We have evaluated the efficiency and insertion characteristics of the piggyBac transposable element in the malaria parasite P. berghei. Insertional mutagenesis approaches have been widely used for genome characterization and transposon-mediated mutagenesis has become a powerful molecular genetic tool for eukaryotic transgenesis [46–50]. It has recently been shown that genomic insertional mutagenesis using piggyBac combined with a phenotypic screen for attenuated growth of the blood stages provided an effective tool for functional analysis of Plasmodium genes . Assuming that appropriate phenotypic screens can be devised, the availability of additional forward genetic technologies holds the great promise for large scale analysis of the function of the many 'hypothetical' Plasmodium proteins. Therefore, in this study we adapted the piggyBac system used for P. falciparum to P. berghei and developed it further. P. berghei, a rodent malaria parasite, is a frequently used model for the functional analysis of Plasmodium genes [10, 35, 51] and it allows for the analysis of Plasmodium gene function both in vitro and in vivo throughout the complete life cycle.
To obtain piggyBac insertions in the P. falciparum genome one million parasites were added to erythrocytes preloaded with the transposon donor and the transposase helper plasmid followed by selection of drug resistant parasite populations . In these 'parent populations' the number of insertions was low ranging between 1-14 as identified by inverse PCR or vectorette PCR reactions. Using these methods 177 unique piggyBac insertions have been identified in 81 independent transfections . To obtain piggyBac insertions in the P. berghei genome we used in this study the standard method of transfection of purified schizonts that result in high transfection efficiency of 10-2-10-3 when parasites are transfected with plasmids [21, 23]. PCR-based detection methods, such as TAIL PCR, have been shown to be highly efficient  and the method of choice in other organisms to identify piggyBac insertions [53–55]. We used an adapted TAIL PCR method here as traditional PCR was inadequate for these purposes in our hands. Using this TAIL PCR method we were able to detect 35 and 40 inserts, respectively, in the two parent populations (P5, P2) that were obtained by drug selection of 1-5 × 106 transfected parasites. These calculations indicate that in our studies 16 to 18 times more inserts could be identified per transfection experiment in P. berghei than currently reported for P. falciparum . It can be expected that a percentage of inserts generated by this approach will not have been identified since the Semi Arbitrary Degenerate (SAD) primers will exercise some specificity and the P. berghei genome is extremely AT-rich (< 80%). Furthermore inserts could have been missed since the ability to detect inserts with PCR based methods is highly dependent on the copy number of the insert . The development and use of additional SAD primer sets for Tail PCR in combination with sequencing strategies of increased efficiency/sensitivity might therefore lead to an increased number of identifiable insertions in the parent populations. In the subpopulations and cloned lines that were obtained from the parent populations by infection of mice with 1-5 parasites we estimated that we were able to identify approximately 65% of the inserts by Tail PCR. This is based on the comparison of the number of visible inserts detected in FIGE-separated chromosomes and the inserts identified by TAIL PCR. In total we have identified 127 inserts by TAIL PCR in parasites of only 2 transfection experiments (P2, P5), indicating that piggyBac integrates efficiently into the genome of P. berghei and that this model permits the generation of piggyBac insertion events significantly more efficiently than the human parasite, P. falciparum.
Analysis of the insert sites in the P. berghei genome showed that insertion occurred exclusively in the expected TTAA insertion site. Like in P. falciparum we found neither a (strong) bias for insertion into a particular chromosome nor a preference for insertion into transcribed/expressed genes indicating a random distribution of inserts. This is in contrast to piggyBac insertion into the genome of several other organisms, including mouse, zebrafish, Schistosoma, Drosophila and mammalian cell lines, where insertions predominantly occur into actively transcribed genes [57–60]. Interestingly we observed a slight insertional bias with regard to the sequence directly flanking the TTAA sequence with a slight preference of T's up- and A's downstream of the insertion site, respectively, which has also been observed in P. falciparum . On the other hand, we found no preference for insertion within or outside CDS, whereas in P. falciparum an increased number of insertions have been observed in the 5'UTR regions, which might also reflect preferential insertion into transcriptionally active genes or subtle differences in genome organization. The fact that piggyBac insertion into the P. berghei genome is for the most part a random process is important for the further development and application of this technology for larger scale forward genetic approaches.
As with P. falciparum we found that in the absence of transposase the piggyBac inserts remained stably integrated at the insertion sites even during prolonged periods of asexual multiplication (84 mitotic divisions in a 3 week period). It has been shown that piggyBac inserts can remobilize in genomes when transposase is present and several studies have estimated the rate of transposon remobilization [43, 61, 45, 60, 62–64]. When we introduced transposase stably into the genome under control of the ama-1 promoter remobilization of piggyBac inserts was detectable during blood stage asexual multiplication of cloned parasite lines. The observed rate of remobilization seems to be low as the majority of parasites before and after the period of asexual multiplication showed the same insert as judged by analysis of FIGE-separated chromosomes. In the three clones of parasite population 5.i we detected a total of 7-10 unique inserts in parasites that had multiplied for a period of 3 weeks (from the start of the cloning procedure). If we assume as above that we detected 65% of the inserts by TAIL PCR the rate of remobilization in these populations is around 15% per mitotic division (7-10 inserts per 84 mitotic divisions).
