Two different approaches were used to achieve insertion of piggyBac elements into the genome of P. berghei. In the first approach parasites were simultaneously 'co-transfected' with piggyBac donor plasmid (pL1302) and helper plasmid (pL1301). The helper plasmid contains the transposase under the control of the constitutive eef1a promoter but does not contain a drug-selection cassette (Figure 1A). Since such plasmids are not retained in parasites during asexual growth without drug-selection 25 the helper plasmid is 'transiently transfected' and will be lost from the parasites during blood stage growth. In the second approach the donor plasmid was transfected into transgenic parasites that contained 2 copies of the transposase stably integrated into the c-ssu-rrna gene locus (Figure 1C). This transgenic line, transposase ama-1 (abbreviated: TPS_ama1), was generated using standard methods for transfection of P. berghei and the transgenic parasites contain the T. gondii dhfr/ts as a selectable marker and transposase under the control of the schizont specific ama-1 promoter (Figure 1B). The construct was integrated into the c-ssu-rrna gene locus by single cross-over integration, resulting in the integration of 2 copies of the transposase gene.

The donor plasmid contains the 5'and 3' inverted terminal repeats of the piggyBac element (Figure 1A) which are the minimal cis elements necessary for piggyBac mobilization. Both inverted repeat sequences consist of a terminal 13 bp and internal 19 bp perfect inverted repeat that are separated by a 3 bp (5'ITR) or a 31 bp (3'ITR) spacer 2627282930. In the donor plasmid the two ITR sequences are located on both sides of a drug-selectable marker cassette and a gfp expression cassette that lacks a promoter region (Figure 1A). The target site for piggyBac insertion is TTAA and it moves by precise insertion and excision mechanisms 3132. Transfection of the donor plasmid would therefore result in insertion of both the drug-selectable marker cassette and the gfp-expression. Insertion of the drug selectable marker, the human dhfr gene, allows for selection of parasites containing the inserts using pyrimethamine or WR99210.

The two approaches described above were used to obtain piggyBac-mediated insertion of the gfp-expression cassette into the P. berghei genome. Co-transfection of wt parasites with both the donor and the helper plasmid, followed by selection with pyrimethamine resulted in selection of two resistant parasite populations (1055 and 1056; parent populations P1 and P2, respectively). Southern analysis of Field Inversion Gel Electrophoresis (FIGE) -separated chromosomes of these parasites showed integration of the donor plasmid (construct) in multiple chromosomes (Figure 2A). No integration was detected when only donor plasmid was transfected (exp. 1057, Figure 2A, lane C). Based on hybridization intensity it appears that a ratio of helper/donor plasmid of 1:2 results in higher insertion frequency then a 1:1 ratio (Figure 2A). To obtain a better insight into the insertion into the different chromosomes we generated 15 'subpopulations' (P1a-P1o) by intravenous injection of 1-5 parasites of parent population P1 in 15 different mice. Southern analysis of FIGE-separated chromosomes of parasites from the subpopulations showed insertion of the constructs in nearly all chromosomes (Figure 2A, right panel).

Similarly, transfection of 3 different amounts (P3 = 15 μg; P4 = 10 μg; P5 = 5 μg) of the donor plasmid into TPS_ama1 parasites that stably express transposase, followed by selection with WR992210, resulted in selection of three resistant parasite populations (1182cl1m1-m3; parent population P3, P4 and P5). Southern analysis of FIGE-separated chromosomes of these parasites again showed integration of the donor plasmid (construct) in multiple chromosomes (Figure 2A, left panel). Based on the relative intensity of the hybridization signals it appears that a lower concentration of the donor plasmid results in higher insertion frequency. Analysing the Southern hybridization data of the FIGE-separated chromosomes from the 10 subpopulations of P5 (generated as described above) insertion of the constructs was observed in nearly all (groups) of chromosomes (Figure 2B, right panel).

