A Novel ‘Gene Insertion/Marker Out’ (GIMO) Method for Transgene Expression and Gene Complementation in Rodent Malaria Parasites

Research on the biology of malaria parasites has greatly benefited from the application of reverse genetic technologies, in particular through the analysis of gene deletion mutants and studies on transgenic parasites that express heterologous or mutated proteins. However, transfection in Plasmodium is limited by the paucity of drug-selectable markers that hampers subsequent genetic modification of the same mutant. We report the development of a novel ‘gene insertion/marker out’ (GIMO) method for two rodent malaria parasites, which uses negative selection to rapidly generate transgenic mutants ready for subsequent modifications. We have created reference mother lines for both P. berghei ANKA and P. yoelii 17XNL that serve as recipient parasites for GIMO-transfection. Compared to existing protocols GIMO-transfection greatly simplifies and speeds up the generation of mutants expressing heterologous proteins, free of drug-resistance genes, and requires far fewer laboratory animals. In addition we demonstrate that GIMO-transfection is also a simple and fast method for genetic complementation of mutants with a gene deletion or mutation. The implementation of GIMO-transfection procedures should greatly enhance Plasmodium reverse-genetic research.


Introduction
Reverse genetic technologies have been widely applied to gain an understanding of the function of genes in Plasmodium and to provide insight into the biology of malaria parasites and interactions with their hosts (for reviews see [1][2][3]. The availability of efficient genetic modification technologies for the rodent malaria parasites P. berghei and P. yoelii and the possibilities for analysis of these parasites throughout the complete life cycle have made P. berghei and P. yoelii the most frequently used models for analysis of gene function [2]. Targeted disruption or mutation of genes coupled with protein tagging has provided insight into Plasmodium gene function and parasite protein expression, localization and transport. Reverse genetics is not only applied to understand Plasmodium gene function by gene deletion but is also increasingly being used to generate parasites that express heterologous proteins, for example parasites having transgenes introduced into their genome to encode fluorescent or luminescent reporter proteins. Such reporter parasites have been instrumental in the visualization and analysis of parasite-host interactions in real-time in vitro and in vivo [4][5][6]. The use of mutant parasites to investigate host-parasite interactions as well as parasite gene function requires genetic modification systems that are flexible and easy to perform. The application of reverse genetics in P. berghei and P. yoelii is however restricted by the limited number of drug resistance genes (permitting the selection of transformed parasites) that are currently available. This low number of selection markers hampers and slows down successive modifications in the genome of the same parasite line. Currently only two resistance gene/drug combinations exist for use in rodent malaria parasites that can be used in successive transfections, specifically dhfr-ts/pyrimethamine and hdhfr/WR99210 [7]. Since both drug-selection markers confer resistance against pyrimethamine, the introduction of consecutive genetic modifications in the same parasite can only be performed by first selecting with pyrimethamine followed by WR99210 selection [7]. In order to circumvent the problem of limited drugselection markers, GFP has been utilized as a selection marker and permits the selection of transformed P. berghei parasites by flow cytometry [8,9]. In addition, a method has been developed for removing drug-selection markers from transformed P. berghei parasites by utilizing the yeast fcu (yfcu) selection marker and negative selection with the drug 5-fluorocytosine (5-FC) [10], which kills all parasites expressing yfcu. In this method transformed parasites expressing the fusion gene hdhfr::yfcu are first selected by positive selection with pyrimethamine. Subsequently, negative selection with 5-FC is applied to select for marker-free parasites that have 'spontaneously' lost the hdhfr::yfcu marker from their genome, achieved by a homologous recombination/excision event around the selection cassette [10]. Both the selection of GFPexpressing mutants by flow cytometry and selection of 'spontaneous' marker-free mutants by negative selection have their limitations. They are laborious and time consuming, and also require the use of many extra animals as additional cloning steps in mice are required; therefore these methods are not commonly used for successive genetic modifications or for complementation studies [11].
Here we report the development and application of a novel 'gene insertion/marker out' (GIMO) system for transfection of two rodent malaria parasites, P. berghei and P. yoelii. For both species we have created reference mother lines that contain the hdhfr::yfcu selection marker stably integrated into the silent 230p genomic locus. We show that transfection of these mother lines with DNAconstructs that target the modified 230p locus, followed by negative selection of transformed parasites with 5-FC is a simple and fast method to generate mutants that stably express heterologous proteins and are free of drug-selectable markers. These mother lines are therefore useful tools to generate a wide range of mutants expressing reporter and/or other heterologous proteins (under the control of different promoters) without restricting subsequent modification of the genome of these parasites. In addition, we demonstrate that GIMO-transfection is a simple and fast method to genetically complement, restoring the wild-type genotype of parasite mutants with a gene deletion or gene mutation. Importantly, GIMO transfection can be easily partnered for use with a recently developed 'recombineering' system for high-throughput, genome wide and highly efficient generation of gene targeting constructs [12].

Results
Generation of the P. berghei and P. yoelii 'gene insertion/ marker out' (GIMO) mother lines For both P. berghei ANKA and P. yoelii 17XNL transgenic parasites were generated that express a fusion of a drug resistance gene and a drug sensitivity gene, the so called postive-negative selectable marker (SM), constitutively expressed by the P. berghei eef1a promoter ( Figure 1A). Specifically, these parasites contain a fusion gene of hdhfr (human dihydrofolate reductase; positive SM) and yfcu (yeast cytosine deaminase and uridyl phosphoribosyl transferase; negative SM) stably integrated into the 230p locus (PBANKA_030600 in P. berghei and PY03857 in P. yoelii) through double cross-over recombination. These lines are named GIMO mother lines (gene insersion/marker out); for P. berghei GIMO PbANKA (line 1596cl1) and for P. yoelii GIMO Py17X (line 1923cl1). Both GIMO mother lines were cloned after transfection by positive selection with pyrimethamine. Correct integration of the hdhfr::yfcu selectable marker cassette in the 230p locus was demonstrated by PCR and Southern analysis of chromosomes separated by pulse-field gel electrophoresis ( Figure 1B and 1C). The multiplication rate of asexual blood stages per 24 h as determined in mice infected with a single parasite [13], gametocyte production and production of oocysts and sporozoites were identical to those of the parent P. berghei and P. yoelii lines (data not shown). These GIMO mother lines are used for introduction of transgenes into the modified 230p locus through transfection with constructs that target the 230p locus. These constructs insert into the 230p locus ('gene insertion'), thereby removing the hdhfr::yfcu selectable marker ('marker out') from the genome of the mother lines. Transgenic parasites that are marker-free are subsequently selected by applying negative drug selection using 5-FC (see below).
