Comparative Proteomics and Functional Analysis Reveal a Role of Plasmodium falciparum Osmiophilic Bodies in Malaria Parasite Transmission*

An essential step in the transmission of the malaria parasite to the Anopheles vector is the transformation of the mature gametocytes into gametes in the mosquito gut, where they egress from the erythrocytes and mate to produce a zygote, which matures into a motile ookinete. Osmiophilic bodies are electron dense secretory organelles of the female gametocytes which discharge their contents during gamete formation, suggestive of a role in gamete egress. Only one protein with no functional annotation, Pfg377, is described to specifically reside in osmiophilic bodies in Plasmodium falciparum. Importantly, Pfg377 defective gametocytes lack osmiophilic bodies and fail to infect mosquitoes, as confirmed here with newly produced pfg377 disrupted parasites. The unique feature of Pfg377 defective gametocytes of lacking osmiophilic bodies was here exploited to perform comparative, label free, global and affinity proteomics analyses of mutant and wild type gametocytes to identify components of these organelles. Subcellular localization studies with fluorescent reporter gene fusions and specific antibodies revealed an osmiophilic body localization for four out of five candidate gene products analyzed: the proteases PfSUB2 (subtilisin 2) and PfDPAP2 (Dipeptidyl aminopeptidase 2), the ortholog of the osmiophilic body component of the rodent malaria gametocytes PbGEST and a previously nonannotated 13 kDa protein. These results establish that osmiophilic bodies and their components are dispensable or marginally contribute (PfDPAP2) to gamete egress. Instead, this work reveals a previously unsuspected role of these organelles in P. falciparum development in the mosquito vector.

Each year more than 200 million people suffer from malaria, with an estimated death toll of 438,000 in 2015 (1). Malaria is caused by apicomplexan parasites of the Plasmodium genus, with Plasmodium falciparum being the most deadly of the five species capable of infecting humans.
Malaria infection in humans is initiated by the bite of an infected Anopheles mosquito, which inoculates hundreds of sporozoites that migrate into the blood stream to reach the liver hepatocytes. Inside the hepatocytes, the parasites multiply asexually yielding thousands of hepatic merozoites which invade red blood cells (RBCs) 1 . Blood stage asexual parasites undergo indefinite cycles of merozoite proliferation, egress from the host cell and reinvasion of new RBCs, causing malaria symptoms. During asexual growth, a fraction of parasites enter sexual differentiation (gametocytogenesis). This process takes place in the bone marrow where the parasites go through five morphologically distinct stages, culminating after 10 days in the formation of the falciform mature gametocytes. Mature sexual stage parasites are then released into the peripheral blood stream where they are ready to be taken up by mosquitoes during the next insect blood meal.
In the mosquito midgut mature female and male gametocytes react to several physical and chemical factors (2), transforming into male and female gametes. Gametogenesis begins with the rapid rounding-up of the elongated parasites and ends in 15-20 min with the egress of gametes from erythrocytes, each male gametocyte producing eight flagellated microgametes and each female forming a single spherical macrogamete. After fertilization, the zygote transforms into an ookinete, which traverses the mosquito midgut epithelium to develop into an oocyst, producing several thousand sporozoites able to migrate to the mosquito salivary glands, ready to infect a new human host (3).
Apicomplexan parasites invade and egress from different cellular types. These complex processes are orchestrated by diverse factors produced by the sets of membranous secretory organelles found in all parasite developmental stages. Conservation of the secretory organelles among apicomplexa highlights the importance of these structures in the life cycle of these diverse parasites.
In Plasmodium, the family of apical secretory organelles include rhoptries, micronemes and dense granules. Rhoptries are elongated paired organelles found in merozoites, whereas micronemes, smaller in size, are present in merozoites, sporozoites and ookinetes. These organelles secrete factors that mediate host cell invasion and egress. Finally, spherical dense granules are found in the merozoites, where they contribute to the formation of the parasitophorous vacuole after RBC invasion (4 -6). Identification of specific organellar components recently enabled the definition of other secretory organelles such as mononemes and exonemes. The mononeme is a thread-like compartment hosting the PfROM1 protein, important in invasion (7), whereas exonemes, similar in shape and distribution to micronemes, release their contents that are indispensable to merozoite egress, e.g. the subtilisin like protease PfSUB-1 (8).
Protein content of these organelles has been identified mainly by cell fractionation experiments where differential and/or gradient centrifugation yielded protein samples enriched in the target organelle. The difficulty in isolating purified individual organelles however made it essential to confirm the localization of the identified candidate proteins, usually by subcellular localization analyses with specific antibodies or transgenic reporter gene fusions.
Sexual blood stages of Plasmodium retain another type of secretory organelles, the osmiophilic bodies (OBs). In P. falciparum, elongated OBs (about 200 nm in diameter) are detected in the periphery of stage III gametocytes (predominantly female) from day four of gametocyte maturation. In the murine malaria model Plasmodium berghei OBs of smaller size are described in female gametocytes and a class of morphologically distinct OBs has been recently described in male gametocytes (9). Very little information is currently available on the protein content of these organelles. The first to be identified and the only specific protein of the P. falciparum OBs to date is Pfg377 (10). Three OB specific proteins have been identified in P. berghei, Pbg377; Male Development-1 (PbMdv-1) and Gametocyte Egress and Sporozoite Traversal (PbGEST) (11,12). Although Pbg377 and Pfg377 localize to OBs in P. berghei and P. falciparum, respectively, PbMdv-1 is specific to OBs only in P. berghei, whereas its P. falciparum homolog has a broader distribution in several gametocytederived membranes (13,14). Finally, the localization of the PbGEST ortholog has not been characterized yet in the human malaria parasite.
