Simultaneous multiple allelic replacement in the malaria parasite enables dissection of PKG function

Abstract

Over recent years, a plethora of new genetic tools has transformed conditional engineering of the malaria parasite genome, allowing functional dissection of essential genes in the asexual and sexual blood stages that cause pathology or are required for disease transmission, respectively. Important challenges remain, including the desirability to complement conditional mutants with a correctly regulated second gene copy to confirm that observed phenotypes are due solely to loss of gene function and to analyse structure-function relationships. To meet this challenge, here we combine the dimerisable Cre (DiCre) system with the use of multiple lox sites to simultaneously generate multiple recombination events of the same gene. We focused on the Plasmodium falciparum cGMP-dependent protein kinase (PKG), creating in parallel conditional disruption of the gene plus up to two allelic replacements. We use the approach to demonstrate that PKG has no scaffolding or adaptor role in intraerythrocytic development, acting solely at merozoite egress. We also show that a phosphorylation-deficient PKG is functionally incompetent. Our method provides valuable new tools for analysis of gene function in the malaria parasite.

Figure 1.. Replacement of pfpkg intron 3 with 2loxPint allows normal PKG expression and parasite replication.

(A) Shown is the integrated 2loxPint sequence derived by nucleotide sequencing of genomic DNA from the P. falciparum pfpkg_2lox line, aligned with the expected sequence (both in red). The loxN and lox2272 sites are in grey boxes, with the unique internal AccI site underlined. Boundaries of the 3′ end of exon 3 (AAAG) and the 5′ end of exon 4 (GGT) are shown in black. (B) Schematic representation of the strategy used to replace intron 3 of the pfpkg gene with 2loxPint in the DiCre-expressing B11 P. falciparum line. Positions of oligonucleotides used for diagnostic PCR are indicated (red and blue arrows). (C) (Left) Diagnostic PCR results showing absence of the endogenous intron 3 in the pfpkg_2lox line (2lox) relative to the parental B11 line (wt). (Right) Diagnostic PCR results and restriction digest of the PCR amplicon with AccI, showing the expected digestion only of the amplicon from the pfpkg_2lox line (2lox). (D) Western blot analysis of extracts of wt and pfpkg_2lox schizonts, showing similar expression levels of PKG (∼98 kD, indicated). Antibodies to the cytoplasmic parasite protein HSP70 were used as a loading control. (E) Growth curves showing replication of DMSO-treated (control) or RAP-treated pfpkg_2lox parasites relative to the parental line B11. Mean values are shown from triplicate experiments. Error bars ± SD (n = 3).

Figure 2.. Stochastic recombination between multiple lox sites creates genetically distinct parasite populations within a single culture.

(A) Schematic of the Cas9-enhanced targeted homologous recombination approach used to create the pkg:wGFP-ckoR line. Positions of the four consensus cyclic nucleotide-binding domains (CNA-CND) and the catalytic domain (open red box) of PKG are shown. The position targeted by the guide RNA used to direct Cas9-mediated cleavage is indicated (blue line), as are positions of the loxN (yellow arrowheads) and lox2272 (green arrowheads) sites. RAP-induced DiCre activity switches expression from wt PKG to either a gene replacement with a partially synthetic pfpkg gene fused to eGFP (recombination event 1; PKGsynth_GFP) or to gene disruption and expression of a truncated protein lacking the cyclic nucleotide-binding and kinase domains, fused to mCherry (recombination event 2; ΔPKG_mCherry). Black arrows; oligonucleotide primers used for identification of both events by diagnostic PCR. Note that tagging of PKG with a C-terminal eGFP was expected to be tolerated because it has previously been achieved in the rodent malaria model P. berghei (29). (B) Diagnostic PCR showing generation of products corresponding to both predicted recombination events upon amplification from genomic DNA of RAP-treated pkg:wGFP-ckoR parasites. Expected sizes of the amplicons corresponding to PKGsynth_GFP and ΔPKG_mCherry are 5.6 and 1.3 kb, respectively. Multiple attempts to amplify the corresponding region from mock-treated parasites (−RAP) failed, probably because of the large size of the predicted amplicon (∼10 kb). (C) Western blot showing loss of expression of PKG upon RAP treatment of pkg:wGFP-ckoR parasites, concomitant with appearance of signals corresponding to the PKG_GFP fusion (expected molecular mass 125 kD) and the mCherry fusion (expected mass 38 kD; red asterisk). After being probed with rabbit anti-PKG antibodies (left-hand blot), the membrane was re-probed with a rabbit anti-mCherry antibody to demonstrate the appearance of mCherry only in the RAP-treated sample and to highlight the difference between the DMSO- and the RAP-treated samples for both PKG (black asterisk) and mCherry. Note that the PKG_GFP fusion is not recognised by the commercially available PKG antibody because of masking of the C-terminal epitope, as previously observed (57). (D) (Left) Quantification by flow cytometry of the relative proportions of PKGsynth_GFP and ΔPKG_mCherry parasites in RAP-treated pkg:wGFP-ckoR parasites (sampled at the end of cycle 0). Individual values (dark circles or squares) are shown from five biological replicate experiments (n = 5), and mean values are indicated as bars. Statistical significance was determined by unpaired t test (P-value < 0.0001). (Right) Representative image from differential inference contrast/fluorescence microscopic examination of RAP-treated pkg:wGFP-ckoR parasites (end of cycle 0), showing both GPF- and mCherry-positive schizonts. Scale bar, 10 μm.

