A Microtubule-Associated Protein Is Essential for Malaria Parasite Transmission

ABSTRACT Mature gametocytes of Plasmodium falciparum display a banana (falciform) shape conferred by a complex array of subpellicular microtubules (SPMT) associated with the inner membrane complex (IMC). Microtubule-associated proteins (MAPs) define MT populations and modulate interaction with pellicular components. Several MAPs have been identified in Toxoplasma gondii, and homologues can be found in the genomes of Plasmodium species, but the function of these proteins for asexual and sexual development of malaria parasites is still unknown. Here, we identified a novel subpellicular MAP, termed SPM3, that is conserved within the genus Plasmodium, especially within the subgenus Laverania, but absent in other Apicomplexa. Conditional knockdown and targeted gene disruption of Pfspm3 in Plasmodium falciparum cause severe morphological defects during gametocytogenesis, leading to round, nonfalciform gametocytes with an aberrant SPMT pattern. In contrast, Pbspm3 knockout in Plasmodium berghei, a species with round gametocytes, caused no defect in gametocytogenesis, but sporozoites displayed an aberrant motility and a dramatic defect in invasion of salivary glands, leading to a decreased efficiency in transmission. Electron microscopy revealed a dissociation of the SPMT from the IMC in Pbspm3 knockout parasites, suggesting a function of SPM3 in anchoring MTs to the IMC. Overall, our results highlight SPM3 as a pellicular component with essential functions for malaria parasite transmission.

Numerous studies in the model apicomplexan organism Toxoplasma gondii, identified several microtubule-associated proteins (MAPs) decorating the SPMTs, including TgSPM1/2 (30), TgTrxL1/2, and TgTLAP1/2/3/4 (31,32). MAPs are expected to define SPMT subpopulations with different properties and to modulate microtubule stability, mechanical properties, and resistance to depolymerization (30,(32)(33)(34)(35). Some of these MAPs, such as TgTrxL1, TgTrxL2, and TgSPM1, are known to localize within the lumen of SPMTs (36), while others might interact with their external surface. Amino acid repeats in the TgMAPs are important for microtubule localization and tubulin binding (30). TgSPM1 and TgTrxL1 are important for stability of the SPMTs but are not essential for parasite development, suggesting remarkable functional redundancy (30,36). TgAC9 and TgAC10 were shown to be required for the organization of the SPMTs at the apical cap and mediate interaction to IMC components (37), while TgDCX has been shown to modify the structure of tubulin in the conoid (38).
Homologues of most of the TgMAPs can be found in the genomes of Plasmodium species, but none has been experimentally characterized for either asexual or sexual developmental stages so far. The presence of Plasmodium MAPs within the SPMT lumen was suggested previously (39,40), but SPMT structures resembling those expected for P. falciparum SPM1 (PfSPM1) and PfTRXL1 in the SPMT lumen only recently identified by cryo-electron microscopy (cryo-EM) and modeling, supporting the idea that these proteins have a role in SPMT function (26).
Although the findings of Ferreira et al. (26) and the presence of several homologues to T. gondii MAPs in the P. falciparum genome suggest that potentially many PfMAPs might have functions similar to those in T. gondii, differences in the malaria parasite life cycle and the structure of microtubules open up the possibility that they have evolved MAPs with unique functions.
Here, we identified SPM3 (PF3D7_1327300/PBANKA_1342500) as a novel MAP that is highly conserved within Plasmodium spp., especially within the clade Laverania, but absent in other Apicomplexa such as T. gondii. Characterization of SPM3 reveals its importance in falciform gametocyte morphology and sexual development in P. falciparum, as well as its role in the transmission of P. berghei sporozoites.

RESULTS
PfSPM3 is a Plasmodium-specific MAP in P. falciparum merozoites. PF3D7_ 1327300 was initially identified as a putative interaction partner of the IMC protein PfPhIL1 (41). Phylogenetic analysis revealed orthologs of PfSPM3 with high similarity (>89% amino acid similarity) in Plasmodium reichenowi and P. gaboni (Fig. 1A). Other Plasmodium species also appear to have homologous proteins, but with significantly lower similarity (,50% amino acid similarity) than in Laverania parasites. The only homologous protein sequence outside the genus Plasmodium was found in Hepatocystis sp. (Fig. 1A).
To test putative IMC association and to investigate the physiological role of this protein in P. falciparum, we first generated transgenic parasites expressing a C-terminal green fluorescent protein (GFP) tag. Genomic integration of an SLI (selection-linked integration)-based plasmid (42) into the spm3 locus of the 3D7 parental strain was verified by PCR (see Fig. S1A in the supplemental material). Expression and localization of the fusion protein were analyzed by fluorescence microscopy. The GFP fusion protein was detectable from the trophozoite stage onward, in agreement with its transcriptional profile (43,44). Unexpectedly, its localization did not resemble the localization of any IMC-resident proteins but was reminiscent of SPMT distribution (45) (Fig. 1B). Localization of GFP-tagged PfSPM3 at the microtubules was visualized by colocalization with TubulinTracker (Fig. 1C, first row), showing partial overlap with a more pronounced signal of PfSPM3 toward one end of the SPMT. This SPMT association was substantiated by colocalization with an overexpressed mCherry-tagged PfSPM1 (PF3D7_0909500) (Fig. 1C, second row), the homologue of TgSPM1 (30), which showed a distribution similar to that of TubulinTracker, with some cytosolic background.
IMC markers shows, in agreement with the PfSPM1 and TubulinTracker investigations, that PfSPM3 lies in close proximity to, but spatially distinct from, the IMC.
