Biosynthesis and Maturation of the Malaria Aspartic Hemoglobinases Plasmepsins I and II*

During the intraerythrocytic stage of infection, the malaria parasite Plasmodium falciparum digests most of the host cell hemoglobin. Hemoglobin degradation occurs in the acidic digestive vacuole and is essential for the survival of the parasite. Two aspartic proteases, plasmepsins I and II, have been isolated from the vacuole and shown to make the initial cleavages in the he- moglobin molecule. We have studied the biosynthesis of these two enzymes. Plasmepsin I is synthesized and processed to the mature form soon after the parasite invades the red blood cell, while plasmepsin II synthesis is delayed until later in development. Otherwise, biosynthesis of the plasmepsins is identical. The proplas- mepsins are type II integral membrane proteins that are transported through the secretory pathway before cleavage to the soluble form. They are not glycosylated in vivo , despite the presence of several potential glyco- sylation sites. Proplasmepsin maturation appears to require acidic conditions and is reversibly inhibited by the tripeptide aldehydes N-acetyl - L - leucyl - L - leucyl -nor-leucinal and N-acetyl - L - leucyl - L - leucyl -methional. These compounds are known to inhibit cysteine proteases and the chymotryptic activity of proteasomes but not aspar- tic proteases. However, proplasmepsin processing is not blocked by other cysteine protease inhibitors, nor by the proteasome inhibitor lactacystin. Processing is also not blocked by aspartic protease inhibitors. This inhibitor profile suggests that unlike most other aspartic

Nearly half the world's population lives in malaria endemic areas. Every year several hundred million people are afflicted and over two million people die from this disease, mainly children (1). Malaria is increasing as a global problem because parasites have developed resistance to drugs used for prophylaxis and treatment at an alarming rate. Thus, there is an urgent need to identify new drug targets (2). Plasmodium falciparum, the organism responsible for the most lethal form of malaria, is an obligate intracellular parasite that resides for most of the course of infection in the host erythrocyte. Hemoglobin degradation is essential for the survival of the parasite and agents that disrupt the hemoglobin catabolic pathway are considered good prospects for antimalarial chemotherapy (3).
During the organism's trophozoite stage most of the host cell hemoglobin is degraded (4 -6), supplying the parasite with amino acids for protein synthesis and energy metabolism (7). Plasmodium ingests hemoglobin by means of the cytostome, an invagination that spans the parasite plasma membrane and the parasitophorous vacuolar membrane (a red cell-derived structure with which the parasite surrounds itself as it invades). The cytostome fills with host cell cytoplasm and pinches off to form transport vesicles. These double membrane-delimited vesicles loaded with hemoglobin fuse with the digestive vacuole and empty their contents. The digestive vacuole, an acidic proteolytic compartment with a pH between 5.0 and 5.4 (8,9), is the site of hemoglobin degradation and heme detoxification (10). Two aspartic proteases, plasmepsins I and II and one cysteine protease, falcipain, have been purified from isolated digestive vacuoles and characterized (11,12). A pepstatin inhibitable activity associated with vacuolar membranes has also been partially characterized and may result from a third aspartic protease (13). Studies with purified plasmepsins and falcipain suggest that hemoglobin degradation is an ordered process. Plasmepsins I and II initiate cleavage of the native hemoglobin tetramer, while falcipain digests denatured substrate (12,14). A peptidomimetic inhibitor of plasmepsin I blocks hemoglobin degradation in vitro and kills parasites in culture (15). Also, cysteine protease inhibitors block hemoglobin degradation, kill cultured parasites, and eradicate parasitemia in a murine malaria model (16,17). These data suggest that targeting hemoglobin digestion could be an effective antimalarial strategy.
The genes for plasmepsins I and II were cloned (18,15) and predicted to encode 51-kDa proteins that are 73% identical and are about 35% homologous to mammalian renin and cathepsin D (19). Typically aspartic proteases are zymogens that are synthesized as pre-proproteins. The pro-piece, characteristically less than 50 amino acids, is removed during protein maturation (20). In contrast, plasmepsin I and II pro-pieces are predicted to be 123 and 124 amino acids, respectively (Fig. 1). Cleavage of the putative pro-piece would yield mature proteins of 37 kDa, in agreement with the size of the isolated proteins. The long pro-pieces of the plasmepsins are distinguished by the absence of a signal sequence. Instead, 65 amino acids upstream of the cleavage site of both plasmepsins is a hydrophobic stretch of 21 amino acids which was proposed to serve as a signal anchor sequence (15). The hydropathy plots of the proplasmepsins are characteristic of membrane proteins (15,18).
