|Home | About | Journals | Submit | Contact Us | Français|
The TTAA-specific transposon piggyBac (PB), originally isolated from the cabbage looper moth, Trichoplusia ni, has been utilized as an insertional mutagenesis tool in various eukaryotic organisms. Here, we show that PB transposes in the fission yeast Schizosaccharomyces pombe and leaves almost no footprints. We developed a PB-based mutagenesis system for S. pombe by constructing a strain with a selectable transposon excision marker and an integrated transposase gene. PB transposition in this strain has low chromosomal distribution bias as shown by deep sequencing-based insertion site mapping. Using this system, we obtained loss-of-function alleles of klp5 and klp6, and a gain-of-function allele of dam1 from a screen for mutants resistant to the microtubule-destabilizing drug thiabendazole. From another screen for cdc25-22 suppressors, we obtained multiple alleles of wee1 as expected. The success of these two screens demonstrated the usefulness of this PB-mediated mutagenesis tool for fission yeast.
Studies using the model organism Schizosaccharomyces pombe have contributed significantly to our understanding of various cellular processes, including cell cycle regulation, genome stability maintenance and cell morphogenesis (1,2). The simplicity of its genome and the ease of genetic manipulations make S. pombe one of the most powerful genetic systems (3).
Chemical mutagens such as ethylmethane sulfonate (EMS) and nitrosoguanidine have often been used for mutagenesis in S. pombe (4). To map the phenotype-causing mutation, plasmid complementation is a favored approach, but it may require significant efforts if the phenotype needs to be scored clone-by-clone, and it is not suitable for dominant mutations or for certain unstable phenotypes such as epigenetic defects (5). Positional cloning can overcome some of these limitations but is time-consuming despite the availability of sophisticated mapping strains (6). Recently, the next-generation sequencing technique has been applied to mapping chemical-induced point mutations in fission yeast (5). Because more than one mutation is usually introduced during chemical mutagenesis, extensive linkage analysis is still required.
In contrast to chemical mutagenesis, insertional mutagenesis allows mutations to be rapidly identified. Illegitimate recombination-based insertional mutagenesis using a PCR fragment containing a selection marker has been developed for S. pombe (7,8). However, multiple and concatemeric integrations of PCR fragments, which are refractory to inverse PCR, may prevent the identification of the sites of integration (9).
Several efforts have been undertaken to develop transposon-mediated mutagenesis tools for S. pombe. The endogenous Tf1 retrotransposon was tested but it showed a strong bias against ORFs (10,11). The only transposon so far successfully applied in S. pombe is the Hermes transposon from Musca domestica (12,13).
Transposons do not insert randomly in the genome, which means there are ‘cold spots’ during mutagenesis if only one transposon is employed (14). A combination of different transposons might provide a solution to reduce insertion bias. For example, a mutagenesis system composed of P element and piggyBac in Drosophila melanogaster was shown to be superior to using a single transposon (15). Therefore, it is worthwhile to develop and utilize multiple transposon-mediated mutagenesis systems in one organism.
piggyBac (PB), originally isolated from the cabbage looper moth, Trichoplusia ni, has been reported to have mobility in a diverse range of organisms, including the budding yeast Saccharomyces cerevisiae, the fruit fly D. melanogaster, the zebra fish Danio rerio and the mouse Mus musculus (16–20). Even though PB prefers to use TTAA as its target sequence, it has little bias against ORFs, as demonstrated by the work in D. melanogaster (15). The broad host range and the low insertion bias make PB an attractive mutagenesis tool. Importantly, when PB is excised from the genome, the double-stranded break left behind is precisely repaired most of the time, thus rendering it easy to verify whether a transposon insertion has caused a phenotype by conducting a reversion analysis (21–23).
In this study, we developed an inducible PB transposition system for S. pombe. With this system, we performed two proof-of-principle genetic screens and obtained the expected mutants.
pPB[ura4] was constructed by replacing the ApaI–SacII fragment of PB[SV40-neo] with an ApaI–SacII fragment containing the ura4+ marker from a pBluescript-based plasmid. PB[SV40-neo] was described previously (20).
pREP1-PBase was constructed by cloning a NcoI restriction fragment of CMV-PBase (20), which contains the coding sequence of the piggyBac transposase, into the polylinker of pREP1 vector, downstream of the nmt1 promoter.