PfSPM3 is dispensable for intraerythrocytic development. To assess the function of PfSPM3 in asexual development of P. falciparum parasites, a glucosamine-inducible degradation system was used to degrade PfSPM3 mRNA (glmS ribozyme system) (53) ( Fig. S1A and B). The glmS ribozyme was introduced upstream of the 39 untranslated region in the modified spm3 locus of the 3D7 parental strain. Although conditional knockdown of PfSPM3-GFP upon addition of 2.5 mM glucosamine resulted in reduction of the GFP signal ( Fig. 2A), it had no measurable effect on either parasite proliferation during asexual blood-stage parasite development (Fig. 2B) or schizont morphology ( Fig. 2A). To exclude the possibility that residual amounts of PfSPM3 upon knockdown might be sufficient for parasite development, we tested an additional approach using targeted gene disruption (TGD) (42) that truncates most of the protein and leads to the expression of only the N-terminal 136 amino acids instead of the 1,795-amino-acid wild-type (WT) PfSPM3 protein (Fig. S1B). For further phenotypical characterization of this truncation, the cell line was generated in a 3D7-iGP background (54). Successful integration and consequent disruption of the Pfspm3 gene were validated by PCR (Fig. S1A). Expression and localization of the truncated protein fragment were monitored by fluorescence microscopy. In contrast to full-length PfSPM3, this truncated version is not connected with the SPMT but exhibits a cytosolic distribution (Fig. 2C). Nevertheless, and in agreement with the knockdown approach, these PfSPM3-deficient parasites showed neither a growth nor a morphological phenotype ( Fig. 2C and D). Congruently, these parasites revealed an SPMT and IMC architecture indistinguishable from that of wild-type parasites, as visualized by coexpression of the IMC marker proteins ALV5-mCherry (48,49) and PF3D7_1345600-mCherry (52) or by costaining with  Live-cell microscopy of 3D7-iGP-PfSPM3-GFP gametocytes (day 7 to day 13 after gametocyte induction) cultured either with or without (control) 2.5 mM GLCN. Microtubules were visualized with TubulinTracker, and nuclei were stained with Hoechst 33342. Bar, 2 mm. (B) Gametocytemia at day 10 after gametocyte induction was determined by counting 1,096 to 2,632 (mean, 1,793) cells per condition in Giemsa-stained thin blood smears. The relative gametocytemia values were obtained by dividing the gametocytemia of GLCN-treated cultures by the gametocytemia of the corresponding untreated cultures. Values are means and SD from independent growth experiments, with the number of experiments (n) indicated. A two-tailed unpaired t test with Welch's and Benjamini-Hochberg corrections was used to calculate multiplicityadjusted P values for 3D7-iGP-PfSPM3-GFP-glmS or 3D7-iGP-PfSPM3-GFP-glmSM9 versus 3D7-iGP parasites all cultured with 2.5 mM GLCN. (C) Classification and quantification of gametocyte morphology within parasite populations of 3D7-iGP, 3D7-iGP-PfSPM3-GFP-glmS, and 3D7-iGP-PfSPM3-GFP-glmSM9 cultured either with or without (control) 2.5 mM GLCN. Between 589 and 2,632 erythrocytes and 19 to 69 (mean, 48) parasites per condition in Giemsa-stained thin blood smears were analyzed. Yellow, falciform; blue, pycnotic; green, abnormal. Representative images are shown on the right at day 10 after gametocyte (Continued on next page) Microtubuli-Associated Protein in Malaria Transmission mBio TubulinTracker (Fig. 2E), confirming the dispensability of PfSPM3 for asexual bloodstage development.
PfSPM3 is important for falciform gametocyte morphology. In addition to the 3D7-based transgenic parasites, we generated a corresponding cell line in the 3D7-iGP (54) background. This cell line (3D7-iGP-PfSPM3-GFP-glmS) allows the robust induction of sexual commitment of the parasite and therefore enables the localization and assessment of PfSPM3 function during gametocytogenesis. As expected, fluorescence microscopy revealed close SPMT localization of PfSPM3 in gametocytes (Fig. 3A). Of note, the PfSPM3-GFP signal disappears earlier than the TubulinTracker signal, indicating loss of detectable PfSPM3 prior to the disassembly of the SPMTs in late gametocyte stages (IV/V gametocytes) (Fig. 3A). Conditional knockdown via the glmS system resulted in significantly reduced numbers of gametocytes (mean of 52.6% 6 16.1% at day 10 postinfection) compared to control cultures (Fig. 3B). Gametocytes were morphologically distinct and appeared as malformed (round) and nonfalciform gametocytes. Importantly, these PfSPM3-depleted gametocytes revealed an aberrant SPMT pattern ( Fig. 3A and C). To exclude detrimental effects of glucosamine on gametocyte development and SPMT architecture, we assessed the phenotype in the control 3D7-iGP-PfSPM3-GFP parasites with an inactive M9 mutant ribozyme sequence (53). No difference in gametocyte morphology or gametocytemia was observed in these parasites in the presence of 2.5 mM glucosamine (GLCN) ( Fig. 3B and C and Fig. S2A). The phenotype was further confirmed by using the 3D7-iGP-PfSPM3-TGD parasite cell line, which expresses the truncated, cytosolic PfSPM3. As for the glmS-based approach, after induction of gametocytogenesis the progression and maturation of gametocytes was heavily impaired, resulting in the absence of falciform gametocytes and leading to the formation of malformed, round gametocytes with aberrant SPMT organization ( Fig. 3D and E and Fig. S2B). Altogether, these results indicate an essential function of PfSPM3 in gametocyte development and the falciform morphology of P. falciparum gametocytes.
