|Home | About | Journals | Submit | Contact Us | Français|
Type 3 fimbriae are expressed by most clinical Klebsiella pneumoniae isolates and mediate adhesion to host structures in vitro. However, the role of type 3 fimbriae in K. pneumoniae virulence has not been evaluated by use of in vivo infection models. In this study, the type 3 fimbrial gene cluster (mrk) of the clinical isolate C3091 is described in detail. The mrk gene cluster was revealed to be localized in close proximity to the type 1 fimbrial gene cluster. Thus, a 20.4-kb fimbria-encoding region was identified and found to be highly conserved among different K. pneumoniae isolates. Interestingly, a homologue to PecS, known as a global regulator of virulence in Erwinia chrysanthemi, was identified in the fimbria-encoding region. Comparison to the previously characterized plasmid encoded mrk gene cluster revealed significant differences, and it is established here that the putative regulatory gene mrkE is not a part of the chromosomally encoded type 3 fimbrial gene cluster. To evaluate the role of type 3 fimbriae in virulence, a type 3 fimbria mutant and a type 1 and type 3 fimbria double mutant was constructed. Type 3 fimbria expression was found to strongly promote biofilm formation. However, the fimbria mutants were as effective at colonizing the intestine as the wild type, and their virulence was not attenuated in a lung infection model. Also, in a urinary tract infection model, type 3 fimbriae did not influence the virulence, whereas type 1 fimbriae were verified as an essential virulence factor. Thus, type 3 fimbriae were established not to be a virulence factor in uncomplicated K. pneumoniae infections. However, since type 3 fimbriae promote biofilm formation, their role in development of infections in catheterized patients needs to be elucidated.
Klebsiella pneumoniae is recognized as an important opportunistic pathogen and is a common cause of urinary tract infections, respiratory tract infections, and septicemia, especially in immunocompromised individuals (37). K. pneumoniae is ubiquitous in the environment, and epidemiological studies have revealed that infections are frequently preceded by colonization of the patient's gastrointestinal tract (33). In recent years an emerging syndrome of community-acquired pyogenic liver abscess caused by highly virulent K. pneumoniae strains has occurred (22, 24, 29, 54). These infections often occur in otherwise healthy individuals and are frequently complicated by devastating dissemination of the infection to the eyes and central nervous system. The rising incidence of serious K. pneumoniae infections stresses the need to elucidate the pathogenic mechanism of this important pathogen.
The ability of bacteria to adhere to tissue surfaces in the host is an important step in the development of infection. Most clinical K. pneumoniae isolates express two types of fimbrial adhesins, type 1 and type 3 fimbriae (37). Type 1 fimbriae are well characterized and found in the majority of enterobacterial species (23). They mediate adhesion to mannose-containing structures present on host cells or in the extracellular matrix. A number of studies have shown that type 1 fimbriae play a significant role in the ability of Escherichia coli to infect the urinary tract (5, 26, 34, 47). In agreement with this, we have recently established that type 1 fimbriae are an important virulence factor in K. pneumoniae urinary tract infection (50).
Compared to type 1 fimbriae, type 3 fimbriae are thin (4 to 5 nm in diameter) nonchanneled fimbriae (11, 25). They are characterized by their ability to agglutinate tannic acid treated, but not native, erythrocytes in a mannose-resistant manner; also referred to as mannose-resistant Klebsiella-like hemagglutination. In addition to Klebsiella species, type 3 fimbriae are common in Enterobacter, Serratia, Proteus, and Providencia isolates (4). Type 3 fimbriae belong to the chaperone-usher class fimbriae and are encoded by the mrk gene cluster, which includes the mrkA gene encoding the major fimbrial subunit and mrkD encoding the fimbrial adhesin responsible for mannose-resistant Klebsiella-like hemagglutination (1). The MrkD adhesin has been shown to mediate adhesion to collagen structures (51); however, the exact identity of the MrkD receptor remains elusive. In vitro studies have revealed that type 3 fimbriae mediate adhesion to different structures in human kidney and lung tissue and epithelial cells from human urine sediments, as well as in endothelial and bladder epithelial cell lines (16, 51, 52, 53). Furthermore, type 3 fimbriae have been established to play a significant role in K. pneumoniae biofilm formation (9, 20, 27). Historically, type 3 fimbriae have not been associated with E. coli; however, most recently two independent studies have reported type 3 fimbria expression in E. coli strains (3, 36). Intriguingly, in both studies type 3 fimbriae were encoded by conjugative plasmids and found to profoundly enhance the ability of E. coli to form biofilm. Type 3 fimbria-mediated biofilm formation and in vitro binding to different host structures indicate that type 3 fimbriae may play a significant role in virulence. However, thus far the influence of type 3 fimbriae on the pathogenicity of K. pneumoniae has not been investigated by use of in vivo models. In the present study, we constructed isogenic fimbria mutants to evaluate the influence of fimbriae in virulence using relevant in vivo infection models. Furthermore, a fimbria-encoding chromosomal region in K. pneumoniae including both the type 1 and the type 3 fimbrial gene clusters was identified.
