Prevention of lymphocyte apoptosis in septic mice with cancer increases mortality

Abstract

Lymphocyte apoptosis is thought to play a major role in the pathophysiology of sepsis. However, there is a disconnect between animal models of sepsis and patients with the disease, since the former use subjects that were healthy prior to the onset of infection while most patients have underlying comorbidities. The purpose of this study was to determine whether lymphocyte apoptosis prevention is effective in preventing mortality in septic mice with pre-existing cancer. Mice with lymphocyte Bcl-2 overexpression (Bcl-2-Ig) and wild type (WT) mice were injected with a transplantable pancreatic adenocarcinoma cell line. Three weeks later after development of palpable tumors, all animals received an intratracheal injection of Pseudomonas aeruginosa. Despite having decreased sepsis-induced T and B lymphocyte apoptosis, Bcl-2-Ig mice had markedly increased mortality compared to WT mice following Pseudomonas aeruginosa pneumonia (85% vs. 44% seven-day mortality, p=0.004). The worsened survival in Bcl-2-Ig mice was associated with increases in Th1 cytokines TNF-α and IFN-γ in bronchoalveolar lavage fluid and decreased production of the Th2 cytokine IL-10 in stimulated splenocytes. There were no differences in tumor size or pulmonary pathology between Bcl-2-Ig and WT mice. To verify the mortality difference was not specific to Bcl-2 overexpression, similar experiments were performed in Bim^-/- mice. Septic Bim^-/- mice with cancer also had increased mortality compared to septic WT mice with cancer. These data demonstrate that despite overwhelming evidence that prevention of lymphocyte apoptosis is beneficial in septic hosts without comorbidities, the same strategy worsens survival in mice with cancer that are given pneumonia.

Introduction

Sepsis is the leading cause of death among critically ill patients in the United States with over 200,000 people dying from the disease annually (1). Despite many advances in understanding the pathophysiology of sepsis, mortality remains unacceptably high (2).

Apoptosis is theorized to play a critical role in the pathophysiology of sepsis (3). Human autopsy studies of sepsis demonstrate increased apoptosis in the spleen and the intestinal epithelium (4). In addition, apoptosis in circulating lymphocytes is markedly increased in septic patients (5-7), and this is associated with poor outcome (8). Animal models of sepsis replicate these findings of increased sepsis-induced lymphocyte and intestinal epithelial apoptosis (9-14). The functional significance of this is demonstrated in animal models (predominantly peritonitis-induced sepsis) demonstrating that prevention of apoptosis in lymphocytes, globally using knockout mice or siRNA or in the intestinal epithelium improves survival following sepsis (15-35). Since apoptosis prevention has been repeatedly successful in improving survival when targeting a wide variety of mediators by a number of investigative groups, there is significant interest in translating these findings to the bedside (36-39).

There has been a longstanding disconnect between animal models of sepsis and therapeutic trials in patients (40). While there are complex reasons why positive preclinical trials have not successfully translated into therapeutic benefit at the bedside, one possibility is the populations studied are different. Typical animal models use mice that were healthy prior to the onset of sepsis. However, the majority of septic patients have one or more pre-existing comorbidities (1). Both patients and animals subjected to a septic insult have increased mortality in the setting of additional comorbidities (41-44). This is consistent with a “two-hit” model of injury, where a chronic comorbidity is the first insult and an acute septic injury represents the second insult. While each of these “hits” independently confers some risk, their combined effects are disproportionately harmful over what might have been predicted from either in isolation.

Cancer is one of the most common comorbidities that can afflict septic patients. It is also associated with a high rate of mortality, with approximately 40% of septic patients with cancer dying from the disease (1,45). In addition, patients with malignancy are nearly ten times more likely to develop sepsis than the general population (46). The factors that affect an individual's susceptibility to developing sepsis may include tumor type, tumor size, presence of metastatic disease, and host immunological response. Our lab recently described an animal model of sepsis and cancer, in which mice that received a transplantable pancreatic adenocarcinoma cell line three weeks prior to the onset of Pseudomonas aeruginosa pneumonia had a 24% increase in mortality compared to septic mice that were previously healthy (44).

In light of a) the extensive literature demonstrating a survival benefit to preventing lymphocyte apoptosis in previously healthy mice and b) the knowledge that mice with cancer (or other comorbidities) behave differently than previously healthy mice subjected to the identical septic insult, this study examined whether preventing lymphocyte apoptosis would improve survival in the clinically relevant model of sepsis in the setting of cancer.