We report the results of biosynthesis studies which indicate that proplasmepsins are integral membrane proteins in the secretory pathway. Instead of the typical aspartic protease autoactivation, an acid pH-dependent ALLN/ALLM inhibitable activity appears to be responsible for cleavage of both proplas-mepsins. Inhibition of proplasmepsin maturation is a new target in the hemoglobin proteolysis pathway that may provide a selective and efficient means to starve the malaria parasite.
Parasite Culture-The P. falciparum clone HB3 (a gift of Dr. W. Trager, Rockefeller University) was grown by the method of Trager and Jensen (21) using human plasma (22). The parasites were grown in A ϩ erythrocytes at 2-5% hematocrit and 5-15% parasitemia. Synchrony was maintained by sorbitol treatment (23).
Recombinant Plasmids-Recombinant plasmids were made using polymerase chain reaction performed in a TC480 thermal cycler with Ampli-Taq (Perkin-Elmer). All reactions were preceded by a single 5-min 94°C denaturation step and followed by a 5-min 72°C final extension step. Proplasmepsin I was amplified by polymerase chain reaction using plasmid (4.4) containing the full-length plasmepsin I cDNA (15) as template and oligonucleotide primers (sense) 5Ј-TCGAG-GATCCATAAAGATGGCTTTATCAATTAAAGAAGATTT-3Ј with (antisense) 5Ј-TCGAGGATCCTTACAATTTTTTTTTGGCAAGGGC-3Ј for 30 cycles (94°C, 1 min, 50°C, 1 min, and extension 72°C, 1 min). The product was ligated into a TA-Cloning vector following the manufacturer's instructions (Invitrogen). Proplasmepsin II was amplified using 12 ng of P. falciparum genomic DNA as template and oligonucleotides designed using published plasmepsin II sequence (18). Polymerase chain reaction of (sense) 5Ј-GCATAAGCTTACAAAATGGATATTA-CAGTAAGAGAACATGAT-3Ј and antisense 5Ј-TCGAGAATTCTCT-TATATAATTATGTTTTTCCTTTG-3Ј generated a product that was digested with EcoRI and HindIII and ligated into pBSSKIIϩ (Stratagene). Plasmepsins I and II used in competition experiments to determine antibody specificity were purified from Escherichia coli transformed with the PET vector expression plasmids described previously (24).
In Vitro Translation and Membrane Translocation of Parasite Proteins-Capped PM I and PM II RNA were prepared using linearized plasmid DNA and T7 polymerase Message Machine kit (Ambion) following the manufacturer's instructions. Transcript was added to nuclease-free rabbit reticulocyte lysate (Promega) in the presence of [ 35 S]methionine/cysteine and 10 g of brewer's yeast tRNA per 25 l of translation reaction. Canine pancreas microsomes (1.8 l) were added to certain in vitro translation reactions as specified in the text.
Antibody Production and Immunoprecipitation-Antibody to plasmepsin I (Ab 574) was generated as described previously (15). Rabbit antiserum against plasmepsin II (Ab 737) was made using a multiple antigen peptide, HSHSSNDNIELNDFQNIMFYGDAEV, derived from the mature NH 2 terminus of native plasmepsin II (12) prepared by Research Genetics.
Pulse-Chase Experiments-Pulse labeling experiments were performed with synchronized cultures of trophozoite stage parasites (between 24 and 36 h after invasion). The cultures were washed one time in RPMI prepared without methionine and cysteine (Met Ϫ Cys Ϫ ), resuspended in the same medium at a density of 2.0 ϫ 10 8 /ml with [ 35 S]methionine/cysteine (350 Ci/ml) and incubated at 37°C for 10 min (3% O 2 , 3% CO 2 ). The cultures were chased following the addition of complete RPMI and 100 g/ml cycloheximide, by incubation for different times depending on the experiment but typically between 0 and 3 h. Inhibitor experiments were performed in the same manner except that cultures were preincubated with inhibitors for 45-60 min before metabolic labeling commenced (except for ALLN and ALLM which were first added during labeling). The inhibitor being tested was added to both labeling and chase media. Free parasites were harvested by saponin lysis (11) and stored at Ϫ70°until analyzed.
ATP requirements for processing in vivo were determined using a variation on these culture conditions. Trophozoites were incubated in glucose-free RPMI medium for 2 h, washed once in RPMI minus glucose, methionine and cysteine, collected by centrifugation, and labeled in the same medium which included 350 Ci/ml [ 35 S]methionine/cysteine for 10 min at 37°C. Cycloheximide (100 g/ml) was added to stop translation and the medium was adjusted to 20 mM deoxyglucose and 0.5 mM 2,4-dinitrophenol. Cells were incubated for 2 h, harvested, and stored frozen.