Previous observations that OBs are discharged, probably into the PVM, at the start of gametogenesis and that are no longer present in extracellular gametes suggested a function of these organelles in gamete egress from the RBCs. Consistently, disruption of either Pfg377, PbMdv1 and PbGEST in transgenic parasites resulted in a mild defect in efficiency and/or a delay in gamete egress (9 -12). In P. falciparum, recent studies however questioned the involvement of OBs and Pfg377 in gamete egress (15), attracting attention to the additional phenotypes described in the pfg377KO sexual stages, namely that the Pfg377-defective gametocytes are almost completely depleted of OBs, and that their capacity to infect mosquitoes is virtually abolished (16). Interestingly, in P. berghei, the pbg377 gene disruption resulted in a reduction in size, but not in number, of OBs in gametocytes, and no significant reduction in mosquito infectivity (9), suggesting existence of species specific difference.
The lack of OBs in the Pfg377-deficient P. falciparum gametocytes constitutes to our knowledge an almost unique case of a parasite developmental stage depleted of a specific organelle. An analogous situation was observed only with the loss of the apicoplast in the P. falciparum asexual stages after a treatment with fosmidomycin, an isoprenoid precursor biosynthesis inhibitor (17).
The work presented here exploited the lack of OBs in Pfg377-deficient gametocytes to identify OB protein content by two label free quantitative proteomic approaches, a global comparative proteomics analysis and an affinity proteomics approach identifying Pfg377 interacting proteins. Functional studies on selected candidate gene products revealed an OB localization for four out of five identified proteins, including two proteases, and the persistence of OB components in gametes after egress from the RBC. These results identify a functional role of OBs in parasite development and transmission in the Anopheles vector.

EXPERIMENTAL PROCEDURES
Experimental Design and Statistical Rationale-The objective of this experiment is to identify osmiophilic body proteins expressed in gametocytes by a cellular comparative proteomics approach between wild type (wt) gametocytes and gametocytes depleted of osmiophilic bodies, and by an affinity proteomics approach identifying Pfg377 interaction partners in wt gametocytes and gametocytes depleted of osmiophilic bodies. Control, wt gametocyte and Pfg377depleted gametocyte samples were produced by culturing P. falcip-arum parasites in human red blood cells. For each sample, three biological repeats were cultured independently resulting in three samples for wt gametocytes, and in six samples for Pfg377-depleted gametocytes. These samples were studied by an gel-LC MS/MS approach (supplemental Fig. S2), and proteins were identified by searching tandem mass spectrometry data against a protein database composed of P. falciparum and human proteins. Protein abundances were estimated by label free quantification (LFQ) based on peptide intensities in MaxQuant. Gametocyte protein abundance data was then filtered to include proteins that are quantified across all samples, that are up-regulated in mature gametocytes (30) and that are up-regulated in female gametocytes with female/male protein ratio Ͼ 1.2 (18). Statistical significant differences in protein levels between wt gametocyte and Pfg377 depleted gametocytes was determined by Student t tests comparing wt group (three expression data sets) versus sample groups (six expression data sets).
For the affinity proteomics experiment single biological repeats were used for measuring protein interaction partners of Pfg377 in wt gametocytes and Pfg377-depleted gametocytes by immunopurification and LC-MS/MS.
The results of the cellular comparative proteomics approach and the affinity proteomics approach were combined to generate a list of putative osmiophilic body proteins.
Plasmodium falciparum Culture-P. falciparum clone 3D7 was used throughout (19). Wild type and transgenic lines were cultured in Oϩ human red blood cells at 5% hematocrit in RPMI 1640 plus hypoxanthine 50 g/ml, HEPES 25 mM, 0.225% sodium bicarbonate and 10 mg/ml gentamicin, supplemented with 10% heat inactivated human serum. Parasites were kept at 37°C, in a 2% O, 5% CO 2 and 93% N 2 atmosphere. For gametocyte production, asynchronous parasites were grown to high parasitaemia (Ͼ8%) and culture medium was doubled at this point. The day after, medium was changed and N-acetylglucosamine 50 mM added. N-acetylglucosamine was maintained for 5 days, until no asexual parasites were detected in the culture. Stage II gametocytes were detected 48 h after the addition of N-acetylglucosamine whereas mature stage V gametocytes appeared 9 days after the start of the treatment. Percoll and MACS purification of gametocytes was performed as described (20). Protocols used to thaw and freeze parasites are described elsewhere (21). In the production of gametocytes for mosquito experimental infections, Nacetylglucosamine and gentamicin were omitted from culture medium not to interfere with the mosquito microbiome.
Production of Transgenic Parasites-Ring stage parasites at 2-3% parasitaemia were transfected by electroporation with 80 -100 g of transfection vectors using the following conditions: voltage: 0.31 kV, capacitance: 960 mF, resistance to infinity (22). Transfected parasites were then added to 5 ml of fresh media, 5% hematocrit. Parasites were cultured without drug selection for 24 h and WR99210 5 nM was then added. Medium was renewed every day during the first 4 days, and every 2 days after that. Fresh 1% hematocrit RBCs were added every 7 days until resistant parasites appeared, which happened from 14 days after transfection. Rounds of drug cycling consisting of 21 days of growth in drug absence before reintroduction of drug pressure were carried out to enrich the population of parasites with integrated plasmid (22).
