|Home | About | Journals | Submit | Contact Us | Français|
Mutant hunts using comprehensive sequence-defined libraries make it possible to identify virtually all of the nonessential functions required for different bacterial processes. However, the success of such screening depends on the accuracy of mutant identification in the mutant library used. To provide a high-quality library for Pseudomonas aeruginosa PAO1, we created a sequence-verified collection of 9,437 transposon mutants that provides genome coverage and includes two mutants for most genes. Mutants were cherry-picked from a larger library, colony-purified, and resequenced both individually using Sanger sequencing and in a pool using Tn-seq. About 8% of the insertion assignments were corrected, and in the final library nearly 93% of the transposon locations were confirmed by at least one of the resequencing procedures. The extensive sequence verification and inclusion of more than one mutant for most genes should help minimize missed or erroneous genotype-phenotype assignments in studies using the new library.
Comprehensive sequence-defined mutant libraries have facilitated the genetic dissection of complex processes in several bacterial species (7). In principle, such libraries can be screened directly for mutants exhibiting a given phenotype to provide relatively complete identification of nonessential functions responsible for the trait. In practice, however, such screens can fail to achieve completeness and may even suggest incorrect genotype-phenotype associations. There are several potential contributing factors that arise from how mutant libraries are created and quality control tested.
Most defined mutant libraries have been generated by large-scale transposon mutagenesis and sequencing. Relatively complete genome coverage requires that an average of 5 to 10 unique insertions per gene be identified (7). Such large primary libraries serve as a source of mutants for smaller secondary libraries that retain genome coverage and facilitate phenotype screening (1–3, 6). Secondary libraries are usually made up of one or more colony-purified mutants per gene, with insertions situated toward the centers of coding regions to help ensure inactivation. A limitation of most secondary libraries is that mutant identities have not been verified by resequencing. In rare cases in which they have been checked, many assignments (typically on the order of 10%) have been found to be incorrect (3). An additional limitation of secondary libraries consisting of only one mutant per gene is that screens using them are at risk of missing genotype-phenotype associations due to mutant cross-contamination and insertion alleles that fail to inactivate. Libraries with more than one mutant per gene provide more than one chance to identify an association.
This report describes a Pseudomonas aeruginosa PAO1 secondary transposon mutant library with two mutants for most genes in which mutant identities have been confirmed by multiple resequencings. The redundancy and verification of mutant identities should make the library particularly useful for genome-scale genetic studies of the bacterium.
Strains are mutants of P. aeruginosa MPAO1 carrying ISlacZ/hah or ISphoA/hah transposon insertions and are derived from a defined mutant library described previously (5). The media used were as follows: Luria-Bertani (LB) agar (tryptone, 10 g liter−1; yeast extract, 5 g liter−1; sodium chloride, 8 g liter−1; Bacto agar, 15 g liter−1), freeze medium (LB medium containing 5% [vol/vol] dimethyl sulfoxide), low-twitch agar (tryptone (20 g liter−1), yeast extract (10 g liter−1), sodium chloride (8 g liter−1), and Bacto agar 30 g liter−1). The media were sometimes supplemented with tetracycline at 5 μg ml−1.
The strains making up the two-allele library were chosen from the primary library (5) to provide, where possible, two different insertion alleles per gene. Mutants were prioritized for inclusion using a scoring method that favored strains with insertions in the centers of genes (between 0.4 and 0.6 of the predicted reading frame), for which the overall sequence quality was high and for which the precise transposon-genome junction was unambiguous. Insertions in which the lacZ or phoA reporter sequences carried on the transposons were in-frame with target genes were also given preference over out-of-frame insertions. Finally, although unique insertion alleles were favored for each gene, 141 duplicates were included for which sequence quality was high. About half of these (66/141) were derived from separate mutageneses and are thus independent.
The strains were rearrayed from the primary library in 96-well format (150 μl of LB medium per well) using a Qpix2 robot (Genetix). After limited growth (2 h, 37°C), the rearrayed strains were serially diluted using a manual replicator, followed by spotting of ~5-μl aliquots on low-twitch agar, a medium that limits twitching motility and thus the cross-contamination of colonies. After overnight incubation, isolated single colonies were picked into deep well blocks (96-well format) carrying freeze medium (1.2 ml) and grown overnight. (Spottings that yielded confluent rather than isolated single colonies were streaked onto LB agar to generate colonies, which were grown separately and then added to the deep-well blocks containing the other strains.) The deep-well cultures served as the source of mutants for copies of the two-allele library, one of which has served as the primary source of individual strains sent to requestors.
Strain-by-strain resequencing of all mutants making up the two-allele library was carried out after rearraying and colony purification. The analysis used Sanger sequencing of amplified transposon-genome junction regions using protocols described previously (5). Mutants were resequenced one to five times.
Tn-seq analysis of a pool of the mutants making up the two-allele library was carried out according to the procedures described by Gallagher et al. (4). Aliquots (~0.5 μl) of each mutant from a 96-well storage plate were spotted onto LB agar, grown overnight, and harvested after flooding the plate with LB medium. Pooled bacteria from different plates were themselves then pooled, and genomic DNA was prepared. Seven Tn-seq assays were performed on genomic DNA from the final pool: two directly and five after additional growth. Insertion locations were considered confirmed if they were detected in at least two different Tn-seq runs. Tn-seq reads that could not be mapped to a unique chromosomal location because they corresponded to repetitive genomic sequences were not included in the analysis.
