Published online before print December 26, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Laboratory for Surgical Infection Research and Therapeutics, University of Chicago, Chicago, Illinois, USA; and
Martin Boyer Laboratories, Department of Medicine, University of Chicago, Chicago, Illinois, USA
1Correspondence: University of Chicago, Laboratory for Surgical Infection Research and Therapeutics, 5841 S. Maryland MC 6090, Chicago, IL 60025, USA. E-mail: jalverdy{at}surgery.bsd.uchicago.edu
ABSTRACT
Recent advances in the ability to genetically interrogate microbial communities within the intestinal tract of humans have revealed many striking findings. That there may be as many as 300 unculturable and unclassified microbes within the human intestinal tract opens the possibility that yet-unidentified microbes may play a role in various human diseases [1 ]. Technologically, the regional and spatial aspects of intestinal microbial communities can now be better appreciated by emerging genetic and in vivo imaging systems using a bioinformatics approach [2 ]. Finally, in situ PCR of tissues and blood now allows the detection of microbes at concentrations that would otherwise remain undetected by culture alone [3 ]. In the aggregate, these studies have empowered clinicians to readdress the issue of how our microbial partners are affected by extreme states of physiologic stress and antibiotic use through the course of critical illness. The role of microbes in systemic inflammatory states, such as systemic inflammatory response syndrome, as well as in primary intestinal mucosal diseases, such as necrotizing enterocolitis, inflammatory bowel disease, and ischemia-reperfusion injury, can now be more completely defined, and the microbial genes that mediate the immune activation during these disorders can be identified. The 2008 roadmap initiative at the National Institutes of Health to fully define the human microbiome is further testament to the power of this technology and the importance of understanding how intestinal microbes, their genes, and their gene products affect the course of human disease and inflammation.
Key Words: genomic screening • quorum sensing • pseudomonas • epithelial barrier
INTESTINAL MICROBES AND CRITICAL ILLNESS: TRANSLOCATION CARRIES THE TUNE, BUT IT IS THE MUCOSA THAT MAKES THE MUSIC
Observations over 30 years ago that microbial pathogens relocate from the intestine to the systemic circulation following experimental models of shock, ischemia, and hemorrhage raised great suspicion that the translocation event itself was a primary and driving force for the systemic inflammatory state. Survival studies in germ-free mice seemed to corroborate these findings, as survival was improved from shock and injury when no intestinal microbes were present [4 ]. Yet a major problem with these observations was that the quantity of bacteria that translocated was trivial (102–104 cfu) and failed to satisfy Kock’s postulates that if the same quantity and species of bacteria were to be injected systemically into untreated animals, the same inflammatory state and mortality outcome should be observed. More confusion has been raised in animal models in which systemic endotoxin administration is used to recapitulate the systemic inflammatory response syndrome (SIRS). The quantity of endotoxin needed to create significant inflammation in animals is excessive, and this amount has never been observed clinically. In trying to understand gut-origin sepsis, we have ignored the observation that for nosocomial pathogens that translocate from the gut into the systemic compartment (i.e., bloodstream), the average quantity of bacteria recovered (<103 cfu/ml) is insufficient to cause any systemic change in inflammation whatsoever [5 ]. For example, a typical blood culture with a nosocomial pathogen contains 30–100 cfu/ml. In animal models, a minimum of 108 cfu systemically administered Escherichia coli, Pseudomonas aeruginosa, or Staphylococcus aureus is needed to induce the sepsis syndrome, again an amount never seen clinically. In fact, is well established that several orders of magnitude less bacteria injected into the lung or gut are necessary to cause sepsis than when injected systemically [6 , 7 ]. This observation has led many researchers in this field to conclude that bacterial translocation itself is a surrogate marker for a high microbial burden at a site of infection or dense colonization but itself, of little clinical relevance. It may be for this reason that multiple studies in critically ill humans matched for severity of illness fail to demonstrate that a positive blood culture carries a higher risk of mortality than patients who are culture-negative [8 ]. In fact, in several studies, the presence of bacteremia carries a lower risk of death, all other things being equal, perhaps as it allows for targeted therapy sooner [9 ]. As most experienced clinicians will attest, most patients dying of severe sepsis have no clear, identifiable pathogen or source of infection at the time of death, yet most receive broad-spectrum antibiotics. It is therefore probably axiomatic to conclude that the location, mechanism, and degree of microbial burden that governs the sustained inflammatory response one sees during critical illness remain elusive.
