«(Über die Bedeutung der bakteriellen Genomplastizität für die Adaptation und Evolution asymptomatischer Bakteriurie (ABU) Escherichia coli ...»
However, PCR-based detection of the F1C fimbriae-encoding gene cluster was also positive for strain 27 and 83972. Similarly, these three strains also differed in their ability to express functional F1C fimbriae. Comparison of the DNA sequence of the encoding foc determinant in these strains (Fig. 15) demonstrated that the A to T transition at the focD nucleotide position 1415 results in exchange of glutamine 472 against a leucine residue in the FocD fimbrial usher of the latter two strains. Mutation of this amino acid alone results in a nonfunctional FocD usher protein (Table 10).
Fig. 15 Amino acid alignment of FocD, SfaF and FimD fimbrial ushers. The conserved Gln472 residue of FocD has been indicated in red. EcN, E. coli Nissle 1917.
5.2.4. Expression of α-hemolysin The ability to express α-hemolysin was checked by plating bacteria on sheep blood agar plates. Only strains 20, 37, 63 and 64 were able to lyze erythrocytes, however, strains 27 and 83972 harboured all genes of the hly gene cluster. Nucleotide sequence comparison of the hly operon of the closely related strains ABU27, ABU37 and ABU83972 revealed that the nonhemolytic phenotype of strains 27 and 83972 relative to ABU37 is due to an A→T transition at the hlyA nucleotide position 416 thus resulting in a premature stop codon and thus a truncated HlyA toxin gene product (Fig. 16).
Fig. 16: Inactivation of the hly determinant in strains 27 and 83972 5.2.5. LPS O side chain expression The ABU isolates exhibited considerable diversity with regard to LPS O side chain expression (Fig. 17): strains that belong to ECOR group B2 either expressed long chain LPS or no side chains. Interestingly, all ST73 isolates tested exhibited a rough LPS phenotype. The O side chain length of the three isolates that belong to ECOR group B1 was shorter than that of smooth strains expressing long O side chains. ABU isolate 5 (ECOR group D) and 57 (ECOR group A) did not express O side chains.
Fig. 17: Analysis of the LPS phenotype among asymptomatic bacteriuria E. coli isolates 5.2.6. Biofim formation The ABU isolates differed markedly with regard to their ability to form biofilms (Fig. 18).
Even among strains of the same sequence type, i.e. the strains of ST73, more than threefold differences were observed in their biofilm formation.
Fig. 18: Analysis of biofilm formation of asymptomatic bacteriuria E. coli isolates in urine. The ability to form biofilms was compared to that of uropathogenic E. coli strain 536 that causes symptomatic urinary tract infections.
5.2.7. Growth characteristics of ABU E. coli isolates In order to characterize ABU isolates in terms of growth properties, the bacteria were grown statically at 37 °C in pooled human urine. As a control UPEC strains CFT073, 536, NU14 and J96, as well as non-pathogenic K-12 strain MG1655 were grown under the same conditions.
With respect to growth rate and their final optical density, all ABU strains grew well in urine (Fig. 19). The isolates 83972, 20, 21, 37 and 62, however, grew better in the early exponential phase than isolates that belong to ECOR group A or D. The ABU model strain 83972 grows as fast as other tested UPEC strains except strain NU14 which grew rather like the E. coli Kstrain MG1655.
Fig. 19: Growth characteristics of E. coli isolates in pooled human urine at 37 °C: A, B) ABU E. coli isolates; C) UPEC strains, K-12 strain MG1655 and ABU strain 83972.
Taken together, a number of phenotypic tests revealed significant differences among investigated isolates with regard to expression of virulence factors. ABU strains often carried virulence determinants on the chromosome but did not express them. DNA amplification and sequencing led to the discovery that in many cases loss of the function was due to point mutations or IS mediated DNA deletions. Regarding LPS expression, no common pattern could be observed, and finally, all ABU isolates grew very well in urine.
5.3. Adaptive flexibility and genome plasticity of model strain 83972 ABU strain 83972 has been successfully used in a medical treatment pilot study (2004-2006) for bladder colonization of patients suffering from recurrent urinary tract infections due to various bladder dysfunctions (Sunden et al., 2006). Access to consecutive re-isolates of strain 83972 from different patients allowed us to study bacterial adaptation in response to host colonization. Microbial genome rearrangements and changes in the global gene expression profiles of the individual re-isolates were analyzed and correlated to host response.