SPM3 deletion in P. berghei impacts sporozoite motility and transmission by mosquitoes. To extend our functional investigation into the role of SPM3 in mosquito stages, we turned to the rodent model parasite P. berghei, for which the full life cycle is readily available. We first generated a P. berghei parasite line in which the gene ortholog of P. falciparum spm3 (PbSPM3, PbANKA_1342500) was tagged with GFP (PbSPM3-GFP) ( Fig. S3A and B). PbSPM3-GFP was detectable in oocysts, as well as in midgut and salivary gland sporozoites. In sporozoites, the fluorescent signal extended from the apical end toward the rear around the nucleus ( Fig. 4A and Fig. S4A). Subsequent costaining with an antitubulin antibody revealed colocalization of PbSPM3-GFP with tubulin ( Fig. 4A) in midgut and salivary gland sporozoites. C-terminal tagging of PbSPM3 with GFP did not affect parasite life cycle progression, as SPM3-GFP parasites showed normal midgut infection and midgut oocyst numbers as well as numbers of both midgut and salivary gland sporozoites comparable to those of the wild type ( Fig. S4B and C).
In order to probe into PbSPM3 function, we generated a clonal PbSPM3 knockout (PbSPM3-KO) cell line ( Fig. S3C and D). Similar to the P. falciparum gene disruption, no apparent difference in blood-stage growth was detected for PbSPM3-KO compared to our previously determined growth rates of P. berghei wild-type parasites (55, 56) (Fig. 4B). Next, to investigate the progression of this parasite line in mosquitoes, we fed naive Anopheles stephensi mosquitoes on mice infected with PbSPM3-KO parasites. Surprisingly, we found that infection and colonization of the mosquito midgut by PbSPM3-KO parasites were similar to those by wild-type parasites, as measured by the induction. (D) Quantification of gametocyte morphology (yellow, falciform; blue, pycnotic; green, abnormal) at day 10 after gametocyte induction for 3D7-iGP (control) and 3D7-iGP-SPM3-TGD parasites. For each condition, the proportion of parasite stages in erythrocytes was determined in three independent experiments (total numbers of erythrocytes screened for 3D7-iGP were 2,694, 1,431, and 2,742; numbers for 3D7-iGP-SPM3-TGD were 1,328, 1,144, and 925) and displayed as a percentage. (E) Representative Giemsa smears of 3D7-iGP and 3D7-iGP-PfSPM3-TGD stage II to V gametocytes (day 5 to day 13 after gametocyte induction). Bar, 5 mm. (B) Deletion of pbspm3 does not affect asexual blood stage growth rate. Asexual blood stage growth rates were calculated based on the parasitemia value at day 8 postinfection with a single infected red blood cell intravenously. PbANKA WT growth rates were plotted as a reference and were determined and previously published by our laboratory (55,56,69). Mean fold change within a single parasite replication cycle (24 h for P. berghei) with SD is presented. (C) Similar oocyst numbers and infection rates of wild-type and PbSPM3-KO-infected A. stephensi. Shown are pooled data from two (WT) and three (PbSPM3-KO) independent cage feeds. A total of 79 (WT) and 151 (PbSPM3-KO) mosquitoes were analyzed on day 11 and day 12 postinfection as technical replicates, of which 78.5% (62) and 88.1% (133) were infected, respectively (P = 0.08, Fisher's exact test). For statistical analysis comparing oocyst numbers in infected midguts, a Mann-Whitney test was performed. The black line indicates the median, with error bars representing the IQR. (D) Ratio of salivary gland-resident versus midgut-resident sporozoites determined at day 18 and 20 postinfection from two (WT) and three (PbSPM3-KO) independent cage infections. The P value was determined with an unpaired t test with Welch's correction. The black line indicates the median, with error bars representing the IQR. (E) Percentage of wild-type and PbSPM3-KO sporozoites isolated from salivary glands displaying movement after attachment or floating in the medium. Pooled data from two independent cage feeds (WT) with 139 sporozoites and three (PbSPM3-KO) with 109 sporozoites, with two technical replicates per cage infection. For statistical analysis comparing differences in gliding categories, a chi-square test was performed. (F) Different types of (Continued on next page)  (Table 1). These findings indicate that neither gametocyte development, exflagellation, ookinete development, nor midgut traversal is significantly affected by PbSPM3 gene knockout. Next, we tested whether PbSPM3-KO might affect the further development of sporozoites. While salivary glands of mosquitoes infected with wild-type parasites showed over 20,000 sporozoites (median, 21,000; IQR, 18,000 to 34,000), only a few hundred were found for PbSPM3-KO sporozoites (median, 300; IQR, 220 to 2,690), indicating a drastic (99%) decrease in the ratio of salivary gland sporozoites to midgut sporozoites ( Fig. 4D; Table 1).
To pinpoint the cause for the reduced number of PbSPM3-KO sporozoites in salivary glands, we assessed the effect of PbSPM3 deletion on sporozoite motility, an essential feature for the parasite entry into salivary glands and for establishing infection in human liver cells. Wild-type P. berghei sporozoites isolated from salivary glands attach and migrate on flat surfaces in a counterclockwise circular manner at speeds exceeding 1 mm/s (40). While 60% of the wild-type sporozoites attached and moved on the glass coverslip, most PbSPM3-KO sporozoites (61%) remained floating in the medium (Fig. 4E). For the majority of PbSPM3-KO sporozoites that managed to attach and move, a striking new phenotype of sporozoite locomotion was revealed: the parasites moved in a helical way ( Fig. 4F and G and Movies S1 to 3) and deviated from the circular path observed in most of the wild-type sporozoites. The few PbSPM3-KO sporozoites that showed persistent circular gliding motility moved with a speed similar to that of wild-type sporozoites (Fig. 4H).