Materials and Methods

Mice

Nine to sixteen week old mice were used for all experiments. Transgenic mice overexpressing human Bcl-2 in both T and B lymphocytes (Bcl-2-Ig mice) were generated as previously described (17,47). Of note, Bcl-2 expression is significantly greater in the spleen than the thymus in these animals. Bim^-/- mice were initially generated at the Walter and Eliza Hall Institute of Medical Research (Melbourne, Victoria, Australia) (48) but were subsequently bred for over three years at Washington University for experiments examining lymphocyte apoptosis and sepsis, where they were a generous gift of Dr. Richard Hotchkiss (21). Transgenic mice that overexpress Bcl-2 in the intestinal epithelium (Fabpl-Bcl-2 mice) were generated on an FVB/N background and were then backcrossed to C57Bl/6 mice for greater than 20 generations (49,50). For all studies in this manuscript, animals were compared to C57Bl/6 littermates or controls bred in the same animal facility. Mice had free access to food and water and were maintained on a 12 hour light-dark schedule in a barrier facility. All studies complied with the National Institutes of Health Guidelines for the Use of Laboratory Animals and were approved by the Animal Studies Committee of both Washington University and Emory University.

Cancer model

The transplantable mouse pancreas adenocarcinoma cell line Pan02 was injected to induce cancer as previously described (51,52). Briefly, Pan02 cells were cultured in RPMI-1640 medium supplemented with 1% glutamine, 1% 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, 10% fetal bovine serum, and 1% penicillin/streptomycin (Cellgro, Herndon, VA). A total of 250,000 Pan02 cells were injected subcutaneously into each animal's right inner thigh. Mice were then housed for three weeks in a barrier facility to allow tumors to grow prior to the induction of sepsis. Animals given Pan02 cells develop well-defined, circumscribed tumors at the injection site, but the tumors are not metastatic and do not cause mortality in the timecourse studied in this set of experiments. The longest axis of the tumor was used for measuring tumor size. In a recent study from our lab (44), animals given Pan02 cells three weeks earlier were compared to healthy mice. Animals had similar body weights, liver function, kidney function, lung histology, wet to dry lung ratios, blood cytokine levels (TNF-α, IL-6, IL-10, IL-12, and MCP-1), BAL cytokines (IL-6, IL-10, IL-12, and MCP-1), gut apoptosis, intestinal villus length, intestinal proliferation, hematocrit, white blood count, absolute neutrophil count, absolute lymphocyte count, and platelet count. In contrast, the presence of cancer caused lower levels of T lymphocyte and B lymphocyte apoptosis compared to healthy mice and additionally induced higher levels of TNF-α in BAL fluid.

Sepsis model

Three weeks after the injection of Pan02 cells, mice were made septic via direct intratracheal injection of Pseudomonas aeruginosa (ATCC 27853) (53). A midline cervical incision was made while under isoflurane anesthesia, and 40μl of a solution of bacteria diluted in 0.9% NaCl (final concentration 6×10⁶ colony-forming units/ml) was injected directly into the trachea using a 29-gauge insulin syringe. Animals were then held vertically for 10 seconds to enhance delivery of bacteria into the lungs. Mice received a single 1 ml subcutaneous injection of 0.9% NaCl to replace insensible fluid losses following incision closure. Animals were either sacrificed 24 hours following induction of pneumonia for tissue harvest or were followed seven days for survival. Of note, mortality is nearly twice as high in mice given Pan02 cells three weeks prior to the onset of Pseudomonas aeruginosa pneumonia compared to previously healthy mice given the identical insult (44). The rationale for studying Pseudomonas aeruginosa pneumonia following cancer as opposed to a more commonly studied model of sepsis such as cecal ligation and puncture was a recent study extensively characterizing the impact of pneumonia superimposed upon cancer (44).

Splenocyte apoptosis

Splenocyte apoptosis was quantified via flow cytometry using antibodies against active caspase-3 (Cell Signaling Technology, Beverly, MA) and via the TUNEL assay (Phoenix Flow Apo-BrdU Kit, San Diego, CA) (21). T- and B- cell populations were identified using fluorescein-labeled anti-mouse CD3 (BD Pharmingen, Franklin Lakes, NJ) and PE-Cy5 conjugated anti-mouse CD45R/B220 (eBioscience, San Diego, CA) respectively. Flow cytometric analysis (50,000 events/sample) was performed on FACScan (BD Biosciences, San Jose, CA) as previously described (54).

