The gut in trauma :

For more than 20 years, the gut has been hypothesized to be the "motor" of multiple organ dysfunction syndrome. As critical care research has evolved, there have been multiple mechanisms by which the gastrointestinal tract has been proposed to drive systemic inflammation. Many of these disparate mechanisms have proved to be important in the origin and propagation of critical illness. However, this has led to an unusual situation where investigators describing the gut as a "motor" revving the systemic inflammatory response syndrome are frequently describing wholly different processes to support their claim (i.e., increased apoptosis, altered tight junctions, translocation, cytokine production, crosstalk with commensal bacteria, etc). The purpose of this review is to present a unifying theory as to how the gut drives critical illness. Although the gastrointestinal tract is frequently described simply as "the gut," it is actually made up of (1) an epithelium; (2) a diverse and robust immune arm, which contains most of the immune cells in the body; and (3) the commensal bacteria, which contain more cells than are present in the entire host organism. We propose that the intestinal epithelium, the intestinal immune system, and the intestine's endogenous bacteria all play vital roles driving multiple organ dysfunction syndrome, and the complex crosstalk between these three interrelated portions of the gastrointestinal tract is what cumulatively makes the gut a "motor" of critical illness.

INTRODUCTION The gastrointestinal tract reservoir is the primary site of colonization of health care-associated pathogens and the site from which most pathogens disseminate to cause serious infections (1 – 5). In addition, the intestinal tract has been considered to be the key site for the emergence of antibiotic resistance and virulence expression among health care-associated pathogens that ultimately cause life-threatening sepsis (6 – 8). This situation is especially prevalent during prolonged critical illness, when the gut undergoes nearly complete ecological collapse owing to the selective pressures imposed by modern intensive care therapy, including multiple antibiotic exposure, provision of all nutrients exclusively through the intravenous route (total parenteral nutrition), and the use of vasoactive agents that alter the intestinal blood supply, acid-reducing agents, and opioids. As a result, a syndrome of gut-derived sepsis can be observed to occur late in the course of critical illness as microbes struggle to survive. Emergence of antibiotic resistance under such circumstances is proposed to occur in the gut due to these aggregate selective pressures. Thus, the term “late-onset sepsis” is being used to describe sepsis late in the course of critical illness involving multidrug-resistant (MDR) health care-acquired pathogens (9 – 14). A better understanding of the ecological perturbations that develop in the microbiota during critical illness, in terms of both composition and function, is crucial to prevent late-onset sepsis during the often-unanticipated prolonged course of recovery from extreme medical interventions such as organ transplant, cancer chemotherapy, and burn injury. Here we characterized the composition and function of microbial communities from several sets of fecal samples from patients suffering from prolonged critical illness. Our results indicate that prolonged critical illness results in near-complete disruption of the normal microbiota, the members of which are replaced by ultra-low-diversity communities of highly resistant pathogens whose virulent or nonvirulent behavior is dependent on the interactions between its members and provocative host factors (i.e., opioids). A more complete understanding of the threat of highly virulent and resistant pathobiomes that emerge as members of the normal microbiota disappear over the course of prolonged critical illness is warranted to develop new strategies to prevent late-onset sepsis in hospitalized patients. RESULTS 16S rRNA analysis of the bacterial composition in stool samples of ICU patients. (i) Phylum level. Stool samples from healthy volunteers demonstrated dominance of Firmicutes and Bacteroidetes among