To identify the location of piggyBac insertions in the genome we initially used a standard method of inverse PCR using the forward primers 3202-3204 in combination with the reverse primers 3205-3207 as described for P. falciparum 18 to amplify re-ligated piggyBac sequences with P. berghei genomic flanking regions. However, this PCR method resulted in a very low yield of inserts when we used DNA extracted from parasites of the subpopulations. Compared to the number of insertions estimated based on the number positive chromosomes, we were able to retrieve less than 20% of the inserts from the different subpopulations (data not shown). We therefore decided to use an adapted method of TAIL-PCR. In this method a large pool of arbitrary degenerate primers designed for use in the AT rich genome of P. berghei (see Methods section) were used in combination with primers specific for both ITR's of the piggyBac element. By analyzing 9 subpopulations that had 16 visible inserts in the 14 P. berghei chromosomes, we were able to identify 11 inserts (~70%) from these chromosomes by TAIL-PCR. Therefore, for further identification of inserts we decided to exclusively use TAIL-PCR and no other methods such as inverse PCR or methods using restriction digestion and ligation 33. Using TAIL-PCR we identified insertions in parasites of parent population 2 and 5 and subpopulations of P1, P2 and P5. In total we identified 127 inserts at unique locations in the genome (Additional file 1 Table S1) and we have not identified any insert at the same location from two different parent populations, suggesting that there is no strong bias in preferred insertion sites between different experiments. However, for two genes: [GeneDB34:PBANKA_060790] (in P2 and P5.b.3) and [GeneDB34:PBANKA_082890] (in P2 and P5.i.3), insertions were identified both in 5'UTR/3'UTR and in CDS/3'UTR, respectively. In all 127 cases, that we identified by sequencing TAIL-PCR products, the insertion occurred via an expected canonical TTAA tetranucleotide (piggyBac) insertion site (Figure 2C, Additional file 1 Table S1).

Since the ability of piggyBac to randomly insert into the genome is an important feature of the piggyBac mutagenesis system in Plasmodium research, we investigated several aspects of the 127 insertions. First, we analysed the immediate 20 nucleotides adjacent to 5´ and 3´ flanking the TTAA site to evaluate if piggyBac insertion exhibited any additional preferences within the insertion flanking sequences. We compared these flanking regions with those of a set of 254 random chosen TTAA sequences from the genome (see Methods section). Overall, a slightly higher AT% was found in the piggyBac TTAA site flanking regions (81.7%) compared with the 254 random TTAA sites (80.4%), however this difference was not statistically significant (two tailed test, P = 0.13). A slight preference for a stretch of seven T's and three A's was observed at the 5´ and 3' of the TTAA piggyBac insertion sites (Figure 2C) when compared to the flanking regions of the randomly selected TTAA sites (Figure 2C). Both the 5'-stretch of seven T's and the 3'stretch A's at the piggyBac TTAA insertion site were significantly different from similar stretches of the random chosen TTAA sites (Chi Square: 5,5E-08 *** and 0,013* for the 5'and 3' stretches, respectively). This indicates that piggyBac may have a slight insertional bias with regard to the TTAA flanking sequences. We next analysed the chromosomal distribution of the 127 inserts. Three inserts (we term repetitive region 1-3) could not be mapped to a specific chromosome but were located on contigs containing genes that belong to known subtelomeric gene families, the bir or Pb-fam families of genes 1035. The 124 remaining inserts were spread over all 14 chromosomes (Figure 2D, left panel) and a weak correlation (R²: 0.56) was observed between the number of inserts and the size of the chromosome. Chromosome 6 was exceptional as more than expected inserts (13) were identified in this chromosome compared to the number identified in the similar sized chromosomes 5, 7 and 8 in which 4-10 inserts were identified.