To produce pfg377KO parasite lines, plasmid pHH-TK-pfg377 (supplemental Fig. S1) was transfected as described above and WR99210 (positive) and ganciclovir (negative) selections were applied to obtain lines GT5 and GT8 from independent transfections. After Southern blot analysis confirmed disruption of the pfg377 locus (supplemental Fig. S1), clones were obtained by limiting dilution. Parasite clones 5-2 and 8 -20, from the GT5 and GT8 lines respectively, were used in this work.
In other experiments, selection for plasmid chromosomal integration was achieved through cycles of off-drug (1 week)/on-drug. In the case of KO plasmids, negative selection was then performed adding 40 M 5-fluorocytosine (5-FCT) to the medium (23). Parasites resistant to both WR99210 and 5-FCT appeared in Giemsa smears after a week. GFP tagged and KO lines were then cloned by limiting dilution (19).
Morphometric Analysis of OB Size in wt and pfg377KO Gametocytes-Micrographs of ultrastructure sections of wt and pfg377KO mature gametocytes were used to count and measure the size of OBs. The eccentricity (e) of OBs in the wt and in the pfg377KO gametocytes was calculated according to the formula e ϭ ͱ 1 Ϫ b 2 a 2 . Student's t test was performed to determine if (e) values were significantly different in wt and pfg377KO OBs.
Measurement of Gametocyte Egress Efficiency-Egress efficiency of gametes was measured as described previously (15). Briefly, mature gametocyte cultures were stained with fluorophore-bound WGA (5 g/ml), which binds to RBCs surface, and then gametogenesis was triggered by adding 20 M xanthurenic acid (XA). After 15 min at room temperature parasites were fixed in 1% paraformaldheide (PFA) and mounted in microscope slides. Gametes were identified and the presence or absence of RBC was assessed examining presence or absence of the fluorescence surrounding the parasites. For each parasite line/clone, at least four independent experiments were done, counting 200 gametes in each case. t-tests were performed on individual pairs of wt-KO parasites.
Mosquito Rearing and Parasite Infections-Anopheles gambiae Ngousso mosquitoes used throughout the study were maintained at 29°C 70 -80% humidity 12/12 h day/night cycle. Mosquitoes were fed on gametocytes cultured in Oϩ human red blood cells (Haema, Berlin,Germany) through a membrane feeder at 37°C for 15 min. 3D7 wild type and pfg377KO clone 5-2 gametocyte cultures were adjusted for approximately the same final stage V gametocytaemia (0.03-0.1%) in each experiment repeat. Unfed mosquitoes were removed shortly after infections, whereas fed mosquitoes were maintained at 26°C in a secured S3 laboratory according to the national regulations (Landesamt fü r Gesundheit und Soziales, project number 411/08). Oocysts were counted in the dissected midguts11 days postinfection.
Gametocyte Purification for Proteomic Analysis-Three biological replicates of this experiment were performed. Magnetic purification with CS columns in a MaxiMACS (Miltenyi Biotec, Teterow, Germany) was used to purify the gametocytes from noninfected RBCs. Prior to purification, Giemsa smears were analyzed to measure gametocyte stage synchronicity and sex ratios and control for the absence of asexual parasites or immature gametocytes. Purified parasites were counted using a cell counter chamber, and 2 ϫ 10 7 parasites were pelleted and frozen for each sample. Parasites were lysed in 4% SDS in PBS, 4ϫ Laemmli Sample Buffer, and reduced with 50 mM DTT at 95°C for 5 min prior to protein gel electrophoresis.
Gametocyte Sample Preparation for LC-MS/MS Experiments-Gel based fractionation of the proteins prior to tryptic digestion was conducted by gel electrophoresis using 10% Criterion XT Bis-Tris precast gels (Bio-Rad inc, Hemel Hempstead, UK) and XT-MES running buffer pH 6.4. Gels were stained with Colloidal Coomassie Blue (Severn Biotech, Kidderminster, UK), and divided into six slices per lane. Gel slices were treated with 10 mM DTT and 55 mM iodoacetamide and digested by trypsin overnight at 37°C as described previously (24). Tryptic digests were acidified to a final concentration of 0.1% TFA and purified by STAGE tips (25).
Liquid chromatography tandem mass spectrometry-Ultimate 3000 UPLC (Thermo Fisher) connected to the Orbitrap Velos Pro mass spectrometer (Thermo Fisher) was used to perform tandem mass spectrometry experiments. Peptides were loaded on a 2 cm Acclaim™ PepMap™100 Nano-Trap Column (Thermo Fisher) and were separated by a 25 cm Acclaim™ PepMap™100 Nano LC column (Thermo Fisher) packed with 3 m C18 beads. Sample run with a flow-rate of 300 nl/min in a 120 min gradient of 95% buffer A/5% buffer B to 65% buffer A/35% buffer B (buffer A: 0.5% acetic acid. Buffer B: 0.5% acetic acid in 100% acetonitrile). Peptides eluting from the column were electrosprayed into the mass spectrometer at 2.3 kV pray voltage. Orbitrap was used for data acquisition in a data-dependent mode automatically switching between MS and MS2. Orbitrap cell was used to acquire full-scan spectra of intact peptides (m/z 35-0 -1500) with automated gain control accumulation value of 1,000,000 ion, with a resolution of 60,000. The ten most abundant ions were then sequentially isolated and fragmented in the linear ion trap, where dissociation was induced through collision, using an accumulation target value of 10,000, a normalized collision energy of 32% and a capillary temperature of 275°C. Dynamic exclusion of ions sequenced within the 45 previous seconds was applied. Unassigned charge states and singly charged ions were excluded from sequencing. For MS2 selection, a minimum of 10,000 counts was required.