Integration plasmid pDUAL-PBase was constructed as follows: a plasmid from the fission yeast ORFeome library (24), pDUAL-Crb2-YFH1c, was digested by BglII and SphI to remove the nmt1 promoter and the DNA sequences encoding Crb2 and the YFH tag, and then ligated with a restriction fragment containing the nmt1 promoter and the PBase ORF from pREP1-PBase.
All the strains used in this study were constructed with standard method or isolated upon PB transposition (listed in Table 1). DY166 was isolated as a stable Ura+ derivative upon PB transposition from pPB[ura4]. Strains with PB insertion at the arg6 and ade6 loci were derived from DY166 by screening for auxotrophic mutants caused by PB transposition upon PBase induction.
Media were prepared according to the standard procedure (4). Thiabendazole (TBZ) was dissolved in DMSO as a stock solution at 20mg/ml and then added to YES medium at the indicated concentrations. 5-Fluoro-orotic acid (FOA) was used at a concentration of 1g/l in YE plates consisting of 0.5% yeast extract, 3% glucose and 2% agar. Vitamin B1 (thiamine) was used in Edinburgh minimal medium (EMM) at a final concentration of 15μM to repress the nmt1 promoter.
In the experiments using PB[ura4+] on an episomal plasmid, Ura− cells harboring a PBase plasmid were first grown in a medium without thiamine to induce PBase expression. Then the plasmid pPB[ura4] was introduced into the cells by transformation with lithium acetate method and cells were spread on plates lacking uracil.
In the experiments using PB[ura4+] integrated at ade6 or arg6 locus, strains containing both integrated PB and nmt1 promoter-driven PBase were first cultured in liquid EMM medium with proper supplements and thiamine. Then the cells were washed and transferred to EMM without thiamine to allow the expression of PBase. At indicated time points, cells were spread on EMM plates with thiamine but lacking adenine or arginine to monitor the numbers of cells that became Ade+ or Arg+, and on YES plates to monitor the total cell numbers. FOA-resistant derivatives were isolated by transferring cells pre-grown in EMM media without thiamine onto FOA plates.
Genomic DNA was prepared with MasterPure yeast DNA purification kit (Epicentre Technologies) and digested with HaeIII overnight. After heat inactivation of HaeIII, digested genomic DNA was ligated with T4 DNA ligase at 16°C for 6h. PCR was performed with two pairs of primers: (i) oligos 12 and 13; (ii) oligos 14 and 15 (Table 2). PCR products were sequenced using oligos 12 or 15 as primers, respectively. All the oligonucleotides used in this study are listed in Table 2.
Purified genomic DNA was analyzed on an ABI Fast 7500 System using the TaKaRa SYBR Green PCR kit according to the manufacturer’s instructions. PCR primers were LD252 and LD253 for act1+, oligo 69 and oligo 70 for PB[ura4+]. PB copy number was determined by ΔΔCT method using genomic DNA of strain DY1432 as a control, which has a single copy of PB (25). Since the ratios of PB versus act1+ were usually not integral numbers, we estimated the copy number to be the integral closest to the ratios.
In the screen for TBZ-resistant mutants, DY1434 cells were mutagenized by culturing in the absence of thiamine in EMM medium supplemented with histidine and arginine for ~40h and then spread on EMM plates containing thiamine and histidine but not arginine to select for Arg+ Ura+ cells (~1 OD600 unit cells for each 9-cm plate). After 20h at 30°C, EMM plates were replica-plated to YES plates containing 40mg/l TBZ. TBZ plates were then incubated at 30°C until resistant mutants formed colonies.
In the screen for suppressors of cdc25-22 at the restrictive temperature, transposition-mediated mutagenesis was performed with strain DY1007 grown at 25°C. After the PBase induction period in liquid medium, cells were spread on EMM plates containing thiamine but not arginine and the plates were incubated at 37°C until visible colonies appeared.
