ABSTRACT
Catheter-associated urinary tract infections are biofilm-mediated infections that cause a significant economic and health burden in nosocomial environments. Using a newly developed murine model of this type of infection, we investigated the role of fimbriae in implant-associated urinary tract infections by the Gram-negative bacterium Klebsiella pneumoniae, which is a proficient biofilm former and a commonly isolated nosocomial pathogen. Studies have shown that type 1 and type 3 fimbriae are involved in attachment and biofilm formation in vitro, and these fimbrial types are suspected to be important virulence factors during infection. To test this hypothesis, the virulence of fimbrial mutants was assessed in independent challenges in which mouse bladders were inoculated with the wild type or a fimbrial mutant and in coinfection studies in which the wild type and fimbrial mutants were inoculated together to assess the results of a direct competition in the urinary tract. Using these experiments, we were able to show that both fimbrial types serve to enhance colonization and persistence. Additionally, a double mutant had an additive colonization defect under some conditions, indicating that both fimbrial types have unique roles in the attachment and persistence in the bladder and on the implant itself. All of these mutants were outcompeted by the wild type in coinfection experiments. Using these methods, we are able to show that type 1 and type 3 fimbriae are important colonization factors in the murine urinary tract when an implanted silicone tube is present.
INTRODUCTION
Catheter-associated urinary tract infections (CAUTIs) are one of the most frequent types of nosocomial infection and result in both increased patient morbidity and health care costs. The Gram-negative, opportunistic pathogen Klebsiella pneumoniae is a prominent cause of infections in individuals with indwelling urinary catheters (1, 2). The insertion of the catheter provides a conduit for bacteria to the normally sterile bladder, and it is hypothesized that the presence of an indwelling catheter provides a site for bacterial attachment and facilitates long-term colonization. Attachment to abiotic surfaces and host cell surfaces is typically mediated in Gram-negative enterobacteria by fimbrial adhesins. In K. pneumoniae, there are two well-characterized fimbrial adhesins that are often detected on the surfaces of clinical isolates, type 1 and type 3 fimbriae (3).
Type 1 fimbriae are one of the best-characterized fimbrial adhesins and are found in many bacteria in the family Enterobacteriaceae (4). These fimbriae are encoded on a gene cluster (fim) containing all the genes required for the fimbrial structure and assembly, with assembly occurring via the chaperone-usher pathway (5). The major component of the fimbrial appendage is made up of repeating FimA subunits with an adhesin molecule (FimH) at the tip of the fimbriae that confers adherence to mannose-containing glycoconjugates on host cells. Type 1 fimbriae in K. pneumoniae are regulated via phase variation in a manner similar to the regulation of type 1 fimbriae in Escherichia coli (6, 7). In uropathogenic E. coli, virulence in the urinary tract is correlated with the expression of type 1 fimbriae, which is attributed to the ability of the fimbriae to mediate adherence to epithelial cells of the urogenital tract (7, 8). Additionally, in a murine model of uncomplicated UTI lacking any indwelling urinary device, a type 1 hyperfimbriate strain forms intracellular biofilm communities within bladder umbrella cells (6).
The type 3 fimbriae are characterized by their ability to agglutinate erythrocytes treated with tannic acid in vitro, and this phenotype has been referred to as the mannose-resistant Klebsiella-like hemagglutination (MR/K) reaction (9). These fimbriae are encoded by the mrk operon and are predicted to also be assembled via the chaperone-usher pathway. While first identified and characterized in Klebsiella, type 3 fimbriae are commonly found in other Enterobacteriaceae, and the mrk gene cluster may be chromosome or plasmid borne (10–12). In vitro studies examining the role of type 3 fimbriae have shown that these fimbriae mediate attachment to endothelial and bladder epithelial cell lines and play a role in biofilm formation on abiotic surfaces, as well as surfaces coated with host-derived materials (13–18). Variants of the adhesin MrkD can bind to type IV and/or type V collagen (19, 20). The crystal structure of one of these variants, MrkD1P, has recently been determined, and the collagen binding pocket has been described (21). Thus far, in vivo models of UTI have not indicated a role for K. pneumoniae type 3 fimbriae in virulence. Given the in vitro evidence demonstrating that the type 3 fimbrial adhesin is necessary for biofilm formation on abiotic surfaces and surfaces coated with host-derived extracellular matrix proteins, it has been suggested that type 3 fimbriae are important in biofilm-mediated infections on indwelling devices, including CAUTIs (22).
