Theodore E. Woodward Award: Host-Pathogen Interactions in Community-Acquired Urinary Tract Infections

Abstract

Community-acquired urinary tract infections continue to be a significant source of morbidity and health care costs. In addition, community-acquired urinary tract infections are an excellent model for studying the interaction of the infecting bacteria and the human host. This review focuses upon five recent areas of progress in understanding host-parasite interactions in urinary tract infection. First, uropathogenic E. coli have been recognized as a specific pathogenic group of organisms characterized by the presence of pathogenecity islands, horizontally-acquired genes encoding various pathogenic phenotypes, including fimbria, other adhesins, lipopolysaccharide, the polysaccharide capsule, various toxins and hemolysins, and siderophores. Second, recent studies indicate that the epithelium plays an active role in the innate host defense against urinary infection, including secretion of chemokines and cytokines, apoptosis of epithelial cells and exfoliation of bacteria-laden epithelial cells. Third, studies in animal models indicate that uropathogenic E. coli invade epithelial cells, forming intracellular communities of organisms and eventually biofilms. These intracellular organisms may persist and form a reservoir from which recurrent infection may develop. Finally, observations suggest that some women may be genetically predisposed to recurrent urinary tract infection and the genes predisposing to recurrent infection are being sought. These new discoveries have improved our understanding of the pathogenesis of these infections and will eventually lead to improved interventions for prevention and therapy.

Introduction

Community-acquired urinary tract infections occur at a rate of more than 7 million cases per year in the United States (1). The incidence of these infections among young sexually active women has been determined in one prospective study to exceed 0.5 episodes per year (2). In addition, 20% to 30% of women have frequent recurring infections at a rate of two or more per year which require medical attention and antimicrobial treatment. The average duration of symptoms associated with acute bacterial urinary tract infections is six days, with 2.4 days of restricted activity and 1.2 days of time lost from work (3). Upper urinary tract infections, namely acute pyelonephritis, are estimated to occur at a rate of approximately 300,000 per year in the United States (1). It is also estimated that 4 million urinary tract infections occur in pregnant women each year, with associated potential morbidity for both the mother and infant (4). Urinary tract infections are a frequent reason for antimicrobial use and as such may foster the increasing occurrence of antimicrobial resistance among E. coli in the community (5). Indeed, community-acquired urinary tract infections due to E. coli represent a subset of the overall fecal flora and thus are a convenient means of monitoring prevalences of antimicrobial resistance in a given community. Finally, due to their relative accessibility and frequency, community-acquired UTI are an excellent model for studying the interaction of the infecting bacteria (typically, E. coli) and the human host since one can acquire the infecting bacteria, monitor their gene expression, sample host epithelial and inflammatory cells and determine their responses to infection by collecting voided urine.

In this review, I will focus on five recent areas of interest in terms of host-parasite interaction in urinary tract infection. These include: (1) the recognition and description of urovirulence characteristics of uropathogenic E. coli (UPEC); (2) the recognition that the epithelium plays an active role in the innate host defense mechanisms employed to control UPEC infection; (3) the finding in animal models that uropathogenic E. coli (under appropriate circumstances) are capable of resisting host defense mechanisms by forming intracellular communities of organisms and biofilms; (4) the observation that some women may be genetically predisposed to urinary tract infections and hence, have an underlying reason for their pattern of frequent recurrent infection; and finally (5), that the new discoveries outlined above have given rise to an alternative hypothesis regarding the mechanism of recurrent infection thought to be operative in most young women, namely that not all recurrences need necessarily be exogenous reinfections from the vaginal reservoir but in some cases may result from persistent organisms within intracellular communities or biofilms in the bladder itself (6).

Uropathogenic E. coli (UPEC)

Approximately 80% of all community acquired urinary tract infections are caused by one bacterial genus, namely E. coli (7). The E. coli responsible for uncomplicated urinary tract infections originate in the gut flora, colonize the vaginal introitus and urethra and subsequently ascend into the bladder where they produce infection. In some cases, they may ascend further to reach the kidney and produce acute pyelonephritis. Most studies of host-pathogen interaction within the urinary tract have thus focused upon E. coli. Uropathogenic E. coli are selected E. coli clones that have evolved specific capabilities of persisting in the fecal flora, colonizing the vaginal introitus and subsequently infecting the urinary tract (8). These groups of uropathogenic E. coli were initially recognized as selected clones based on their serological characteristics (O, K, and H antigens). However, the factors that actually distinguish them from garden-variety fecal E. coli are primarily multiple genetic determinants that these strains have acquired through horizontal transfer and then retained. These determinants are present in pathogenecity islands on the chromosomes of these organisms (9). Uropathogenic E. coli, in contradistinction to normal fecal strains of E. coli, are strongly associated with particular disease entities. Thus, 20–30% of fecal strains but 50–60% of cystitis strains and 75–100% of strains from women with acute pyelonephritis or urosepsis exhibit the genetic determinants characteristic of uropathogenic E. coli (10). Conversely, women with complicated urinary tract infections associated with catheterization or urologic abnormalities are generally infected with strains that do not have these virulence determinants and seemingly are capable of causing infection largely due to the presence of impaired host defense mechanisms.

