Genetic disruption of the Plasmodium falciparum digestive vacuole plasmepsins demonstrates their functional redundancy.

The digestive vacuole plasmepsins PfPM1, PfPM2, PfPM4, and PfHAP (a histoaspartic proteinase) are 4 aspartic proteinases among 10 encoded in the Plasmodium falciparum malarial genome. These have been hypothesized to initiate and contribute significantly to hemoglobin degradation, a catabolic function essential to the survival of this intraerythrocytic parasite. Because of their perceived significance, these plasmepsins have been proposed as potential targets for antimalarial drug development. To test their essentiality, knockout constructs were prepared for each corresponding gene such that homologous recombination would result in two partial, nonfunctional gene copies. Disruption of each gene was achieved, as confirmed by PCR, Southern, and Northern blot analyses. Western and two-dimensional gel analyses revealed the absence of mature or even truncated plasmepsins corresponding to the disrupted gene. Reduced growth rates were observed with PfPM1 and PfPM4 knockouts, indicating that although these plasmepsins are not essential, they are important for parasite development. Abnormal mitochondrial morphology also appeared to accompany loss of PfPM2, and an abundant accumulation of electron-dense vesicles in the digestive vacuole was observed upon disruption of PfPM4; however, those phenotypes only manifested in about a third of the disrupted cells. The ability to compensate for loss of individual plasmepsin function may be explained by close similarity in the structure and active site of these four vacuolar enzymes. Our data imply that drug discovery efforts focused on vacuolar plasmepsins must incorporate measures to develop compounds that can inhibit two or more of this enzyme family.

Malaria remains a public health problem of enormous magnitude in the tropical and subtropical regions of the world, annually afflicting an estimated 500 million people and killing nearly 2 million, mostly children (1,2). Of the four species of malaria parasites that infect humans, the most lethal one, Plasmodium falciparum, is becoming increasingly resistant to the available drugs, making it essential to pursue the development of new antimalarial compounds that act on essential parasite pathways unencumbered by current mechanisms of drug resistance (3). One validated pathway whose metabolic function is unique to the parasite is hemoglobin digestion, which results in the sequestration of free ferriprotoporphyrin IX as an inert, crystalline form (hemozoin) in the digestive vacuole (DV) 1 of the parasite.
Previous studies by others suggest that hemoglobin digestion is a semiordered process initiated in the DV by two aspartic proteinases named plasmepsins (plasmodium pepsins) I and II (PfPM1 and PfPM2, respectively). These enzymes can mediate the first cleavage of the Phe 33 -Leu 34 bond in the hinge region of the ␣ chain, causing the whole molecule to unravel in the acidic DV environment (4 -7). Genome sequencing and subcellular localization studies have identified two additional DV plasmepsins, a histoaspartic proteinase (PfHAP) and plasmepsin 4 (PfPM4) that are thought to be involved in hemoglobin degradation (8 -11). These enzymes share ϳ60% sequence identity with PfPM1 and PfPM2, including similar targeting proregions (11), and have been found localized to the DV by both immunoelectron microscopy (10) and co-purification with the DV (11). Their capacity to digest native hemoglobin is significantly less than that of PfPM1 or PfPM2, but both actively cleave globin as demonstrated in vitro (10). Cysteine proteases falcipain 2, 2*, and 3, plus the metalloprotease falcilysin, are believed to be involved in further degrading large hemoglobin fragments into short peptides and free heme (12)(13)(14) and may themselves be able to initiate hemoglobin degradation (15)(16)(17). The released toxic heme product is neutralized by the formation of coordinated hematin dimers that generate hemozoin, also known as malaria pigment (18 -20). Hemoglobin degradation provides an abundant source of amino acids that can be used by the parasite during protein assembly. In addition, this process may provide the means for the parasite to regulate intracellular osmolarity during its maturation and replication in the red blood cell (21)(22)(23). Plasmepsins have been thought to play a critical role in hemoglobin degradation; furthermore, the finding that aspartic protease inhibitors can kill malaria parasites (16, 19, 24 -27) has made them prime candidates for antimalarial drug development. However, these candidate targets remain unvalidated. To address whether these DV plasmepsins are critical, we performed gene disruption experiments and report our findings herein.

