Michael S. Gilmore

The Problem: Antibiotic resistance in enterococci, streptococci and staphylococci has rendered many common and severe infections treatable only with last line drugs.

The emergence in the 1980's of vancomycin resistant enterococci among leading causes of hospital acquired infection, raised the specter of untreatable bacterial infection. Since then, enterococci have transferred vancomycin resistance to methicillin-resistant S. aureus , creating the potential for untreatable community acquired infection as well. The problem of antibiotic resistance in common causes of infection has emerged in the last few decades as a leading threat to public health. Therefore a main arm of our research focuses on the exchange among enterococci (and between enterococci and other genera) of auxiliary elements encoding antibiotic resistance and virulence traits; and determination of how these variable traits impact the virulence and ecology of the organism.

In addition to being leading causes of antibiotic resistant infection, the enterococci, streptococci and staphylococci are also highly adapted members of the human commensal flora. As a result, these organisms possess sophisticated mechanisms for colonizing human mucosal surfaces and skin, and interaction with the host immune system is finely balanced. Understanding the pathogenesis of infection caused by these bacteria requires an understanding of how host and microbial factors contribute to this fine balance. A second arm of our research has focused on identifying factors that undermine the host/microbe dynamic, leading to disease. The overarching goal of this research is to develop new strategies for preventing, mitigating or otherwise treating infections that are increasingly refractory to available antibiotics. Toward that end, we examine the pathogenesis of infection as a process, and work to identify the critical points on which outcome depends. The approaches used to identify critical points range from genomic studies to developing new models to examine the biology of infection.

Enterococci

We were part of the team that described the first vancomycin-resistant Enterococcus isolated in the US (1). This isolate was found to be closely related to another that caused a large infection outbreak in a hospital ward (2). Members of this lineage were shown to possess a number of previously unknown traits, including a capsule (3, 4, 5), and a pathogenicity island (6). We showed that the pathogenicity island harbored a new adhesin that contributed to colonization of the bladder and biofilm formation (6, 7, 8), in addition to the novel cytolysin toxin.

The enterococcal cytolysin increases virulence of enterococci by about 100 fold in murine peritoneal challenge models. We showed that it is novel both in terms of its structure and regulation. The toxin consists of two dissimilar, post-translationally modified, peptides. Both peptides include lanthionine-type modifications, and both are secreted through a dedicated transport channel. During secretion, both subunits are cleaved by a protease domain of the transporter. Outside of the cell, the subunits remain inactive until cleaved a second time by another dedicated protease – one that is encoded by the cytolysin operon, but is independently secreted via the secA pathway. This second cleavage of both subunits results in their being rendered competent to insert into membranes and effect target cell lysis (9).

In addition to being structurally novel, the cytolysin is regulated by a quorum sensing mechanism that is capable of sensing host cells in the environment. The smaller cytolysin subunit, CylL S ”, not only participates in target cell lysis, it feeds back to the producing organism as a quorum-sensing autoinducer (10). This feedback system requires two regulatory genes – one encoding a helix-turn-helix protein that binds to the cytolysin promoter, and a second encoding an apparent membrane protein. Neither have conserved homologs of known function.

We recently found that enterococci actively probe the environment, using a chemical type of sonar, to detect the presence of target cells for the cytolysin (11).   The larger toxin subunit, CylL L ”, has the ability to inhibit quorum sensing autoinduction by forming a stoichiometric complex with the smaller subunit, CylL S ” (the subscript designates the large ( L ) or small ( S ) toxin subunit, the double prime indicates that it has been twice modified during and after secretion). In the absence of target cells, CylL L ”-CylL S ” form a complex in solution that prevents CylL S ” from functioning as an autoinducer. In the presence of target cells, however, CylL L ” binds membranes, initially to the exclusion of CylL S ”, because of its higher membrane affinity, leaving a residual pool of free CylL S ” which can then induce high level expression of the cytolysin operon. In a slower reaction, CylL S ” enters the target cell membrane, and joins with resident CylL L ” to form an oligomeric pore, resulting in target cell lysis. This induction mechanism is unique in allowing the bacterium to sense the presence of target cells in the environment (11). These observations formed the subject of the American Society for Microbiology Division B Lecture (2006), along with invited talks at the Microbial Toxins and Pathogenicity (2006), and Bacterial Cell Surfaces (2006) Gordon Conferences.

