1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Klebsiella pneumoniae is a Gram-negative enterobacterium that has historically been, and currently remains, a significant cause of human disease. It is a frequent cause of urinary tract infections and pneumonia, and subsequent systemic infections can have mortality rates as high as 60%. Despite its clinical significance, few virulence factors of K. pneumoniae have been identified or characterized. In this study we present a mouse model of acute K. pneumoniae respiratory infection using an intranasal inoculation method, and examine the progression of both pulmonary and systemic disease. Wild-type infection recapitulates many aspects of clinical disease, including significant bacterial growth in both the trachea and lungs, an inflammatory immune response characterized by dramatic neutrophil influx, and a steady progression to systemic disease with ensuing mortality. These observations are contrasted with an infection by an isogenic capsule-deficient strain that shows an inability to cause disease in either pulmonary or systemic tissues. The consistency and clinical accuracy of the intranasal mouse model proved to be a useful tool as we conducted a genetic screen to identify novel virulence factors of K. pneumoniae. A total of 4800 independent insertional mutants were evaluated using a signature-tagged mutagenesis protocol. A total of 106 independent mutants failed to be recovered from either the lungs or spleens of infected mice. Small scale independent infections proved to be helpful as a secondary screening method, as opposed to the more traditional competitive index assay. Those mutants showing verified attenuation contained insertions in loci with a variety of putative functions, including a large number of hypothetical open reading frames. Subsequent experiments support the premise that the central mechanism of K. pneumoniae pathogenesis is the production of a polysaccharide-rich cell surface that provides protection from the inflammatory response.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Klebsiella pneumoniae is a non-motile Gram-negative bacterium that is ubiquitous in the environment and is often found as a commensal resident of the human gastrointestinal tract (Seaton, 2000). When inoculated parenterally, it has a remarkable ability to cause a wide range of human diseases, from urinary tract infections to pneumonia (Sahly and Podschun, 1997). If a localized K. pneumoniae infection is allowed to disseminate, the subsequent systemic infection is often rapid and overwhelming. Among immunocompromised, hospitalized or elderly patients K. pneumoniae infections are particularly devastating, with resulting mortality rates between 25% and 60% (Ellis, 1998). The increasing prevalence of antibiotic-resistant strains only serves to compound this species’ clinical importance (Timko, 2004).

Since the first identification of Klebsiella as a cause of pneumonia by pathologist Karl Friedlä nder in 1882, capsular polysaccharide has been established as the species’ most distinguishing characteristic and most studied virulence factor (Eisenstadt and Crane, 1994). Previous studies have led to a greater understanding of the potential functions capsule may be performing for this species during infection. In vitro, the presence of capsule significantly inhibits the deposition of complement components onto the bacterium, and has been shown to measurably reduce phagocytosis of the bacterium by macrophages (Favre-Bonte et al., 1999; Cortes et al., 2002). The production of capsule also appears to inhibit the proper assembly of type 1 fimbriae on the bacterial surface, and may lead to a transcriptional inhibition in the production of another adhesin (Matatov et al., 1999). Accordingly, isogenic capsule-negative strains show higher levels of adherence to, and invasion of, a variety of cultured cells when compared with wild-type strains (Favre-Bonte et al., 1999; Sahly et al., 2000).

The importance of capsule as a virulence factor for K. pneumoniae has also been examined using in vivo models of colonization and pathogenesis. Using a catheter inoculation method, a capsule-negative derivative failed to colonize mouse bladder tissue to the same density as wild type (Struve and Krogfelt, 2003). Mutants lacking capsule also fail to cause similar lethality via intraperitoneal (i.p.) injection (Meno and Amako, 1991). Cortes et al. (2002) demonstrated that an isogenic capsule mutant is less successful at colonizing the lungs of infected mice than a wild-type strain and fails to disseminate to the spleen following pulmonary inoculation.

Lipopolysaccharide (LPS) is a major component of Gram-negative bacterial cell walls and possesses potent immunomodulatory properties (Strieter et al., 1990). However, in the context of wild-type capsule production the significance of LPS as a virulence factor for K. pneumoniae is unclear. Three studies have examined the importance of LPS during pulmonary infection, and their conclusions vary significantly. Two studies find no attenuation of disease due to LPS deficiency (Cortes et al., 2002; Izquierdo et al., 2003), while the other study observes that an LPS mutant is unable to disseminate beyond the respiratory tract (Shankar-Sinha et al., 2004). Confounding these data is the observation that some studies on LPS mutants of K. pneumoniae have not verified wild-type capsule production in these strains; conversely, evaluations of capsule mutants have not always determined whether the mutants in question produce wild-type levels of LPS. With the potential overlap among polysaccharide biosynthetic pathways, in addition to the known interactions between these structures in the bacterial cell surface, it is not unreasonable to assume that insertional mutations which affect the production of one of these molecules would impact the production of both (Izquierdo et al., 2003).

There are few other K. pneumoniae components which have been implicated to be necessary during infection. A chromosomal region associated with allantoin metabolism is also suspected to play a role in vivo as a potential contributing factor to invasive liver infections, a common complication during K. pneumoniae systemic infection (Chou et al., 2004). A number of adhesins have been suggested as potential virulence factors, including type 3 fimbriae and the adhesin CF29K; however, their importance has yet to be demonstrated in vivo (Di Martino et al., 1995; 2003). Finally, a subset of insertional mutants defective for biofilm formation has been shown to cause reduced lethality in vivo, although the specific mechanism of attenuation has not yet been explored (Lavender et al., 2004).

Since its introduction in 1995, the signature-tagged mutagenesis (STM) screen has become a useful tool for identifying genes necessary for microbial survival (Hensel et al., 1995; Chiang et al., 1999). This method allows for efficient and high-throughput studies of a large number of randomly generated mutants. Among both Gram-positive and Gram-negative organisms a variety of genes have been identified using the STM technique, including studies on several bacterial pathogens which cause pneumonia (Shea et al., 2000).

Three previous in vivo mutagenic screens have been carried out to identify virulence factors of K. pneumoniae. The first of these examined the ability of 2200 STM mutants to colonize intestinal tissue. Only one mutant was found to be attenuated in both in vitro adherence to intestinal epithelial cells and in vivo mouse intestinal colonization; this mutant contained a transposon insertion in a putative homologue of Hmw1A, an adhesin from Haemophilus influenzae (Maroncle et al. 2002). The second screen examined 1440 mutants, 13 of which were attenuated in mouse models of both gastrointestinal colonization and urinary tract infection. Three of these mutants had insertions in genes for LPS synthesis, one contained a mutation in outer membrane protein A (OmpA), and the remaining mutants had insertions in previously unidentified loci (Struve et al., 2003). The third screen evaluated approximately 1000 K. pneumoniae genes using in vivo expression technology (IVET) to identify those that are only expressed during i.p. infection. Twenty-one genes were induced in vivo, two of which were putatively involved in iron acquisition, and the remainder were conserved hypothetical open reading frames (ORFs) (Lai et al., 2001). The roles of these in vivo induced genes during infection has not been determined.

Despite its prevalence as a cause of pneumonia, no published studies have investigated K. pneumoniae genes necessary for bacterial survival in the lungs. In order to better understand the bacterial factors important for K. pneumoniae pathogenesis, we began by establishing a mouse model of pulmonary infection. This model was subsequently employed to identify genetic loci required for bacterial survival in vivo. Our findings represent the first use of an in vivo mutagenic screen to identify genes necessary for this pathogen to cause pneumonia. In addition, analysis of spleen colonization provided insight into genes involved in the transition from pneumonia to systemic infection.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Intranasal model of K. pneumoniae infection

In light of the limited body of knowledge regarding the virulence factors of K. pneumoniae, we sought to establish a simple, reproducible animal model that recapitulated the major aspects of clinical K. pneumoniae disease. This study began with the characterization of wild-type K. pneumoniae infection in C57Bl/6j mice via intranasal inoculation. KPPR1 is a rifampicin-resistant derivative of a clinical pneumonia isolate that previously has been used for in vivo studies (Standiford et al., 1999). Infection via intranasal inoculation yielded a 50% lethal dose (LD50) value of 3.0 × 103 colony-forming units (cfu), which is similar to previously reported results for this strain (Ye et al., 2001; Yadav et al., 2003).

To further characterize the progression of disease, mice were inoculated with 1.8 × 104 cfu of KPPR1, and five mice were sacrificed at 12, 24, 48 and 72 h after infection to determine bacterial loads in each tissue (Fig. 1). The small variation in the number of bacteria isolated from different mice at the 12 h time point illustrates the reproducibility and consistency of this inoculation method. Between 12 and 48 h after infection, there was an approximately 100-fold increase in the quantity of bacteria recovered from the lung. All infected mice developed a disseminated infection by 48 h, as measured by viable bacterial counts in the spleen (Fig. 1). After 72 h after infection an increasing variability can be seen among the quantity of bacteria in these tissues, with some mice successfully clearing the infection and others succumbing to it (data not shown).


