Coopersmith CM, Stromberg PE, Dunne WM, Davis CG, Amiot II DM, Buchman TG, Karl IE, Hotchkiss RS. Inhibition of Intestinal Epithelial Apoptosis and Survival in a Murine Model of Pneumonia-Induced Sepsis. JAMA. 2002;287(13):1716-1721. doi:10.1001/jama.287.13.1716
Author Affiliations: Departments of Surgery (Drs Coopersmith and Buchman, Mssrs Stromberg and Amiot), Anesthesiology (Mr Davis and Dr Hotchkiss), Pathology (Dr Dunne), and Medicine (Dr Karl), Washington University School of Medicine, St Louis, Mo.
Context Increased intestinal epithelial apoptosis is present in both human autopsy
studies and animal models of sepsis. Whether altering gut apoptosis decreases
mortality in sepsis induced by pathogenic bacteria outside the gut is unknown.
Objective To determine if decreasing levels of intestinal cell death improves
survival in a murine model of Pseudomonas aeruginosa
Design and Materials Prospective study in which transgenic mice that overexpress the antiapoptotic
protein Bcl-2 in their intestinal epithelium (n = 25) and control littermates
(n = 26) were subjected to intratracheal injection of P
Main Outcome Measures Survival at 7 postoperative days, compared between the 2 groups. Secondary
outcomes included quantification of gut epithelial apoptosis.
Results Survival in transgenic mice that overexpress Bcl-2 in the intestinal
epithelium was 40% (10/25) compared with 4% (1/26) in control littermates
7 days after intratracheal injection of P aeruginosa
(P = .001), with differences in survival noted within
24 hours of surgery. Overexpression of Bcl-2 was associated with a decrease
in gut epithelial apoptosis demonstrated by active caspase 3 staining, the
terminal deoxynucleotidyltransferase-mediated dUTP nick end-labeling assay,
and hematoxylin-eosin staining.
Conclusions In this murine model, inhibiting gut epithelial apoptosis by overexpression
of Bcl-2 was associated with a survival advantage in P aeruginosa pneumonia–induced sepsis. These results suggest that intestinal
epithelial apoptosis may play a role in sepsis-related mortality.
Sepsis is the leading cause of mortality in critically ill patients
nationwide. A recent analysis estimated that 750 000 individuals develop
sepsis annually and more than 210 000 die of the disease.1
The national hospital cost associated with care of patients with severe sepsis
is $16.7 billion.1 The rate of death from sepsis
has increased more than 90% in the last 20 years.2
Previous studies have implicated alterations of intestinal function as critical
to the development of sepsis,3- 15
leading to the theory that the gut represents the "motor" of the systemic
Gut epithelial apoptosis is increased in human autopsy studies and animal
models of sepsis.18- 21
An autopsy study comparing samples from multiple organ systems in 20 patients
who died in the surgical intensive care unit of sepsis and multiple organ
dysfunction with those from 16 critically ill, nonseptic patients demonstrated
increased intestinal epithelial apoptosis in septic patients.18
The gut epithelium represented 1 of only 2 cell types (along with lymphocytes)
in septic patients with prominent apoptosis. Mice that underwent cecal ligation
and puncture (CLP), a murine model of ruptured appendicitis, had increased
gut epithelial apoptosis compared with controls.19
In other studies, septic transgenic mice that overexpress the antiapoptotic
protein Bcl-222- 25
in their intestinal epithelium had increased survival following CLP compared
with their septic control littermates that underwent the same procedure.21 These studies suggest that decreasing gut epithelial
apoptosis is associated with a survival advantage in sepsis of intestinal
origin but do not evaluate whether gut epithelial apoptosis plays a role in
mortality in sepsis caused by pathogenic bacteria focused outside the gut.
In this study, we examined survival in transgenic mice that had Pseudomonas aeruginosa pneumonia–induced sepsis.