The observation that virulent, multiple antibiotic resistant strains of enterococci possess a pathogenicity island encoding over 120 new traits that could affect niche selection – e.g., the cytolysin (which in addition to being a toxin is a bacteriocin active against gram positive bacteria), Esp (which contributes to biofilm formation), and a bile acid hydrolase, among others – suggested to us that colonization of the GI tract of hospitalized patients by multiple antibiotic resistant enterococci may be non-competitive with indigenous enterococci that lack this island (12). This prospect was reinforced by the observation that leading predisposing factors for multiple antibiotic resistant enterococcal infection include the prior use of antibiotic with little activity even for commensal enterococcal strains. We therefore hypothesized that multiple antibiotic resistant enterococcal infection is a two step process: The first step is asymptomatic infection of the GI tract through oral acquisition of small numbers of multiple antibiotic resistant enterococci from the contaminated hospital environment. Once in the GI tract, these few organisms multiply, potentially by many logs, as the result of filling niches vacated not by indigenous enterococci, but by highly antibiotic susceptible species – and these niches are inaccessible to indigenous enterococci because they lack the traits encoded by the pathogenicity island. With large numbers of antibiotic resistant enterococci in intimate contact with the host, then, the second step is symptomatic infection resulting from contamination of patient surfaces from the GI tract reservoir, and inadvertent introduction of organisms into sterile sites through catheters, suture lines, etc. (12). According to this model, two types of traits have contributed to the emergence of multiple antibiotic resistant enterococci as leading causes of nosocomial infection – traits that permit non-competitive GI tract colonization, and traits that facilitate or exacerbate infection. We therefore are working on strategies to prevent the initial colonization, and to mitigate infections once they occur. 

Future Directions

Our research stemming from this line of investigation is designed to: 1) identify other differences between commensal and pathogenic enterococci, by examining enterococcal species diversity using a comparative genomic approach, 2) determine whether the pathogenicity island functions as a genetic unit, and characterize the contribution of traits encoded within it in altering the ecology and pathogenicity of the microbe, and 3) determine the molecular mechanisms by which the cytolysin lyses prokaryotic and eukaryotic cells, and by which its expression is regulated. We   recently completed what we believe is the most comprehensive characterization of E. faecalis species diversity undertaken (in preparation). We examined over 100 strains, including strains isolated from the preantibiotic era as long as 100 years ago, strains previously used to develop serotype differentiation sets thought to represent the diversity of the species, strains from notable infection outbreaks, and strains from environmental sources. These strains were analyzed for phylogenetic relatedness by multilocus sequence typing (MLST), and the occurrence of pathogen relevant genes was determined. We then developed a custom Affymetrix microarray, and using the 6 deepest nodes from the resulting MLST unrooted cladogram, identified the core genome of the organism.

To assess the role of auxiliary elements, such as the pathogenicity island, in altering the ecology or pathogenicity of enterococci, we have developed several new models. The vast preponderance of enterococcal existence, and most of enterococcal evolution, occurs as a member of a complex consortium in the GI tract of man, animals and insects. Considering the highly competitive nature of this environment, it is counterintuitive that enterococci are fastidious in their growth requirements, being auxotrophic for about 20 amino acids, vitamins and other micronutrients. This strongly suggests that enterococci normally exist in the GI tract, not as individuals, but as part of a complex and interdependent web – and these auxotrophies serve to define its position within this web. As the pathogenicity island, in addition to harboring known virulence determinants, also was observed to include new biosynthetic and metabolic pathways, we hypothesize that strains possessing the island may have interdependences in the consortium that are different from commensal strains lacking the island. This would also grow out of the hypothesis that multiple antibiotic resistant, virulent enterococcal lineages colonize the GI tract by a mechanism that is not competitive with commensal enterococcal strains. We therefore are comparing the behavior of commensal enterococci and multiple antibiotic resistant virulent strains in a model of the complex GI tract consortium, in vitro. We have found that, depending upon how the consortium is perturbed, it can be either conducive or antagonistic to the outgrowth of these virulent enterococcal strains. Interestingly, and as predicted, commensal strains are not impacted in the same way, supporting the prospect that they fill a separate niche within this complex consortium. We are using FACS to identify organisms in this consortium that physically interact with commensal and clinical isolates of E. faecalis , and are using other strategies to enrich the population and identify organisms that specifically promote the outgrowth of, or are antagonistic for, various enterococcal strains.