the four stool samples (H1, H2, H4, and H5). A predominance of Firmicutes and very low levels of Bacteroidetes were found in one (H3) stool sample. Proteobacterial abundance remained below 1% in healthy volunteers (Fig. 1A). In 50% of intensive care unit (ICU) patients (ICU1, ICU4, ICU6, ICU9, ICU11, and ICU15), either Proteobacteria or Firmicutes organisms were totally dominant from at least one time point of stool collection (Fig. 1B and C). In patients ICU1, ICU11, and ICU15, during their course of hospitalization, we observed dramatic changes in microbial composition, with Firmicutes being completely replaced by Proteobacteria. Proteobacteria-dominated communities that persisted during the course of hospitalization were found in the stool samples of patients ICU6 (Fig. 1B) and ICU4 (Fig. 1C). Importantly, Proteobacteria and Firmicutes dominated in the stool of both deceased (Fig. 1B) and discharged (Fig. 1C) ICU patients. FIG 1  The taxonomic composition of the gut microbiome at the phylum level determined by 16S rRNA analysis of stool samples collected from healthy volunteers (A), ICU patients dying with signs of severe sepsis (as indicated by black circles on the time line) (B), and ICU patients who had recovered (as indicated by green circles on the time line) (C). Dates of stool collection are displayed in numbered quadrants. (ii) Genus level. In the group of healthy volunteers, the microbial communities comprised ~40 bacterial genera (see Fig. S1 in the supplemental material), including Coprococcus (phylum Firmicutes [p_Firmicutes], family Lachnospiraceae [f_Lachnospiraceae]), Bacteroides (p_Bacteroidetes, f_Bacteroidaceae), Roseburia (p_Firmicutes, f_Lachnospiraceae), Prevotella (p_Bacteroidetes, f_Prevotellaceae), a nonidentified genus of f_Lachnospiraceae (p_Firmicutes), a nonidentified genus of f_Coprobacillaceae (p_Firmicutes), and Blautia (p_Firmicutes, f_Lachnospiraceae) as the seven most abundant. The microbial composition in critically ill patients was profoundly different from that in healthy volunteers. We observed decreased diversity in >80% ICU stool samples as seen by Chao1 index values below 50 and extremely low diversity in ~50% of ICU stool samples as seen by Chao1 diversity index values below 10 (Fig. 2). Unusually high abundances of Pseudomonadaceae were observed in the stool of ICU2 and ICU3 (10% and 30%, respectively) (see Fig. S2). Decreased diversity and domination by Escherichia and Enterococcus were found in seven patients (ICU5, ICU8, ICU10, and ICU12 to ICU15) (see Fig. S3). FIG 2  Chao1 diversity index of bacterial composition in fecal samples. Chao1 index values for stool samples of healthy volunteers (black circles) and ICU patients (colored circles) are ordered as follows (left to right): 1 to 5, healthy volunteers; 6 and 7, ICU1-1 and ICU1-2; 8 to 11, ICU2-1 to ICU2-4; 12 to 15, ICU3-1 to ICU3-4; 16 to 19, ICU4-1 to ICU4-4; 20, ICU5-1; 21 and 22, ICU6-1 and ICU6-2; 23, ICU8-1; 24, ICU9-1; 25, ICU10-1; 26 to 30, ICU11-1 to ICU11-5; 31, ICU12-1; 32, ICU13-1; 33, ICU14-1; and 34 and 35, ICU15-1 and ICU15-2. The horizontal dashed lines corresponding to Chao1 index values 50. Mean Chao1 index value, 33; median, 25; range, 1.46 to 100.24. The emergence of extremely low bacterial diversity (>90% abundance of one bacterial taxon) was determined in samples from five patients (ICU1, ICU4, ICU6, ICU9, and ICU11) (Fig. 3; see also Fig. S4 in the supplemental material). A predominance (>90% abundance) of Enterococcus (p_Firmicutes, f_Enterococcaceae), Staphylococcus (p_Firmicutes, f_Staphylococcaceae), and f_Enterobacteriaceae with an absence of p_Bacteroidetes was the most striking feature of this group. Another striking feature of this group was fast dynamic replacement of one dominated phylum by another (i.e., Firmicutes by Proteobacteria). The taxonomic composition in the microbiomes in the members of these groups was found to be similar to the composition of cultured bacteria (Fig. 3; see also Fig. S4), a unique situation that may develop only when the microbiome diversity is profoundly decreased. We also noticed the presence of the eukaryotic pathogen Candida in four of five patients. One of the five patients (ICU9, who was discharged after 10 