EXPERIMENTAL PROCEDURES
Parasite Cultivation, Synchronization, and Release from Erythrocytes-P. falciparum parasites were cultured as described (28,29) at 4% hematocrit in RPMI 1640 (Invitrogen) medium supplemented with 0.5% Albumax (Invitrogen), 0.225% sodium bicarbonate, and 0.01 mg/ml gentamycin. Parasites were synchronized using 5% sorbitol when they were mainly in the ring stage. The parasites used for RNA and protein preparations were collected at ϳ12-h intervals. The parasitemia and the synchronicity of each culture was determined by microscopic examination of Giemsa or Protocol Hema (Fisher)-stained smears.
FIG. 1. Genetic strategy for disruption of the four DV plasmepsins. A, schematic of the constructs (pHD-⌬PM1, pHD-⌬PM2, pHD-⌬PM4, or pHD-⌬HAP) designed to integrate into the chromosomal copy of the targeted plasmepsin gene via a single crossover event. The generic plasmepsin coding region is represented by an open box with stippled bands. The arrow designates the direction of transcription. Integration of the pHD-⌬PM constructs produces two nonfunctional fragments of the targeted gene. The 5Ј-fragment lacks the gene region encoding Psi loop 2 at the C-terminal end of the mature enzyme, whereas the 3Ј-fragment lacks the 5Ј-end encoding the targeting peptide, the proregion, and the catalytic loop 1. Plasmid integration into the plasmepsin genes was detected by PCR using primers p1/p4 and p3/p2 (numbered arrows). PCR with primers p1/p2 detected the wild-type gene locus, whereas primers p3/p4 detected plasmid DNA. B-E, these indicate the sizes and origins of restriction fragments containing each DV plasmepsin gene from the parental line (upper representation) as well as the predicted sizes for the knockout locus (lower representation). A, AccI; Bm, BsmI; Bt, BstB1; M, MfeI; P, PacI; S, StyI. B, AccI digestion of the PfPM1 locus yields a 1.5-kb fragment containing the coding sequence for this gene. Gene disruption was predicted to produce 4.5-and 1.8-kb fragments. StyI was predicted to produce 9.2-and 15.8-kb bands from the uninterrupted and disrupted PfPM1 loci, respectively. C, for PfPM2, AccI digestion was predicted to yield a 5.7-kb fragment for Dd2 versus 8.2-and 2.3-kb bands for the disrupted locus. Insertion of a single copy of the knockout plasmid into PfPM2 was predicted to increase the size of the BsmI fragment from 5.3 kb in Dd2 to 11.8 kb in the knockout line. D, for PfHAP, AccI was predicted to yield a 5.8-kb band for Dd2 versus 5.9-and 4.6-kb bands for the PfHAP knockout line. Insertion of a single plasmid copy was predicted to increase the size of the BstB1 fragment from 7.4 kb in Dd2 to 14.0 kb in the knockout line. E, for PfPM4, PacI was predicted to yield a 2.0-kb fragment for Dd2 versus 5.5and 3.0-kb bands for the knockout, with insertion of a single plasmid copy resulting in an increase in size of the MfeI fragment from 11.3 kb in Dd2 to 17.0 kb in the knockout line.
Probes for Southern and Northern Blot Analysis-These were prepared from the 0.7-kb PCR product used to prepare each knockout construct by random primer labeling (Stratagene) with [␣-32 P]dATP. These PCR products were found not to cross-hybridize with one another when used as probes under stringent hybridization and wash conditions.