We recently identified a tractable model that permits us to study the relationship between commensal colonization of the GI tract and pathogenesis of infection in vivo. The requirements for this model were that the host possess a GI tract that is natively colonized by enterococci, that the GI tract consortium of this host be simpler than that of man, that the host be housed and reared under highly reproducible conditions, and that the host itself be genetically tractable. We found that Drosophila fulfills each of these criteria. Interestingly, enterococci represent approximately 0.1 % to 1% of the Drosophila microflora, a level of colonization similar to that in man. Further, the most common species found to colonize the fruit fly was E. faecalis , with rarer colonization by E. faecium and E. durans – again, findings similar to humans. Despite these similarities, the Drosophila GI tract consortium was found to consist of 20 – 50 species, depending upon whether they were reared in the laboratory or in the wild. Finally, Drosophila are an extremely facile model – large numbers of organisms can be reared in a common, closed system; specific pathogen free or axenic flies can be reared easily and in abundance; and mutants exist in innate defense pathways that are common to man. We recently found that virulent strains of enterococci that are orally acquired by Drosophila result in substantial mortality within the population (submitted). This model will be used to examine elements of the pathogenicity island for their ability to influence the position of E. faecalis within the consortium, or for the ability to alter its virulence.

While studying new factors that are found to be important determinants of enterococcal ecology and virulence, we also continue to examine the structure, function and expression of the cytolysin in greater detail. Ongoing studies have identified a novel membrane metalloprotease that renders the producing E. faecalis strain immune to the cytolytic effects of the toxin. Membrane metalloproteases represent a new family of zinc metalloproteases characterized by a catalytic center that is imbedded within the lipid bilayer. In the case of the cytolysin immunity protease, it appears to function by specifically cleaving the large subunit of the toxin once it has inserted into the cytoplasmic membrane. This is a novel mechanism for bacteriocin resistance. We continue to investigate the regulation of the cytolysin operon. We know that two proteins are essential for regulation, and that each functions as a corepressor; mutation in either results in derepression. One protein, the helix-turn-helix binds the 5' end of the cytolysin promoter, and the structure of this protein has been solved (13, 14). The requirement of this protein for a second, integral membrane protein for function remains unexplained. We hypothesize that the membrane protein and the heliz-turn-helix protein form a ternary membrane complex with the cytolysin promoter, resulting in inhibition of transcription. This complex would then be relieved by binding of the small toxin subunit inducer. We are testing this model by fluorescently labeling the membrane component and the helix-turn-helix protein, and looking for colocalization and possibly fluorescent resonance energy transfer (FRET). Finally, as evolution selects for efficiency, we anticipate that the complex post-translational modification, secretion, and activation process, and probably also the immunity function, do not occur at random sites in and around the cell. Instead, we anticipate that there is a higher order architecture to a toxin “maturosome.” We are using standard protein-protein association strategies to collect data to test this hypothesis.

Staphylococci

In addition to studying enterococci, we also have an interest in S. aureus. S. aureus tends to be more overtly virulent than enterococci, it causes many hospital and community acquired infections, and it is only nominally less antibiotic resistant than enterococci. We were the first to show that 5 distinct genetic lineages accounted for most S. aureus infections (15).