Assessing the efficiency of GIMO-transfection to select transgene expressing, drug-selectable marker-free P. berghei parasites We generated a test DNA-construct containing a transgene expression-cassette to test the efficiency of selection of transgenic mutants through the application of negative selection using 5-FC after transfection into the GIMO PbANKA mother line. This construct contains the mCherry gene under the control of the constitutive eef1a promoter and 230p targeting sequences ( Figure 2A) and lacks a drug selectable marker cassette. This DNA-construct, pL1628, targets the same regions in the 230p locus in which hdhfr::yfcu selection cassette was introduced in the GIMO PbANKA mother line (Figure 2A). Transfection of GIMO PbANKA (exp. 1645) was performed using standard procedures [14] except that after transfection negative drug selection was applied instead of positive drug selection. This negative selection was performed by treating mice that were infected with transfected parasites with the drug 5-FC for 4 consecutive days (one dose per day of 10 mg), starting 24 hours after transfection.
Transfected parasites of line 1645 were collected at day 7 and 8 after transfection (at a parasitemia of 0.5-3%) for phenotype and genotype analyses. Diagnostic PCR and Southern analysis of separated chromosomes confirmed the correct integration of the test construct and simultaneous removal of the hdhfr::yfcu selection cassette ( Figure 2B). Analysis of mCherry expression by fluorescence microscopy in blood stage parasites of line 1645 showed that .90% of the parasites expressed mCherry ( Figure 2C). Quantification of the percentage of mCherry-expressing parasites was performed by FACS analysis of mature schizonts collected from overnight blood stage cultures. Expression of transgenes, such as mCherry, under the control of the eef1a promoter increases with the maturation of parasites inside blood cells and therefore FACS quantification is improved by analysing mature schizont stages (these stages are selected based on Hoechst-fluorescence) [15]. FACS analysis confirmed that .90% (93%61.1 Figure 2D) of the schizonts were mCherry positive. Since episomal constructs cannot be maintained during selection in GIMO-transfected parasites (see Discussion), these analyses demonstrate that GIMO-transfection permits the selection parasites that express transgenes and are marker-free.
To further investigate the efficiency of the GIMO system, we performed a set of independent transfections with the DNAconstruct pL1628 (exp. 1794-1799) in the GIMO PbANKA mother line. In these experiments transfected parasites were selected using negative selection as described above and mCherry expression analysed by FACS ( Figure 3A). In 5 out of 6 transfection experiments, the percentage of mCherry-expressing parasites was higher than 75%, whereas in one experiment (exp. 1798) 32% of schizonts were mCherry positive ( Figure 3A). The presence of mCherry negative parasites in the drug-selected population indicates that non-transformed parasites survived the drug-selection but presumably still carry the hdhfr::yfcu cassette. We therefore analysed the genotype of the selected populations of all experiments by quantitative real-time PCR (qPCR) and Southern analysis of separated chromosomes to determine the ratio between parasites with and without hdhfr::yfcu. For qPCR, C T values of amplification of mCherry, hdhfr::yfcu and the control hsp70 gene were determined and the percentage of mCherry positive parasites was calculated as the relative ratio between mCherry and hdhfr::yfcu using the 2 2DDCT method [16]. The percentage of mCherry positive parasites based on qPCR correlated well with the percentage determined by FACS analysis ( Figure 3A). Southern analysis also showed that in the selected populations a low percentage of parasites still contain the hdhfr::yfcu gene ( Figure 3B). These observations indicate that the application of negative selection after transfection of GIMO PbANKA , while it highly enriches for transformed parasites, it does not generate a pure population of marker-free parasites. Therefore, parasite cloning after negative selection is an essential step in GIMO-transfection in order to obtain correctly transformed parasites that express the transgene and are drugselectable marker free.
Generation of a P. yoelii reporter line, PyGFP-luc con , which is marker-free and expresses a GFP-luciferase fusion protein, by GIMO-transfection The application of negative selection to genetic modification of P. yoelii has not been reported. To test the possibility to select P. yoelii parasites lacking hdhfr::yfcu from a population of hdhfr::yfcucontaining parasites by negative selection, we generated a construct (pL1847) that targets the modified py230p locus of the Figure 1. Generation and genotype analyses of P. berghei and P. yoelii GIMO mother lines. (A) Schematic representation of the constructs used to introduce the positive-negative selectable maker cassette in the P. berghei (PbANKA) or P. yoelii (Py17XNL) 230p locus. DNA constructs pL1603 (targeting P. berghei 230p, PBANKA_030600) and pL1805 (targeting P. yoelii 230p, PY03857) containing a fusion of the positive drug selectable marker hdhfr (human dihydrofolate reductase) and negative marker yfcu (yeast cytosine deaminase and uridyl phosphoribosyl transferase) under the control of the eef1a promoter target the 230p locus at the target regions (hatched boxes) by double cross-over homologous