Local, systemic, and stimulated cytokine analysis

Bronchoalveolar lavage (BAL) and whole blood were used to assess local and systemic cytokine levels respectively (44). The trachea was lavaged with 1ml of sterile 0.9% NaCl to obtain BAL fluid. Whole blood was collected retro-orbitally and centrifuged at 5000 rpm for 5 minutes in serum separator tubes. BAL and whole blood cytokine concentrations were then evaluated using the Bio-Plex Pro Mouse Cytokine Standard Group I Kit (Biorad, Hercules, CA) according to manufacturer's protocol. All samples were run in duplicate.

Stimulated cytokines were evaluated by harvesting the spleen 24 hours after septic injury. Lymphocytes contained in splenocyte suspensions were cultured in vitro and stimulated with anti-CD3 and anti-CD28 (BD Biosciences, San Jose, CA) as previously described (19). The supernatant was removed 24 hours later and cytokine levels were measured using the Bio-Plex cytokine kit.

Pneumonia severity

Hematoxylin and eosin (H&E)-stained lung sections were evaluated by a pathologist (ABF) blinded to sample identity to determine the severity and distribution of pneumonia. Samples were first graded to determine percent of tissue with inflammation. An inflammation score ranging from 0 to 5 (0 = None, 1 = <5%, 2 = 5-10%, 3 = 10 - 19%, 4 = 20 - 49%, 5 = > 50%) was then calculated. Next, degree of atelectasis was assessed as follows: none, focal representing < 10%, moderate representing 10-50%, and severe representing > 50%. Vascular congestion was then assessed utilizing the same scale as atelectasis. Finally, a subjective analysis of each individual section was performed.

Myeloperoxidase (MPO) activity

Bronchoalveolar lavage (BAL) fluid was obtained after tracheal instillation with 1 ml sterile saline. BAL fluid was collected and centrifuged at 5,000 rpm for 5 minutes. Following addition of substrate buffer containing O-dianisidine and 0.0005% hydrogen peroxide, MPO activity was measured at 470 nm wavelength over 6 minutes (Bio-Tek Instruments-μQuant Microplate Spectrophotometer, Winooski, VT). MPO activity was calculated as ΔOD/minute per μl of BAL fluid.

Bacterial cultures

Blood was collected from the inferior vena cava and BAL fluid was obtained as above. BAL samples were serially diluted in sterile saline and plated on sheep's blood agar plates. Plates were incubated overnight at 37°C and examined for colony counts 24 hours later.

Intestinal epithelial apoptosis

Apoptotic cells were quantified in 100 contiguous well-oriented crypt-villus units using H&E-staining and active caspase-3 staining (53). Apoptotic cells were identified on H&E-stained sections by morphologic criteria. Cells with characteristic nuclear condensation and fragmentation were considered to be apoptotic. For active caspase-3 staining, intestinal sections were deparaffinized, rehydrated, and incubated in 3% hydrogen peroxide for 10 minutes. Antigen retrieval was performed by placing slides in Antigen Decloaker (Biocare Medical, Concord, CA) and heating them in a pressure cooker for 45 minutes. Slides were then blocked with 20% goat serum (Vector Laboratories, Burlingame, CA), and incubated with rabbit polyclonal anti-active caspase-3 (1:100; Cell Signaling Technology, Beverly, MA) overnight at 4°C. The following day, slides were incubated with goat anti-rabbit biotinylated secondary antibody (1:200; Vector Laboratories) for 30 minutes at room temperature followed by Vectastain Elite ABC reagent (Vector Laboratories) for 30 minutes. Slides were developed with diaminobenzidine and counterstained with hematoxylin.

Statistics

All data were tested for Gaussian distribution using the Shapiro-Wilk normality test. If data were found to have a Gaussian distribution, comparisons were performed using the Student's t-test and results are presented as mean ± SEM. If data did not have a Gaussian distribution, comparisons were performed using the Mann Whitney test and presented as median± range. Survival studies were analyzed using the Log-Rank test. Data were analyzed using the statistical software program Prism 4.0 (GraphPad, San Diego, CA). A p value of <0.05 was considered to be statistically significant.