K. pneumoniae C3091 is a streptomycin-resistant human urinary tract infection isolate (35, 48). The C3091 type 1 fimbria mutant was previously described (50). Construction of a C3091 isogenic type 3 fimbria mutant and a type 1 and type 3 fimbria double mutant is described below. The K. pneumoniae bacteremia isolates C132-98 and C4712, and the urinary tract isolate C747 have been previously described (49). The K. pneumoniae strains 3858 and 3928 were isolated from patients with liver abscess, whereas strains H79192 and W215 are human fecal isolates from healthy individuals.
For cloning purposes, the nonfimbriated E. coli strain HB101 was used (38). Bacteria were cultured on solid or in liquid Luria-Bertani (LB) medium. For infection studies, the wild-type and mutant strains were grown overnight at 37°C in LB medium with shaking (200 rpm). When appropriate, media were supplemented with the following concentrations of antibiotics: apramycin, 30 μg/ml; tetracycline, 8 μg/ml; kanamycin, 25 μg/ml; and streptomycin, 100 μg/ml.
Expression of type 1 fimbriae was detected by the ability of bacterial cells to agglutinate a freshly prepared solution of 5% washed guinea pig erythrocytes in a mannose-sensitive manner. Expression of type 3 fimbriae was detected as mannose-resistant agglutination of tannic acid-treated ox erythrocytes prepared by treating ox erythrocytes with 0.03% tannic acid for 10 min at 37°C. Subsequently, the tannic acid-treated blood was washed twice with 0.9% saline and resuspended to a concentration of 5%. The agglutination assays were performed on glass slides with or without 5% mannose.
For assessment of hemagglutination titers, bacterial suspensions containing 1010 CFU/ml were serially diluted in microtiter plates, mixed with tannic acid-treated ox erythrocytes, and incubated at 4°C overnight. The hemagglutination titer was read as the last dilution where visible agglutination had occurred.
Plasmid DNA isolation was carried out by use of a Qiaprep spin miniprep kit (Qiagen) according to the manufacturer's instructions. Sequencing was conducted commercially at MWG-Biotech AG (Germany). Restriction endonucleases were used as recommended by the manufacturer (New England Biolabs).
Negative staining was carried out as follows. A Formvar-coated carbon-reinforced copper grid was applied, film side down, on a droplet of a culture suspension for 2 min. The excess liquid was then removed with a piece of filter paper, and the grid was then stained for 30 s on droplets of 2% ammonium molybdate (pH 7.4) or 15 s on droplets of 1.25% phosphotungstic acid (pH 6.5). Electron microscopy was carried out on a Philips 201C electron microscope at 60 kV.