To further dissect the physiological consequences of this aberrant cell motility, we investigated the transmission capacity of mosquitoes infected with PbSPM3-KO parasites. We found that only 3 of 12 mice became infected after being bitten by PbSPM3-KO infected mosquitoes, while all mice bitten by mosquitoes infected with wild-type  Table 1. Note that mice from two replicates remained negative over the course of parasitemia monitoring, and hence, the parasitemia curve of replicate 2 was slightly nudged for better visibility. Also, lines in Kaplan-Meier plots in panels J and I were slightly nudged for better visibility. (J) Parasitemia curves and Kaplan-Meier plots showing increases in blood-stage parasitemia (left) and mouse survival after injection of 1,000 sporozoites intravenously. Two independent infections for the wild type (gray and black) and PbSPM3-KO (red and orange) using 3 or 4 mice per experiment (totals: WT, n = 7; PbSPM3-KO, n = 8). % parasitemia at day 6 (infected only) WT 34,000 6 6,000 21,000 6 1,000 3.6 (7/7) 1.4 3.7 (7/7) 1.5 55,000 6 16,000 27,000 6 10,000 PbSPM3-KO 70,000 6 22,000 3,000 6 1,000 4.6 (8/8) 0.8 5.7 (3/12) 0.3 90,000 6 9,000 200 6 40 67,000 6 18,000 300 6 50 a n = 2 for WT, n = 3 for PbSPM3-KO transmitted by bite, and n = 2 for PbSPM3-KO transmitted by intravenous (i.v.) injection. Parasitemia values are given for all mice that became positive within 20 days; C57/BL6 mice were used. parasites became infected ( Fig. 4I; Table 1). Furthermore, we noticed a delay in appearance (prepatency) of 2 days in mice that did become infected with PbSPM3-KO parasites. This drop in infectivity is likely due to the much lower load of sporozoites within the salivary glands. We next tested mouse liver infectivity of PbSPM3-KO sporozoites by direct intravenous inoculation of an equal number of purified sporozoites, as the lower level of sporozoites in salivary glands made direct transmission from mosquito to mice unreliable for this assay (57). All mice infected through the intravenous injection of 1,000 PbSPM3-KO sporozoites became infected (Fig. 4J). These results suggest that the reduced number of PbSPM3-KO sporozoites in the salivary glands and their aberrant motility affect the transmission from the mosquito to the mammalian host, but not the invasion of liver cells.
PbSPM3 deletion disturbs SPMT architecture in P. berghei. To identify the cause of the aberrant motility of PbSPM3-KO sporozoites, we investigated the SPMT organization in fixed midgut sporozoites by transmission electron microscopy (TEM). Preparation of thin sections from oocysts allows the imaging of sporozoites in cross sections (Fig. 5A). We found that near the apical end of PbSPM3-KO sporozoites, the arrangement of SPMTs largely followed that of wild-type sporozoites, with a single SPMT being somewhat apart from the others and all SPMTs being closely associated with the inner membrane complex (Fig. 5A). However, we found that further from the apical end (as discernible by an increased sporozoite diameter, the presence of apical organelles such as micronemes and rhoptries, and an increased intermicrotubule distance), the SPMTs were not associated with the IMC and often appeared deep within the cytoplasm (Fig. 5A and B and Fig. S5A to C). This suggests a dissociation from the IMC. We next quantified the distance between SPMTs and IMC by measuring from microtubule center toward the inner side of the IMC. Also, the SPMT-IMC distance was increased in the SPM3-KO parasites, with only 1% of SPMTs being more than 40 nm apart from the IMC in the wild type compared to 19% in the PbSPM3-KO mutants ( Fig. 5C; Fig. S4D). We also observed SPMTs associated with non-IMC membranes in the cytoplasm for both WT and SPM3-KO parasites (Fig. 5A, magenta). There was no difference between wild-type and SPM3-KO parasites in the number of SPMTs associated with these membranes or the distance between SPMTs and the membranes ( Fig. S4E and F). Taken together, these observations point to a role of PbSPM3 in anchoring the SPMT to the IMC in sporozoites.

DISCUSSION
In this study, we identified SPM3, a novel Plasmodium-specific microtubule-associated protein critical for gametocyte development and morphology of P. falciparum gametocytes and motility of P. berghei sporozoites. PfSPM3 is a high-molecular-weight protein with no predicted transmembrane domain, initially identified as a potential IMC component by PfPhIL1-based proximity-dependent biotinylation in schizonts (41) and by PfBLEB-based proximity-dependent biotinylation in stage II-III gametocytes, a protein (PfBLEB) associated with the basal complex of the parasite (58). By using colocalization approaches, we showed that PfSPM3 colocalizes with the SPMTs in asexual and sexual stages and only partially with the IMC compartment. In asexual stages, PfSPM3 appears to be more pronounced at one pole of the SPMTs, in contrast to PfSPM1, a homologue of a well-characterized T. gondii MAP, which showed a homogenous pattern along the microtubule. MAPs display restricted localization patterns along the SPMTs and may compartmentalize different regions of the microtubule by interacting with pellicular structures such as the IMC (32). Of note, similar to the pronounced apical localization observed for PfSPM3, TgTLAP3 has been shown to selectively coat the SPMTs in the apical cap region and the intraconoidal microtubules of both mother and daughter cells (32).