About 400000 Arg+ Ura+ colonies derived from DY1432, DY1433 and DY1434 upon transposition induction for 2 days in liquid EMM medium without thiamine were harvested and pooled. Genomic DNA was extracted from 30 OD600 units of cells and fragmented to the size of 250–1000bp by sonication. The fragmented DNA in the size range of 250–750bp was purified from 2% agarose gel with GFX PCR DNA and Gel Band Purification Kit (GE Healthcare). A DNA adaptor composed of oligo 496 and 497 was ligated to the DNA fragments with NEBNext DNA Sample Master Mix Set 1 (New England Biolabs). To amplify the PB insertion site-flanking sequences, we performed 30 cycles of PCR with a PB-specific primer close to the terminal repeats (oligo 128) and a primer annealing to the adaptor (oligo 498). After purification using the GFX kit, a further round of 30-cycle PCR was performed with primers oligo 145 and Seqr to add sequences needed for Illumina sequencing. Forty-two cycle single-end sequencing was carried out using an Illumina Genome Analyzer II. The sequencing primer oligo 549 was designed such that the first three bases of PB-specific reads were GGG from PB terminal repeats followed by PB-flanking genomic DNA sequence.
About 7 million sequence reads starting with GGG were obtained. After trimming the first three bases from these reads, the remaining 39bp was mapped to S. pombe genome sequence (downloaded from Sanger Center genome database, 15 October 2010 version) with Bowtie version 0.12.7 (26). Only perfectly matched and uniquely mapped reads (~5.3 million) were kept for further analysis. For multiple reads having the same sequence, we conservatively considered them coming from a single insertion event, as read numbers can be strongly influenced by the timing of transposition events during the 2-day induction, the proliferation rates of the cells, and amplification bias during the PCR steps. Insertion site data analyses were carried out using Perl, Matlab and Microsoft Excel. The sequencing data have been deposited at Sequence Read Archive (accession number SRA027355).
To test whether PB has mobility in S. pombe, we constructed a binary transposition system (Figure 1A). In this binary system, the transposon was provided by the donor plasmid pPB[ura4], in which a ura4+ marker was flanked by inverted terminal repeats (ITRs) of PB. The transposase PBase under the control of the thiamine-repressible nmt1 promoter was expressed from a plasmid pDUAL-PBase integrated into chromosome 2 at the leu1 locus. Strain DY1010 containing the integrated PBase plasmid was cultured in liquid medium without thiamine for ~36h to induce PBase expression and then was transformed with the donor plasmid pPB[ura4]. Since there is no S. pombe autonomous replication sequences (replication origins) in pPB[ura4] and the host strain is uracil auxotrophic, only cells which have integrated the ura4-containing sequence into their genome can proliferate robustly on media without uracil. Circular plasmids are not good substrates for recombination and thus rarely get integrated into the genome. As expected, only low levels of Ura+ transformant colonies were observed with a control strain that had no PBase, probably due to the infrequent integration of the whole plasmid (Figure 1B). In contrast, the presence of PBase dramatically increased the transformation frequency (Figure 1B). A similar result was obtained when the PBase was expressed from the episomal plasmid pREP1-PBase (data not shown). If PBase-dependent transformation events happened through transposition, the PB-flanking sequences in the Ura+ transformants should be different from the PB-flanking sequences in pPB[ura4]. To examine the PB-flanking sequences, we randomly picked 10 stable Ura+ derivatives and performed inverse PCR using primers annealing to PB sequences close to the ITRs. For nine out of the ten transformants, a single distinct PCR product was obtained; and the other transformant gave rise to two distinct bands (data not shown). The sizes of these PCR products were different from each other, suggesting that PB-flanking sequences are likely to be unique sequences of fission yeast genome. Sequencing of the PCR products confirmed that PB indeed integrated into either chromosome 1 or 2 at 11 distinct sites (Supplementary Table S1). In each of these 11 mapped insertion events, the PB terminal sequences were next to sequences from the fission yeast genome rather than the plasmid backbone. The four terminal nucleotides were duplicated upon insertion and had a bias strongly favoring TTAA (10/11 TTAA; 1/10 TTAT). Together, these data indicated that the PBase-dependent ura4+ integration events stemmed from transposition from the plasmid to the chromosomal DNA. All of the 11 insertion sites are located in intergenic regions, suggesting a possible bias against ORFs for PB transposition from a plasmid donor to chromosomes.