Figure 1. Dynamics of wild-type K. pneumoniae growth in mouse tissues. Mice were inoculated intranasally with 1.8 × 104 cfu of the KPPR1 strain in a total volume of 20 µl. At each time point trachea, lung and spleen tissues were dissected and plated for cfu per gram of tissue. Five mice were examined at each time point. Data from one representative experiment are shown. The limit of detection is approximately 10 organisms.

Download figure to PowerPoint

To examine the pathology induced during KPPR1 infection, infected tissues were also examined histologically. At 12 h after infection, marked changes in the bronchial epithelium were apparent, with hypertrophy and hyperplasia noticeable when compared with uninfected lungs (Fig. 2A and B). Between 12 and 48 h this infection progressed to a classic bronchopneumonia, with an increasing inflammatory infiltrate focused around the bronchioles (Fig. 2C and D). As expected during a Gram-negative bacterial pneumonia, polymorphonuclear leukocytes (PMNs) constituted the majority of this cellular influx. By 72 h the characteristic lobar pneumonia of Klebsiella infection had developed, with large areas of the lung becoming consolidated. In addition, areas of apparent cellular destruction could also be visualized (Fig. 2E). As the infection intensified, large numbers of PMNs and bacteria could be seen in aggregates throughout the airways (Fig. 2F).


Figure 2. Histological examination of wild-type K. pneumoniae lung infection. Each bar represents a distance of 50 microns. A. Uninfected bronchiole (20× magnification). B. Hypertrophy and hyperplasia in the bronchial epithelium at 12 h after infection (20× magnification). C. Some inflammation visible at 12 h after infection. Arrow indicates area of PMN infiltration (4× magnification). D. Increasing amount of cellular infiltrate at 48 h after infection. Arrow indicates an area which is becoming consolidated (4× magnification). E. At 72 h after infection, most of the lung tissue has been overwhelmed by an influx of PMNs. Arrows indicate areas of clearance and destruction of lung tissue around blood vessels (4× magnification). F. High bacterial concentrations can be visualized in the airways at 72 h. Arrows point to aggregates of PMNs and bacteria in the airway (100× magnification).

Download figure to PowerPoint

Isolation of capsule-deficient mutant

The above results indicated that the KPPR1 strain is able to successfully infect the mouse respiratory tract, with subsequent disseminated infection and mortality. To verify the utility of this model in distinguishing between virulent and avirulent strains of K. pneumoniae, an isogenic mutant deficient in capsular polysaccharide synthesis was isolated. Wild-type bacteria were mutagenized with a mini-Tn5Km2 transposon, and candidate capsule-deficient mutants were identified by colony morphology on Luria–Bertani (LB) agar (Fig. 3A). Sequence analysis of one mutant identified a transposon insertion in cpsB, a phosphomannomutase gene located at the 3′ end of the type 2 capsule synthesis locus in K. pneumoniae (Arakawa et al., 1995). The inability of this mutant to produce capsular polysaccharide was verified by measuring the production of uronic acid, a significant constituent of K. pneumoniae capsule. A marked decrease in uronic acid production by this cpsB mutant strain was observed in vitro when compared with KPPR1 despite a similar rate of growth (Fig. 3B and C).


Figure 3. Identification of a K. pneumoniae capsular polysaccharide mutant. A. Colony morphology of KPPR1 and the cpsB mutant after overnight growth on LB agar at 37°C. B and C. (B) Growth curves and (C) quantification of uronic acid production by both strains. Experiments were performed in duplicate, and data shown are from two independent experiments. Graphs represent mean values ± SEM. *P < 0.05, **P < 0.01 (two-way anova).

Download figure to PowerPoint

Evaluation of the cpsB mutant via intranasal infection

Initial studies in vivo demonstrated that the cpsB mutant was severely attenuated in its ability to cause disease via intranasal inoculation. All infected mice survived intranasal challenge with as many as 9 × 108 cfu of the cpsB mutant (data not shown). Furthermore, when inoculated at a dose close to the wild-type LD50 (1 × 104 cfu), the cpsB mutant bacteria were cleared from all tissues by 48 h (data not shown). To provide a more relevant comparison to the wild-type infection, all subsequent infections with the cpsB mutant were performed with a dose almost 1000-fold higher than wild type (∼1 × 107 cfu for the cpsB mutant, as compared with ∼1 × 104 cfu for KPPR1). At these inocula both wild type- and cpsB-infected lungs contain equivalent bacterial loads at the 12 h time point.

Analysis of the cpsB mutant revealed significantly lower bacterial counts in all tissues examined throughout the course of infection (Fig. 4). This capsule mutant appeared to be unable to persist in the trachea, as bacterial counts were below the limit of detection by the 48 h time point (Fig. 4). The wild-type strain grew significantly in the lung environment, while the cpsB mutant was increasingly cleared from the lungs at every time point, despite its dramatically higher dose of inoculation. However, the cpsB mutant was never completely cleared from the lungs, and could still be found in the tissue 5 days after infection (Fig. 4). The cpsB mutant was also never detected beyond the respiratory tract; bacterial counts were below the limit of detection in the spleen at every time point (Fig. 4).


Figure 4. Comparison of intranasal infection by wild-type KPPR1 and an isogenic cpsB mutant. Mice were inoculated with either 1.8 × 104 cfu of the wild-type strain, or 8.7 × 106 cfu of the mutant. Data shown are from five infected mice per time point. Mean values are plotted ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 (two-way anova). Data from one representative experiment are shown. The limit of detection is approximately 10 organisms.

Download figure to PowerPoint

Intranasal STM screen

The previous experiments demonstrated that the intranasal mouse model could distinguish significant differences between wild-type and attenuated K. pneumoniae strains. The larger goal of this study was to identify novel genes important in K. pneumoniae pathogenesis. To this end we performed an STM screen to identify genetic loci necessary for K. pneumoniae growth in vivo. A total of 4800 independent transposon mutants of the KPPR1 strain were inoculated intranasally in pools of 48 mutants each, and examined for their ability to be recovered from the lungs or spleen. A total of 128 mutants were not recovered from either tissue, 308 were not recovered from the spleen but were recovered from the lungs, and 23 were only recovered from the spleen and were not found in the lungs. Overall, 9.6% of the mutants tested were attenuated in at least one of the tissues examined during this primary screening step, which is similar to the overall rates of attenuation for STM screens involving Gram-negative pathogens and in vivo models of disease (Shea et al., 2000).

In order to narrow our further investigations to genes which harboured the greatest in vivo importance, we focused on those mutants that were not recovered from either lung or spleen tissues. Southern hybridizations revealed that only 22 of these mutants appeared to have insertions in similar regions of DNA. Transposon locations were determined for the 106 unique mutants as described in Experimental procedures. Insertions were discovered in genes encoding a variety of different systems, including regulators, outer membrane components, transporters, and a large collection of genes with homology to hypothetical ORFs or genes of unknown function (Table 1).

Table 1.  Identification of genetic loci required for K. pneumoniae growth in both lung and spleen tissues.a
Functional classStrain No.NCBI GenBank homology putative functionGenusCompetitive indices mean values
In vitro Lung n Spleenn
  • a

    . A total of 106 mutants which were not recovered from either lung or spleen tissues were sequenced, and categorized by their putative functions.

  • The first part of each strain number indicates the pool which each mutant belongs to, and the second part indicates which tagged transposon was the cause of the insertional mutation. Sequence at the transposon insertion site was determined by using the chromosome capture technique as described, and the putative functions of these genes were determined by comparing sequences with NCBI GenBank database. If the K. pneumoniae sequence showed a particular homology to a sequence from one genus, then that genus is indicated. Those loci with homology to hypothetical ORFs, or with poor homology to genes in the database (e-values at or less than e-5) are grouped together. Mean in vitro and in vivo competitive index values are also shown for both lung and spleen tissues (n = number of mice infected).

