Probiotic Administration Reduces Mortality and Improves Intestinal Epithelial Homeostasis in Experimental Sepsis

SEPSIS is a common cause of death in children and adults despite advances in the supportive care for the critically ill patients. Centers for Disease Control data shows that death rates from sepsis have increased at a rate greater than from any other common cause of mortality in the last year, and sepsis is now one of the top 10 causes of death in the United States.¹  Annually, above 1 million deaths worldwide are associated with sepsis within the pediatric and neonatal population alone.^2,3  Even when pediatric and adult patients survive, they face substantial long-term adverse consequences after sepsis and critical illness.^4–6 

Critical illness and sepsis are systemic syndromes that lead to a hostile environment in the gut, resulting in an imbalance of the intestinal microbiota in favor of pathogenic species.⁷  The intestine plays a central role in the pathogenesis of sepsis and has been referred to as the “motor” of the systemic inflammatory response.^8,9  Perturbations of intestinal epithelial homeostasis in sepsis result in increased proinflammatory cytokine production,¹⁰  barrier dysfunction,^11–13  and increased apoptosis^14–17  which may lead to multiple organ failure.

Probiotic therapy represents a promising intervention for the treatment of nosocomial infections in the intensive care unit, which often lead to sepsis and multiple organ failure.¹⁸  Probiotics are living nonpathogenic bacteria that colonize the intestine and provide a benefit to the host.¹⁹  Results from recent randomized controlled trials in pediatric and adult populations suggest a benefit to the use of probiotics in the intensive care unit.^18,20–22  In an age of increasing antimicrobial resistance, the use of probiotics as a potential treatment for severe infections and subsequent septic shock may be a promising, cost-effective, preventative strategy.

The potential beneficial role(s) of probiotics on survival given in early systemic polymicrobial sepsis, and the mechanisms by which probiotics may function to protect against experimental sepsis, particularly pediatric sepsis, are currently not well understood. Therefore, we sought to delineate the clinical utility and potential mechanistic targets of acutely administered probiotic therapy at initiation of experimental pediatric sepsis via a weanling mouse peritonitis model. We selected the two commonly clinically used probiotic strains, Lactobacillus rhamnosus GG (LGG) and Bifidobacterium longum (BL), given at clinically relevant doses, to evaluate possible strain-related differences in clinical outcome and mechanistic targets in the gut.

Lactobacillus spp. and Bifidobacterium spp., alone or in combination, are the most frequently used probiotic strains in the treatment of various gastrointestinal disorders^23–25  or as therapy for clinical conditions including antibiotic-associated diarrhea,^26,27  ventilator-associated pneumonia,^18,21,28  sepsis, and postoperative infections.^29,30  Although probiotics seem to be an effective treatment in various clinical conditions, the specific mechanisms responsible for their beneficial action are complex and not fully understood.³¹  On the basis of the results from several in vivo and in vitro studies, we hypothesized that LGG and BL would prevent bacterial translocation,^32,33  reduce the overgrowth of pathogenic bacteria, decrease apoptosis in intestinal epithelial cells,^34–37  and reduce inflammation.^38–41 

The animal protocol used in these studies was approved by the Institutional Animal Care and Use Committee of the University of Colorado Anschutz Medical Campus. Briefly, 3-week-old FVB/N mice were orally gavaged with 200 µl of either LGG (1 × 10⁹ colony forming unit [CFU]/ml), BL (1 × 10⁷ CFU/ml), or sterile water (vehicle) immediately before initiation of the cecal ligation and puncture (CLP) procedure.⁴²  Briefly, a small midline abdominal incision was made, the cecum was ligated just distal to the ileocecal valve, and was then punctured twice with a 23-gauge needle. The cecum was squeezed to extrude a small amount of stool, replaced in the abdomen, and the peritoneum and skin were closed in layers. Sham mice were treated identically, except the cecum was neither ligated nor punctured. All mice received 1.0 ml normal saline subcutaneously after the surgery to compensate for fluid loss. Animals were euthanized at either 24 h (for acute studies) or followed 7 days for survival (LGG, n = 14; BL, n = 10, or vehicle, n = 13; shams, n = 3 per group). For survival studies, mice were treated with probiotics daily for 7 days. For acute studies, mice received a single dosage of probiotics before tissue collection.