days in the ICU) did not harbor Candida species. Reads corresponding to Enterococcus determined by 16S rRNA amplicon sequencing in the stool of ICU9 comprised 98.82% of all sequence reads and represented 2 cultivable Enterococcus species, MDR E. faecium and E. gallinarum (see Fig. S4). Blood cultures from this patient yielded Escherichia coli but were culture negative for Enterococcus. The other 4 patients (ICU1, ICU4, ICU6, and ICU11) were admitted to the ICU for longer than a month; for these patients, a temporal series of stool samples were acquired. Thus, we could correlate the 16S rRNA amplicon profiles with the clinical histories, antibiotic therapy, and microbiological culture data from nongastrointestinal sources (i.e., blood, sputum, and urine) and gastrointestinal sources (feces) for each of these patients (Fig. 3). FIG 3  Taxonomic and culturing composition of 2-member pathogen communities in correlation to antibiotic regimen, antibiotic resistance of stool isolates, and microbial data of cultured species from nonintestinal samples. (A) Patient ICU1. (B) Patient ICU6. (C) Patient ICU11. (D) Patient ICU4. Cultured species from sources other than the gut refer to species cultured from extraintestinal sources such as blood, urine, and lung. The arrows indicate the time points along the course of antibiotic exposure at which these organisms were cultured. Vanc, vancomycin; cipro, ciprofloxacin; difluca, fluconazole (Diflucan); AmphoB, amphotericin B; flagyl, metronidazole (Flagyl); Pip, piperacillin; TMP/SUX, trimethoprim/sulfamethoxazole; Quinupristin/dalf, quinupristin/dalfopristin; Staph, Staphylococcus. The 16S rRNA profiling is in agreement with culture analysis of stool samples in ICU patients harboring low-diversity pathogen communities. (i) Patient ICU1 (Fig. 3A). Patient ICU1 was admitted for a liver transplant but died from severe sepsis before undergoing transplantation following 8 weeks of ICU confinement. In this patient’s stool sample collected at time point 1 (i.e., ICU1-1), Enterococcus reads comprised 99.9% of the 16S rRNA sequenced reads. Culture analysis of the first stool samples (ICU1-1) presented a 2-member community of MDR E. faecium and Candida albicans. A retrospective analysis of the clinical history of this patient demonstrated that the urine and paracentesis fluid of this patient were culture positive for C. albicans and E. faecium (Fig. 3A). In the second stool sample (ICU1-2) taken 23 days later, Enterobacteriaceae and Enterococcus reads comprised 96% and 3.3% of the 16S rRNA reads, respectively. Culture results from the feces demonstrated a 4-member community consisting of MDR Serratia marcescens, MDR Klebsiella oxytoca, tetracycline-resistant (Tetr) E. faecalis, and C. albicans. At that time, the urine and blood of the patient were culture positive for C. albicans and Enterococcus. The bacterial members of pathogen communities were found to be resistant to multiple antibiotics (Fig. 3A). A retrospective analysis of the clinical history demonstrated that antibiotics were aggressively administered to treat the presumed infection with combinations of up to six antibiotics administered daily. C. albicans was sensitive to the fungal antibiotics used, but the bacteria present were resistant to several of the antibiotics used, including ciprofloxacin, piperacillin-tazobactam (Zosyn), and linezolid. The observation that microbial pathogens persisted in the gut suggests that they might have been a continuing source of infection. (ii) Patient ICU6 (Fig. 3B). Patient ICU6 was transferred from an outside hospital, where he was treated for a presumed pneumonia without any improvement. He was suspected of having a postprocedure esophageal perforation. The patient was placed on broad-spectrum antibiotics during his hospital stay because of mediastinitis and bilateral empyema. Throughout his stay, methicillin-resistant Staphylococcus aureus (MRSA) was cultured from his sputum as well as from his chest tube sites bilaterally and from his pleural fluid, and he eventually died of severe sepsis. Two stool samples (ICU6-1 and ICU6-2) collected 25 days apart were analyzed using 16S rRNA amplicon sequencing. In sample ICU6-1, 86.83% of sequence reads showed the