Cell metabolism 58-32Acetoin dehydrogenase      
 85-28Chaperone Escherichia      
 66-45Cytochrome D, subunit B Klebsiella      
 11-34Cystathionine beta-lyase Vibrio      
 94-34Dehydrogenase Shigella      
 14-6Glutamate racemase      
 49-1Glutamate racemase      
 81-38Glutamate racemase      
 26-23Intracellular protease Pseudomonas      
 11-25 l-fuculose kinase Salmonella      
 11-28 l-fuculose kinase Salmonella      
  8-41Methionine sulphoxide reductase 2.690.662 0.502
 99-44MobB-like protein Escherichia      
 12-3Phosphoenolpyruvate synthase Salmonella      
 11-21ppGpp synthase  0.622 0.142
 97-28Recombination associated protein (rdgC) Haemophilus      
 32-2Spheroplast protein Y Escherichia      
 44-48Sulphatase Salmonella      
  1-25Thiamine pyrophosphate-requiring enzyme  0.012 0.022
 12-25Transthyretin-like protein Salmonella      
 63-29Virulence-associated protein (vacB) 1.921.334 2.164
Outer membrane and cell surface components 51-29cps locus ORF2 Klebsiella  1.10E-042 1.35E-052
  2-35GlcNAc transferase; ECA synthesis 6.480.055 5.40E-065
  9-35GlcNAc transferase; ECA synthesis 4.19    
 21-35GlcNAc transferase; ECA synthesis 3.030.033 0.043
 32-25GlcNAc transferase; ECA synthesis 0.762.143 3.983
 32-35GlcNAc transferase; ECA synthesis  0.022 1.40E-032
 33-35GlcNAc transferase; ECA synthesis      
 51-35GlcNAc transferase; ECA synthesis      
 92-35GlcNAc transferase; ECA synthesis  6.90E-032 02
 22-7O antigen export (rfbB) Klebsiella 0.975.10E-042 02
 87-32O antigen export (rfbB) Klebsiella 1.721.50E-033 1.70E-043
 31-29OmpK37 porin Klebsiella  1.132 0.941
 97-29Siderophore receptor (iroN) Klebsiella 0.760.863 1.503
 26-2UDP-galactopyranose mutase      
Regulators  1-7 cyn operon transcriptional activator Escherichia 1.930.273 2.123
  9-1mdr transcriptional regulator (ramA) Klebsiella  0.64217.222
 16-28mdr transcriptional regulator (ramA) Klebsiella 0.94    
 43-42mdr transcriptional regulator (ramA) Klebsiella      
 43-48mdr transcriptional regulator (ramA) Klebsiella      
 49-37mdr transcriptional regulator (ramA) Klebsiella      
 89-32mdr transcriptional regulator (ramA) Klebsiella      
 81-25Phosphate transport regulator (phoU)      
 21-37Putative sensor kinase (yojN) Escherichia 2.000.384 0.843
 81-45Anti-σE Salmonella 2.494.73E-043 7.00E-043
 43-39tetR regulator (ybiH) Escherichia 1.310.672 0.712
 43-22Virulence protein S (sensor for evgA) Klebsiella 1.170.342 2.942
Transporters 57-8ABC transport, ATP-binding potA Escherichia 0.710.794 0.443
 14-28ABC transporter dppB Pseudomonas 9.972.243 0.973
  7-13ABC transporter, inner membrane Yersinia 0.910.102 1.701
 81-24Glutamate transport ATPase Salmonella      
 14-12Major facilitator family transporter Pseudomonas 1.330.332 0.522
 21-6Maltose transport permease malG Escherichia      
 77-25Potassium channel trkA Yersinia 1.150.984 0.834
 26-20Putative PTS system protein Streptococcus      
 43-30Putative PTS system protein Clostridium      
 43-40Transport protein ybjL Salmonellla      
Hypothetical ORFs/homology < e−5 80-38ATPase subunit Corynebacterium      
 39-13Cell surface protein swmA Synechococcus      
 49-36CheY-like histidine kinase Burkholderia      
 43-1eps11p Streptococcus 1.31    
 14-38FImbrial subunit (fimA) Salmonella 7.381.552 0.332
  7-37GLutamate racemase      
  1-3Hypo. phage protein Shigella      
 16-12Hypo. protein 3′ to ramA Klebsiella      
 29-23Hypo. protein 3′ to ramA Klebsiella      
 78-28Hypo. protein 3′ to ramA Klebsiella      
 98-28Hypo. protein 3′ to ramA Klebsiella      
100-44Hypo. protein 3′ to ramA Klebsiella 0.760.222 0.582
 73-32Hypo. protein Photorhabdus      
 94-32Hypo. protein Escherichia  0.843 0.863
 71-44Hypo. protein CBG11431 1.120.773 0.083
 91-38Hypo. protein COG2128 Burkholderia 0.990.433 0.333
 39-24Hypo. protein CV1384 Chromobacterium  1.053 1.203
 25-42Hypo. protein ECs2270 Escherichia      
 39-42Hypo. protein ECs2270 Escherichia      
 80-34Hypo. protein jhp0089 Helicobacter      
100-28Hypo. protein STY0934 Salmonella  1.282 0.772
 71-42Hypo. protein STY3048 Salmonella  1.263 1.003
 18-22Hypo. protein STY4089 Salmonella 4.531.72311.513
 94-42Hypo. protein VC1317 Vibrio      
 32-1Hypo. protein ybdJ Escherichia  1.193 2.203
104-24Hypo. protein ybdN Escherichia  0.733 0.623
 77-47Hypo. protein ycdB Escherichia      
 98-34Hypo. protein yjeJ Escherichia      
 80-47Hypo. protein ydiY Escherichia  1.113 0.432
 21-26Hypo. protein YP01467 Yersinia 1.451.552 0.331
 56-19Hypo. protein YP01467 Yersinia 1.341.363 0.412
  7-25HP1 phage Haemophilus  11.552 1.121
 16-8None  1.223 7.013
 63-25None 2.780.592 4.062
 51-2PTS system protein Streptococcus      
 63-34Putative phospholipase Chromobacterium      
 32-6Ribonucleotide reductase Leishmania      
 32-14Surface antigen p30 Toxoplasma      

An STM screen identifies mutants unable to successfully compete with a large pool of other bacterial strains under a selective condition. Several mechanisms can account for an attenuated phenotype; therefore, it is necessary to perform secondary screens to verify those mutants with genuine defects for growth in vivo. The competitive index (CI) assay is often used to examine the growth of an individual mutant in the presence of the wild-type strain. In vitro CI experiments were performed first to verify that strains did not have defects under normal laboratory conditions. None of the tested strains showed a significant deficiency when grown in the presence of the wild-type strain, with all exhibiting CI values greater than 0.5 (Table 1). These results suggest that any in vivo phenotypes are the result of a specific defect for survival in the host.

When the cpsB mutant was examined via in vivo CI, CI values in both lung and spleen tissues were found to be significantly below 1 (lung CI of 10−6, and spleen CI of 0). The large majority of mutants tested by the CI assay exhibited values not significantly less than 1, suggesting that these mutants were not deficient for in vivo growth. However, several mutants identified in the STM screen showed significant attenuation by CI, among them mutants in LPS synthesis, enterobacterial common antigen (ECA) production and a mutant in an anti-σE homologue (Table 1).

In order to test the utility of the CI assay as a secondary screen, we carried out a series of independent colonization experiments with a small number of STM-identified mutants. Mice were infected with either wild-type or mutant strains, and bacterial counts in lung and spleen tissues were determined at 72 h after infection. Some mutants which had shown a significant CI value also had a significant independent infection phenotype (Fig. 5, Class I). However, for a number of mutants there was a poor correlation between the CI results and the ability of a particular mutant to independently cause disease. Class II mutants demonstrated low CI values in the lung, and Class III mutants possessed low CI values in the spleen. After examining these mutant strains using independent infections, both of these groups showed no significant deviation from the wild-type infection in either tissue (Fig. 5B). The converse was also observed; despite having shown a negligible phenotype in CI experiments, some mutants were limited in their ability to grow during independent in vivo infection (Fig. 5, Class IV).


Figure 5. Competitive index (CI) data do not always correlate with independent infection results for K. pneumoniae mutants. A. Competitive index scores for a number of different STM mutants, classified by their putative virulence defect. CI values are considered significant if they are below 0.5. B. Results from independent infections. Mice were inoculated with approximately 1 × 104 cfu, and at 72 h after infection lung and spleen tissues were assayed for bacterial colonization per gram of tissue.

Download figure to PowerPoint

These results demonstrated that the CI assay would only be useful for identifying the most dramatically attenuated mutants from this screen. In order to evaluate mutants with more subtle in vivo phenotypes, the majority of our secondary screening experiments involved the use of small-scale independent infections. The peak of infection at 72 h was used as the time point for these comparisons (Fig. 6). Among the mutants examined several unique phenotypes were identified, and these mutants will provide a number of opportunities for further study.


Figure 6. Independent intranasal infections of STM mutants. At least three mice were inoculated with approximately 1 × 104 cfu of each mutant strain, and trachea, lung and spleen tissues were harvested at 72 h after infection. Wild-type data are compiled from 10 infected mice, and are shown ± SEM. Shown in boxes are those mutants with a mean cfu g−1 measurement at least one log lower than wild type. *P < 0.01 (one-way anova, as compared with wild-type counts).

Download figure to PowerPoint

ECA synthesis mutants: polysaccharide synthesis

Eight independent insertions were identified in the ECA synthetic locus, and demonstrated attenuation in both lung and spleen tissue in the primary screen (Table 1, Fig. 7). To further examine the importance of ECA during infection two mutants were chosen for detailed study. Mutant 2-35 has an insertion in wecA, the first ORF of the ECA synthetic locus and a gene which has previously been shown to be required for both LPS and ECA production (Meier-Dieter et al., 1992; Toth et al., 1999). Mutant 32-25 has a transposon insertion in the ECA polymerase enzyme wzyE, which is necessary for ECA synthesis in Escherichia coli (Keenleyside and Whitfield, 1999).