PfSPM3 appears to be-in agreement with genome-wide mutagenesis screens (59, 60)-dispensable for asexual blood stage proliferation, since depletion of the protein showed no growth effect, indicating redundancy of MAPs or a stage-specific function. In accordance with the latter, we provide experimental evidence that SPM3 is a critical factor for gametocyte development and morphology in P. falciparum. Depletion of PfSPM3 led to an aberrant organization of the SPMTs in gametocytes, a reduction in the number of gametocytes, and the formation of round gametocytes differing from the typical falciform shape of P. falciparum.
The drastic morphological transformation observed during gametocytogenesis from a committed ameboid trophozoite to a banana-shaped, falciform gametocyte is orchestrated by the expansion of the IMC network and the extension of SPMT arrays (22). These pellicular components maintain the elongated shape of P. falciparum and other Laverania gametocytes (24). SPMTs also confer rigidity to the immature gametocytes, which is proposed to be involved in the sequestration of gametocytes in the microvasculature and tissues of the human host to avoid splenic clearance (25). Later depolymerization and disassembly of the SPMT are expected to increase cellular deformability and release into the bloodstream, where gametocytes are taken up by the feeding mosquitoes. The falciform morphology is the hallmark of Plasmodium species within the clade Laverania and is expected to favor the transmigration from mature gametocytes from the bone marrow reservoir to the bloodstream (22,24). Nevertheless, the molecular components that determine this morphology remain largely unknown.
The phenotype observed in gametocytes upon depletion of PfSPM3 and the low conservation of the protein outside the subgenus Laverania could indicate that Laverania SPM3 proteins are somehow involved in the falciform architecture of the late-stage gametocytes. Consistent with this notion, we observed no evident phenotype in the roundish P. berghei gametocytes from rodents. Previous studies with nonfalciform round gametocytes of P. berghei and P. knowlesi show a discontinuous IMC network and pellicle architecture that does not surround these stages (61). This reflects the pronounced structural differences in gametocyte development between the rodent-infecting P. berghei and the human-infecting P. falciparum parasites (25). The stage-specific phenotypes observed in P. berghei and P. falciparum indicate that in the different species, MAPs can be repurposed according to requirements of the different stages and might be explained by the limited conservation outside the clade Laverania. However, the phenotype in P. falciparum SPM3-depleted gametocytes and mosquito stages could not be further assessed in this study due to the intrinsic deficiency in the ability of the 3D7 strain to exflagellate in vitro and to develop in the mosquito host. Further studies with NF54 would be necessary to elucidate the role in mosquito stages, but such investigations are beyond the scope of this study.
Nevertheless, to investigate the role of SPM3 in mosquito stages, we switched to the malaria mouse model. There, P. berghei sporozoites showed a severe defect in motility and the ability to enter salivary glands and consequently a lower transmission efficiency. This phenotype could be explained by the impaired SPMT organization due to the lack of microtubule anchoring to the IMC (Fig. 5D). Microtubules in sporozoites are organized in a polar fashion, with all the microtubules underlying the IMC and all but one extending over two-thirds of the circumference and a single microtubule being located in the remaining third (62). How this asymmetry is formed is not clear, but it has been speculated that this polarity contributes to the counterclockwise continuous circular movement of Plasmodium sporozoites (40,63). This suggests that displacement of microtubules from the IMC could affect persistent circular motility, as we found here for PbSPM3-KO sporozoites ( Fig. 4E to G). In line with this, the observed helical movement is reminiscent of the motion observed for P. berghei ookinetes and T. gondii tachyzoites (64,65), where the microtubules are equally spaced around the circumference.
Curiously, the deletion of SPM3 causes a problem for P. berghei sporozoites but not ookinetes, while SPM3 is expressed in both stages (66,67). We found previously that mutations in either actin or actin-associated proteins that were expressed in both ookinetes and sporozoites invariably caused problems for sporozoites and not ookinetes (68)(69)(70). Those differences could arise due to the different requirement for speed and force generation by these parasite forms. Likely, the sporozoites are more susceptible to subtle changes. In keeping with this line of thought, one possible explanation for an absence of an SPM3-KO phenotype in ookinetes could be their larger number of microtubules (around 50 compared to the sporozoites' 16) (7,26,62,71) and their lower speed as they penetrate the midgut (72)(73)(74). Additionally, or alternatively, there might be a redundant protein expressed in ookinetes that is absent from sporozoites. More work will be required to understand these intriguing processes in more detail.
MAPs influence several properties and define individual subpopulations of SPMT in the different stages of apicomplexan parasites. The exact role of SPM3 in microtubule organization and pellicle architecture remains to be elucidated. The close association of the SPMTs with the cytosolic face of the parasite pellicle is expected to be mediated by MAPs (11,40,75). We can speculate that in P. falciparum, SPM3 in gametocytes provides specialized contact sites between SPMTs and the IMC that support SPMT organization, cell elongation, and ultimately falciform cell shape. Consequently, loss of SPM3 may disrupt the interaction with IMC components such as PhIL1, alveolins, and other IMC proteins, leading to SPMT disorganization and rounding of the cell. Additional PfSPM3-specific interaction partners in gametocytes remain to be identified. In accordance, our ultrastructural analysis in P. berghei sporozoites suggests that PbSPM3 might connect the SPMTs to the IMC. Notably, the function of SPM3 appears to be early in gametocyte development, since the PfSPM3 signal vanishes prior to the disassembly of SPMTs in stage IV/V gametocytes. Association of PfSPM3 to the SPMT may also influence stability and depolymerization of SPMT as shown for T. gondii SPM1, TLAP2, and TLAP3, MAPs that protect the stability of the SPMT in a region-dependent manner (32).