The eye as a model    In looking for models to study the subtle pathogenesis of infections caused by enterococci, streptococci and staphylococci, we initially used chemical agents to render mice neutropenic. However, many controls were succumbing to environmental organisms because of the immune suppression. At about that time, the molecular basis of immune privilege in the eye began to emerge. We therefore tested the eye as an actively immune limited tissue for its ability to support enterococcal and staphylococcal infection at low inocula. We found that although it requires 10 9 cfu of a moderately virulent Enterococcus to achieve an LD 50 via the intraperitoneal route in mice, as few as 50 cfu injected into the vitreous of a rabbit resulted in an infection that consumed the organ in about 96 h (16). The advantages of this model are that it: 1) permits a small inoculum to be used, allowing for a quorum to develop in situ; 2) because the medium is clear and visible in a living animal, it permits direct, real time examination of the infection using the standard tools of ophthalmology [which in turn allows both the growing nidus to be visualized as well as its effects on the adjacent capillary bed, with resolution at a level that permits the observation of neutrophil migration]; and 3) provides for a perfectly matched control in the contralateral eye. This model allowed us to demonstrate that in the absence of cytolysin expression, E. faecalis infection resulted in substantial inflammation by no direct tissue damage. Moreover, this infection could be successfully treated with a combination of antibiotics and anti-inflammatory agents. However, if the strain expressed cytolysin, no therapeutic regimen mitigated the course of infection because of the direct toxin damage to the retina (16). As S. aureus is a leading cause of postoperative infections of the eye, as well as a leading cause of infection of the mucosal surface of the eye, we have adapted this, and other eye-based models, which give us unique insights into the pathogenesis of both enterococcal and S. aureus infection.

Using stereotaxis and microinjection to place S. aureus in the vitreous of mice, we found that mice genetically defective in activating complement as a result of a C3 deficiency (C57B6 C3-/-), in contrast to what had been shown using cobra venom factor (CVF) complement depleted mice, were no more susceptible to infection (17) than wild type. The sensitivity of this test showed that the substantial pleitropic effects of CVF complement activation resulted previously in a misestimation of the role of complement. This finding casts a significant shadow on previous studies that employed CVF to examine host-microbe interactions.  

In a second line of study, and again contrary to expectation, we found that Fas ligand plays an important positive role in host defense in immune privileged tissues. It was thought previously that the ability of Fas ligand to induce apoptosis in infiltrating immune cells was critical to the establishment of immune privilege in the eye. We therefore reasoned that Fas ligand deficient mice ( gld ), having a reduced ability to establish immune privilege, would be less susceptible to intraocular infection. We found the opposite is true – Fas ligand deficient animals were more susceptible to infection, and Fas ligand appears to play an important role in neutrophil activation at this site. This finding is consistent with the results of previous studies that aimed to display Fas ligand on transplanted tissues to enhance graft acceptance. Contrary to the universal graft acceptance that was hoped, Fas ligand displaying tissues led to rapid granuloma formation. These observations have led to a re-examination of the role of Fas ligand in contributing to immune privilege, and it is now believed that there are two forms (soluble and membrane bound) with distinct functions (17).

Future Directions

Because S. aureus is a leading cause of postsurgical infection, and because the visual tract permits direct observation of infection as well as objective assessment of organ function by electrophysiology (electroretinography), we continue to use this model to study the pathogenesis of S. aureus infection. We are taking advantage of the ability to directly observe these infections at the cellular level of resolution to develop (in collaboration with Charles Lin, PhD, Massachusetts General Hospital) a system for the continuous real time imaging of the evolution of an S. aureus infection over 72 – 96 h.

