|Home | About | Journals | Submit | Contact Us | Français|
Here we report an efficient, site-specific system of genetic integration into Plasmodium falciparum malaria parasite chromosomes. This is mediated by mycobacteriophage Bxb1 integrase, which catalyzes recombination between an incoming attP and a chromosomal attB site. We developed P. falciparum lines with the attB site integrated into the glutaredoxin-like cg6 gene. Transfection of these attB+ lines with a dual-plasmid system, expressing a transgene on an attP-containing plasmid together with a drug resistance gene and the integrase on a separate plasmid, produced recombinant parasites within 2 to 4 weeks that were genetically uniform for single-copy plasmid integration. Integrase-mediated recombination resulted in proper targeting of parasite proteins to intra-erythrocytic compartments, including the apicoplast, a plastid-like organelle. Recombinant attB × attP parasites were genetically stable in the absence of drug and were phenotypically homogeneous. This system can be exploited for rapid genetic integration and complementation analyses at any stage of the P. falciparum life cycle, and it illustrates the utility of Bxb1-based integrative recombination for genetic studies of intracellular eukaryotic organisms.
In response to the worsening burden of Plasmodium falciparum malaria, intense research efforts have led to the development of various tools for the genetic manipulation of this organism. The application of techniques such as gene disruption and allelic replacement has afforded important insights into parasite mechanisms of drug resistance, pathogenesis, host cell invasion and transmission1. Yet genomic integration studies are plagued by low transfection and recombination efficiencies. In addition, transgene expression assays have been restricted to episomally replicating plasmids that parse unequally during mitotic segregation and hence cannot be selected in sexual stages and propagated through the life cycle1.
Integrative recombination offers a powerful approach to target foreign DNA into specific genomic locations. Bacteriophage serine integrase–based systems are particularly suitable because of their intrinsically high efficiency of recombination between host attB and phage attP sites, with integration and excision having distinct reaction requirements that preclude spontaneous reversion events2. These recombinases, unlike their tyrosine counterparts such as lambda integrase (used to develop the Invitrogen GATEWAY site-specific recombination system), are sufficient to catalyze recombination and do not require bacterial host factors. Among the serine integrases, the most extensively used has been Streptomyces sp. phage phiC31, which efficiently integrates plasmid vectors into prokaryotic or eukaryotic hosts containing att or pseudo-att sites2,3.
The serine integrase of the mycobacteriophage Bxb1, which is a temperate phage whose genome recombines into an attB site located within the Mycobacterium smegmatis groEL1 gene4, has been used to catalyze recombination in M. smegmatis, in Escherichia coli5,6 and in vitro. Recombination occurs between short attB and attP substrates, each of which contain regions of inverted symmetry flanking a conserved 5′ GT dinucleotide, whose nonpalindromic nature determines the polarity of integration. Both Bxb1 and phiC31 integrases mediate integrative recombination without a requirement for divalent cations, cofactors or DNA supercoiling, making them ideally suited for targeted gene delivery2,5.
In this study, we describe the development of a new means of genetic integration into P. falciparum. In this system, transient expression of the mycobacteriophage Bxb1 integrase mediates recombination between an attP–bearing plasmid and an attB site integrated into the P. falciparum genome. Using a dual-plasmid approach, we achieved rapid site-specific integration of single-copy plasmids that properly targeted proteins to intracellular parasite compartments. Integrase-mediated recombination produced genetically and phenotypically homogeneous populations of transgenic parasites within 1 month, and these remained stable in the absence of drug.