Strain FVB/N transgenic mice containing nucleotides −596 to +21
of a rat fatty acid binding protein (Fabpl) linked to human Bcl-2 (a gift
from Jeffrey I. Gordon, MD, Washington University, St Louis, Mo) were generated
and genotyped using polymerase chain reaction protocols detailed elsewhere.26Fabpl-Bcl-2 animals have
no detectable abnormalities when aged to 18 months and appear phenotypically
identical to nontransgenic littermates.21,26
All studies complied with National Institutes of Health guidelines for the
use of laboratory animals and were approved by the Washington University Animal
A total of 103 six- to eight-week-old mice had a midline cervical incision
performed under halothane anesthesia. 27- 29
Each animal received an intratracheal injection of 40 µL of a P aeruginosa solution, after which the mouse was held vertically
for 10 seconds to enhance delivery into the lung. The incision was closed
in 2 layers. Sham-operated mice were handled identically, but had intratracheal
injection of 40 µL of 0.9% NaCl.
The ATCC 27853 strain of P aeruginosa was grown
overnight in trypticase soy broth with constant shaking. A 10-mL volume of
the culture medium was placed in a 50-mL tube and centrifuged for 10 minutes
at 6000g. The resulting pellet was resuspended in
an equal volume of saline and centrifuged again. The final density of the
inoculum was adjusted to 0.3 A600nm, corresponding to a cell density
ranging between 5 × 108 and 1 × 109 colony-forming
units (CFUs)/mL as determined by serial dilution and colony counts. Based
on the cell density and the volume injected intratracheally, the dose of P aeruginosa administered to each animal was between 20
million and 40 million CFUs per injection.
For quantification of apoptosis, 21 mice (12 nontransgenic and 9 transgenic)
were euthanized 24 hours after injection with P aeruginosa. Each animal's entire small intestine was immediately removed. The
intestine was opened along the length of its cephalocaudal axis, washed in
10% buffered formalin (to remove luminal contents) and then fixed in the same
Apoptotic cells were identified using 3 complementary techniques: active
caspase 3 staining, the terminal deoxynucleotidyltransferase-mediated dUTP
nick end-labeling (TUNEL) assay,30 and hematoxylin-eosin
(H&E) staining. Serial sections were scored for apoptosis by a single
investigator. For active caspase 3 staining, a minimum of 100 well-oriented
crypt-villus units were scored per section. Well-oriented was defined as a crypt sectioned parallel to the crypt-villus axis
with Paneth cells at the crypt base and an unbroken epithelial column extending
to the villus tip. For H&E staining and the TUNEL assay, a minimum of
10 random fields were evaluated in each sample. Cells undergoing apoptosis
were identified by characteristic morphology including nuclear fragmentation
(karyorrhexis) and cell shrinkage with condensed nuclei (pyknosis) on H&E-stained
sections and by immunohistochemical detection of positive cells using the
TUNEL assay. In comparing animals subjected to intratracheal injection with
either P aeruginosa or 0.9% NaCl, identifying information
on sections was obscured, and the slides were extensively shuffled by an investigator
different from the person scoring for apoptosis (P.E.S.). When scoring on
all slides was entirely complete, the numerical code (devised by C.M.C.) was
broken to reveal the identity of all slides counted.
Staining for active caspase 3 was performed as previously described.18,21 Briefly, paraffin-embedded tissues
were dewaxed and rehydrated. Endogenous peroxidase activity was blocked by
incubating in 3% H2O2 in methanol at 23°C for 15
minutes. Sections were then microwaved in citrate buffer (pH 6.0) for 9 minutes
to facilitate antigen retrieval. Polyclonal rabbit antiactive caspase 3 was
applied, diluted 1:100 in phosphate-buffered saline for 60 minutes at 23°C
(Cell Signaling Technology Inc, Beverly, Mass), followed by a secondary biotinylated
goat anti–rabbit antibody for 30 minutes (1:200) (Vector Laboratories,
Burlingame, Calif). Slides were then incubated with VECTASTAIN ABC-AP (Vector
Laboratories), developed with alkaline phosphatase substrate solution, and
counterstained with hematoxylin.
Tissue sections were stained for the TUNEL assay using a commercially
available kit according to manufacturer specifications (Roche Diagnostics
Corp, Indianapolis, Ind). After rinsing, a streptavidin-biotin complex (VECTASTAIN
ABC, Vector Laboratories) was applied at 23°C for 30 minutes. After rinsing,
metal-enhanced diaminobenzidine (Pierce Chemical Co, Rockford, Ill) was added,
and slides were counterstained with hematoxylin.