Figure 7. Enterobacterial common antigen synthesis locus. Assignments of each ORF were made based on blastx similarity of sequence from the K. pneumoniae genome sequencing project ( with the NCBI Entrez database. Sites of the transposon insertions are indicated with the identifying number of each mutant strain. The lines above wecE and wecG indicate the approximate size and position of the two deletion strains that were constructed (see Experimental procedures for details).

Download figure to PowerPoint

To determine that in vivo attenuation of ECA mutants was not the result of defects in production of known virulence factors, both capsular polysaccharide and LPS were purified from these strains. A mutant in the LPS export locus rfbB (strain 22-7) was also used in these experiments. Measurements of capsule production in vitro revealed that both ECA mutants produced wild-type levels of uronic acid (Fig. 8A). However, the LPS mutant showed a significantly lower level of capsule production, reflecting earlier observations in K. pneumoniae that disruptions in LPS structure can also affect capsule production (Izquierdo et al., 2003). Another STM-derived mutant with an insertion in an O antigen synthesis gene demonstrated a similar capsule deficiency (strain 87-32; data not shown).


Figure 8. Polysaccharide production by various K. pneumoniae mutants. A. Determination of capsular polysaccharide production. Uronic acids were extracted from each strain and measured by a colorimetric assay. B. Lipopolysaccharide preparations from various K. pneumoniae strains. Hot phenol/chloroform extracts of LPS were separated on a 12.5% SDS-PAGE gel and stained to visualize bands corresponding to the LPS O antigen. C. ECA production determined via both passive haemagglutination and colony immunoblot assays for each strain.

Download figure to PowerPoint

Extracts from wild-type and capsule mutant K. pneumoniae strains produced a banding pattern on SDS-PAGE gels consistent with proper O antigen synthesis (Fig. 8B). The insertional LPS mutant completely lacked O antigen production as expected. As previous studies would indicate, the two ECA mutants exhibited different LPS synthesis phenotypes. The wecA mutant failed to produce O antigen, while the wzyE mutant produced O antigen that is visually identical to wild type. Two other wecA insertional mutants were tested (9-35 and 21-35), and both showed an identical LPS defect to mutant 2-35 (data not shown). To insure that this defect was due specifically to the transposon insertion in wecA, the mutant strain 2-35 was complemented with a plasmid encoding wecA, and wild-type LPS production was restored (data not shown).

In addition, these strains were examined for their ability to produce ECA in vitro (Fig. 8C). The results for each strain were consistent using both passive haemagglutination and colony immunoblot assays. The wild type, capsule mutant and LPS mutant strains were positive for ECA production (Fig. 8C). The wecA mutant was unable to produce ECA as expected. However, despite the reported role for WzyE in ECA polymerization, the insertional mutant in this gene was still able to produce ECA. In order to better understand the role of ECA in the context of other K. pneumoniae polysaccharides, two further mutants with in-frame deletions of the wecE and wecG genes were constructed. Both of the enzymes encoded by these genes have been confirmed to be necessary for ECA production in E. coli (Danese et al., 1998; Erbel et al., 2003). The K. pneumoniae wecE and wecG mutants failed to produce ECA, but did produce wild-type levels of capsule and LPS in vitro (Fig. 8C; data not shown).

ECA synthesis mutants: in vivo studies

In order to explore the roles that ECA and LPS may play during infection, intranasal infections were carried out with the above mutants. Two mutant strains were first examined for bacterial growth at the peak time points of 48 and 72 h (Fig. 9). The polysaccharide defects of the rfbB mutant were reflected in an inability of this strain to survive in any tissues examined. Other intranasal infections with this mutant show a dramatic decrease in bacterial concentration in trachea and lung tissues as early as 24 h after infection (data not shown). In contrast, the wzyE mutant could be found at wild-type concentrations in all tissues examined at both 48 and 72 h time points (Fig. 9). These data corroborate our in vitro results, which would have us predict severe attenuation for the rfbB mutant and no attenuation for the wzyE mutant.


Figure 9. The roles of ECA and LPS during intranasal infection. Mice were inoculated intranasally with 104 cfu of either the KPPR1, rfbB or wzyE strains. Five mice were infected per strain per time point. At 48 and 72 h after infection tissues were harvested and bacterial counts per gram of tissue were calculated.

Download figure to PowerPoint

A very different in vivo phenotype was observed with the wecA mutant. In both tracheal and lung tissues the wecA strain grew at a similar rate to the KPPR1 strain early in infection (Fig. 10). However, after 48 h this mutant appeared to decrease in bacterial concentration when compared with wild type, and there was a greater variability in bacterial counts in the respiratory tract for wecA-infected mice. As demonstrated earlier, wild-type infections do not show substantial variability in bacterial counts between mice, and while the variability in wecA mutant bacterial loads was not statistically significant, it may reflect an increase in the susceptibility of the wecA strain to the immune response.


Figure 10. Examining the impact of a mutation in the ECA locus on the growth of K. pneumoniae via intranasal inoculation. A total of 104 cfu of either wild type (KPPR1, black diamonds) or ECA mutant (wecA, open diamonds) were inoculated intranasally into mice. At each time point mice were sacrificed and tissues were measured for quantities of bacteria per gram of tissue. Data shown are combined from two different experiments. Each mouse that died during the course of the infection is indicated with an X.

Download figure to PowerPoint

A more substantial difference was found during systemic infection, as the wecA mutant demonstrated a lag in disseminating beyond the respiratory tract (Fig. 10). At 24 h, viable bacteria were recovered from the spleen in 40% of wild type-infected mice, compared with only 20% of wecA mutant-infected mice. This trend continued at 48 h, with 60% of wild type-infected mice positive for bacteria in the spleen, compared with 30% of mice infected with the wecA mutant (Fig. 10). Splenic colonization peaked for both strains at the 72 h time point; however, the mean concentration of wecA bacteria was approximately 100-fold lower than wild type (Fig. 10).

The next goal was to determine whether this defect in systemic dissemination was reflected in an attenuation in host lethality following intranasal inoculation. No significant difference was observed between wild-type and wecA LD50 values (data not shown). However, the rates of lethality were very different; the mean day to death (MDD) for the wild-type infection was 4 days, while the wecA mutant infection had an MDD measurement of 11.5 days (data not shown). This delay is consistent with the bacterial dissemination data, where at the 96 h time point 10 mice had died from wild-type infection compared with only five wecA mutant-infected mice (Fig. 10).

As the wecA mutant is defective in both ECA and LPS production, it was not clear which polysaccharide was playing the more important role in affecting bacterial survival in vivo. However, intranasal infections with both wecE and wecG mutants showed no significant differences in bacterial concentrations in lung or spleen tissues when compared with a wild-type infection (data not shown). Overall, these results suggest that ECA does not play a role in virulence that can be detected via an intranasal inoculation (Table 2).

Table 2.  Summary of in vivo and in vitro phenotypes of K. pneumoniae strains.
 Polysaccharide productionIntranasal inoculation
cpsB ++Avirulent
rfbB (22-7)++Avirulent
wecA (2-35)++Dissemination defect
wzyE (32-25)++++Virulent
wecE + ++Virulent
wecG+ ++Virulent

LPS and K. pneumoniae infection

These findings strongly indicate that the LPS defect present in the wecA mutant is responsible for its attenuation, and that its ECA deficiency does not affect its ability to survive in vivo. Previous studies on the role of LPS during pulmonary infection by K. pneumoniae have presented inconsistent conclusions. This mutant strain provided an opportunity to examine the importance of LPS for K. pneumoniae pathogenesis in the context of verified wild-type capsule production (Fig. 8A). Intraperitoneal infections were used to determine whether reduced splenic colonization of the wecA mutant lay either in an inability of this strain to successfully exit the lungs, or in an increased susceptibility to the systemic immune response.

Mouse survival curves following i.p. inoculation were significantly different (P = 0.0003), and LD50 values showed a corresponding difference. The i.p. LD50 of the wecA mutant was calculated to be 3.5 × 104 cfu, while the LD50 of the wild-type strain was greater than 1000-fold lower (the exact value could not be calculated, as 80% of wild type-infected mice did not survive infection with the lowest dose (15 cfu)). Following i.p. inoculation bacterial counts from both peritoneal lavage fluid and the bloodstream were measured, further demonstrating the significant attenuation of the wecA strain. The mutant bacteria were present at significantly lower concentrations than the wild-type strain in lavage samples and were not detected in the bloodstream of infected mice, while a majority of wild type-infected mice had developed bacteraemia at this time point (Fig. 11).


Figure 11. Intraperitoneal growth of K. pneumoniae. Bacteria were inoculated via i.p. injection, and 24 h later peritoneal lavage fluid and whole blood samples were taken for bacterial counts. Data are representative of two independent experiments. Statistical significance was calculated using a Mann–Whitney t-test.

