In this study we have successfully designed and evaluated a tiling microarray which targets the whole chromosome of the facultative pathogen P. aeruginosa PAO1. Given PAO1's large genome size and a limited construction capacity, one microarray depicted the genome with 25-base-long oligonucleotides tiled with an average 29 base pair spacing which corresponded to an average gap between the oligonucleotides of 4 base pairs.
The alignment of all PAO1 derived 25 base pair sequences present on the P. aeruginosa
microarray with the published genome sequence of strain PA14 identified regions specific to PAO1 that largely correspond the results of previous studies based on ORF-by-ORF alignment [16
]. Experimental data from comparative hybridization of PA14 and PAO1 DNA on the PATA1 chip reproduced this pattern at the chromosomal scale (Figure ), indicating that genetic differences affecting multiple consecutive oligonucleotides on the array can be detected at very high sensitivity and specificity, even when data were obtained from only one comparative hybridization.
The comparative hybridization of a reference PAO1 obtained from the Washington Genome Centre and the PA14 strain, furthermore lead to the identification of a 1 kb deletion in the Washington PAO1 strain, as compared to the published PAO1 sequence. This deletion affected part of the PA4684 and almost the whole PA4685 gene and explains why in the Washington transposon mutant collection no insertions within the PA4685 gene have been found.
Whereas genetic differences affecting multiple consecutive oligonucleotides on the array can easily be detected, we were also interested in the performance of the P. aeruginosa
genome array in the identification of single-base pair changes. A similar microarray has previously been developed and successfully applied for the identification of even single-point mutations leading to metronidazole resistance in H. pylori
]. To evaluate the performance of the PATA1 array we compared an in silico
alignment with experimental comparative hybridization data in great detail and identified three factors influencing the sensitivity in the detection of single nucleotide mismatches: i) positional bias – mismatches located at the 4 marginal positions from either side of the oligonucleotide were detected with a much lower sensitivity than mismatches at central locations, leaving a core region of 17 nucleotides; ii) interaction sites bias – sensitivity was decreased for mismatches where a C or G (3 possible H bonds) on the PAO1 probe sequence was corresponding to an A or T (2 possible H bonds) in the PA14 sample sequence and vice versa; and iii) neighbor bias – sensitivity was increased if the nucleotides neighboring the mismatch showed weaker binding, i.e. fewer H bonds (A or T), and vice versa.
Because the sensitivity towards point mutations was significantly reduced at the outer 4 positions at either end of an oligonucleotide, the core regions of all oligonucleotides covered only ~58% of the PAO1 chromosome effectively. This limitation leaves space for a next microarray generation which should depict the whole genome with 25-base-long oligonucleotides tiled with a 16 base pair or less spacing (corresponding to more than 400.000 oligonucleotides). However, despite this low effective coverage, about 50% of all single nucleotide polymorphisms (SNPs) theoretically covered by the PATA1 array could be detected in a comparative hybridization of PAO1 and PA14 genomes using a threshold hybridization ratio of Dth = -0.5. Variations larger than SNPs were detected with a much higher sensitivity (up to 85%) leading to an overall detection of about 60% of all theoretical variations with high specificity.
Our results clearly indicate that the microarray hybridization presented in this study represents a very robust method to screen whole sets of P. aeruginosa
strains bearing unknown genetic variations. One attractive future application of these microarrays could be to identify and depict the P. aeruginosa
genome organization of clinical strains. Most of the CF patients acquire P. aeruginosa
from the environment early in life and suffer from transient airway infections with diverse strains of P
, whereas at a later stage the patients become permanently colonized with one or few P. aeruginosa
]. Although diverse environmental P. aeruginosa
isolates cause chronic infections in the CF lung, it has recently been reported that there are dominant clones in the environment and disease and that individual clones prefer a specific repertoire of accessory segments [21
]. In contrast to the core genome, which is mostly shared by all P. aeruginosa
strains, the accessory genome is highly strain specific [18
The microarray could be used as a highly sophisticated fingerprinting method to significantly advance the question of whether there are specific genome organizations or certain genetic elements in clinical strains that are more frequently associated with chronic disease and with adverse clinical outcome. These traits may serve as important prognostic markers and may be targets of future drugs designed specifically for action against chronic infections. The recent development of next generation microarrays will offer the opportunity to include strain specific markers (pathogenicity islands) of common P. aeruginosa clones.
Furthermore arrays that cover the whole P. aeruginosa
genome (with tighter tiling) might serve as an alternative cost-effective method and a clear alternative to whole genome sequencing strategies for the identification of genetic variations. The availability of an easy to perform mutation discovery method in P. aeruginosa
will make a very important contribution and will significantly advance the field of patho-adaptive P. aeruginosa
evolution during the chronic infection process. P. aeruginosa
undergoes an intense genetic adaptation processes during the establishment of chronic pulmonary infections in the CF lung in which (mainly single-base pair change) mutations leading to the loss of function of multiple genes are positively selected [10
]. This adaptive behavior seems to be crucial in the development of chronic persistent disease, were P. aeruginosa
resides in a protected niche within biofilms and hides from the host immune responses. A detailed knowledge on general patho-adaptive mutations will lead to the identification of infection relevant bacterial traits which might be very interesting targets for the development of alternative treatment strategies effective against chronic persistent diseases.
Moreover, since the P. aeruginosa
microarray presented in this study depicts the whole PAO1 genome including the intergenic regions, the array could also serve as a valuable tool to identify binding sites of transcriptional regulators via the "ChIP-on-chip" technique. ChIP-on-chip (Ch
recipitation), is a genome-wide location analysis and a technique for isolation and identification of the DNA sequences occupied by specific DNA binding proteins in cells [22
]. More than 9% of the open reading frames in P. aeruginosa
PAO1 encode for (putative) transcriptional regulators or two-component systems which facilitate efficient adaptation to varied habitats [17
]. The identification of the transcriptional regulons involved in the regulation of functions required for bacterial persistence could significantly advance knowledge on specific P. aeruginosa
adaptation to the environment of the CF lung.