The survival studies were conducted in a separate group of mice (n =
51) in the following fashion: offspring of FVB/N and heterozygous Fabpl-Bcl-2 mice were weaned at 3 weeks of age. Prior to identifying
the genotype of each of these animals, the tails of mice in the resulting
litters (up to 5 mice per cage of approximately 50% transgenic and 50% nontransgenic
animals as would be expected by mendelian genetics) were uniquely marked by
circumferential inscription with indelible ink, 1 to 5 circles per mouse.
Each animal was subsequently genotyped using polymerase chain reaction techniques
Three weeks later, an investigator (P.E.S.) blinded to the identity
of the mice performed intratracheal injections of 40 µL of P aeruginosa in both age-matched and sex-matched Fabpl-Bcl-2 mice and FVB/N littermates. Injections were done sequentially
according to the tail marks (animals with 1 mark were injected first, followed
by animals with 2 marks, etc). The identifying tail marks were made prior
to genotyping of the animals and the investigator was unaware of the identity
of any animal injected. Mice were allowed free access to food and water throughout
the course of the experiment. Animal survival was recorded for 7 days postoperatively.
In a different group of animals (n = 12) than those used for either
quantitation of apoptosis or survival curves, mesenteric lymph node and spleen
were removed 16 hours following intratracheal instillation of P aeruginosa, then weighed and homogenized in glass tissue grinders
containing 1 mL of sterile phosphate-buffered saline. Blood (100 µL)
was diluted 1:10 in sterile phosphate-buffered saline. Serial dilutions of
homogenate and blood were then cultured on blood agar (total aerobes) and
MacConkey (gram-negative facultative aerobes) plates. Plates were incubated
at 37°C and examined 24 hours later.
At 2, 16, or 24 hours after intratracheal injection with P aeruginosa solution, whole blood was drawn by cardiac puncture. Animals
in the cytokine experiments were separate from those used in either apoptosis
quantitation or survival experiments, although the bacteriologic analysis
described above was performed on the same subgroup of animals examined here
at 16 hours after bacterial injection. Blood was centrifuged for 5 minutes
at 6000g to separate out plasma. Concentrations of
tumor necrosis factor α (TNF-α), interleukin 6 (IL-6), interferon γ
(IFN-γ), and IL-10 were measured by enzyme-linked immunosorbent assay
using commercially available kits (R&D Systems, Minneapolis, Minn) according
to manufacturer specifications.
Data analysis was performed using Prism v3.0 (GraphPad Software, San
Diego, Calif). Differences in group survival were analyzed using the χ2 test. Comparisons of apoptosis levels, bacteriologic data, and cytokine
levels between transgenic and nontransgenic animals were analyzed using the t test. A P value of less than
.05 was considered statistically significant.
Mice that received P aeruginosa were judged
to be septic by the presence of bacteremia and objective clinical findings
(eg, tachypnea, lethargy) while sham-operated animals that received 0.9% NaCl
had no gross abnormalities. There was no evidence of wound infection in any
animal studied. Animals that received intratracheal bacteria had a 50% mortality
24 hours postoperatively, while no animals that received saline died. Septic
animals that received intratracheal injection of P aeruginosa (n = 8) had significantly greater numbers of active caspase 3 cells/100
crypts (mean, 71.4 [SEM, 1.6]) (Figure 1A),
compared with sham-operated animals that received intratracheal saline (n
= 4) that had 10.8 (1.4) active caspase 3 cells/100 crypts (Figure 1D; P<.001). Similar results
were qualitatively obtained using both the TUNEL assay and H&E staining
(Figure 1B, C, E, F).
Transgenic mice that overexpress human Bcl-2 under control of the Fabpl
promoter have detectable levels of this antiapoptotic protein in all epithelial
lineages of the intestine, without expression in the surrounding gut-associated
lymphoid tissue.21,26 Following
intratracheal injection of P aeruginosa, and compared
for gut epithelial apoptosis as described above, transgenic Fabpl-Bcl-2 mice (n = 9) had a decrease in mean (SEM) intestinal cell
death—23.9 (2.0) active caspase 3 cells/100 crypts in Fabpl-Bcl-2 animals vs 71.4 (1.6) active caspase 3 cells/100 crypts
in nontransgenic septic littermates (n = 8) (Figure 2A and C; P<.001). Similar qualitative
results were seen with H&E staining and the TUNEL assay (Figure 2B and D; also, data not shown).
