«(Über die Bedeutung der bakteriellen Genomplastizität für die Adaptation und Evolution asymptomatischer Bakteriurie (ABU) Escherichia coli ...»
Taken together, genotypic and phenotypic properties of the strain 83972 after long term in vitro continuous culture were assessed. Consequently, the restriction pattern determined by pulsed-field gel electrophoresis revealed that the genetic structure of all in vitro re-isolates was not altered and was identical to that of strain 83972. This is in contrast to the finding that the genome structure of in vivo E. coli 83972 re-isolates, even after less generations of growth in the human bladder, was affected in multiple re-isolates. The growth characteristics of the in vitro re-isolates tested in pooled human urine did not differ markedly from those of their parent strain 83972. Similarly, the comparison of biofilm formation by the in vitro re-isolates resulted in a very few significant differences when compared to that of parent strain 83972.
These observations are in striking contrast to the results seen with the in vivo re-isolates.
5.4. Transcriptome analysis of 83872 re-isolates Multiple pheno- and genotypic experiments uncovered striking differences among in vivo 83972 re-isolates. In contrast, in vitro re-isolates exhibited fewer alterations. To evaluate changes on the transcriptional level, microarray experiments were performed. The transcriptome of parent strain 83972 was compared with those of the in vivo re-isolates SR12, CK12, KA25 and one randomly chosen in vitro re-isolate 4.9 upon growth in pooled human urine. RNA was extracted from the mid-logarithmic growth phase followed by reverse transcription using fluorescent labelled nucleotides. The resulting cDNA was hybridized to an oligonucleotide DNA microarray (OPERON). For each re-isolate three independent competitive hybridisations were performed.
5.4.1. Significant changes in the expression pattern
The statistical analysis using t-test showed a significant reproducibility of the triplicate hybridisations (Table 13). All changes in the expression patterns of the re-isolates must be due to prolonged growth either in the human bladder or in the continuous in vitro culture and could be still detectable after in vitro growth in pooled human urine. Exception for strain SR12, the vast majority of significantly affected genes in the other re-isolates were downregulated when compared to the parent strain 83972 (Table 13). Interestingly, in all in vivo reisolates the number of significantly affected genes was on average four-fold higher than in the in vitro-grown strain.
De-regulated genes were allocated to different functional categories (Fig. 35). The distribution of down-regulated genes among the re-isolates was very diverse, however, strains CK12 and KA25 exhibited a certain similarity in the expression pattern of de-regulated genes. Strain SR12 differed not only in the total number of down-regulated but also in the high number of up-regulated genes, from which the vast majority could be grouped as coding for hypothetical
proteins, carbon and energy metabolism, and motility proteins. The high energy consumption due to flagella expression may be one reason for the increased expression of many genes involved in transport and energy production.
Apart from genes coding for hypothetical proteins and genes with unknown function, the second largest group of genes comprises those related to mobile elements like phages, transposons and plasmids (Fig. 35).
5.4.2. Individual adaptation of re-isolates Hierarchical cluster analysis of all the genes affected as determined by microarray hybridisation demonstrated remarkable differences among the investigated re-isolates. The cluster analysis grouped together genes with the same expression pattern and resulted in six subclusters, of which the first two subclusters represent groups of differently up-regulated genes in the re-isolates. The last four subclusters pinpoint genes which are only downregulated in the individual strain (Fig. 36).
Interestingly, in each strain distinct genes were affected. This illustrates the very unique environmental conditions faced by the different bacteria. Although the re-isolates SR12, KA25 and CK12 originate from the human colonization study, not many commonly deregulated genes could be observed. However, a subgroup of genes from the subcluster 1 and 4 show similarity exclusively in between the in vivo re-isolates (Fig. 36). In contrast, only a few genes which were grouped into subcluster 4 were de-regulated in the in vitro re-isolate 4.9.