Figure S1.. Generation and genotyping of transgenic P. falciparum lines pkg:wGFP-ckoR, pkg:ckoR_mutGFP, and pkg:3cKO.

(A, B, C) Modification strategies and genotyping data for generation of parasite lines pkg:wGFP-ckoR, pkg:ckoR_mutGFP, and pkg:3cKO, respectively. Positions of oligonucleotides used for genotyping by diagnostic PCR are indicated (coloured arrows), and agarose gel electrophoresis of corresponding PCR products are shown. Positions of lox sites are indicated with coloured arrowheads (yellow, loxN; green, lox2272; purple, lox71; brown, lox66).

Figure 3.. PKG-null parasites undergo normal intraerythrocytic development but arrested egress.

(A) Growth curves showing replication of DMSO-treated (control) or RAP-treated pkg:wGFP-ckoR and pfpkg_2lox parasites. Percentage parasitaemia values are shown (quantified by flow cytometry). Error bars ± SD (n = 3). (B) Differential inference contrast/fluorescence images of schizonts from a RAP-treated pkg:wGFP-ckoR culture, showing virtual disappearance of ΔPKG_mCherry parasites by the end of cycle 1. (C) Two-parameter dot plot representation of flow cytometry data monitoring the relative proportions of PKGsynth_GFP schizonts (green box, Q3; lower right-hand quadrant) and ΔPKG_mCherry schizonts (red box, Q1; upper left-hand quadrant) with time. Monitoring was initiated ∼44 h after RAP treatment of a highly synchronous pkg:wGFP-ckoRc culture. The percentage of each population at each time point is shown within the relevant quadrant. The Q4 population predominantly represents uninfected erythrocytes. Parasitaemia at the point of RAP-treatment (the start of cycle 0) was 6.5%. (D) Histogram depiction of flow cytometry analysis of schizonts enriched ∼44 h after RAP treatment of a pkg:wGFP-ckoRc culture, showing time-dependent accumulation of ΔPKG_mCherry schizonts and loss of PKGsynth_GFP schizonts over a 3-h time period. (E) Stills from time-lapse differential inference contrast/fluorescence microscopy of isolated, RAP-treated pkg:wGFP-ckoR schizonts after release of a compound 2–mediated egress block, showing that only the PKGsynth_GFP schizonts undergo rupture and merozoite egress. No rupture of the ΔPKG_mCherry schizonts was observed even after prolonged imaging. Scale bars, 10 μm.

Figure S2.. ΔPKG_mCherry parasites are defective in egress.

Dot plot representation of flow cytometry data monitoring the fate of mock (DMSO-) or RAP-treated pkg:wGFP-ckoR schizonts over a time period of 3 h. DMSO-treated parasites were used as a control for gating. In the RAP-treated parasites, the proportion of PKGsynth_GFP parasites in the population (quadrant Q3, green box) decreased over time as these schizonts underwent egress, whereas in contrast, the proportion of ΔPKG_mCherry schizonts (quadrant Q1, red box) increased over time as these parasites accumulated because of a defect in egress. The Q4 population predominantly represents contaminating uninfected erythrocytes. 50,000 events were recorded for each plot (n = 2).

Figure 4.. Phosphosite mutations render P. falciparum PKG inactive.