Results

In all experiments, mice were injected with the transplantable pancreatic adenocarcinoma cell line Pan02. Three weeks later, animals were given Pseudomonas aeruginosa pneumonia. As such, all animals developed sepsis in the setting of pre-existing cancer.

Bcl-2 overexpression in lymphocytes increases mortality despite preventing sepsis-induced lymphocyte apoptosis

Transgenic mice overexpressing human Bcl-2 in both T and B lymphocytes (Bcl-2-Ig mice) had decreased levels of T lymphocyte apoptosis compared to WT littermates by both the active caspase-3 staining and TUNEL assay as measured by flow cytometry (Fig. 1A, B). Bcl-2-Ig mice also had decreased levels of B lymphocyte apoptosis (Fig 1C). A similar decrease in apoptosis was also observed on H&E-stained splenic sections (Fig. 1D, E).

Fig. 1.

Effect of Bcl-2 overexpression on lymphocyte apoptosis in septic mice with cancer. T lymphocyte (A, B, n=13-19/group) and B lymphocyte apoptosis (C, n=14-19/group) was decreased in Bcl-2-Ig mice compared to WT mice whether assayed by active caspase 3 staining (A,C) or the TUNEL assay (B) by flow cytometry. Similar results were qualitatively seen on H&E-stained sections where splenic apoptosis was elevated in septic WT mice with cancer (D) but lower in Bcl-2-Ig mice subjected to the same insult (E). Representative histomicrograph is shown at magnification 400x.

A different set of Bcl-2-Ig and WT cancer mice were then followed for survival after the induction of sepsis. Despite the fact that Bcl-2-Ig mice had lower lymphocyte apoptosis, 85% of Bcl-2-Ig mice died 7 days following the onset of sepsis compared to 44% of WT mice (p=0.004, Fig. 2).

Fig. 2.

Effect of Bcl-2 overexpression in lymphocytes on survival in septic mice with cancer. Mortality was nearly twice as high in Bcl-2-Ig mice compared to WT mice 7 days after induction of Pseudomonas aeruginosa pneumonia, n=20-27/group.

Bcl-2 overexpression in lymphocytes in septic mice with cancer induces an upregulation of Th1 cytokines in BAL fluid but not plasma

In order to determine whether the local host response played a role in the elevated mortality in septic Bcl-2-Ig mice, BAL cytokines were measured in transgenic and WT mice (Table 1). Concentrations of the Th1 cytokines TNF-α (p=0.008) and IFN-γ (p=0.03) were significantly higher in Bcl-2-Ig than WT mice.

Table 1. Bronchoalveolar lavage cytokines.

Bronchoalveolar lavage
Cytokine (pg/ml)	WT (n=10)	Bcl-2-Ig (n=11)	p-value
TNF-α	436±48	734±115	*p=0.03
IFN-γ	17±1	43±17	**p=0.008
IL-10	179±54	82±33	p=ns
IL-6	11813±3632	13469±2325	p=ns

In contrast, when plasma cytokines were assayed, no significant differences were detected between septic Bcl-2-Ig and WT mice with cancer (Table 2).

Table 2. Blood cytokines.

Blood
Cytokine (pg/ml)	WT (n=10-11)	Bcl-2-Ig (n=9-10)	p-value
TNF-a	3445±392	2908±352	p=ns
G-CSF	14200±1236	14349±860	p=ns
IL-10	187±41	298±138	p=ns
IL-6	3197±584	3828±701	p=ns

Bcl-2 overexpression in lymphocytes in septic mice with cancer decreases production of IL-10 in stimulated lymphocytes

Lymphocytes isolated from the spleens of septic Bcl-2-Ig mice that were incubated ex vivo with anti-CD3/28 had lower production of the Th2 cytokine IL-10 compared to WT littermates (p=0.004, Table 3). Other stimulated cytokines were similar in septic mice with cancer, regardless of whether they overexpressed Bcl-2 in their lymphocytes.

Table 3. Stimulated splenocyte cytokines.

Stimulated Splenocytes
Cytokine (pg/ml)	WT (n=9)	Bcl-2-Ig (n=9)	p-value
IL-10	187±15	117±15	**p=0.004
TNF-α	56±4	52±5	p=ns
IFN-γ	10492±1375	12493±1364	p=ns
IL-6	1922±189	1509±209	p=ns

Bcl-2 overexpression in lymphocytes does not alter tumor size

In order to determine whether overexpression of an anti-apoptotic protein in lymphocytes caused a secondary effect on cancer growth, tumor size was measured in Bcl-2-Ig and WT mice. No difference in tumor size was identified, regardless of whether an animal overexpressed Bcl-2 in its lymphocytes (Fig. 3).