recombination. Location of primers used for PCR analysis and sizes of PCR products are shown (see Table S2 for all primer sequences). (B) Diagnostic PCR and Southern analysis of PFGseparated chromosomes confirming correct integration of the construct in the P. berghei mother line GIMO PbANKA : 59 integration PCR (59 int; primers 5510/3189), 39 integration PCR (39 int; primers 4239/5511), amplification of hdhfr::yfcu marker (SM; primers 4698/4699) and the original P. berghei 230p (230p; primers 1637/5600). Primer location (black arrows) and product sizes are shown in A. For Southern analysis, PFG-separated chromosome were hybridized using a 39UTR pbdhfr probe that recognizes the construct integrated into P. berghei 230p locus on chromosome 3 and the endogenous locus of dhfr/ts on chromosome 7. (C) Diagnostic PCR and Southern analysis of PFG-separated chromosomes confirming correct integration of the construct in the P. yoelii mother line GIMO Py17X : 59 integration PCR (primers 6527/4770), 39 integration PCR (primers 4771/6528), amplification of hdhfr::yfcu marker (primers 4698/4699) and the P. yoelii 230p original locus (primers 6529/6530). Primer location (grey arrows) and product sizes are shown in A. For Southern analysis, chromosomal hybridization using a 39UTR pbdhfr probe recognizes the construct integrated into P. yoelii 230p locus on chromosome 3 and the endogenous locus of dhfr/ts on chromosome 7. doi:10.1371/journal.pone.0029289.g001 Figure 2. Generation of a marker-free mCherry-expressing parasite using GIMO-transfection. (A) Schematic representation of the introduction of a mCherry-expression cassette into the GIMO PbANKA mother line. Construct pL1628 containing the eef1a-mCherry-39pbdhfr cassette (mCherry; red box) is integrated into the modified P. berghei 230p locus containing the hdhfr::yfcu selectable marker cassette (black box) by double cross-over homologous recombination at the target regions (hatched boxes). Negative (Neg) selection with 5-FC selects for parasites (line 1645) that have mCherry reporter introduced into the genome and the hdhfr::yfcu marker removed. Location of primers used for PCR analysis and sizes of PCR products are shown (see Table S2 for primer sequences). (B) Diagnostic PCRs and Southern analysis of PFG-separated chromosomes confirms the correct integration of construct pL1628 in line 1645 parasites shown by the absence of the hdhfr::yfcu marker and the presence of the mCherry gene: 59 integration PCR (59 int; primers 5510/4958), 39 integration PCR (39 int; primers 5515/5511), amplification of hdhfr::yfcu (SM; primers 4698/4699) and the eef1a-mCherry (EF-mC; primers 3173/5514). Primer locations and product sizes are shown in A (primer sequences in Table S2). Hybridization of separated chromosomes of GIMO PbANKA and line 1645 using a hdhfr probe recognizes the hdhfr::yfcu marker in the 230p locus on chromsomse 3 in GIMO PbANKA but is absent in line 1645. Hybridization with 39UTR dhfr probe recognizes both modified the 230p locus on chromosome 3 (both marker and mCherry expression cassettes contain the 39pbdhfr sequence) and the endogenous dhfr/ts gene on chromosome 7 as loading control. (C) Fluorescence microscopy of a live mCherry-expressing trophozoite of line 1645; bright field (BF), DNA staining (Hoechst; Blue) and mCherry expression (red). (D) FACS analysis of mCherry-expressing blood stages of line 1645. The percentage of mCherry-expressing parasites was performed by FACS analysis on cultured blood stage. Mature schizonts (12-16 N) were selected based on their Hoechst fluorescent intensity (gate P2) and mCherry-expressing schizonts were selected in gate P3 (right panel). doi:10.1371/journal.pone.0029289.g002 GIMO Py17X mother line by double cross-over homologous recombination. Plasmid pL1847 contains a fusion gene of gfp and luciferase under the control of the P. berghei eef1a promoter ( Figure 4A). Integration of this construct will result in the introduction of the gfp-luc expression cassette and a simultaneous removal of the hdhfr::yfcu gene from GIMO Py17X ( Figure 4A). Transfection of GIMO Py17X parasites and negative selection was performed as described above for GIMO PbANKA . Comparable to results obtained with the transfection of GIMO PbANKA , two mice (exp. 1970 & 1971) that were infected with GIMO Py17X transfected parasites became positive at day 6 (parasitemia 1-2%) after selection with the drug 5-FC. Analysis by fluorescence microscopy showed that ,30% and ,70% of the parasites of line 1970 and 1971, respectively, were GFP positive ( Figure 4B). Southern analysis of PFG-separated chromosomes confirmed that most drug-selected parasites of line 1971 had removed the hdhfr::yfcu selectable marker ( Figure 4C). We obtained three clones of line 1971 and all three expressed luciferase as shown by in vivo imaging of mice infected with 1971cl1-3 blood stages parasites ( Figure 4D). PCR analysis confirmed the correct integration of the fusion gene gfp-luciferase and removal of hdhfr::yfcu ( Figure 4E). The results demonstrate that GIMO-transfection and the negative selection procedure can be applied to P. yoelii in order to generate parasites that express transgenes and are free of drug-selectable markers. In addition, these marker-free P. yoelii 1971 cloned lines (PyGFP-luc con ), are excellent tools to quantitatively analyse P. yoelii development in blood and liver stages using both in vivo and in vitro luminescent assays as has been achieved with P. berghei reporter parasites [17,18].