Southern and Northern Blot Analysis-Parasites were prepared free of excess erythrocytes by treatment with 0.05% saponin and washed twice with phosphate-buffered saline. Genomic DNA was extracted using TELT lysis buffer (50 mM Tris (pH 8.0), 62.5 mM EDTA (pH 8.0), 2.5 M LiCl, 4% Triton X-100) (34). The DNA was further purified by phenol/chloroform/isoamyl alcohol (25:24:1, Fisher) deproteination, and isopropyl alcohol precipitation. Following restriction enzyme digestion, DNA samples were separated on 1% agarose gels and blotted onto Hybond N ϩ membranes (Amersham Biosciences). Total parasite RNA was extracted using Trizol (Invitrogen) and purified by isopropyl alcohol precipitation. Quantification used both absorbance at 260 nM and agarose gel electrophoresis followed by ethidium bromide staining. For Northern blots, ϳ6 g of total RNA was loaded per well, resolved on 1% formaldehyde-agarose gels, blotted onto Hybond N ϩ membranes, and hybridized with 32 P-labeled DNA probes.
Isolation of DV-These were prepared from sorbitol-synchronized cultures of P. falciparum trophozoites, using a method from Saliba et al. (35) with major modifications. Briefly, 100 ml of culture containing trophozoites at 5-10% parasitema was treated with 0.05% saponin, and the free parasites were immediately collected by centrifugation. Parasites were washed twice by resuspension in ice-cold wash buffer (10 mM Tris-HCl (pH 7.5), 250 mM sucrose) followed by centrifugation. Washed pellets were resuspended in 1 ml of ice-cold wash buffer, transferred to 1.5-ml microcentrifuge tubes, and triturated four times through a 27gauge needle (4 s per trituration). Samples were centrifuged at 13,800 ϫ g at 4°C for 2 min. Supernatants were discarded, and pellets were resuspended in 1 ml of uptake buffer (2 mM MgSO 4 , 100 mM NaCl, 25 mM HEPES, 25 mM NaHCO 3 , 5 mM Na 3 PO 4 (pH 7.4)) to which was added 50 l of 1 unit/l DNase. After 10 min of incubation at 37°C, samples were centrifuged at 13,800 ϫ g at 4°C for 2 min. Supernatants were discarded, and the pellets were resuspended in 0.1 ml of ice-cold wash buffer and then mixed with 1.3 ml of ice-cold 42% Percoll prepared in 0.25 M sucrose, 1.5 mM MgSO 4 (pH 7.4). Mixtures were triturated two times through a 27-gauge needle (9 s per trituration) and centrifuged at 13,800 ϫ g at 4°C for 10 min. The DV appeared as a dark band in the bottom 50 l of the Percoll gradient. These were collected and transferred to clean 1.5-ml microcentrifuge tubes and washed twice with ice-cold wash buffer, involving centrifugation at 13,800 ϫ g at 4°C for 2 min and removal of the supernatant each time. After the second wash, the pellet was resuspended in 100 l of IEF sample buffer (9 M deionized urea, 4% CHAPS, 60 mM DTT, 0.8% ampholytes (pH 3-10)) and 10 l of 10ϫ protease inhibitor mixture (Sigma). Samples were stored at Ϫ80°C. Prior to use, samples were thawed, vortexed, and centrifuged at 16,000 ϫ g at room temperature for 5 min. Supernatants were used in two-dimensional gel electrophoresis as described below.
Two-dimensional PAGE-This was performed by using a Multiphor II Electrophoresis Unit equipped with a MultiDrive EPS 3501XL gradient power supply (Amersham Biosciences). Precast Immobiline Drystrips (IEF strips, pH 4 -7 linear, 180 ϫ 3 ϫ 0.5 mm) were rehydrated overnight (in a re-swelling cassette (Amersham Biosciences)) with 50-l DV extracts and brought up to 125 l with re-swelling buffer (8 M urea, 2% (w/v) CHAPS, 2% (v/v) IPG buffer (Amersham Biosciences), and 0.002% bromphenol blue). Just prior to use, DTT was added at a final concentration of 3 mg/ml. Rehydration buffer and stock DTT (62.5 mg/ml) were stored at Ϫ20°C. Rehydrated IEF strips were isoelectrically focused at 15°C under low viscosity oil (Fisher) with a gradient voltage of 0 -200 V for 1 min, 200 -3,500 V for 1.5 h, and a constant voltage of 3,500 V for 1.5 h. The IEF strips were equilibrated for 15 min in 10 ml of SDS equilibration buffer (50 mM Tris-HCl (pH 8.8), 6 M urea, 30% (v/v) glycerol, 2% (w/v) SDS, and 0.002% (w/v) bromphenol blue), and DTT was added at a final concentration of 10 mg/ml immediately before use. The second dimension was SDS-PAGE, performed according to the method of Laemmli (36) with precast 10% polyacrylamide gels (Bio-Rad). The equilibrated IEF strips were each placed on a preparative well and sealed using 1% agarose plus 0.002% (w/v) bromphenol. After electrophoresis the two-dimensional gels were silver-stained by using a MALDI-TOF-compatible silver-staining protocol (37). Spots corresponding to each plasmepsin were excised and confirmed by MALDI-TOF analysis (38) (performed by Kendrick Laboratories Inc., Madison, WI). Gel images were documented with an AlphaImager IS-1000.