It has been estimated that up to 80% of nosocomial infections are associated with an indwelling medical device, and many of these types of infections are predicted to be biofilm mediated (23, 24). The insertion of these devices provides a site for biofilm formation and blocks some of the natural host immune defenses. In addition, the mechanical insertion of these devices causes host cellular damage and exposes additional sites, such as basement membranes, for the attachment of bacteria. These niches can represent targets for type 3 fimbrial adherence (19). The biofilm-forming deficiency of type 3 fimbrial mutants in vitro and the availability of attachment sites in a catheterized host suggested that type 3 fimbriae could be important virulence factors in these types of infections.
To test this hypothesis, we used an implant-associated UTI murine model representing the effects of urinary catheterization (25). This model was originally developed to investigate Gram-positive bacterial infections (25, 26) and has recently been used to evaluate small-molecule inhibitors to prevent E. coli CAUTIs (27). By using this model of CAUTI, we were able to evaluate the contribution of both fimbrial types to the establishment and persistence of colonization. We found that mutants lacking the ability to produce type 1 or type 3 fimbriae or a combined double mutant were impaired in colonization and subsequent persistence under specific experimental conditions. Our results suggest that type 3 fimbriae, in addition to type 1 fimbriae, are indeed an important colonization factor in biofilm-mediated infections associated with CAUTIs.
MATERIALS AND METHODS
Bacterial strains and growth conditions.Table 1 lists the strains, plasmids, and primers that were used in this study. Bacteria were routinely cultured in Luria broth (10 g/liter tryptone, 5 g/liter yeast extract, 5 g/liter NaCl) with the addition of the appropriate antibiotics. The construction of K. pneumoniae TOP52 ΔT1 has previously been described (6). TOP52 ΔT3 and TOP52 ΔT1ΔT3 were constructed using the technique of allelic exchange as described in detail elsewhere by our group (28). In the procedure to construct TOP52 fim::Knr, the kanamycin resistance cassette was amplified from pKD4 using the primer pair Kn1 and Kn2 (29), and the chromosomal DNA regions flanking the fim gene cluster were amplified using the primer pair UpfimB-F (28)/CNM115 and CNM116/DwfimK-R (28). One hundred nanograms of each of the three fragments was added to a PCR mixture containing the primers UpfimB-F and DwfimK-R. The purified PCR product was electroporated into TOP52 containing the lambda Red recombinase-encoding, thermosensitive plasmid pKOBEGpra (29). The mutant was selected by growth on kanamycin at 37°C and inability to grow on apramycin. All mutants were verified using PCR analysis as described elsewhere (28) and serum agglutination as described elsewhere (30).
Strains, plasmids, and primers used in this study
Mouse strain.For this study, 6- to 7-week-old C57BL/6NCr female mice were purchased from the National Cancer Institute (NCI) and were housed in the University of Iowa Animal Facility for 7 days prior to infection. All animal procedures and protocols used in this study were approved by the Institution Animal Care and Use Committee.
Mouse model implantation.Silicone tubing was inserted into the bladders of 7- to 8-week-old C57BL/6NCr mice as previously described (25). Briefly, a 4- to 5-mm segment of SIL025 (0.025-mm outside diameter [o.d.] by 0.012-mm inside diameter [i.d.]) silicone tubing (Braintree Scientific, Inc., MA) was placed on a 30.5-gauge needle containing a 7-mm-long piece of polyethylene tubing (Becton, Dickinson, MD). Following overnight UV sterilization of the implant-needle assembly, mice were anesthetized by inhalation and the needle was inserted into the urethral opening. Using tweezers, the polyethylene tubing was advanced into the urethra and then removed, resulting in implantation of the silicone into the murine bladder. For each experimental group in these studies, a minimum of 10 (for monoinoculations) or 8 (for competitions) mice were used.