The type 3 fimbria gene cluster in C3091 was deleted by allelic exchange with a kanamycin resistance-encoding cassette flanked by regions homologous to the regions up- and downstream the mrk gene cluster (Fig. (Fig.1).1). All primers used are listed in Table Table1.1. The cassette was generated by a modification of a three-step PCR procedure as previously described (8, 50). As the first step, the kanamycin resistance-encoding gene (kan) was amplified from pKD4 by using the primer pair Kn1 and Kn2 (7). Second, from C3091 chromosomal DNA 309-bp and 239-bp regions flanking the mrk gene cluster were amplified by PCR using the primer pairs UpmrkA-F/UpmrkA-R and DwmrkF-F/DwmrkF-R, respectively. At their 5′ ends, primers UpmrkA-R and DmrkF-F contained 20-bp regions homologous to the extremities of the kan gene. In the third step, the flanking regions were added on each side of the kan gene by mixing 100 ng of each fragment, followed by PCR amplification using the primer pair UpmrkA-F and DwmrkF-R. The PCR product was purified and electroporated into C3091 harboring the thermosensitive plasmid pKOBEGApra encoding the lambda Red recombinase. The C3091Δmrk mutant was selected by growth on LB plates containing kanamycin at 37°C. Loss of the pKOBEGApra plasmid was verified by the inability of the mutant to grow on LB agar plates containing apramycin. Correct allelic exchange was verified by PCR analysis using combinations of primers inside the kan gene, K1 and K2 (7), and primers flanking the deleted mrk gene cluster, i.e., Upmrk and Dwmrk (Fig. (Fig.11).
To construct the type 1 and type 3 fimbria double mutant, the type 1 fimbrial gene cluster in the Δmrk mutant was deleted by allelic exchange with a tetracycline resistance-encoding cassette. The procedure was essentially as described above and in Fig. Fig.1.1. The tetracycline resistance-encoding cassette was amplified from pAR82 using the primer pair Ucas and Dcas (39). All primers used to construct the double mutant are listed in Table Table11.
Biofilm formation was assayed by adding overnight cultures to 24-well multidish polystyrene plates (Nunc) containing 1 ml of M9 minimal medium (with 0.02 M glucose added) per well, followed by incubation for 48 h at 37°C with agitation (100 rpm). Biofilm formation was quantified by crystal violet staining. After removal of medium and two washes with phosphate-buffered saline, surface-attached cells were covered with 0.1% crystal violet (Sigma-Aldrich) for 15 min. After two washes with phosphate-buffered saline, crystal violet was dissolved by the addition of ethanol, and the absorbance measured at 590 nm (Ultrospec 1100 Pro; Amersham Biosciences). Each strain was tested in quadruplicate.
Six- to eight-week-old female CFW1 mice (Harlan) were used for intestinal colonization experiments as described previously (30). Briefly, mice were individually caged and provided with drinking water containing 5 g of streptomycin sulfate per liter for 24 h. Prior to inoculation, fecal samples were collected for testing for growth of streptomycin-resistant bacteria on selective media. All experiments were performed as coinfections by feeding 100-μl bacterial suspensions, containing ~105 CFU of the wild-type and mutant strains mixed 1:1, to the mice by pipette. During the colonization experiment, cages were changed daily, and the mice continuously received water containing streptomycin. At the indicated times, 0.5 g of feces were collected and homogenized in 5 ml of 0.9% NaCl, and dilutions were plated on selective media. Randomly picked colonies were verified to be K. pneumoniae by biotyping using an Minibact E kit (Statens Serum Insitut).
An intranasal infection model previously described was applied for coinfection studies (13, 41). Six- to eight-week-old female NMRi mice (Harlan) were anesthetized. The mice were hooked on a string by the front teeth, and 50 μl of bacterial suspension, containing approximately 5 × 107 CFU of the wild-type and mutant strains mixed 1:1, was dripped onto the nares. The mice readily aspirated the solution and were left hooked on the string for 10 min before being returned to their cages. The mice were sacrificed 2 days after inoculation. For recovery of bacteria, lungs were collected in 1 ml of 0.9% NaCl and homogenized, and serial dilutions were plated on selective media.
Six- to eight-week-old female CFW1 mice (Harlan) were used for coinfection studies. The model has been described in detail previously (19). Three days before inoculation and throughout the experiment, 5% glucose was added to the drinking water since this has been shown to promote urinary tract infection in mice (21). The mice were anesthetized and inoculated transurethrally with 50 μl of bacterial suspension containing approximately 5 × 108 CFU of the wild-type and mutant strains mixed 1:1 by using plastic catheters. The catheter was carefully pushed horizontally through the urethral orifice until it reached the top of the bladder, and the bacterial suspension was slowly injected into the bladder. The catheter was immediately removed, and the mice were subjected to no further manipulations until sacrifice. The mice were sacrificed 3 days after inoculation. For the recovery of bacteria, bladders and kidneys were collected in 1 ml of 0.9% NaCl and homogenized, and serial dilutions were plated on selective media.