The recently described system of virulence gene expression in bacteria, called quorum sensing (QS), could clarify much of the observed confusion in this area [10 ]. QS is a hierarchical system of virulence gene regulation in bacteria, in which bacterial virulence genes are expressed only after a critical bacterial density is achieved, presumably that amount necessary to overcome and overwhelm the host. The secretion and uptake of QS molecules provide a cell-to-cell communication system for bacteria to synchronize their behavior and coordinate a toxic offensive against their host should they perceive a threat to their survival or nutrient resources. In ideal growth conditions, bacteria constantly make a cost-benefit analysis in terms of virulence gene activation, and their virulence genes always turned off when nutrient supply is abundant, and there is no perceived threat to their elimination. Interestingly, in nutrient-rich broth, most bacteria do not release QS molecules to activate their virulence genes until they reach late exponential phases of growth, usually representing 107–108 cfu/bacteria [11 ]. In fact, it is virtually impossible to activate the QS system in bacteria below these threshold bacterial densities, thus accounting for the term "quorum sensing." Work from our lab has shown that soluble compounds released at local tissues sites, such as the intestinal epithelium during ischemia, hypoxia, or tissue injury, are potent activators of QS circuitry in the opportunistic pathogen P. aeruginosa. This is in direct distinction to the observation that whole blood from stressed animals actually suppresses the QS circuitry activation signal, suggesting that virulence activation by P. aeruginosa may be an asset in one tissue site (i.e., the intestinal epithelium) and a liability in another (the systemic compartment) [12 ]. From the standpoint of a bacterium colonizing an epithelial surface of a critically injured patient, this would be a logical and cost-effective strategy, given that the decision to invade and harm the host represents a fundamental tradeoff in that the bacteria must expend resources to contend with the immune system and the presence of antibiotics. Sense and respond circuitry such as the QS systems has been described for many pathogens and has likely evolved to be able to allow pathogens to coexist as colonizing pathogens one moment and switch to highly virulent pathogens the next. A synchronized program to kill its host could become activated using the QS signaling system once a microbial community senses that the physiology of its host has reached an evolutionarily, unprecedented level for survival, and/or the host environment is overtly hostile to the nutrient and growth potential of its microbial community.
This latter concept is best documented in a recent study from Japan, in which the fecal flora was cultured in conjunction with analysis of the intestinal short-chain fatty acids and pH levels (Fig. 1 ). This study is of particular relevance, as the care of severely injured patients involves the use of multiple interventions, which not only alter the defense functions of the gastrointestinal tract (peristalsis, pH, IgA, mucus production) but also cause a relative deficiency of preferred microbial nutrients, as most foods fed to these patients are completely absorbed before they reach the distal intestine or are administered i.v. (Fig. 2 ). Results of this study demonstrate three critical findings: a several log-fold decrease in the protective probiotic intestinal microflora including bifidobacteria and lactobacilli; the presence of highly pathogenic bacteria within the intestine, such as P. aeruginosa and S. aureus; and a dramatic decrease (>90%) in the concentration of short-chain fatty acids, which are the preferred respiratory fuels for epithelial cells and have been shown to induce the cytoprotective heat shock proteins (hsp) in epithelial cells [13 ]. Results of this study provide a stark example that the human intestinal environment of critical illness not only displays an imbalanced flora in favor of highly lethal pathogens but also is an environment scarce in the preferred nutrients, which bacteria and epithelia require for homeostasis.
 View larger version (44K): [in a new window] |
Figure 1. Reprinted with permission from Shimizu et al. [13
]. Fecal flora examined for the presence of probiotic/pathogenic bacteria, pH, and the concentration of free fatty acids in patients with SIRS. Marked imbalance in favor of pathogenic growth is observed, and protective probiotic organisms display several orders of magnitude decrease. Fecal fatty acid concentration is decreased by >75%.
|
 View larger version (29K): [in a new window] |
Figure 2. Schematic of the various factors that influence the altered intestinal microflora during severe stress and critical illness management.