presence of Enterobacteriaceae, while 12.9% showed the presence of Staphylococcus; these were confirmed by culture to represent K. pneumoniae and methicillin-resistant S. aureus. In sample ICU6-2, 97.78% of the 16S rRNA reads indicated the presence of Enterobacteriaceae, which was in agreement with culture results indicating that the reads exclusively represented K. pneumoniae. The eukaryotic pathogen C. albicans was cultured in both stool samples (ICU6-1 and ICU6-2). Despite the sensitivity to fluconazole, C. albicans persisted in the gut of patient ICU6 throughout treatment with multiple fungal antibiotics, including fluconazole. K. pneumoniae was found to produce extended-spectrum beta-lactamase, defining it as strain ESBL, and also to be resistant to multiple antibiotics. It was sensitive, however, to tobramycin and imipenem, but despite the patient being treated with these antibiotics, K. pneumoniae ESBL persisted in his gut during his ICU stay. C. albicans, MRSA, and K. pneumoniae were cultured from the pleural fluid and tracheal aspirates over the course of hospitalization in patient ICU6. (iii) Patient ICU11 (Fig. 3C). Patient ICU11 suffered from intestinal bleeding, requiring several operations to control the bleeding. The patient had several underlying comorbidities, including a nonischemic cardiomyopathy. Following the last operation, the patient developed progressive decompensation and multiorgan system failure and eventually died from severe sepsis after being confined to the ICU for 2 months. Stool samples were collected at 5 time points (Fig. 3C). ICU11-1 and ICU11-2 16S rRNA reads indicated 89.7% and 99.4% levels of Staphylococcus, respectively, which was in agreement with culturing analyses yielding only coagulase-negative (CoA−) staphylococcus. Stool samples ICU11-3 and ICU11-4 demonstrated the presence of Enterobacteriaceae, corresponding to 99.26% and 94.98% of sequence reads, respectively, which was in agreement with culture identifying K. pneumoniae. Later in the course of the patient’s illness (ICU 11-5), coagulase-negative staphylococci emerged again. The eukaryotic pathogen Candida glabrata persisted at all points of stool collections from this patient. CoA− Staphylococcus, K. pneumoniae, and C. glabrata were cultured from blood and urine at various times during the course of ICU confinement. Similarly to the cases described above, the gut microbial pathogens persisted regardless of their sensitivity or resistance to antibiotics applied to the patient. (iv) Patient ICU4 (Fig. 3D). Patient ICU4 was the only patient of the four with ultra-low-diversity pathogen communities who survived and was discharged after 40 days in the ICU. This patient was admitted to the ICU with a history of anaplastic astrocytoma, including 2 weeks of decreased oral intake, altered mental status, and fevers. The analysis of four stool collections at various time points (Fig. 3D) demonstrated the persistence of the same composition microbial community dominated by Enterobacteriaceae, comprising 99.8% to 99.9% of the 16S rRNA sequences, which was in agreement with the culture results showing the presence of Enterobacter aerogenes resistant to multiple antibiotics. Additionally, the eukaryotic pathogen C. albicans was persistently cultured from feces of patient ICU4 during the ICU stay. C. albicans was found to be sensitive to fluconazole and E. aerogenes to be sensitive to piperacillin-tazobactam and imipenem—antibiotics with which the patient was treated. Despite the sensitivity to these antibiotics, the community comprising C. albicans plus E. aerogenes persisted during this course of treatment (Fig. 3D). E. aerogenes was recovered from the bronchoalveolar lavage samples and endotracheal aspirates of this patient. To conclude, in all cases listed above (ICU1, ICU4, ICU6, and ICU11), there were similar striking features that included a strong agreement of 16S rRNA amplicon sequencing profiles with the bacterial composition shown by culture, an agreement of the results determined using microbes cultured from intestinal and nonintestinal sources, and a failure of aggressive antibiotic treatment to eliminate these pathogens regardless of their sensitivities to administered