Nucleotide sequences were analyzed by use of the Lasergene Software v.5.03 (DNAStar, Inc.). The competitive index (CI) was calculated as the proportion of wild-type to mutant bacteria recovered from infected organs divided by the proportion of wild-type to mutant bacteria in the inoculum. The Wilcoxon rank sum test was used to analyze the data of the infection studies, with zero as the comparator for the log10 of the CI values. P values below 0.05 were considered to be statistically significant.
The nucleotide sequence of the fimbria-encoding region in K. pneumoniae strain C3091 reported in the present study have been assigned the GenBank accession number EU682505. The nucleotide sequences of the region interspacing the type 1 and type 3 fimbrial gene clusters in strains C132-98, C747, and C4712 have been assigned GenBank accession numbers FJ514821, FJ514822, and FJ514823, respectively.
We recently reported the isolation and characterization of the K. pneumoniae type 1 fimbrial gene cluster (fim) by screening a clone library of chromosomal DNA from the clinical isolate C3091 (50). Further analysis of a fim containing clone by sequencing revealed that the type 3 fimbria-encoding gene cluster (mrk) was located in close proximity to the fim gene cluster. Thus, a 20.4-kb region encoding both types of fimbriae was identified (Fig. (Fig.22).
The two fimbrial gene clusters were found to be interspaced by a 4.6-kb region. This region comprised five open reading frames (ORFs), including homologues of the pecM and pecS genes in Pectobacterium atrosepticum (previously Erwinia carotovora subsp. atroseptica), with 63 and 65% amino acid identity, respectively. Furthermore, an ORF similar to ORF KPN_0384 in the genome of the sequenced K. pneumoniae strain MGH78578, encoding a putative high-affinity nickel transporter of the NicO superfamily, and two additional ORFs encoding hypothetical proteins were identified.
Sequence comparison of the clinical isolate C3091 with the MGH78578 genome revealed that a similar 20.4-kb fimbria-encoding region with 99.5% identity at the nucleotide level is present in the sequenced reference strain MGH78578. To investigate whether the fimbria-encoding region is commonly present in different K. pneumoniae strains, PCR was performed with the primers mrkF and fimR, located in the mrk and fim gene clusters, respectively, on chromosomal DNA obtained from seven isolates with different origins. From each isolate PCR products of identical size were obtained. Thus, the 4.6-kb region between the two fimbrial gene clusters was found to be highly conserved in different K. pneumoniae isolates. Furthermore, sequence analysis of the region from strains C132-98, C747, and C4712 revealed greater than 99% nucleotide identity among the different isolates. Thus, all three sequenced isolates had homologues to the five ORFs described in C3091.
A type 3 fimbrial gene cluster isolated from the K. pneumoniae strain IA565 has previously been characterized (1). However, subsequent studies have revealed that this gene cluster is plasmid encoded and only present in a small subset of K. pneumoniae strains (17, 42, 43). Indeed, PCR analysis using the primer pair CAS210 and CAS211 specific for the IA565 variant revealed that none of the eight K. pneumoniae strains investigated in the present study possess this type 3 fimbria variant. Most recent studies have also described two plasmid encoded mrk gene clusters isolated from E. coli (3, 36). A comparison of the different mrk gene clusters is shown in Table Table22.
The mrk gene cluster includes genes encoding the major fimbrial subunit MrkA, a chaperone (MrkB), an usher (MrkC), the adhesin (MrkD), and a minor fimbrial subunit (MrkF). However, only the plasmid-encoded gene cluster from K. pneumoniae strain IA565 contains the putative regulatory gene mrkE. In strain IA565, mrkE is located directly upstream of mrkA (1); however, in neither of the isolates investigated here or MGH78578, was a mrkE homologue located in the fimbria-encoding region. Thus, mrkE is generally not a part of the chromosomal type 3 fimbrial gene cluster.