GIMO-transfection is a rapid and simple method for gene complementation
Gene complementation is used to prove that the phenotype of a gene deletion/modified parasite is the direct result of the gene mutation and not a consequence of an unintended alteration of the parasites genome [11]. Complementation is performed by reintroduction of a wild-type copy of the gene into the genome of a mutant in order to restore the wild-type phenotype, thereby establishing the association of the phenotype to the deletion genotype. We analysed whether GIMO-transfection can be used for gene complementation using a published gene deletion mutant of P. berghei with a defined phenotype. Complementation of a mutant using GIMO-transfection requires that the mutant contain the negative selectable marker yfcu in its genome. We therefore choose to complement a P. berghei mutant (Dgr) which lacks expression of glutathione reductase [19]. In this mutant, the glutathione reductase (gr) has been deleted using a construct containing the hdhfr::yfcu marker and the mutant becomes arrested in the mosquito during oocyst development with a complete absence of sporozoite production [19]. For complementation of the Dgr mutant we generated a restoration DNAconstruct by simply amplifying the gr gene from wild-type P. berghei genomic DNA and therefore avoided any cloning steps. Using the same primers that amplified the 59 and 39 targeting regions for the DNA construct used to generate the Dgr gene deletion mutant [19] (See Table S1), specifically the forward primer of 59 targeting region and reverse primer of 39 targeting region, a 2.8 kb PCR product that contained the complete gr gene and both targeting regions was amplified by a high fidelity proof reading polymerase (see Figure 5A). This PCR product was used to transfect Dgr parasites, with the aim to introduce the complete gr gene ('gene insertion') and thereby replacing the deleted gr locus, containing the hdhfr::yfcu ('marker out') as shown in Figure 5A. Selection of transfected parasites, using negative selection was as described above for other GIMO-transfections, and resulted in the selection of parasites (exp. 1761; Dgr(+gr)) in which the deleted gr had been replaced by the wild-type gr gene as confirmed by both diagnostic PCR and Southern analysis of digested genomic DNA ( Figure 5B). We next analysed the phenotype of the complemented Dgr(+gr) parasites by comparing oocyst and sporozoite development of Dgr(+gr) and Dgr parasites in Anopheles stephensi mosquitoes. As previously reported [19], Dgr produced oocysts that abort development resulting in small degenerated oocysts without any signs of sporoblast or sporozoite formation ( Figure 5C) at day 12 post infection (p.i.). The Dgr infected mosquitoes are not able to infect naive mice at day 21 p.i. In contrast, the complemented Dgr(+gr) have normal development in mosquitoes producing normal sized mature oocysts, which contain sporozoites at day 12 p.i. and salivary glands contained sporozoites at day 21 p.i. (Figure 5D). The Dgr(+gr) sporozoites are infectious as shown by injection of 10 4 salivary gland sporozoites in two naïve Swiss mice. Both mice developed a blood stage infection with a prepatency period of 5  Figure 2) after negative selection. The percentage of mCherry-positive parasites in six independent transfections (1794-1799) was determined by FACS analysis (see Figure 2D) and quantitative PCR (qPCR). By qPCR the ratio of mCherry and hdhfr::yfcu marker positive parasites was determined relative to the presence of a control gene hsp70, using the 2 2DDCT method (primers used in qPCR are described in Table S2). (B) Efficiency of selection of hdhfr::yfcu marker-free determined by Southern analysis of PFG-separated chromosomes. Hybridization performed using a mixture of two probes, one specific for pb25 (chromosome 5) and one for hdhfr (chromosome 3) showing the efficiency of selecting hdhfr::yfcu marker-free parasites in the different experiments. doi:10.1371/journal.pone.0029289.g003 days which is comparable to the prepatancy of mice infected with 10 4 wild type sporozoites. Genotype analysis of Dgr(+gr) blood stage parasites after mosquito passage and sporozoite infection, by diagnostic PCR and Southern analysis of digested genomic DNA, confirmed that gr was indeed restored (i.e. complemented) in the Dgr(+gr) parasites and no deletion mutants were present ( Figure 5B). The restoration of the phenotype of Dgr parasites using a PCR-amplified construct in combination with negative selection demonstrates that GIMO transfection is a fast method for gene complementation (see also the Discussion section). In addition it is a relatively simple method, requiring only PCRamplified DNA-constructs that can be used as the constructs do not require a drug-selectable marker cassette.

Discussion
Genetic modification of malaria parasites is limited by the paucity of drug-selection markers that permit selection of transformed mutants, which in turn hampers the generation of multiple genetic modifications in the same mutant. The novel GIMO-transfection method reported in this study permits the generation of mutants stably expressing heterologous proteins free of drug-selectable markers, facilitating further genetic modification of the transgenic parasites. In addition, it provides a fast and simple way for gene complementation of gene deletion/mutation mutants. We have generated reference mother lines and standard 'knock-in' constructs for both P. berghei ANKA and P. yoelii . Generation of a P. yoelii reporter line, PyGFP-luc con that is marker-free and expresses a fusion protein of GFP and luciferase. (A) Schematic representation of the introduction of a gfp-luciferase-expression cassette into the GIMO Py17X mother line. Construct pL1847 containing the eef1a-gfp::luciferase-39pbdhfr cassette is integrated into the modified P. yoelii 230p locus containing the hdhfr::yfcu selectable marker cassette (black box) by double cross-over homologous recombination at the target regions (hatched boxes). Negative selection with 5-FC results in selection of parasites that have the gfp-luciferase reporter introduced into the genome and the hdhfr::yfcu marker removed. Location of primers used for PCR analysis and sizes of PCR products are shown (see Table S2 for primer sequences). (B) Fluorescence microscopy of a live schizont of PyGFPluc con ; bright field (BF), DNA staining (Hoechst; Blue) and GFP expression (green). (C) PFG-separated chromosomal Southern analysis of two independent GIMO transfection parasite lines (exp. 1970 and 1971). Hybridization performed with a mixture of two probes, one specific for pb25  17XNL, which we will make available for the research community. In GIMO-transfection of these mother lines, transgenes are introduced in the 230p locus of both P. berghei and P. yoelii. For P. berghei ANKA it has been shown that 230p is a 'silent' locus [20] and different reporter lines with transgenes introduced in this locus has been generated that show wild-type progression through the complete life-cycle [8,21]. Whether 230p is also a 'silent' locus in P. yoelii has not been reported before. Our observations of normal development of asexual stages, mosquito development and sporozoite infectivity of the P. yoelii mother line  Table S2. Southern blot was hybridized with 39UTR gr probe (i.e. 39 targeting region). The localization of the restriction enzymes used and the expected size of the fragments are shown in A: wt (wild type); Dgr (gr deletion mutants); Dgr(+gr) (complemented Dgr); mp (blood stages after mosquito passage). (C) Oocyst development of Dgr and Dgr (+gr) parasites. Only small, aberrant oocysts with no signs of sporozoite formation are present in Dgr infected mosquitoes at days 10-21 after feeding. In Dgr (+gr) infected mosquitoes sporozoite-containing oocysts with wild-type morphology are visible at day 12. (E) Salivary gland sporozoites of Dgr(+gr) examined by immuno-fluorescence microscopy: bright field (BF) and anti-CS antibody staining (CS, green). doi:10.1371/journal.pone.0029289.g005 and PyGFP-luc con indicates that p230 is also a suitable locus to introduce transgenes in P. yoelii.