We introduced the attB site (from M. smegmatis) into the P. falciparum cg6 gene7, the disruption of which is not detrimental for asexual blood stage growth (D.F., unpublished data). To achieve this, we transfected Dd2 and 3D7 parasite strains with a plasmid (pCG6-attB) containing a 40-bp attB sequence, a truncated cg6 insert and the human dihydrofolate reductase (DHFR) selectable marker (Fig. 1a and Supplementary Fig. 1 online). We selected transfected parasites with WR99210 (a specific inhibitor of P. falciparum DHFR) and used PCR to screen for plasmid recombination into the cg6 locus. This led to the identification of lines that had integrated the attB site downstream of a disrupted cg6 lacking its 3′ coding sequence and 3′ untranslated region (UTR). We obtained attB+ clones, referred to asDd2attB and 3D7attB, by limiting dilution (Table 1). Southern blots confirmed insertion of a single attB site into the cg6 locus in these Dd2attB and 3D7attB clones (Fig. 1b,c). To generate the controlDd2C1 clone, we transfectedDd2 with the pCG6 plasmid (without attB) that also integrated into cg6. This clone served to test the site specificity and relative efficiency of Bxb1 integrase–mediated recombination. We also generated clones in Dd2 and 3D7 (denoted Dd2attB/hrp3 and 3D7attB/hrp3) that had undergone pCG6-attB integration into the hrp3 5′ UTR used to drive the human DHFR selectable marker, as confirmed by Southern blot analysis (Table 1 and Supplementary Fig. 2 online).
To test whether mycobacteriophage Bxb1 integrase could catalyze recombination between an attP plasmid and an attB site resident in the P. falciparum genome, we transfected Dd2attB and 3D7attB clones with a plasmid (pBSD-GFP-INT-attP) containing a 50-bp attP element, the integrase gene and a blasticidin S-deaminase (bsd) selectable marker fused with GFP (Fig. 2a and Supplementary Fig. 1). This produced the lines Dd2attB/BSD-GFP and 3D7attB/BSD-GFP (Table 1). We also electroporated pBSD-GFP-INT-attP into the Dd2C1 clone lacking attB, to produce the control line Dd2C1/BSD-GFP. Parasites appeared in the Dd2attB/BSD-GFP and 3D7attB/BSD-GFP lines within 9–14 d of dual drug selection with WR99210 and blasticidin (BSD), and after a few days they attained a growth rate comparable to nontransfected parasites. In contrast, we first observed parasites in the Dd2C1/BSD-GFP control line on day 25, and the parasitemia remained low until 36 d post-electroporation, when the growth rate became typical of episomally transfected lines. These observations provided initial indications that attB × attP recombination had markedly increased the transfection efficiency and decreased the time required for the appearance of transfected parasites.
PCR screening for Bxb1-specific attB × attP recombination junctions (known as attL and attR) showed integration of the attP plasmid into the chromosomal cg6-attB site within 14 d in both the Dd2attB/BSD-GFP and 3D7attB/BSD-GFP lines (Fig. 2b). Recombination occurred via a directional single-site crossover, resulting in insertion of the entire plasmid containing the integrase and bsd-GFP expression cassettes. We also transfected Dd2attB and 3D7attB parasites with either one plasmid containing a single attP element or an equal mixture of plasmids containing one to three attP elements. This suggested preferential integration of plasmids harboring two to three attP sites (Fig. 2b). Based on these results, we subsequently used plasmids bearing two tandem attP sites.
Southern blot analysis of the Dd2attB/BSD-GFP and 3D7attB/BSD-GFP lines confirmed integration of a single (non-concatameric) plasmid into the attB site. In these lines, the absence of episomal plasmids was indicative of a parasite population that had uniformly undergone integrative recombination (Fig. 2c). This key finding obviated the need for cloning to obtain a genetically homogeneous population of integrants. Sequence analysis of recombination junctions showed a precise crossover between attB and attP sites through the central dinucleotide (5′-GT) within the 8-bp common core sequences (Fig. 2d), as observed with Bxb1 integration into M. smegmatis5. Fluorescence microscopy of live Dd2attB/BSD-GFP parasites showed strong GFP expression in the cytoplasm (Fig. 2e), indicating that the bsd-GFP fusion was functional both as a selectable marker and as a reporter construct. Continual expression of integrase had no adverse effect on the growth or development of recombinant parasites and did not lead to attP plasmid excision from the integrated locus.
Results with the pBSD-GFP-INT-attP plasmid (Fig. 2a) showed that Bxb1 integrase could efficiently and rapidly catalyze site-specific recombination in P. falciparum. However, the integrase and bsd expression cassettes left little room to introduce a third cassette without compromising plasmid stability in E. coli. We therefore developed a dual-plasmid cotransfection system that expressed integrase from a separate plasmid (pINT, which used the neomycin selectable marker that confers G418 resistance8,9; Supplementary Fig. 1), thereby enabling larger transgenes to be introduced into a more compact attP plasmid. With this strategy, only the transgene expression cassette and bsd selectable marker are permanently incorporated into the parasite genome.