Download figure to PowerPoint

Several mechanisms could be responsible for attenuation of the wecA mutant. Initial experiments indicate that this mutant does not demonstrate an increased rate of killing in the presence of macrophages (data not shown). In addition, serum resistance experiments in vitro have shown no defects in survival of the wecA mutant, even when performed in undiluted normal human serum (data not shown).


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

In this study we have presented a comprehensive description of an intranasal mouse model of K. pneumoniae infection. By using an intranasal inoculation method with small volume aliquots, we observed remarkably consistent results between mice and between experiments. Infection with a clinical pneumonia isolate bears striking similarities to human K. pneumoniae infection, including bacterial persistence in the trachea, a remarkable rate of bacterial replication in the lungs, a dramatic cellular infiltrate predominantly composed of neutrophils and a high rate of lethality upon subsequent bacterial dissemination. In particular, histopathological examination illustrates that the infection begins as a bronchopneumonia and progresses to an overwhelming lobar pneumonia and subsequent consolidation of much of the lung tissue.

The choice to use an intranasal rather than an intratracheal inoculation was influenced by several factors. The intranasal inoculation minimizes procedural trauma, and more closely mimics a natural route of infection. In addition, recent work in group A Streptococcus indicates that the immune response against a pulmonary bacterial pathogen can be initiated by the nasal associated lymphoid tissue (Park et al., 2003). Clinical reports on K. pneumoniae infection also suggest that the trachea and nasopharynx are an important site of colonization and persistence (Eisenstadt and Crane, 1994; Hollander et al., 2001). As tracheal integrity is not disturbed during the inoculation, this tissue can also be used as another site for measuring bacterial growth and persistence.

Results from infections with a capsule polysaccharide mutant confirm that the loss of capsule significantly retards the ability of K. pneumoniae to cause both pneumonia and disseminated infection. The cpsB mutant is not able to grow in lung or tracheal tissues, and is unable to progress to a disseminated infection. This capsule mutant is also unable to cause host mortality even at extremely high inoculum concentrations. Preliminary comparisons also suggest that wild-type and cpsB mutant infections induce remarkably different host responses. Fluorescence activated cell sorter (FACS) analysis indicates that at 48 h after infection wild type-infected lungs are comprised of 75% neutrophils, while cpsB mutant-infected lungs contain only about 5% neutrophils (M. Lawlor and V. Miller, unpubl. data). Further work is necessary to compare the cellular and cytokine responses among infections by different K. pneumoniae mutant strains.

As other investigators have shown, capsule polysaccharide plays an important role in protecting K. pneumoniae from intracellular killing, complement, surfactant, oxidative stress and phagocytosis (Domenico et al., 1994; 1999; Sahly et al., 2000). However, the cpsB mutation does not render the bacteria completely incapable of establishing infection. While it cannot proliferate in the lungs, the mutant can survive in this tissue, as small numbers of capsule mutant bacteria are still detectable 5 days after infection. This result suggests either that other virulence factors are important for the persistence of capsule-deficient Klebsiella, or that even non-pathogenic bacterial strains are difficult to completely clear from the lung environment.

Previous studies of K. pneumoniae have made attempts to draw parallels between in vitro and in vivo studies of pathogenesis. However, it has been observed that in vitro adherence to tissue culture cells does not correlate well with gastrointestinal or urinary tract colonization in mice (Struve and Krogfelt, 2003). The present study confirms the need for restraint in correlating in vivo and in vitro results in terms of pulmonary infection. An increase as large as 10-fold has been observed in the ability of K. pneumoniae strains to adhere to cultured lung cells after the deletion or suppression of capsule production (Favre-Bonte et al., 1999). In our hands we have observed a similar increase of in vitro adherence of the cpsB mutant to an A549 (human pneumocyte) cell line (M. Lawlor and V. Miller, unpubl. data), but during intranasal infection the cpsB mutant fails to persist in the trachea. Furthermore, histological examination reveals a significant quantity of wild-type bacteria binding to the apical surface of the mouse tracheal epithelium at all time points (M. Lawlor and V. Miller, unpubl. data). Favre-Bonte et al. (1999) have suggested that encapsulated K. pneumoniae may either adhere better to cells producing mucus, or suppress the natural mechanism of ciliary clearance. Examining the interaction of K. pneumoniae with primary mouse tracheal cells in vitro may approximate the tracheal environment more accurately, and allow examination of this site of bacterial persistence in greater detail.

Despite the well-documented epidemiological impact of K. pneumoniae disease, there is limited knowledge of the genes necessary for K. pneumoniae to cause pneumonia. Here we have combined our intranasal infection model with an STM approach to examine genes necessary for growth in lung tissue, and those required for the subsequent dissemination of K. pneumoniae beyond the respiratory tract. In preparing for the STM screen, the mutagenesis protocol was tailored to enrich for mutants in putative virulence factors. The transposon mutant selection was carried out on minimal media to minimize the number of strains with mutations in essential metabolic and biosynthetic pathways. In addition, mutants were selected for study that appeared phenotypically similar to wild-type K. pneumoniae, minimizing the number of mutants with insertions in the capsule polysaccharide synthesis locus. Through these efforts we hoped to bias our screen in favour of identifying virulence factors which had not been characterized previously in vivo.

In other STM screens, the body of knowledge about the pathogen of interest has allowed investigators to categorize mutants based on their deficiencies in known virulence factors. For example, investigators examining Pseudomonas aeruginosa mutants identified during an in vivo genetic screen used protease secretion, motility and a number of other assays to narrow down the number of mutants selected for further study (Potvin et al., 2003). Due to the paucity of knowledge regarding K. pneumoniae virulence factors, there were no in vitro assays which could be readily used as a secondary screen to evaluate the virulence of our bank of mutants. Instead, we first performed CI experiments to examine the in vivo growth of our mutants in the presence of the wild-type strain. This assay has often been used following a primary STM screen, as it recapitulates the in vivo competition between strains which occurs within each pool. After testing a large number of mutants by this method, we discovered that our mutants fell into two categories. A small number of mutants showed mean CI values well below 1; these mutants almost all had insertions in known or previously speculated virulence factors, including LPS synthesis genes and an anti-σE homologue. The second group of mutants included the large majority of strains identified in the screen, and exhibited CI values not significantly below 1. This large number of mutants with minor or insignificant CI results is not unprecedented, as the CI assay has been shown to be ineffective for several other mutagenesis studies (Autret et al., 2001; Sheehan et al., 2003).

Among the large number of published genetic screens to identify virulence factors, most have utilized a secondary screen to confirm mutants with genuine defects in pathogenesis. Competitive indices have traditionally been the secondary screening method of choice. However, very few studies have examined mutants further by using independent infection to verify the degree of attenuation. In the 62 published STM screens to date, there has not been a detailed comparison between CI assays and independent infections using the same group of mutants. In the course of this study it became apparent that while CI assays can identify mutants with a high degree of attenuation, this assay was not useful for analysing mutants with more subtle defects in virulence. This may be due to the dramatic production of capsule by K. pneumoniae, which could interfere with host–pathogen interactions and alter the dynamics of in vivo competition. Future in vivo screens may aim to use more stringent secondary and tertiary screening methods in order to clearly identify those genetic loci which play a genuine role in pathogenesis.

Therefore, as our secondary screen we took advantage of the sensitivity and reproducibility of the intranasal model. Experiments to establish this model indicated that the 72 h time point represents the peak of infection, with very high bacterial counts in trachea, lung and spleen tissues. Due to the consistency of infections between mice, we could infect as few as three mice per mutant and determine differences in the progression of disease. Mutants with varying degrees of attenuation in both lung and spleen growth were identified using this method. It proved to be particularly useful by indicating an attenuation of several mutants with insertions in hypothetical ORFs, which will provide a number of opportunities for further study.

Those mutants defective for growth in both lung and spleen tissues were selected for more detailed experiments. A glance at this group of mutants would suggest that K. pneumoniae relies on a complex cell membrane to survive in the adverse environment of the host. As a largely environmental species, K. pneumoniae's production of a variety of polysaccharides would allow survival in a range of harsh environments, and may prevent dessication in more arid conditions. When inoculated into a mammalian host, these characteristics also serve it well in protection from the immune response. In addition, K. pneumoniae strains produce nearly twice as much capsular polysaccharide when grown in minimal media as opposed to nutrient-rich media (Lai et al., 2003).

Several putative virulence factors have been described in K. pneumoniae, including a major outer membrane protein (OmpA), type 3 fimbriae and a handful of adhesive factors including CF29K. Most of these factors have been previously examined using in vitro assays, including tissue culture adherence and invasion, and biofilm formation (Di Martino et al., 1995; 2003; Matatov et al., 1999; Lavender et al., 2004). However, these factors have not been shown to play roles in pathogenesis using animal models of disease. In addition, none of these loci were identified in the group of mutants that were deficient for growth in both lung and spleen tissues. It is possible that these loci are necessary for bacterial survival in a tissue-specific manner, and that mutants in these genes are present in the remainder of mutant strains that we have yet to analyse further.