Several P. berghei reference lines exist that express reporter proteins, such as GFP and luciferase, and do not contain drugselection markers. Most of these parasites have been obtained by FACS-sorting where GFP expression is used as the selectable marker [8,9]. However, selection of transgenic fluorescentexpressing parasites by FACS-sorting has been only reported for selecting GFP-expressing parasites and not with parasites that express other fluorescent proteins. In our hands, FACS-sorting of GFP-expressing parasites is not a highly efficient selection method as often the selected population consists of both mutant and wild type parasites. Moreover, introducing a GFP-selection cassette increases the size of the transfection construct. This limits the size of the heterologous DNA that can be cloned into these vectors as it is difficult to maintain Plasmodium transfection vectors with a size larger than 14 kb in E. coli. Therefore, in comparison with FACSsorting, the GIMO-transfection system is a more flexible and simpler system to introduce a wide range of heterologous genes into the parasite genome with the additional advantage that GIMO transfection constructs are far smaller since a selectionmarker cassette is not required.
In addition to the use of FACS-sorting for the generation of marker-free P. berghei mutants a 'marker-recycling' method has also been employed in P. berghei [10]. Specifically, transformed parasites expressing the fusion gene hdhfr::yfcu are first selected by positive selection with pyrimethamine; subsequently negative selection with 5-FC is applied to select parasites that have lost the resistance genes. The efficiency of selection of marker-free parasites is dependent on the frequency of the loss of the hdhfr::yfcu marker from the genome by homologous recombination and excision [10]. This method has been successfully used to generate marker-free reporter lines [22], to introduce two independent genetic modifications in the same parasite lines [22][23][24] and for complementation [10]. However, this marker-recycling method is relatively laborious and time consuming since it involves both positive and negative selection procedures and two parasitecloning steps, a procedure requiring at least 9 weeks to complete. Further, marker-recycling method requires at least 24 mice in order to obtain a marker-free mutant ( Figure 6A), in part a consequence of essential cloning procedures [19,22]. In contrast, the generation of marker-free mutants with GIMO-transfection can be achieved in only 4 weeks and requires only 11 mice ( Figure 6A). The marker-recycling transfection constructs consist of the hdhfr::yfcu drug-selectable marker cassette, a transgene expression and two targeting sequences for integration into the genome (See Figure S2A). In addition, they have two identical regions of DNA sequence that can recombine (in the parasite genome) and excise the selectable marker cassette. In contrast the GIMO-constructs contain only the two genome targeting sequences and the transgene expression cassette (see Figure S2 for a comparison of the marker-recycling and GIMO constructs). The simple structure of GIMO constructs permits the cloning of larger transgenes (the GIMO constructs are smaller as the selectable marker cassette is absent) and improves the retention of plasmids in bacteria as internally repetitive regions of AT-rich Plasmodium DNA are absent. Further, after transfection with the GIMO construct, the selection of integration mutants is improved as no episomal construct DNA is maintained in the parasites and negative selection kills parasites expressing yfcu.
GIMO-transfection is dependent on the transgene-expression construct replacing the hdhfr::yfcu selection cassette present in the mother line genome and the efficiency of the drug 5-FC to kill all parasites where this integration has not occurred and that are still expressing yFCU. Interestingly, in both P. berghei and P. yoelii GIMO-transfection experiments we always observed that populations of 5-FC selected parasites contain (low numbers of) parasites that still have the hdhfr::yfcu selection cassette in their genome. Further research is required to determine whether these parasites express yFCU but are able to survive 5-FC drug treatment or if these parasites have lost expression of yFCU through the mutation of hdhfr::yfcu selectable marker cassette. Experiments in our laboratory are now focused on improving the application of negative selection to mutant parasites in mice by providing 5-FC in the drinking water, which may permit treatment with higher concentrations of 5-FC and for longer periods. Notwithstanding the presence of non-transformed parasites after selection of GIMO-transfected parasites, the high percentages of transformed parasites in the populations permit the collection of the desired mutants by cloning. Using GIMO-transfection we have already been able to successfully generate multiple marker-free lines that express a variety of heterologous proteins (unpublished data JWL and SK).
GIMO-transfection was used to generate a P. yoelii GFPluciferase reporter parasite and is the first report describing the use of negative selection with 5-FC in combination with the yFCU marker for genetic modification of this parasite species. Moreover, the PyGFP-luc con line is the first P. yoelii reporter line that is marker-free and can be easily further genetically modified. Similar P. berghei reporter lines have been used to visualize and quantify host parasite interactions in vivo [13,21,25,26], analysis of drugsusceptibility [17,27,28] and in vivo quantification of liver stage development [18,29].
In this study we demonstrate that GIMO-transfection can not only be used to introduce heterologous genes but also is a fast and simple method for gene complementation. Restoration of the wild type phenotype by gene complementation is the most optimal strategy to show that a mutant phenotype is the result of the intended deletion (or mutation) and is not due to unrelated alterations in the parasite genome [2,11]. Genetic complementation has not been widely applied in Plasmodium due to difficulties in making successive genetic modifications in the same parasite, and to problems inherent in cloning full-length AT-rich Plasmodium genes into bacterial plasmid vectors [11]. Till now two methods have been used to complement gene deletion mutants in P. berghei. The first method re-introduces the wild-type gene using a construct containing hdhfr as a positive selectable marker [30,31]. The encoded protein confers resistance to WR99210, and can be used to transfect gene deletion mutants that already contain the pyrimethamine resistance markers dhfr/ts from P. berghei or the dhfr from Toxoplasma gondii (tgdhfr) [7]. However, selection with WR99210 is not straightforward because of problems with dissolving this drug and because there is a reduced sensitivity to WR99210 of parasites that already contain the dhfr/ts or tgdhfr marker [7,10] (unpublished observations CJJ). The second complementation method is based on the marker-recycling, as described above. Gene deletion mutants (containing hdhfr::yfcu) are first subjected to negative selection to select for marker-free parasites, cloned and then transfection is performed with constructs containing the gene for complementation and a drug selection cassette [10] (see Figure S3A). This method requires generally 7 weeks and 14 mice to perform ( Figure 6B). In contrast, complementation with GIMO-transfection takes only 2 weeks and 1 mouse ( Figure 6B). Not only is the GIMO method much faster, requiring far fewer mice, but also a big advantage is that a simple PCR amplicon containing the wild-type gene can be used for complementation as no drug selectable needs to be used in the construct (see Figure S3 for schematics of the marker-recycling and GIMO methods).