Infection and CFU determination.Bacterial cells were harvested from LB agar following overnight growth and resuspended in 1× phosphate-buffered saline (PBS) to an optical density of 0.5 at 600 nm. Mice with or without an implant were anesthetized by inhalation of isoflurane and inoculated by transurethral inoculation with 50 μl of bacterial suspensions (≈2.3 × 107 to 3.3 × 107 CFU) or PBS alone. After an allotted time, mice were sacrificed by CO2 inhalation and cervical dislocation, and the urinary bladder was excised and homogenized in 1 ml of PBS. When present, the silicone implant was recovered and placed in 0.5 ml of PBS. During recovery, the exact location of the implants varied, and in some cases, they were completely contained within the bladder, and in other animals, the implants were partially in the bladder and partially in the urethra. The implants were vortexed at maximum speed for 2 h at 4°C for removal of attached bacteria using a Tomy Microtube Mixer MT-360 (CS Bio Co., CA). PBS was also flushed through the tubing to remove adherent bacteria. To determine the number of bacteria present, samples were serially diluted and plated on LB for recovery of bacteria for the monoinfection experiments and on LB and LB supplemented with 100 μg/ml kanamycin for the coinfection experiments. CFU were quantified after incubation for 16 h at 37°C.
Determination of competitive index.For competition experiments, mice were inoculated with a 1:1 ratio of the wild type to fimbrial mutants. Mice were sacrificed 24 h postinoculation, the bladder and implant were collected separately from each animal, and dilutions were plated. After enumeration of the recovered bacteria, the competitive indices were calculated by dividing the ratio of the number of mutant bacteria to that of the wild-type bacteria in the output (bladder homogenates) by the ratio of the mutant to the wild type in the inoculum [(CFUmut/CFUwt)output/(CFUmut/CFUwt)input].
Statistical determination.GraphPad Prism (GraphPad Software, version 6.0) was used to evaluate statistical significance of groups of experiments. The Mann-Whitney U test was used for statistical analyses, and a P value of less than 0.05 was considered to be significant.
RESULTS
Murine CAUTI model.Initially, in order to determine the optimal conditions for infection using the implanted silicone tubes, a dose of approximately 2 × 107 to 3 × 107 bacteria was used to inoculate mice transurethrally immediately after insertion of silicone implants. Colonization of the bladder and implant up to 4 days postinfection was determined (Fig. 1). After 6 h, there was an initial decline in bacterial numbers followed by growth in vivo over the next 4 days. These results indicated that following implantation and inoculation, K. pneumoniae TOP52 is able to persist in the host for over 96 h. Bacteria can be recovered out to as far as 7 days postinfection when implants are present; however, the ability to recover silicone tubing after 96 h decreases. The recovery of catheters from mice is about 90% for the first 96 h; after that, the recovery rate drops to less than 50%. The implants are thought to be lost from the animal during urination, and over time the chances that the animal has voided the implant increase.
Examination of bacterial infection using K. pneumoniae TOP52 at various time points postinoculation. CFU were determined in the bladder (A) and on the recovered implant (B). The dashed line indicates the experimental limit of detection. Each symbol represents the total CFU recovered from a single animal. The solid bar indicates the median number of bacteria in each experimental group at one time point postinoculation.
The results shown in Fig. 1 indicate bacterial loads in infected animals when K. pneumoniae TOP52 is used to inoculate the mice immediately after implantation of the silicone tubing. To determine if the duration of time between implantation of the silicone tubes and bacterial inoculation played a role in the rate and degree of infection, as determined by recovered CFU, implants were also inserted 24 h prior to inoculation. These experiments demonstrated that the CFU recovered from the bladder of mice inoculated 24 h after implantation is significantly higher than the recovery from mice that were inoculated immediately following implantation (Fig. 2). As seen in Fig. 2A, the number of bacteria in the bladder of mice with no inserted tubes was essentially at the limit of detection (1.2 × 103) in comparison to mice inoculated immediately after implantation (1.9 × 105). However, if mice were infected 24 h after insertion of the silicone tube, the number of bacteria in the bladder at 48 hours postinfection was even greater (2.2 × 107). Bacterial counts on the inserted tubing itself were also higher from implants that had been in site for 24 h prior to inoculation (Fig. 2B). Additionally, mice mock infected with sterile PBS were clear of any sign of bacterial colonization (not shown). In a series of subsequent experiments, mice were inoculated both immediately after implantation and also 24 h after silicone tube implantation, and bacterial counts were performed at 6 and 48 h postinfection.