To test this system, we used the pfenr (P. falciparum enoyl acyl carrier protein (ACP) reductase) gene, the product of which has been predicted to be transported to the apicoplast and participate in de novo fatty acid type II biosynthesis10. We inserted a pfenr-GFP fusion into an attP expression plasmid to form pLN-ENR-GFP (Fig. 3a and Supplementary Fig. 1).We cotransfected two separate Dd2attB cultures with pLN-ENR-GFP and pINT and placed them under triple drug selection (WR99210, BSD and G418), generating two Dd2attB/ENR-GFP lines (Table 1). These showed evidence of plasmid integration into the attB site within 18 d of electroporation, as shown by PCR and sequencing (data not shown). Upon detection of integration, we removed G418 selection pressure to cure these lines of the integrase plasmid pINT.
Southern blot analysis of genomic DNA from both Dd2attB/ENR-GFP lines harvested on day 30, without prior cloning, showed pLN-ENR-GFP integration into the chromosomal attB site (Fig. 3b,c). Hybridization did not uncover any signs of episomes in either line within 1 month of electroporation, consistent with a predominant population of recombinant parasites. We did not observe any integration into the endogenous pfenr locus (Fig. 3c). Western blot analysis with GFP-specific monoclonal antibodies detected a 68-kDa protein in Dd2attB/ENR-GFP parasites, as predicted (Fig. 3d). Immunofluorescence images of Dd2attB/ENR-GFP parasites colocalized PfENR-GFP with the apicoplast protein ACP (Fig. 3e,f) but not with the mitochondrial marker MitoTracker Red (MTR; Fig. 3g,h). Thus, the Bxb1 integrase–mediated recombination system permitted precise targeting of a parasite protein to its host organelle.
Earlier studies have shown that Bxb1 integrase–mediated attB × attP recombination produces asymmetric attL and attR sites that cannot recombine without an exogenous excision factor5. However, the final recombinant locus generated in the attB × attP parasites contained duplicate cg6 sequences that could potentially recombine to excise the entire segment of exogenous DNA and restore the original cg6 locus (Fig. 1a). To test the stability of the attB × attP locus, we propagated Dd2attB/ENR-GFP parasites in the absence of drug selection, or with either BSD or WR99210 only. As a control experiment, we cultured an episomally transfected Dd2 line (referred to as Dd2/ENR-GFP and replicating the pLN-ENR-GFP plasmid; Table 1) without BSD and assayed it for the rate of plasmid loss. cg6 probe hybridization of DNA from Dd2attB/ENRGFP parasites cultured without drug pressure for 23 d showed no change in the organization of the recombinant locus, with no evidence of DNA rearrangement (Supplementary Fig. 3 online). In contrast, the episomal pLN-ENR-GFP plasmid was rapidly lost from Dd2/ENR-GFP parasites cultured without drug selection, as shown by cg6 probe hybridization (Supplementary Fig. 3), consistent with previous observations11. After 3–4 weeks of non-selection, we placed parasite cultures back under drug pressure. Whereas the integrant Dd2attB/ENR-GFP line did not show any growth delay, the episomal Dd2/ENR-GFP line was cleared of living parasites within 3 d of reexposure to BSD (indicating loss of the plasmid-borne selectable marker). The above data demonstrate an intrinsic stability of the integrated transgene as compared with its episomal form in the absence of drug pressure.
