piggyBac insertions into the P. falciparum genome were obtained by co-transfection of parasite erythrocytic stages with a transposon plasmid and a transposase-expressing helper plasmid as described previously 21. To optimize the piggyBac system for maximum efficiency, several transposon and transposase plasmids were tested in P. falciparum (Fig. 1). The transposon plasmids tested contained different regulatory elements and drug selectable markers, which, however, resulted in similar transformation efficiencies (interpreted as the number of piggyBac insertions obtained per transfection). As piggyBac transposase is the functional enzyme catalyzing the integration event, we hypothesized that increased expression of the transposase with a stronger promoter would result in increased transformation efficiency. The hsp86 promoter in the helper plasmid, pHTH 21, was therefore replaced with a previously described dual Plasmodium promoter, containing 5' calmodulin and 5' dhfr-ts regions in head to head orientation 22. Corroborating our theory, significantly higher transformation efficiencies (an average of 3.1 × 10^-6) were obtained using the dual promoter for transposase expression as compared to using pHTH (an average of 1.6 × 10^-6) in approximately 40 transfections each (χ² test, df 1, P = 0.015).

Following transfection with piggyBac plasmids, drug resistant parasite populations were established rapidly, within 2–3 weeks and the total number of piggyBac insertions obtained per transfected parasite population varied from 1 to 14. Through 81 independent transfections, we generated 177 unique mutant clones of P. falciparum with piggyBac insertions in their genomes. Southern blot hybridization analysis of parasite clones, derived by limiting dilution of drug-resistant populations, revealed single piggyBac insertions in all except two clones that had two insertions each (data not shown). Also, none of the mutant clones retained the piggyBac plasmid as episomes indicating highly efficient transposition events (data not shown). Out of the 179 piggyBac insertions identified, 165 could be mapped unambiguously on the P. falciparum genome by performing BLAST searches using NCBI http://www.ncbi.nlm.nih.gov/genome/seq/BlastGen/BlastGen.cgi?taxid=5833 and PlasmoDB 23 databases. The remaining 14 insertions either mapped to telomeric repetitive elements or could not be mapped to a chromosomal location through BLAST searches of public databases. The identified piggyBac insertion sites were distributed throughout the genome in all 14 P. falciparum chromosomes (Fig. 2a) with no bias for any particular chromosome (Fig. 2b). All piggyBac insertions were obtained in the expected TTAA target sequences except two that integrated into TTAT and TTAG sequences. As in other organisms 1720, piggyBac preferentially inserted into predicted transcribed units of P. falciparum genome (Fig. 3a), affecting 178 transcription units. Thirty-six of the insertions resulted in direct disruption of open reading frames (ORFs) and 3 insertions were mapped to introns. A vast majority of insertions (119) occurred in 5' untranslated regions (UTRs) whereas only a few (22) were obtained in 3' UTRs (Additional file 1).

Obtaining unbiased insertions into the genome is critical for whole-genome mutagenesis and other large-scale analyses. Hence, we evaluated the randomness of piggyBac insertions into the P. falciparum genome by comparing the functional categories of genes with piggyBac insertions to all annotated genes in the genome. An identical functional distribution of genes was seen in both piggyBac insertion loci and the genome (Fig. 3b) except for fewer insertions in genes involved in DNA metabolism/DNA-binding and invasion/pathogenesis (Fisher's exact test, P = 0.038 and P = 0.04, respectively). Since the parasite erythrocytic stages were used for piggyBac transformation, we further investigated the bias for piggyBac insertions in erythrocytic stage genes relative to genes expressed in other stages of development. By utilizing the gene expression profiling data for P. falciparum 3, we classified all annotated genes based on their expression in different parasite life cycle stages and confirmed unbiased piggyBac insertions in genes expressed in all parasite stages (Fig. 3c). A separate comparison of genes with piggyBac insertions in coding sequences only to all genes also revealed no significant insertion bias for any functional category or stage of expression (data not shown).

Even though transposon-mediated mutagenesis is a relatively random process, preferential insertion into genomic hotspots is characteristic of some transposons 20. In our studies, we observed a significantly higher number of piggyBac insertions in 5' UTRs and a significantly lower number in coding sequences, relative to a distribution of 214 randomly selected genomic TTAA sequences (Fig. 3d).