SDS-PAGE and Western Blot Analysis-Parasites were lysed in 1ϫ RIPA Lysis buffer (Upstate Biotechnology Inc., Lake Placid, NY) containing the protease inhibitors pepstatin (2 g/ml), leupeptin (1.5 g/ ml), and phenylmethylsulfonyl fluoride (1 mM). Lysates were then treated with an equal volume of 2% SDS as described (39), and their concentrations were adjusted to ensure equal protein loading. Bio-Rad sample buffer was added to the solubilized protein to a final concentration of 1ϫ prior to boiling for 5 min and loading onto 12% polyacrylamide gels (39). After electrophoresis, proteins were transferred to nitrocellulose (Bio-Rad). Membranes were hybridized with primary antibodies as described (10), followed by horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit IgG (1:10,000 dilution, Pierce). Immunoreactivity was detected using the SuperSignal chemiluminescent substrate (Pierce).
Electron Microscopy-For electron microscopy, infected red blood cells or DV preparations were fixed with 2.5% glutaraldehyde in 0.05 M phosphate buffer (pH 7.4), containing 4% sucrose, for 30 min at room temperature followed by 90 min at 4°C. Cells were then postfixed in 1% osmium tetroxide for 1 h. After a 30-min en bloc stain with 1% aqueous uranyl acetate, cells were dehydrated in ascending concentrations of ethanol and embedded in Epon 812. Ultrathin sections were stained with 2% uranyl acetate in 50% methanol and lead citrate and examined using a Zeiss CEM902 electron microscope (Oberkochen, Germany).
Southern Blot Analysis-To confirm the integration of the separate constructs into the plasmepsin gene loci, we carried out Southern blot analysis with gene-specific probes (Fig. 2). For every knockout clone, the banding patterns were consistent with integration of a single plasmid copy leading to disruption of the targeted plasmepsin. Details are provided in the Fig. 1 legend. In view of the identical results obtained between the independent clones, for HAP and PfPM4, we chose to pursue our investigations using a single representative clone for each locus, namely Tx1.A3, Tx2.B6, Tx3.C3, and Tx4.F8. These are referred to hereafter simply as Tx1, Tx2, Tx3, and Tx4.