We have frequently observed that episomally transfected cells express fluorescent proteins at varying intensities and that there can be a substantial subpopulation of cells that lack fluorescence altogether (data not shown, and ref. 11). We postulated that integrase-mediated insertion of a single-copy GFP gene into the parasite genome would result in a more homogeneous pattern of GFP fluorescence. Accordingly, we generated lines in which an hrp3 5′ UTR–driven GFP transgene was expressed either episomally (the Dd2/GFP line) or from an integrated attB site (the Dd2attB/GFP line), as confirmed by PCR analyses and plasmid rescue (Table 1). We analyzed these lines by flow cytometry to determine their GFP expression levels. We incubated cells with Hoechst 33342 and gated them for nuclear staining to separate the infected from uninfected erythrocytes (Fig. 4a). Hoechst staining of nontransfected Dd2, episomally transfected Dd2/GFP and integrated Dd2attB/GFP parasite lines showed very similar profiles that corresponded to predominantly singly infected, ring-stage parasites12 and were distinct from uninfected red blood cells (RBCs; Fig. 4b). GFP expression profiles showed a noticeably higher percentage of GFP-expressing parasites in the integrant Dd2attB/GFP line than the episomal Dd2/GFP line (94% versus 74% in this representative experiment, with the GFP-negative cells overlaying the profile observed with nontransfected Dd2). Although both transfected lines showed similar median intensities, the Dd2attB/GFP integrants showed a single peak of GFP expression with minimal variance, whereas the episomally transfected Dd2/GFP line had multiple peaks of GFP fluorescence intensity with a considerably wider variance (Fig. 4b).
We developed a 96-well assay to quantitatively compare the efficiency of transfection and integration of the Bxb1 integrase– mediated system with that of episomal transfection (Table 2). We cotransfected Dd2attB parasites with the pLN-ENR-GFP and pINT plasmids or separately transfected them with the single plasmid pBSD-GFP-INT-attP. We also used these plasmids to transfect the control Dd2C1 clone, representing a standard episomal approach. We determined parasitemias 2 d after electroporation and plated parasites over a range of concentrations (0.63 to 10 × 105 infected RBC/well). We aliquoted each concentration into eight wells, and we detected positive wells 24 d later by their expression of P. falciparum lactate dehydrogenase13. The transfection efficiency was calculated as the total number of parasites needed to obtain a viable, transfected parasite culture (that is, two times the starting parasite numbers that yielded four positive wells out of eight).
Dd2attB parasites electroporated with pBSD-GFP-INT-attP yielded one transfected parasite per 2.5 × 105 infected RBC (Table 2). Electroporation of this same plasmid into the Dd2C1 line required 106 parasites to produce the same efficiency. In the case of cotransfection with pLN-ENR-GFP and pINT, the Dd2attB line yielded one transfected parasite per 1.5 × 106 infected RBC, whereas the Dd2C1 line yielded no parasites at any starting inoculum. We obtained comparable efficiencies from dual-plasmid transfection of Dd2attB parasites and single-plasmid transfection of Dd2C1 parasites. PCR assays identified plasmid integration into 16/16 transfected Dd2attB wells, versus 0/16 Dd2C1 wells (data not shown), confirming that this integrase system mediated rapid and highly efficient transgene integration.
This specific integration system, which we demonstrate is applicable to the intracellular eukaryotic organism P. falciparum, has a number of attractive features that distinguish it from existing methods of P. falciparum transfection. First, it is now possible to rapidly and specifically target transgenes directly into a known site in the P. falciparum genome, thereby eliminating the need to screen multiple candidate integration sites and unnecessarily culture lines that undergo unwanted integration events. Plasmodium is particularly suitable as a host genus for Bxb1 serine integrase–based recombination, as its AT-rich genomes have no sequences that bear any detectable homology to the GC-rich attB or attP sites. The site specificity inherent to serine integrases is distinct from the piggyBac andmini-Tn5 transposon systems, recently adapted to Plasmodium, which exploit the existence of frequently occurring transposition target sites in the parasite genome and offer promise for random mutagenesis, gene delivery or systematic gene knockout studies14,15.