(A) Cartoon of the P. falciparum PKG x-ray crystal structure (PDB ID: 5DYK) in its apo form with rainbow colouring (N terminus in dark blue; C terminus in red). Cyclic nucleotide-binding domains A (dark blue), B (cyan), C (green), and D (lime) are shown, whereas the central kinase domains are in yellow/orange/red. Phosphosites identified by mass spectrometry are indicated and shown as sticks within colour-matching transparent spheres. The image was ray-traced in the PyMOL Open-Source Molecular Graphic System (https://pymol.org/2/). (B) Schematic of the modified pfpkg locus in the pkg:cKOR-mutGFP parasite line. Upon DiCre induction with RAP, recombination event 1 leads to conditional gene disruption (ΔPKG_mCherry), whilst recombination event 2 leads to replacement of the endogenous allele with a partially synthetic full-length allele containing Ala substitutions of all seven phosphosites (asterisks), fused to GFP (recombination event 2; PKGmut_GFP). (C) Diagnostic PCR results showing detection of the two distinct recombination events after DiCre activation. The amplicon specific for ΔPKG_mCherry (denoted by the black and red arrows) is ∼1 kb in the RAP-treated sample and ∼4.9 kb in the mock-treated (non-excised) sample. The amplicon specific for PKGmut_GFP is ∼4 kb in the RAP-treated sample. Amplification from mock-treated samples was unsuccessful, likely because of the large size of the predicted fragment. (D) Quantification of the ratio between PKGmut_GFP and ΔPKG_mCherry schizonts in the RAP-treated parasite population at the end of cycle 0. Data shown are from five independent experiments; individual and mean values are shown. Error bars ± SD (n = 5). (E) Giemsa-stained images of Percoll-enriched schizonts isolated at the end of cycle 0 of DMSO- and RAP-treated pkg:cKOR-mutGFP parasites, showing no discernible morphological differences. Scale bar, 10 μM. (F) Replication of DMSO- and RAP- treated pkg:ckoR_mutGFP parasites over three erythrocytic cycles. Parasitaemia values shown (obtained by flow cytometry) are averages of three independent experiments. Error bars ± SD (n = 3). (G) Both PKGmut_GFP and ΔPKG_mCherry schizonts are defective in egress. Still images of the first and final frame of a 30-min time-lapse video of mock-treated (grey) or RAP-treated (green and red) pkg:ckoR_mutGFP parasites. Schizonts were synchronised by incubation with the reversible PKG inhibitor compound 2, then washed, mixed in equal proportions, and monitored for egress over a period of 30 min. Neither the red (ΔPKG_mCherry) nor the green (PKGmut_GFP) schizonts underwent egress, whereas most of the parental mock-treated pkg:ckoR_mutGFP schizonts ruptured. Scale bar, 10 μΜ. (H) Western blot showing that no SERA5 P50 was released into culture supernatants of RAP-treated pkg:ckoR_mutGFP schizonts, consistent with impaired egress in both the PKGmut_GFP and ΔPKG_mCherry schizonts.

Figure S3.. Replacement of pfpkg intron 3 with 3loxPint.

(A) Schematic representation of the approach used to replace intron 3 of pfpkg with 3loxPint. (C) Red arrows indicate relative positions of oligonucleotides used for genotyping in panel (C). The unique AccI site within 3loxPint is underlined. (B) Sequence alignment of 3loxPint from genomic DNA of the pfpkg_3lox parasites, relative to the expected sequence. The sera2 intron sequence is in red, whereas the lox71, loxN, and lox2272 sites are boxed. Boundaries of the 3′ end of exon 3 and 5′ of exon 4 are shown in black. (C) Diagnostic PCR and restriction digest of the product with AccI, showing the expected two fragments only in the pfpkg_3lox line (3lox). The equivalent PCR amplicon from B11 parasites (wt) is undigested. (D) Western blot analysis of Percoll-enriched schizonts showing similar expression levels of PKG protein expression in pfpkg_3lox, pfpkg_2lox, and parental B11 (wt) schizonts. Expression of parasite HSP70 was used as a loading control.

Figure 5.. Simultaneous generation of three distinct allelic exchange events.

(A) Schematic of the approach used to conditionally disrupt pfpkg and create three distinct knockout parasite populations expressing either mTagBFP2 (ΔPKG_BFP2), eGFP (ΔPKG_GFP), or mCherry (ΔPKG_mCherry). Positions of lox sites are indicated with coloured arrowheads (yellow, loxN; green, lox2272; purple, lox71; brown, lox66). Positions of oligonucleotide primers used for diagnostic PCR are indicated (coloured arrows). (B) Confirmation by diagnostic PCR of the three recombination events (RE1, RE2, and RE3, respectively) in RAP-treated pkg:3cKO parasites. Coloured arrows represent identify of the primers used. (C) Growth assay of the pfpkg_3lox and pkg:3cKO parasite lines after mock-treatment (DMSO) or treatment with RAP. Parasitaemia was measured by flow cytometry. Error bars ± SD (n = 6). (D) Fluorescent microcopy images of live RAP-treated pkg:3cKO schizonts from the end of cycle 0, confirming the presence of all three fluorescent populations. Scale bars, 10 μΜ.

Figure S4.. Correct recombination between the different lox sites in 3loxPint.

(A) Data from nucleotide sequencing of PCR amplicon RE1 (Fig 5B) showing correct recombination leading to formation of the unique lox72 site (green letters) and deletion of both loxN and lox2272. (B) Data from nucleotide sequencing of PCR amplicon RE2 (Fig 5B) showing the presence of an unmodified lox71 and correct recombination between the loxN sites. (C) Data from nucleotide sequencing of PCR amplicon RE3 showing recombination only between the lox2272 sites, whereas the other 2 lox sites remain unaltered. In all three cases, the chromatograms are shown. The last nucleotides of exon 3 are shown in black letters whilst recombined 3loxPint modules are shown with red letters. Relevant lox sites are within grey boxes. (A, B, C) The start of the coding regions of mTagBFP2, eGFP, and mCherry (A, B, C, respectively) following the 3loxPint module are shown in black letters.