The plastic nature of
K. pneumoniae genomes is well described and an increasing number of studies have elucidated the function of various components of the accessory genome of the pyogenic liver abscess-associated strain
K. pneumoniae NTUH-K2044. However, functional characterization of the accessory genome of strains associated with other types of infection is lacking. In order to investigate the plasticity of
K. pneumoniae associated with other infections, we previously interrogated the
pheV locus of sixteen clinical isolates from patients without pyogenic liver abscesses for the presence of foreign DNA elements [
13]. In this study, further tRIP-PCR interrogation of
K. pneumoniae KR116 using
met56-specific primers identified a novel GI, KpGI-5, inserted within its
met56 gene. KR116 had been isolated from the blood of a patient with pneumonia and neutropenic septicaemia. KpGI-5 was sequenced in this study and found to encode a putative γ
1-type CU fimbrial operon that has been named
fim2.
The genetic organization of
fim2 resembles that of the
K. pneumoniae fim operon and contains homologs of all eight
fim genes.
fim2 is predicted to code for a major fimbrial subunit (Fim2A), three minor fimbrial subunits (Fim2F, Fim2G, Fim2H) and homologs of the FimC and FimD chaperone and usher proteins, respectively, thus classifying this locus as a novel γ
1-type CU operon that putatively encodes a fimbrial appendage [
20]. A seventh predicted protein, Fim2I, exhibited 82% identity to FimI, a protein required for fimbrial biogenesis; however, the exact nature of this dependence remains unknown [
42]. Amino acid sequences of the eight
fim2 gene products showed 60 to 92% identity to cognate Fim proteins. Indeed, the two clusters would appear to be pseudoparalogs, homologs that appear to be paralogous but have ended up in the same genome by both vertical and horizontal gene transfer [
43]. The unique evolutionary origins of the
fim and
fim2 cluster are further highlighted by differences in transcriptional control. The
fim cluster is largely controlled by the FimB and FimE recombinases which together switch transcription on and off by inverting a 314 bp promoter-containing sequence called
fimS that lies upstream of
fimA[
22]. Exact copies, genetic remnants or potential functional-replacements of the 9 bp
fimS-flanking inverted repeats could not be identified within the putative
fim2 promoter region that lies upstream of
fim2A. Furthermore, as KpGI-5 lacks homologs of the FimB and FimE recombinases we conclude that
fim2 expression is not controlled via a
fimS-like switch mechanism. Additionally, the
fim2K gene within the
fim2 cluster encodes an EAL domain-containing protein that is similar to FimK, which has previously been shown to regulate type 1 fimbrial expression [
31]. FimK was hypothesised to exert its influence via the hydrolysis of the intracellular messenger c-di-GMP, which is known to regulate expression of virulence genes, motility and biofilm formation in other bacteria [
29]. The
in vitro and
in vivo function of Fim2K is currently under investigation.
Bacterial adhesion to and colonization of host cells is frequently mediated by a diverse assortment of afimbrial and fimbrial adhesins, each thought to possess a particular tissue tropism [
19]. The vast majority of
K. pneumoniae strains are able to produce type 1 fimbriae [
37,
44]. These structures are associated with mannose-sensitive agglutination of guinea pig red blood cells, a phenotype caused by interaction of the adhesin subunit FimH with terminally-exposed mannose residues in N-linked oligosaccharides on cell surfaces [
45]. Previously it has been shown that the FimH residues partaking in binding to mono- and tri-mannose moieties are highly conserved [
45]. The specific binding properties of Fim2H, the putative Fim2 adhesin, remain to be identified but it is unlikely to bind to mannose since only four out of the 13 mono- and tri-mannose binding residues of FimH are strictly conserved in Fim2H [
45]. This is also in agreement with the inability of
E. coli HB101 expressing
fim2 to agglutinate guinea pig red blood cells (data not shown), though the relevance of these data remain uncertain given the lack of visualisable fimbriae in this model.
