Essentiality of Plasmodium falciparum plasmepsin V

The malaria parasite replicates within erythrocytes. The pathogenesis of clinical malaria is in large part due to the capacity of the parasite to remodel its host cell. To do this, intraerythrocytic stages of Plasmodium falciparum export more than 300 proteins that dramatically alter the morphology of the infected erythrocyte as well as its mechanical and adhesive properties. P. falciparum plasmepsin V (PfPMV) is an aspartic protease that processes proteins for export into the host erythrocyte and is thought to play a key role in parasite virulence and survival. However, although standard techniques for gene disruption as well as conditional protein knockdown have been previously attempted with the pfpmv gene, complete gene removal or knockdown was not achieved so direct genetic proof that PMV is an essential protein has not been established. Here we have used a conditional gene excision approach combining CRISPR-Cas9 gene editing and DiCre-mediated recombination to functionally inactivate the pfpmv gene. The resulting mutant parasites displayed a severe growth defect. Detailed phenotypic analysis showed that development of the mutant parasites was arrested early in the ring-to-trophozoite transition in the erythrocytic cycle following gene excision. Our findings are the first to elucidate the effects of PMV gene disruption, showing that it is essential for parasite viability in asexual blood stages. The mutant parasites can now be used as a platform to further dissect the Plasmodium protein export pathway.


Introduction
Malaria was responsible for approximately 445,000 deaths and 216 million clinical cases in 2016, an increase of~5 million cases over the previous year [1]. Vital to the growth and pathogenicity of the parasite is host cell remodelling in which the parasite modifies the host erythrocyte by the synthesis and export of over 300 parasite proteins beyond the bounds of the parasitophorous vacuole (PV) within which it replicates (for reviews see [2,3] protein complex known as the translocon of exported proteins (PTEX), located within the PV membrane (PVM) [4][5][6][7][8][9]. The exported proteins extensively alter the mechanical and adhesive properties of infected erythrocytes, resulting in vascular sequestration of the infected cells and eventual destruction of the host erythrocyte [10]. Protein export in Plasmodium has been most intensively studied in P. falciparum (reviewed in [11,12]), where the majority of known exported proteins contain the pentameric localisation motif RxLxE/Q/D, termed Plasmodium export element (PEXEL), typically located downstream of the N-terminal secretory signal sequence that regulates entry into the ER [13,14]. Some proteins lacking a PEXEL motif, referred to as PEXEL-negative exported proteins (PNEPs), can also be exported through a distinct mechanism that may however share features with PEXEL-mediated export [15,16]. P. falciparum plasmepsin V (PfPMV) is an ER-located aspartic protease comprising 590 amino acid residues (~68 kDa) [17]. PfPMV expression levels in the parasite progressively increase throughout schizogony [17]. Numerous studies have now established that PMV is directly responsible for cleavage of the PEXEL motif within exported proteins [18][19][20][21]. Cleavage occurs on the C-terminal side of the conserved Leu residue (RxL#), revealing a new N-terminus that is rapidly acetylated ( Ac -xE/Q/D) [22,23]. A recent x-ray crystal structure of Plasmodium vivax PMV (PvPMV) revealed a canonical aspartyl protease fold with several important features [24]. These include a nepenthesin (NAP)-like insertion within the N-terminal part of the enzyme, which may control substrate entry into the active site and influence enzyme specificity, as well as a helix-turn-helix (HTH) motif near the C-terminus of the enzyme, which is conserved in PMV from other Plasmodium species but not found in other parasite plasmepsins involved in hemoglobin digestion. PMV also possesses an unpaired Cys residue (C140 in PvPMV, equivalent to C178 in the P. falciparum enzyme) which is located in the flap of the structure and is restricted to Plasmodium species. In attempts to establish the druggability and function of PMV, several peptidomimetic inhibitors based on the PEXEL motif have been developed [25][26][27]. A co-crystal structure of PMV bound to one of these compounds, the PEXEL-mimetic WEHI-842, showed that although the unpaired Cys points into the active site of the enzyme, it does not appear to make contact with the inhibitor [24,28]. Another inhibitor, WEHI-916, inhibits the activity of purified PMV isolated from both P. falciparum and P. vivax, and higher concentrations of WEHI-916 were required to kill parasites engineered to over-express PfPMV indicating on-target efficacy [25,27]. Compound 1, a hydroxyl-ethylamine PEXEL-mimetic, inhibited PfPMV activity in vitro with picomolar potency but failed to block parasite growth due to poor stability and membrane permeability [26]. Collectively, these findings indicate that PMV has a number of distinguishing features that could be exploited in drug design, and that it can potentially be targeted with suitable inhibitory compounds.
An important component in the validation of any enzyme as a drug target is genetic ablation of enzyme expression. Unfortunately, previous attempts to genetically delete or knockdown PMV expression have been only partially successful. Work by Klemba and colleagues [17] and Boddey et al [18] failed to disrupt the pfpmv gene or the P. berghei PMV gene (pbpmv) respectively, suggesting (but not proving) that the gene is essential. Russo et al [21] similarly attempted to disrupt the pfpmv gene using an allelic replacement approach. Single crossover homologous recombination into the endogenous pfpmv locus was possible only when the catalytic dyad aspartate codons were preserved. Conditional knockdown of PfPMV expression was also attempted using the RNA-degrading glmS ribozyme system [27]. Although this resulted in around 75-90% knockdown of PMV expression, the remaining PMV levels were presumably sufficient to enable export and sustain parasite development. Gambini et al [26] generated an inducible PMV knock-down by fusing PMV with a destabilising domain (DD), but only a 4 to 10-fold knock-down of PMV cellular levels was achieved using this approach, which did not affect parasite viability. In summary, whilst some of these data are consistent with an essential role for PMV, in the absence of any reports of complete ablation of PfPMV expression, direct genetic proof that PMV is an essential protein is lacking.
Here we have used a robust conditional genetic approach to truncate the pfpmv gene. Our findings are the first to confirm genetically that disruption of PfPMV has a significant effect on ring-to-trophozoite development and ultimately results in arrest of parasite growth.