Peptide Identification and Quantification by MaxQuant-Andromeda (26) search engine integrated in the MaxQuant (Version 1.3.0.5) (27) was used to identify proteins. Peak lists were generated for the top 6 most intense MS peaks in 100 Da windows by MaxQuant prior to the database search. The protein database contained 91,901 protein sequences from P. falciparum (http://plasmodb.org/common/ downloads/release-27/Pfalciparum3D7/) and human IPI version 3.68 (ftp://ftp.ebi.ac.uk/pub/databases/IPI/) supplemented with frequently observed contaminants. Andromeda search parameters for protein identification were set to tolerance of 6 ppm for the parental peptide and 0.5 Da for fragmentation spectra and trypsin specificity allowing up to 2 miscleaved sites. Deamination of glutamine, oxidation of methionine, and asparagine and protein N-terminal acetylation were set as variable modifications, carboxyamidomethylation of cysteines was specified as a fixed modification. Minimal required peptide length was specified at 6 amino acids. MaxQuant was used as described (27) to perform internal mass calibration of measured ions and peptide validation by the target decoy approach and to enable the 'match between run option' for an elution time window of 2 min. Peptides and proteins detected by at least two peptides in one of the samples with a false discovery rate (FDR) of 1% were accepted. Excluded from validation were proteins identified by site only, external contaminants and reversed proteins. Proteins were quantified by normalized summed peptide intensities (28,29) computed in MaxQuant with the label free quantification (LFQ) option switched on.
Identification of Osmiophilic Body Proteins by Global Proteomics-Osmiophilic body proteins were identified by a comparison in protein abundances between wt gametocytes and pfg377KO gametocytes. To minimize variability from differences between parasites strains, gametocyte samples from two independent pfg377KO parasite strains and from wt gametocytes were cultured in parallel. A total of three biological replicate samples were generated and analyzed. Morphological analysis of the samples ensured similarity across them through evaluation of absence of asexual parasites, sex ratio, maturity, percentage of infected RBCs, and percentage of gametes. Mass spectrometry data were acquired sequentially, completing the acquisition of data from each batch individually before the analysis of subsequent samples. Data from each sample was acquired in triplicate and processed by MaxQuant.
Perseus (Version 1.3.7.1) (30) was used to perform statistical analysis for the identification of a list of putative osmiophilic body proteins. Prior to analysis in Perseus the gametocyte protein abundance data was filtered to include proteins that are quantified across all samples, that are up-regulated in mature gametocytes (31) and pro-teins that are up-regulated in female gametocytes with female/male protein ratio Ͼ 1.2 (18). Protein abundance data was normalized by median expression values and transformed to relative expression data for each biological repeat and subsequently imported into Perseus and classified into a group of pfg377KO gametocyte samples and a group of wild type gametocytes. Statistical significant differences between both groups were calculated by two-tailed t-tests with results presented in a Volcano plot depicting protein ratios versus statistical significance.
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (32) partner repository with the dataset identifier PXD004124.
Identification of Osmiophilic Body Proteins by Affinity Proteomics-Prior to the immunopurification experiment, antibodies against Pfg377 B fragment were crosslinked to agarose beads: Agarose beads (50 l) were washed three times with PBS 0.1% Tween-20 (Sigma-Aldrich) and resuspended in 100 mM HCl, 0.5 M NaCl (pH 8.0). Antibodies (20 l) were added and the mix incubated for 3 h at room temperature. Beads were then washed (500 g centrifugations for 5 min was performed to pellet beads throughout the experiment) three times with 200 mM sodium borate pH9.0 to eliminate unbound antibody. Beads were then resuspended in 3 volumes of methylpimelimidate (DMP, the crosslinking agent) 20 mM in 200 mM borate pH9.0 and incubated for 90 min at room temperature in a shaker. Crosslinking solution was washed away and beads resuspended washed three times in TrisHCl 50 mM pH7.5, and then incubated at room temperature for 2 h, to terminate the crosslinking reaction. Noncrosslinked antibodies were removed by washing three times with 1 ml of 0.58% v/v acetic acid 150 mM NaCl. Beads were then washed three times in PBS and stored at 4°until experiment began. Correct crosslinking was evaluated through an acrylamide protein gel in which crosslinked beads showed to contain a protein of the expected length of the ␣-Pfg377B antibodies.