The virulence factors characteristic of UPEC strains include both bacterial surface structures and secreted proteins (11). A critical set of surface structures are fimbria or pili (Table 1). These hairlike appendages enable E. coli to adhere to the uroepithelium and matrix. While at least five separate types of fimbria have been recognized in UPEC, the two most important are Type 1 (or mannose sensitive) fimbria and P (or mannose resistant) fimbria. Other surface structures that are characteristic of UPEC strains include flagella; the polysaccharide capsule which is antiphagocytic and plays a role in serum resistance; lipopolysaccharide, which mediates endotoxin effects, liberates cytokines and also participates in serum resistance; and other outer membrane proteins (8).

TABLE 1.

Virulence Factors in Uropathogenic E. coli: Surface Structures

Virulence Factor	Function
Type 1 fimbriae	Adherence (epithelium, matrix), invasion, biofilm
P fimbriae	Adherence (epithelium, matrix), cytokine secretion
S fimbriae	Adherence (epithelium, endothelium, matrix)
F1C fimbriae	Adherence (epithelium, endothelium)
Curli	Adherence (epithelium, matrix), biofilm
Flagellum	Motility, fitness
Capsule	Antiphagocytic, anticomplement, serum resistance
LPS	Endotoxin effects, cytokines, serum resistance
OPMS	Receptor, transport

Additional virulence determinants seen in UPEC strains are exported molecules (Table 2). These include hemolysins, cytotoxins, and iron binding proteins (or siderophores). At least three different toxins (α-Hemolysin, CNF-1 and SAT-1) and at least three different groups of siderophores have been recognized in uropathogenic E. coli (Table 2). The genes for these molecules, accounting for the uropathogenic properties of E. coli, are acquired horizontally from other organisms. While most UPEC strains contain many of the same horizontally acquired virulence determinants, each strain may differ in terms of the precise urovirulence determinants it demonstrates, and the number of pathogenecity islands it has (8). In addition, pathogenecity islands may be distributed differentially throughout the chromosome. The repertoire of pathogenicity islands each strain possesses likely determines the nature of the clinical presentation and pathology that particular strain produces.

TABLE 2.

Virulence Factors in Uropathogemic E. coli: Exported Factors

Virulence Factor	Function
α-Hemolysin	Cytotoxicity, hemolysin
CNF-1	Interference with phagocytosis
SAT-1	Cytotoxicity
Enterobactin	Siderophore
Aerobactin	Siderophore
Yersiniabactin	Siderophore

The Role of the Epithelium in Host Response to Urinary Tract Infection with UPEC Strains

The bladder and kidney epithelium plays an active role in the innate immune response to invading uropathogens. The two major adhesins, as mentioned previously, are the pap (or P) fimbria and the Fim or Type-1 fimbria. Type-1 fimbria are present on almost all E. coli strains, whether they are uropathogenic or not (8). Thus, they are ubiquitous and are not epidemiologically associated with UTI. However, expression of the fim genes is induced immediately upon bladder entry (8). Fim H, the adhesin protein at the tip of the Type-1 fimbrial structure, binds to uroplakin (a glycosylated glycoprotein) on the surface of the bladder epithelium (8). Fim H negative isogenic mutants do not infect animal models of urinary tract infection (8). The consequences of UPEC attachment mediated by Fim H include bacterial adherence, induction of urothelial cell exfoliation (a means of ridding the bladder of infecting organisms) and internalization of the organism to form intracellular bacterial colonies (12).

The Pap or pyelonephritis associated pili, as the name implies, are strongly associated with strains causing pyelonephritis or urosepsis (10). These strains utilize the Pap G adhesin to attach to digalactose moieties embedded in glycosphingolipids on the surface of the uroepithelium (8). Binding via this adhesin stimulates the secretion of IL-6 and IL-8 from epithelial cells. Secretion of IL-6 and IL-8 is also induced in part by bacterial LPS, which binds to TRL4 on epithelial cells (8). Pap G binding also mediates attachment to the renal epithelium and initiates pyelonephritis, probably due to the denser concentration of glycosphingolipids in renal epithelium.