Fig. 3.

Effect of Bcl-2 overexpression in lymphocytes on tumor size. Bcl-2-Ig mice and WT mice were both injected with 250,000 Pan02 cells and then given sepsis with Pseudomonas aeruginosa three weeks later. Overexpression of Bcl-2 in lymphocytes had no affect on the growth of tumors, n=9-10/group.

Bcl-2 overexpression in lymphocytes does not alter pneumonia severity

In light of the differences identified in BAL cytokines, pneumonia severity was assessed to determine whether alterations in pulmonary pathology were responsible for the mortality difference between Bcl-2-Ig and WT mice. Both pneumonia severity and percentage of lung tissue with inflammation was similar between Bcl-2-Ig and WT mice (Fig. 4A,B). No differences were detected between the groups in atelectasis (ranging from none to moderate in Bcl-2-Ig mice, ranging from focal to moderate in WT mice) or vascular congestion (focal to moderate in all animals, without differences between groups). Additionally, there were no differences in neutrophil infiltration and degranulation in BAL fluid as measured by myeloperoxidase (MPO) activity between Bcl-2-Ig and WT mice (Fig. 4C). Bacteria were not detectable in BAL fluid 24 hours after tracheal instillation of P. aeruginosa (n=5-6 mice/group).

Fig. 4.

Effect of Bcl-2 overexpression on pulmonary pathology. Bcl-2-Ig mice and WT mice were both injected with 250,000 Pan02 cells and then given sepsis with Pseudomonas aeruginosa three weeks later. Overexpression of Bcl-2 in lymphocytes had no affect on pneumonia severity (A) or percentage of lung tissue with inflammation (B). MPO activity in BAL fluid was also unaffected (C), n= 6/group.

Septic Bim^-/- mice with cancer have increased mortality compared to WT mice

In order to determine whether this finding was specific to Bcl-2-Ig mice or potentially more generalizable, survival studies were repeated in Bim^-/- mice that were injected with Pan02 cells as above and made septic via Pseudomonas aeruginosa pneumonia 3 weeks later. Septic Bim^-/- mice with cancer had an 18% increase in mortality compared to WT mice with cancer (p=0.046, Fig. 5).

Fig. 5.

Effect of knocking out the pro-apoptotic factor Bim on survival in septic mice with cancer. Mortality was higher in Bim^-/- mice than WT mice (n=11-14/group) 7 days after induction of Pseudomonas aeruginosa pneumonia.

Bcl-2 overexpression in the intestinal epithelium has no effect on mortality

To determine if the effect of altering sepsis-induced apoptosis was lymphocyte specific, a survival study was performed on transgenic mice that overexpress Bcl-2 in the intestinal epithelium (Fabpl-Bcl-2 mice). There was no difference in survival between septic Fabpl-Bcl-2 mice with cancer and septic WT mice with cancer (Fig. 6A). However, unlike either septic Bcl-2-Ig mice with cancer or previously healthy septic Fabpl-Bcl-2 mice, no protection against sepsis-induced apoptosis was identified in the cell type where the transgene is expressed in this set of experiments (Fig. 6B, C).

Fig. 6.

Effect of Bcl-2 overexpression on intestinal epithelial apoptosis and survival in septic mice with cancer. No difference in mortality was observed between Fabpl-Bcl-2 mice and WT mice 7 days after induction of Pseudomonas aeruginosa pneumonia (A, n=27-29/group). Despite the fact that Bcl-2 was overexpressed in the intestinal epithelium in transgenic mice, intestinal epithelial apoptosis was similar by both H&E staining (B) and active caspase-3 staining (C) in Fabpl-Bcl-2 mice and WT mice, n=12-13/group.

Discussion

This manuscript demonstrates that prevention of sepsis-induced lymphocyte apoptosis in mice with cancer increases mortality. These results contrast significantly from a very extensive literature demonstrating that prevention of lymphocyte apoptosis in a peritonitis model of sepsis improves survival in mice that were healthy prior to their septic insult. In addition, animals with the identical genetic alteration in apoptosis signaling respond to sepsis in contradictory manners depending on whether an animal has cancer prior to the onset of sepsis and the model of sepsis used -- both Bcl-2-Ig mice and Bim^-/- mice have increased survival if peritonitis is initiated in previously healthy mice (17,21,55) but have decreased survival if pneumonia is initiated in mice with pre-existing cancer.