Samples containing 3 ϫ 10 7 mature synchronous gametocytes from wt and the pfg377KO clone 8 -20 were isolated via magnetic columns, pelleted and frozen. Lysis occurred for 30 min at 4°C in lysis buffer (50 mM Tris-HCl pH7.5, 150 mM sodium chloride, 1% Nonidet P40 (Sigma-Aldrich, Milan, Italy), 0.5% sodium deoxycholate, Complete protease inhibitor (Roche, Monza, Italy)). Debris was pelleted and supernatants were incubated for 3 h at 4°C with noncrosslinked agarose beads to preclear the lysate of proteins binding to immunobeads in a nonspecific manner. Noncrosslinked beads were pelleted and discarded, and precleared supernatant was then incubated overnight at 4°C with ␣-Pfg377B-crosslinked beads, to allow the binding of Pfg377 and any associated proteins to the specific antibodies. Beads were then pelleted and washed three times in lysis buffer. Elution of the bound proteins was achieved through addition of 2ϫ Laemmli sample buffer (Bio-rad, Hemel Hempstead, UK) to the beads and incubating the sample at 95°C for 5 min. Beds were centrifuged and the supernatant containing the purified proteins used in subsequent steps. Gel based fractionation of the proteins in the supernatants was conducted by gel electrophoresis in a 4 -12% Bis-Tris precast gel (Invitrogen, Waltham, MA) and MES-SDS buffer. The gel was stained with Colloidal Coomassie Blue and a single slice containing each sample was excised. Gel slices were treated with 10 mM DTT and 55 mM iodoacetamide and digested by trypsin overnight at 37°C as described previously (24). Tryptic digests were acidified to a final concentration of 0.1% TFA and purified by STAGE tips (25).
Immuno-purified complexes of Pfg377 binding proteins in P. falciparum wt gametocytes were measured with liquid chromatography tandem mass spectrometry and analyzed by MaxQuant as described above. Immuno-purified complexes of Pfg377 binding proteins in P. falciparum pfg377KO gametocytes functioned as negative controls. Protein abundance data (LFQ) was loaded into Perseus for further analysis, where only proteins detected with at least 2 razor peptides in one of the samples and detected in more than 50% of the samples were considered. Missing LFQ data were imputed with intensities from the lower part of normal data distribution, with an imputation with of 0.3 and a shift of 0.8. Proteins showing LFQ wt/LFQ KO ratio Ͼ 30 were considered as candidates. Ribosomal proteins, a common nonspecific contamination in affinity purification, were ignored as candidates.
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (32) partner repository with the dataset identifier PXD004088.

Depletion of Osmiophilic Bodies in Pfg377-defective Gametocytes Does Not Affect Their Egress from Erythrocytes but
Compromises Their Infectivity to Mosquitoes-As some reversion to wt was reported in the previously described pfg377 knock out (pfg377KO) parasites (16), new independent pfg377 defective parasites were obtained through gene disruption by double crossover recombination. Two parasite clones, pfg377KO 5-2 and 8 -20, were derived from independent transfections and the structure of the pfg377 disrupted locus was confirmed by molecular analysis (supplemental Fig. S1). Inspection of independent ultrastructure sections from stage IV-V gametocytes from clones 5-2 and 8 -20 confirmed that gametocytes defective in the production of Pfg377 are depleted of OBs (Fig. 1A) (16). In addition, a morphometric analysis on the residual electron dense granules in the Pfg377-defective gametocytes revealed differences in both the size and the shape of these organelles from those observed in wt gametocytes, being smaller and lacking the typically elongated shape (supplemental Table S1). Immuno-  Table S2. fluorescence (IFA) and Western blot (WB) analyses with ␣Pfg377 antibodies confirmed disruption of Pfg377 expression in the late gametocyte stages of the pfg377KO clones (Fig. 1B, 1C).
As the observed reduction in efficiency to egress erythrocytes of the Pfg377-defective gametes described in (16) was not reproduced in subsequent experiments (15), this phenotype was analyzed also in the newly obtained pfg377KO gametocytes. Again, no difference in gametocyte egress efficiency was detected between wt and pfg377KO (Fig. 1D), thereby confirming that Pfg377 is dispensable for gametocyte egress.
To investigate whether pfg377 disruption affected developmental steps after gamete egress, transmission of the pfg377KO clone 5-2 to mosquitoes was analyzed in experimental infections of Anopheles gambiae. Striking decrease in the prevalence of infection and in the number of developing oocysts was observed for pfg377KO parasites, indicating that OBs are required for Plasmodium transmission ( Fig. 1E; supplemental Table S2).
In summary, these experiments demonstrated that Pfg377 defective sexual stages egress normally from erythrocytes at gametogenesis, despite lacking their complement of OBs, and confirmed a severe defect in mosquito infectivity, thus revealing an unsuspected role of OBs and/or the OB specific Pfg377 protein in mosquito stages after gamete egress.
Comparative Proteomics of wt and pfg377KO Gametocytes Identifies Candidates for Organelle Resident Proteins-Attempts to isolate or to enrich for OBs in gametocyte cell fractionation experiments from wt parasites in order to identify resident proteins by mass spectrometry were unsuccessful (data not shown). As pfg377KO gametocytes are virtually depleted in OBs, this opportunity was exploited to isolate OB proteins by a comparative mass spectrometry analysis between wt and OB defective gametocytes. The underlying assumption of this experiment was that OB proteins failing to be trafficked to these organelles may be degraded and consequently underrepresented in the proteome of the defective gametocytes.
Three biological replicates of this experiment were performed, using, in each case, parallel synchronous gametocyte samples from wt and both independent pfg377KO clones. Prior to gametocyte purification for mass spectrometry analysis, morphological examination confirmed that stage of maturation, sex ratio and fraction of spontaneously activated "rounded-up" parasites were similar in the wt and the pfg377KO gametocyte cultures (data not shown).