Eight attenuated mutants featured insertions in the ECA synthetic locus. ECA is an oligosaccharide complex which has been identified in a number of other in vivo STM screens, but has not been studied in further detail (Darwin and Miller, 1999; Bahrani-Mougeot et al., 2002). As a polysaccharide that could contribute to the complexity of the K. pneumoniae cell membrane, this locus is a prime candidate for a putative virulence factor. In CI assays, ECA mutants isolated in this screen appeared to be among the most attenuated mutants tested. During independent infections, the wecA mutant disseminated from the respiratory tract to the spleen at a markedly slower rate than wild type, and also caused a lower level of host lethality. WecA is an enzyme that has been shown in other species to play a key role in the production of both LPS and ECA polymers (Meier-Dieter et al., 1992). Concurrent studies with other mutants indicated that the wecA mutant attenuation results from an LPS defect rather than an ECA defect. It is also likely that the large number of mutants recovered with insertions in the 5′ region of the ECA synthesis locus are a reflection of the importance of LPS in the pathogenesis of K. pneumoniae. However, it remains an open question why ECA synthesis genes continue to be identified in these types of in vivo screens among a number of different enterobacterial species. Nothing is known about how ECA is presented on the surface of Klebsiella, or in what niches it may be playing a role to enhance bacterial survival. One possibility is that ECA may be important for the structure or presentation of other surface antigens.

The wecA mutant has proved useful for examining the importance of LPS during infection in the context of wild-type capsule production; other LPS-deficient mutants that have been isolated are defective in capsule production. Characterization of this mutant strain shows a defect in systemic colonization following intranasal inoculation, along with a reduction in host mortality, lower bacterial growth and decreased bloodstream dissemination following i.p. inoculation. Unlike some LPS mutants, this strain does not exhibit an increased susceptibility to serum killing in vitro. This may be reflective of the synergistic effects of humoral and cellular defences that could combine to make LPS mutants less successful for growth in vivo. Alternatively, the protection that the capsule provides to the bacterium may have a much more significant role in protecting K. pneumoniae from soluble factors than LPS. As the wecA mutant is defective in transiting out of either the peritoneal cavity or the pulmonary tract, it is tempting to speculate that the susceptibility of this strain is most pronounced in the lymph node, where there is an increased concentration and diversity of immune cells.

With the importance of cell surface structure to the growth of K. pneumoniae in a number of diverse environments, further work on the relative importance of ECA, LPS and capsule for K. pneumoniae is important. These studies may explore the structural linkages among these polysaccharide complexes in the outer membrane, along with the characteristics of independent mutants in each component using both in vitro and in vivo assays.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Bacterial strains and plasmids

KPPR1 is a rifampicin-resistant derivative of a K. pneumoniae subspecies pneumoniae clinical pneumonia isolate featuring a type 1 O antigen and a type 2 polysaccharide capsule (ATCC 43816). For all experiments, bacteria were grown overnight at 37°C, either on LB agar or shaking in LB broth. Medium was supplemented with 30 µg ml−1 rifampicin, 50 µg ml−1 kanamycin and/or 25 µg ml−1 chloramphenicol as needed.

Deletions strains were constructing using the pKO3 vector as described previously (Link et al., 1997). This vector is a temperature-sensitive plasmid that allows for homologous recombination of flanking sequences and construction of clean deletion mutants. Sequences flanking the wecE and wecG loci were obtained from the K. pneumoniae genome sequencing project at Washington University in St Louis ( For the wecE mutant, the following primers were used: A5′IN – AATATCTGCTGGATCTGCT; A3′IN – TAATCAAGCTCGG TTCCA; B5′IN – GTTAGCAAAAGCCTCACTGT; B3′IN – GATAGCCTCCCAACACGTA. For the wecG mutant, the following primers were used: A5′IN – GCTGCAGAACTACGA CAAG; A3′IN – CCCTCTTAACAGGTACAACG; B5′IN – ACT GTAACCAGGCGAGTTT; B3′IN – ACCAGAACTGCACATA GACC.

Intranasal infection model

Five- to seven-week-old female C57Bl/6j mice (Jackson Laboratories, Jackson, ME) were anaesthetized by i.p. injection with a mixture containing ketamine (8 mg ml−1) and xylazine (1.6 mg ml−1). Overnight bacterial cultures were diluted in phosphate-buffered saline (PBS), and 20 µl of the bacterial suspension was inoculated intranasally in four 5 µl aliquots. To facilitate consistent inoculations, mice were held vertically during inoculation, and were placed on a 45° incline while recovering from anaesthesia.

To determine the LD50, groups of five mice were intranasally inoculated with various doses of bacteria. Surviving mice were counted twice each day for 10 days, and LD50 values were calculated using the method of Reed and Muench (1938). MDD calculations were determined by statistical software.

In order to examine bacterial growth in the host mice were inoculated intranasally, and at various time points after infection sacrificed by a lethal injection of 300 µl of sodium pentobarbital (20 mg ml−1). Trachea, lungs and spleens were dissected, weighed and homogenized in 500 µl of PBS. Homogenates were plated on LB agar with rifampicin to determine cfu per gram of tissue.

To obtain samples for histological examination, mice were infected and sacrificed as described above. Organs were dissected and placed in 10% neutral buffered formalin. After 24 h, tissues were moved into 70% ethanol and stored until paraffin embedding. Tissues were sectioned and stained with haematoxylin and eosin by the Digestive Diseases Research Core at Washington University.

All animal experiments were performed under the guidance of the Department of Comparative Medicine at Washington University, and protocols were approved by the Animal Studies Committee.

Capsule-deficient mutant construction

To obtain a capsule-deficient mutant of KPPR1, a plasmid carrying the mini-Tn5Km2 transposon was introduced into KPPR1 via conjugation. K. pneumoniae bacteria containing a transposon insertion were selected using LB agar containing rifampicin and kanamycin. Small colonies were chosen as candidate capsule-deficient mutants. To identify mutants with an insertion in the locus responsible for capsular polysaccharide synthesis, chromosomal DNA flanking the transposon was recovered using chromosome capture via integration of the pKanπless vector as described below. VK20 (KPPR1 cpsB) has an insertion in cpsB, a phosphomannomutase at the 3′ end of the capsule synthesis locus.

Uronic acid quantification

To determine the quantity of uronic acid produced by K. pneumoniae, cultures were grown at 37°C in LB broth and at various time points 500 µl of culture was removed and stored at 4°C. Extraction and measurement of uronic acids were carried out as described previously (Favre-Bonte et al., 1999). A standard curve was calculated using known concentrations of glucuronolactone (Sigma Chemical, St Louis, MO).

STM screen: generation of mutant pools

Mini-Tn5Km2 transposons containing 96 unique signature tags in the pUT vector were acquired as a gift from David Holden at Imperial College, London (Hensel et al., 1995). These plasmids were transferred into E. coli strain S17-λpir, and those 48 transposons showing the greatest levels of hybridization were selected. Each one of 48 E. coli strains containing a signature-tagged transposon was mated with KPPR1 overnight on M9 minimal agar containing rifampicin and kanamycin to select for K. pneumoniae insertional mutants (Mazodier et al., 1989). In order to identify mutants with insertions outside the capsular polysaccharide synthesis locus, only colonies of wild-type morphology were picked. One hundred pools were assembled with 48 mutants each, and within each pool every mutant contained an insertion with a different tagged transposon. Pools were grown overnight in LB broth containing 25% glycerol, and then stored at −80°C. A total of 4800 mutants were generated, assembled and eventually screened in vivo.

STM screen: in vivo selection

Pools were grown overnight in LB broth containing rifampicin and kanamycin. All 48 mutants from each pool were combined, washed once in PBS and diluted to an approximate concentration of 1 × 109 cfu ml−1. Twenty microlitres of each pool were inoculated intranasally into four mice. Approximately 40 h after infection, mice were sacrificed and lungs and spleens were harvested from three mice. Organ homogenates were diluted onto LB agar containing rifampicin and kanamycin. Approximately 2000 colonies from each organ were combined, and chromosomal DNA was extracted using a standard phenol/chloroform method. Unique sequence tags were amplified from chromosomal DNA samples of both input pools and organs, and resulting products were digested to remove invariant sequences as previously described (Hensel et al., 1995). A second polymerase chain reaction (PCR) step using 32P-labelled dCTP was used to create radioactive tags from each pool, which were subsequently hybridized to nylon membranes containing samples of all 48 tags.

Southern hybridization analysis

Bacterial mutants that showed reduced hybridization in both lung and spleen tissues were chosen for Southern hybridization (Sambrook et al., 1989). Both EcoRV and KpnI restriction enzymes were used to digest genomic DNA samples from each mutant. The kanamycin cassette was used to probe DNA blots, and blots were labelled using the ECL Direct Nucleic Acid Labelling Kit (Amersham Biosciences, Buckinghamshire, UK). Mutants that contained insertions in similar size fragments using both restriction digests were classified as having mutations in the same locus, and only one of each of these mutants was analysed further.