A differential host response appears to play a role in explaining this discrepancy. Bcl-2-Ig mice with cancer have increases in their Th1 cytokines TNF-α and IFN-γ in bronchoalveolar lavage fluid 24 hours after the onset of pneumonia compared to WT septic mice with cancer. Additionally, when their splenocytes are stimulated ex vivo, they have decreased production of the Th2 cytokine IL-10. These suggest there is an exaggerated inflammatory response in septic mice with cancer when lymphocyte apoptosis is prevented. This increased pro-inflammatory response is both local and systemic in nature given the findings in both BAL fluid and in stimulated splenocytes.

The inflammatory response following sepsis is complex. Either too much or too little inflammation can damage the host by causing unintended organ failure or leaving the host susceptible to secondary infections respectively. Some degree of inflammation is beneficial as immunoparalyzed hosts are at risk for fatal infectious complications. However, the “right” amount of inflammation varies with both the clinical scenario and the length of time elapsed after the initial insult. Too much inflammation can be every bit as harmful as too little inflammation, and it is likely that the increased inflammatory state in the lungs and the decreased ability of splenocytes to secrete compensatory anti-inflammatory cytokines played a role in the excessive mortality seen.

In our prior description of why septic mice with cancer were more likely to die than septic previously healthy mice (both WT C57Bl/6 mice), we found that sepsis induced an increase in T and B lymphocyte apoptosis in all animals (44). However, septic mice with cancer had decreased T and B lymphocyte apoptosis compared to previously healthy septic mice. Together with these results, this implies that some degree of lymphocyte apoptosis may be beneficial in a subset of septic hosts in light of the fact that a) mice with cancer have lower levels of sepsis-induced apoptosis than previously healthy mice despite higher mortalities and b) preventing this apoptosis leads to a further increase in mortality. Thus, an important implication of this study is that prevention of lymphocyte apoptosis may not always be a beneficial therapeutic strategy and, in fact, may be a harmful strategy in select patient groups with a specific disease process. This is directly relevant because of the large amount of interest in translating anti-apoptotic therapies to the bedside of septic patients based upon overwhelming evidence in animal models as well as observational human studies suggesting that lymphocyte apoptosis is harmful in sepsis.

One possible explanation for the discrepancy between our results and those published in the existing literature is the difference in the model used. The vast majority of mouse studies demonstrating the benefit of preventing lymphocyte apoptosis use the cecal ligation and puncture model, and there is overwhelming evidence supporting apoptosis prevention in previously healthy mice with peritonitis. In contrast, this study uses a pneumonia model. To the best of our knowledge, there is only a single published study examining lymphocyte apoptosis prevention in pneumonia, where overexpression of Bcl-2 in lymphocytes resulted in a trend toward improved survival in Pseudomonas aeruginosa pneumonia (55). A similar experiment performed for this manuscript also demonstrates a non-significant 16% improvement in survival in Bcl-2-Ig mice subjected to Pseudomonas aeruginosa pneumonia compared to WT mice (data not shown). Based upon this, it is possible that Bcl-2 overexpression confers either a small survival benefit or has no meaningful benefit in previously healthy mice subjected to pneumonia. Each of these results would be at least somewhat different from previously healthy mice subjected to peritonitis, where apoptosis prevention confers a marked survival benefit. Importantly, the results also differ significantly from mice with cancer that subsequently are given pneumonia, demonstrating the addition of cancer as a comorbidity alters the host response by changing an intervention (lymphocyte apoptosis prevention) that may ordinarily be either beneficial or neutral into one that is detrimental to host survival. Although both peritonitis and pneumonia models induce sepsis, they do not induce an identical inflammatory response. Both Bcl-2-Ig mice and Bim^-/- mice have increased basal numbers of lymphocytes compared to WT animals. In a pneumonia model, this baseline increase in effector cells could lead to profound changes in a stressed host. As such, this increase in effector cells might be more physiologically significant than prevention of lymphocyte apoptosis and functionally overwhelm it, resulting in the observed pro-inflammatory response. This explanation could help explain the differential result between mice subjected to peritonitis vs. those subjected to pneumonia, although it would not explain the difference in survival between Bcl-2 overexpression in mice that were healthy prior to the onset of pneumonia vs. those that had cancer prior to the onset of pneumonia.