antibiotics. Antibiotic susceptibilities of all the cultured microbes from the patients in this study are displayed in Table S1 in the supplemental material. As might be predicted, a high frequency of antibiotic-resistant microorganisms colonizing the gut of critically ill patients was observed. Ex vivo analysis of potential virulence of low-diversity intestinal communities. Given the close correlation of 16S rRNA amplicon sequencing profiles to microbial composition revealed by culture, we decided to determine the virulence potential of these pathogen communities by reassembling them ex vivo and allowing C. elegans to feed on them. C. elegans normally feeds on bacteria as a food source, and therefore its response to pathogen communities can be viewed as a proxy for the virulence potential of the bacterium. Six communities were included in the analysis: ICU1-1 (C. albicans: Enterococcus faecium), ICU6-2 (C. albicans: Klebsiella pneumoniae), ICU4-4 (C. albicans: Enterobacter aerogenes), ICU11-2 (C. glabrata: CoA− Staphylococcus), ICU11-4 (C. glabrata: K. pneumoniae), and ICU9-1 (Enterococcus faecium: Enterococcus gallinarum). Based on results of C. elegans experiments, we made several important observations recorded below. (i) Commensal lifestyle of 2-member pathogen communities. In the first round of the experiments, to mimic the nutrient-limited environment typical of the gut of a critically ill host, we grew the individual members of the communities as well as assembled whole communities in nutrient-poor 0.1× TY liquid media (tryptone, 1 g/liter; yeast extract, 0.5 g/liter, potassium phosphate buffer, 0.1 mM [pH 6.0]) (referred to here as 0.1× TY). After 4 h of growth in 0.1× TY, prestarved and antibiotic-treated worms were transferred to the microbial cultures. Antibiotic treatment was used to eliminate all residual bacteria from the intestinal tubes of the worms. We found that all but one Candida species (C. albicans ICU4-4) caused high levels of mortality in C. elegans (Fig. 4). None of the individual bacteria demonstrated a high killing capacity, and they were able to attenuate the C. elegans mortality caused by the Candida species, resulting in a commensal assemblage. FIG 4  The commensal behavior of the 2-member pathogen communities. Data represent Kaplan-Meier survival curves of C. elegans fed on 2-member pathogen communities and their individual members (n = 10/plate, 3 plates/group). Data represent the cumulative results of 2 experiments. (ii) Replacement of a comember can switch commensal organisms to display pathogenic behavior. As seen in Fig. 5, the killing ability of the ICU1-2 community significantly exceeded the killing ability of the ICU1-1 community. This correlated with the morphology of C. albicans inside the community, with the yeast form being prevalent in the ICU1-1 sample and the hyphal form being prevalent in the ICU1-2 sample. These data confirm our previous demonstration that the ICU1-2 community consisting of C. albicans plus S. marcescens plus K. oxytoca plus E. faecalis induces a high level of mortality in C. elegans and mouse models (15). Therefore, the sudden replacement of a comember may provoke pathogenic behavior in a newly created community. FIG 5  Replacement of bacterial comember switches the commensal behavior to pathogenic. (A) Kaplan-Meier survival curves of C. elegans fed on ICU1-1 and ICU1-2 pathogen communities (n = 10/plate, 3 plates/group). Data represent the cumulative results of 2 experiments. (B and C) Scanning electron microscopy images of ICU1-1 (B) and ICU1-2 (C) grown for 20 h in 0.1× TY. (iii) Critical care therapy can break commensalism. As was previously demonstrated, critical care factors used for treatment of critically ill patients (i.e., steroids, hormones, and opioids) can induce bacterial virulence via interkingdom signaling through their interaction with quorum-sensing signaling systems (15 – 23). Opioids can also be released in the gut during physiological stress such as that seen following ischemia, which commonly occurs during critical illness (16). To determine if host-derived factors can affect the commensalism, we included a synthetic opioid (U-50,488) in the growth media. We have previously shown that this potent