The position of the 127 inserts in and around predicted genes show that piggyBac inserted both in coding sequence (CDS) and untranslated regions (UTR's) of genes (Figure 2D, right panel). For these regions a comparable proportion of piggyBac TTAA insertion sites and 254 randomly chosen TTAA sequences were observed: 24% (piggyBac integration) and 23% (random TTAA sites) in 5'UTR's; 45% and 54% in CDS (including introns); 23% and 16% in 3'UTRs; 8% and 7% in intergenic regions. For all regions the distribution of the piggyBac insertion sites was not significantly different from the random selected TTAA sites (Fisher's test: 5'UTR: p = 0.90, CDS (including introns): p = 0.15, 3'UTR: p = 0.12; intergenic regions: p = 0.84). These results indicate that piggyBac insertion in P. berghei occurs randomly and there is no preference for insertion either within the CDS or non-CDS of genes. We analysed whether piggyBac had a preference for insertion into transcribed/expressed genes, by analyzing published data on expression of genes with piggyBac insertions. For this analysis we used the new P. berghei gene models/systematic id's provided by the 'GeneDB 2010 release' and included only insertions that were located at a distance of more than 1 kb from the CDS (see Additional file 1 Table S1 for the included/excluded genes). Expression data were obtained from published proteomes from different P. berghei life cycle stages 3 and gametocytes 7. In addition, all available P. berghei EST databases present in the PlasmoDB database 36 were used (using the gene overlap function at default settings to retrieve ESTs matching genes assigned to different life cycle stages). Of the 124 analysed piggyBac inserts, 91 (73%) had proteome/EST based expression evidence compared with expression assigned to 3594 of the 4479 (80%) P. berghei gene models ('2010 Sanger P. berghei release'). Of the genes with piggyBac inserts, 55% (68 of 124) were transcribed/expressed in the asexual blood stages (ABS) compared to 54% of all P. berghei gene models (2420 of 4479) with evidence for transcription/expression in ABS (not significant, Fishers test: p = 0.07). Since genes that are inactive in ABS might be less accessible for piggyBac insertion, we compared the proportion of inserts in genes which are inactive in ABS but are expressed in other stages. There is transcription/proteome evidence that 26% (1174 of 4479) of all genes are expressed exclusively in stages other than the ABS. For genes with piggyBac inserts, 20% (25 of 124) of the genes are expressed in other life cycle stages but not in ABS (not significant, Fishers test: p = 0.15). In none of the comparisons mentioned above were significant differences found, indicating that piggyBac insertion is not linked to the expression pattern or activity of the genes.

A large proportion of the inserts were found inside CDS, CDS introns or in the 5'UTR regions (86 out of 127) of genes and such insertions most likely affect or completely disrupt the expression of these genes. The presence of an insert in the CDS/introns/5'UTR may therefore provide indirect evidence that the gene is not essential for ABS development. This is because Plasmodium is haploid during ABS development; therefore disruption of a gene essential for ABS would result in parasites that can not be selected after transfection as the deletion is lethal. We analysed 73 genes in more detail that contained inserts either in the CDS, CDS intron or in the 5'UTR with a maximum distance of 500 bp to the start codon of the CDS (Additional file 2 Table S2). Interestingly for at least 4 genes evidence already existed that expression of these genes is not essential in ABS as have been demonstrated by standard targeted deletion of these genes (See references in Additional file 2 Table S2). In addition, published data on expression in blood and mosquito stages 78 showed that 8 genes (or their P. falciparum orthologs) are either gametocyte- or mosquito-stage specific and are not expressed in ABS, indicating that the lack of expression should have no effect on the survival of ABS. However, for many genes with piggyBac inserts evidence is available indicating expression of these genes in ABS (collated and available at PlasmoDB 36) suggesting that these genes have a non-essential and/or redundant function during asexual blood stage development. We confirmed the non-essential role of two of such genes, p41 ([GeneDB 34:PBANKA_100260]; interrupted in its 5' UTR by piggyBac) and metacaspase2 ([GeneDB 34:PBANKA_130230]; interrupted in a CDS intron by piggyBac), through targeted disruption of these genes. P41 belongs to proteins encoded by the 6-cysteine family of genes and in P. berghei is transcribed in ABS 37. In P. falciparum P41 is intimately associated with GPI-anchored proteins that are located to the surface of merozoites (i.e. in lipid rafts). It has been reported that all P. falciparum GPI-anchored proteins (apart from MSP5) on the surface of merozoites are essential 383940. We targeted P. berghei p41 by disruption of the CDS through double cross-over recombination (Additional file 3 Figure S1) and confirmed that this protein was not essential in P. berghei ABS. The gene encoding P. falciparum metacaspase 2 [PF14_0363] is transcribed throughout ABS development 3641. As with P41, we were able to disrupt the CDS of P. berghei metacaspase2 by targeted double cross-over recombination (Additional file 3 Figure S1), confirming the non-essential role of this protein in ABS. These data indicate that analysis of large scale piggyBac mutagenesis can provide evidence for the dispensability of certain, albeit ABS-expressed, genes for Plasmodium ABS. All 73 genes with piggyBac inserts in the CDS, CDS introns or in the 5'UTR region (within 500 bp from the start codon) have been deposited in the publically accessible database of genetically modified rodent malaria parasites (RMgmDB; 4222) and this information on piggyBac insertion is linked to the information on individual genes in PlasmoDB 36 and GeneDB 34. For identification of the exact location of the insertion, the sequence of 20 bp up-and downstream of the TTAA sequence, is provided for all genes in RMgmDB 42 (See also Additional file 2 Table S2).