LGG (ATCC, Manassas, VA) was incubated in de Man, Rogosa, and Sharpe broth (Becton Dickinson, Sparks, MD) for 24 h at 37°C and 5% CO₂. BL (ATCC) was cultured in Trypticase soy broth (Becton Dickinson) for 72 h in an anaerobic chamber at 37°C. A₆₀₀ was measured to determine the number of CFU per 1 ml. BL and LGG were pelleted from the broth (10,000 rpm; 10 min) and resuspended in distilled water.

Blood was collected at 24 h from the inferior vena cava of anesthetized mice, serially diluted in sterile 0.9% saline and cultured on Trypticase soy agar plates with 5% sheep blood (Becton Dickinson) for 24 h at 37°C/5% CO₂. CFU were then enumerated for each animal (shams, n = 8 per group; septic, n = 10; LGG, n = 7; and BL, n = 8).

DNA was extracted from collected frozen colon samples obtained from a separate study to ensure sufficient amount of tissue (shams, n = 4 per group; septic, n = 12; LGG, n = 10; and BL, n = 9; luminal and intestinal wall contents; 24-h time point) using UltraClean Fecal DNA kit (MO BIO Laboratories, Inc., Carlsbad, CA). The concentration, integrity, and purity of DNA were determined using a NanoDrop (Thermo Fisher Scientific, Wilmington, DE). Quantification was performed by quantitative polymerase chain reaction (PCR) using standard curves derived from cloned 16S ribosomal RNA genes⁴³  using the following primers: Lactobacillus⁴⁴ : LactoF (5′ TGGAAACAGRTGCTAATACCC) and LactoR (5′ GYCCATTGTGGAAGATTCCC). Bifidobacterium⁴⁵ : Bif1F (5′ TCG CGT CYG GTG TGA AAG) and Bif1R (5′ CCA CAT CCA GCR TCC AC). The following cycling protocol was used: denaturation at 95°C (10 min) and 40 cycles of 95°C (15 s), 60°C (30 s), and 65°C (1 min). Reporter dye emission (SYBR green, Applied Biosystems, Foster City, CA) was detected by an automated sequence detector combined with ABI Prism 7300 Real Time PCR System (Applied Biosystems).

Gram staining was performed on tissue sections and evaluated by microscopy for presence of Gram-positive or Gram-negative bacteria.

Colon was collected from each animal at 24 h and fixed overnight in 10% formalin, paraffin-embedded, and sectioned at 4–6 µm. Serial sections were stained (shams, n = 5 per group; septic, n = 5; LGG, n = 4; and BL, n = 4). After deparaffinization and rehydration, sections were blocked with 1.5% rabbit or goat serum (Vector Laboratories, Burlingame, CA) in phosphate-buffered saline for 30 min, then incubated with either rabbit polyclonal cleaved caspase-3 (1:100; Cell Signaling, Danvers, MA) or mouse monoclonal proliferating cell nuclear antigen (1:100; Invitrogen, Camarillo, CA) antibody for 1 h, washed with phosphate-buffered saline, and incubated with either goat anti-rabbit or anti-mouse biotinylated secondary antibody (Vector Laboratories) for 30 min. Vectastain Elite ABC reagent (Vector Laboratories) was then applied, followed by diaminobenzidine as substrate. Sections were counterstained with hematoxylin, dehydrated, and cover-slipped.