Early implantation 24 h prior to infection leads to a higher bacterial burden of K. pneumoniae as seen from the recovery of CFU from the bladder (A) and the implant (B) 48 h postinoculation. Bacterial counts were determined in mice with no silicone implants (no catheter), in animals inoculated immediately following implantation (catheter), and in animals where an implant was inserted 24 h prior to inoculation (24-h catheter). The solid bar indicates the median CFU recovered. Dashed lines indicate the limit of detection. P values were calculated using a Mann-Whitney U test (*, P < 0.05; ***, P < 0.001; ****, P < 0.0001).
Role of type 1 and type 3 fimbriae in infection.To investigate the role of K. pneumoniae type 1 and type 3 fimbriae in bladder colonization, we first examined the role of these appendages in facilitating colonization in both an uncomplicated UTI murine model in which mice received no bladder implants and then subsequently in mice with implants. These experiments show that in both the implanted and unimplanted bladder, both fimbrial types play a role in initial colonization and persistence up to 48 h postinfection. However, in the animals with no implanted tubing (Fig. 3A), the bacterial counts of fimbriate or nonfimbriate mutants were significantly lower than those for animals with an indwelling tube (Fig. 3B and C). Also, detectable numbers of K. pneumoniae TOP52 were typically present in animals without silicone tubes only at 6 h postinoculation, and the number of TOP52 bacteria recovered dropped significantly by the 48-h time point. As indicated above, no significant recovery of the parental strain in the bladder of infected mice was observed 48 h after inoculation. In addition, the fimbrial mutants did not persist in the bladders of mice with no implants. The median number of bacteria (1.9 × 104 bacteria) detected in the bladders of mice after 6 h in the animals with implanted tubes was relatively low. This number increases to approximately 3.9 × 105 at 48 h postinfection. So, it appears that the presence of implanted silicone tubes facilitates the growth of K. pneumoniae TOP52 in the murine bladders (Fig. 3B and C).
Colonization and persistence in the murine bladder. Transurethral inoculation of C57BL/6NCr mice with K. pneumoniae TOP52 (WT), the ΔT1 or ΔT3 mutant, or the ΔT1ΔT3 double mutant in uncatheterized animals at 6 and 48 h postinfection (A), animals infected immediately following catheterization (B), or animals infected 24 h after the insertion of the catheter implant (C). Each symbol represents one animal, and the number in parentheses indicates the number of animals per group. The solid bar indicates the median CFU recovered. Dashed lines indicate the limit of detection. Significant differences between groups when subjected to a Mann-Whitney U test are shown (*, P < 0.05; **, P < 0.005; ***, P < 0.001) above the data sets.
Using this implant-associated model, we were able to detect significant differences in the persistence of K. pneumoniae TOP52 compared to that of mutants lacking type 1, type 3, or both fimbrial types. In experiments 48 h after transurethral inoculation of all mice with ∼107 CFU of bacteria, mice infected with wild-type bacteria were colonized by a median count of 3.9 × 105 CFU/bladder. Mice infected with the type 3 fimbrial mutant (K. pneumoniae ΔT3) and double mutant (K. pneumoniae ΔT1ΔT3) yielded median counts of 6.0 × 103 and 1.0 × 103 bacteria, respectively. For the type 1 fimbrial mutant (K. pneumoniae TOP52 ΔT1), the number of bacteria isolated from murine bladders was below the limit of detection (1.0 × 103 CFU/bladder). In all of these experiments, inoculation was immediately following insertion of the tubes into the urinary tract (Fig. 3B).
A series of experiments was performed in which mice were inoculated 24 h after implantation of silicone tubes (Fig. 3C). In these experiments, bacterial colonization of the bladder after 6 h was relatively low for all strains, ranging from a median of 5.0 × 104 CFU/bladder with the parental strain to levels of K. pneumoniae TOP52 ΔT3 typically below the limit of detection. At 48 h postinfection, the ability of K. pneumoniae TOP52 ΔT3 and TOP52 ΔT1ΔT3 to colonize the bladder was significantly reduced compared to those of the wild-type strain and TOP52 ΔT1 (Fig. 3C).
The numbers of bacteria recovered from the implanted silicone tubing are shown in Fig. 4. Following inoculation, immediately after implantation the median number of K. pneumoniae TOP52 bacteria recovered at 6 h postinfection was approximately 8.3 × 104 CFU/implant, and this increased to approximately 8.6 × 105 at 48 h postinfection. For all fimbrial mutants, the growth of bacteria on the tubing was significantly reduced compared to that for the parental strain at 48 h postinfection (Fig. 4A).
