«(Über die Bedeutung der bakteriellen Genomplastizität für die Adaptation und Evolution asymptomatischer Bakteriurie (ABU) Escherichia coli ...»
Complete solubilisation of the membrane proteins is critical for proteomic analysis of membrane fractions. Using the carbonate extraction method, 19 different membrane associated proteins were identified (Table 17). This represents 73 % of the 26 predicted outer membrane proteins of E. coli strain K-12 (Molloy et al., 2000). However, many of the detected proteins (42 %) represented components of iron uptake systems which are not present in the strain K-12 strain MG1655.
The protein profiles of the three in vivo re-isolates were comparable. However, minor changes either in the composition or in the expression level could be detected. In general, protein separation was of a good quality and allowed precise alignment of the profiles from parent strain and its consecutive re-isolates (Fig. 47; Fig. 48; Fig. 49). Different protein isoforms could not always be perfectly aligned during in silico proteome comparison and are therefore either underlined or boxed as the same protein. All changes between the re-isolates and the wild type strain are summarised in Table 17.
While the outer membrane protein profiles of the strains KA25 and 83972 were very similar (Fig. 47), the outer membrane proteome of re-isolate CK12 exhibited minor changes (Fig.
48). The most prominent difference in the strain CK12 was the lack of the ferric aerobactin receptor protein IutA. Interestingly, the iron(III) dicitrate transport protein (FecA) was found to be located in almost the same position on the gel. Because of the almost identical
isoelectric points and molecular weights of these two proteins (Table 17), it is difficult to distinguish them when both proteins IutA and FecA are expressed together.
Fig. 47: Comparison of the outer membrane proteome of ABU strain 83972 and the in vivo re-isolate KA25. Red-channel, outer membrane proteins of re-isolate KA25; Green-channel, outer membrane proteins of strain 83972. Proteins with the same expression level are shown in yellow. On each gel 100 µg of proteins were separated and Coomassie-stained. Differently expressed proteins were identified by MALDI-TOF mass spectrometry.
These results are consistent with the microarray data, indicating that the iutA gene was 2-fold down- and the fecABCDE operon was 2-fold up-regulated relative to parent strain 83972 (Table 21). Moreover, the protein profiles showed that the amount of Iha, the exogenous ferric siderophore receptor was significantly decreased in strain CK12, while another siderophore receptor, IroN, seemed to be present in higher amounts.
Fig. 48: Comparison of the outer membrane proteome of ABU strain 83972 and the in vivo re-isolate CK12. Red-channel, outer membrane proteins of re-isolate CK12; Green-channel, outer membrane proteins of strain 83972. Proteins with the same expression level are shown in yellow. On each gel 100 µg of proteins were separated and Coomassie-stained. Differently expressed proteins were identified by MALDI-TOF mass spectrometry.
One more interesting protein that was found to be expressed in higher amounts in re-isolate CK12 than in strain 83972 is Tsx, the nucleoside-specific channel-forming protein. As already described in section 5.5.3, this protein facilitates the uptake of ribo- and deoxyribonucleosides from the environment. The tsx gene was 2-fold up-regulated in strain CK12 when compared to 83972. Moreover, other proteins involved in that uptake processes were identified to be present in higher amounts in the intracellular proteome of strain CK12 (Fig. 44).
Fig. 49: Comparison of the outer membrane proteome of ABU strain 83972 and the in vivo re-isolate SR12. Red-channel, outer membrane proteins of re-isolate SR12; Green-channel, outer membrane proteins of strain 83972. Proteins with the same expression level are shown in yellow. On each gel 100 µg of proteins were separated and Coomassie-stained. Differently expressed proteins were identified by MALDI-TOF mass spectrometry.
The most striking difference was the presence of multiple isomeric forms of the protein FliC.
As already shown, this strain is motile (Fig. 26) and many genes of the flagella regulon were up-regulated when compared to the parent strain 83972 (Table 18).
