

Some portion of the inhibition of production of these inflammatory cytokines by ethanol could be mediated by stress hormones such as corticosterone that can be induced by ethanol. However, our recent studies indicate that ethanol-induced glucocorticoids and catecholamines do not contribute to the inhibition of cytokine production induced through TLR3 or TLR4, with the possible exception of a partial role in inhibition of IL-6 production (Glover and Pruett, 2006; Glover et al., 2009). Thus, it seems unlikely that the stress response is critical to the effects of ethanol on the pathogenesis of sepsis in this model system.

Several investigators have reported that ethanol inhibits TLR signaling (Dai et al., 2005; Szabo et al., 2007), and it seemed likely that this was involved in the decreased resistance to infection associated with acute ethanol exposure. However, the results presented here indicate that survival was enhanced in the absence of fully functional TLR4, so inhibition of TLR4 signaling is unlikely to be the major mechanism by which ethanol suppresses resistance to lethality in this experimental system. This is also suggested by the observation that ethanol decreases survival time and/or percentage in TLR4-knockout mice to a similar degree as in wild-type mice, indicating that targets other than TLR4 are involved in lethality. There are cytoplasmic (Cartwright et al., 2007) as well as membrane-bound receptors that respond to LPS and other TLRs that respond to other components of gram-negative bacteria. It is possible that in the absence of TLR4, these receptors mediate sufficient response to lead to bacterial clearance but not to a lethal overproduction of inflammatory mediators. Our results with regard to cytokine and chemokine production in TLR4-mutant and wild-type mice support this idea (Fig. 7). However, it is also possible that decreased cytokine responses in the first few hours after challenge in mice treated with ethanol allow overgrowth of bacteria which becomes lethal after the ethanol has been eliminated. This lethality may be related to TLR4-induced overproduction of inflammatory mediators, but this cannot be the only mechanism because TLR4-mutant mice treated with ethanol have a similar mortality profile as wild-type mice treated with ethanol. This strongly suggests that lethal effects of sepsis are mediated through other receptors and may not involve well-recognized inflammatory mediators because many of these are decreased in concentration in TLR4-mutant mice. Other receptors that may be involved in the pathogenesis include TLR3 and TLR2. There are reports that TLR3 contributes to lethality in sepsis by sensing the RNA released from necrotic cells during sepsis and amplifies the secondary inflammatory responses (Cavassani et al., 2008). Similarly, TLR2-dependent signaling has been shown to be critical for host resistance to bacterial infection both in animal studies and humans (Alves-Filho et al., 2009; Ferwerda et al., 2009; Mancuso et al., 2004; Murphey et al., 2008). Interestingly, ethanol also suppresses TLR3 and TLR2 signaling as reported in our previous study (Pruett et al., 2004a,c).

It remains possible that inhibition of TLR4 signaling by ethanol (or TLR4 mutation) does play an important role in decreased resistance to lower dosages of bacteria, as reported for C3H/HeJ mice (Alves-Filho et al., 2006), and that clearance of higher dosages of bacteria could be delayed (van Westerloo et al., 2005) by inhibition or lack of TLR4, even if ultimately effective. Results for enteropathogenic E. coli indicate that C3H/HeJ mice do not survive as well as wild-type mice when a low dose of bacteria is administered (Cross et al., 1995). Similar results were obtained by another group when mice were treated with a sublethal challenge dose of bacteria (Alves-Filho et al., 2006). However, when mice were treated with a greater dose of bacteria (lethal for a major percentage of wild-type mice), a much higher percentage of C3H/HeJ mice survived than wild type, as noted in our study. Another group very recently reported similar results, indicating that TLR4-mutant mice have increased resistance to a lethal outcome in E. coli sepsis caused by a high dosage of E. coli (Roger et al., 2009). Thus, it seems that the role of TLR4 in resistance to sepsis and lethality in sepsis depends on the initial challenge dose of bacteria.