CFU recovered from the implants at 6 and 48 h postinfection. Implants were inserted either immediately prior to inoculation (A) or 24 h in advance of inoculation (B), and then implants were collected from mice sacrificed at 6 or 48 h, at which point bacterial counts were determined. Each symbol represents one animal, and the number in parentheses indicates the number of animals per group. The dashed line represents the limit of detection. Black bars indicate the median. Significant differences as calculated using a Mann-Whitney U test are shown (*, P < 0.05; **, P < 0.005; ***, P < 0.001).
The results of bacterial counts on the implants in studies when inoculation was performed 24 h after implantation are shown in Fig. 4B. In this case, both the parental strain and K. pneumoniae ΔT1 were recovered from the tubes in approximately the same numbers regardless of the time postinoculation. However, the growth of TOP52 ΔT3 as well as TOP52 ΔT1ΔT3 was significantly reduced under these conditions. For K. pneumoniae TOP52 and ΔT1, the median count increased in number from approximately 4.7 × 104 to 6.4 × 104 at 6 h to approximately 5.5 × 105 to 1.1 × 106 after 48 h. For the remaining two strains, the highest median bacterial count observed was 1.0 × 104 for the TOP52 ΔT3 mutant at 6 h postinfection. At all other time points, the median bacterial load on the implants ranged from 8 × 103 to less than 1 × 103 (the lower limit of detection in the assay), as shown in Fig. 4B.
Coinfection studies.To further evaluate the role of K. pneumoniae fimbriae in infections of the urinary tract in vivo, mice were coinfected with the parental strain and fimbrial mutants. As described above, silicone tubes were implanted into the bladder of mice, and animals were immediately infected with equal numbers (∼3 × 107 for each strain) of bacteria. In all experiments, bacteria were harvested from bladders and implants at 24 h postinfection and the viable CFU were calculated for each strain by plating on LB and LB with kanamycin to select for the mutants.
As shown in Fig. 5A, competition between the wild-type bacterium and the type 1 fimbrial mutant resulted in an inability of the type 1 mutant to colonize the infected host; consequently, no competitive index was determined. In all but two of the samples in which animals were inoculated, no kanamycin-resistant colonies were recovered. In experiments with the wild type and the type 3 fimbrial mutant, as shown in Fig. 5B, growth of TOP52 ΔT3 could be observed. When the mutant was recovered, however, the competitive indices were 0.388 for the bladder and 0.150 for the catheter. This indicates that while the type 3 fimbrial mutant is able to persist in vivo, bacterial growth is reduced compared to that of K. pneumoniae TOP52. Colonization of the bladder and implant was reduced approximately 3-fold and 8-fold, respectively, by TOP52 ΔT3 compared to TOP52. In competition with the wild-type strain, TOP52 ΔT1ΔT3 was also significantly impaired in its ability to grow in vivo. Figure 5C indicates that in the bladder, the TOP52 ΔT1ΔT3 colonization of the bladder is outcompeted by the wild-type bacteria. In the bladder, only a single animal had any detectable colonies of TOP52 ΔT1ΔT3. In 4 out of the 8 animals colonized, TOP52 ΔT1ΔT3 was able to persist in low levels on the catheter. However, in the four animals where TOP52 ΔT1ΔT3 was recovered, the growth of TOP52 ΔT1ΔT3 was less than that of the wild-type strain, with a competitive index of <0.5 in all cases.
Relative colonization of the fimbrial mutants compared to the wild-type bacteria in the mouse model. Mouse bladders were implanted with silicone tubes and immediately inoculated with equal numbers of 107 (A to C) or 105 (D to F) wild-type and mutant bacteria. The competitions compared the colonization abilities of the wild type and the type 1 fimbrial mutant (A and D), the wild type and the type 3 fimbrial mutant (B and E), and the wild type and the fimbrial double mutant (C and F). The competitive index is the proportion of mutant to wild-type bacteria recovered from the bladder or catheter divided by the proportion of mutant to wild-type bacteria that was delivered in the inoculum. If no mutant bacteria were recovered from the bladder or implant, it is indicated on the graph. In experiments where fewer than three mice were colonized by the mutant, no competitive index was determined. Solid lines indicate the mean competitive index. Dashed lines indicate a competitive index where competition between strains is equal.