To test the stability of piggyBac inserts in the genome, we generated parasite clones from both the subpopulations P1 (transient transfection of transposase) and P5 (stable expression of transposase) by the method of limiting dilution. For cloning of both P1 and P5 parasites we performed two sequential cloning procedures in order to obtain 'pure clones'. For P1 this resulted in generation of two clones P1.d.1 and P1.d.2. For P5 we obtained 4 clones P5.b.3, P5.i.1, P5.i.2 and P5.i.3. For both clones of P1 FIGE analysis of separated chromosomes showed evidence for piggyBac insertion into a single chromosome (chromosome 12 and 13/14 respectively; Figure 3A). Also by TAIL-PCR analysis only one insertion was detected for both clones. One being 633 bp into the 3'UTR of the gene [GeneDB 34:PBANKA_132830], located on chromosome 13 (clone P1.d.2) and the other being a mapped to a repetitive region, Repetitive region 1 (clone P1.d.1; see Additional file 1 Table S1). After mechanical passage of parasites of both clones for a period of 2 weeks (14 asexual multiplication cycles), no new insertions were detected by FIGE analysis of chromosomes (Figure 3A) and TAIL-PCR also identified only the same single insertion in each parasite clone as identified before mechanical passage (Additional file 1 Table S1). In contrast, in clones obtained from P5 lines that stably express transposase we found evidence for remobilization of the piggyBac insert, both by FIGE analysis of separated chromosomes and by TAIL-PCR. In clone P5.b.3 a dominant insert is observed in chromosome 13/14 (Figure 3A) but by TAIL-PCR we were able to identify 10 different inserts in different chromosomes (Figure 3B and Additional file 1 Table S1). Also in the three clones P5.i.1-3 one dominant insert was observed in the group of chromosomes 9-11 but faint hybridization signals are also observed in other chromosomes (Figure 3A), indicating the presence of additional inserts. TAIL-PCR confirmed the presence of inserts into chromosomes 9-11 as well as additional inserts in a number of different chromosomes in these three cloned lines (Additional file 1 Table S1). These results indicate that parasites of clones derived from the transposase parasite line are not homogeneous populations with respect to piggyBac inserts but they consist of mixed populations of parasites containing inserts in different chromosomes. After mechanical passage of parasites of the three P5.i.1-3 clones for a period of 2 weeks (14 asexual multiplication cycles), we detected by both TAIL-PCR and FIGE analysis 12 novel inserts within the parasite populations (Figure 3B, Additional file 1 Table S1). Together these results indicate that in the presence of transposase the piggyBac inserts are able to remobilize in the Plasmodium genome, as has been also reported in Drosophila melanogaster, Bombyx mori and mouse embryonic stem cells. 434445. In the absence of transposase remobilization was not detected and inserts remained stably integrated in the genome.

Experimental validation of P. berghei gene models would significantly improve annotation of the genome. Therefore, we designed the piggyBac donor plasmid for promoter trapping experiments by introducing the gfp-expression cassette without a promoter region next to the 5'ITR1 sequence. Integration of the construct downstream of a promoter that is active in blood stages would therefore result in GFP expression in blood stage parasites. We analysed GFP-expression by fluorescence microscopy of parasites of primary transfection populations P1 and P2 that were obtained by transfection with the donor and helper plasmid. In both populations low numbers of GFP-positive blood stage parasites (< 0.1%) were detected whereas in the control parasites that were transfected with only the donor plasmid no GFP-positive cells were found (Figure 2). To isolate GFP-expressing parasites from population P2, we performed FACS sorting of GFP-positive blood stages from tail blood of mice with asynchronous infections. Three populations (F1-3) each of 10-50 GFP-positive cells were FACS-sorted by setting three different gates based on GFP-fluorescence intensity as shown in Figure 4A. These parasites were intravenously injected into mice to generate expanded parasite populations for further analysis. FACS analysis of blood stages of the F1-F3 populations (obtained from overnight cultures to enrich for maturing schizonts) confirmed GFP expression in blood stages (Figure 4B) and demonstrates different GFP-expression patterns between the sorted populations. However, since the initial sorting of GFP-expressing parasites was performed using asynchronous blood stages from tail blood and the confirmation of GFP-expression was derived from cultured schizonts, the GFP-fluorescence intensities cannot directly be compared between the populations shown in Figure 4A and B. Southern analysis of separated chromosomes showed integration of the gfp-expression cassette into multiple chromosomes (7 and 1/2 in F1; 12 in F2; 7 and 9/11 in F3; Figure 4C). Using TAIL PCR we confirmed integration into these chromosomes: chromosome 7 in F1, chromosome 12 in F2, chromosome 7 and 10 in F3 (Additional file 1 Table S1). In addition to FACS sorting, we analysed GFP-expression by fluorescence microscopy in blood stages of the 15 subpopulations (a-o) that were obtained from P1 (Figure 2A). Four of the subpopulations contained low numbers of GFP-positive (results not shown), three of which showed low fluorescence intensity (P1f, P1g, P1m) and one, P1e, exhibited a stronger GFP-fluorescence (Figure 4B).