The gene products of the C3091 and the MGH78578 mrk gene cluster all exhibited >99% identity (Table (Table2).2). In contrast, the IA565 gene cluster varied in similarity from 96% for the major fimbrial subunit MrkA to only 60% for the MrkD adhesin, confirming that the plasmid encoded mrk cluster of strain IA565 is markedly different from the chromosomally encoded variant. Interestingly, both plasmid-encoded mrk gene clusters isolated from E. coli exhibited a much higher similarity to the chromosomally than the plasmid-encoded K. pneumoniae mrk gene cluster.
A 5.6-kb region containing the C3091 mrk gene cluster was subcloned into the cloning vector pUC18 to yield pCAS625. To verify that the cloned mrk gene cluster conferred phenotypic expression of functional type 3 fimbriae, pCAS625 was transformed into the nonfimbriated E. coli strain HB101. Expression of functional type 3 fimbriae was verified by the ability of HB101(pCAS625) to agglutinate tannic acid treated ox erythrocytes in a mannose-resistant manner. As expected, the HB101 parent strain was unable to agglutinate tannic acid treated ox erythrocytes, whereas HB101(pCAS625) exhibited pronounced hemagglutination with a hemagglutination titer of 1:128. Furthermore, transmission electron microscopy confirmed that HB101(pCAS625) expressed fimbriae on the cell surface, whereas the HB101 parent strain as expected was nonfimbriated (Fig. (Fig.33).
To investigate the role of type 3 fimbriae in K. pneumoniae virulence, an isogenic type 1 fimbria mutant and a type 1 and type 3 fimbria double mutant were constructed by allelic exchange as described in Materials and Methods and Fig. Fig.1.1. To verify that the type 3 fimbria mutant did not express type 3 fimbriae, the Δmrk mutant and the wild-type strain were analyzed for the ability to agglutinate tannic acid-treated ox erythrocytes. The wild-type strain exhibited mannose-resistant hemagglutination with a hemagglutination titer of 1:32, whereas the Δmrk mutant was unable to agglutinate tannic acid-treated ox erythrocytes. As expected, the type 1 and type 3 fimbria double mutant was also unable to agglutinate guinea pig erythrocytes, verifying abolishment of type 1 fimbria expression.
Comparison of biofilm formation by C3091 and its isogenic type 3 fimbria mutant revealed that type 3 fimbriae play a pronounced role in K. pneumoniae biofilm formation (Fig. (Fig.4A).4A). Thus, the type 3 fimbria mutant exhibited a striking reduction in biofilm formation compared to the wild-type strain (P = 0.0002). Furthermore, introduction of pCAS625 encoding the type 3 fimbrial gene cluster cloned from C3091 into E. coli HB101 resulted in remarkably enhanced biofilm formation compared to HB101 carrying the empty pUC18 cloning vector (P = 0.0001), confirming that expression of type 3 fimbriae strongly promotes biofilm formation (Fig. (Fig.4B4B).
To assess the influence of type 3 fimbriae on the ability of K. pneumoniae to colonize the intestine, a group of three mice was inoculated with equal numbers of the wild type and the type 3 fimbria mutant, and the bacterial counts in feces were monitored for 13 days (Fig. (Fig.5A).5A). The type 3 fimbria mutant was found to be as effective at colonizing the intestine as the wild-type strain. In all mice, each inoculated with ~105 bacteria, the bacterial counts in feces day 1 after inoculation were ~108 to 109 bacteria per g of feces, and the counts of the Δmrk mutant were found to be similar to those for the wild-type strain. Thus, the lack of type 3 fimbriae did not influence the ability of the mutant to become established in the intestinal tract. Furthermore, the wild-type strain and Δmrk mutant were found to cocolonize the intestine since similar numbers of the wild type and mutant were detected in fecal samples throughout the colonization period.
To further characterize the role of fimbriae on K. pneumoniae intestinal colonization, a similar experiment was performed with the type 1 and type 3 fimbria double mutant. This revealed that neither type 1 nor type 3 fimbriae affect the ability of K. pneumoniae to colonize the intestine since similar numbers of the wild type and the Δmrk Δfim mutant were detected in fecal samples throughout the experiment (Fig. (Fig.5B5B).