Glycosidase Treatment-Proplasmepsins I and II were prepared by in vitro translation in the presence of microsomes described above or by pulse labeling P. falciparum trophozoites for 10 min in RPMI (Met Ϫ Cys Ϫ ) with [ 35 S]methionine/cysteine (350 Ci/ml). Immunocomplexes were collected as described above and resuspended in denaturing buffer (0.5% SDS, 1% ␤-mercaptoethanol), then heated for 10 min at 100°C. The samples were split into four equal portions. One portion was digested with 500 units of endoglycosidase H (Endo H) in 50 mM sodium citrate (pH 5.5) for 3 h at 37°C. A second portion was treated identically except that Endo H was not added (mock). The third portion was digested with 500 units of N-glycosidase F (PNGase F) in 50 mM sodium phosphate (pH 7.5), 0.1% Nonidet P-40. The final portion was untreated. Proplasmepsins I and II were also translated without microsomes to generate unglycosylated products which were used both for size comparison and as controls for glycosidase digestions. Endo H and PNGase F have no activity on unglycosylated substrates (not shown). The products were analyzed by 12% SDS-PAGE with fluorography.
Association of Proplasmepsins I and II with Membranes-Trophozoites (5 ϫ 10 8 ) were labeled with 150 Ci/ml [ 35 S]methionine/cysteine for 3 h at 37°C, BFA (5 g/ml) was added and labeling continued for 2 h. This method generated roughly equal amounts of labeled proplasmepsins and mature plasmepsins. Free parasites were harvested as above, lysed in 10 mM phosphate buffer (pH 7.1), sonicated, and split into three parts. One part served as a control for the total amount of processed and unprocessed plasmepsins I and II in the sample, the other two parts were centrifuged for 214,000 ϫ g-h, 4°C. The resulting membrane fractions were extracted with either phosphate-buffered saline or sequentially with 0.5 M KCl and 100 mM Na 2 CO 3 (pH 11.0). All extractions were for 30 min on ice followed by centrifugation for FIG. 1. Schematic of plasmepsins I and II. The proforms of both enzymes are predicted from the nucleotide sequence to encode 51-kDa proteins (18,15). 37 amino acids from the predicted initiator methionine of proplasmepsin I and 38 amino acids from the initiator methionine of proplasmepsin II is a 21-amino acid hydrophobic stretch predicted to be a signal anchor sequence (filled box). Both enzymes are processed to a mature form of approximately 37 kDa. The NH 2 -terminal sequences were determined from plasmepsins I and II isolated from digestive vacuoles (12). The cleavage site and the 4 adjacent amino acids (P4, P3, P2, P1, P1Ј, P2Ј, P3Ј, P4Ј) of both proteins is shown. The arrow indicates an autocatalytic cleavage site used by plasmepsin II expressed in E. coli (33). This site is 12 amino acids upstream from the cleavage site determined for native plasmepsin II. The active site aspartic acids of both plasmepsins are denoted by asterisks (*). 214,000 ϫ g-h. Membrane fractions were solubilized with 1% SDS and immunoprecipitated along with supernatants and the unfractionated control. Immunoprecipitated complexes were analyzed by SDS-PAGE with fluorography.
Membrane orientation of proplasmepsins I and II was determined by labeling 5 ϫ 10 8 trophozoites in 150 Ci/ml RPMI (Met Ϫ Cys Ϫ ) in the presence of BFA (5 g/ml). Cells were placed on ice for 15 min, collected by centrifugation at 2,000 ϫ g for 15 min, washed twice in cold phosphate-buffered saline, and stored at Ϫ70°C. The pellet was resuspended gently in cold phosphate-buffered saline in the presence of protease inhibitors (10 M pepstatin, 10 M E-64, 1.0 mM 1,10-phenanthroline, 100 M ALLN, 1 mg/ml aprotinin) on ice (25). The lysate that resulted from this treatment was adjusted to 5 mM CaCl 2 and divided into three parts: 1) untreated; 2) 1 or 2 mg/ml proteinase K; 3) 1 or 2 mg/ml proteinase K plus 0.3% Triton X-100, and the lysates were incubated on ice for 60 min. Proteinase K activity was inhibited by the addition of 10 mg/ml PMSF. The samples were immunoprecipitated as above after two preclearings with protein A-Sepharose. Analysis of immunocomplexes was by 12% SDS-PAGE with fluorography.
Expression of Plasmepsins I and II during the Intraerythrocytic Cycle-Parasites were carefully synchronized by sorbitol treatment (23) and cultured at 2% hematocrit to 3-5% parasitemia. The resulting cultures (approximately 10 8 parasites) were labeled at specified times after invasion with 50 Ci/ml [ 35 S]methionine/cysteine for 90 min at 37°C. Ring-stage cultures (5 h after invasion) were labeled in the same manner in the presence of ALLN (50 M) or bafilomycin A 1 (1.3 M). Labeled parasites were harvested by saponin lysis and immunoprecipitated in series with Ab 737 and Ab 574 using the conditions described above.