In summary, it is clear that the uroepithelium plays a key role in interaction with uropathogenic bacteria. The epithelium initiates two host defense mechanisms, namely (1) secretion of cytokines and chemokines, with subsequent recruitment of leukocytes into the mucosa and (2) exfoliation of infected or colonized cells. However, the epithelium may not always be successful in totally repelling the bacterial challenge with subsequent development of intracellular bacteria which are able to evade the host response by invading epithelial cells and forming microbial communities (12).

Intracellular Invasion of Uropathogenic E. coli

In animal models and cell cultures, uropathogenic E. coli can be seen to induce their own uptake into quiescent uroepithelium. Subsequently, it has been shown by Hultgren and coworkers that bacterial communities termed pods are present on the surface of experimentally infected mouse bladders (13). Within these pods are large numbers of E. coli embedded in a polysaccharide matrix coated by uroplakin. Inside these pods, transmission electron microscopy demonstrates individual organisms surrounded by matrix polysaccharides and a clear space (13). A specific maturational process appears to underlie the formation of an intracellular bacteria-filled pod. Initially, intracellular organisms are seen as single bacilli or as small numbers of bacteria on time lapse video fluorescence microscopy (14). Subsequently, with the replication time of the bacterium being approximately 30 minutes, a small cluster of bacteria is seen intracellularly. Eventually, the organisms become smaller and more coccoid in appearance and a pod-like structure is formed (14). Over time, it is also evident that some organisms leave the pod through a fluxing mechanism (15). This enables the UPEC strains to infect either adjacent cells or to be shed in the urinary stream. It is also evident that some of these organisms are filamentous during the fluxing stage, appearing as long, thin non-septated bacteria that resist phagocytosis (15).

Mechanism of Recurrence

A new paradigm underlying recurrent UTI can be proposed based on the studies just summarized. Namely, that UPEC organisms may bind to the uroepithelium, invade superficial uroepithelial cells, induce the host mechanisms and responses outlined above but despite them form intracellular bacterial colonies and eventually biofilms (15). These intracellular colonies may be quiescent but represent a reservoir that can reactivate and produce recurrent infection at a subsequent time through fluxing and filamentation of the organisms out of mucosal pods. Thus, intracellular uptake of uropathogenic E. coli is a Fim H-dependent process, and its rate and frequency also depend upon the animal host that is utilized. A developmental process then results in an intracellular pod-like biofilm structure that provides UPEC a refuge from antibiotics and may allow organisms to subvert innate host mechanisms and serve as a source of recurrent urinary tract infection. While the data to date supporting this hypothesis are largely derived from animal models, ongoing studies in humans are assessing whether this mechanism also may be present in women with recurrent urinary tract infections.

Genetic Determinants of Urinary Tract Infections

It has been known for some time that women with recurrent urinary tract infections are more likely to be nonsecretors of blood group antigens than are control women (16). These women also exhibit 2–3 fold greater adherence of uropathogenic E. coli to their vaginal epithelial cells as compared with cells from normal women (17). Finally, a history of UTI in mothers or sisters is more common among women with a history of recurrent UTI or pyelonephritis than among those without urinary tract infections (18). Taken together, these pieces of evidence suggest that there may be underlying mechanisms accounting for an inherited predisposition to recurrent urinary tract infection. In studies conducted by Ann Stapleton and colleagues (19,20), it was possible to collect vaginal epithelial cells from women with recurrent infection and from controls and to then extract gangliosides, the suspected receptors on vaginal epithelial cells for the adhesins of P pili. The gangliosides were then separated by extraction and fractionation from the epithelial cells and underwent high resolution thin layer chromatography. In a bacterial overlay assay, E. coli R45, a uropathogenic strain labeled with S-35 methionine, was used in an audoradiography assay to identify the specific glycolipids to which the bacteria could bind. It was shown that specific gangliosides extracted from vaginal epithelial cells of nonsecretors but not those of secretors were a target to which the radiolabelled E. coli R45 bound. The identity of the target gangliosides was surmised to be sialosyl gal-globoside (SGG) and disialosyl gal-globoside based on co-migration of these molecules with controls on the chromographic analysis. Subsequently, it was possible to isolate various gangliosides from the kidney and demonstrate that uropathogenic E. coli strains bound avidly to SGG, which is present in nonsecretor but not secretor epithelial cells. This mechanism may underlie the apparently increased risk of uncomplicated urinary tract infections among nonsecretors, accounting for an inherited susceptibility to UTI.