To evaluate the role of type 3 fimbriae in K. pneumoniae lung infection, two groups of four mice were intranasally inoculated with an equal number of the Δmrk mutant and the wild-type strain. The two groups of mice were sacrificed one and 2 days after inoculation, respectively. The lungs of all mice were found to be infected with high numbers of K. pneumoniae (Fig. (Fig.6A).6A). Type 3 fimbriae did not influence the virulence since the bacterial counts of the mrk mutant were similar to those of the wild-type strain.
Also, the type 1 and type 3 fimbria double mutant was found to be as virulent as the wild-type strain (Fig. (Fig.6B).6B). Thus, neither type 1 nor type 3 fimbriae play a significant role in the ability of K. pneumoniae to cause lung infection.
The influence of type 3 fimbria expression on K. pneumoniae urovirulence was evaluated by inoculating a group of eight mice with equal numbers of the Δmrk mutant and the wild-type strain. The mice were sacrificed 3 days after inoculation. All bladder samples were found to be infected with a median bacterial count of 2.2 × 106 CFU per bladder, and in all but one mouse the infection had spread to the kidneys. No influence of type 3 fimbriae on the urovirulence was observed since similar numbers of the type 3 fimbria mutant and wild-type strain were detected in infected bladders and kidneys (Fig. (Fig.7A7A).
We recently established that type 1 fimbriae are a significant virulence factor in K. pneumoniae urinary tract infection (50). In agreement with this, the type 1 and type 3 fimbria double mutant was found to be markedly attenuated in the urinary tract infection model (Fig. (Fig.7B).7B). The wild-type strain had significantly outcompeted the Δmrk Δfim mutant in the bladders by a median of 5.0 log10 (P = 0.008), and in seven of eight infected kidneys only the wild-type strain was detected.
It could be speculated that the profound influence of type 1 fimbriae on the urovirulence may mask a putative influence of type 3 fimbriae. To investigate this hypothesis, the urovirulence of a previously constructed type 1 fimbria mutant (50) was compared to the Δmrk Δfim double mutant. The pronounced importance of type 1 fimbriae was again confirmed since the bacterial counts in infected bladders were very low (median bacterial count = 1.2 × 102 CFU), and in none of the mice had the infection spread to the kidneys. The ability of the type 1 fimbria mutant to express type 3 fimbriae was found not to influence the virulence since it was detected in similar numbers as the Δmrk Δfim double mutant (Fig. (Fig.7C7C).
Even though adhesion is generally considered to play a significant role in bacterial pathogenicity, few studies have investigated the influence of different adhesion factors on K. pneumoniae virulence in vivo. We recently demonstrated that type 1 fimbriae are an important virulence factor in K. pneumoniae urinary tract infection but does not influence intestinal colonization ability or virulence in a lung infection model (50). In addition to type 1 fimbriae most clinical K. pneumoniae isolates also express type 3 fimbriae (37). However, although type 3 fimbriae are generally considered as a virulence factor, to our knowledge the influence of type 3 fimbriae in K. pneumoniae virulence has not been evaluated by use of in vivo infection models.