A total of 3 inserts were identified by TAIL-PCR survey in GFP-expressing parasites selected by FACS sorting. Two of these inserts occurred in a direction compatible with GFP expression (Figure 4D and See Additional file 1 Table S1). In addition, 2 inserts were identified from the P1e subpopulation derived by cloning of parent population P1 (See Table Additional file 1 Table S1 for inserts). The identification in a parasite population of one insert that is not compatible with GFP expression may indicate that this population contains parasites with gfp inserted into two different loci. FIGE-analysis of this population indeed shows evidence for integration into two loci, one in chromosome 6 and the other in one of the chromosomes of group 9-11 (Figure 4C). We determined transcription of gfp by Northern analysis in blood stages of the three FACS-sorted and P1e populations and in all parasites gfp transcripts were detected (Figure 4E). Transcription of gfp insert into the 5'UTR of gene [GeneDB 34:PBANKA_062360] (population F1/F3) was confirmed by RT-PCR (Figure 4F) demonstrating that our piggyBac approach had successfully trapped the promoter of this gene which belongs to the bir multi gene family. Interestingly, PCR analysis showed that the size of the RT-PCR amplified gfp-transcript in F1/F3 was smaller in comparison with the PCR-amplified genomic fragment (Figure 4F). Sequencing of the RT-PCR amplified product revealed that the smaller size was due to the unexpected removal of an intron in the piggyBac 5'ITR, not affecting the CDS of gfp (Additional file 4 Figure S2). This splicing event shows that the endogenous piggyBac 5'ITR can be recognized by the splicing machinery of P. berghei.

The pXL-BACII-GFP-HDHFR piggyBac transposon donor plasmid (pL1302) was created by cloning a gfp-expression cassette without promoter region and a hdhfr selectable marker cassette into the minimal piggyBac vector pXL-BACII 28. The gfp-expression cassette, containing gfpm3 and the 3'pbdhfr/ts was excised from plasmid pL0024 by BamHI/EcoRI digestion while digesting the pXL-BACII vector with BglII/EcoRI resulting in the intermediate plasmid pXL-BACII- gfpm3-3'pbdhfr flanked by XhoI/EcoRI restriction sites. The hdfr selectable marker cassette (5' pbeef1aa-hdhfr-3'pbdhfr) was excised from plasmid pL0009 with EcoRI digestion and subcloned into the intermediate pXL-BACII- gfpm3-3'pbdhfr plasmid, generating the final piggyBac donor plasmid, pXL-BACII-gfp-hdhfr. (pL1302). The pHTH transient helper plasmid (pL1301) was constructed by first excising the piggyBac transposase coding sequence from the pHTH plasmid used for P. falciparum 18; kindly provided by Bharath Balu) with BamHI and cloned into the intermediate plasmid pL0011 under the control of 5'-pbeef1aa and 3'pbdhfr, resulting in plasmid pL1303. The 5' pbeef1aa-transposase-3'pbdhfr cassette was excised from pL1303 with BamHI digestion and subcloned into the plasmid pL0004, generating pHTH_pb (pL1301) for transient transposase expression. The plasmids pL0004, pL0009, pL0011 and pL0024 can be obtained from MR4 66.