Although the vast majority of preclinical studies use healthy six to twelve week old mice, this actually models a patient population that almost never gets septic and is very unlikely to die from the disease in the rare instance that they develop it. Sepsis is a disease that most commonly affects the elderly or patients with comorbidities (or both); however, the typical study using young, previously healthy mice is the equivalent of studying the disease in a 13 year old (56,57) without any medical history. Interventions that may be beneficial in this age group may lose their efficacy or actually be harmful in either aged patients of those with chronic comorbidities, who have markedly different inflammatory milieus at baseline.

The findings in septic Fabpl-Bcl-2 mice reinforce the importance of the cancer as a comorbidity. Previously healthy Fabpl-Bcl-2 mice do not have the increase in sepsis-induced gut epithelial apoptosis seen in WT animals subjected to the same insult and have improved survival following sepsis (24,58). In contrast, when these animals have cancer prior to the induction of sepsis, they neither prevent sepsis-induced apoptosis nor improve survival. Although the finding that Bcl-2 was ineffective in preventing sepsis-induced gut apoptosis was surprising, we have previously described that these animals fail to prevent sepsis-induced intestinal apoptosis in the absence of lymphocytes (59), which suggests that the strength of apoptotic signaling (in the intestine at least) may be modifiable by host factors outside of the local environment. Further, while the loss of the survival benefit conferred by gut-specific Bcl-2 was not as striking as the worsening of survival conferred by lymphocyte-specific Bcl-2 in septic mice with cancer, the results are consistent in that both the host response to sepsis and attempts to alter this response may be determined, at least in part, prior to the onset of infection.

This study has a number of limitations. Since all nonsurvival studies were performed at a single timepoint (24 hours) the experimental design did not allow for a dynamic assessment of the temporal changes in the host response. Next, even though they were subjected to the same model of sepsis, mortality in WT C57Bl/6 mice was significantly higher in experiments examining the effect of Bim^-/- mice than in those examining the effect of Bcl-2-Ig mice for which we do not have a clear explanation as they were performed by the same person (ACF). Next, the decrease in sepsis-induced lymphocyte apoptosis in Bcl-2-Ig mice was small. In large part, we believe this is because sepsis-induced lymphocyte apoptosis is significantly lower in mice with cancer than previously healthy mice (44). Previous work from our laboratory has shown that T lymphocyte apoptosis in Pan02-bearing septic mice is 10.4% and that T lymphocyte apoptosis in sham mice with cancer is 3.5% (44). Comparing these prior results to data shown in figure 1B, Bcl-2 therefore decreased sepsis-induced T lymphocyte apoptosis to levels seen in sham mice. The fact that mortality was decreased in both Bcl-2-Ig mice and Bim^-/- mice, both of which have improved survival following CLP in previously healthy mice (17, 21, 55), strengthens our findings. However, all transgenic and knockout mice used in this study have lifelong genetic alterations which could lead to chronic changes in other immune effector cells which could affect the outcome when mice are subjected to pneumonia. As such, an alternate strategy initiating either antibodies or drugs that inhibit apoptosis after the onset of pneumonia might have more clinical relevance and in theory have a different impact on mortality.

Despite these limitations, these results lead to a new paradigm in understanding lymphocyte apoptosis in sepsis. There continues to be overwhelming evidence that lymphocyte apoptosis is harmful in hosts that were healthy prior to the onset of septic peritonitis. However, prevention of lymphocyte apoptosis by chronic genetic alterations in the mitochondrial pathway is harmful in mice with cancer that subsequently develop sepsis from pneumonia. Thus, the results do not support the view that sepsis-induced apoptosis is, by definition, a maladaptive response independent of the host or insult. Rather, they suggest a response that is modifiable by pre-existing host comorbidities and model of sepsis utilized in which a basal level of lymphocyte cell death may be beneficial under the correct circumstances. These results should be repeated using other types of tumors, other comorbidities, other models of sepsis and other apoptosis inhibitor strategies to determine how generalizable they are. They should also be considered in designing entry criteria for future clinical trials aimed at preventing lymphocyte apoptosis in sepsis.