In summary, we have developed a novel method that simplifies and speeds up both the generation of marker-free parasites expressing heterologous proteins and for the genetic complementation of gene deletion/mutation mutants. Moreover the application of this method greatly reduces the numbers of animals required to generate and complement mutants. We have also generated the first marker-free P. yoelii reporter line and established the successful use of negative selection in transfection of P. yoelii parasites. The GIMO-transfection is a simple, fast and efficient approach to generate mutants permissive to subsequent genetic modification. Therefore we recommend that, where possible, transfection of P. berghei and P. yoelii parasites be performed with constructs that contain the postive-negative selectable marker cassette, hdhfr::yfcu. The presence of this marker in mutants permits subsequent GIMO transfection that not only simplifies the creation of additional deletions or modifications but also gene complementation experiments. A recent study has reported high-throughput, genome wide and highly efficient 'recombineering' system, for high-throughput, genome wide and highly efficient generation of gene targeting constructs [12]. This exciting development can be partnered with GIMO transfection by ensuring all these targeting constructs have a positive-negative (hdhfr::yfcu) selectable marker cassette. Consequently all resulting mutants would be receptive to GIMO transfection thereby permitting further modification (e.g. reporter protein expression) and complementation.

Experimental animals and parasites
Female Swiss OF1 mice (6-8 weeks old; Charles River/Janvier) were used.
All animal experiments of this study were approved by the Animal Experiments Committee of the Leiden University Medical Center (DEC 07171; DEC 10099). The Dutch Experiments on Animal Act is established under European guidelines (EU directive no. 86/609/EEC regarding the Protection of Animals used for Experimental and Other Scientific Purposes).
Generation of GIMO mother lines in P. berghei ANKA and P. yoelii 17XNL To generate the GIMO mother line in P. berghei, a DNAconstruct pL1603 was generated for integration into the 230p gene (PBANKA_030600) by cloning the 59 and 39 regions of 230p as previously described [8]. The targeting sequences were amplified from genomic DNA using primer sets 5585/5586 and 5587/5588 (See Table S1 for the sequence of all primers) and cloned into the restriction sites of HindIII/KspI and Asp718I/EcoRI of the standard cloning vector pL0034 (MRA-849, www.mr4.org), which contains the hdhfr::yfcu selectable marker under the control of the eef1a promoter [10]. The hdhfr::yfcu marker is a fusion gene of the positive selection marker human dihydrofolate reductase and the negative selection marker which is a fusion gene of yeast cytosine deaminase and uridyl phosphoribosyl transferase [10]. Prior to transfection the DNA-construct pL1603 was linearized with HindIII and EcoRI.
To generate the GIMO mother line in P. yoelii, a modified two step PCR method [33] was used to generate DNA-construct pL1805 for integration into the 230p gene (PY03857) of P. yoelii ( Figure S1A). In the first PCR reaction two fragments (59-and 39targeting sequences, both ,1 kb) of 230p were amplified from P. yoelii 17XNL genomic DNA with the primer sets 6523/6524 and 6525/6526 (Table S1). Primers 6524 and 6525 have 59-extensions homologues to the hdhfr::yfcu selectable marker cassette (CATC-TACAAGCATCGTCGACCTC in 6524 and CCTTCAA-fTTTCGGATCCACTAG in 6525). This selectable marker cassette was excised by digestion with XhoI and NotI from a plasmid (pL0048) that contains the P. berghei eef1a-hdfhr::yfcu-39dhfr/ts (i.e. promoter-drug selectable marker-39 terminator sequence) selection cassette. Primers 6523 and 6526 have 59terminal extensions with an anchor-tag suitable for the second PCR reaction. In the second PCR reaction, the amplified 59-and 39-targeting sequences were annealed to either side of the selectable marker cassette, and the joint fragment was amplified by the external anchor-tag primers 4661/4662, resulting in the PCRbased targeting construct with an expected size of 4.7 kb (2.7 kb of the selectable marker cassette plus two targeting sequences of 1 kb). Before transfection, the PCR-based construct was digested with Asp718I and ScaI (in primers 6523 and 6526, respectively) to remove the 'anchor-tag' and with DpnI that digests any residual pL0048 plasmid. Transfection in P. berghei ANKA and P. yoelii 17XNL, selection and cloning of the mother lines were performed by standard procedures described for transfection of P. berghei [14]. DNAconstruct pL1603 was introduced into P. berghei generating mother line, GIMO PbANKA (1596cl1), and DNA construct pL1805 was introduced into P. yoelii generating mother line, GIMO Py17X (1923cl1). Correct integration of the constructs was verified by diagnostic PCR analysis (see Table S2 for primers used) and Southern blot analysis of pulse-field gel (PFG) electrophoresisseparated chromosomes probed with the 39 untranslated region (UTR) of the dhfr/ts gene of P. berghei.
Generation of basic constructs without selection marker and that target the 230p locus of the GIMO PbANKA and GIMO Py17X mother lines To generate a basic P. berghei 230p-targeting construct (pL0043), the 230p targeting regions as well as the ampicillin resistance gene were amplified from plasmid pL1063 (MRA-852, www.mr4.org) using primers 5116/5117 (Table S1). A multiple cloning site (MCS) was amplified from pCRII-Blunt-TOPO vector (Zero Blunt TOPO PCR Cloning Kit, Invitrogen, Groningen, The Netherlands) using M13 forward and reverse primers. The two PCR products were digested with Asp718I and NotI restriction enzymes and ligated together creating the targeting construct pL0043.
A basic P. yoelii 230p-targeting construct (pL1849) was generated using a modified 2-step PCR method ( Figure S1B). In the first PCR reaction, 59-and 39-targeting sequences (both ,1 kb) of 230p were amplified from P. yoelii 17XNL genomic DNA with the primer set 6523/6534 and 6525/6526 (Table S1). As described above these primers contain 59-extensions homologues to the hdhfr::yfcu selectable marker cassette and 59-terminal extensions with an anchor-tag suitable for the second PCR reaction. A 55 nt oligo (oligo 6598; GAGGTCGACGATGCTTGTAGATGC-CCGGGCCTTCAATTTCGGATCCACTAG) containing a XmaI restriction site flanked by 2 sequences homologues to the hdhfr::yfcu selectable marker cassette was used to join the two 230p targeting regions ( Figure S1B). In the second PCR reaction an fragment containing both 230p targeting sequences interrupted by the XmaI site was amplified, using the external anchor-tag primers 4661/4662, resulting in the PCR product of ,2 kb. The PCR product was cloned into TOPO TA vector (TOPO TA CloningH Kit, Invitrogen, Groningen, The Netherlands) resulting in construct pL1849.