Identification of transposon insertion sites

The chromosome capture method was used to recover the chromosome–transposon junction from mutants of interest. In order to do this, K. pneumoniae mutants were mated with an E. coli strain that contained the pKanπless vector, which features a chloramphenicol resistance cassette, a promoterless kanamycin cassette and a λpir-dependent origin of replication (K.H. Darwin and C. Nathan, unpubl. data). K. pneumoniae mutants that integrate the plasmid into the kanamycin cassette of the mini-Tn5Km2 transposon are selected for using media containing chloramphenicol. Chromosomal DNA was then extracted and digested, and ligation reactions were carried out in large volumes. The ligated DNA was then electroporated into E. coli strain S17-λpir, and colonies were selected for on LB agar containing both chloramphenicol and kanamycin. The resulting plasmids that were recovered contain varying amounts of chromosomal DNA neighbouring the transposon insertion site. These plasmids were subsequently sequenced from the ends of the pKanπless vector using primers P6 (CCTAGGCGGCCAGATCT GAT) and P7 (GCACTTGTGTATAAGAGTCAG). Sequencing was performed by the Protein and Nucleic Acid Chemistry Laboratory at Washington University.

In vitro competitive indices

Wild-type and mutant K. pneumoniae were grown separately overnight, and diluted to approximately 1 × 104 cfu ml−1. Ten microlitres of the wild-type strain and 10 µl of the mutant were inoculated into 5 ml of LB broth containing rifampicin and grown overnight at 37°C. This culture was diluted onto plate media containing rifampicin with and without kanamycin. CI values were calculated as previously described (Darwin and Miller, 1999).

In vivo competitive indices

Wild-type and mutant K. pneumoniae were grown separately overnight, and diluted to approximately 1 × 106 cfu ml−1. Ten microlitres of the wild-type strain and 10 µl of the mutant were combined and inoculated into mice intranasally. Two to five mice were infected per mutant strain. Dilutions of the inoculum were plated on LB agar containing rifampicin with and without kanamycin to determine the relative concentrations of wild-type and mutant bacteria. At 48 h after infection, lung and spleen tissues were harvested and homogenized. Dilutions of tissue homogenates were plated on media containing rifampicin with and without kanamycin. CI values were calculated as previously described (Darwin and Miller, 1999).

Lipopolysaccharide purification

Lipopolysaccharide purification was performed using a hot phenol and water protocol developed by Westphal and Jann (1965), using lysozyme and ultracentrifugation modifications as previously described (Johnson and Perry, 1976). Samples were fractionated using 12.5% SDS-PAGE gels, and stained using the Pro-Q Emerald 300 Lipopolysaccharide Gel Stain Kit (Molecular Probes, Eugene, OR). LPS samples were also detected using a silver stain as previously described (Tsai and Frasch, 1982).

ECA detection

The presence of ECA on the surface of bacteria was detected by both passive haemagglutination and colony immunoblot assays. Briefly, cells were grown overnight at 37°C in 10 ml of LB broth containing sodium salicylate at a final concentration of 10 mM. The cells in a given volume of culture were harvested by centrifugation, washed with 0.9% saline and resuspended to the original volume in 0.9% saline. Passive haemagglutination assays were carried out using 200 µl of the washed and resuspended cells employing polyclonal rabbit anti-ECA antiserum as previously described (Rick et al., 1985). Colony immunoblot assays were carried out using 2 µl of the washed and resuspended cells employing mouse anti-ECA monoclonal antibody mAb898 as described elsewhere (Meier-Dieter et al., 1989).

Intraperitoneal infections

Bacteria were grown overnight and diluted to a concentration of 105 cfu ml−1. One hundred microlitres of either wild-type or mutant bacteria were inoculated into each mouse by i.p. injection. Peritoneal lavage samples were obtained by injecting mice with 5 ml of cold PBS and removing 1 ml of lavage fluid. Blood samples were obtained via cardiac puncture.

Statistical analysis

All figures were prepared and subsequent statistical analysis was performed using Prism3 for Macintosh, version 3.0cx (GraphPad Software, San Diego, CA).