Northern Blot Analysis-To determine the effect of gene disruption on the expression and stability of the altered mRNAs of each DV plasmepsin in the respective knockout clones, we performed Northern blot analyses by using genespecific probes. For this, total RNA was collected at four time points from synchronized parasite clones. The normal size of the mRNA for PfPM1 is 4.0 kb, for PfPM2 is 2.4 kb, for PfHAP is 2.4 kb, and for PfPM4 is 4.0 kb (Fig. 3). None of the four knockout mutants expressed a comparably sized transcript from the disrupted gene locus in the asexual cycle. Instead, all four expressed an abnormally large RNA present in nearly all stages of the asexual cycle (Fig. 3). This product also hybridized to a probe prepared from the pHD22Y plasmid vector (without plasmepsin sequence; data not shown). Thus, the large mRNA transcript had both plasmepsin and plasmid sequences. The significance of the slight fluctuations in the intensity of this band from different stages of various knockouts is uncertain. Two other transcripts also hybridized with the plasmepsin probes, one larger and one slightly smaller than the 2.37-kb marker. These are most prominent on the Tx3 RNA blot, which had been washed at moderate stringency. Corresponding bands may be seen at lower intensity on Tx1, Tx2, and Tx4 blots washed at high stringency (Fig. 3, A, B, and D). All of these abnormal RNAs that hybridized with the plasmepsin probe also hybridized to a probe prepared from the pHD22Y vector alone, indicating transcription from the chromosome-integrated truncated plasmepsin plasmid sequences. For all knockout clones, additional Northern blot analyses revealed that individual gene disruptions were not accompanied by changes in transcript sizes of the other DV plasmepsins. Representative data are shown in Fig. 3C, where RNA from Tx2 was included as a control to demonstrate the normal size of the PfHAP mRNA (which was the same for Tx2 and Dd2).
Two-dimensional PAGE and Western Blot Analysis-To compare the plasmepsins present in DV preparations from Dd2 and each of the four knockout clones, we used two-dimensional gel electrophoresis. The first dimension was isoelectric focusing performed on an Immobiline gel strip (pH 4 -7), and the second dimension was electrophoresis on a 10% SDS-PAGE slab gel. As shown in Fig. 4, PfPM1, PfPM2, PfHAP, and PfPM4 were all identified in Dd2 DV preparations. The identity of each of these spots in Dd2 was confirmed by MALDI-TOF mass spectroscopic analysis. For each knockout clone, the expected individual plasmepsin spot was absent. We note that whereas these results demonstrate the presence of the expected individual plasmepsins in these DV preparations, the overall protein complexity is somewhat overestimated because of the co-purification of some free merozoites as detected by electron microscopy (data not shown).
Because each knockout construct was designed to generate two truncated gene fragments, and thereby potentially result in two partial protein fragments with differing molecular masses and isoelectric points, whole parasite preparations of each knockout line were further examined by Western blot analysis using antibodies specific for each plasmepsin. No form of PfPM1 was detected by this technique in ring, early trophozoite, late trophozoite, or schizont stages of Tx1, for which proteins in the range of 14 -200 kDa could be observed by Coomassie staining (Fig. 5A). Similarly, no forms of PfPM2, PfHAP, and PfPM4 were detected by the antibodies employed in blots of Tx2, Tx3, and Tx4, respectively, over this same size range (data not shown). These analyses demonstrated the re-activity of each antibody against the corresponding plasmepsin in Dd2 (Fig. 5B). As a control, anti-PfPM2 antibodies were reacted with blots of the same extracts, demonstrating the presence of PfPM2 in all lines except for Tx2. In this instance, anti-PfPM1 antisera was used as a control to demonstrate the presence of PfPM1.
Comparative Growth in Culture-During the propagation of these recombinant lines, we observed a noticeable effect of certain plasmepsin gene disruptions on rates of growth. To study this, we sorbitol-synchronized our knockout clones and the parental line, initiated cultures at 0.5% parasitemia, and quantified the increase in parasitemia over 96 h. In similar experiments, the parasite rate of multiplication was followed for 2 weeks, adjusting the parasitemia every 48 h to begin at 1%. Both types of assays revealed a 30 -50% decrease in parasite multiplication in Tx1 and Tx4 relative to Dd2. This prompted us to analyze growth rates from parasitemia measurements recorded over an 18-month period, which should minimize any short term differences arising from sorbitol treatment or culture care. Data points were retained only when parasitemias were calculated at the beginning and end of a 48-h generation period, and measurements were only selected from periods when parasites were growing robustly, typically in the range of 0.7-6.0% parasitemia. Measurements meeting these criteria totaled 27-45 independent data points over 18 months of culture. These revealed a statistically significant 30 -35% decrease in growth rates of the Tx1 and Tx4 lines relative to Dd2 (Table I). A slight drop in growth rates was also observed for Tx3, although this was not statistically significant, whereas Tx2 grew at least as well as Dd2 and over some periods appeared to grow even faster.