Second, recombination between the non-identical attB and attP sites produces asymmetric attL and attR sites that are refractory to excision by Bxb1 integrase5. The stability of the attB × attP locus, well-documented with the phiC31 integrase system in bacteria and yeast16,17, was evidenced here by the lack of rearrangement or reversion to the wild-type locus in recombinant parasites cultured without drug selection (Supplementary Fig. 3). Plasmid excision, leading to reconstitution of the attB locus, may nevertheless be desirable to confirm a transgene phenotype or to recycle markers. This may become feasible with the recent discovery that the Bxb1 gp47 gene product can interact with attL and attR to mediate excision18. The stability of Bxb1 integrase-mediated integration stands in marked contrast to tyrosine-based recombinases such as E. coli P1 phage Cre or Saccharomyces cerevisiae FLP (which target lox or frt sites, respectively). These recombinases mediate reversible integration and excision events between two identical target sites but thermodynamically favor the latter and are therefore more often used to delete a genomic locus19. Of note, the FLP/frt system was recently developed as a tool for conditional mutagenesis in P. berghei20.
Third, transgenes can now be rapidly integrated and carried throughout the life cycle, whereas transgene expression assays based on episomal replication are compromised by the rapid loss of episomes in the absence of selection (as happens, for example, when attempting to express transgenes in sexual stage parasites)11,21. For this, the 3D7 strain used here should be beneficial, as it is known to be able to produce infectious gametocytes and to be transmitted through Anopheles22. Our studies indicate that 3D7attB parasites produce mature gametocytes (data not shown), although their transmissibility has not been confirmed. In this line, the attB site is engineered into the cg6 coding sequence, the disruption of which does not confer any growth disadvantage, at least for asexual blood-stage parasites. We also engineered an attB site into the hrp3 5′ UTR, which does not disrupt the coding sequence and thus can be expected to have no adverse effect on parasite progression through the life cycle (Supplementary Fig. 2). Even if there were an effect, parasites lacking hrp3 are capable of self-fertilization and can complete the invertebrate and vertebrate life cycle23.
Fourth, the systematic integration of a single plasmid copymakes gene regulation studies more comparable by ensuring similar numbers of integrated plasmids (versus the episomal assays, in which episomal copy numbers typically range from 0 to 15 (refs. 8,11)). Our flow cytometry studies, comparing integrant versus episomal parasites, confirm this by showing a greater homogeneity both in the signal strength and in the percentage of parasites that express detectable levels of protein (Fig. 4). Indeed, gene regulation and protein trafficking studies may prove to benefit the most from this new system.
We have now used the Bxb1 integration system to mediate 93 separate integration events into the attB+ parasite lines described here. In every instance, PCR analyses identified attB × attP recombination (see, for example, Figs. 2 and and33 and Supplementary Fig. 4 online). Routinely, parasites are obtained between 14–25 d post-electroporation. Nevertheless, we have infrequently observed a lengthening of these times (up to 50 d, corresponding to reduced transfection efficiencies). In such cases, we have observed that integrated parasites constitute the major population and have successfully isolated the desired integrants by limiting dilution cloning. Furthermore, for these transgene studies, the size limitation of plasmids that can be successfully integrated is not yet known. Thus far, we have integrated plasmids as large as 10.2 kb, containing 5.4 kb of transgene (Supplementary Fig. 4). Evidence from mycobacteriophage, which use this method of recombination to specifically integrate 50- to 55-kb genomes into small (40- to 50- bp) attB sites of the host2,5, suggests that attB × attP integrative recombination will be just as efficient with larger P. falciparum transgenes. Cloning of large P. falciparum transgenes in E. coli is notorious for plasmid scrambling and sequence deletions24 and is likely to be the more frequent obstacle.
A limitation of the integration system we describe is its dependency on the use of host cell lines engineered to harbor an integrated attB site. The current study established such lines in both 3D7, whose genome has been sequenced25, as well as the multidrug-resistant Dd2 line. We are continuing to develop new attB+ lines and have obtained attB integration into a subtelomeric var gene in the cytoadherent A4 line26. Ultimately, the choice of line and attB recipient site remains subject to the limitations of traditional methods of homologous recombination to insert an attB site.
This system adds to a repertoire of recent genetic advances in Plasmodium studies, which include insertional mutagenesis mediated by the type II transposable element piggyBac14 and mini-Tn5-mediated shuttle transposon mutagenesis15. Regulated transgene expression has also been described in P. falciparum, in which a tetracycline analog is used to repress transactivated gene expression27. Combining our integrase system with these other techniques offers the prospect of developing efficient methods for complementation and change or loss-of-function studies. These expanded genetic techniques should benefit studies investigating biological processes in P. falciparum as we strive to discover and exploit weaknesses in this ecologically highly successful and lethal pathogen.