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

We wish to thank Krystal Onatolu and Mike Mahwold for their valuable technical assistance in studying the wecA mutant, and Kim Walker for her assistance during the STM screen. We also wish to thank Chris O’Connor, Matt Lawrenz and Bill Goldman for evaluating the manuscript. In addition, we are grateful to George Church for his generous donation of the pKO3 vector, and to David Holden for his gift of the signature-tagged transposons and helpful advice. Histological sections were prepared by the Washington University Digestive Diseases Research Core Facility, funded by Grant P30-DK52574 from the National Institutes of Health. M.S.L. was supported by a Lucille P. Markey predoctoral fellowship. This work was supported in part by funds to V.L.M. as a Pew Scholar in the Biomedical Sciences.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  • Arakawa, Y., Wacharotayankun, R., Nagatsuka, T., Ito, H., Kato, N., Ohta, M. (1995) Genomic organization of the Klebsiella pneumoniae cps region responsible for serotype K2 capsular polysaccharide synthesis in the virulent strain Chedid. J Bacteriol 177: 17881796.
  • Autret, N., Dubail, I., Trieu-Cuot, P., Berche, P., Charbit, A. (2001) Identification of new genes involved in the virulence of Listeria monocytogenes by signature-tagged transposon mutagenesis. Infect Immun 69: 20542065.
  • Bahrani-Mougeot, F.K., Buckles, E.L., Lockatell, C.V., Hebel, J.R., Johnson, D.E., Tang, C.M., Donnenberg, M.S. (2002) Type 1 fimbriae and extracellular polysaccharides are preeminent uropathogenic Escherichia coli virulence determinants in the murine urinary tract. Mol Microbiol 45: 10791093.
  • Chiang, S.L., Mekalanos, J.J., Holden, D.W. (1999) In vivo genetic analysis of bacterial virulence. Annu Rev Microbiol 53: 129154.
  • Chou, H.C., Lee, C.Z., Ma, L.C., Fang, C.T., Chang, S.C., Wang, J.T. (2004) Isolation of a chromosomal region of Klebsiella pneumoniae associated with allantoin metabolism and liver infection. Infect Immun 72: 37833792.
  • Cortes, G., Borrell, N., de Astorza, B., Gomez, C., Sauleda, J., Alberti, S. (2002) Molecular analysis of the contribution of the capsular polysaccharide and the lipopolysaccharide O side chain to the virulence of Klebsiella pneumoniae in a murine model of pneumonia. Infect Immun 70: 25832590.
  • Danese, P.N., Oliver, G.R., Barr, K., Bowman, G.D., Rick, P.D., Silhavy, T.J. (1998) Accumulation of the enterobacterial common antigen lipid II biosynthetic intermediate stimulates degP transcription in Escherichia coli. J Bacteriol 180: 58755884.
  • Darwin, A.J., Miller, V.L. (1999) Identification of Yersinia enterocolitica genes affecting survival in an animal host using signature-tagged transposon mutagenesis. Mol Microbiol 32: 5162.
  • Di Martino, P., Bertin, Y., Girardeau, J.P., Livrelli, V., Joly, B., Darfeuille-Michaud, A. (1995) Molecular characterization and adhesive properties of CF29K, an adhesin of Klebsiella pneumoniae strains involved in nosocomial infections. Infect Immun 63: 43364344.
  • Di Martino, P., Cafferini, N., Joly, B., Darfeuille-Michaud, A. (2003) Klebsiella pneumoniae type 3 pili facilitate adherence and biofilm formation on abiotic surfaces. Res Microbiol 154: 916.
  • Domenico, P., Salo, R.J., Cross, A.S., Cunha, B.A. (1994) Polysaccharide capsule-mediated resistance to opsonophagocytosis in Klebsiella pneumoniae. Infect Immun 62: 44954499.
  • Domenico, P., Tomas, J.M., Merino, S., Rubires, X., Cunha, B.A. (1999) Surface antigen exposure by bismuth dimercaprol suppression of Klebsiella pneumoniae capsular polysaccharide. Infect Immun 67: 664669.
  • Eisenstadt, J., Crane, L.R. (1994) Gram-negative bacillary pneumonias. In Respiratory Infections: Diagnosis and Management. Pennington, J.E. (ed.). New York: Raven Press, pp. 374378.
  • Ellis, M.E. (1998) Gram-negative bacillary pneumonia. In Infectious Diseases of the Respiratory Tract. Ellis, M.E. (ed.). Cambridge: Cambridge University Press, pp. 136139.
  • Erbel, P.J., Barr, K., Gao, N., Gerwig, G.J., Rick, P.D., Gardner, K.H. (2003) Identification and biosynthesis of cyclic enterobacterial common antigen in Escherichia coli. J Bacteriol 185: 19952004.
  • Favre-Bonte, S., Joly, B., Forestier, C. (1999) Consequences of reduction of Klebsiella pneumoniae capsule expression on interactions of this bacterium with epithelial cells. Infect Immun 67: 554561.
  • Hensel, M., Shea, J.E., Gleeson, C., Jones, M.D., Dalton, E., Holden, D.W. (1995) Simultaneous identification of bacterial virulence genes by negative selection. Science 269: 400403.
  • Hollander, R., Ebke, M., Barck, H., von Pritzbuer, E. (2001) Asymptomatic carriage of Klebsiella pneumoniae producing extended-spectrum beta-lactamase by patients in a neurological early rehabilitation unit: management of an outbreak. J Hosp Infect 48: 207213.
  • Izquierdo, L., Coderch, N., Pique, N., Bedini, E., Corsaro, M.M., Merino, S., et al. (2003) The Klebsiella pneumoniae wabG gene: role in biosynthesis of the core lipopolysaccharide and virulence. J Bacteriol 185: 72137221.
  • Johnson, K.G., Perry, M.B. (1976) Improved techniques for the preparation of bacterial lipopolysaccharides. Can J Microbiol 22: 2934.
  • Keenleyside, W., Whitfield, C. (1999) Genetics and biosynthesis of lipopolysaccharide O-antigens. In Endotoxin in Health and Disease. Brade, H., Opal, S., Vogel, S., Morrison, D. (eds). New York: Marcel Dekker, pp. 331358.
  • Lai, Y.C., Peng, H.L., Chang, H.Y. (2001) Identification of genes induced in vivo during Klebsiella pneumoniae CG43 infection. Infect Immun 69: 71407145.
  • Lai, Y.C., Peng, H.L., Chang, H.Y. (2003) RmpA2, an activator of capsule biosynthesis in Klebsiella pneumoniae CG43, regulates K2 cps gene expression at the transcriptional level. J Bacteriol 185: 788800.
  • Lavender, H.F., Jagnow, J.R., Clegg, S. (2004) Biofilm formation in vitro and virulence in vivo of mutants of Klebsiella pneumoniae. Infect Immun 72: 48884890.
  • Link, A.J., Phillips, D., Church, G.M. (1997) Methods for generating precise deletions and insertions in the genome of wild-type Escherichia coli: application to open reading frame characterization. J Bacteriol 179: 62286237.
  • Maroncle, N., Balestrino, D., Rich, C., Forestier, C. (2002) Identification of Klebsiella pneumoniae genes involved in intestinal colonization and adhesion using signature-tagged mutagenesis. Infect Immun 70: 47294734.
  • Matatov, R., Goldhar, J., Skutelsky, E., Sechter, I., Perry, R., Podschun, R., et al. (1999) Inability of encapsulated Klebsiella pneumoniae to assemble functional type 1 fimbriae on their surface. FEMS Microbiol Lett 179: 123130.
  • Mazodier, P., Petter, R., Thompson, C. (1989) Intergeneric conjugation between Escherichia coli and Streptomyces species. J Bacteriol 171: 35833585.
  • Meier-Dieter, U., Acker, G., Mayer, H. (1989) Detection of enterobacterial common antigen on bacterial cell surfaces by colony-immunoblotting: effect of its linkage to lipopolysaccharide. FEMS Microbiol Lett 50: 215219.
  • Meier-Dieter, U., Barr, K., Starman, R., Hatch, L., Rick, P.D. (1992) Nucleotide sequence of the Escherichia coli rfe gene involved in the synthesis of enterobacterial common antigen. Molecular cloning of the rferff gene cluster. J Biol Chem 267: 746753.
  • Meno, Y., Amako, K. (1991) The surface hydrophobicity and avirulent character of an encapsulated strain of Klebsiella pneumoniae. Microbiol Immunol 35: 841848.
  • Park, H.S., Francis, K.P., Yu, J., Cleary, P.P. (2003) Membranous cells in nasal-associated lymphoid tissue: a portal of entry for the respiratory mucosal pathogen group A Streptococcus. J Immunol 171: 25322537.
  • Potvin, E., Lehoux, D.E., Kukavica-Ibrulj, I., Richard, K.L., Sanschagrin, F., Lau, G.W., Levesque, R.C. (2003) In vivo functional genomics of Pseudomonas aeruginosa for high-throughput screening of new virulence factors and antibacterial targets. Environ Microbiol 5: 12941308.
  • Reed, I.J., Muench, H. (1938) A simple method of estimating fifty per cent endpoints. Am J Hyg 27: 493497.
  • Rick, P.D., Mayer, H., Neumeyer, B.A., Wolski, S., Bitter-Suermann, D. (1985) Biosynthesis of enterobacterial common antigen. J Bacteriol 162: 494503.
  • Sahly, H., Podschun, R. (1997) Clinical, bacteriological, and serological aspects of Klebsiella infections and their spondylarthropathic sequelae. Clin Diagn Lab Immunol 4: 393399.
  • Sahly, H., Podschun, R., Oelschlaeger, T.A., Greiwe, M., Parolis, H., Hasty, D., et al. (2000) Capsule impedes adhesion to and invasion of epithelial cells by Klebsiella pneumoniae. Infect Immun 68: 67446749.
  • Sambrook, J., Fritsch, E.F., Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press.
  • Seaton, D. (2000) Pneumonia. In Crofton and Douglas's Respiratory Diseases, Vol. 1. Seaton, A., Seaton, D., Leitch, A.G. (eds). Malden, MA: Blackwell Science, pp. 406407.
  • Shankar-Sinha, S., Valencia, G.A., Janes, B.K., Rosenberg, J.K., Whitfield, C., Bender, R.A., et al. (2004) The Klebsiella pneumoniae O antigen contributes to bacteremia and lethality during murine pneumonia. Infect Immun 72: 14231430.
  • Shea, J.E., Santangelo, J.D., Feldman, R.G. (2000) Signature-tagged mutagenesis in the identification of virulence genes in pathogens. Curr Opin Microbiol 3: 451458.
  • Sheehan, B.J., Bosse, J.T., Beddek, A.J., Rycroft, A.N., Kroll, J.S., Langford, P.R. (2003) Identification of Actinobacillus pleuropneumoniae genes important for survival during infection in its natural host. Infect Immun 71: 39603970.
  • Standiford, T.J., Wilkowski, J.M., Sisson, T.H., Hattori, N., Mehrad, B., Bucknell, K.A., Moore, T.A. (1999) Intrapulmonary tumor necrosis factor gene therapy increases bacterial clearance and survival in murine gram-negative pneumonia. Hum Gene Ther 10: 899909.
  • Strieter, R.M., Lynch, J.P., 3rd, Basha, M.A., Standiford, T.J., Kasahara, K., Kunkel, S.L. (1990) Host responses in mediating sepsis and adult respiratory distress syndrome. Semin Respir Infect 5: 233247.
  • Struve, C., Krogfelt, K.A. (2003) Role of capsule in Klebsiella pneumoniae virulence: lack of correlation between in vitro and in vivo studies. FEMS Microbiol Lett 218: 149154.
  • Struve, C., Forestier, C., Krogfelt, K.A. (2003) Application of a novel multi-screening signature-tagged mutagenesis assay for identification of Klebsiella pneumoniae genes essential in colonization and infection. Microbiology 149: 167176.
  • Timko, J. (2004) Changes of antimicrobial resistance and extended-spectrum beta-lactamase production in Klebsiella spp. strains. J Infect Chemother 10: 212215.
  • Toth, I.K., Thorpe, C.J., Bentley, S.D., Mulholland, V., Hyman, L.J., Perombelon, M.C., Salmond, G.P. (1999) Mutation in a gene required for lipopolysaccharide and enterobacterial common antigen biosynthesis affects virulence in the plant pathogen Erwinia carotovora subsp. Atroseptica. Mol Plant Microbe Interact 12: 499507.
  • Tsai, C.M., Frasch, C.E. (1982) A sensitive silver stain for detecting lipopolysaccharides in polyacrylamide gels. Anal Biochem 119: 115119.
  • Westphal, O., Jann, K. (1965) Bacterial lipopolysaccharides: extraction with phenol-water and further applications of the procedure. In Methods in Carbohydrate Chemistry, Vol. 5. Whistler, R. (ed.). New York: Academic Press, pp. 8392.
  • Yadav, V., Sharma, S., Harjai, K., Mohan, H., Chhibber, S. (2003) Induction & resolution of lobar pneumonia following intranasal instillation with Klebsiella pneumoniae in mice. Indian J Med Res 118: 4752.
  • Ye, P., Rodriguez, F.H., Kanaly, S., Stocking, K.L., Schurr, J., Schwarzenberger, P., et al. (2001) Requirement of interleukin 17 receptor signaling for lung CXC chemokine and granulocyte colony-stimulating factor expression, neutrophil recruitment, and host defense. J Exp Med 194: 519527.