ligand of the kappa-opioid receptor acts similarly to the endogenous kappa-opioid dynorphin (16), which is released during intestinal ischemia. C. elegans was allowed to feed on bacteria grown under conditions of nutrient limitation (0.1× TY) and exposure to opioids (50 µM of U-50,488). In two communities, ICU6-2 and ICU11-2, the addition of 50 µM U-50,488 significantly increased the killing capacity of bacterial members K. pneumoniae ICU6-2 and CoA− Staphylococcus ICU11-2; accordingly, the virulence of each assemblage was shifted from a commensal behavior to a pathogenic one (Fig. 6). However, not all the pathogen communities responded with enhanced killing capacity during opioid exposure. There was no significant increase in the mortality of worms fed on ICU1-1, ICU11-4, ICU4-4, and ICU9-1 communities (Fig. 6; see also Fig. S5 in the supplemental material), a result similar to the baseline control of mortality seen with C. elegans fed on a nonpathogen food source such as E. coli OP 50 (see Fig. S6). FIG 6  Opioids can break the commensalism of 2-member pathogen communities. Kaplan-Meier survival curves of C. elegans fed on ICU6-2 (A), ICU11-2 (B), ICU11-4 (C), and ICU4-4 (D) 2-member communities in the presence of an opioid (50 µM U-50,488) (n = 10/plate, 3 plates/group; P 30 days (6 of 14). When possible, stool samples were collected at several time points during the ICU course. Demographic and clinical history. Demographic and clinical histories of ICU patients, including past medical history, admitting diagnosis, hospital course, therapy, especially antibiotic therapy, and discharge summary, were available in the patients’ charts in accordance with the IRB protocol. Among the 14 ICU patients, 10 were admitted to the ICU for a period of 20 days or longer. Of these 10 patients, 7 died despite continuous intensive care. The patients who died all died with signs of severe sepsis, although there was considerable variability in their comorbidities and underlying illnesses. 16S rRNA analysis of stool samples. The microbial composition of stool samples was analyzed by 16S rRNA V4 iTAG amplicon sequencing of DNA recovered from stool samples. A BiOstic Bacteremia DNA isolation kit from MoBio Laboratories was used to extract DNA. DNA was analyzed using next-generation Illumina sequencing. Reads containing low-confidence base calls (phred value, <20) were dropped, and the remaining 150-bp-long sequences were clustered into 97%-similarity operational taxonomic units (OTUs) using QIIME 1.6.0’s pick_otus.py script with default parameters (36, 37). OTUs composed of fewer than 10 sequences were discarded, and the remaining OTUs were assigned taxonomic annotations by searching representative members of each cluster against the Greengenes 12_10 (38) database using the RDP classifier (39). Sequencing depth among samples was normalized by rarefaction to 1,200 sequences per sample prior to calculating taxonomic relative abundances. Microbial-community diversity was calculated for each sample using the Chao1 alpha diversity metric, as implemented by the QIIME alpha_rarefaction.py script. The samples isolated from ICU patients (patients ICU1-1, ICU1-2, ICU2-1, ICU2-2, ICU3-1, ICU3-2, ICU4-1, ICU4-2, ICU4-3, and ICU4-4) were independently processed for 16S rRNA analysis at Michigan State University. DNA was extracted from the stool samples using a PowerSoil DNA isolation kit from MoBio Laboratories following the manufacturer’s instructions with the exception that Inhibitor Removal Solution from MoBio’s Ultra-Clean Fecal DNA isolation kit was added to the PowerBead Tubes along with solution C1. PCR amplification and purification were similar to those described by Rosenzweig et al. (40). The amplification and purification procedures differed in that samples were first amplified for 15 cycles using primers 577f and 927r without bar codes or adaptors, and then 2 µl of each reaction mixture was amplified for 15 more cycles with primers containing the bar codes and Roche adaptors. The thermocycler program for the first PCR was 95°C for 3 min and then 15 cycles of 95°C for 45 s, an annealing temperature of 44°C (determined from optimization experiments) for 45 s, and 72°C for 90 s, followed by elongation at 