Generation of modified pfpmv parasites using CRISPR-Cas9
Previous attempts to disrupt the pfpmv gene using conventional genetic techniques were unsuccessful [18,21], and conditional knockdown approaches did not significantly affect PEXEL processing or parasite viability [26,27], presumably due to relatively low levels of PfPMV expression being sufficient to sustain parasite viability. To explore the consequences of complete functional inactivation of PMV, we therefore took advantage of the DiCre conditional recombinase system, recently adapted to P. falciparum [29]. Using Cas9-mediated genome editing [30] we first introduced synthetic introns containing loxP sites (loxPint sequences; [31]) into the endogenous pfpmv locus such that they flanked (floxed) an internal segment of the gene encoding Asp133 to Thr590. Importantly, one of the PfPMV catalytic dyad residues (Asp365) is included within this floxed region. At the same time, the modified gene was fused to a C-terminal HA3 epitope tag to aid detection of the transgene product ( Fig  1A). The repair plasmid used for the manipulation contains a modified selection-linked integration (SLI) region [32] which allows selection for integration events with neomycin. This is enabled by the integration of a 2A peptide sequence immediately downstream of the modified gene, which causes ribosome 'skipping' to produce 2 two distinct polypeptides (the modified PMV-HA and the neo-R gene product) from a single mRNA (note that neomycin selection was in fact unnecessary in our work due to the high efficiency of the Cas9-mediated homologous recombination). The genomic modification was made in the DiCre-expressing P. falciparum B11 parasite clone [33] such that excision of the floxed sequence could be induced by treatment of the transgenic parasites with rapamycin (RAP). DiCre-mediated excision was predicted to generate an internally-truncated mutant form of PMV lacking one of the catalytic dyad residues. Excision would also remove the neo-R gene and induce expression of GFP as a fluorescent reporter of excision by virtue of a second downstream 2A peptide sequence ( Fig  1A).
Successful modification of the pfpmv gene in the transfected parasite population following the introduction of the targeting vector was confirmed by diagnostic PCR. Limiting dilution cloning of the modified parasites resulted in the isolation of parasite clones PMV-C5 and PMV-F9, which were derived from independent transfections using different guide RNAs. Modification of the native pfpmv locus was confirmed in both transgenic parasite lines by diagnostic PCR (Fig 1B). Both clones replicated at a rate indistinguishable from that of the parental B11 parasites, suggesting that the modifications did not affect parasite viability. Transgenic parasite clone PMV-C5 was used for all subsequent experiments.