RESULTS
Proplasmepsins Are Secretory Proteins That Are Processed with Similar Kinetics-To address whether the proplasmepsins are made as membrane proteins and to examine how the proteins are processed in vivo, biosynthesis of both proteins was followed using specific antibodies. A plasmepsin I-specific antibody (Ab 574) was prepared against the first 224 NH 2terminal amino acids of the mature enzyme. The PM II antibody (Ab 737) was prepared against a 22-amino acid peptide corresponding to the mature NH 2 terminus of plasmepsin II. To assess the specificity of Ab 574 (PM I) and Ab 737 (PM II), proplasmepsins I and II radiolabeled during in vitro translation were immunoprecipitated with both antibodies. No crossreactivity was observed (Fig. 2). Immunoprecipitations performed in the presence of competing unlabeled plasmepsins I or II also indicate that the antibodies are specific for their respective proteins.
To examine the maturation of the plasmepsins, intraerythrocytic P. falciparum trophozoites were pulse-labeled with [ 35 S]methionine/cysteine for 10 min, followed by a chase in unlabeled medium with cycloheximide to prevent new protein synthesis. Processing was monitored at specific time intervals. As predicted, proplasmepsins I and II migrate at about 51 kDa (Fig. 3). The time course of processing for both proteins is virtually identical. A single cleavage occurs generating mature proteins of 37 kDa with a t1 ⁄2 of about 20 min. Processing is complete by 60 min and the proteins are stable for at least 3 h after synthesis.
The mechanism of protein secretion by Plasmodium is not well characterized (26 -28). As is seen for other eukaryotic cells (29), the addition of brefeldin A (BFA) to cultures impairs protein secretion by Plasmodium, resulting in a significant reduction in protein export (30). However, some proteins are secreted even in the presence of BFA. When BFA (5 g/ml) was added to trophozoite cultures during a 3-h metabolic labeling, maturation of proplasmepsins I and II was completely blocked. In the absence of BFA, processing proceeded normally and mature proteins accumulated (Fig. 4A). Inhibition with BFA suggests that the plasmepsins are secretory proteins that are transported through the Golgi apparatus. Proplasmepsin processing is likely to occur in a downstream compartment.
Plasmodium lacks a morphologically well defined Golgi apparatus. Earlier reports indicated that in contrast to all other eukaryotes, N-glycosylation of proteins does not occur (31). Recently N-glycoproteins were found in intraerythrocytic parasites (32), so it was of interest to determine whether the plasmepsins are N-glycosylated and if this modification plays any role in delivery of proteins to the digestive vacuole. Plasmepsin I has 7 consensus N-glycosylation sites, 5 on the putative intralumenal part of the protein and plasmepsin II has 5 consensus sites, 3 on the predicted lumenal side of the protein.
When proplasmepsins I and II, translated in vitro in the presence of canine pancreas microsomes, are examined by SDS-PAGE, higher molecular weight forms (3 for plasmepsin I and 2 for plasmepsin II) are seen for both proteins (Fig. 4B). Treat-FIG. 2. Plasmepsin I and II antibodies are specific for their respective proteins. Proplasmepsins I and II were prepared by in vitro translation and immunoprecipitated with an antibody prepared against mature plasmepsin I (Ab 574) denoted here as PM I or with an antibody generated against the NH 2 -terminal 21 amino acids of native plasmepsin II (Ab 737, denoted here as PM II). Competition assays were performed with soluble expressed plasmepsins I or II, prepared as described (24). ment of these products with Endo H or PNGase F (not shown) results in trimmed products that comigrate with proteins synthesized without microsomes. To examine native proplasmepsins, trophozoites were metabolically labeled for 10 min and harvested. Immunocomplexes containing plasmepsins I and II were isolated, denatured, and treated with either Endo H or PNGase F. Radiolabeled proplasmepsins translated in vitro without microsomes were prepared for size comparison. The native proplasmepsins comigrate with the in vitro translated controls, with or without glycosidase treatment, as is expected if they are not N-glycosylated (Fig. 4C).