To determine apoptosis in the colonic epithelium, apoptotic epithelial cells were quantified in 100 consecutive crypts by using two complimentary methods: morphological analysis of hematoxylin and eosin–stained sections where apoptotic cells were identified by characteristic morphology of nuclear fragmentation (karyorrhexis) and cell shrinkage with condensed nuclei (pyknosis)⁴⁶  and second by enumeration of cleaved caspase-3–positive cells. Colonic epithelial proliferation was determined by quantifying proliferating cell nuclear antigen-positive cells in 100 consecutive crypts. All counting was performed by a blinded evaluator.

Total RNA was isolated from colonic tissue (snap frozen in liquid nitrogen and collected at 24 h) using the RNeasy Plus Mini Kit (Qiagen, Santa Clarita, CA) as described in the manufacturer’s protocol. RNA concentrations were quantified at 260 nm, and the purity and integrity were determined using a NanoDrop. Reverse transcription and real-time PCR assays were performed to quantify steady-state messenger ribonucleic acid (mRNA) levels of interleukin (IL)-6, tumor necrosis factor-α (TNF-α), IL-1β, IL-10, MyD88, Toll-like receptor (TLR)-4, and TLR-2 (shams, n = 4 per group; septic, n = 7; LGG, n = 6; and BL, n = 4). Complementary DNA was synthesized from 0.2 μg of total RNA. Predeveloped TaqMan primers and probes (Applied Biosystems) were used for detection. Reporter dye emission was detected by an automated sequence detector combined with ABI Prism 7300 Real Time PCR System (Applied Biosystems). Real-time PCR quantification was performed with TaqMan glyceraldehyde-3-phosphate dehydrogenase controls.

Individual frozen colon samples (shams, n = 3 per group; septic, n = 8; LGG, n = 4; and BL, n = 4; 24-h time point) were homogenized with a hand-held homogenizer in a 5x volume of ice-cold homogenization buffer (Tris HCl, 50 mm; pH, 7.4; NaCl, 100 mm; EDTA, 10 mm; Triton X-100, 0.5%) with added protease inhibitors (Roche Diagnostics, Mannheim, Germany). The homogenates were centrifuged at 10,000 rpm for 5 min at 4°C and the supernatant was collected. Total protein concentration was quantified using the Bradford protein assay.⁴⁷  For protein analysis, 40 μg of protein was added to an equal volume of 2x Laemmli sample buffer and boiled for 5 min. The samples were run on 4–15% polyacrylamide gels (Bio-Rad Laboratories, Hercules, CA) at 200 V for 30 min. Protein was transferred to Immuno-Blot PVDF membranes (Bio-Rad Laboratories) at 65 V for 4 h. Membranes were blocked with 5% nonfat milk in Tris-buffered saline with 0.1% Tween 20 (Sigma-Aldrich, St. Louis, MO) for 1 h at room temperature and then incubated overnight at 4°C with one of the following rabbit polyclonal antibodies: Bax (1:1,000; Cell Signaling), Bcl-w (1:1,000; Cell Signaling), anti-Akt, (1:1,000; Cell Signaling), and anti-P-Akt (1:1,000; Cell Signaling). After extensive washing, the membranes were incubated for 1 h at room temperature with horseradish peroxidase–conjugated anti-rabbit immunoglobulin G (Santa Cruz Biotechnology, Dallas, TX). Proteins were visualized with a chemiluminescent system (Pierce, Rockford, IL) by using Epi Chemi II Darkroom (UVP BioImaging System, Upland, CA).

Enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN) was used to determine the concentrations of TNF-α and IL-6 in serum (shams, n = 5 per group; septic, n = 7; LGG, n = 5; and BL, n = 6; 24-h time point) according to the manufacturer’s instructions. Serum was collected after centrifuging blood for 10 min at 5,000 rpm and stored at −80°C until the assay was performed.