Conditional truncation of the pfpmv gene
Expression of the recodonised pfpmv gene in transgenic parasite clone PMV-C5 was expected to produce an epitope-tagged PMV product (called PMV-HA), as well as expression of the neo-R gene product. Note that in this event GFP is not produced because of the presence of a translational stop codon directly downstream of the neo-R gene. Upon RAP-treatment, sitespecific recombination between the introduced loxP sites in the modified pfpmv locus of the PMV-C5 parasites was anticipated to reconstitute a single functional loxP-containing intron. Splicing of this intron results in a truncated form of PMV lacking the HA epitope tag, as well as allowing expression of free GFP (Figs 1A and 2A). Diagnostic PCR analysis of DNA extracted from RAP-treated and mock-treated (DMSO-treated control) PMV-C5 parasites confirmed the expected excision event (Fig 2B), although a signal indicative of some residual non-excised DNA was also detectable, as observed with some previously-described DiCre mutants (e.g. [29,31]). To further examine this, we therefore monitored the appearance of GFP-positive parasites following treatment of synchronous ring-stage parasites with DMSO or RAP. As expected, GFP fluorescence was only detected in the RAP-treated parasites ( Fig 2C). Indeed, no schizonts lacking GFP fluorescence could be detected upon exhaustive microscopic examination of the RAP-treated cultures, indicating highly efficient excision of the floxed pfpmv sequence. This was supported by FACS analysis (Fig 2D), showing that nearly 99% of the RAP-treated parasites displayed GFP fluorescence. These results demonstrated close to complete conditional truncation of PMV within a single erythrocytic cycle in the PMV-C5 parasite clone.

Disruption of the pfpmv gene leads to a developmental arrest in the subsequent erythrocytic cycle
To initially explore the effects of PMV truncation on parasite viability, we examined growth and development of the ΔPMV mutant by microscopic examination of Giemsa-stained cultures. This revealed that although RAP-treated PMV-C5 parasites appeared morphologically normal throughout the erythrocytic growth cycle in which the parasites were treated (henceforward referred to as cycle 0), development of the mutant parasites stalled at the late ring/ early trophozoite stage in the following cycle (cycle 1; Fig 3A). By the end of cycle 1, only arrested pycnotic forms were observed which did not develop further upon prolonged incubation. These results suggested that ablation of PMV expression causes a severe growth defect and developmental arrest between ring and trophozoite stage.
Whilst these results showed that truncation of PfPMV affected parasite development in cycle 1, it did not rule out the possibility of defects in egress and invasion of fresh erythrocytes at the end of cycle 0 (the cycle of RAP-treatment). To evaluate and quantify this, we used flow cytometry to monitor increases in parasitaemia at the transition between cycle 0 schizonts and cycle 1 rings. As shown in Fig 3B, the increase in parasitaemia of PMV-C5 and PMV-F9 schizont cultures at the end of cycle 0 was unaffected by RAP-treatment, indicating no effects of PMV ablation on schizont development in cycle 0, or the egress and invasion capacity of the released merozoites. However, upon further development of the new parasite generations in cycles 1 and 2, parasitaemia levels in the RAP-treated culture did not increase further. This confirmed that disruption of PfPMV did not prevent egress and invasion at the end of cycle 0, and were consistent with the Giemsa data indicating that the enzyme is essential for the ringto-trophozoite developmental transition of the intracellular parasite in cycle 1, the cycle immediately following gene disruption.
To further explore the PMV-null phenotype, we examined the expression of PMV-HA as well as other non-exported and exported parasite proteins. Extracts of DMSO and RAP-treated PMV-C5 schizonts were analysed by immunoblot~42 h following treatment (i.e. near the end of cycle 0), using antibodies specific for the HA epitope tag as well as GFP, the merozoite plasma membrane surface protein MSP1, and the exported protein KAHRP. As shown in Fig  4A, only a faint HA signal was detected in the RAP-treated sample, whilst a corrrespondingly strong GFP signal was observed only in this sample, as expected. In contrast, no differences between DMSO-and RAP-treated parasites were detectable in expression levels of the nonexported merozoite surface protein MSP1 (thought to play an important role in egress and invasion [34]) There were no obvious significant differences in expression or migration on SDS-PAGE of KAHRP, an example of a PEXEL-containing protein that is cleaved by PMV [18,19,21].
We next determined the effects of PfPMV truncation on its subcellular localization within the parasite, as well as on the trafficking of other exported proteins in cycle 0. Immunofluorescence analysis (IFA) showed that, as expected, PMV-HA showed a perinuclear localisation in schizonts of control PMV-C5 parasites but the signal was lost in RAP-treated parasites ( Fig  4B). Mock-and RAP-treated PMV-C5 parasites were also probed with anti-MSP1 and anti-KAHRP2 antibodies, as well as antibodies to another exported protein, HRP2. The localization and intensity of the MSP1 signal and the KHARP signal did not alter upon RAP-treatment, correlating well with the western blot analysis. The HRP2 signal remained detectable in both control and the RAP-treated parasites, but in contrast to the other proteins examined, the fluorescence intensity of the HRP2 signal was substantially decreased in RAP-treated parasites, suggesting a partial effect of PMV depletion on HRP2 trafficking (Fig 4B). Together, these data confirmed the loss of PMV-HA in RAP-treated parasites whilst indicating that both exported proteins (KAHRP and HRP2) and merozoite surface proteins still localised correctly in cycle 0. However, the decreased HRP2 signal intensity was consistent with a partial impact on protein export in cycle 0.