Typically, excision of PB from a donor site leaves no footprint and this has been tested true for all host systems examined thus far (21,22). In mammalian cells, it has been observed that PB excised from a donor site sometimes failed to re-insert into the genome (23,27). We took advantage of this phenomenon to collect PB excision events to test whether PB transposition leaves footprints in fission yeast. Because the ura4+ marker can be counter-selected by FOA, cells that lose the PB[ura4+] transposon after excision can be selected as colonies growing on FOA plates. DY2533 had PB[ura4+] inserted 40 base pairs (bp) upstream of the ade6 ORF (see ‘Materials and Methods’ section for strain construction) and this insertion prevented cells from growing on plates without adenine and rendered the colonies red on low adenine medium such as YE (Ade− phenotype) (Figure 2A). Upon the induction of PBase, cells were spread on YE+FOA plates and allowed to grow into colonies. We analyzed 49 FOA-resistant derivatives and found that 95.9% (47/49) were white (Ade+) and 4.1% (2/49) were dark red (Ade−). Therefore, vast majority of PB excision events were accompanied by restoration of a functional ade6+ locus, consistent with a clean excision from the insertion site. From the 47 white colonies, we randomly picked 22 and confirmed by PCR that they all lacked the PB insertion (Figure 2A). We detected no deviation in the size of the PCR products from the expected 534bp. This suggests that PB excision either left no footprint or ones that were too small to be discerned from the agarose gel. We sequenced the PCR products from four Ade+ derivatives and found no base alternation around the excision site compared to the reference genome sequence, indicating that PB excision occurred in a precise fashion and left no footprint. In both Ade− derivatives (the FOA-resistant dark red colonies described above), we detected a deletion of the ade6-proximal region of the transposon (Figure 2B). Because this deletion removed part of the ura4 marker, the cells were converted to Ura− without the loss of the entire PB[ura4+]. Interestingly, the deleted sequence was flanked by two copies of PB preferred target motif, TTAA. Thus it is likely that two sequential transposition events might have generated this type of footprint (Supplementary Figure S1).
The precise excision of the PB transposon allowed us to develop a selectable marker to quantitatively monitor the excision events. In strain DY1432, PB[ura4+] is located in the middle of the arg6 ORF and consequently the cells are auxotrophic for arginine and prototrophic for uracil (Arg− Ura+) (see ‘Materials and Methods’ section for strain construction). PB transposition from the arg6 locus to another site will restore the function of arg6+, thereby converting the cell from Arg− Ura+ to Arg+ Ura+ (Figure 3A). Thus, the percentage of Arg+ Ura+ cells induced by PBase could represent the frequency of transposition events. Time-course studies conducted with DY1432 and two other strains with the same genotype showed that, the percentage of Arg+ Ura+ cells dramatically increased after PBase expression and peaked at ~10% around 36h after the removal of thiamine (Figure 3B). This result suggests that PB integrated at a chromosomal site can undergo high frequency transposition, and PB inserted at an auxotrophic marker gene provides a useful means for detecting and enriching transposition events.