In addition to the above advantages of the eye for studying host-pathogen interactions, the cornea and conjunctiva represent the most accessible wet mucosal surfaces of any mammalian anatomical site. As most human infections involve interaction with and crossing a mucosal surface, accessibility is a great asset for probing host/pathogen interactions. Beyond serving as a model for examining how microbes bind and invade mucosal tissues, S. aureus infection of the ocular surface is a very common complication of trauma to the cornea in humans. We therefore used the cornea and cells derived therefrom, to examine how S. aureus binds and invades cells (18), and how the epithelium participates in the innate immune response by upregulating antimicrobial factors, chemokines, and adhesion molecules (19). We are currently examining how mucin gene expression modulates S. aureus binding to corneal epithelial cells (in collaboration with Ilene Gipson, Ph.D.), we are asking how S. aureus and its toxins may influence mucin gene expression and shedding by epithelial cells, creating sites for binding and cell invasion. Finally, since many more epithelial cells are influenced by toxins expressed at the site of infection, than are overtly killed by toxins, we believe that toxin-induced derangement of efficient clearance mechanisms may represent an important factor in the pathogenesis of infection at mucosal surfaces. We therefore are currently examining the effect of sublethal concentrations of S. aureus toxins on the behavior of human epithelial cells from several sites (in preparation).

Streptococci and biofilms

To this point, our research on streptococci has been limited to studies of gene expression in biofilms formed by the oral Streptococcus , S. gordonii . As oral biofilms form over years (although most models examined biofilms for hours or days), we decided to develop a simple model that permitted us to examine biofilms formed by S. gordonii over the period of a month. This led to a surprising observation that provided us with a new perspective on biofilms. The model we used consisted of farming a biofilm on a glass coverslip. This was done by simply emersing a coverslip in standard culture medium that had been inoculated with S. gordonii . At 24 h intervals, the coverslips would be rinsed to dislodge adventitiously associated cells, and the coverslip was placed in fresh sterile medium (we believed that in addition to being facile, this model better represented the cycles of nutrient availability experienced by oral biofilm microbes in nature than a continuous flow system). This feeding and rinse cycles were repeated every day for a month. Coverslips were examined daily by confocal microscopy, and viable cells were quantified by plating. Over the first week, we noticed that cell numbers adherent to glass increased as did biofilm thickness. However, shortly thereafter, we observed a profound crash of the population, with viable cfu being reduced by 3 logs, hitting a trough by day 10. To our surprise, this was then followed by a resurgence in the population, which within a few days eclipsed the previous maximum by 10 fold, reaching a plateau by day 17. This plateau was stable through the remaining 2 weeks of the experiment. When the confocal microscopic images were analyzed, it was noted that biofilms prior to the population crash, possessed only microstructure. In contrast, those following the crash appeared as billowing, coralline architectures (20). We deduced that a death program had been triggered in the population by an apparent quorum mechanism, but we had no evidence for a mechanism. The recent description of the induction of a bacteriocin and accompanying immunity gene by the S. pneumoniae competence peptide, and resulting “allolysis” of siblings, provided a possible mechanism for non-motile gram positive organisms to generate biofilms and their architectures through an orchestrated lysis of members of the population (21). The presence of DNA as a leading component of the matrix of many biofilms is now established. Its role in determining biofilm structure, and in protecting the biofilm from the immune system, remain to be established.

Reference for Key Observations

1. Sahm DF, Kissinger J, Gilmore MS , Murray PR, Mulder R, Solliday J, Clarke B. 1989.   In vitro susceptibility studies of vancomycin-resistant Enterococcus faecalis. Antimicrob Agents Chemother. 33(9):1588-91.  

2. Huycke MM, Spiegel CA, Gilmore MS . 1991. Bacteremia caused by hemolytic, high-level gentamicin-resistant Enterococcus faecalis . Antimicrob Agents Chemother. 35(8):1626-34.

3. Hancock LE, Gilmore MS . 2002. The capsular polysaccharide of Enterococcus faecalis and its relationship to other polysaccharides in the cell wall. Proc Natl Acad Sci U S A. 99(3):1574-9.

4. Hancock LE, Shepard BD, Gilmore MS . 2003. Molecular analysis of the Enterococcus faecalis serotype 2 polysaccharide determinant. J Bacteriol. 185(15):4393-401.

