Colonisation of the GI tract plays a key role in the ability of K. pneumoniae to cause disease, stressing the need for an increased understanding of the mechanisms underlying this important feature. In this study, we employed a genomic-library-based approach to identify K. pneumoniae genes promoting GI colonisation. We demonstrated that screening of a K. pneumoniae C3091-based fosmid library, expressed in E. coli strain EPI100, in a mouse model led to the positive selection of clones containing genes which promote GI colonisation. Thus, oral ingestion of pooled library fosmid clones led to a successful selection of single clones capable of persistent colonisation of the mouse GI tract. This is a testament to the remarkably competitive environment of the GI tract where only clones having obtained a colonisation advantage will be able to colonise and persist in high numbers due to the presence of the endogenous microflora. When tested individually in growth competition experiments against EPI100 carrying the empty fosmid vector, each of the selected fosmid clones rapidly outcompeted the control strain. Based on these clones, we were able to identify C3091 genes and gene clusters conferring enhanced GI colonisation, including recA, galET and arcA.
Notably, EPI100 harbours deletions in recA
, suggesting that the selection of K. pneumoniae
reflects complementation of this missing E. coli
gene. RecA plays an essential role in chromosomal recombination and repair, and E. coli
RecA mutant strains have been shown to exhibit attenuated growth rates in vitro
]. Indeed, EPI100 carrying pACYC184-recA
also showed a clear growth advantage compared to the vector control when grown in LB broth. This finding verifies that RecA plays a significant role in bacterial growth in general and thus the GI colonisation promoting effect of recA
is most likely due to a generally enhanced growth rate of the recA
containing clone. Nevertheless, while the selection of RecA in the mouse model is not a surprising finding it serves as a proof of principle, regarding the validity of the screening approach.
The fact that pACYC184-galET
was unable to ferment galactose in vitro
was to be expected since EPI100 harbours deletions in galactokinase (GalK) and UTP-glucose-1-phosphate uridylyltransferase (GalU), both of which are necessary for growth on galactose [24
]. Instead, we observed an intriguing decreased sensitivity to bile salts in vitro
conferred by C3091-derived GalET.
Further studies are needed to characterise the mechanism underlying this phenotype and its physiological implications. However, we speculate that incorporation of C3091 GalET-mediated sugar-residues into the bacterial membrane, i.e. as a part of LPS as previously described [20
], may have an enhancing effect on the membrane stability, thus promoting decreased sensitivity to bile salts and possibly other compounds such as antimicrobial peptides present in the mouse GI tract. In support of this, enterohaemorrhagic E. coli gal
mutant strains have been shown to be 500-fold less able to colonise the GI tract of rabbits and 100-fold more susceptible to antimicrobial peptides than the parent strain [26
Together with the sensor transmitter protein ArcB, ArcA constitutes a two-component ArcAB system which functions as a global regulator of genes involved in metabolism in response to oxygen availability, primarily favouring anaerobic growth [27
]. ArcA homologues have, moreover, been implicated in regulating the expression of virulence factors and proteins involved in serum resistance [28
]. To our knowledge, the EPI100 strain does not harbour mutations in ArcAB, thus indicating a cumulative effect of native and K. pneumoniae
-derived ArcA activity promoting enhanced colonisation. To assess whether this effect was due to enhanced adaption to anaerobic growth in general, we tested EPI100 carrying pACYC184-arcA
for its potential enhanced ability to grow under anaerobic conditions in LB broth in competition with the EPI100 vector control. We did not observe any significant differences in the growth rate between the two strains. Thus, although a growth promoting effect of ArcA in the intestinal environment cannot be excluded from these in vitro
assays, the effect of ArcA on GI colonisation may instead be via the regulation of colonisation factors not related specifically to anaerobic growth. Notably, during screening of a K. pneumoniae
C3091 mutant library we previously found an arcB
transposon mutant to be markedly attenuated in GI colonisation [13
]. The identification of the arcAB
regulon by two fundamentally different screening approaches emphasizes the key role of ArcAB in GI colonisation and furthermore underscores the validity of the screening approaches.
Our screening assay also identified a Klebsiella
two-gene cluster of unknown function, here designated kpn_01507
, which conferred enhanced GI colonisation ability to EPI100. KPN_01507 is a putative membrane protein, whereas the use of SignalP 4.0 predicted the presence of a secretory signal peptide in KPN_01508, a signal targeting its passenger domain for translocation across the bacterial cytoplasmic membrane [30
]. These findings, therefore, suggest that KPN_01508 may be translocated and/or secreted from the cell. Interestingly, homologues of both genes are found among several sequenced strains of K. pneumoniae
but do not appear to be present in E. coli
. Future studies may reveal the function of these genes in GI colonisation.
The fact that genes associated with metabolism were selected in the in vivo
screening assay is not surprising since the ability to obtain nutrients for growth is essential for any GI colonizing organism. However, many highly conserved proteins involved in metabolism are increasingly recognized as having additional roles, some of which are related to bacterial virulence [31
]. The GalET cluster may be viewed as an example of such so-called moon-lighting proteins as the colonisation enhancing effect was not associated with galactose fermentation per se but was due to increased resistance against bile salt possibly mediated by the modification of LPS core synthesis.
A key limitation of the library-based technique is its inability to identify interactions among distant genetic loci. This limitation could be circumvented by using co-expressed plasmid- and fosmid-based genomic libraries as recently described [16
]. Thus, future studies combining the C3091 fosmid library with a co-expressed plasmid-based C3091 library may lead to the selection of more GI-enhancing genes than those obtained in this study.
The fact that our screening method is based on a laboratory E. coli strain, as opposed to a commensal E. coli isolate, raises another important point. Genes mutated in the laboratory strain, e.g. recA, would most likely not have been selected if the screening had been carried out using a commensal strain. However, since commensal E. coli are already excellent GI colonisers, it is possible that genes which are important for K. pneumoniae GI colonisation but also present in E. coli commensal strains will not be selected in the screening. However, if the objective is to specifically identify K. pneumoniae virulence genes, using a commensal E. coli strain as a host in the screening will be a favourable approach.
Using E. coli as a host has several advantages when it comes to construction, cloning, and expression of the fosmid library. However, a shortcoming is that although the assay identifies genes promoting colonisation in E. coli, additional studies involving the construction of specific mutants are warranted to verify the role of these genes in K. pneumoniae. Although future studies are needed to characterise the role of galET and kpn_01507/01508 in K. pneumoniae colonisation, as discussed above, both recA and arcA are expected to play a significant role in K. pneumoniae colonisation.