|
Work from our laboratory in close collaboration with the Eugene Chang laboratory has demonstrated two important findings relevant to the above observations about the microflora composition that characterizes the critically ill patient. The first is that feared pathogens, such as P. aeruginosa, can cause a defect in the intestinal epithelium, resulting in lethal gut-derived sepsis that occurs, independent of bacterial dissemination, and the presence of beneficial probiotic organisms such as Bacillus subtilis can increase intestinal epithelial hsp abundance to a degree that renders the mucosal epithelium resistant to injury [14
, 15
]. Intriguingly, the opposing actions of these two colonizing bacteria are mediated by secretion of QS molecules that are specific to each organism. For example, in the case of the human opportunistic pathogen P. aeruginosa, this organism has a membrane biosensor termed OprF that binds IFN-
, the ligation of which results in activation of QS molecules, virulence expression, binding of the organisms to the epithelium, and epithelial barrier disruption [10
]. Activation of the OprF membrane protein by IFN- causes the release of the QS molecule C-4 HSL (homoserine lactone), which then dimerizes transcriptional activators that begin the cascade of virulence gene expression. This same OprF outer membrane porin has been demonstrated to facilitate binding to lung epithelial cells, demonstrating the bidirectional nature of signals between the host cells and microbes [16
]. That P. aeruginosa can recognize and respond to IFN- and activate its QS signaling system of virulence expression demonstrates how bacteria have evolved systems of countermeasures to host immune elimination that involve harmful counter-counter measures of their own. It may be for this reasons that clinical and animal trials of IFN- supplementation during infection have not resulted in enhanced survival [17
]. We have described a similar effect on P. aeruginosa virulence activation with another host factor that is secreted into the gut during ischemia reperfusion injury, adenosine, which is taken up into the cytoplasm of P. aeruginosa and converted to inosine, resulting in direct activation of the QS system [8
]. In addition, we have demonstrated that many of these host compounds released during stress can activate P. aeruginosa to produce a QS-dependent chemical, called 2-heptyl-4-hydroxyquinoline-N-oxide, which directly suppresses the growth of Lacotobacilli species [18
]. Therefore, another virulence response of opportunistic pathogens, when activated by host stress signals, is to suppress growth of the protective probiotic flora, further enhancing their ability to invade the host.
The importance of the probiotic flora has been demonstrated recently by the Chang laboratory in conjunction with our lab in experiments that demonstrate the mechanisms by which these organisms participate to promote health of the intestinal epithelium [11 ]. Using the probiotic organism B. subtilis, they demonstrated that this microbe secretes a QS molecule, called colony sporulation factor (CSF), which is taken up in epithelial cells by the carnitine transporter organic cation transporter-2 (OCTN2), resulting in a cytoprotective increase in hsp via a number of intracellular regulatory proteins (Figs. 3 and 4 ). A similar, protective effect of the intestinal microflora on host health has been demonstrated when the normal flora interacts with epithelial TLRs [19 ]. These latter two examples are in stark contrast to those in which following hemorrhagic shock, elimination of the intestinal microflora results in improved outcome [4 ]. The variance between the observation that the normal flora is useful one moment and a liability the next speaks to the complexity of the intestinal microbial community when it encounters a healthy host one moment and adversity the next. A final factor in this complex and fragile balance is the genetic polymorphisms in the immune system, which play an important role in the response to shock, infection, and tissue injury [20 ].
 View larger version (66K): [in a new window] |
Figure 3. The effect of CSF from B. subtilis, a QS molecule, on induction of cytoprotective hsp at the intestinal epithelial surface. CSF is transported by OCTN2 and activates multiple arms of the hsp induction system. p-Akt, Phosphorylated Akt.
|
 View larger version (60K): [in a new window] |
Figure 4. The effect of growth of pathogenic bacteria on the response of the mucosal epithelium. Activation of bacterial virulence genes in response to soluble compounds released by the host during catabolic stress alters the phenotype of bacteria to induce proinflammatory cytokine release by the intestinal mucosa. Absence of probiotic organisms further compounds this problem. GTPase, Guanosinetriphosphatase; pFAK, phospho-focal adhesion kinase; HMGB1, high-mobility group box-1.
|
We have created a unique model of in vivo virulence activation in using P. aeruginosa and a surgical injury model of a 30% surgical hepatectomy. Mice undergoing a 30% hepatectomy followed by direct injection of P. aeruginosa into the cecum have a near 100% mortality rate, whereas sham-operated, control mice injected with the identical strain and quantity of P. aeruginosa have 100% survival and show no signs of sepsis [21 ]. We discovered that soluble host compounds released into the intestinal lumen of mice following hepatectomy directly activate QS circuits, leading to expression of the adherence protein and epithelial barrier-dysregulating protein the PA-I lectin in P. aeruginosa. PA-I lectin alone causes an increase in tight junctional permeability, leading to systemic dissemination of the potent ribosylating toxin exotoxin A, causing systemic and lethal sepsis. In vivo expression of the PA-I lectin can be shown to occur as a direct result of soluble compounds released into the mouse intestine following surgical injury, thus demonstrating how colonizing pathogens can shift their phenotype from indolent colonizer to lethal pathogen in direct response to compounds released during physiologic stress [22 ]. In some cases, these host-derived bacterial signaling compounds bind to membrane biosensors (i.e., IFN) and activate the QS system [14 ]; in other cases, they diffuse into the bacterial cytoplasm (i.e., dynorphin) and act as surrogate bacterial QS molecules [18 ].