Primary mutant library insertion assignments had been classified as “exact” if the sequence quality of the transposon-chromosome junction region was high in the corresponding Sanger sequence read and as “estimated” if the sequence quality precisely at the junction was low, but high enough nearby to allow an estimate of the junction location (5). In the verification tests, an insertion location was considered confirmed by Sanger sequencing if resequencing identified an insertion site within 500 bp of the initial site with the transposon in the same orientation, regardless of whether the initial site was exact or estimated (see Table S1 in the supplemental material). A location was considered confirmed by Tn-seq if two or more Tn-seq runs identified a junction corresponding exactly to an exact Sanger sequence location (93.2% [7,993/8,572] of Tn-seq confirmations) or situated within 100 bp of an estimated Sanger sequence location (6.8% [579/8,572] of Tn-seq confirmations) (see Table S1 in the supplemental material). Primary library insertion assignments were changed if high-quality sequence from Sanger resequencings identified a new location and failed to verify the original location (770 strains, 8.2%).
We describe here the construction and characterization of a sequence-defined transposon mutant library of P. aeruginosa PAO1 designed to provide redundant genome coverage with a minimum of strains. This library is derived from a much larger primary library (5) and includes insertions in nearly 84% of the predicted genes of the bacterium. The library differs from most secondary libraries in two regards (Fig. 1). First, there are two insertion alleles (rather than one) for the majority of genes to provide redundancy. Second, most of the insertion locations have been confirmed by resequencing to correct errors in original mapping assignments and in the generation of the secondary library.
Strains making up the two-allele library were chosen where possible to carry unique transposon insertions situated near the centers of genes (which should tend to be inactivating), for which the quality of the original sequence data was high (Fig. 2). In addition, insertions were favored that led to in-frame fusions of chromosomal genes with the ′lacZ and ′phoA reporter genes carried on the transposons (ISlacZ/hah and ISphoA/hah) used to generate the library (5).
Two different resequencing procedures were used to check the identities of mutants making up the two-allele library. First, all mutants of the library were individually resequenced by Sanger sequencing of amplified transposon-genome junctions using the methods originally used to define the primary library (5). Individual mutants were subjected to at least one and as many as five independent Sanger sequencing runs. Second, the strains of the library were pooled, and the pool was subjected to seven independent runs of Tn-seq analysis using a procedure we developed (4) (Table 1). The total number of positions identified by Tn-seq (n = 9,295) agreed well with the expected number of unique insertions (n = 9,031) (see below). Although the Tn-seq analysis is less definitive than individual mutant resequencing, it provides an efficient way to confirm that a predicted mutation is represented in the library.
Approximately 86% (8,096/9,437) of the original, primary library insertion assignments were confirmed by one or both of the two sequencing procedures. About 8% (n = 770) of the insertions were assigned new locations, most of which (664/770) were themselves confirmed. For the final two-allele library, nearly 93% of the assignments were confirmed by at least one of the sequencing methods, and 69% were confirmed by both (Table 2). This level of verification is on a par with the highest yet achieved for such a transposon mutant library (3). The identity and confirmation results for all 9,437 mutants of the two-allele library are summarized in Table S1 in the supplemental material.
The two-allele library was found to include 406 strains (4.3%) with insertions that duplicated those in other strains. Although ca. 35% (141/406) of the duplicates had been included on purpose (Materials and Methods), the remainder were apparently introduced inadvertently during generation of the library. The two-allele library thus includes 9,031 (i.e., 9,437 − 406) unique mutants.
The new library provides redundant genome coverage to help overcome four limitations associated with screens using libraries with one mutant per gene. First, a phenotype associated with loss of a gene's function may be missed if the single mutant representing it is cross-contaminated with a second mutant, such as commonly arises during the generation of copies of mutant libraries. Second, a genotype-phenotype association will be missed if the mutation is silent, e.g., due to insertion in only one copy of a tandem-duplicated gene. Third, genotype-phenotype assignments may be incorrect if an insertion site has been misassigned, e.g., due to low quality of the sequence used to identify the site. Fourth, incorrect assignments can also be made if a strain carries an unidentified mutation in addition to the assigned mutation. The consequences of these limitations should be minimized in screens using libraries with two mutants per gene since even if one mutant is compromised, a second will exhibit the phenotype and identify the corresponding genotype-phenotype association. Screening libraries with more than one mutant per gene can also provide immediate confirmation that a genotype-phenotype association is valid.
The redundant coverage and the high level of sequence verification of the two-allele library should make it a reliable resource for identifying the nonessential genes responsible for virtually any process for which a suitable phenotypic screen can be devised. Mutants making up the P. aeruginosa PAO1 two-allele library are available individually, as complete library copies, or in a pool. Instructions for requesting mutants and additional information about individual mutants are available (http://www.gs.washington.edu/labs/manoil/libraryindex.htm).
We thank Cheri Turner for assisting with assembly of the two allele mutant library and Pradeep Singh for helpful discussions.
This study was funded by grants from the Cystic Fibrosis Foundation and the National Institutes of Health (P30 DK089507).
Published ahead of print 14 September 2012
Supplemental material for this article may be found at http://jb.asm.org/.