72°C for 7 min before the reaction mixture was held at 4°C. The thermocycler program for the second PCR was 95°C for 3 min and then 15 cycles of 95°C for 45 s, 56°C for 45 s, and 72°C for 90 s, followed by elongation at 72°C for 7 min before the reaction mixture was held at 4°C. Equal mass amounts of the amplicon libraries were mixed and submitted for sequencing to the Research Technology Support Facility at Michigan State University. The sequence results were obtained and processed using the tools on the Ribosomal Database Project’s Pyrosequencing Pipeline at http://pyro.cme.msu.edu/. Initial processing parameters were a forward primer maximum edit distance of 2, a reverse primer maximum edit distance of 0, a maximum number of N’s of 0, a minimum read Q score of 20, and a minimum sequence length of 300. Trimmed sequences averaging 331 bp were aligned and clustered at a distance of 3%, and the cluster results were reformatted to be read into R (41) as an OTU table. Representative sequences for each OTU were submitted to the RDP Classifier for identification and to the Decipher Find Chimeras tool of the University of Wisconsin—Madison at http://decipher.cee.wisc.edu/ using the Short-Length Sequences option. OTUs identified as chimeric sequences were removed from the data. Further analysis was done using the BioDiversityR (42) package within the R program. The 16S rRNA analysis of the aliquots of stool samples from the same patients was performed by the Argonne National Laboratories and Michigan State University with almost identical results, confirming extremely low diversity of the stool samples from patients ICU1 and ICU4. Isolation, identification, and antibiotic resistance of cultured microorganisms. Microbial species were isolated by plating bacterial communities on plates selective for Gram-negative (−) bacteria (MacConkey II), for Gram-positive (+) bacteria (Columbia CNA agar with 5% sheep’s blood [SB]) (Becton and Dickinson), for Pseudomonas aeruginosa (Pseudomonas isolation agar [PIA]), and for fungi (saturated dextrose agar with chloramphenicol and gentamicin) and on Trypticase soy agar with 5% sheep blood (TSA II 5% SB) for cultivating fastidious microorganisms. Gram (−) bacteria were identified by the use of a Vitek2 system (bioMérieux, Inc., Durham, NC). Gram (+) bacteria were identified by standard manual methods. Enterococcus species identification was performed by the use of a Vitek2 system. A positive germ tube test identified Candida albicans. Susceptibility testing was performed by testing on a Vitek2 system or by disk diffusion. Susceptibility testing of Gram-negative bacilli was performed by the use of a Vitek2 system or by disk diffusion. Susceptibility testing of MRSA and enterococci was performed by the use of a Vitek2 system. Susceptibility testing of the other Gram-positive cocci was performed using a combination of disk diffusion and E test strips. Susceptibility testing of Candida was performed using a Sensititre YeastOne MIC panel (TREK Diagnostic Systems Inc., Cleveland, OH). Susceptibility results were interpreted using Clinical and Laboratory Standards Institute (CLSI) guidelines. Caenorhabditis elegans killing assay. The 2-member whole communities were compared to their individual members for killing capacity by assessing mortality of C. elegans transferred into liquid microbial cultures as in our previously described experiments (15, 17). Nematodes of C. elegans N2 provided by the Caenorhabditis Genetic Center (CGC), University of Minnesota, were used in all experiments. Nematodes were prepared as follows: synchronized nematodes (larval stage 4 [L4] to young adults) were transferred from E. coli OP 50 stock plates onto plain agarized plates, followed by a second transfer onto new plain agarized plates (Falcon) (60-mm diameter). Next, 1 ml of 100 µg/ml kanamycin was poured on the agar surface; after 3 h, worms were retransferred into experimental microbial cultures. Microbial cultures were prepared as follows: microbes from frozen individual stocks were plated on Trypticase soy broth (TSB) solid plates and incubated at 37°C overnight. Cells grown overnight were used to prepare stock microbial suspensions in 0.1× TY liquid media (tryptone, 1 