Gametocyte extracts from the nine samples were analyzed by label free quantitative proteomics (29). Briefly, parasite protein extracts were fractionated through gel electrophoresis, trypsin-digested and purified with Stage tips. Samples were then analyzed by LC MS/MS and mass spectrometry data was processed with MaxQuant by the procedure depicted in supplemental Fig. S2. This resulted in the identifica-tion of 29,904 unique P. falciparum peptide sequences (supplemental Table S3) mapping to 1942 protein groups in all gametocyte samples (supplemental Table S4). This protein set was filtered for proteins described to be up-regulated in stage V gametocytes versus stage I (31) and up-regulated in female versus male gametocytes (18) because OBs appear on day four of sexual differentiation and accumulate in female gametocytes (33). Differential protein expression in the resulting list of 386 proteins (supplemental Table S5) between wt gametocytes and OB depleted gametocytes was determined by Student's t tests and visualized by a Volcano plot, where statistical significance is plotted versus protein ratios (Fig. 2). This analysis identified 16 potential OB protein candidates (OBCs) significantly enriched in wt gametocytes compared with the pfg377KO parasites (Table I). This list includes as expected Pfg377, reduced more than 1000-fold in the pfg377KO gametocytes compared with wt.
In a second comparative proteomics approach, immunopurified Pfg377 complexes were obtained in wt and OBdepleted gametocytes to identify Pfg377 interaction partners in wt gametocytes as candidates for OB associated proteins. The mass spectrometric analysis of the purified Pfg377 complexes identified 462 unique P. falciparum peptide sequences (supplemental Table S6) mapping to 66 protein groups (supplemental Table S7). The comparative proteomics analysis revealed three proteins overrepresented more than 30-fold in the wt samples: the bait protein Pfg377, as expected; PfPSOP12, which was identified also in the global comparative approach described above and PF3D7_1214800, a hypothetical 13 kDa protein. Altogether, the two comparative proteomic approaches identified 16 parasite gene products as putative OBCs. Five OBCs were chosen for further characterization based on (1) their conservation in P. berghei (PfGEST); (2) the results of the immunopurification experiments (the 13 kDa protein PF3D7_1214800, named hereafter OBC13, and PfPSOP12, which was identified also in the global proteomics approach), and (3) the potential role in proteolysis revealed by functional annotation PfSUB2 (Subtilisin 2) and PfDPAP2 (Dipeptidyl aminopeptidase 2).
All of these proteins display a predicted transmembrane domain and a signal peptide. Additionally, the selected genes show similar transcriptional profiles to that of pfg377, indicating functional association of the OBC proteins (34) (supplemental Fig. S3).
PfGEST is a Conserved Component of the Gametocyte Osmiophilic Bodies in Human and Rodent Malaria Parasites-An antibody against the C-terminal portion of PfGEST was produced and used to study the localization of this protein in gametocytes (supplemental Fig. S4). In WB analysis, the ␣-PfGEST antibody specifically reacted with a band of the expected molecular weight of ϳ25 kDa in samples from gametocytes and its intensity increased in stage V gametocytes as compared with the immature stage II-III (Fig. 3A). The band was not observed in extracts from equivalent number of uninfected red blood cells (RBCs) or asexual parasites. This antibody was used to investigate the PfGEST subcellular localization by IFA and immuno-electron microscopy analyses on stage IV-V gametocytes. In IFA, conspicuous ␣-PfGEST positive signal colocalized with the specific granular fluorescent pattern obtained with antibodies against the OB marker Pfg377, strongly indicating that PfGEST resides in these organelles (Fig. 3B). This was confirmed by immuno-electron microscopy, where the ␣-PfGEST-reactive colloidal gold particles were detected inside or on OBs (Fig. 3C).
These results demonstrate PfGEST presence in the OBs. However, detection of PfGEST in Pfg377 negative gametocytes, the male sexual stages that lack OBs, suggests that distribution of this protein is not restricted to OBs and that it may be present in other membranous structures (supplemental Fig. S5).
Functional Analysis Identifies New Protein Constituents of the Osmiophilic Bodies-To investigate the subcellular localization of the identified OBC proteins, reporter gene fusions between the coding sequences of respective genes and Green Fluorescent Protein (GFP) were generated and transfected in 3D7 parasites. Four lines were obtained, whose endogenous chromosomal loci were successfully tagged by the fluorescent reporter (Fig. S6). Clonal lines pfsub2GFP, pfdpap2GFP and pfobc13GFP and the uncloned line pfpsop12GFP were used for further analysis. GFP expression of the tagged proteins was first investigated in live asexual blood stages and gametocytes. In asexual stages, fluorescence was observed only in schizonts of the pfsub2GFP clone. Further experiments on these parasites showed a full colocalization of the GFP signal with that from ␣-PfAMA1 antibodies on the merozoite apical portion, which confirmed that the PfSUB2GFP fusion is correctly trafficked and localized in asexual stages (supplemental Fig. S7). The analysis of live stage V gametocytes from all GFP tagged lines showed a weak but appreciable granular GFP fluorescence only in gametocytes from the 3D7pfsub2GFP and the 3D7pfdpap2GFP clones (Fig. 4A), whereas fluorescence of the pfobc13GFP and pfpsop12GFP gametocytes was barely detectable.