In conclusion, we identified here a Plasmodium-specific microtubule-associated protein that has essential functions for gametocyte and sporozoite development and hence is critical for malaria transmission.

MATERIALS AND METHODS
P. falciparum culture. Blood stages of P. falciparum 3D7 (76) were cultured in human red blood cells (O 1 or B 1 ; Blood Bank, Universitätsklinikum Hamburg-Eppendorf). Cultures were maintained at 37°C in an atmosphere of 1% O 2 , 5% CO 2 , and 94% N 2 using RPMI complete medium containing 0.5% Albumax according to standard protocols (77). In order to obtain highly synchronous parasite cultures, late schizonts were isolated by Percoll gradient centrifugation (78) and cultured with fresh erythrocytes for 4 h. Afterwards, sorbitol synchronization (79) was applied in order to remove remaining schizonts, resulting in a highly synchronous ring-stage parasite culture with a 4-h age window.
Induction of gametocytogenesis was done as previously described (54,80). Briefly, GDV1-GFP-DD expression was achieved by addition of 2 or 4 mM Shield-1 to the culture medium, and gametocyte cultures were treated with 50 mM N-acetyl-D-glucosamine (GlcNAc) for 5 days starting 72 h after Shield-1 addition to eliminate asexual parasites (81). Alternatively, asexual ring-stage cultures with >10% parasitemia, cultured in the presence of choline, were synchronized with sorbitol (79) and washed twice in choline-free RPMI medium. Cells were kept in choline-free medium for the entirety of the assay. After one reinvasion cycle, cultures at the trophozoite stage were treated with 50 mM GlcNAc (81) and kept on this for 5 days. Gametocytes were maintained in RPMI complete medium containing 0.25% Albumax and 0.25% sterile filtered human serum (Interstate Blood Bank, Inc., Memphis, TN, USA).
For SLI-based targeted gene disruption (SLI-TGD) of PfSPM3, a 399-bp homology region was amplified using 3D7 gDNA and cloned into the pSLI-TGD plasmid (42) using NotI and MluI restriction sites.
For overexpression constructs, the full-length sequences of PhIL1 (PF3D7_0109000), SPM1 (PF3D7_ 0909500), and PF3D7_1345600 were amplified from parasite cDNA and cloned into pARL-ama1 AIP-mCherry-yDHODH (41) using the XhoI and KpnI restriction sites. Oligonucleotides used to generate the DNA fragments are summarized in Table S1.
For gene deletion of PbSPM3 (PBANKA_1342500), both a 39 and a 59 homology region (750 and 717 bp) were amplified from PbANKA wild-type-genomic DNA and cloned into the Pb262 vector (84) using HindIII/XhoI and EcoRI/EcoRV restriction sites. The Pb262 vector contains the hDHFR gene, which allows for positive selection using the drug pyrimethamine. Prior to transfection, the vector was linearized using SacII and PmeI followed by ethanol precipitation.
For endogenous tagging of PbSPM3 (PBANKA_1342500), a 992-bp 39 homology region was amplified from PbANKA wild-type genomic DNA. The reverse primer encoded the linker sequence that was used for GFP tagging of PfSPM3, as described by Birnbaum et al. (42). The amplicon was cloned into the pL18 vector (85) using EcoRI/BamHI restriction sites. Prior to transfection, the vector was linearized using the SwaI restriction enzyme followed by ethanol precipitation.
All oligonucleotides used to generate DNA fragments as well as those used for genotyping PCRs are listed in Table S1.
Transfection of P. falciparum. For transfection, Percoll-purified (78) parasites at the late schizont stage were transfected with 50 mg plasmid DNA using Amaxa Nucleofector 2b (Lonza, Switzerland) as previously described (86). Transfectants were selected using either 4 nM WR99210 (Jacobus Pharmaceuticals) or 0.9 mM DSM1 (87) (BEI Resources). In order to select for parasites carrying the genomic modification via the SLI system (42), G418 (Thermo Fisher, USA) at a final concentration of 400 mg/mL was added to a culture with about 5% parasitemia. The selection process and integration test were performed as previously described (42).
Generation of P. berghei parasite lines. The linearized PbSPM3-KO vector was transfected into the unmodified P. berghei strain ANKA using standard protocols (88). Parasites that integrated the desired DNA construct were selected by administration of pyrimethamine (0.07 mg/mL) via the mice's drinking water. Parasites from transfections are mixed populations, and hence, a limiting dilution was performed to isolate clonal lines. For this, a single blood-stage parasite was injected into each of 8 Swiss mice intravenously. Once mice reached 1 to 3% parasitemia, blood was collected via cardiac puncture from anesthetized mice (100 mg/kg ketamine and 3 mg/kg xylazine; Sigma-Aldrich). Whole-blood aliquots were stored in liquid nitrogen. Genomic DNA was isolated from whole blood and tested for correct construct integration by PCR. Erythrocytes were lysed in 1.5 mL phosphate-buffered saline (PBS) containing 0.03% saponin. After centrifugation and washing, the genomic DNA was isolated using a blood and tissue kit (Qiagen Ltd.) according to the manufacturer's protocol. All generated isogenic parasite lines were analyzed via PCR. Parasite lines with the correct genotype were assumed to be isogenic.