Proplasmepsins Are Type II Integral Membrane Proteins-Previous immunoelectron microscopy studies using a plasmep-sin I-specific antibody (Ab 574) suggested that the enzyme is membrane-associated in delivery vesicles, at the parasite cell surface and in the cytostome (15). Trophozoites were metabolically labeled for 3 h to accumulate mature plasmepsins, then BFA was added and labeling continued for 2 h to accumulate proplasmepsins. This treatment produced approximately equal amounts of pro and mature plasmepsins (Fig. 5). High speed centrifugation of trophozoite lysate produced a pellet containing trophozoite membranes from which the proforms of plasmepsins I and II were immunoprecipitated. An additional membrane pellet prepared in the same manner was extracted successively with high salt and high pH solutions. These treatments did not dissociate proplasmepsins from the membrane fraction, suggesting that the 21-amino acid hydrophobic stretch in the proplasmepsins serves as a signal anchor sequence. Immunoprecipitation of the supernatant fraction indicates that mature plasmepsins are soluble proteins produced by cleavage of the membrane-bound proforms (Fig. 5).
To examine the membrane topology of the proplasmepsins, trophozoites were metabolically labeled in the presence of BFA for 3 h and the parasitized erythrocytes were collected. The cells were frozen and thawed in cold. This treatment lyses cells and permits microsomes to reseal with proteins in their original orientation (25). Proteinase K was added with or without Triton X-100 and the digested products were immunoprecipitated and examined by SDS-PAGE (Fig. 6). A clear difference in the size of proplasmepsin I is observed after protease digestion of intact microsomes. This 4-kDa difference is consistent with the predicted cleavage of 35 amino acids NH 2 -terminal to the signal anchor. Disruption of microsome integrity by Triton X-100 permitted complete plasmepsin digestion by proteinase K, indicating that 47 kDa of plasmepsin I is contained within the microsome. Similar results were obtained for plasmepsin II (not shown).
Proplasmepsin Processing Is Inhibited by ALLN and ALLM in Vivo-Comparison of the putative cleavage sites deduced from the NH 2 termini of isolated mature enzymes (12), suggests that there is conservation at the Ϫ2, Ϫ1, and ϩ4 positions (Fig. 1). However, in vitro recombinant proplasmepsin II activates autocatalytically by cleavage at another site, 12 amino FIG. 4. Plasmepsins I and II are secretory proteins. A, trophozoites were radiolabeled with 35 S-labeled amino acids for 3 h in the presence (ϩ) or absence (Ϫ) of BFA (5 g/ml). Trophozoite lysate was immunoprecipitated in series using Ab 737 (PM II) followed by Ab 574 (PM I). B, proplasmepsins I and II were in vitro translated in the presence or absence of canine pancreas microsomes. Digestion with Endo H was performed, demonstrating that the slower mobility of protein translated in the presence of microsomes is due to glycosylation. N-Glycosidase F (PNGase F, not shown) gave similar results. C, proplasmepsins I and II are not glycosylated in vivo. Trophozoites were pulse labeled for 10 min, harvested, and immunoprecipitated as above. For comparison, proplasmepsins I and II were translated in vitro without microsomes to generate the 51-kDa unglycosylated forms (IVT). The immunoprecipitated products were incubated with Endo H buffer alone (Ϫ), Endo H, or PNGase F. No shift in mobility was observed following glycosidase digestion of native proplasmepsins I and II compared with undigested proplasmepsins I and II translated in vitro with microsomes. In other experiments (not shown) in vitro translated proplasmepsins prepared without microsomes were not digested by Endo H or PNGase F. acids upstream from the end identified for the native enzyme (33). This difference might be explained if autocatalytic cleavage occurs in vivo, and is followed by an additional processing event or by exopeptidase-mediated trimming during protein isolation (33). Alternatively, autocatalytic processing of proplasmepsin II might be an in vitro artifact. Cleavage of proplasmepsin I to mature protein has not been demonstrated in vitro (24), suggesting either that the plasmepsins are processed differently in vivo or that expressed proplasmepsin I resists activation because of improper folding. Interestingly, the site of proplasmepsin II autocatalytic cleavage is Phe-Leu; cleavage of hemoglobin at ␣33/34 Phe-Leu by both plasmepsins initiates its breakdown (11,12). Proplasmepsin I has Phe-Phe at the corresponding position in the pro-piece.