5.3.1. Patient colonization
In the Department of Urology of the Lund University hospital (Sweden), the patients participating in the pilot study of deliberate bladder colonization with ABU strain 83972 were treated with appropriate antibiotics to sterilize their urinary bladder. After an antibiotic-free interval, the patients were catheterized and the bladder was emptied. Thirty millilitres of E.
coli strain 83972 (105 colony-forming units (CFU)/ml) were instilled in the bladder and the catheter was removed. This procedure was repeated on the next 2 days. According to individual study protocols, subsequent urine samples were taken to assess host response parameters: interleukin 6 (IL 6) and interleukin 8 (IL 8) levels, numbers of polymorphonuclear cells (PMN), and to prove the success or failure of the colonization procedure. Urine samples have been obtained from the patients in monthly intervals and bacterial samples were stored in agar stamps at room temperature. Several re-isolates of strain 83972 from different time points were subjected to phenotypic and genotypic characterisation.
Fig. 20: Schematic representation of the experimental design. Patients P1 to P6 were colonized with ABU strain 83972. Blue arrows illustrate the time of colonization. Re-isolates were collected at different time points. Re-isolates obtained from different inoculations of the same patient are represented on opposite sides of an arrow.
5.3.2. Patients’ immune response upon colonization with strain 83972
The colonized patients, four males and two females, differed with respect to their host immune response towards bacterial colonization of the bladder. The mean of IL 8 expression in the strongest “responder” KA was more than 8-fold higher than in patient SR and 6-fold higher in patient SN (Fig. 21). While IL 8 expression was very diverse among colonized patients, their IL 6 expression did not differ drastically. In patient IJ, however, the IL 6 expression was more than 2-fold lower compared to that of the other patients. Noteworthy, in some patients only low IL 6 levels were detected in their urine, while their IL 8 expression was very high. The influx of PMNs into the bladder could be correlated with IL 8 expression, except for patient CK (Fig. 21; Fig. 22).
Fig. 21: Mean of host response parameters in urine samples collected from patients during the time of colonization with strain 83972: A). IL 8 expression; B) PMNs influx into the bladder; C) IL 6 expression. These data were kindly provided by Dr. B. Wullt, Lund.
To access the dynamics of host response upon asymptomatic bladder colonization by strain 83972, the levels of IL 8 and PMN influx in the bladder at each sampling time point were compared (Fig. 22). Only patients KA and POS were permanently colonized by 83972 during the duration of the colonization study, whereas in patients SR, CK, IJ and SN the bladder was cleared once from strain 83972 so that they had to be re-colonized. Interestingly, the deviation of the IL 8 expression from the mean value in the first group of patients was very low and less than 2-fold and 4-fold for KA and POS respectively. In contrast, a 10- and 16-fold increase in IL 8 expression was followed by rapid bacterial clearance in patients IJ and SN, respectively.
In patient CK, bacterial colonization of the bladder was lost already after a 3.5-fold IL 8 induction, however, the bacteria were lost about one month after that event.
Moreover, in patient SR a 3.8-fold induction of IL 8 (after 3 months) did not result in immediate bacterial clearance. However, bacteria were lost after eight months of bladder colonization and the host response at that time point was at the mean level.
Fig. 22: Levels of IL 8 and PMNs at each sampling time point. Black bars indicates time points corresponding to all investigated re-isolates, dotted double lines show bacteria clearance from the bladder. Single dotted line across the graph shows the mean of IL 8 expression. These data were kindly provided by Dr. B. Wullt, Lund.
Taken together, the individual host response described as PMN influx into the bladder, IL6 and IL8 production differed from patient to patient. At different time points of the colonization experiment, the levels of the response parameters were different. An increase in IL 8 expression generally correlated with an increase in PMN influx into the bladder. In a certain group of patients bacterial clearance was a direct follow up of increased host response.
5.3.3. Verification of the re-isolates Bacteria recovered from the patient’s urine samples were confirmed to be derivatives of strain 83972 by means of PCR. Two genetic markers specific for the E. coli strain 83972 were amplified: a DNA fragment covering a 4.7-kb deletion in the fim gene cluster (Fig. 23) and a fragment of a cryptic plasmid specific for strain 83972.