Type 3 fimbriae are encoded by the mrk gene cluster which has previously been cloned from the K. pneumoniae strain IA565 (1). The mrk gene cluster from strain IA565 was reported to contain six genes, mrkABCDFE, encoding a major fimbrial subunit MrkA, the chaperone MrkB, the usher MrkC, the MrkD adhesin, a suggested anchorage protein MrkF (most recently identified as a minor fimbrial subunit ) and a putative regulatory protein MrkE. However, studies using an IA565-specific mrkD gene probe revealed that only a few K. pneumoniae isolates possess the IA565 mrkD variant which, in contrast, appeared to be closely related to the mrkD gene of K. oxytoca (17, 42, 43). In agreement, none of the K. pneumoniae strains included in the present study were found to possess the IA565 variant. It was subsequently shown that strain IA565 possesses two different copies of the mrk gene cluster, a chromosomal and a plasmid-encoded variant, and that the sequenced gene cluster was the plasmid-encoded variant (17, 43). We report here the sequence of the chromosomally encoded mrk gene cluster from the clinical K. pneumoniae strain C3091. As expected, the chromosomally encoded MrkD adhesin had a low degree of identity (60%) to the plasmid-encoded MrkD adhesin of strain IA565. Furthermore, the other gene products of the C3091 mrk gene cluster were found to differ significantly from IA565, although the major fimbrial subunit (MrkA) was found to be 96% identical (Table (Table2).2). In contrast, sequence analysis revealed >99% similarity of all mrk gene products, including MrkD, when C3091 was compared to the genome-sequenced strain MGH78578, indicating a high degree of conservation of chromosomally encoded K. pneumoniae mrk gene clusters. Moreover, the recently identified plasmid-borne type 3 fimbrial gene clusters isolated from E. coli were found to be highly similar to the C3091 mrk gene cluster. This indicates that the E. coli plasmid-encoded mrk gene clusters likely have originated from the K. pneumoniae chromosome rather than from plasmid-encoded K. pneumoniae mrk variants.
In the IA565 mrk gene cluster the putative regulatory gene mrkE is located directly upstream of mrkA (1). However, we found that the mrkE gene is absent from the chromosomally encoded mrk gene cluster. Likewise, mrkE is absent from the plasmid-encoded mrk gene clusters isolated from E. coli (3, 36). Nonetheless, the E. coli and the mrk gene cluster cloned from C3091 express functional type 3 fimbriae. In IA565, mrkE has been suggested to encode a regulator of type 3 fimbrial expression (1). However, no evident change in the level of fimbrial expression was detected between an E. coli transformant expressing the cloned IA565 mrk gene cluster and an E. coli transformant expressing a derivate of the IA565 mrk gene cluster lacking the mrkE gene (1). Thus, the putative role of mrkE in regulation of type 3 fimbria expression is yet to be determined.
The mrk gene cluster was found to be localized in close proximity to the type 1 fimbrial gene cluster on the K. pneumoniae chromosome. Thus, a 20.4-kb fimbria-encoding region encoding both fimbrial types was identified. The chromosomal region interspacing the two fimbrial gene clusters was highly conserved in all K. pneumoniae isolates investigated and contains five putative ORFs. In addition to two ORFs encoding hypothetical proteins and an ORF encoding a putative high-affinity nickel transport protein, homologues to the pecM and pecS genes, were identified. The pecS gene product belongs to a subset of the Mar/SlyA family of transcriptional regulators which are involved in regulation of virulence factors (12). The pecS locus, consisting of the two divergently transcribed genes pecM and pecS, was first identified as a regulator of virulence genes in the phytopathogenic enterobacterium Erwinia chrysanthemi (40). In E. chrysanthemi, the pecS gene was shown to regulate the expression of several different virulence factors, including plant cell wall-degrading enzymes, flagella, type II and type III secretion systems, toxinlike proteins, a high-affinity iron uptake system, and factors involved in resistance to oxidative stress (40, 15). Thus, PecS was most recently proposed to be a global regulator of the symptomatic phase in E. chrysanthemi infections (15). It is obvious to speculate that the K. pneumoniae PecS homologue may have a similar role as a regulator of virulence gene expression in K. pneumoniae. Thus, the location of the pecS locus in the fimbria-encoding region may indicate that PecS could be involved in regulation of fimbrial expression. However, future studies are required to establish whether PecS has a role in K. pneumoniae virulence and the regulation of fimbrial expression.
To investigate the role of type 3 fimbriae in K. pneumoniae virulence, a well-defined type 3 fimbria mutant and a type 1 and type 3 fimbria double mutant was constructed and evaluated in three relevant mouse models of K. pneumoniae infection. The majority of K. pneumoniae infections are preceded by colonization of the patient's intestinal tract (33); thus, the role of type 3 fimbriae in intestinal colonization was investigated. Strain C3091 was found to be an excellent colonizer of the mouse intestine; however, type 3 fimbriae did not influence the ability of C3091 to either establish itself or long-term colonize the intestine since similar numbers of the wild type and type 3 fimbria mutant were detected in fecal samples throughout the colonization period. Also, the Δmrk Δfim double mutant was as effective at colonizing the intestine as the wild-type strain. Thus, neither type 1 or type 3 fimbriae enhance K. pneumoniae colonization ability, an observation in agreement with the recent finding that K. pneumoniae actually downregulates type 1 fimbria expression in the large intestine (50). Since adhesion is generally believed to be a key factor in the ability of bacteria to reside in the intestinal tract, it is possible that K. pneumoniae express unknown adhesins that mediate adhesion to mucosal surfaces in the gastrointestinal tract. Indeed, the sequenced K. pneumoniae MGH78578 genome (accession number CP000647) includes putative fimbrial operons that may encode as-yet-uncharacterized fimbriae.