5. Hufnagel M, Hancock LE, Koch S, Theilacker C, Gilmore MS , Huebner J. 2004. Serological and genetic diversity of capsular polysaccharides in Enterococcus faecalis . J Clin Microbiol. 42(6):2548-57.

6. Shankar N, Baghdayan AS, Gilmore MS . 2002. Modulation of virulence within a pathogenicity island in vancomycin-resistant Enterococcus faecalis . Nature. 417(6890):746-50.

7. Shankar V, Baghdayan AS, Huycke MM, Lindahl G, Gilmore MS . 1999. Infection-derived Enterococcus faecalis strains are enriched in esp , a gene encoding a novel surface protein. Infect Immun. 67(1):193-200.

8. Shankar N, Lockatell CV, Baghdayan AS, Drachenberg C, Gilmore MS , Johnson DE. 2001. Role of Enterococcus faecalis surface protein Esp in the pathogenesis of ascending urinary tract infection. Infect Immun. 69(7):4366-72.

9. Coburn PS, Gilmore MS . 2003. The Enterococcus faecalis cytolysin: a novel toxin active against eukaryotic and prokaryotic cells. Cell Microbiol. 5(10):661-9.

10. Haas W, Shepard BD, Gilmore MS . 2002. Two-component regulator of Enterococcus faecalis cytolysin responds to quorum-sensing autoinduction. Nature. 415(6867):84-7.  

11. Coburn PS, Pillar CM, Jett BD, Haas W, Gilmore MS . 2004. Enterococcus faecalis senses target cells and in response expresses cytolysin. Science. 306(5705):2270-2.  

12. Gilmore MS , Ferretti JJ. 2003. The thin line between gut commensal and pathogen. Science. 299(5615):1999-2002.

13. Rumpel S, Razeto A, Pillar CM, Vijayan V, Taylor A, Giller K, Gilmore MS , Becker S, Zweckstetter M. 2004. Structure and DNA-binding properties of the cytolysin regulator CylR2 from Enterococcus faecalis. EMBO J. 23(18):3632-42.

14. Razeto A, Giller K, Haas W, Gilmore MS , Zweckstetter M, Becker S. 2004. Expression, purification, crystallization and preliminary crystallographic studies of the Enterococcus faecalis cytolysin repressor CylR2. Acta Crystallogr D Biol Crystallogr. 60(Pt 4):746-8.

15. Booth MC, Pence LM, Mahasreshti P, Callegan MC, Gilmore MS .   2001. Clonal associations among Staphylococcus aureus isolates from various sites of infection. Infect Immun. 69(1):345-52.

16. Jett BD, Jensen HG, Atkuri RV, Gilmore MS . 1995. Evaluation of therapeutic measures for treating endophthalmitis caused by isogenic toxin-producing and toxin-nonproducing Enterococcus faecalis strains. Invest Ophthalmol Vis Sci. 36(1):9-15.

17. Engelbert M, Gilmore MS . 2005. Fas ligand but not complement is critical for control of experimental Staphylococcus aureus Endophthalmitis. Invest Ophthalmol Vis Sci. 46(7):2479-86.  

18. Jett BD, Gilmore MS . 2002. Internalization of Staphylococcus aureus by human corneal epithelial cells: role of bacterial fibronectin-binding protein and host cell factors. Infect Immun. 70(8):4697-700.

19. Ruan X, Chodosh J, Callegan MC, Booth MC, Lee TD, Kumar P, Gilmore MS , Pereira HA. 2002. Corneal expression of the inflammatory mediator CAP37. Invest Ophthalmol Vis Sci. 43(5):1414-21.  

20. Gilmore KS, Srinivas P, Akins DR, Hatter KL, Gilmore MS . 2003. Growth, development, and gene expression in a persistent Streptococcus gordonii biofilm. Infect Immun. 71(8):4759-66.  

21. Gilmore MS , Haas W. 2005. The selective advantage of microbial fratricide. Proc Natl Acad Sci U S A. 102(24):8401-2.