Transcriptome changes of in vitro re-isolate 4.9 relative to parent strain 83972
Interestingly, the two component system TorSR involved in the regulation of the carbon metabolism and anaerobic respiration (Jourlin et al., 1997) was down-regulated in re-isolate
4.9. This regulatory system also controls the transcriptional regulation of the gadAX operon,
which was also down-regulated in strain 4.9 according to the array data. GadA, a glutamate decarboxylase, is part of the glutamate-dependent acid resistance system 2 (AR2) which confers resistance to extreme acid conditions. AR2 also protects the cell during anaerobic phosphate starvation when glutamate is available by preventing damage from weak acids produced from carbohydrate fermentation (Moreau, 2007). In addition to that, all genes of the regulon gcl-hyi-glxR-ybbVW-allB-ybbY-glxK involved in glyoxylate catabolism and the allantoin assimilation pathway turned out to be down-regulated. These reactions take place during anaerobic respiration.
Taken together, among hypothetical proteins and phage-related genes, most of the changes in the global gene regulation pattern of the in vitro re-isolate 4.9 are implicated in anaerobic respiration (Table 24 and Table 25).
Transcriptome changes of in vivo re-isolate SR12 relative to parent strain 83972 Genes with higher expression levels in the in vivo re-isolates were grouped into the subclusters 1 and 2. Most of them are exclusively up-regulated in strain SR12. As already shown in Fig. 35, the affected genes encode for the flagella apparatus and chemotaxis, or are involved in carbohydrate transport and metabolism, as well as energy production and conversion (Table 18).
Pentose and glucuronate interconversions as well as sialic acid, arabinose, mannose and xylose uptake and metabolism seem to be main pathways involved in extra energy delivery during growth in urine (Fig. 37). Most likely, up-regulation of these genes is directly connected to the nutrient availability during in vivo growth in the bladder. These sugars are taken up by a number of up-regulated uptake systems and degraded thus supplying the glycolysis with glyceroaldehyde-3-phosphate.
Fig. 37: Altered expression of sugar transport and degradation pathways in the in vivo re-isolate SR12 compared to strain 83972. Red arrows indicate up-regulated genes during in vitro growth in pooled human urine.
Moreover, a number of genes involved in the citrate cycle (TCA cycle) and glutamate metabolism were found to be affected in strain SR12 (Table 18 and Table 19). The two genes, glnG encoding the response regulator (NtrC, synonymous name GlnG) and glnL encoding the sensor kinase (NtrB, synonymous name GlnL) were down-regulated. NtrB functions as a membrane-associated protein kinase that phosphorylates NtrC in response to nitrogen- and carbon-limitation. This reflects adaptation to urine, which is a nitrogen-rich environment. This two component system regulates the expression of the glnALG and glnHPQ operons, which indeed were down-regulated according to the array results. In addition to that, purine degradation turned out to be down-regulated. Interestingly, D-serine uptake and deamination pathway to pyruvate were significantly up-regulated in re-isolate SR12 relative to parent strain 83972.
Transcriptome changes of in vivo re-isolates KA25 and CK12 relative to parent strain 83972 The hierarchical cluster analysis grouped most of the down-regulated genes together (subclusters 3-6; Fig. 36). Interestingly, major fractions of the genes with the same expression pattern were strain-specific. After functional classification, it turned out that most of the genes
encode for hypothetical, unclassified and unknown proteins. The second largest group of genes comprises those related to mobile elements like phages, transposons and plasmids (Table 21 and Table 23). The significance of this result might be questionable, since many of the de-regulated genes don’t represent complete transcriptional units. Nevertheless, each reisolate represents a unique gene expression pattern that is due to distinct environmental niches that bacteria have been growing in.
5.4.3. Common adaptive patterns in re-isolates
To identify general adaptation strategies to prolonged bacterial growth in urine, hierarchical cluster analysis was performed to group genes of at least two re-isolates that are commonly de-regulated. Altogether, 35 genes turned out to be similarly expressed in more than one reisolate relative to strain 83972 (Fig. 38).