To increase detection levels of the reporter proteins, IFA experiments were performed with antibodies against the GFP reporter, used alone or in combination with those against the OB marker Pfg377 and the newly identified OB component PfGEST. The ␣-GFP antibody produced a granular fluorescence pattern in gametocytes of the pfsub2GFP, pfdpap2GFP and pfobc13GFP lines (Fig. 4B, left). A different pattern was found instead in the pfpsop12GFP line, with gametocytes showing in some experiments a more condensed cytoplasmic signal, or no signal at all (data not shown). Importantly, the GFP signals in the gametocytes of 3D7pfsub2GFP, 3D7pfdpap2GFP, and 3D7pfobc13GFP transgenic clones colocalized with those of the OB marker Pfg377 and of PfGEST (Fig. 4B, right, supplemental Fig. S8), indicating that all three fusion proteins were present in OBs. In the double IFA with ␣-Pfg377 and ␣-GFP antibodies, a quantitative assessment of signal colocalization using the Costes approach (35) confirmed colocalization with Pearson's R values of 0.83 Ϯ 0.16, 0.83 Ϯ 0.13 and 0.85 Ϯ 0.03 in the pfsub2GFP, pfdpap2GFP, and pfobc13GFP gametocytes, respectively. In the pfsub2GFP line, available antibodies against a recombinant portion of the PfSUB2 protein (36) were used to confirm the localization of the endogenous protein to the Pfg377-positive structures (supplemental Fig. S9). Finally, the above double IFA experiments were conducted using a different combination of antibodies, a mouse ␣GFP and a rabbit ␣Pfg377 against the B portion of the protein (10) (supplemental Fig. S10). These experiments produced results comparable to those shown in Fig. 4B, confirming colocalization of the GFP tagged OBC proteins with Pfg377. Importantly, in all experiments the granular patterns of the GFP tagged proteins were observed exclusively in the Pfg377 positive female gametocytes and not in the Pfg377 negative male gametocytes (supplemental Fig. S11; supplemental Table  S8). These results show that, similar to Pfg377, PfSUB2GFP, PfDPAP2GFP, and PfOBC13GFP proteins localize specifically to the OBs.
Role of the Newly Identified OB Proteins in Gametocyte Egress and Persistence in the Extracellular Gametes-To investigate the function of the identified OB proteins in gametogenesis, and more specifically their role in the release of gametes from the infected erythrocyte, a phenotypic analysis was carried out on transgenic parasites where the OB genes were disrupted. This analysis was not possible for PfSUB2, reported to be essential for parasite proliferation (37). Production of pfdpap2KO parasites was previously described (38), and parasites disrupted in the pfobc13 gene were generated for this purpose (supplemental Fig. S12).
Efficiency of egress of the pfdpap2KO and pfobc13KO gametes was measured at 15 min after induction. Although a twofold decrease in egress efficiency was observed for the pfdpap2KO gametes, no difference was observed in the egress of the pfobc13KO gametes (Fig. 5).
As no obvious function in gamete egress could be ascribed to two out of three OB proteins analyzed, i.e. Pfg377 and PfOBC13, the fate of the newly identified components of these secretory organelles was investigated after gamete egress. Mature gametocytes from the pfsub2GFP, pfdpap2GFP and pfobc13GFP lines were triggered to undergo gametogenesis and the presence of the fusion proteins was investigated in live extracellular gametes. The signal from the PfSUB2GFP protein was detected in the  periphery of gametes at least 30 min after egress from RBCs, suggesting a surface localization of the fusion protein. A relatively weaker fluorescence intensity in the extracellular gametes was observed for DPAP2GFP. In contrast, the OBC13GFP protein was not detected in the egressed gametes, which was expected from the lack of GFP signal in live gametocytes from this line (Fig. 6).
In summary, these analyses indicated that both OB associated proteases PfSUB2 and PfDPAP2 are detectable in extracellular gametes at least 30 min after egress from the infected erythrocytes in live extracellular gametes, further substantiating the hypothesis that function of OB proteins may be required for the mosquito stages of parasite development.

DISCUSSION
Although secretory organelles of apicomplexan parasites play key roles in their biology, they remain poorly characterized. This is partially because of the difficulties in their isolation that prevent thorough biochemical analyses. In this work, using P. falciparum mutants that lack osmiophilic bodies and comparative proteomics, 16 proteins enriched in P. falciparum wt gametocytes were identified and four out of the five analyzed were functionally shown to be OB components.
Earlier work suggested that OBs play a role in the gametocyte egress from the red blood cells (16). However, recent studies in P. falciparum (15) and P. berghei (9) did not confirm this observation. Here, analyses of the newly established independent pfg377KO lines confirmed the role of this gene in OBs biogenesis and in parasite transmission to the mosquito vector but not in gamete egress. Results suggest that a block in the parasite development takes place at some point between gametocyte egress and the oocyst maturation in the mosquito midgut. Further studies are needed to elucidate the exact stage at which development of the OBs depleted parasites is arrested.
The almost complete lack of OBs in pfg377KO gametocytes was exploited here to develop a comparative proteomic study to identify several new OB proteins. The observed modest differences in the abundance of the OB components PfGEST, PfSUB2, PfDPAP2, and PfOBC13 suggest that pfg377 deficiency does not impact expression of OB proteins but impairs their trafficking and stability. The comparative proteomics analysis also showed that abundance levels of 26 proteins were increased in the OB-depleted parasites. It is thus possible that the defect in OB biogenesis may cause accumulation of some parasite proteins. Current functional annotation of these proteins did not provide clues on the mechanisms underlying accumulation in the defective gametocytes.