Fig. 23: Genetic organization of the fim loci in E. coli K-12 and E. coli 83972. P1 and P2 - primers used for PCR amplification.
The two designed primer pairs did not result in amplification of a PCR product with several prototypic non-pathogenic and pathogenic E. coli strains. Accordingly, these primer pairs were confirmed to be specific for strain 83972. In contrast, all tested patient re-isolates were positive for both E. coli 83972-specific genetic markers (Fig. 24) thus proving that they indeed represent derivatives of strain 83972.
Fig. 24: Verification of the patient re-isolates of E. coli strain 83972. A) PCR amplification of a 4.7-kb internal deletion region of the fim operon; B) PCR amplification of a DNA region of the cryptic plasmid found in strain 83972.
5.3.4. Genome structure of in vivo 83972 re-isolates Changes in the genome structure were assessed by PFGE following digestion with the endonucleases XbaI and AvrII, respectively (Fig. 25A and B). Depending on the enzyme used, different changes in the restriction pattern could be observed. Although most of the re-isolates exhibited the same DNA fingerprint as their parent strain 83972, significant changes in the genomic restriction pattern could be observed in five out of 16 re-isolates. Upon digestion with XbaI, the strains IJ15 and SR6 resulted in similar changes of their restriction pattern indicating that similar rearrangements occurred in these strains. Only strain CK12 showed differences in the restriction pattern relative to that of strain 83972 upon digestion with XbaI and AvrII. Interestingly, in none of the re-isolates from patient POS modifications of the restriction pattern could be observed, although the latest re-isolate, POS18, colonized the bladder for 536 days. In contrast, strain SR12 exhibited significant changes in the PFGE pattern already after 54 days of bladder colonization.
Moreover, the analysis of the genomic I-CeuI restriction fragments by PFGE demonstrated that strain CK12 is the only re-isolate with a reduced genome size compared to strain 83972 (Fig. 25D).
In order to roughly assess the complexity of the E. coli 83972 population colonizing the bladder, different colonies from the same urine samples were subjected to PFGE following digestion with AvrII (Fig. 25C). It turned out, that all three tested re-isolates represented at least a major fraction of the bacterial population in the bladder of their hosts. While all four independent colonies tested from the urine sample KA25 and CK12 exhibited a uniform restriction pattern, one colony from the SR12 sample had the same restriction pattern as the parent strain 83972, the other three colonies had an identical DNA fingerprint that differed from that of strain 83972.
Taken together, genomic alternations of in vivo re-isolates were accessed. Five out of 16 reisolates of strain 83972 showed changed genome structure as accessed by PFGE. Moreover, bacterial complexity in the bladder was low and in most cases analyzed re-isolate represented a major fraction of the bacterial population in the urine sample.
5.3.5. Phenotypes of different in vivo re-isolates Motility As the first test, re-isolates were stabbed on urine swarm agar plates to assess their motility (Fig. 26). The parent strain 83972 exhibited a very low motility, however, was not completely non-motile when compared to strain 536Δfli used as a negative control. A number of reisolates showed a similar low motility. However, the strains IJ15, SN16, SN25, CK6, SR6 and SR12 exhibited higher motility than the strain 83972 (Table 11) and also differed among each other. Moreover, strains POS6 and POS9 were less motile than strain 83972, comparable to strain 536Δfli. These data demonstrate that strain 83972 is capable of modulating the swarming ability in response to the growth environment.
Fig. 26: Motility of in vivo re-isolates of strain 83972 on urine soft agar plates incubated overnight at 37 °C.
Growth characteristics The growth characteristics of the re-isolates in vitro were assessed in pooled human urine as well as in LB medium and compared to those of strain 83972 (Fig. 27; Table 11). Generally, almost all re-isolates did not reach the same final bacterial number and had lower growth rates in the urine compared with the parent strain. Only growth of strains CK3, CK9 and CK12 was similar to that of E. coli 83972. The most significant decrease in growth was observed in case of isolates from patient POS. Interestingly, only the re-isolate SN16 grew better than strain 83972.