Continuous data sets were tested for Gaussian distribution by using a Shapiro–Wilk Test for normality, and the Levene F-Test for equality of variances. Although results of these tests showed some departures from normality and equality of variances, it is well documented that the one-way ANOVA is robust with respect to violations of these assumptions. Thus, multiple group comparisons were performed with one-way ANOVA followed by the Newman–Keuls post hoc test. Survival studies were analyzed via the log-rank test. No measurements or animals were lost for observation or missing in the analysis. Data were analyzed using Prism 4.0 (GraphPad Software, San Diego, CA) and reported as means ± SD. A P value less than 0.05 was considered to be statistically significant.

To determine whether treatment with LGG or BL have an effect on mortality in peritonitis-induced sepsis, a separate cohort of mice (shams, n = 3 per group; septic, n = 13; LGG, n = 14; and BL, n = 10) were subjected to CLP and followed 7 days for survival (fig. 1). Mice treated with either LGG or BL had significantly improved 7-day survival compared with untreated septic mice. All sham animals survived.

To examine whether the probiotics LGG or BL could prevent bacteremia in sepsis, whole blood was cultured and bacteria were counted. The data are expressed as the Log of CFU per 1 ml (fig. 2, A–C). Septic animals had significantly increased bacteremia compared with shams. Although both probiotic strains reduced bacteremia, septic mice treated with LGG exhibited normalization to sham mouse blood bacterial counts whereas septic mice treated with BL exhibited significantly decreased bacteremia compared with untreated septic mice, but still had an increased bacterial load in the blood compared with shams.

DNA isolated from the colon, including luminal and intestinal wall content, was analyzed by PCR to quantify the presence of Lactobacillus spp. and Bifidobacterium spp. There was a significant increase in Lactobacilli in septic animals treated with LGG compared with untreated septic mice. Similarly, there was an increase in Bifidobacteria in septic mice treated with BL compared with untreated septic mice (table 1). These data demonstrate that LGG and BL are indeed able to survive and propagate in the gastrointestinal tract for at least 24 h after oral administration. Therefore, LGG and BL may help to prevent the overgrowth of pathogenic bacteria. This hypothesis was confirmed by Gram staining of intestinal wall specimens, which demonstrated an increased number of both Gram-positive and Gram-negative bacteria in the colon of septic mice compared with shams, and most importantly, less bacteria is seen in the colons of mice treated with either LGG or BL (fig. 3).

As measured by quantifying proliferating cell nuclear antigen-positive cells in 100 consecutive crypts, septic mice exhibited a significant decrease in proliferation of the colonic epithelium compared with sham mice. In contrast, the proliferative response in septic animals treated with either probiotic strain was normalized to levels observed in sham mice (fig. 4).

Colonic epithelial apoptosis was increased in untreated septic mice compared with shams, both when assayed by cleaved caspase-3 staining and also by morphological criteria in hematoxylin and eosin–stained sections. In contrast, septic mice treated with either LGG or BL exhibited decreased colonic apoptosis, with levels similar to those seen in the colons of sham mice (fig. 5, A–D). The ratio of proapoptotic to antiapoptotic molecules is often used as an indicator of sensitivity to apoptosis.⁴⁸  Untreated septic mice exhibited significantly increased ratio of Bax/Bcl-w protein expression compared with shams, suggesting a shift toward increased programmed cell death. In contrast, septic mice treated with either LGG or BL exhibited a decrease in the ratio of Bax/Bcl-w compared with untreated septic mice, suggesting a shift toward increased cell survival (fig. 5E).

Previously published data by Yan et al.³⁷  showed that LGG promotes survival of intestinal epithelial cells through regulation of the antiapoptotic Akt/protein kinase B signal transduction pathway. Likewise, our results show a significant increase in antiapoptotic Akt in the colonic tissue of both probiotic groups, LGG and BL, compared with the untreated septic group (fig. 6, A–B). This indicates a possible involvement of the pathway in the protective mechanism of both tested probiotic strains against sepsis.