Determination of P. berghei asexual blood stage growth rate. Blood smears from all mice used for limiting dilution and carrying an isogenic parasite line were done. The parasitemia determined on day 8 postinfection was used to back-calculate the growth rate, as described elsewhere (89,90).
Mosquito infection. Frozen parasite stocks were thawed and injected intraperitoneally (100 to 150 mL) into one mouse. Parasites were allowed to grow for 4 to 6 days, with the infection rate being monitored by blood smears. Once the infected mouse reached 2 to 3% parasitemia, the mouse was anesthetized (100 mg/kg ketamine and 3 mg/kg xylazine; Sigma-Aldrich) and bled by cardiac puncture. From this donor mouse, 20 million parasites were transferred into two naive recipient mice by intraperitoneal injection. Three days after fresh blood transfer, exflagellation of parasites was checked first; mice were anesthetized (120 mg/kg ketamine and 16 mg/mL xylazine) and then placed in a cage containing around 200 female Anopheles stephensi mosquitoes, which had been starved for 3 to 5 h by removal of sugar and salt pads. Mosquito were allowed to feed on mice for at least 15 min up to 30 min. Infected mosquitos were kept at 21°C and 70% humidity.
Exflagellation assay in vitro. A drop of tail blood from an infected mouse was placed onto a glass slide, tightly covered with a cover glass, and placed at 21°C for 12 min. Exflagellation centers were observed by light microscopy at Â40 magnification with a phase-contrast ring on a Zeiss Axiostar microscope. The number of exflagellation centers was roughly counted in several fields with comparable erythrocyte densities.
Analysis of oocyst development by mercurochrome staining. Midguts of infected mosquito were isolated on day 11 and day 12 after a mosquito blood meal from three independent mosquito feeds. A minimum of 50 mosquitos per feed replicate were dissected. Midguts were dissected in 100 mL of PBS and permeabilized with 1% Nonidet P40 in PBS (AppliChem GmbH) for 20 min at room temperature (RT). The supernatant was removed, and midguts were stained for 30 min to 2 h in 0.1% mercurochrome in PBS (NF XII; Sigma-Aldrich). The staining solution was discarded, and midguts were washed several times with PBS until the supernatant became clear. Stained midguts were transferred onto a glass slide, leaving some PBS on top, and sealed with a coverslip. Midguts were imaged using an epifluorescence microscope (CellObserver; Zeiss) using a 10Â (numerical aperture [NA], 0.5; air) objective with a green filter, and both the infection rate and the number of oocysts were determined.
Sporozoite isolation, counting, and imaging. Midgut and salivary gland sporozoites of infected mosquitoes were isolated on day 18 and day 20 after a mosquito blood meal from three independent mosquito feeds. A minimum of 10 mosquitoes was dissected per count. The midguts and salivary glands were dissected on ice in RPMI and crushed with a pestle before counting sporozoites using a Neubauer counting chamber. For imaging, the solution containing crushed salivary glands was filled up to 1 mL with RPMI, and was carefully underlaid with 3 mL of 17% Accudenz. Centrifugation at 2,800 rpm for 20 min at room temperature separated sporozoites from cell debris. The sporozoite-containing interphase was collected (total volume of 1.4 mL), and sporozoites were pelleted for 3 min at 1,000 rpm (Biofuge Primo; Thermo Fisher Scientific). The supernatant was removed; RPMI containing 3% bovine serum albumin (Carl Roth) was added to the sporozoites, and the resulting mixtures were transferred into the wells of an optical-bottom 96-well plate (Thermo Fisher Scientific). The plate was centrifuged for 3 min at 1,000 rpm (Multifuge S1-R; Heraeus) and imaged using an epifluorescence microscope (CellObserver; Zeiss). Movies were obtained using a 25Â (NA, 0.8; water) objective at a speed of 1 frame every 3 s with 100 cycles imaged per movie. Movies were taken until 1 h after sporozoite activation. Videos were analyzed for 5 min with Fiji (version 2.0.0 rc 64/1.51s) (91). Gliding assays were carried out as technical replicates on day 19 and day 22 for each cage infection. For analysis, data from two (WT) and three (PbSPM3-KO) biological replicates were pooled and analyzed in a single-blind fashion. Gliding motility was categorized into different patterns as described before (92). Briefly, sporozoites' motility behavior can be distinguished as follows: continuous (persistently moving sporozoites for at least 50 of 100 frames without stopping more than 10 frames in a row), partial (sporozoites moving in a circular manner but not persistently), helical (sporozoites moving in a new type of motility as illustrated in Fig. 4E and F), attached (sporozoites adhering to the substrate but showing no sign of movement), patch (sporozoites moving actively over a single adhesion spot in a back-and-forth manner), wavers (sporozoites attached with one end and the other end moving actively through the medium), and floating (sporozoites floating in the medium without contact with the substrate).
Mouse infection by natural transmission and intravenous injection of sporozoites. For natural transmission via mosquito bite, 10 female mosquitos that had had their blood meal 21 days before the experiment were transferred into cups the day before the experiment. Mosquitoes were starved for 3 to 5 h the next day by removal of sugar and salt pads. Naive C57/BL6 mice were anaesthetized with 120 mg/kg ketamine and 16 mg/mL xylazine, and one mouse was placed per cup, allowing mosquitoes to feed for at least 15 min up to 30 min with a minimum number of 6 mosquitos having fed. To confirm that mosquitos were infected, salivary glands of all blood-fed mosquitos were dissected as described above, and sporozoite numbers were determined.
For intravenous injection of sporozoites, salivary gland sporozoites were dissected as described above, and 1,000 sporozoites (in 100 mL PBS) were injected into the lateral tail vein of a naive C57/BL6 mouse.