These experiments were repeated using a lower inoculum to evaluate if bacterial numbers in the initial inoculum played a role in the ability of the wild type and mutants to colonize the host and to investigate if a smaller inoculum size would affect bacterial growth of the fimbrial mutants and wild type. Experiments performed using an inoculum of 103 bacteria did not result in any colonization of any animals by any strains (n = 30), as no bacteria were recovered at 24 h postinfection (data not shown). Inoculations using 105 bacteria did result in colonization of some animals but not others. The results of these experiments are shown in Fig. 5D to F, and in all cases, at least 8 animals were inoculated. The results suggested that TOP52 ΔT1 (Fig. 5E) and TOP52 ΔT3 (Fig. 5F) were outcompeted by the wild-type bacteria, and in some animals, no fimbrial mutants were recovered. This was also observed for TOP52 ΔT1ΔT3 in the bladder and in competition experiments on the catheter where only three animals were colonized. The recovery of TOP52 ΔT1ΔT3, however, was again significantly reduced compared to that of K. pneumoniae TOP52 (Fig. 5F).
DISCUSSION
In the investigations described above, we examined the role of K. pneumoniae type 1 and type 3 fimbriae in facilitating colonization and subsequent growth in the murine bladder in the presence of silicone tubing. Silicone is commonly used in the preparation of urinary catheters for hospitalized patients, where CAUTIs are a significant problem. Infections by multiple-antibiotic-resistant bacteria, such as K. pneumoniae, are occurring with increasing frequency and present a serious problem in these environments. Previous studies have indicated that type 1 fimbrial mutants of K. pneumoniae are impaired in their infectivity in a mouse model of uncomplicated UTI when there is no indwelling device (28). Thus far, in vivo infection models have not shown a role for type 3 fimbriae in urinary tract infections, but results from studies in vitro have indicated that these fimbrial types are crucial for biofilm formation on abiotic and biotic surfaces (13–17). Additionally, characterization of clinical isolates has shown that both type 1 and type 3 fimbriae are frequently produced by these strains (3, 31, 32). To our knowledge, no studies to date have tested the importance of these fimbrial adhesins in vivo using a model of infection in which silicone implants are placed into the bladder prior to infection.
Our initial studies using this model of CAUTI demonstrated that the implantation of silicone tubing in the murine bladder leads to colonization of the bladder by higher numbers of bacteria than those in animals with no implants. Guiton et al. (25) have shown that implantation of silicone tubing results in histological and immunological changes in the murine bladder. In addition to these changes in the host, we postulate that the implant itself becomes coated with host-derived extracellular matrices and proteins. It has been described that the surface of foreign bodies inserted in a host become coated in various host proteins, including fibrin and collagen (33, 34). Damage to the host from the insertion of the catheter and the presence of host-derived material on the surface of the silicone tubes are likely to provide niches for bacterial attachment and biofilm formation that are not present in the “uncatheterized” bladder. Consistent with these observations, we detected a greater colonization of the bladder and silicone by K. pneumoniae TOP52 if the tubes were implanted 24 h prior to infection. Under these circumstances, both a greater degree of damage to the urinary tract and greater coating of the tubes would occur than in infection immediately after implantation.
In order to investigate the role of type 1 and type 3 fimbriae in colonization and persistence, we constructed defined mutations in each of these systems and assessed their ability to affect growth in vivo. By using a “catheterized” (silicone tube implant) model of infection, it was possible to detect a role for fimbria production over the course of an infection. Over a prolonged (48-h) period of infection, type 3 fimbriae appeared to play a significant role in mediating persistence in the bladders of animals with implanted tubes. Depending upon the time of infection following tube implantation, type 1 fimbriae were also implicated in the colonization process. Type 3 fimbria-producing enterobacteria are commonly associated with nosocomially acquired CAUTIs (3, 31, 32). Clearly, the site of insertion and location of urethral catheters in humans are different from those for implantation of silicone tubes into the murine bladder. However, both cases are characterized by the presence of silicone tubes in the urinary tract in vivo.