We propagated P. falciparum parasites in human RBC under 5% O2 and transfected as previously described28.We selected plasmids harboring the human DHFR selectable marker (pCG6-attB and pCG6), the bsd marker (pBSD-GFP-INT-attP and pLN-ENR-GFP) or the neomycin marker (pINT) using 2.5 nM WR99210, 2.5 μg/ml BSD or 125 μg/ml G418, respectively.
We performed western blot analysis as described29. We probed membranes with monoclonal antibodies to GFP (Clontech) or polyclonal antibodies to PfERD2 (ref. 30; obtained from MR4, the malaria research and reference reagent resource center), which were diluted 1:20,000 and 1:7,500, respectively. We detected signals using chemiluminescence (ECL Plus; Amersham Biosciences).
For GFP imaging, live infected erythrocytes were adhered to polylysine-coated MatTeck glass-bottom Petri dishes. We captured images on an Olympus IX81 electronically motorized inverted microscope at 90× magnification. We performed immunofluorescence microscopy as described31. We used antibodies to ACP32 at a 1:500 dilution and used 1 μg/ml Hoechst 33342 (Sigma) for nuclear staining. We deposited cells on coverslips coated with 1% polyethylenimine (Sigma) and mounted them on glass slides. We captured images on an Olympus IX71 upright microscope at 100× magnification. Microscopes were equipped with 12-bit Cooke Sensicam QE air-cooled CCD cameras and were operated by IP Lab Spectrum 3.6.1 at the Analytical Imaging Facility of the Albert Einstein College of Medicine.
We stained mitochondria by treating live pfenr-GFP expressing parasites with 50 nM MitoTracker Red CM-H2XRos (Molecular Probes) for 20 min at 37 °C, followed by cell fixation and Hoechst 33342 staining as described above. We imaged cells using differential interference contrast.
We incubated sorbitol doubly synchronized, late–ring stage parasites (harvested 16–20 h post-invasion) with 6.4 μg/ml Hoechst 33342 for 10 min at 37 °C. After washing in 1× PBS, we immediately analyzed cells for both GFP and Hoechst fluorescence (by exciting with a 488-nm or 350-nm laser and detecting with a 530/30 or 440/40 band pass filter, respectively) using an LSR-II flow cytometer (BD Biosciences). After compensation of the fluorescent signals, we collected one million events from each parasite line on three separate occasions over 6 weeks. We displayed data on bi-exponential plots to represent the total Hoechst- and GFP-negative cell populations, and analyzed the data using FlowJo 8 software (Tree Star).
pCG6-attB, pBSD-GFP-INT-attP, pLN-ENR-GFP and pINT plasmid details have been deposited into GenBank (accession numbers DQ813651–DQ813654). PlasmoDB: cg6, PF07_0036; hrp3, MAL13P1.480.
We thank P. Wang and J. Hyde (University of Manchester, UK) for their gift of the pPKDSneoII plasmid and G. McFadden (University of Melbourne, Australia) for his kind gift of the antibodies to ACP. We also gratefully acknowledge the Burroughs Wellcome Fund for providing a New Initiatives in Malaria Research Award (to W.R.J. and D.A.F.) that provided funding for this work, as well as US National Institutes of Health grants AI59114 to G.F.H. and AI060342 to D.A.F. and W.R.J. (P01; principal investigator, James C. Sacchettini).
Note: Supplementary information is available on the Nature Methods website.
AUTHOR CONTRIBUTIONSL.J.N., R.A.M., P.A.M., W.R.J. and D.A.F. designed the experiments, which were performed by L.J.N., R.A.M. and P.A.M. P.G. and G.F.H. provided integrase reagents and W.R.J., G.F.H. and D.A.F. conceived of the plan to test Bxb1 integrase in P. falciparum. All authors contributed to the manuscript, which was written primarily by L.J.N., R.A.M. and D.A.F. L.J.N. was co-mentored by D.A.F. and W.R.J.
COMPETING INTERESTS STATEMENT
The authors declare that they have no competing financial interests.