Knockout Clone Morphology-Each knockout clone and Dd2 was examined by electron microscopy and compared at various stages of their asexual growth in erythrocytes, with particular attention paid to the morphology of the DV (Fig. 6). For Tx1, Tx2, and Tx3, no significant differences were noted in either the size or organization of the DV or the arrays of hemozoin crystals when compared with Dd2. However, in numerous images of Tx4, there was a notable accumulation in the DV of smaller more electron-dense, single-membrane vesicles (Fig.  6E). Yet in other Tx4 images the DV and their arrays of hemozoin crystals appeared normal, indicating that the phenotype shown was not observed for all cells. From examining all other subcellular structures visible by electron microscopy, the knockout mutants appeared normal except for Tx2. For this clone, the mitochondria were enlarged in approximately onethird of the mature-stage parasites (Fig. 6C).

DISCUSSION
Two early observations have assigned key roles to aspartic proteinases in essential metabolic functions critical to the survival of the malaria parasite (reviewed in Ref. 24). The first is that pepstatin, a classical aspartic proteinase inhibitor, kills malaria parasites in culture (25); and the second is that PfPM1, localized in the DV by immunolabeling (40), is efficient in vitro at initiating the degradation of native hemoglobin at pH 5, which corresponds to the estimated DV pH (6). These observations stimulated the proposal that hemoglobin digestion occurred via a semi-ordered pathway initiated by PfPM1, which cleaved hemoglobin in the hinge region of the ␣ chain at Phe 33 -Leu 34 , thereby unraveling the molecule and rendering it accessible for further proteolysis in the acidic environment of the DV. Cysteine proteases and other plasmepsins (e.g. PfPM2) were hypothesized to contribute at later stages of the degradation pathway. Support for this model came from studies with the inhibitor SC-50083 (Roche Applied Science), which was comparable in efficiency to pepstatin in blocking parasite growth in vitro (40). This compound, which also blocked the majority of hemoglobin digestion in DV preparations, was reported to be more efficient in inhibiting PfPM1 than PfPM2 (5). Based on these observations, several projects have been undertaken to identify specific inhibitors for PfPM1 and PfPM2 (41)(42)(43)(44)(45)(46). A search for inhibitors has also been performed with the Plasmodium vivax DV plasmepsin, PvPM4, for which active recombinant protein was available (47). Elucidation of the P. falciparum genome has now led to the discovery of 10 plasmepsin genes (24), including the 4 investigated here that encode the DV enzymes (10,48). Our comparative analyses of the available genome data from other Plasmodium spp. indicate that whereas all six of the plasmepsins located outside the DV (10) are present in the primate, rodent, and chicken malaria parasite species, only a single DV plasmepsin is present (11). Phylogenetic analysis clearly places this DV plasmepsin as being an ortholog of PfPM4 (11). The only exception is Plasmodium reichnowi, which infects chimpanzees and gibbons and which is very closely related to P. falciparum. This species harbors all four DV plasmepsin paralogs. 2 Our present studies enable us to ask whether any of the specialized functions that may have been acquired for PfPM1, PfPM2, PfHAP, or PfPM4, during the course of their evolution (presumed to have begun with sequence duplication and divergence from PfPM4 (11)), are essential to the survival of P. falciparum asexual blood stages. Identifying one or more as being essential would validate them as rational drug design targets. The most important finding of this study is that P. falciparum asexual parasites can propagate in vitro with any three of the four DV plasmepsins still functioning, without presenting a highly deleterious phenotype. Our data establish that no single DV plasmepsin is essential to the P. falciparum intracellular growth. This includes PfPM1, which had been thought to play the key role in initiating hemoglobin digestion, yet cannot be so on the basis of our data (5). Our results clearly suggest redundancy in the functions performed by the DV plasmepsins. The function of the inactivated gene product may be performed by either one or more of the other three plasmepsins, or possibly the cysteine proteases falcipain 2, 2*, and 3 that are also present in the DV (17,49,50). Parasites in which falcipain 2 has been knocked out also lack any obviously deleterious phenotype, suggesting that the cysteine proteases along the hemoglobin digestion pathway also have redundant functions (51). The observation that the falcipain 2-knockout parasites become highly sensitive to pepstatin (51) suggests that its loss of function was compensated for by an increased dependence on plasmepsin activity and raises the possibility that cooperative, partially redundant roles exist for these two different mechanistic classes of DV proteases.