The fact that four out of five candidates analyzed from the list of 16 proteins overrepresented in the wt gametocytes were confirmed as OB proteins indicates that this comparative proteomic approach was successful. The exception was represented by PfPSOP12, whose tagged gene product did not colocalize with the signal from ␣-Pfg377 and ␣-PfGEST antibodies. This is surprising as this protein was independently identified in both the global comparative proteomics and the co-IP approaches. Moreover, similar to other OBC proteins identified in this study, the ortholog of this protein in the murine parasite P. berghei, PbPSOP12, has been detected on the gametocyte surface (39). Therefore, failure to detect PfPSOP12 in OBs could be an artifact of incorrect trafficking of the chimeric GFP-tagged fusion protein and further experiments using anti PfPSOP12 antibodies are needed to clarify this discrepancy.
The localization of proteases PfSUB2 and PfDPAP2 in the gametocyte OBs is intriguing. In asexual parasites, PfSUB2 is expressed in very mature schizonts and trafficked into the merozoite micronemes together with other invasion factors. When micronemes discharge their contents after egress from the host RBC, PfSBU2 relocates to the posterior part of the parasite (40), where its activity during RBC invasion sheds the surface proteins AMA1, MSP1 and PTRAMP from the merozoite surface in a process essential for invasion (41,42). The proteolytic activity of PfSUB2 has been shown to be sequence independent in asexual parasites, which makes it difficult to predict its substrates in sexual stage parasites (37,42). Previous attempts to disrupt PbSUB2 (43) or PfSUB2 have failed (44), suggesting an essential role for the protein in the asexual cycle. Attempts to introduce a GFP-tag in the endogenous gene had also proven unsuccessful (40), with only an HA-tag being successfully inserted into the chromosomal gene locus (41). The discovery of PfSUB2 in the secretory organelles of female gametes presented here open key questions on the role of PfSUB2 in the parasite sexual stages. As a role of the OB content in the release of gametes from the infected erythrocyte membrane seems ruled out, the presence of PfSUB2 in gametes after egress proposes a role for this enzyme in later stages of the mosquito infection. Interestingly, in P. berghei it has been observed that the rodent parasite SUB2 is secreted into the cytoplasm of the mosquito host cells invaded by ookinetes, hinting that this sheddase could have a role in this mosquito stage of the parasite (45). Recent advances in the genetic manipulation of P. falciparum (46,47) makes it now conceivable to generate parasites with a conditional PfSUB2 expression to elucidate its function in the different developmental stages.
The other OB protease identified here, PfDPAP2, is expressed only in gametocytes and its possible role in gametocytogenesis has been previously studied (38). PfDPAP2 belongs to the small conserved family of parasite DPAP proteases with PfDPAP1 and PfDPAP3. Phenotypically, DPAP2 defective parasites in the human and in the rodent parasites show a normal gametocytogenesis. The pbDPAP2KO parasites were tested for male gamete exflagellation, showing no difference compared with wt parasites, and the protease resulted to be not required for mosquito transmission (38). The work presented here reveals that PfDPAP2 is localized in OBs and that its ablation in pfDPAP2KO parasites results in a mild defect in gamete egress. Interestingly, this phenotype is the only observation that proposes a relationship between P. falciparum OBs and gamete egress. In this respect it is possible that PfDPAP2 has a role similar to that of PfDPAP3, which is localized in the micronemes of the asexual schizont where it processes proteins essential for the egress from RBCs (48).
The PfSUB2, PfDPAP2, and PfOBC13 proteins appear here to be specific components of OBs, as the GFP tagged proteins are not detectable in the Pfg377 negative male gametocytes. In contrast, PfGEST, whose localization in OBs was here confirmed by specific antibody reactivity in immunoelectron microscopy analysis, was also found in male gametocytes. The distinct granular pattern observed also in these gametocytes (data not shown) opens the possibility that other secretory organelles, negative to Pfg377, exist in P. falciparum. One instance of such organelles in male gametocytes of P. berghei has been recently described (9).
The confirmed organellar localization of the candidate proteins described here suggests that other OB components are potentially present in the list of the 16 up-regulated OBC proteins. Candidates with signal peptide and transmembrane domains remain to be explored, including the nonannotated PF3D7_1222100 gene product, the putative signal peptidase complex subunit 3, SPC3 (PF3D7_0904400) and the GPIanchored antigen PfGAMA (PF3D7_0828800). PfGAMA has been characterized in asexual parasites as a specific component of the micronemes (49), present on the merozoite surface after secretion from these organelles, and its expression was described also in ookinetes and sporozoites, the other invasive stages of the parasite (49,50). This draws a parallelism with PfSUB2 and suggests that the storage organelle of PfGAMA in the sexual stages could be the OBs, a hypothesis to be explored with available specific antibodies (50). Interestingly, gene disruption of the murine parasite pfGAMA ortholog, named pbPSOP9, phenotypically resulted in parasites with a reduced oocyst and sporozoite production, in which mosquito transmission to the mouse was abolished (51). Another shortlisted protein, RAB5b, is part of the rab GTPase protein family, associated to membrane trafficking (52). This points to a possible role of RAB5b in trafficking of components of the OBs to their cellular target.
Results of the present study expand our knowledge of the secretory organelles of P. falciparum by identifying novel proteins specific to the gametocyte OBs. The persistence in the extracellular gametes of the P. falciparum SUB2 and DPAP2 proteases, assigned here to these organelles, and the indication that OBs and their specific component Pfg377 are necessary to efficient oocyst formation identify a function of OBs in P. falciparum infection to mosquitoes and indicate OBs and their components as new targets for disrupting malaria transmission.