Concluding Remarks

The study of uncomplicated urinary tract infections caused by E. coli has provided an excellent model allowing investigators to unravel the interactions of uropathogenic E. coli (which have acquired specific genes to facilitate infection) with the host epithelium and the innate immune response that often successfully repels such infections. Studies done primarily in animals suggest that uropathogenic E. coli are capable of intracellular infection and subsequent development of biofilms. These characteristics may allow the infecting organism to resist host defense mechanisms, including induction of a leukocyte response, chemokine and cytokine secretion, and epithelial cell exfoliation. Persisting intracellular bacterial colonies may be responsible for some recurrences, which would necessitate rethinking how recurrent infections should be managed from a clinical perspective. A joint collaborative study undertaken by Mac Hooton, myself and colleagues at the University of Washington and Scott Hultgren and colleagues at Washington Saint Louis University is prospectively following a cohort of women to monitor recurrence rates, determine the virulence characteristics of UPEC strains causing recurrent infection, assess the innate host defense mechanisms that are activated and seek to identify intracellular biofilms of E. coli in exfoliated epithelial cells. In addition, investigators at the University of Washington are in the middle of a study comparing women with recurrent urinary tract infection and/or pyelonephritis versus control women for alterations in specific target genes, namely, toll-like receptors, chemokine receptors (CXCR1 and CXCR2), interferon gamma receptors and secretor status. These studies will help us to better understand host pathogen interactions in the urinary tract and perhaps devise novel prevention strategies.

DISCUSSION

Luke: Cincinnati: Is it possible that there will be a readily available clinical test that will tell us how long to treat these ladies?

Stamm: Seattle: Well, that’s a good question. The last attempt to do that utilized the antibody-coated bacteria test, which failed. At this point in time, there’s nothing on the horizon so far as I can see. Looking at the uropathogenic properties of the organism doesn’t help in terms of selecting length of therapy as far as we can tell.

Luke: In follow-up, you said that pathogenecity was associated with a certain marker which is associated with pylonephritis. If you knew that was there in the urine that might suggest longer treatment.

Stamm: Right, agreed, but specificity of these markers is not high since many cystitis strains carry them as well.

Cohen: North Carolina: My question has to do with—you’ve made so many observations about method of attachment over the years, I can’t believe you aren’t developing a therapeutic intervention that embraces that knowledge. Because the antibiotics are a mess and the poor people are taking them forever, I am trusting that you are going to reveal the secret right now.

Stamm: That’s a very important and pertinent question. But it hasn’t been as easy as you might think to do that. The most concerted effort to do that really was undertaken by Medimmune and Scott Hultgren. They tried to develop a vaccine based on the Type I adhesin and chaperone proteins. Basically, although it looked promising in animals (in primates and mice), it did not seem to protect humans. The reason for lack of protection seemed to be that the vaccine didn’t induce much antibody in urine. And so we are back to the drawing boards with trying to amplify the production of antibody in the urine.

Bartlett: Baltimore: I wanted to ask you about the attachment. Attachment of bacteria to cell surfaces is essential to infection as shown with many infections like endocarditis and strep throat. Has a genetic predisposition been shown with any other infection?

Stamm: Yes, its association was actually first demonstrated with meningococcal and gonococcal infections.

Tweardy: Houston: I was wondering if you could extrapolate to complicated or catheter-related UTI’s, in particular, if any of the mechanisms that you’ve discovered and discussed for E. coli apply to the other pathogens we see in catheter-related UTI’s. In particular, I wonder if you would speculate upon whether the catheter can serve as a surrogate pod focus or biofilm?

Stamm: That’s a good question. Certainly the formation of biofilms is important in catheter-associated infections; that’s come to be appreciated over the last few years, for not just E. coli but for other species as well. The formation of the biofilm does make infections much more difficult to treat, because antibiotics don’t work as effectively when organisms are in biofilms. Looking at the uropathogenicity of catheter-associated organisms is a little more difficult because the organisms are such a heterogeneous group. Interestingly, most of the E. coli that cause catheter-associated infections are not actually the virulent types. Thus, looking at the biofilm and how it is made is perhaps more important here than are specific sets of genes that regulate virulence, as is the case in community-acquired infections.

Tweardy: But just to extrapolate from what you have said, preventing biofilm formation by pathogenic strains, be they E. coli or others, in the catheter setting, would it seem to you to be a rational sort of approach?

Stamm: It definitely would be.