Data from Bozdech et al. (52) and Le Roch et al. (53) indicate that PfPM1 and PfPM4 are transcribed early in the asexual cycle and the steady-state level of the mRNA falls during the latter half of the cycle, whereas for PfPM2 and PfHAP, the FIG. 5. Western blot analysis of whole parasite lysates of Dd2 and plasmepsin knockout clones. Total parasite protein extracts were prepared from each knockout clone at four stages in the asexual development cycle: rings, early trophozoites, late trophozoites, and schizonts. Equal amounts of protein from each knockout line at each stage, and from Dd2 schizonts, were assayed. A, protein extracts of Tx1 and control, Dd2, probed with anti-PfPM1 antibody demonstrated the presence of the ϳ37-kDa PfPM1 protein in Dd2; however, no protein or partial protein fragment was detected in the Tx1 line at any developmental stage. Similar results were found for Tx2, Tx3, and Tx4 samples probed with antibodies specific for PfPM2, PfHAP, and PfPM4, respectively. B presents data from a restricted section of those blots probed with antibodies against two different plasmepsins. In each case the antibodies specific for the plasmepsin targeted by the knockout strategy bound to an ϳ37-kDa band in Dd2 but failed to react with bands of any size in any stage of development of the knockout clones.  amount of transcript present is low in the first half of the cycle and rises prominently in the second half. This suggests that perhaps PfPM1 and PfPM4 could complement for each other, whereas PfPM2 and HAP could functionally complement as the other pair. Most interestingly, Tx4 displayed in some, although not all, parasites a noticeable accumulation of multiple smaller vesicles inside the DV, which suggests that perhaps PfPM4 plays a role in vesicle-mediated transport of hemoglobin into the DV and subsequent elimination of these discrete vesicles, either by their incorporation into the DV membrane or their proteolytic degradation. Most interestingly, the loss of PfPM4 in Tx4 did not eliminate the production of hemozoin. A subtle phenotype was also observed for the PfPM2 knockout, which in about a third of the parasites displayed unusually enlarged mitochondria. We have yet to be able to define this phenotype further. A pronounced detrimental effect on growth rates was also observed for both the PfPM1 and PfPM4 knockout lines, which was statistically significant in comparison with the parental Dd2 line. This defect therefore indicates that these two plasmepsins, although not essential, nonetheless are required for normal robust rates of parasite propagation in vitro. Ongoing studies of the molecular changes together with any subtle phenotypes connected with the loss of function of these DV plasmepsins should provide further insight into their functions and how that relates to hemoglobin degradation and antimalarial drug susceptibility.
The implications of our work are clear: rational anti-plasmepsin drug design strategies must be revised to include approaches to identify compounds capable of inhibiting at least two if not all four of these enzymes, without toxicity to the host. This will only be feasible if these enzymes share structurally very similar active sites that can be bound by pantothenatespecific inhibitors. Some evidence for this being the case comes from structural studies that reveal close active site similarity between PfPM2 and the P. vivax ortholog of PfPM4 (54, 56 -58). 3 Most importantly, these DV plasmepsins differ sufficiently from the most closely related human aspartic proteinase, cathepsin D (59), making it potentially feasible to develop selective anti-plasmodial inhibitors (41,44). Indeed, the practicality of a strategy to design inhibitors capable of inhibiting the whole family of DV plasmepsins has been reported recently (55). These genetic, structural, and biochemical data therefore all point toward a close degree of functional and structural conservation that could be exploited by the identification of plasmepsin inhibitors effective against multiple DV plasmepsins, making this a key consideration for subsequent inhibitor screening against this family of enzymes.