A common problem with transposon-based mutagenesis is multiple insertions in the genome, which complicate identification of the mutation causing the phenotype. One reason we designed the arg6 donor system that launches PB transposition from a chromosomal site is to avoid multiple insertions associated with launching the transposition from a multicopy episomal plasmid. However, even with a single copy chromosomal donor site, the copy number of a DNA transposon in individual cells can increase if transposition occurs in S or G2 phase of the cell cycle (28). In S or G2, a sister-chromatid can serve as the homologous recombination template, and DNA double-strand breaks formed upon PB transposition can be repaired by copying the PB from the sister, thus restoring the PB-containing sequence at the donor locus. This will lead to a net increase of the copy number of PB (Figure 3C). Besides inter-sister recombination, segregation of sister chromatids during mitosis can also lead to copy number change in the daughter cells (Supplementary Figure S2). In all these depicted schemes, the daughter cells with two copies of PB harbor one copy at the original donor site. We have indeed observed such type of copy number increase (J.-M. Zhang and L.-L. Du, unpublished observations). The use of a selectable excision marker like arg6::PB can in theory select against such events because only cells that lose the PB at the arg6 donor site can grow on media lacking arginine. To analyze the extent of PB copy number change, we determined the copy number of PB[ura4+] in the Arg+ Ura+ derivatives by quantitative PCR. After 36h of PBase induction in DY1432, cells were spread on plates without arginine and uracil, allowing only Arg+ Ura+ cells to form colonies. Among the Arg+ Ura+ colonies examined, 91.7% (22/24) had only one copy of transposon, while 8.3% (2/24) contained two copies. This result suggested that if we only use transposition-induced Arg+ Ura+ cells for screening, the copy number of PB will not increase significantly, thus facilitating the utilization of PB as a mutagenesis tool in S. pombe.
Ideally an effective mutagen should hit genes everywhere in the genome with similar frequency. To examine whether PB insertion events selected by the excision marker have a distribution bias, we applied Illumina sequencing technique to map the insertion sites of PB transposed from the arg6 locus. From ~7 million raw sequence reads, we identified 33 218 independent insertion events that are supported by uniquely and perfectly matched reads. We counted PB insertions at the same position in opposite orientations as two independent insertions.
Consistent with our inverse PCR and Sanger sequencing analysis of 11 PB integration events (Supplementary Table S1), among deep sequencing-mapped insertions, the vast majority (96.7%) used TTAA as the target sequence, and most of the remaining ones used TTAA-like variant sequences such as TTAT and TTAG (Figure 4A).
Previous studies in insects, mouse, and Plasmodium falciparum suggested that, besides the nearly obligatory requirement for TTAA sequence, PB also has a slight preference for Ts in the five bases upstream of TTAA and As in the five bases downstream of TTAA (20,29,30). We observed a similar pattern in our deep sequencing-mapped insertions (Supplementary Figure S4).
To visualize the distribution of PB insertion sites in the genome, we plotted the numbers of PB insertion sites in 10-kb intervals along the three chromosomes (Figure 4B). We only plotted the insertions targeting TTAA because a great majority of PB insertion sites mapped to TTAA sites, and because we wanted to compare the insertion site distribution with a reference dataset of all the TTAA sequences whose flanking sequences can be uniquely mapped to the genome sequence. PB insertion site distribution among the three chromosomes largely correlates with TTAA distribution, with a slight bias favoring chromosome 2 where PB was launched from (38.2% for PB insertion sites versus 36.8% for reference TTAAs). There is a noticeable but moderate enrichment of PB insertion sites in a 100-kb region surrounding the arg6 locus, suggesting a mild local hopping effect. By and large, PB insertion does not appear to strongly favor or avoid any particular regions of the fission yeast genome.
Seventy-nine percent of TTAA-targeting PB insertion sites are located in intergenic regions, compared to 55% for reference TTAA sites. The intergenic regions (IGRs) can be divided into three classes depending on the transcription directions of flanking ORFs: convergent IGRs, divergent IGRs and tandem IGRs. Among these three classes, we found that PB prefers tandem and divergent IGRs (Figure 4C).
Our insertion site mapping was done with haploid strains. Thus, the preference for intergenic regions can be at least partially explained by fitness loss resulted from PB insertion into ORFs in the haploid cells. Consistent with this idea, when we analyzed the dispensability information of ORFs hit by PB (31), we found a much stronger bias against essential genes (1% of PB insertions are in essential genes versus 12.4% of reference TTAAs), compared to the non-essential ones (19.3% of PB insertions versus 31.2% of reference TTAAs) (Figure 4C).