Our results are consistent with the role of type 3 fimbriae mediating the initiation of biofilm formation, an observation that has been described by numerous investigators examining this process in vitro (14–18). When testing the fimbrial mutants in animals that were implanted with silicone tubing 24 h prior to inoculation, we saw that type 1 fimbriae are not needed to facilitate colonization and persistence in the bladder and on the implant. The predicted coating of the implant with host-derived proteins as described above likely provides additional sites for type 3 fimbrial adherence that are not present when inoculation occurs immediately after implantation. Our studies also cannot rule out a function for type 1 fimbriae in this model of infection under some experimental conditions, and clearly, bacteria that are unable to produce both types are at a distinct disadvantage during infection. From these findings, we conclude that under certain circumstances in vivo both type 1 and type 3 fimbriae can play a complementary role in biofilm formation, which has been suggested by some recent investigations in vitro (35).
The coinfection studies indicate that type 1 fimbriae play a crucial role in colonization, which was suggested with the monoinoculation experiments conducted under similar conditions. Our investigations also showed that K. pneumoniae TOP52 ΔT1ΔT3 could persist on the silicone tubing in a few animals. Possibly, in the absence of both fimbrial types, an additional adhesin that allows for detectable but weaker colonization of the implant than that by the wild-type strain is present. Like those of other enterobacteria, the genome of K. pneumoniae possesses multiple fimbrial gene clusters. We have suggested that these gene clusters are controlled by integrated regulatory circuits (36, 37). The deletion of type 1 and/or type 3 fimbriae may have an effect on the expression of other ill-defined adhesins. However, taken together, our results suggest that both fimbrial types confer an advantage on the bacteria in the presence of an indwelling bladder device.
As indicated above, K. pneumoniae strains possess multiple fimbrial gene clusters predicted to encode appendages assembled by the chaperone-usher pathway (38). Also, as for other species of enteric bacteria, the function of many of these appendages is unknown. K. pneumoniae produces two well-characterized fimbrial types, type 1 and type 3 fimbriae. Type 1, or mannose-sensitive, fimbriae have been shown in many strains to play a role in the initial stages of infection and specifically in E. coli to facilitate adherence to uroepithelial cells (39). The type 1 fimbriae of both K. pneumoniae and E. coli facilitate the formation of intracellular bacterial communities (IBCs) during infection in vivo using a murine model of UTI (6). Our studies also indicate that these fimbriae can influence colonization in the urinary tract possessing an implanted tube. Type 3 fimbriae are not produced by all E. coli strains but are expressed by strains implicated in nosocomially acquired CAUTIs (11). Possibly, the presence of this fimbrial type among a variety of nosocomially acquired enterobacterial pathogens may facilitate the colonization of indwelling devices in vivo. This would be consistent with the proposed role of type 3 fimbriae in facilitating biofilm formation on solid surfaces arising from investigations using in vitro analyses (15, 17). The ability to isolate bacteria from the implanted silicone will enable us to investigate and characterize gene products that facilitate K. pneumoniae growth in vivo. For example, future studies will involve K. pneumoniae fimbrial gene expression using the model described here in order to determine whether an array of fimbriae are involved in this process, since many gene clusters may be expressed only in vivo.
In summary, this report indicates that K. pneumoniae type 1 and type 3 fimbriae are important colonization factors in UTIs. Under all conditions, we determined that one or both fimbrial types are required for colonization and persistence. Additionally, this work has shown that both type 1 and type 3 fimbriae contribute to bacterial adherence in a murine model of CAUTI, supporting the hypothesis that fimbrial adhesins are an important attachment factor in CAUTIs caused by K. pneumoniae. The development of this model for K. pneumoniae may be utilized in the future for further studies of virulence factors or as a suitable system in testing the efficacy of vaccine candidates or antimicrobials. Given the high incidence of CAUTIs caused by K. pneumoniae compared to the incidence of uncomplicated UTIs caused by K. pneumoniae, this model may serve to better test prevention strategies than the previously used uncomplicated UTI model.
ACKNOWLEDGMENTS
This work was supported by a grant from the NIH to S.C. (RO1AI 050011). M.S.M. was supported in part by grants from the following foundations: the Jan og Erna Loegstrups Fund, Ingenioerforeningen Danmarks Laane- og Hjaelpefond, and the Reinholdt W. Jorck og Hustrus fund.
We thank Scott Hultgren and Pascale S. Guiton, Washington University in St. Louis, for allowing one of us (Caitlin Murphy) to visit the Hultgren laboratory and the opportunity to learn how to use the implantation model.
FOOTNOTES
- Received 18 March 2013.
- Returned for modification 23 April 2013.
- Accepted 1 June 2013.
- Accepted manuscript posted online 10 June 2013.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