g/liter; yeast extract, 0.5 g/liter, potassium phosphate buffer, 0.1 mM [pH 6.0]) (referred to here as 0.1× TY) to reach an absorbance of 1.0 at an optical density of 600 (OD600). Stock microbial suspensions were diluted in fresh 0.1× TY to an OD600 of 0.1 to prepare individual microbial cultures. To prepare community cultures, the respective stocks were mixed to reach an OD600 of 0.1 for each individual microbe in the mixture. In the experiments performed with Pi-PEG15-20, the stock microbial suspensions and final microbial cultures were prepared in 0.1× TY supplemented with 10% Pi-PEG15-20, and the pH was adjusted with KOH to that of 0.1× TY (pH 5.8). Microbial cultures were grown individually and as a mixture for 5 h at 37°C with shaking at 200 rpm. Then, 2 ml of the microbial culture was poured into a 35-mm-diameter empty sterile plate and adjusted to room temperature, and 5 to 10 worms were transferred into each plate. For each variant, 3 to 5 plates were prepared. C. elegans were fed on microbial cultures for up to 72 h under static conditions at room temperature (~23°C). The mortality of worms was followed dynamically. SUPPLEMENTAL MATERIAL Figure S1 The taxonomic composition of the gut microbiome at the genus level determined by 16S rRNA analysis of stool samples collected from healthy volunteers. The top 7 genera, consisting of Coprococcus (no. 36, p_Firmicutes, f_Lachnospiraceae), Bacteroides (no. 6, p_Bacteroidetes, f_Bacteroidaceae), Roseburia (no. 40, p_Firmicutes, f_Lachnospiraceae), Prevotella (no. 8, p_Bacteroidetes, f_Prevotellaceae), a nonidentified genus of Lachnospiraceae (no. 32, p_Firmicutes, f_Lachnospiraceae), a nonidentified genus of Coprobacillaceae (no. 62, p_Firmicutes, f_Coprobacillaceae), and Blautia (no. 35, p_Firmicutes, f_Lachnospiraceae), are shown. Download Figure S1, PDF file, 0.04 MB Figure S2 The taxonomic composition of the gut microbiome in stool samples collected from patients ICU2 and ICU3. Pseudomonadaceae (no. 77) reached levels of 10% (ICU2-4) and 30% (ICU3-4). Download Figure S2, PDF file, 0.2 MB Figure S3 The taxonomic composition of gut microbiome in stool samples with decreased bacterial diversity. Escherichia (no. 74) and Enterococcus (no. 20) were major pathogens in this group. Download Figure S3, PDF file, 0.05 MB Figure S4 Taxonomic and culturing composition of the ICU4 communities in correlation to antibiotic regimen, antibiotic resistance of stool isolates, and microbial data of cultured species from nonintestinal samples. Download Figure S4, PDF file, 0.04 MB Figure S5 Kaplan-Meier survival curves of C. elegans fed on 2-member communities whose response to an opioid (50 µM U-50,488) was negligible (n = 10/plate, 3 plates/group; P < 0.01). Data represent cumulative results of 2 experiments. Download Figure S5, PDF file, 0.1 MB Figure S6 Kaplan-Meier survival curves of C. elegans fed on E. coli OP 50 with and without exposure to an opioid (50 µM U-50,488) (n = 10/plate, 3 plates/group). Download Figure S6, PDF file, 0.04 MB Table S1 Antibiotic susceptibility of microbes isolated from ICU patients and healthy individuals. Table S1, DOCX file, 0.04 MB.

Sepsis remains one of the leading causes of death in burn patients who survive the initial insult of injury. Disruption of the intestinal epithelial barrier has been shown after burn injury; this can lead to the translocation of bacteria or their products (e.g., endotoxin) from the intestinal lumen to the circulation, thereby increasing the risk for sepsis in immunocompromised individuals. Since the maintenance of the epithelial barrier is largely dependent on the intestinal microbiota, we examined the diversity of the intestinal microbiome of severely burned patients and a controlled mouse model of burn injury. We show that burn injury induces a dramatic dysbiosis of the intestinal microbiome of both humans and mice and allows for similar overgrowths of Gram-negative aerobic bacteria. Furthermore, we show that the bacteria increasing in abundance have the potential to translocate to extra-intestinal sites. This study provides an insight into how the diversity of the intestinal microbiome changes after burn injury and some of the consequences these gut bacteria can have in the host.