PB insertion into ORFs may not always generate null alleles, as suggested by our observation of several hundred insertion sites inside of essential ORFs. We plotted the relative positions of insertion sites in essential and non-essential genes and found that, PB insertion in essential ORFs appeared to strongly favor the regions close to the stop codon, suggesting that most of these insertions resulted in functional gene products with small C-terminal truncations (Supplementary Figure S4). A similar but weaker distribution bias was observed for non-essential genes as well, suggesting there were fitness losses associated with PB insertions in non-essential genes.
For mutagenesis purpose, PB insertions outside of ORFs can be useful, as insertion into transcriptional regulatory sequences may result in hypomorphic alleles. Such hypomorphic alleles are particularly valuable for essential genes whose null alleles are inviable. When we analyzed the PB insertion sites located in the 400-bp regions upstream of ORFs and 400-bp regions downstream of ORFs, we noticed that PB insertion sites were found less frequently in regions immediately upstream of essential genes, indicating that insertion into 5′ flanking region can result in fitness loss (Supplementary Figure S5).
To examine whether gene expression level influences PB distribution, PB insertions in non-essential genes were tabulated according to the mRNA levels of their target genes (32). We did not observe a close relationship between PB insertion frequency and gene transcription level (Supplementary Figure S6).
A necessary task in a transposon-based forward genetic screen is to verify whether the phenotype is caused by the transposon insertion or due to spontaneous mutations not related to the transposon. This can be addressed in our PB system by analyzing the linkage between the phenotype and the ura4 maker, but it is time-consuming and cannot exclude the possibility that a spontaneous mutation had occurred near the inserted PB. Thus, we examined the feasibility of performing reversion analysis based on the ‘no-footprints’ feature of PB. The rationale is as follows: if the insertion of PB is responsible for the phenotype, precise excision of PB will abolish the phenotype; if the phenotype is not caused by PB insertion, excision of PB will not alter the phenotype. To obtain colonies derived from cells that had experienced an excision event, FOA counter-selection was used again.
DY3038 and DY3039 were isolated as TBZ-resistant mutants in a pilot genetic screen and each had a single copy of PB as determined by qPCR. In DY3038, it was found that PB remained at the arg6 locus, i.e. no transposition had occurred. Thus, the TBZ resistance phenotype of DY3038 was likely due to a spontaneous mutation. As expected, the FOA-resistant derivatives of DY3038 were all as resistant to a high concentration of TBZ as the parent strain (Figure 5B). In contrast, DY3039 had PB inserted at a TTAA target site within the klp6 gene whose null mutation is known to cause TBZ resistance (33). Consistently, all eight independent FOA-resistant derivatives of DY3039 lost the TBZ-resistance phenotype (Figure 5B). This demonstrates that reversion analysis is able to distinguish whether the mutant phenotype is caused by PB insertion.
One potential concern with this verification method is that when PB inserts into a non-TTAA site, its subsequent excision might leave a footprint, which then might cause the same phenotype as the PB insertion. This would give a false negative result, making the phenotype appear to be not associated with PB insertion. To address this concern, we analyzed two TTAG-targeted TBZ-resistant mutants and found no footprint left by the excision of PB from the TTAG sites and not surprisingly, the phenotype was reverted to that of the wild-type upon excision (Figure 5C and D).
Our general PB-based screening strategy is as follows: first, a strain containing the arg6::PB selectable excision marker and the integrated transposase gene is constructed; then after the induction of transposition, Arg+ Ura+ cells are enriched and further screened under certain conditions (Figure 6A). To investigate the feasibility of forward genetic screens with PB in S. pombe, we undertook two proof-of-principle screens: (i) for mutants resistant to the microtubule depolymerization drug, thiabendazole (TBZ); and (ii) for mutants that suppress the temperature-sensitive growth defect of cdc25-22.