In the lung infection model, type 3 fimbriae were found not to influence the virulence, and the Δmrk Δfim double mutant was established to be as virulent as the wild-type strain. Fimbriae are known to be recognized by phagocytic cells (2, 45); therefore, it may be that fimbrial expression is a disadvantage in phagocyte-rich infection sites such as the lungs, whereas factors protecting the bacterium against the immune system, such as capsule and lipopolysaccharide, have been shown to be essential virulence factors in this host environment (6, 28, 44).
The type 3 fimbria mutant was found to be as virulent in the urinary tract model as the wild-type strain. In contrast, the virulence of the Δmrk Δfim double mutant was significantly attenuated, a finding consistent with the fact that type 1 fimbriae are an important virulence factor in the urinary tract (50). It could be speculated that the inability of the mrk mutant to express type 3 fimbriae may be compensated for by enhanced expression of type 1 fimbriae. Indeed, regulatory cross talk and coordinated expression of different fimbrial gene clusters have been described in E. coli (14, 31, 46, 55). Alternatively, a minor role of type 3 fimbriae in urovirulence may be masked by the profound influence of type 1 fimbriae. However, when the virulence of the Δmrk Δfim double mutant was directly compared to a type 1 fimbria mutant, no influence of the ability of the type 1 fimbria mutant to express type 3 fimbriae was detected.
In conclusion, type 3 fimbriae were shown not to influence the ability of K. pneumoniae to colonize the mouse intestine or infect the lung and urinary tract, although in all experiments the wild-type strain expressed type 3 fimbriae in the bacterial suspension used to inoculate the mice as revealed by hemagglutination assays. Previous studies have revealed that type 3 fimbriae are important in K. pneumoniae biofilm formation (9, 20, 27). This was verified in the present study since the type 3 fimbria mutant was found to be severely attenuated in biofilm formation compared to the wild-type strain. Also, in E. coli plasmid-encoded type 3 fimbriae have been linked to significantly enhanced biofilm formation on abiotic surfaces, including catheters (3, 36). Indeed, we found that E. coli HB101 transformed with the cloned type 3 fimbrial gene cluster from C3091 exhibited a remarkable increase in biofilm formation. K. pneumoniae is one of the most common causes of nosocomial infections; thus, K. pneumoniae infections frequently occur in patients with indwelling devices, e.g., catheters or endotracheal tubes (37). Bacterial biofilm formation on indwelling devices is a significant medical problem and often results in infections difficult to eradicate due to enhanced resistance to antibiotic treatment and the host immune system (10). Furthermore, bacteria can disseminate from biofilms, causing systemic disease. Since type 3 fimbriae mediate biofilm formation, it is altogether plausible that type 3 fimbriae play a significant role in biofilm-associated infections. Indeed, expression of type 3 fimbriae has previously been shown to correlate with long-term persistence of Providencia stuartii in patients with catheter-associated bacteriuria (32). Therefore, although the present study indicates that type 3 fimbriae are not a virulence factor in uncomplicated K. pneumoniae infections, the expression of type 3 fimbriae may play an important role in development of infections in catheterized patients.
We thank Christina Kofoed Johnsen, The Electron Microscopy Unit, Statens Serum Institut, Denmark, for performing the electron microscopy experiments.
C.S. was partially financed by Danish Research Agency grant 2052-03-0013. M.B. was financed by Danish Research Agency grant 22-04-0725.
Editor: V. J. DiRita
Published ahead of print on 24 August 2009.