Inhibitors were added to trophozoites during metabolic labeling to identify the class of enzyme responsible for plasmepsin processing. Proplasmepsins I and II were affected identically by all reagents and conditions tested. The results shown are for plasmepsin I. Trophozoite cultures were preincubated with inhibitor for 45 min and then cells were pulse-labeled with 35 S-labeled amino acids for 10 min, followed by a 1-2 h chase in complete medium containing inhibitor and cycloheximide. Since autocatalytic cleavage of the pro-piece seemed likely, cultures were tested with the aspartic protease inhibitor pepstatin, as well as the plasmepsin I-specific inhibitor SC-50083 (15). No inhibition of processing was observed even at concentrations greater than the IC 50 for parasite growth for these compounds (12) (Fig. 7A). A variety of other standard protease inhibitors were tested including the cysteine protease inhibitors leupeptin and E-64 (also Z-Phe-Ala-CHN 2 , not shown) and serine protease inhibitor PMSF. No effect on plasmepsin processing was observed. The metalloprotease inhibitor 1,10-phenanthroline (1 mM) killed cells rapidly, so the effect on processing could not be determined (not shown). Additional inhibitor screening revealed that the tripeptide aldehyde inhibitors ALLN and ALLM completely block proenzyme cleavage in a reversible manner (Fig. 7, A and B). These inhibitors target cysteine proteases (including calpains) and the chymotryptic activity of proteasomes (ALLNϾ ϾALLM) (34 -36). Titration of ALLN and ALLM indicated that both compounds inhibit processing equally well with an IC 50 of 5 M for each (not shown). To further define the processing activity, the proteasome inhibitor lactacystin (37) was tested and found to be ineffective. Most proteins degraded by proteasomes require ATP-dependent ubiquitination (38). No evidence for higher molecular weight, ubiquitinated forms of proplasmepsin are observed following ALLN treatment even when cells are labeled for several hours (not shown) in contrast to what is seen when proteasome hydrolysis of ubiquitinated proteins is blocked (39). Incubation of cells with deoxyglucose and dinitrophenol (DEOX, Fig. 7A) uncouple oxidative phosphorylation and deplete cellular ATP.
These agents had no effect on processing, suggesting that ATP is not required.
To test the possibility that the tripeptide aldehyde blockade could be inhibition of autocatalytic plasmepsin processing, recombinant proplasmepsin II was incubated at pH 5 and analyzed by SDS-PAGE (Fig. 8). Pepstatin blocked autocleavage well, while ALLN was ineffective. Up to 100 M ALLN failed to block autoprocessing. Also, in enzyme assays using globin as substrate, mature plasmepsin II was not inhibited by ALLN, while picomolar concentrations of pepstatin blocked catalysis (data not shown). These results suggest that intracellularly, ALLN is not blocking an autocatalytic event but may be inhibiting an unusual processing enzyme.
Proplasmepsin Processing Requires Acid pH in Vivo-Lyso- FIG. 6. Proplasmepsins are type II integral membrane proteins. Trophozoites were labeled with [ 35 S]methionine/cysteine for 3 h. Parasitized erythrocytes were lysed and digested with proteinase K (K) at either 1 or 2 mg/ml. Triton X-100 (TX) was added to some reactions. An aliquot from reactions with and without proteinase K was mixed (MIX) to demonstrate the size difference between digested and undigested proplasmepsin. Proplasmepsins I and II were immunoprecipitated in series. Proplasmepsin I is shown. motropic agents were added to trophozoite cultures during metabolic labeling to determine whether acid conditions are required for proplasmepsin processing. Maturation was inhibited at micromolar concentrations of chloroquine known to raise the digestive vacuole pH (40,41) (Fig. 9). As an alternative method of increasing digestive vacuole pH, the fungal metabolite bafilomycin A 1 was used to specifically inhibit the proton pump (42,43). When proplasmepsin processing was assayed in the presence of bafilomycin A 1 , it was completely inhibited (Fig. 9). These results suggest that proplasmepsin processing is a pH-dependent event, although the effect could also be explained by a blockade of proplasmepsin transport to the compartment where processing occurs.
Proplasmepsin I Is Synthesized and Processed Very Early in the Parasite Life Cycle-Northern blot analysis of total RNA shows that the plasmepsin I message is most abundant at the early (ring) stage, while plasmepsin II message peaks at the later (trophozoite) stage (Fig. 10A). When different stage parasites were 35 S-pulse-labeled, plasmepsin I synthesis and processing could be detected as early as 3 h after erythrocyte invasion (Fig. 10B). Plasmepsin II was first detected in late rings (12 h). Synthesis of both proteins peaked in the trophozoite stage (24 -32 h). By the schizont stage (36 -42 h), synthesis dropped off markedly (not shown). Early stage parasites were fully capable of processing proplasmepsin I and the processing was inhibited by BFA as well as ALLN (Fig. 10C), similar to the results observed when both plasmepsins I and II are synthesized and processed later in development (Figs. 4 and 7).