In the screen for mutants resistant to 40mg/l of TBZ, 3 OD600 units of mutagenized DY1434 cells were screened. The proportion of Arg+ Ura+ cells was ~6.3% as measured by plating assay. As 1 OD600 unit culture contains ~107 cells, we estimated that transposition occurred in ~2×106 cells. In the end, we obtained 73 TBZ-resistant mutants.
Fission yeast Klp5 and Klp6 form a heterocomplex and are involved in the microtubule disassembly (33,34). Mutants harboring a deletion of either klp5 or klp6 have been reported to be resistant to high concentrations of TBZ (33). Because initial inverse PCR analysis suggested that these two genes were frequent hits of our screen, we used colony PCR to check whether these two genes were intact in the resistant clones. From the 73 TBZ-resistant mutants, we found 49 with PB[ura4+] inserted in klp5, and 14 in klp6 (Figure 6B). The junctions between PB and the fission yeast genome were then amplified and sequenced for each mutant. All insertions were located in the coding regions. There are 20 and 11 TTAA sites in the coding regions of klp5+ and klp6+, respectively. Of these, 60% (12/20) and 45.5% (5/11) of the potential insertion sites were targeted in our screen hits (Figure 6B). Besides TTAA, two different TTAG sites were also targeted in klp5 ORF.
Apart from klp5 and klp6, we also had 10 other mutants that might identify additional genes required for TBZ sensitivity. Through reversion analysis using the FOA selection of excision events, we determined that in seven of these mutants, the TBZ-resistant phenotype was due to spontaneous mutations rather than PB insertion. Thus, we performed inverse PCR to identify the PB insertion sites in the remaining three mutants and found that they all had an insertion at the only TTAA site in the ORF of dam1 (Figure 6B). Fission yeast Dam1 is a subunit of the DASH complex required for proper chromosome segregation (35). Deletion of dam1 makes cells hypersensitive to TBZ, thus the PB insertion allele is a gain-of-function mutation. The translational product of our dam1::PB allele consists of the N-terminal 102 amino acids of Dam1 followed by 31 amino acids derived from the PB terminal sequence (Figure 6B). The dam1::PB mutant showed similar resistance to TBZ as a klp6 deletion strain (Figure 6C). It has previously been reported that dam1 mutants with altered C-terminus, either in a truncated form or with two amino acid substitutions, are TBZ resistant (36,37).
Fission yeast Cdc25 dephosphorylates and activates cyclin-dependent kinase Cdc2, which is phosphorylated by the Wee1 kinase. Activation of Cdc2 is required for entry into mitosis. At 37°C, a temperature-sensitive mutant of cdc25, cdc25-22, is arrested at the G2/M stage. A saturated screen of cdc25-22 suppressors by chemical mutagenesis showed that, besides very rare cdc2 point mutations, the vast majority of extragenic suppressors were loss-of-function wee1 alleles (38). We screened for suppressors of cdc25-22 at the restrictive temperature and identified four PB-induced mutants. Each of them had one PB insertion in wee1 ORF, covering 50% (3/6) of TTAA sites (Figure 6D).
In this study, we show that PB undergoes transposition in the presence of PBase in the fission yeast S. pombe and it can serve as an effective mutagenesis tool for genetic screens.
We developed a PB-based mutagenesis system for S. pombe. In this system, PB transposition is launched from an auxotrophic marker gene locus, thus allowing transposition events to be easily monitored and enriched. A major advantage of our system is that the copy number of the transposon stays at one in most of the mutagenized cells selected by the excision marker, thus avoiding the complication caused by multiple insertions. In addition, we took advantage of the precise nature of PB excision to implement an efficient reversion analysis procedure for verifying the causal relationship between the transposon insertion and the mutant phenotype.
A potential drawback of initiating transposition from a fixed chromosomal locus is the distribution bias caused by local hopping effect. Compared to other transposons used for mutagenesis, PB is known to have low tendency of local hopping (27). Our deep sequencing result showed that there was little chromosomal bias of PB insertions selected by the excision marker. The limited local hopping we observed is not likely to pose a problem for mutagenesis.