Reduction of inflammation as well as improvement of innate immunity are hypothesized to be protective mechanisms after probiotic administration.^38–41  To determine the effect of probiotics on these parameters, levels of the proinflammatory cytokines such as IL-6 and TNF-α in serum and colonic tissue were measured by enzyme-linked immunosorbent assay. In parallel, gene expression of IL-6, TNF-α, IL-1β, and IL-10 in the colon was quantified using real-time PCR. All cytokine levels were measured 24 h after CLP.

Systemic levels of IL-6 significantly increased in untreated septic mice compared with shams (fig. 7), whereas treatment with LGG or BL in septic mice led to significantly reduced systemic IL-6 levels compared with septic mice not treated with probiotics. Interestingly, systemic levels of TNF-α remained unchanged among all studied groups.

Local, colonic gene expression of TNF-α, IL-1β, and IL-10 (fig. 8, A–C) was significantly increased in the colon of untreated septic animals, whereas levels of both were markedly decreased in colons of septic mice treated with LGG or BL. Colonic gene expression of IL-6 was increased in untreated septic mice and decreased in both probiotic-treated groups although the differences did not reach statistical significance (data not shown).

TLRs signal through the MyD88 pathway, which includes NFκB transcriptional factors, activating various cytokines involved in the innate immunity response. MyD88 has an important role in early recruitment of inflammatory cells and in the control of bacterial infection.⁴⁹  Expression of the TLR-2, TLR-4, and MyD88 genes were markedly increased in the colon of untreated septic mice compared with shams, while LGG or BL treatment significantly reduced these levels (fig. 9, A–C).

This study demonstrates for the first time that two different probiotic strains, LGG and BL, confer a significant survival benefit in weanling mice subjected to septic peritonitis. To our knowledge, this is the first description of improved survival after live probiotic therapy in a pediatric or adult polymicrobial sepsis model. This advantage of survival was associated with decreased bacteremia, decreased colonic apoptosis and increased colonic proliferation, decreased systemic and local expression of inflammatory cytokines, and reduced colonic expression of TLR-2/TLR-4 and MyD88.

Critical illness and its treatments (vasopressors, antibiotics, opiates, etc.) create a hostile environment in the gut by altering the microbiota, and thereby favor the growth of pathogens.⁵⁰  This is partially due to the loss of the beneficial lactic acid bacteria,⁵¹  which can inhibit the overgrowth of pathogens by producing bacteriocins, hydrogen peroxide, organic acids, ammonia and by increasing the competition for adhesion sites on intestinal epithelia.^52,53  In critical illness, enhanced virulence gene expression in bacteria, called quorum sensing, leads to aggressive bacterial behavior, toxin expression, and ultimately translocation into the gut wall and/or gut barrier dysfunctions. This impairment of the gut barrier can lead to gut-derived sepsis, progression to organ failure, and ultimately mortality in the critically ill patients.⁵⁰  Interestingly, the administration of beneficial probiotic organisms has been shown to enhance gut epithelial resistance to injury in in vitro models.⁵⁰  Consistent with these previous findings, our data demonstrate statistically significant reduction of bacteremia, improvement of colonic epithelial homeostasis, and enhancement of survival in animals treated with either probiotic strain compared with the untreated septic animals.

Existing clinical trials indicate that currently used probiotic strains may not be administered early enough to optimize benefits on prevention and therapeutic efficacy against severe infections and sepsis.^54–57  In our mouse model of sepsis, probiotics were administered early, at the onset of infection, to better imitate the clinical setting of early treatment to prevent progressive infection and sepsis (i.e., before or at onset of nosocomial pneumonia or postoperative abdominal sepsis).

The observed overgrowth of potentially pathogenic bacteria has been shown to cause intestinal cell apoptosis and disruption of epithelial tight junction permeability.⁵⁸  As previously shown, intestinal proliferation and intestinal epithelial apoptosis are altered in the CLP sepsis model.¹⁴  Our data are consistent with previous reports of reduced apoptosis with probiotics in intestinal injury.^34,36 

TLRs play a central role in the initiation of innate immune responses and in the development of subsequent adaptive immune responses to microbial pathogens.⁵⁹  In existing data from murine CLP models, TLR-2 and TLR-4 expressions are significantly up-regulated in multiple organs^60,61  including the intestine⁶²  when compared with sham mice. Our data are consistent with the hypothesis that down-regulation of TLR-2/TLR-4 via MyD88 in the colon of LGG- or BL-treated mice may play a protective role in attenuating the local and systemic inflammatory response and ultimately the pathophysiology of polymicrobial sepsis.