Discussion
Malarial proteolytic enzymes play regulatory and effector roles in multiple key biological processes in this important pathogen and have long been of interest as potential drug targets. In this study, we have shown for the first time that functional ablation of the pfpmv gene leads to a block at the ring-to-trophozoite transition and failure of the parasite to proliferate. To achieve this, we employed the robust conditional DiCre approach in combination with Cas9-induced gene modification. This system provides a rapid and efficient means of generating transgenic parasites as well as enabling complete disruption of the gene of interest [35,36]. Our genetic strategy also incorporated a method to permit the selection of parasites in which genomic integration of the input constructs had taken place, termed selection-linked integration (SLI), previously developed by Birnbaum et al [32]. For this, our repair plasmid contained an SLI-resistance marker (neomycin phosphotransferase), which was linked to the modified pfpmv gene separated by a 2A 'skip' peptide [37,38]. This approach allows selection for correct integration using neomycin selection. In our case, with the assistance of Cas9-mediated integration we in fact did not need to use neomycin selection in our experiments to select for correct integration. However, our plasmids could be useful for future genetic complementation studies since the neo-R gene is removed upon DiCre-mediated excision, effectively recycling the marker for subsequent use.
When the PMV-modified parasite clone PMV-C5 was RAP-treated at early ring stage for just 3 h, followed by removal of RAP, parasite development proceeded normally in the first erythrocytic cycle (cycle 0). Whilst expression of most of the parasite proteins examined appeared unaffected, we did observe some impact on expression of the exported parasite protein HRP2. This is reminiscent of previous observations of reductions in expression of PMV substrates following inhibition of PMV [21,27]. We suspect that the very limited effects of pfpmv disruption on parasite development in cycle 0 may be due to low levels of PMV expression early in cycle 0, perhaps due to transcription early in the erythrocytic cycle prior to gene excision. That low levels of PMV are sufficient to sustain development has also been indicated by previous conditional protein knockdown experiments [27]. Egress and invasion at the end of cycle 0 was not affected in the RAP-treated PMV-C5 parasites; however, parasite development in the next cycle irreversibly stalled early following invasion. This phenotype is very similar to that observed upon treatment with the PMV inhibitor WEHI-916 [27], where the drug-  treated parasites showed a growth defect at the ring-to-trophozoite transition from approximately 20 h post-invasion. A similar parasite developmental arrest at ring stage was also observed following conditional ablation of the PTEX component HSP101 [6,7].
In summary, using a DiCre-mediated conditional gene editing approach to selectively disrupt the pfpmv gene, we have shown that the gene is essential for the ring-to-trophozoite transition of intracellular growth. This engineered platform will be useful for further study of PEXEL-protein export as well as for dissection of PMV domain interactions, providing further impetus for focusing on PMV as a new potential antimalarial drug target.