In contrast to the clear protective effect of the absence of functional TLR4 in the C3H/HeJ mice, the lack of MyD88 in MyD88-knockout mice did not significantly improve survival (although there was a tendency in that direction). This leaves open the interesting possibility that the lethal effects of sepsis involve signals transmitted through the alternate adaptor molecule used by TLR4 (and by TLR3 as its only adaptor), TIR-domain-containing adaptor inducing interferon-beta (TRIF). Results indicating that mice lacking TRIF survive sepsis better than wild-type mice are consistent with this possibility (Weighardt and Holzmann, 2007). Also consistent with this idea is a recent report indicating that IFN-β (the production of which is mediated by the TRIF pathway) is required for release of high-mobility group box 1 protein and lethality in sepsis (Kim et al., 2009). We previously reported that IL-12 and IL-10 production in response to LPS is decreased almost to the lower limit of detection (Pruett et al., 2005). The finding that MyD88 is dispensable in survival of E. coli infection is consistent with findings in humans, in which individuals with a genetic defect in MyD88 were less resistant only to pyogenic bacterial infections, not other to other types of infections (von Bernuth et al., 2008). In apparent contrast to these findings, Peck-Palmer et al. (2008) reported that MyD88-knockout mice do not survive as well as wild-type mice in a cecal ligation and puncture model of sepsis. Mortality in that study was 40% in wild-type C57Bl/6 mice, whereas it was 100% in our study (Fig. 6). Thus, the difference in results may simply reflect different outcomes with high versus lower dosages of bacteria, as reported by other investigators (Alves-Filho et al., 2006; Mancuso et al., 2004) or differences between polymicrobial sepsis and E. coli sepsis.

The observation that mice lacking MPO were not significantly more susceptible to sepsis-induced mortality than wild-type mice was not entirely unexpected because similar results have been reported previously (Brovkovych et al., 2008). However, the finding that wild-type and MPO-knockout mice are similarly and significantly susceptible to ethanol-induced mortality in sepsis indicates that ethanol does not act primarily by inhibiting expression or function of MPO. We had considered this as a potential mechanism because our microarray analysis indicated that early after E. coli challenge MPO expression was decreased by ethanol (data not shown).

The effects of ethanol on survival were similar for wild-type mice, MyD88-knockout mice, TLR4-mutant mice, and MPO-knockout mice. Significantly decreased survival percentage or survival time was noted in all cases. These results indicate that inhibition of TLR4 signaling through MyD88 (which does occur) is not the major mechanism by which ethanol decreases host resistance. They also demonstrate that inhibition of MPO by ethanol is not a major mechanism for decreased resistance to sepsis in this system.

The cellular targets for decreased antibacterial effectiveness seem to be macrophages and/or neutrophils. Although the results reported here indicate that phagocytosis occurred in ethanol-treated mice (Fig. 3), it was evident that killing and degradation of the bacteria in both macrophages and neutrophils were substantially decreased (Fig. 4). Microarray results using this experimental model indicate inhibition by ethanol of a variety of genes coding for proteins involved in antibacterial effects of phagocytic cells (manuscript in preparation). Others have reported that acute ethanol exposure decreases host resistance by inhibiting upregulation of key antimicrobial effectors (e.g., nitric oxide synthase 2) (Greenberg et al., 1999). Future studies will be conducted to determine which of these are involved in the inhibition of phagocyte antimicrobial function.

There are at least two obvious potential applications suggested by these results for treatment of sepsis in humans. For E. coli-mediated sepsis, inhibition of TLR4 signaling may be a useful therapeutic approach because decreased response to TLR4 in a mutant mouse strain improved survival substantially (Fig. 6). This conclusion is consistent with another recently published report (Roger et al., 2009). This approach has the advantage of decreasing responses of several mediators of inflammation, not just one. Also, greater efficacy of therapy in sepsis might be obtained using treatments that inhibit combinations of the cytokines overproduced in the later stages of sepsis (e.g., IL-1β and IL-6) (Fig. 5) and that augment key mediators that are not known to contribute to lethality at intermediate stages of sepsis (e.g., granulocyte-macrophage colony-stimulating-factor) (Fig. 5). It is clear that this animal model is associated with rapid lethality, which is different from the typical course of sepsis in humans, which lasts 15–17 days, possibly due to antibiotic therapy that is not included in the mouse model (Angus et al., 2001). However, there is evidence that intervention of various types early after the diagnosis of sepsis is more effective than later intervention (Moore et al., 2009). Thus, the patterns noted in the animal model may be useful, but the results reported here would suggest that the timing of intervention is critical and will need to be determined in studies with human subjects.