Previous studies in other organisms had observed some AT-richness around piggyBac insertion sites 1724. However, it was somewhat surprising that our analysis of a 100 bp flanking region showed a significantly higher AT-content around piggyBac inserted TTAA sequences (average AT content of 85.56%) as compared to random TTAA sequences (average AT content of 80.24%), in the already AT-rich P. falciparum genome (two-tailed t-test, P = 2.95 × 10^-13). A closer look at the piggyBac insertion sites revealed their presence in the middle of an AT-rich core of 10 nucleotides predominantly with 'T's upstream and 'A's downstream (Fig. 4a, upper panel). No such signature motif was present around the randomly selected TTAA sequences either from the genome (Fig. 4a, lower panel). Even when only analyzing the genomic 5' UTRs, a similar bias in the insertion site selection existed (Fig. 4b).

The preferential insertion of piggyBac into transcription units prompted us to investigate the feasibility of forward genetic studies in P. falciparum that have been completely lacking thus far. Little is known about what metabolic pathways and processes are essential for parasite growth and survival in the blood of the vertebrate host, and therefore we screened the erythrocytic stages of P. falciparum mutant clones for attenuated growth phenotypes. We first screened for mutant clones that appeared to have aberrant growth rate by standard light microscopy methods and then studied them further by performing more precise growth assays. The mutant clones selected for growth analysis contained single piggyBac insertions in their genomes in either coding sequences or 5' UTRs and were associated with several metabolic pathways (Fig. 5a). To confirm that piggyBac insertion into the genome alone does not affect growth, additional mutant clones were included as controls. An exponential growth curve was generated for each mutant clone by estimating parasitemias every 24 hrs for 7 days using flow cytometry as described before 2526 with some modifications. Four mutant clones (A5, B7, E6 and F3) displayed significantly reduced growth rate as compared to five other insertional mutants (B3, B4, F10, G1, and H11) and the wild type (WT) clones (Fig. 5b). The experiment was performed three times, with two sub-clones for each mutant and similar results were obtained in all experiments (data not shown). The parasite exponential growth curve was further used to estimate the individual doubling times of the mutant clones as described previously 26 that confirmed the observed attenuated phenotypes (Table 1, See Fig. S1 in Additional file 2). Knock out of gene expression was confirmed in clones with insertions in coding sequences by RT-PCR (See Fig. S2 in Additional file 3). Clones A5 and F3 with insertions in the coding sequences of PFF0770c and MAL8p1.104, respectively, were the most affected with an approximate growth rate of only 30% as compared to the WT clones (Fig. 5c). The attenuated growth rates observed in these mutant clones substantiate their significance in intra-erythrocytic development of the parasite, though additional studies are required to characterize the attenuation mechanisms.

piggyBac plasmids used for transfections were derived from previously reported plasmids pXL-BACII-DHFR and pHTH 21.

pLBacII-HDH-pXL-BacII-DHFR was digested with XhoI and the site was removed by filling in the overhangs with klenow and religation to yield pLBacII-DHFR. The human DHFR selection cassette in pLBacII-DHFR was then replaced with a different human DHFR drug selection cassette from the plasmid pHD22Y 43 using EcoRI/BamHI to yield pLBacII-HDH.

pLBacII-HDH-GFP- The gfp coding sequence along with 3' Pbdhfr was amplified as a single fragment from the vector pHH2 44 by PCR with extensions for restriction sites SpeI and ApaI using primers F-ACTAGTGCGGCCGCCTACCCT and R-GGGCCCGGTACCCTCGAGATCTTAGAATGAAGATCTTATTAC. The PCR product was then cloned into pGEM-Teasy vector (Promega) and sub-cloned into pLBacII-HDH using ApaI and SpeI.

pLBacII-HDH-eGFP- A 200 bp region of 5' eba-175 was amplified from the P. falciparum genome using primers F-ATCGATGAATATAATTGATTGATTGTAATAAAAAGTG and R-GGGCCCTGTATGCACATTGAATATATTTATATGTTATTATC and cloned into pLBacII-HDH-GFP as a ClaI/ApaI fragment.