Cloning of repair plasmid pT2A-5´UTR-3´-PMV-ΔDHFR
pT2A-5´UTR-3´-PMV-ΔDHFR plasmid is based on vector pT2A-DDI-1cKO (a kind gift of Dr. Edgar Deu, the Francis Crick Institute). The plasmid comprised the native 5´UTR of pfpmv as well as nucleotides 1-397 of the native pfpmv sequence (Met1 to Lys132) as 5' targeting sequence for homologous recombination. This was followed by a synthetic DNA segment comprising heterologous loxP-containing sera2 and sub2 introns (loxPint) [31] flanking a segment of pfpmv sequence encoding residues Asp133 to Thr590 fused to an HA3 epitope tag sequence, a 2A 'skip' peptide (2A) [44], and the neomycin resistance gene (neo). The synthetic pfpmv sequence encoded the wild-type protein sequence but was 'recodonised' (i.e. with codon usage different to that of the native genomic parasite gene) in order to reduce its similarity at the nucleotide level with the native gene and thus prevent undesirable homologous recombination within this segment. The entire synthetic segment was produced by GeneArt. The second loxPint site was immediately followed by another 2A sequence, a gfp gene and nucleotides 1333 to 1773 of native pfpmv sequence which encode Lys445 to Thr590, as a 3´targeting sequence for homologous recombination. Overlapping PCR was used to generate the loxPint-2A-gfp fragment from the template pT2A-DDI-1-cKO-complement (obtained from Dr. Edgar Deu, the Francis Crick Institute). The resulting PCR product was digested with XhoI and ligated into pT2A-DDI-1cKO pre-digested with the same restriction enzymes, yielding plasmid pT2A-DDI-1-cKO-modified GFP (S1 Fig). The 5´-targeting sequence (903 bp) was PCR amplified from 3D7 genomic DNA using Phusion HF DNA polymerase (NEB) with forward primer P3.1-BglII_5´UTR_F-mod and reverse primer P2-Int_PMV_R. The recodonised fragment (1,602 bp) containing loxPint, recondonised pfpmv, HA3, and 2A was amplified from the plasmid 17ACRILIP_2177297_endoPMV (GeneArt) with primers P4-PMV_int_F and P5-T2A_HA_R. PCR products of 5´-target region and the recodonised fragment were ligated with pT2A-DDI-1-cKO-modified GFP pre-digested with BglII and SalI using In-fusion HD cloning kit (Clontech, Mountain View, CA), producing the plasmid pT2A-5´UTR-PMV-cKO. The 3´-target region (441 bp) was amplified from 3D7 genomic DNA using Phusion HF DNA polymerase (NEB) with forward primer P16-EcoRV-PMV-F and reverse primer P17-EcoRV-PMV-R. This fragment was then ligated into pT2A-5´UTR-PMV-cKO predigested with EcoRV, yielding the plasmid pT2A-5´UTR-3´PMV-cKO. The hdhfr gene was removed from pT2A-5´UTR-3´PMV-cKO by digesting the plasmid with BamHI and EcoRI. The plasmid backbone was blunt-end using T4 DNA polymerase (NEB) then religated using T4 DNA ligase (NEB), giving rise to the repair plasmid pT2A-5´UTR-3´-PMV-ΔDHFR.