Survival experiments were performed on 3 separate days using 8 to 10
mice per group (total: 25 Fabpl-Bcl-2 mice and 26
FVB/N mice). P aeruginosa pneumonia–induced
sepsis was highly lethal in wild-type mice, with only 1 (4%) animal surviving
5 days. Among Fabpl-Bcl-2 mice, 10 (40%) survived
for 7 days (Figure 3, P<.005). The survival curves began diverging within the first postoperative
day, and nearly all mortality in both groups occurred within the first 3 days
of the study.
The majority of both transgenic and nontransgenic mice were still alive
16 hours postoperatively. However, a substantial portion of wild-type animals
injected with P aeruginosa died within 24 hours.
All animals in both groups had P aeruginosa
bacteremia and had organisms detectable in their spleens. Nontransgenic animals
(n = 7) had the greatest mean (SEM) organism burden in their blood—2 274 000
(1 497 000) CFU/mL. Although Fabpl-Bcl-2
mice (n = 3) had a 120-fold decrease in bacterial counts (19 000 (8185)
CFU/mL), there was no statistical difference between groups, probably due
to the small sample size and large variation of P aeruginosa counts in the blood of nontransgenic animals (range, 20 000-10
million CFU/mL). A similar trend was seen in splenic cultures, with mean (SEM)
629 300 (8185) CFU/g in nontransgenic animals compared to 29 970
(19 040) CFU/g in septic transgenic animals.
Mesenteric lymph node cultures obtained 16 hours after injection with P aeruginosa and tested for bacterial translocations had
low levels of P aeruginosa in the mesenteric lymph
nodes of all mice tested (n = 12 animals), but evidence of other enteric organisms
in 2 animals—2600 CFU/g in a nontransgenic animal and 560 CFU/g in a
transgenic animal, both of which grew Klebsiella pneumoniae.
Plasma from transgenic and nontransgenic animals was measured for the
proinflammatory cytokines TNF-α, IL-6, and IFN-γ, and the anti-inflammatory
cytokine IL-10 at 2, 16, and 24 hours after intratracheal injection of P aeruginosa. All cytokines were either undetectable or
present at low levels at the 2-hour time point (transgenic: n = 4; nontransgenic:
n = 7). Levels of each cytokine increased by 16 hours in both groups of animals.
There was a nonsignificant trend toward higher quantities of all cytokines
in nontransgenic animals compared with Fabpl-Bcl-2
mice, especially for IL-10 measurements, with mean (SEM) levels of 233 (76)
pg/mL in nontransgenic animals (n = 6) vs 105 (46) pg/mL in transgenic animals
(n = 4). Smaller differences were observed in levels of TNF-α (179 
pg/mL for nontransgenic vs 99  pg/mL for transgenic) and IL-6 (3344 
pg/mL for nontransgenic vs 2461  pg/mL for transgenic).
By 24 hours postinjection, cytokine levels were relatively static or
decreasing in all animals. The sole exception was TNF-α in wild-type
animals (n = 4), which had a greater than 2-fold increase compared with septic
animals of the same genotype at 16 hours, increasing to a level of 376 (145)
pg/mL. Levels of TNF-α in Fabpl-Bcl-2 mice
(n = 4) were minimally changed (158  pg/mL]) at the 24-hour time point.
In this study, sepsis induced by P aeruginosa
pneumonia resulted in an increase in gut epithelial apoptosis. Although sepsis
originated from a pulmonary source, decreasing intestinal epithelial cell
death by overexpression of Bcl-2 was associated with improved survival.
The results described herein suggest that decreasing intestinal cell
death with Bcl-2 may play a role in improving survival for sepsis that originates
from an extra-abdominal source. This finding may be clinically significant
because approximately 40% of septic episodes in humans begin in the lung.1,31
While this study shows an association between increased Bcl-2 and survival
in sepsis, we cannot conclude that a decrease in gut epithelial apoptosis
is directly responsible for the decrease in mortality observed in transgenic
animals. Although the role of Bcl-2 as an antiapoptotic protein has been demonstrated,22- 25 it
is possible that a separate effect of the transgene that we are unable to
detect is responsible for the survival benefit conferred. If, however, the
lower death rate in transgenic animals is due to an alteration in gut epithelial
apoptosis, the mechanisms that underlie this link must be further clarified.