pLBacII-HDH-KanOri- The kanamycin resistance gene and pUC origin of replication were amplified as a single fragment by PCR from the vector pEGFP-C1 (Clontech) using primers F-ATGATGATGGGATCCAAATGTGCGCGGAACCCC and R-ATGATGATGGGATCCGCAAAAGGCCAGCAAAAGG and cloned into pGEM-Teasy vector (Promega). The fragment was then sub-cloned into the plasmid pLBacII-HDH as a BamHI fragment.

pLBacII-HBH- The hDHFR coding sequence was first cut out from the vector pHD22Y using NsiI and HindIII and replaced with the blasticidin-S-deaminase (BSD) coding sequence that was cut out from the vector pCBM-BSD 45 using NsiI and HindIII. The BSD selection cassette in pHD22Y was then moved as an EcoRI/BamHI fragment into the vector pL-BacII-DHFR to yield pLBacII-HBH.

pLBacII-HDGH- The hDHFR-GFP fusion gene was cut out from the vector pHDGFP2 46 using NsiI and HindIII and cloned into pHD22Y replacing the human DHFR coding sequence. The whole selection cassette was then moved as an EcoRI/BamHI fragment into the vector pLBacII-DHFR to yield pLBacII-HDGH.

pDCTH- The plasmid with a dual promoter for transposase expression was created by PCR amplifying 5' Pcdhfr-ts and 5' calmodulin as an EcoRI fragment from the plasmid pHC1-CAT 22 using primers F-ATGATGGAATTCCCTGATATATTTCTATTAGGTATTTATTA; R-ATGATGGAATTCTTTGTAAGTTTTAGGTGTGTGTAT and swapping it with the 5' hsp86 region in the helper plasmid, pHTH.

P. falciparum clone NF54 was cultured in human erythrocytes at 5% hematocrit in RPMI1640 medium containing 0.5% Albumax II, 0.25% sodium bicarbonate and 0.01 mg/ml gentamicin. Transfections were performed using red blood cells as described previously 21. Briefly, mature blood-stage parasites were purified on a MACS magnetic column (Miltenyi Biotec) and 1 million purified parasites were added to erythrocytes loaded with 100 μg of the transposon plasmid and 50 μg of the transposase plasmid to start a 5 ml parasite culture. Individual mutant clones were obtained by limiting dilution of parasites post-drug selection.

Genomic DNA (2 μg) extracted from transformed parasites was digested with 10 units of either Dra I or Rsa I and used either in inverse PCR 21 or vectorette PCR reactions according to manufacturer's instructions (UVS1 Vectorette™ Genomic Systems, Sigma). The amplified PCR products were sequenced with primers in piggyBac inverted terminal repeats 21 and analyzed using MACVECTOR software (MacVector, Inc.). Insertion sites were identified by performing BLAST searches using NCBI http://www.ncbi.nlm.nih.gov/genome/seq/BlastGen/BlastGen.cgi?taxid=5833 and PlasmoDB databases 23.

Growth assays were performed by maintaining asynchronous cultures of P. falciparum wild type and mutant clones at parasitemias 0.5–2% in 96-well plates by diluting every 48 hrs. Parasite cultures were plated in triplicates for each time point and samples were taken every 24 hrs for 7 days and fixed in 0.05% glutaraldehyde after removal of culture medium. Flow cytometry was used to estimate parasitemia as described before 2547 by staining parasites with Ethidium bromide and analyzing using FACSCanto™ II flowcytometry system (Becton, Dickinson and Company) in a high throughput format. A total of 20,000 cells were counted for each sample. The data were analyzed using FACSDIVA™ software (Becton, Dickinson and Company). Growth rate (defined as the change in parasite numbers every 24 hrs over a period of 7 days) analyses were performed using SAS (9.1). The total number of parasites (y) (parasitemia × dilution factor), was plotted against time (×) and fitted to the exponential growth curve

where, D is the intrinsic parasite doubling time and m0 is the theoretical parasite number at time 0. To compare directly the growth rate of parasite clones with slightly different starting parasitemias, the -fold increase of the parasite number, normalized to have a single theoretical parasite for each culture at time 0, was used for graphing the growth curve 26. One hundred parameter initiation values ranging from 5 to 105 were tested and the best converging model with the smallest Sum Square of Error (SSE) was chosen for estimation of doubling time.