Biosynthesis and Delivery of Plasmepsins I and II to the
Digestive Vacuole-Like other aspartic proteases, the plasmepsins are secretory proteins. Remarkably little is known about the transport of secretory proteins in Plasmodium (26,28). When brefeldin A, a fungal metabolite known to block secretion in eukaryotes, is added to Plasmodium cultures, processing of proplasmepsins to mature forms is inhibited, presumably because delivery to the processing compartment is impaired. Whether this results because BFA disrupts and redistributes the Golgi or because it blocks secretion from the endoplasmic reticulum is unclear. Both models have been proposed to explain the effect of BFA on Plasmodium (26,28). In fact, it has been suggested that early stage parasites may lack a Golgi altogether (28) as is the case for certain stages of the protozoan parasite Giardia lamblia (44). Stacked Golgi cisternae have not been seen in Plasmodium although some Golgi markers have been identified (26,45). Proplasmepsins have N-glycosylation sites, some of which were recognized during our in vitro translation with canine pancreas microsomes. However, despite reports of N-glycosylation of some malarial proteins (32), native proplasmepsins were not found to be Nglycosylated, so whether they transit the Golgi could not be ascertained. How the plasmepsins escape this post-translational modification is not clear.
Typically proforms of aspartic proteases are soluble, although peripheral membrane association is sometimes observed (46 -49). In contrast, proplasmepsins I and II are integral membrane proteins that are processed to form soluble mature enzymes. Cleavage of a membrane-spanning precursor In other experiments, up to 100 M ALLN had no effect.
FIG. 9. Proplasmepsin processing requires acid conditions. Trophozoites were metabolically labeled as in Fig. 7A and chased for 2 h in the presence of agents that raise pH. Chloroquine was tested at 1, 10, and 50 M and bafilomycin A 1 at 1.3 M (BAF). Unprocessed (P) and processed (C) proplasmepsin controls are shown. Immunoprecipitations done in series for plasmepsins I and II indicated that processing of both was identical. The results shown are for plasmepsin I. to generate soluble enzyme also occurs for mammalian lysosomal acid phosphatase (50). Proenzyme processing is different from the proplasmepsins in this case, since acid phosphatase is transported to the lysosome several hours after synthesis and is released from the membrane by sequential proteolytic steps (51).
Proplasmepsins have type II membrane topology. This is consistent with the positive inside rule (52). Positively charged NH 2 -terminal residues have been shown to predict type II membrane topology (53,54). Proplasmepsin I has 3 lysines and 1 arginine within 8 amino acids NH 2 -terminal to the signal anchor (15) and proplasmepsin II has 4 lysines (18). Plasmepsin I was previously hypothesized to traverse the parasite plasma membrane to the parasitophorous vacuolar membrane and intersect with the hemoglobin ingestion pathway (15). This delivery route places plasmepsin I near the parasitophorous vacuole membrane with the carboxyl terminus of the protein (from which the active enzyme is derived) in the erythrocyte cytoplasm. The orientation is reversed from that predicted for type II proteins that traffic through the secretory pathway by successive membrane fusion and vesiculation events (55). A simplified secretory model has been proposed for Plasmodium in which local parasite plasma membrane and parasitophorous vacuole membrane fusion occurs forming joint domains with a single membrane bilayer (26,56). Contact sites between parasite plasma membrane and parasitophorous vacuole membrane have been observed microscopically (55,57). Fusion of delivery vesicles containing proplasmepsins with these joint domains could yield the orientation suggested by immunoelectron microscopy for plasmepsin I. However, additional targeting studies will be required to delineate the route taken by plasmepsins to the digestive vacuole.
Proplasmepsin processing appears to require acidic conditions. Two different types of lysosomotropic agents, a proton pump inhibitor and a weak base alkalinization compound were able to block proplasmepsin maturation. It is possible that this is an indirect effect, whereby these agents prevent the proenzymes from getting to their site of processing. We have recently observed proplasmepsin processing in a cell-free system and find that cleavage also requires acid conditions in vitro. 2 Digestive vacuole proteins have similar pH requirements (11,12). These data are consistent with a model for activation in which the membrane-bound proplasmepsins are trafficked to the acidic digestive vacuole where they are cleaved by a processing enzyme. This model predicts that transport through the secretory pathway would have a t1 ⁄2 for delivery less than or equal to the t1 ⁄2 for processing (20 min). Extracellular secretion of newly synthesized proteins by Plasmodium has been observed within 5 min, suggesting that transit through the secretory apparatus is fast (58). In yeast, transport of soluble proteins proteinase A (59) and carboxypeptidase Y (60), as well as the membrane protein alkaline phosphatase (61), to the acidic food vacuole, has been observed with a t1 ⁄2 of 6 min. Thus, rapid delivery of the plasmepsins to the digestive vacuole seems entirely reasonable, but this model will require further testing.