In addition to FliC, the expression of different components of iron uptake systems was affected. First of all, the protein FhuE could not be detected and FhuA was much lesser
expressed than in strain 83972. In contrast to re-isolate CK12, the Iha protein was present in higher amounts and IroN was less prevalent than in the ancestor strain. Also the ferrienterobactin receptor FepA turned out to be present in higher amounts in the re-isolate SR12. However, many differences on the protein level were not as clear as for strain CK12 and KA25. This might be due to the expression of flagella and a thus significant fraction of the FliC protein in the protein preparation.
The fact that in all outer membrane protein proteomes of the three re-isolates and the parent strain 83972 some prototypic outer membrane proteins like OmpF, OmpW, YfiO and ZipA could not be detected indicates the limitation of 2D gel analysis of outer membrane proteins which can frequently not be resolved due to their high hydrophobicity. Gel-free analysis of outer membrane protein fractions may be a suitable alternative to get a more complete picture of the outer membrane proteome.
Taken together, to assess changes in the outer membrane proteome and adaptation of ABU strain 83972 during growth in the human bladder, the membrane protein fractions of this strain and its in vivo re-isolates were compared. Together, 19 different membrane associated proteins were identified. Many of the detected proteins (42 %) represented components of iron uptake systems. Outer membrane protein profiles of the strains KA25 and 83972 were very similar, however, the outer membrane proteome of re-isolate CK12 and SR12 exhibited already minor changes. In isolate CK12, nucleoside uptake system was up-regulated what clearly correlated with microarray data. In addition to that, the expression of different components of iron uptake systems was affected in isolates SR12 and CK12.
6. Discussion Asymptomatic bacteriuria (ABU) probably represents the most frequent form of urinary tract infections. Up to 6 % of healthy individuals and up to 50 % of elderly patients are estimated to be colonized by ABU strains (Colgan et al., 2006). Despite ABU isolates frequently reach cell densities of 105 bacteria/ml urine, colonization of the urogenital tract usually occurs in these cases without symptoms as the urogenital epithelia are rarely damaged and no inflammation is induced (Lindberg et al., 1978; Wullt et al., 2003). Several virulenceassociated factors of uropathogenic E. coli that contribute to colonization of the urinary tract and symptomatic UTI have been well characterized. However, little is known about the virulence determinants of ABU isolates. Moreover, gene expression of asymptomatic isolates has not been studied in detail and there is no study concerning the host-driven bacterial evolution within the urinary tract, promoting bacteria to enter a commensal-like state.
In the first part of this work, a more general approach was used. To learn more about the characteristics of ABU isolates that may account for the ABU lifestyle, the genome content and phenotypic traits of eleven ABU isolates were studied in detail and compared to those of uropathogenic and non-pathogenic E. coli isolates. Later, the ABU model E. coli strain 83972 was used to analyze the genome flexibility and adaptive changes during human bladder colonisation.
6.1. Asymptomatic bacteriuria is caused by a heterogeneous group of E. coliisolates
Interestingly, ABU isolates represent a rather heterogeneous group of organisms with regard to their phylogenetic lineage and repertoire of typical virulence-associated genes of UPEC.
Although the majority of strains tested belongs to the ECOR groups B2 and D to which UPEC can be usually affiliated, some isolates could be grouped to the ECOR groups A and B1.
These lineages normally include non-pathogenic as well as intestinal pathogenic variants.
Accordingly, and in contrast to the strains of ECOR group B2 and D, typical UPEC virulenceassociated genes could not be detected in these ABU isolates and their pathoarray CGH barcode differs markedly from those ABU strains which seem to be more closely related to UPEC (Fig. 10; Table 7; Table 9)
The ABU isolates could be allocated to different phylogenetic lineages and different clonal groups. These results demonstrate that asymptomatic bacteriuria is not caused by one specialized clonal group of organisms. Instead, bacteria with different independent phylogenetic backgrounds are able to efficiently colonize the urinary tract without causing symptoms. Furthermore, comparison of the genome structure accessed by PFGE uncovered that even strains falling into the same ST differed markedly in the genome structure. It underlines the diversity and genome flexibility among strains causing ABU.
6.2. Impaired ability of ABU isolates to express typical UPEC virulencefactors
As expected, the ABU strains of ECOR group B2 and D exhibited a pathoarray CGH barcode similar to those of archetypal UPEC variants which cause symptomatic UTI. This indicates that UPEC virulence-associated genes are present in the genomes of these ABU isolates.