Despite multiple attempts we were unable to visualize fimbrial structures using electron microscopy when the
fim2 operon was over-expressed in
E. coli HB101 and
K. pneumoniae C3091Δ
fimΔ
mrk. Paradoxically, biofilm forming ability appeared to be enhanced in this
fim2-expressing
E. coli strain. These results are similar to those of a study in which constitutive expression of four of seven
E. coli CU fimbrial operons was shown to cause phenotypic alternations despite the fact that fimbrial appendages could not be visualized by electron microscopy [
36]. Difficulty in visualisation of fimbriae by electron microscopy has also been described for the enterotoxigenic
E. coli fimbriae CS3 and CS6 and the putative Stg fimbriae of
Salmonella enterica serovar Typhi [
46-
48]. Most interestingly, when the latter was expressed in a bald
E. coli strain an enhanced ability to adhere to INT-407 epithelial cells was noted despite the absence of EM-observable fimbriae [
48]. It is possible that the
fim2 operon may code for a short and/or thin fimbrial structure that is not readily visualized by electron microscopy, or one that is extremely fragile. Conceivably, the hypothesized Fim2 appendages may be best expressed under biofilm-forming conditions, potentially explaining the enhanced biofilm-forming phenotype exhibited by HB101/pFim2-Ptrc, or in other specific
in vivo environments. Alternatively, the putative phosphodiesterase Fim2K may regulate
fim2 transcription and/or that of an unknown
E. coli adherence factor via a c-di-GMP-dependent pathway. Indeed, heterologous expression of
fim2K has been shown to complement a mutant lacking an EAL-bearing protein (van Aartsen and Rajakumar, unpublished data). Proposed future anti-Fim2A-based immunofluorescence and immunogold electron microscopy studies in addition to detailed characterisation of Fim2K will ultimately help determine the mechanism by which
fim2 contributes to biofilm formation.
The genomes of
E. coli K-12,
E. coli O157:H7 and
Salmonella Typhi possess numerous cryptic CU fimbrial operons that are tightly regulated and not expressed under the majority of
in vitro conditions tested [
35,
36,
49]. In this work,
fim2-specific transcript was identified in standard laboratory culture but the amount detected was 30- to 90-fold lower than that identified for
fim and
mrk, respectively. Compared to the
K. pneumoniae genome-averaged A
+
T content (~43%),
fim2 is AT-rich (53%) and the putative promoter region upstream of
fim2A possesses an even higher AT-content (73%). As moderate-to-marked upregulation of seven CU fimbrial operons has been reported in an
E. coli K-12 H-NS mutant [
36], the finding of an AT-rich
fim2 promoter region suggests that the H-NS protein may play a role in controlling this operon as well. Moreover, H-NS has been shown to bind preferentially to regions of horizontally-acquired DNA in
Salmonella Typhimurium and it is therefore possible this also occurs with KpGI-5 [
50]. Furthermore, in addition to Fim2K, KpGI-5 also encodes two other potential regulators one or more of which could alter
fim2 expression. By analogy with other CU systems, we propose that upregulation of
fim2 expression and biosynthesis of Fim2 fimbriae is likely to be triggered by specific environmental conditions and involve a complex interplay of multiple transcriptional regulators such as H-NS, Fim2K and/or FimK, and levels of expression of other surface components, such as the capsule [
31,
36,
38,
51]. It is important to note that even though
fim2 lacks an invertible promoter switch, it may still be stochastically controlled by a bistable regulatory circuit such as the DNA methylation-based system described in detail for
E. coli Pap fimbriae and it is therefore possible that single cell variants expressing
fim2 may exist [
51].