Four clusters of genes were formed according to similar changes of their expression pattern among at least two re-isolates. The clusters one and two represent genes which are similarly expressed in re-isolates CK12 and KA25. The last two clusters comprise commonly regulated genes of re-isolates SR12 and KA25.
The first cluster comprises nine genes that are less expressed in re-isolates CK12 and KA25 relative to the parent strain 83972. Except for the gene glpC, the remaining eight genes code for hypothetical proteins with predicted function. The meaning of this result is unclear as these nine genes represent single genes of polycistronic operons. Subcluster 2 consists of eight genes whose expression was similarly affected in strains CK12 and SR12. As already shown by phenotypic tests (Fig. 26), the in vivo re-isolates differed in their ability to express flagella. Although strain CK12 was as motile as the parent strain 83972, microarray data indicated that the expression of two genes, flgB and tar, which are involved in motility and chemotaxis, was down-regulated. In contrast, in strain SR12 these genes were up-regulated.
This is consistent with the corresponding phenotype (section 5.3.5.).
Fig. 38: Hierarchical cluster analysis of commonly de-regulated genes in at least two re-isolates of strain 83972 relative to their parent strain. Numbers from 1 to 4 indicate the four main subclusters of commonly de-regulated genes. Clustered values are mean values of the expression ratio from at least three independent microarray experiments. Results with a p-value 0.05 are indicated in black colour.
Moreover, together with flgB and tar, almost the entire determinants coding for the flagellar apparatus were significantly up-regulated in re-isolate SR12 (Table 18). Several genes of subcluster 2 belong to gene clusters, e.g. yeiCNM as well as iucD, iutA and yfbA.
The last two subclusters include 18 commonly de-regulated genes of re-isolates SR12 and KA25. While 16 genes were up-regulated, expression of the remaining two genes was
repressed relative to the parent strain 83972. Interestingly, in the group of up-regulated genes the whole frmRAB gene cluster could be found. The FrmA protein is a glutathione-dependent alcohol dehydrogenase that is, together with FrmB, involved in the metabolism of endogenously formed formaldehyde and detoxification of exogenous formaldehyde (Gutheil et al., 1997). However, it was also shown that this alcohol dehydrogenase (also designated AdhC) is conserved from humans to bacteria and apart from the metabolic functions mentioned before is involved in the protection against nitrosative stress (Liu et al., 2001). In addition to that, the gene hmp encoding for flavohemoglobin was grouped right next to the frmRAB operon. Multiple studies describe the implication of Hmp in protection against NO released upon bacterial infection (Crawford and Goldberg, 1998; Poole et al., 1996).
Moreover, many genes encoding proteins involved in glycine, serine, threonine and methionine transport and metabolism were found within the same subcluster (Fig. 38). The MetR protein was already shown to bind and modulate the glyA-hmp intergenic region. This results in induction of flavohemoglobin expression encoded by hmp gene (MembrilloHernandez et al., 1998). Altogether, this supports the significance of the transcriptome data and functional relationship of genes as displayed in the cluster analysis.
In contrast to the previously described up-regulated genes, two genes coding for putative dehydrogenases were down-regulated in strains SR12 and KA25 relative to strain 83972 (subcluster 4; Fig. 38). Interestingly, both of them are located within a polyketide biosynthesis gene cluster (Nougayrede et al., 2006).
5.4.4. Verification of microarray results by quantitative RT-PCR
Quantitative RT-PCR was performed to verify changes of transcript levels of selected genes as determined by DNA array hybridization. In general, the trend of de-regulation of gene expression as determined by microarray results could be confirmed by quantitative RT-PCR.
Transcript levels of the genes metR and hmp were about two times higher in strain SR12 than in strain KA25, while in strains CK12 and 4.9 the differences were not significant (less than 2-fold) relative to strain 83972. The expression level obtained for the tar gene was 200-fold higher and 4-fold lower than in strains SR12 and CK12, respectively. A significant downregulation could be observed for expression of iutA and yeiC in strain CK12 (15- and 12fold), whereas in strain SR12 these genes were up-regulated 5- and 8-fold, respectively. The