Limitations of this research include that we did not administer antibiotics as part of this initial evaluation of early probiotics to prevent progression of infection and sepsis. This was to ensure the adequate colonization of our probiotic therapy, and we plan to do future studies administering antibiotics after treatment with probiotics. Furthermore, time-point studies are being planned currently to examine the effects of both probiotic strains more closely and to explore other possible pathways involved in their protective mechanisms in sepsis. It is important to note that previous clinical trials that have administered antibiotics with probiotic therapy have still noted significant reduction of nosocomial infections (such as ventilator pneumonia) despite antibiotic therapy concomitant with probiotic therapy.²¹  Furthermore, killed probiotic organisms have been shown to be as clinically effective in some studies as live probiotic therapy.⁶³  Finally, in our statistical analysis, we did not adjust our analysis for multiplicity (i.e., multiple comparisons). Given our small sample sizes, and that these are the first experiments of this type reported, such adjustments would decrease the statistical power of our tests. This would lead to an increase in the probability of making a type II error, i.e., the failure to reject a false null hypothesis. In any experimental setting, there is always a balance in type I and type II errors. Presently, for this line of research, we are more concerned with missing a potentially valuable finding that we hope will lead to further investigation.

In conclusion, our data show that early therapy with either LGG or BL can reduce mortality and systemic bacterial translocation in experimental sepsis in weanling mice. These results indicate that potential mechanistic explanations for this clinical benefit may include reduced intestinal epithelial apoptosis and restoration of colonic epithelial cell proliferation, which may be mediated by Akt/protein kinase B signal transduction pathway. Another potential mechanism may be probiotic-mediated attenuation of local and systemic inflammatory response, mediated by down-regulation of TLR-2/TLR-4 signaling pathway via MyD88 in the colon.

We believe our data adds to existing clinical trial data supporting the potential efficacy of probiotics in critical illness. To this point, the design of larger scale clinical trials and/or more routine clinical administration of live probiotics to children and adults have recently been limited by concerns regarding safety and potential increased risk of infection from the administered probiotics. Although we did not specifically evaluate for bacteremia from our administered probiotic strains, we did observe a statistically significant reduction in overall bacteremia and mortality in our pediatric mice treated with live probiotic bacteria. Furthermore, the results of the recent American Health Care Research and Quality report on the safety of probiotic therapy in over 600 published clinical trials and case reports are reassuring with regard to the safety of probiotic administration,⁶⁴  although isolated adverse effects of probiotic administration have been reported.⁶⁵  One recent clinical trial studying probiotics in severe pancreatitis (the PROPATRIA trial) found an unexpected increase in mortality in probiotic-treated patients.⁶⁶  This trial was unique as it administered multiple strains of probiotic bacteria and prebiotic-like fiber via a postpyloric feeding tube (placed in the small bowel). This postpyloric method of administration was associated with an increase in small bowel necrosis, which was subsequently associated with death in a number of patients receiving the prebiotic fiber/probiotic mixture. It is possible that the postpyloric administration of this fiber/multiple probiotic strain mixture in pancreatitis patients may carry significant risk and should likely be avoided.⁶⁷  In addition, methodological and safety concerns regarding the conduct of this trial have been raised.⁶⁷  In any case, careful and appropriate safety monitoring in all future probiotic clinical trials should be conducted. Perhaps the time has come to proceed with carefully designed, carefully monitored, multicenter randomized clinical trials of probiotic therapy to attempt to reduce the risk of infection, sepsis, and mortality in critically ill children and adults.