Remobilization might actually offer benefits by increasing the number of unique inserts in an experimental population towards the desired saturation levels of mutagenesis . For P. falciparum it has been calculated that ~15.000 mutations/inserts will represent about 50% saturation and obtaining such a level of saturation is seen as a difficult but realistic possibility for the P. falciparum genome . Remobilization could help to significantly reduce the number of transfections necessary to produce true saturation mutagenesis in a population. For instance, while a 50% coverage library would require ~380 individual transfections using the transient transposase expression strategy (~50 detectable transpositions per transfection) the same level could be obtained with far fewer transfections in parasites containing a stably expressed transposase. Since we observed a 7-10× increase in inserts during 3 mechanical blood passages as a result of remobilisation, a comparable 50% coverage could be obtained by as few as ~50 transfections that are passaged for a period of 3 weeks in mice. The integrated transposase in our experiments is controlled by the P. berghei ama-1 promoter, which is active only briefly in the schizont stage. Remobilisation will be especially beneficial if remobilisation can be controlled by regulating transposase activity. Encouragingly the use of inducible expression systems has been shown to greatly improve control of piggyBac insertion and remobilization rates in other organisms .
PiggyBac insertion into CDS or 5'UTRs of genes may provide indirect evidence that the gene is essential for blood stage development. Therefore the data on the location of inserts from large scale piggyBac mutagenesis experiments can provide additional evidence for the dispensability of Plasmodium genes for blood stage development, information that will be of use for example for validation of drug and vaccine targets. We confirmed the non-essential nature of two of the genes interrupted by piggyBac by standard targeted gene deletion. Of these pb41 is an orthologue of pf41 that encodes a GPI anchored protein found on the merozoite surface and as such as been proposed as a vaccine candidate. Strategies of vaccination targeting non-essential proteins have been attempted in the human infectious parasite P. knowlesi in the past and resulted in variant parasites that escaped the vaccination regime. In some cases the escaped parasites failed to express the target antigen . Therefore, knowledge of the essential nature of a protein proposed as a vaccine candidate is potentially significant. The use of the model P. berghei to determine the essential nature of conserved genes is relevant due to the relative ease of genetic manipulation in this system and in the cited example we have subsequently learned that PF41 is non-essential in P. falciparum (B. Crabb, personal communication).
We have therefore deposited all genes with piggyBac inserts in the CDS or in the 5'UTR region (500 bp from the start ATG) in the publically accessible database of genetically modified mutant parasites, RMgmDB  and this information on piggyBac insertion is linked to the information on individual genes in PlasmoDB  and GeneDB . In addition, we demonstrate that piggyBac insertion can be used to identify promoters that are active during blood stage development. FACS sorting of GFP expressing blood stage parasites appears to be an efficient method to collect parasites that have GFP inserted downstream of an active gene promoter region (i.e. 5'UTR) initiating GFP expression. One such identified insert was located in the promoter region of a member of the bir multigene family [GeneDB :PBANKA_062360]. FACS analysis of blood stage GFP expression of this FACS-sorted population, demonstrated a pattern of GFP expression that is highly comparable to expression of BIR proteins tagged with either GFP or mCherry, specifically showing highest levels of expression in maturing trophozoites and schizonts (results not shown). The most significant application of random mutagenesis is the ability to perform forward genetic screens to select mutants of a desired phenotype. Recently the feasibility of such an approach has been shown for P. falciparum by screening for mutants with attenuated growth of the blood stages. This relied upon parasite cloning and phenotype characterisation soon after piggyBac integration and screening method resulted in the identification of several parasite genes and pathways critical for intra-erythrocytic development . Such an approach is applicable to both P. falciparum and P. berghei offering the possibility to develop and apply forward genetic screens for additional and important phenotypes such as virulence, drug resistance, commitment to and successful completion of sexual development. In addition, phenotypic screens might be developed during mosquito transmission and pre-erythrocytic development. However, bottlenecks in parasite numbers during transmission in and out of the mosquito may reduce the efficiency of selecting the desired mutants from a large pool of piggyBac-mutants. For P. berghei efficient methods exist for production of gametocytes and ookinetes and production of mosquitoes containing large numbers of oocysts (> 500) and salivary gland sporozoites can relatively easily be scaled up. However, application of efficient phenotype screens during liver stage development will require the development of more efficient in vitro cultivation systems for the analysis of sporozoites into viable blood stage merozoites.