Analysis of three sequenced
K. pneumoniae strains revealed that, in addition to the
fim and
mrk operons, these genomes collectively encode at least six other CU fimbrial systems [
22,
23], one or more of which may perform an as-yet uncharacterised role in adhesion to target tissues. To investigate the role of
fim2 in virulence, isogenic
fim2 mutants were constructed and examined in three murine models, each focussed on primary infection of a distinct clinically-relevant anatomical site. Surprisingly, despite many fimbrial systems having been clearly implicated in virulence, we detected no clear evidence of attenuation (murine lung and urinary tract infection models) or reduction in colonizing ability (murine intestinal colonization model) in the
fim2-negative strains studied.
Intriguingly, examination of bladder CFU count-based CIs for the urinary tract infection experiments hinted at a subtle role for
fim2 in the colonization of bladder and kidney tissues. In both tissues, median wildtype CFU counts were approximately ten-fold higher than those of the
fim2 mutant, although when performed in a
fim negative background this difference was reversed and reduced in bladder and kidney samples, respectively. Nevertheless, the latter conflicting results may due to the markedly lower CFU counts obtained in the
fim negative background. As shown by neutral CI values in the lung tissue but an approximately 100-fold higher median liver CFU count for KR2107 as compared to its isogenic
fim2 mutant, the
fim2 locus would appear to be involved in systemic dissemination and/or survival of
K. pneumoniae following primary infection of the respiratory tract. However, given the noted lack of statistical significance, low numbers of mice examined and substantial mouse-to-mouse variation for these liver CFU data, no firm conclusions can be derived at present. As an aside, the previously demonstrated dramatic positive contribution of
fim to urovirulence in this murine model was also shown to be the case in the KR2107 background [
22,
23]. At an overview level, based on total CFU counts per liver and per kidney for the lung infection and ascending urinary tract infection models, respectively, there was a suggestion, though not supported statistically, of an ordered gradation amongst the four isogenic strains with the most-to-least virulent as follows: KR2107, KR2107
![[increment]](/corehtml/pmc/pmcents/x2206.gif)
fim2, KR2107
fim and KR2107
fimfim2. We speculate this relates to a Fim2-mediated enhancement of bacterial biofilm-forming-, adhesive- and/or invasive-potential under the
in vivo conditions tested. In addition, the predicted influence of Fim2K on the c-di-GMP regulatory circuit, may itself impact on virulence via regulation of Fim2, Fim and/or other virulence factors.
The
fim2 cluster was also assessed for its ability to contribute to biofilm formation. Gene knock-out experiments in KR2107 failed to reveal a role for
fim2 in biofilm formation. However, the function of the product of
fim2 may have been masked due to physical interference by the
K. pneumoniae capsule, a phenomenon previously observed with type 1 fimbriae [
38,
39]. Alternatively, it may be a function of limited
fim2 expression under the
in vitro conditions examined. Therefore, heterologous expression of
fim2 in the afimbriate
E. coli strain HB101 and the bald
fim2-negative
K. pneumoniae C3091
fimmrk mutant was pursued
. Yet again evidence of a
fim2-associated phenotype was elusive and only apparent in HB101 and then too only when crystal violet-staining data were standardised for total pre-wash cell numbers. Attempts to alleviate the observed growth retardation associated with over-expression of
fim2 in a HB101 background by reducing incubation temperature to 30°C and by providing rare tRNAs
in trans were unsuccessful. Furthermore, the observed growth retardation was highly reproducible even when newly generated HB101 strains possessing independently-constructed pFim2-Ptrc plasmids were used instead (van Aartsen and Rajakumar, unpublished data). Thus, it would appear that over-expression of
fim2 in HB101 was specifically responsible for this phenotype, though no comparable effect occurred with over-expression of
fim.
The presence of fim2 in more than one species and its global spread suggests that this horizontally acquired locus has been maintained within a subset of the Klebsiella population due to positive selection. Hence, although the role fim2 remains elusive, given the glimpses of functionality hinted at by our data and the evolutionary survival of this multi-gene entity, we hypothesize that putative Fim2 contributes to pathogenesis of infection and/or environmental persistence, at least under highly specific conditions.
1,4