Their inability to cause symptomatic UTI cannot be attributed to the absence of such virulence-associated genes per se.
Therefore, comparison of the foc determinants of the F1C-fimbriae-negative ABU strains 83972 and 27 relative to F1C fimbriae-positive isolate 37 led to the discovery that one particular amino acid exchange in FocD (glutamine 472 → leucine) is responsible for the loss of the usher activity of FocD and thus the absence of functional F1C fimbriae in strains 83972 and 27 (Table 10). This glutamine residue is conserved among the related usher subunits FocD, FimD and SfaF (Fig. 15) and its exchange probably results in an altered conformation or stability of the usher protein thus impairing its function.
The accumulation of point mutations resulting in a loss of gene function is further corroborated by the DNA sequence comparison of pap determinants coding for P fimbriae. P fimbriae are considered as one of the most important virulence factors contributing to UTI (Plos et al., 1995; Vaisanen et al., 1981). Five out of eleven isolates tested were pap-positive (Table 9). Interestingly only one strain, ABU64, was able to express functional P fimbriae.
The fact that P fimbriae trigger mucosal inflammatory responses to Escherichia coli in the human urinary tract (Bergsten et al., 2005) may explain the frequency of P-fimbrial inactivation in ABU isolates.
The type 1 fimbriae-encoding gene cluster, besides its inactivation by point mutations in strains 5 and 57, seems to represent a rather unstable genomic region involved in partial chromosomal deletions resulting in loss of a central 4.2-kb part of the operon or in larger 29kb deletions including adjacent DNA stretches. It is an interesting observation that in all cases of partial fim gene cluster deletion the fimH gene which is frequently used in generally accepted screening tests as a marker for the presence of the type 1 fimbrial gene cluster (Johnson and Stell, 2000), stays intact. Furthermore, it is tempting to speculate that also the loss of functional type 1 fimbriae may be correlated with the ABU lifestyle as it is nonfunctional in the majority of strains tested.
The comparison of other virulence-associated characteristics such as motility, LPS phenotype and biofilm formation further supports the finding that E. coli ABU isolates are not characterized by a common phenotypic appearance. Consequently, the establishment of asymptomatic bacteriuria not solely depends on a specific set of bacterial traits, but results from different bacterial colonization strategies. In this context, it has been recently suggested, that increased growth rates in urine enable ABU isolate 83972 to outcompete, e.g.
uropathogenic E. coli isolates from symptomatic urinary tract infections (Roos et al., 2006b).
However, growth characteristics of examined eleven ABU isolates were very diverse, in a range from the non-pathogenic strain K-12 to that of 83972, underlying complexity of the ABU phenomena.
The detailed phenotypic and genotypic comparison demonstrated, that important virulenceassociated determinants such as those coding for α-hemolysin (hly), type 1- fimbriae (fim), Pfimbriae (pap) and F1C-fimbriae (foc) have been frequently inactivated in ABU isolates by point mutations and (IS element-mediated) deletions. These findings are in accordance with results published recently by Klemm and co-workers who described the inactivation of fimbrial adhesin determinants in ABU model strain 83972 by point mutations and deletions (Klemm et al., 2006; Roos et al., 2006a). Our data suggest that, in addition to the bacterial traits, also host factors which allow urinary tract colonization by less specialized E. coli variants and even by those harbouring a functional hly and pap determinant contribute to the development of asymptomatic bacteriuria.
The loss of virulence factors has been shown to reduce the host response to infection in animal models and specifically, the loss of fimbriae decreases the innate host response and
bacterial clearance from the urinary tract. More than 80 % of UPEC strains express P fimbriae, 14-30 % of UPEC strains express F1C fimbriae (Pere et al., 1987) and type 1 fimbrial expression is quite frequent. P fimbriae enhance the establishment of bacteriuria and trigger the innate defence by stimulating the production of cytokines, which orchestrate the subsequent recruitment of inflammatory cells. Type 1 fimbriae have a similar function in mice and have also been shown to enhance intracellular persistence in the mouse bladder mucosa, but these effects have not been reproduced in the human urinary tract (Bergsten et al., 2005;