One possibility we tested for was the presence of bacterial translocation.
However, the microbiologic data demonstrate that bacterial translocation is
not the predominant mechanism of the survival benefit conferred by Bcl-2.
Although low levels of P aeruginosa were present
in the mesenteric lymph nodes of all animals examined 16 hours postoperatively,
only 2 animals demonstrated evidence of enteric organisms at this time point.
In addition, although all animals were bacteremic, no enteric organisms were
cultured from the blood of either transgenic or nontransgenic animals.
Our results are consistent with human studies that challenge the role
of bacterial translocation in the origins of sepsis,32
but differ from a recent study by Yu and Martin29
that demonstrated 67% of rats subjected to intratracheal injection of P aeruginosa had evidence of translocation of intestinal
bacteria to the mesenteric lymph system. However, that study examined translocation
40 hours after surgery, whereas the cultures in our study were obtained 24
hours earlier. It is possible that had the animals in our study been cultured
40 hours postoperatively, a similar increase in bacterial translocation may
have been observed. This would be of little physiologic significance since
75% of nontransgenic animals and 50% of transgenic animals did not survive
Even though bacterial translocation does not seem to be responsible
for the survival benefit conferred by Bcl-2, it is possible that altered intestinal
permeability plays a role in a potential link between gut epithelial apoptosis
and survival. Yu and Martin29 have shown that
sepsis from P aeruginosa pneumonia is associated
with decreased gut barrier function. This is consistent with rat studies that
show that gut permeability is increased in intestinal ischemia/reperfusion.33 Increased gut epithelial apoptosis is also associated
with gut hyperpermeability in animal models.34
Importantly, treatment with the caspase inhibitor z-VAD in an intestinal cell
line prevented both gut epithelial apoptosis and barrier dysfunction.35 Since translocation of gut-derived factors other
than intact bacteria to the mesenteric lymph has been shown to be detrimental
in animal models of critical illness,32,36
additional studies are needed to assess the role of Bcl-2 on gut barrier function.
The cellular signaling pathways responsible for increased gut apoptosis
in sepsis are unclear. In this study, apoptotic cells were present in nearly
every crypt in pneumonia from P aeruginosa, whereas
they are a more focal phenomenon in CLP21;
this may be attributable to the differing mortalities between the models.
The role cytokines play in the relationship between Bcl-2 overexpression
and survival is incompletely answered by this study. Multiple studies have
shown alterations in cytokine levels of intestinal4,5,7,9,11,15,16
or immune37,38 origin in critical
illness. The data presented here, however, fail to show any statistically
significant differences in cytokine levels between transgenic and nontransgenic
animals. Potential explanations for this include alterations in cytokine levels
not examined for this study and true differences in local mediator levels
not adequately detected by measuring systemic blood levels.
Our study has several limitations. It is possible that even if alterations
in gut cell death are responsible for the effect presented in this murine
study, decreasing intestinal apoptosis would therefore not be beneficial to
septic patients in the intensive care unit. To have clinical applicability,
an approach must be developed to decrease gut epithelial apoptosis after a
patient becomes septic. While caspase inhibitors administered via intraperitoneal
injection improve survival in CLP,37,38
it is unclear what role the intestine plays in this response, and no data
exist on gut-directed anticaspase therapy. The role intestinal apoptosis plays
in either gram-positive or fungal septicemia also is not known, since both
this study and earlier work on CLP used gram-negative bacteria in murine models
of sepsis. In addition, animals in this study were not given antibiotics,
which is standard treatment for patients with pneumonia, sepsis, or both.31 Although both transgenic and nontransgenic animals
were bacteremic, it is possible that antibiotics would have improved survival
disproportionately in the nontransgenic group.
In summary, similar to other animal models and human studies of critical
illness, monomicrobial sepsis induced by P aeruginosa
pneumonia is associated with an increase in gut epithelial apoptosis. Overexpression
of Bcl-2 was associated with reduced apoptosis in the intestinal epithelium
and conferred a survival advantage, with 40% of transgenic mice surviving
their septic insult. Gut epithelial apoptosis may play a role in sepsis-related
mortality, and reduction of intestinal cell death may represent a potential
therapeutic approach toward improving survival in sepsis.