As shown for other organisms, TTAA is the preferred target site of PB in fission yeast while other TTAA-like 4-nt sequences could be targeted occasionally. There are ~110000 TTAAs across the fission yeast chromosomal genome and ~40% of them are found in annotated genes (39). More than 98% of the fission yeast genes contain TTAA in their coding sequences and are susceptible to PB insertion (Supplementary Figure S7). Thus, it is theoretically possible to utilize PB to perform exhaustive screens of the fission yeast genome. Experimentally, from a mutant pool of 400000 Arg+ Ura+ colonies, we detected PB insertions in 54% of the TTAA-containing ORFs using high-throughput sequencing (2647 out of 4914 genes). Insertions in the essential genes were severely under-represented, most likely due to insertion-caused lethality in haploid cells. For TTAA-containing non-essential ORFs, we found insertions in 66% of them (2306 out of 3468 genes). We believe this ratio is still an underestimate of the level of gene coverage that can be achieved by our system, because not all PB insertions isolated in our TBZ-resistant screen were detected by deep sequencing. For example, in klp5, only five of the 12 TTAAs identified as PB insertion sites in the TBZ screen were identified in sequencing-based profiling. The failure to detect all available insertion sites may be due to lower-than-saturation complexity of the mutant pool used for our sequencing analysis, and/or underrepresentation of certain sites caused by amplification bias of the PCR procedure.
We performed two proof-of-principle genetic screens to validate our system. In the screen for TBZ-resistant mutants, klp5, klp6 and dam1 mutants were found. Our literature search suggested that all the known mutants that can tolerate 40mg/l of TBZ affect one of these three genes (33,34,36). Thus, our screen has identified all the expected genes. The number of Arg+ Ura+ cells screened (2×106) was about 20 times of the number of TTAA sites in the fission yeast genome. However, the numbers of TBZ-resistant isolates of the three hit genes were not as high as 20 times of the numbers of TTAA sites in these genes. Instead, for klp5, klp6 and dam1, the numbers of isolates were 2.4, 1.3 and 3 times of the numbers of TTAA sites in their coding regions, respectively. This lower-than-expected recovery rate may be due to a PB insertion bias. We also observed uneven utilization of the TTAA sites within the same gene. For example, 20 insertion events were mapped to one TTAA site in klp5 whereas no insertion was mapped to eight other TTAA sites in klp5. Most likely, there are both ‘hot spots’ and ‘cold spots’ among the TTAA sites in the same gene. To cover as many genes as possible, especially for the genes with fewer TTAA sites, it would be advisable to screen at least 106 Arg+ Ura+ cells using our system.
Based on current annotation, ~5000 genes are encoded by the S. pombe genome (39). A genome-wide haploid deletion library is commercially available from the Bioneer Corporation (http://pombe.bioneer.co.kr/). This library covers ~80% of non-essential genes and each mutant has two unique barcodes that can be monitored by microarray or high-throughput sequencing technique (31,40). This reverse genetics tool makes it possible to quickly identify genes responsible for a desired phenotype. However, a deletion library-based screen is not without limitations. For example, it would be difficult to carry out a screen in a genetic background different from that of the library because such an undertaking involves crossing the desired background into every single deletion strain in the library. In comparison, it is much easier to introduce the transposon elements into a specific genetic background to obtain a strain from which a transposon-based screen is initiated. Besides, all the mutations in the haploid deletion library are null alleles of non-essential genes, whereas different types of alleles can be expected from insertional mutagenesis, including hypermorphic alleles like the dam1 mutants reported here as well as hypomorphic alleles of essential genes.
National Basic Research Program of China (973 Program; 2006CB806700); Chinese Ministry of Science and Technology 863 grant (2007AA02Z1A5). Funding for open access charges: Chinese Ministry of Science and Technology 863 grant.
Conflict of interest statement. None declared.
Supplementary Data are available at NAR Online.
We thank Dr Xiaohui Wu at Fudan University for kindly providing the PB[SV40-neo] and the CMV-PBase plasmids. We thank Dr Meng-Qiu Dong for critically reading the article.