Generation of pmv-loxPint parasites and conditional PMV truncation
The repair plasmid pT2A-5´UTR-3´-PMV-ΔDHFR was linearized with ScaI prior to electroporation. Percoll-enriched mature schizonts of P. falciparum clone B11 were electroporated with 20 μg of pSgRNA1 or pSgRNA2 and 60 μg of linearized pT2A-5´UTR-3´-PMV-ΔDHFR using an Amaxa P3 primary cell 4D Nucleofector X Kit L (Lonza) as described [29]. Twentyfour hours post-transfection, the electroporated parasites were treated with 2.5 nM WR99210 for 96 h to select for transfectants harboring pSgRNA plasmids before returning the cultures to medium without drug. Integrant parasites generally reached parasitaemia levels suitable for cryopreservation within 2-5 weeks. Detection of the pfpmv-loxPint modified locus was carried out by diagnostic PCR using primer pairs P6-5´UTR_Screen_F and P23-rcPMV-5´integr-R, P6-5´UTR_Screen_F and 198_GFP_start_seq_R, and P24-GFP-3´integr-F and P25-3 UTR-PMV-R. The wild-type pfpmv locus was detected by diagnostic PCR using primers P6-5 UTR_Screen_F and P22-PMV-endo-R. Transgenic parasite clones were obtained by limiting dilution cloning by plating a calculated 0.3 parasite per well in flat-bottomed 96-well microplate wells as described [46]. Wells containing single plaques were subsequently expanded into round-bottomed wells. Transgenic parasite clones (PMV-C5 and PMV-F9) were finally checked by diagnostic PCR for integration and modification of the endogenous pfpmv gene. Once established, all transgenic clones were maintained in medium without any drug.
Recombination between the loxPint sites was induced in tightly synchronised ring-stages of parasite clone PMV-C5 by incubation for 3 h in the presence of 100 nM RAP in 1% (v/v) DMSO; mock treatment was with 1% (v/v) DMSO only. DiCre-mediated excision of the floxed pfpmv was detected by PCR analysis of parasite genomic DNA using primers P6-5´UTR_ Screen_F and P23-rcPMV-5´integr-R, and P6-5´UTR_Screen_F and Deu198_GFP_start_ seq_R. Truncation of PfPMV was evaluated by immunoblot analysis of SDS extracts of mature Percoll-enriched schizonts, probing with the anti-HA antibody 3F10 (Roche), followed by horseradish peroxidase-conjugated secondary antibodies.

Nucleic acid extraction and polymerase chain reaction
For DNA extraction, total cell pellets were first treated with 0.15% saponin in PBS for 10 min, then washed with PBS before DNA was extracted using a DNeasy Blood & Tissue Kit (QIAGEN). For diagnostic PCR amplification, GoTaq (Promega) DNA master mix was used. Amplification of fragments used in construct design was carried out using Phusion high fidelity DNA polymerase (NEB).

Parasite growth assay
Parasitaemia measurement by FACS was as described previously [47]. Briefly, parasites recovered at various time-points were fixed in 8% paraformaldehyde 0.04% glutaraldehyde, pH 7.4 and stained with 2 μM Hoechst 33342 (Invitrogen, Waltham, MA). Parasitaemia was calculated using the FACS BD Fortessa flow cytometer (BD Biosciences). Cultures to be analysed were initially screened using forward and side scatter parameters and gated for erythrocytes. From this gated population, the proportion of Hoechst-stained cells in 100,000 cells was determined using ultraviolet light with a violet filter (450/50 nm). Samples were analysed using FlowJo software.

Statistical analysis
Prism 7 (GraphPad) was used for statistical analysis of growth assay data by unpaired t-test of biological triplicate experiments. A P value of <0.05 was considered statistically significant.
Supporting information S1 Table. Oligonucleotide primer sequences used in this study. (PDF) S1 Fig. Construction of pT2A-DDI-1-cKO modified GFP. The loxPint and gfp fragments were amplified from pT2A-DDI-1cKO-complement. Overlapping PCR was performed to join the two fragments. The AflII and NheI restriction sites were introduced to facilitate further cloning. pT2A-DDI-1cKO was digested with XhoI. The digested fragment was ligated with lox-Pint-modified GFP fragment at the XhoI site. (PDF)