Generation of a mCherry reporter test construct and GIMO-transfection in the P. berghei mother line, GIMO PbANKA A test construct (pL1628) for GIMO-transfection in the GIMO PbANKA mother line was generated by transferring the mCherry-expression cassette (59pbeef1a-mCherry-39pbdhfr) from plasmid pL0017-mCherry [34] into the basic 230p targeting construct pL0043 (see above) using restriction sites EcoRV/Asp718I. This plasmid was linearized with KspI before transfection. Transfection was performed as described [14]. Transformed parasites were selected by negative selection by the administration the drug 5-FC (Sigma) to mice infected with transfected parasites. Specifically; 0.4 g/kg bodyweight of 5-FC (stock: 20 mg/ml in 16PBS) administered by intra-peritoneal injection; one dose per day; for a period of 4 days, starting at 24 hours after transfection. Transformed parasites were collected at day 6/7 (infected tail blood) for phenotype analysis by fluorescence microscopy and FACS (see below) and at day 7/8 (infected heart blood) for genotype analysis using standard methods of diagnostic PCR and Southern analysis of PFG-separated chromosomes [14].
Transfection of GIMO Py17X parasites and negative selection of transformed parasites was performed as described above for transfection of GIMO PbANKA . Transformed parasites were collected for genotype analyses using standard methods of diagnostic PCR and Southern analysis of PFG-separated chromosomes [14]. Cloned parasites were analysed for luciferase expression using the in vivo imaging technology described below.

Gene complementation using GIMO transfection
Gene complementation was performed using the published glutathione reductase deletion mutant (Dgr) of P. berghei [19]. In this mutant (Dgr4; 1531cl1) the glutathione reductase (gr) gene has been deleted by a replacement construct (pL1538) that contains the postive-negative hdhfr::yfcu selectable marker cassette [19]. The pL1538 construct contains 59 and 39 targeting regions of gr. We used two of the primers that have been used to generate the replacement construct pL1538 to amplify gr gene from P. berghei genomic DNA using a proof reading polymerase (PhusionH, Finnzymes, Espoo, Finland). These primers (4049; forward primer for 59 targeting region and 3681; reverse primer for 39 targeting region) amplify the complete gr gene including the 59 and 39 targeting regions (see Table S1 for primer sequences). PCR resulted in amplification of a 2.8 kb fragment which was used to transfect Dgr parasites using standard transfection procedures [14]. Transformed parasites were selected by negative selection as describe above. Transformed parasites were collected for genotype analyses using standard methods of diagnostic PCR and Southern analysis of digested genomic DNA. Analysis of the phenotype of the complemented parasites, Dgr(+gr), was analysed by mosquito transmission experiments (see below).

Fluorescence microscopy and FACS analysis
For analysis of GFP-or mCherry-expression in blood stages of transgenic parasites, infected tail blood was collected in PBS and examined by microscopy using a Leica DMR fluorescent microscope with standard GFP and Texas Red filters. Parasites nuclei were labeled by staining with Hoechst-33258 (2 mmol/L, Sigma, NL). Images were recorded with the digital camera CoolSNAP HQ 2 (Photometrics, NL) and processed with the ColourProc software [35]. The percentage of blood stages parasites that express mCherry was determined by FACS analysis of cultured blood stages. In brief, infected tail blood (10 mL) with a parastemia between 0.5 and 1% was cultured overnight in 1 mL complete RPMI1640 culture medium at 37uC under standard conditions for the culture of P. berghei blood stages [36]. Cultured blood samples were then collected and stained with Hoechst-33258 (2 mmol/L, Sigma, NL) for 1 hr at 37uC in the dark and analysed using a FACScan (BD LSR II, Becton Dickinson, CA, USA) with filter 440/40 for Hoechst signals and filter 610/20 for mCherry fluorescence. For FACS analysis the population of mature schizonts were selected based on the their Hoechstfluorescence intensity [37]; see gate P2 in the left panel of Figure 2D. The percentage of mCherry-expressing parasites was calculated by dividing the number of mCherry-positive schizonts (gate P3 in right panel of Figure 2D) by the total number of schizonts (gate P2).
Quantitative real-time PCR (qPCR) analysis of transformed parasites Genomic DNA extracted from blood stage parasites was used for qPCR analysis. To determine the ratio of transformed/nontransformed parasites in the selected parasite populations, PCR amplifications of the mCherry gene (only present in transformed parasites) and the hdhfr::yfcu selectable marker (only present in non-transformed) were carried out using the QuantiTect SYBR Green PCR Kit (Qiagen, Hilden, Germany) on a CFX96 thermal cycler (Bio-Rad Laboratories, The Netherlands). The housekeeping gene, P. berghei hsp70, was used as reference (see Table S2 for primers used). Real-time PCR cycle thresholds (C T ) were calculated as the average of triplicate analyses (per genomic DNA from transgenic parasite). The ratio between mCherry and hdhfr::yfcu was calculated by the 2 2DDCT method relative to hsp70 [16]. The amplification efficiencies of mCherry and hsp70 did not violate assumptions of the DDC T method (data not shown).
Real time in vivo imaging of the PyGFP-luc con reporter parasites in whole bodies of live mice Expression of luciferase and imaging of distribution of luciferase-expressing PyGFP-luc con parasites in whole bodies of live mice was determined by measuring bioluminescent activity using the IVIS100 in vivo imaging system (Caliper Life Sciences, USA) as described previously [21,38]. Bioluminescence of blood stage parasites was imaged in Swiss mice with asynchronous infections of PyGFP-luc con parasites at a parasitemia of 0.5-2%.