To stably integrate the piggyBac transposase gene into the P. berghei genome plasmid pL1307 was generated. The transposase coding sequence was excised from the pHTH plasmid (see above) with BamHI and cloned into the expression plasmid pL0010 (MR4; 66), which contains the toxoplasma gondii dhfr-ts encoding selectable marker cassette and the dssu-rrna target sequence for single cross-over integration into the non-essential c/d-rrna locus locus. The transposase was cloned with BamHI into the 5'pbama1 - 3'pbdhfr expression cassette of pL0010.

Transfection, selection and cloning of mutant parasite lines were performed as described 21. For experiments for transient transfection of parasites with the transposase-containing helper plasmid, circular donor plasmid pL1302 and helper plasmid pL1301 were mixed in a 1:1 (exp. 1056) or 1:2 (exp. 1056) ratios using 5 μg donor plasmid per transfection prior to transfection of the parasites (P. berghei ANKA, cl15cy1). Selection of resistant parasite populations after transfection was performed by treatment of the infected mice with pyrimethamine. As a control parasites were transfected with 5 μg donor plasmid (exp 1057). For generation of the transgenic line containing transposase stably integrated into the genome, parasites (P. berghei ANKA, cl15cy1) were transfected with plasmid pL1307 after linearization with Apa-1. Selection and cloning of transgenic parasites were performed as described 21. One clone, 1042cl1 (TPS_ama1) was used for further experiments. After transfection of TPS_ama1 with donor plasmid pL1302 a resistant parasite population (1182) was selected by treatment with WR9221 67. The genotype of transfected parasites was analysed by standard PCR analysis and Southern blot analysis of digested genomic DNA or of FIGE separated chromosomes 23.

RNA extraction and Northern blot analysis was performed according to standard methods 6869. RNA of mixed blood stages was hybridized to a gfp-probe that was PCR amplified from plasmid pL1302 using the primer set 3552-3553 (Additional file 5 Table S3) that is specific for the gfp coding sequence (CDS). As a loading control, the ethidium bromide stained gel was used. RNA used for RT-PCR analysis was first digested with DNaseI for 1 hour, followed by phenol/chloroform extraction. Reverse transcription was performed using SuperScriptIII™ (Invitrogen) according to the manufacturer's instructions after splitting the RNA into two samples, one to which reverse transcriptase was added (+RT) and one without (-RT). RT-PCR was performed using primer 4571 targeting a P. berghei genomic sequence located 117 bp upstream of the confirmed piggyBac insertion site in 5' UTR of the bir gene [GeneDB 34:PBANKA_062360]. Primer 4571 was used in combination with primer 3209, located in the gfp CDS of the piggyBac insert (442 bp from the terminal end of the 5'ITR; see Additional file 5 Table S3 for primer sequences). The expected PCR product size on gDNA using the primer pair 4571/3209 is 559 bp. As a control for reverse transcription and DNaseI digestion, the gene [GeneDB 34:PBANKA_133840] was amplified from both cDNA and genomic DNA with the primer set 3902/3903 located on either side of two confirmed introns (expected sizes are 530 bp on gDNA and 232 bp on cDNA). Thirty-five PCR amplification cycles were performed with Amplitaq DNA polymerase (Roche) using 4 mM MgCl₂ and an annealing temperature of 50°C and 1 minute extension time at 72°C.

For FACS sorting of GFP-positive parasites, we collected 10 μl of infected tail blood (with a parasitemia of 3%) in complete culture medium from mice infected with parasites from exp. 1056 (see above). GFP-positive cells were sorted on the FACSAria Cell Sorter (Becton Dickinson). Selection of blood cells on forward/side light scatter and filter settings for detection of GFP-fluorescence was performed as described 70. GFP-positive cells were sorted using three different gates as shown in the Results Section (Figure 4A) and sorted cells (50-300 cells per sorting experiment) were collected in ~300 μl complete culture medium at room temperature. This cell-suspension was injected intravenously (tail vein) into a single mouse. 7-8 days after injection, parasites were collected from the mice and were analysed for GFP expression. GFP-fluorescence intensity of live blood stages was examined after staining with Hoechst-33258 71 using a Leica-fluorescence MDR microscope and pictures recorded using a DC500 digital camera.