TRACKING THE HUMAN MICROBIOME AND INFLAMMASOME THROUGH THE COURSE OF CRITICAL ILLNESS: CLIMBING A COMPUTATIONAL MOUNT EVEREST
Currently, it is possible to use a 16-s RNA PCR-based analysis of the fecal flora to determine the diversity of the intestinal microflora without culture [1 ]. It is now possible to apply metagenomics (metabolism and genomics) to fully appreciate the metabolic function, diversity, phylotype, and speciation of the intestinal flora in a single fecal sample or in intestinal tissues using bioinformatics and computational approaches. At the same time, genomic analysis of the immune system could provide information about the state of the immune response relative to the microbial flora. However, how this information can be aligned with the spatial, regional, and tissue differences in the intestinal microflora along the entire 200 M2 of intestinal surface will present a formidable challenge. The need for signal-to-noise reduction will limit such an analysis, as will the ability to sample along various regions and depths (lumen, mucus, epithelium) of the intestinal tract. As such, attempts to comprehensively understand fluctuations in the intestinal microbiome that can be synchronized to those in the immune response will be a challenge. Tracking the bidirectional molecular chatter among the more than 400 bacterial species in the intestinal tract, their 5000–8000 genes, and the more than 30,000 human genes that they may activate and be activated by through the course of shock, tissue injury, and infections seems impossible. Yet, there are already several large genome-sequencing centers prepared to handle this amount of computational analysis, making what seems impossible a reality. As a first step in this process has been the use of PCR screening of blood for rapid microbial identification. Already, the challenge of this technique is to understand when the presence of microbial DNA in the blood represents a true infection. Perhaps a first step in this process will be to define a molecular signature in the blood that can identify when blood-borne microbial pathogens or their DNA fragments represent a real danger signal.
CONCLUSIONS
What causes gut-derived sepsis? Gut-derived sepsis, a rare but often fatal cause of SIRS in patients with severe injury or infection, occurs when the right set of microbes, expressing the right set of genes, interacts with the mucosal epithelium to activate and sustain the expression of the right set of proinflammatory mediators in a genetically susceptible host. Whether a short-lived exposure of the intestinal epithelium to such an array of microbial virulence gene products is enough to trigger and sustain a lethal proinflammatory response or whether continuous microbial epithelial signaling is necessary remains to be determined. The ability of highly lethal, nosocomial pathogens to inject their gene products into epithelial cells at an arm’s length from the immune system using a molecular syringe called the Type III secretion apparatus, coupled with their ability to produce and exist in antibiotic-impenetrable biofilms makes their mere presence on and within the intestinal epithelium a clear and present danger to the host [23 , 24 ]. Newer techniques in interrogating microbial communities within the intestinal tract using a genome-wide approach offer the possibility of understanding gut-derived sepsis at a more molecular level. Use of capsule endoscopy to sample the intestinal flora and the underlying mucosa might be possible someday to produce a rapid and preliminary profile of the balance of the microbial communities and the response of the mucosal immune system during critical illness.
Received June 9, 2007; revised September 10, 2007; accepted October 8, 2007.
REFERENCES
activation by Pseudomonas aeruginosa Am. J. Physiol. Gastrointest. Liver Physiol. 292,G134-G142 Shock 27,289-295[CrossRef][Medline]This article has been cited by other articles:
 |
|
 |
|
  G. D. Pullinger, S. C. Carnell, F. F. Sharaff, P. M. van Diemen, F. Dziva, E. Morgan, M. Lyte, P. P. E. Freestone, and M. P. Stevens Norepinephrine Augments Salmonella enterica-Induced Enteritis in a Manner Associated with Increased Net Replication but Independent of the Putative Adrenergic Sensor Kinases QseC and QseE Infect. Immun., January 1, 2010; 78(1): 372 - 380. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Sharma, C. Young, M. Mshvildadze, and J. Neu Intestinal Microbiota: Does It Play a Role in Diseases of the Neonate? NeoReviews, April 1, 2009; 10(4): e166 - e179. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. R. Wikoff, A. T. Anfora, J. Liu, P. G. Schultz, S. A. Lesley, E. C. Peters, and G. Siuzdak Metabolomics analysis reveals large effects of gut microflora on mammalian blood metabolites PNAS, March 10, 2009; 106(10): 3698 - 3703. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Lopez, N. Li, J. Kataria, M. Russell, and J. Neu Live and Ultraviolet-Inactivated Lactobacillus Rhamnosus GG Decrease Flagellin-Induced Interleukin-8 Production in Caco-2 Cells J. Nutr., November 1, 2008; 138(11): 2264 - 2268. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