Analysis of the phenotype of Dgr and complemented Dgr(+gr) parasites during mosquito transmission Infection of Anopheles stephensi mosquitoes with Dgr and Dgr(+gr) parasites as well as determination of production of oocysts and salivary gland sporozoites was performed as previously described [39]. Infectivity of sporozoites was tested by intravenous injection of Swiss OF1 mice with 10 4 hand dissected salivary gland sporozoites. The prepatent period was determined by light microscopy analysis of Giemsa-stained thin smears of tail blood. Prepatency (measured in days after sporozoite inoculation) is defined as the day when parasitemia reaches 0.5-2%.
Indirect Immunofluorescence assay 10 4 Dgr(+gr) salivary gland sporozoites in 10 mL were allowed to adhere to polylysine coating slides, fixed for 15 minutes with 4% PFA, and washed 365 minutes with PBS. Sporozoites were then permeabilized with 0.5% Triton-X100 for 15 minutes followed by a 365 minutes wash with PBS. Slides were blocked 30 minutes at room temperature in 10% FCS and incubated over night with polyclonal rabbit anti-CS antiserum [40] (dilution 1:1000, kindly provided by Dr M. Yuda) at 4uC. Slides were washed 365 minutes in PBS and incubated with donkey anti-rabbit, Alexa 488-conjugated secondary antibody (dilution 1:500), 1 hr in room temperature. Slides were washed 365 minutes in PBS, and then incubated 15 minutes with Hoechst 33342 in room temperature. Prior to mounting, slides were washed for 5 minutes and analysed with were analyzed using a Leica DMR fluorescence microscope at 10006 magnification. Figure S1 Generation of P. yoelii 230p targeting constructs using a PCR method. A. The DNA construct (pL1805) used to generate the P. yoelii GIMO mother line was created using a modified two-step PCR method. In the first PCR reaction, 59-and 39-targeting sequences of 230p were amplified from P. yoelii 17XNL genomic DNA with the primer sets 6523/ 6524 and 6525/6526 (Table S1). Primers 6524 and 6525 have 59extensions homologues to the hdhfr::yfcu selectable marker cassette (hatched boxes). This selectable marker cassette was excised from plasmid pL0048 digested with XhoI and NotI. Primers 6523 and 6526 have 59-terminal extensions (black boxes) for the second PCR reaction. In the second PCR reaction, the 59-and 39targeting sequences annealed to either side of the selectable marker cassette, and the joint fragment was amplified by the external anchor-tag primers 4661/4662. Before transfection, the PCR construct was digested with Asp718I and ScaI to remove the anchor-tag and with DpnI to digest any residual pL0048 plasmid. B. The basic P. yoelii 230p targeting construct (pL1849) was generated by a modified PCR method. In the first PCR reaction, 59-and 39-targeting sequences with homologous sequences (hatched boxes) and anchor-tag sequences (black boxes) were amplified as shown in A. Oligo no. 6598 that contains the joint homologous sequences interrupted by an XmaI site (hatched boxes) was used as template for the second PCR reaction. Using the external anchor-tag primers 4661/4662, a PCR product containing both targeting sequences now with the XmaI site in the middle was amplified and subsequently cloned into TOPO TA vector resulting in construct pL1849. (EPS) Figure S2 Schematic representation of the generation of marker-free gene insertion (GI) mutants using GIMOtransfection method or using the marker-recycling method. A. Generation of marker-free gene insertion mutants expressing a gene of interest (GOI; grey box) using the standard marker-recycling method. The construct containing the hdhfr::yfcu selectable maker (black box) flanked by the recombination sequences (rc, shaded boxes) targets the 230p locus by double cross-over homologous recombination at specific target regions (hatched boxes). GI mutants are obtained after transfection, using positive selection with pyrimethamine and then cloning. Subsequently, marker-free GI mutants are selected by negative selection using 5-FC. Only those mutants that have 'spontaneously' lost the hdhfr::yfcu marker from their genome, achieved by a homologous recombination/excision (see arrow), survive negative selection. B. Generation of marker-free gene mutants that express a GOI (grey box) using GIMO-transfection. The construct that contains no selectable marker cassette and targets the modified GIMO mother line 230p locus that contains the hdhfr::yfcu (black box) marker, by double cross-over homologous recombination at the target regions (hatched boxes). Marker-free GI mutants, that have GOI expression cassette introduced into the 230p locus replacing the hdhfr::yfcu marker, are obtained by negative selection with 5-FC. (EPS) Figure S3 Schematic representation of gene complementation using GIMO-transfection and the markerrecycling method. A. Gene deletion and complementation using the marker-recycling method. The gene deletion construct, containing the hdhfr::yfcu selectable maker (black box) flanked by the recombination sequences (rc; shaded boxes), targets the gene of interest (GOI) by double cross-over homologous recombination at the target regions (hatched boxes). Gene deletion mutants are obtained after transfection and positive selection with pyrimethamine, and cloning. Subsequently, marker-free gene deletion mutants are selected by negative selection using 5-FC. Only those mutants that have 'spontaneously' lost the hdhfr::yfcu marker from their genome, achieved by a homologous recombination/excision event (see arrow), survive negative selection.Complementation of the (cloned) marker-free gene deletion mutant is performed using constructs that contain a GOI expression cassette and a positive selectable marker cassette. These constructs can target either the original deleted locus or a locus that is redundant or functionally silent. Complemented parasites are selected by positive selection. B. Gene deletion and complementation using the GIMOtransfection method. The gene deletion construct containing the hdhfr::yfcu selectable maker fusion (black box) targets the GOI by double cross-over homologous recombination at specific target regions (hatched boxes). Gene deletion mutants are obtained after transfection using positive selection with pyrimethamine and then cloning. These constructs do not include recombination (rc) sequences (see A). Complementation of the gene deletion mutant is performed using a PCR fragment amplified from genomic DNA using the same outer primers used to generate the gene deletion construct (i.e. the forward primer of the 59UTR and the reverse primer of 39UTR, indicated by arrows). Integration of the PCR fragment by homologous recombination restores the deleted gene locus replacing the hdhfr::yfcu maker. Complemented parasites are selected by negative selection. (EPS)