To identify piggyBac inserts, two sets of each three primers specific for the piggyBac 5' and 3'ITR regions (Additional file 5 Table S3) were used successively in combination with 8 Semi-Arbitrary Degenerate (SAD) primers (Additional file 5 Table S3) with 140 to 192 fold degeneracy. Each SAD primer contains a Sau3Ai site (4 with GATC and 4 with the reverse complement CTAG at the 3' end) since this sequence is widely distributed in the P. berghei genome (approximately 2.3 × 10⁴ occurrences identified by text searches against all contigs present in PlasmoDB 36. The SAD primers were designed to reflect the variability surrounding the GATC sites in P. berghei from a multiple alignment of around 1100 genomic P. berghei sequences obtained from PlasmoDB 36. TAIL-PCR was performed as described 5272 with some modifications of the protocol (See Additional file 6 Table S4 for the different conditions used for TAIL-PCR). Briefly, eight primary TAIL-PCR reactions were performed, each with one of the SAD primers (3649-3656) and in combination with either the piggyBac 5'ITR primer 3202 or the 3'ITR primer 3726. After the initial PCR reaction, products were diluted 1:2000 with distilled water and used in the secondary reaction, containing the piggyBac primers 3203 (5'ITR) or 3727 (3'ITR) together with the same SAD primer as used in the primary reaction. The secondary PCR products were diluted 1:500 and used in the tertiary reaction mixture. The tertiary PCR reaction was performed with piggyBac primers 3204 (5'ITR) or 3728 (3'ITR) again with the same SAD primers as before. The tertiary amplification fragments were visualized through electrophoresis in 1% agarose gels, where the average fragment size was usually between 300 and 500 bp. Immediately after the tertiary reaction, the reaction mixture was purified using a PCR-purification kit (Roche Applied Science, Germany) according to manufacturers instructions. Subsequently the reaction mixture (containing several TAIL PCR products) was ligated into the TOPO TA vector (Invitrogen) and transformation and blue/white screening was performed according to manufacturer's instructions. DNA from 15 to 20 of positive colonies were analysed by sequencing. The insertion site in the genome of the piggyBac cassette was determined by BLASTN (vs. PlasmoDB 36 and GeneDB 34 databases) analysis of the sequences outside of the piggyBac cassette followed by precise identification of the insertion site using BioEdit alignment and mapping (using Artemis) of each TAIL-PCR sequence to the closest gene. If an insertion occurred more than 1000 bp from the CDS of a gene, it was not considered to be located in its (potential) 5' or 3'UTRs. For comparison of characteristics of the piggyBac insertion sites, 254 randomly TTAA sites and flanking regions were extracted from P. berghei genomic sequences (obtained from the Sanger Centre 34). Visualization of the TTAA insertion site and flanking site conservation was performed with Weblogo 3.0 73.

Using the Boolean operators, implemented in the PlasmoDB site 36, sets of genes with proteome/EST evidence were generated using the databases implemented on the PlasmoDB site 36 including published P. berghei gametocyte and proteomes from all developmental stages 37. All datasets were converted to the updated gene models available (released by Sanger Centre as of 15 April 2010) and a comparison was made between all expressed/transcribed genes and genes with piggyBac insertions for each set.

Two replacement DNA constructs were made for targeted disruption of p41 [GeneDB 34:PBANKA_100260] and metacaspase2 [GeneDB 34:PBANKA_130230] by double cross-over homologous recombination. Details of primers used to generate double-cross over replacement constructs (i.e. to introduce 'targeting regions' in the replacement construct pL0001; (See MR4 66 and also Additional file 3 Figure S1 and Additional file 5 Table S3). The p41 gene has been disrupted in the ANKA reference line cl15cy1 21 and metacaspase2 has been disrupted in the GFP-expressing reference line of the ANKA strain (line 507cl1, identifier RMgm-7 in 42). Transfection, selection and cloning of mutant parasites was performed as described 21. Correct integration of the constructs was analysed by diagnostic PCR and Southern analysis of FIGE-separated chromosomes and transcription by standard Northern analysis of mRNA collected from mixed blood stage parasites 21. Primers for amplification of the probes used for Northern analysis are shown (See Additional file 3 Figure S1 and Additional file 5 Table S3). As a loading control Northern blots were hybridized with primer L644R that hybridizes to the blood stage large subunit ribosomal RNA 74.