Proplasmepsin Processing Has an Unusual Inhibitor Profile and May Be Mediated by an Unusual Enzyme-The aspartic protease inhibitors pepstatin and SC-50083 did not block proplasmepsin processing. This could result from poor penetration of the inhibitors. However, SC-50083 appears to be readily taken up by cells (15) and while pepstatin membrane permeability is poor, concentrations well above the IC 50 were found to be ineffective at blocking processing. These data, combined with the effectiveness of the tripeptide aldehydes ALLN and ALLM (which do not block autoprocessing of proplasmepsin II in vitro) suggest that the activating cleavage of the proplasmepsins is not autocatalytic. This is atypical for acidic aspartic proteases (20). Instead, cleavage of both proenzymes appears to be mediated by a tripeptide aldehyde-inhibitable activity. The tripeptide aldehydes ALLN and ALLM inhibit cysteine proteases including calcium-activated cysteine proteases (calpains) (34). Such enzymes may be envisioned to encounter unprocessed plasmepsins in the parasite. Falcipain has been isolated from the digestive vacuole and was shown to have substrate specificity that matches the proplasmepsins cleavage site (12). Also, calpains I and II have been purified from erythrocyte cytoplasm (62,63) and might interact with plasmepsins in the hemoglobin ingestion pathway. However, the processing enzyme does not appear to be any known cysteine protease since the inhibitors E-64, leupeptin, and Z-Phe-Ala-CHN 2 were ineffective at inhibiting processing at high concentrations.
ALLN inhibits the chymotryptic activity and to a lesser extent the peptidylglutamic peptide hydrolase and tryptic activities of proteasomes (35,64). Most of our experimental data suggest that proteasome involvement in proplasmepsin cleavage is unlikely. We find that ALLN and ALLM inhibit processing equally well. This is not the case for the chymotryptic activity of proteasomes against which ALLN is 5-42 times more potent than ALLM (35,65). In addition, proteasomes generally hydrolyze proteins into peptides. Although proteasomes have infrequently been implicated in protein processing (66,67) this appears to result when the precursor is ubiquitinated in an ATP-dependent manner and then partially degraded (66). We find no evidence of proplasmepsin ubiquitination and ATP is not required for proplasmepsin cleavage. Additionally, proteasomes work optimally at neutral pH, while processing appears to involve an acid hydrolase. Finally, the Streptomyces antibiotic lactacystin inhibits the chymotryptic activity of proteasomes (37) but does not inhibit proplasmepsin processing.
Proplasmepsin I Synthesis and Processing Commences Early in Development-At the ring stage of parasite development, plasmepsin I mRNA is abundant and protein is synthesized. At the trophozoite stage even more plasmepsin I protein is made, although the quantity of message has declined substantially. This suggests that there is better translation efficiency later in development, but what accounts for this is not clear. Also unexpected is the observation that plasmepsin I is processed in very early parasites and appears to require acidic pH for this to occur. In early ring stage parasites there is no identifiable digestive vacuole (3) and little hemoglobin degradation is thought to occur. It seems possible that plasmepsin I has some function in addition to hemoglobin metabolism. Alternatively, a low level of hemoglobin digestion may be necessary for parasite survival even at the earliest stages of development. At this time, plasmepsin I alone may suffice. Digestion of a greater mass of hemoglobin required for parasite maturation could be augmented by plasmepsin II later in development.
Proplasmepsin Activation as a Potential Target for Antimalarial Chemotherapy-Hemoglobin breakdown by the malaria parasite P. falciparum is a massive catabolic process that appears to be essential for the survival of the parasite. Plasmepsins I and II have been shown by in vitro assay to comprise 80% of the total globin degrading activity in the digestive vacuole (12). They have different substrate specificities except for cleavage at ␣Phe 33 -Leu 34 which is thought to initiate hemoglobin proteolysis (12). Previously, we have shown that malaria parasites in culture are killed by a peptidomimetic inhibitor of plasmepsin I (SC-50083) and proposed that it could serve as a good lead compound for developing more potent anti-plasmepsin agents (15). This goal is complicated by the degree of difficulty involved in synthesis of SC-50083 and its derivatives. In this study, we have determined that processing of both plasmepsins I and II is inhibited by ALLN and ALLM. These inhibitors are nonspecific and obviously not suitable for use as antimalarials, however, the notion that both plasmepsins could be completely inactivated by a single compound is appealing. ALLN and ALLM are permeable tripeptide aldehydes that are readily taken up by cells. They are relatively easy to synthesize which should facilitate identification of related compounds specifically tailored to inhibit processing. The cleavage sites of proplasmepsins I and II have been determined and shown to have conserved sequence (12). Both have glycine at P2, leucine at P1, and aspartic acid at P4Ј (Fig. 1). Additional insight into the processing event may be derived from the observations that ALLN and ALLM inhibit equally well, but replacing norleucinal or methional with argininal (leupeptin) results in the loss of processing blockade. This hint of selectivity shows promise for the development of easily synthesized plasmepsin inhibitors that efficiently prevent hemoglobin degradation and are lethal to the parasite.