Daily blood smears were taken from day 3 postinfection at approximately the same time as the infection experiment. Once the infected mice reached 2 to 3% parasitemia, they were sacrificed by cervical dislocation. Mice that did not become positive 20 days postinfection were considered negative.
Fluorescence imaging of fixed P. berghei sporozoites. Microtubules and SPM3 localization of fixed Plasmodium berghei sporozoites were visualized using a Zeiss Airyscan 2 LSM900 laser scanning confocal microscope. Both midgut and salivary gland sporozoites were isolated as described above but without subsequent Accudenz purification. Instead, sporozoites were directly pelleted for 3 min at 13,000 rpm (Biofuge Primo; Thermo Fisher Scientific). The supernatant was removed, and sporozoites were resuspended in RPMI containing 3% bovine serum albumin (Carl Roth). For imaging, 30,000 to 50,000 sporozoites were transferred into a well of a Labtek slide (m-slide, 8-well ibiTreat, no. 80826). Sporozoites were pelleted for 3 to 4 min at 800 rpm (Multifuge S1-R; Heraeus) and then incubated for 10 min at room temperature, allowing sporozoites to attach and start gliding while leaving the supernatant on top. Sporozoites were washed three times with RPMI before being fixed in 4% paraformaldehyde (PFA) in PBS at room temperature. After 1 h of incubation, fixative was removed by washing sporozoites three times with RPMI followed by permeabilization with 0.5% Triton X-100 in PBS for 1 h at room temperature. Before incubation with primary antibody, sporozoites were again washed three times with RPMI. Sporozoites were incubated for 1 h at room temperature with mouse antitubulin (1:500; Sigma-Aldrich, no. T5168) to stain microtubules, followed by three washing steps with RPMI. Alexa Fluor 546-conjugated goat anti-mouse immunoglobulin (1:500; Invitrogen, no. A11030) was used as the secondary antibody and incubated for 30 min up to 1 h at RT before incubation of sporozoites for 15 min in RPMI containing 1:1,000-diluted Hoechst 33342 from a 10 mM Hoechst 33342 stock solution (Thermo Fisher Scientific). Sporozoites were washed three times in RPMI and subsequently imaged. Z-stacks were acquired with a z-spacing of 0.5 mm and a total of 29 layers being taken. Images were directly 3D processed by the internal 3D Airyscan processing option.
Fluorescence imaging of live P. berghei parasites. Midguts were isolated as before and transferred onto a glass slide, leaving some PBS on top, and sealed with a coverslip. Sporozoites were activated in 3% BSA in RPMI and transferred to an optical 96-well plate (Thermo Fisher Scientific). The plate was centrifuged for 3 min at 1,000 rpm (Multifuge S1-R; Heraeus) and imaged using an epifluorescence microscope (CellObserver; Zeiss). Images from both oocysts and sporozoites were obtained using a 63Â (NA, 1.4; oil) objective with an exposure of 80 milliseconds.
TEM. Highly infected midguts were fixed by incubation in 4% paraformaldehyde and 4% glutaraldehyde diluted in 100 mM sodium cacodylate buffer at 4°C overnight. Fixed samples were washed three times in 100 mM sodium cacodylate buffer at RT for 5 min. A secondary fixation was done in 1% osmium (in 100 mM sodium cacodylate buffer) at RT for 60 min. Samples were washed twice with 100 mM sodium cacodylate buffer and twice in double-distilled water (ddH 2 O) and then contrasted with 1% uranyl acetate (in ddH 2 O) at 4°C overnight. Samples were washed twice with ddH 2 O for 10 min and then dehydrated by incubating in increasing concentrations of acetone (30%, 50%, 70%, and 90%) for 10 min and twice in 100% acetone for 10 min. Samples were adapted to Spurr's solution (23.6% epoxycyclohexylmethyl-3,4-epoxycyclohexylcarboxylate [ERL], 14.2% ERL-4206 plasticizer, 61.3% nonenylsuccinic anhydride, 0.9% dimethylethanolamine) by incubating in increasing concentrations (25%, 50%, and 75%) at RT for 45 min and at 100% at 4°C overnight. Midguts were resin embedded with Spurr's solution at 60°C overnight. Embedded midguts were trimmed, and 70-nm-thick sections were imaged on a transmission electron microscope at 80 kV (JEOL JEM-1400) using a TempCam F416 camera (Tietz Video and Image Processing Systems GmbH, Gauting, Germany). Images were analyzed with Fiji. For contrast purposes, a Gaussian blur was applied.
Microtubule numbers and distances were measured in a single-blind fashion. After all sporozoites with clearly visible microtubules had been defined, the number of microtubules per sporozoite was manually counted. The distance of microtubules to either the IMC or a membrane of a yet-unknown origin were measured by taking the distance from the center of the microtubule to the inner side of the IMC or to the membrane, respectively. A cutoff value of 40 nm was set, with distances of 40 nm or less being considered IMC-close, according to Ferreira et al. (26).
Fluorescence imaging of P. falciparum-infected erythrocytes. All fluorescence images were captured using a Zeiss Axioskop 2 Plus microscope with a Hamamatsu digital camera (model C4742-95) or a Leica D6B fluorescence microscope equipped with a Leica DFC9000 GT camera and a Leica Plan Apochromat 100Â/1.4 oil objective.
Microscopy of live-parasite-infected erythrocytes was performed as previously described (93). Briefly, parasites were incubated in standard culture medium with 1 mg/mL Hoechst 33342 (Invitrogen) for 15 min at 37°C prior to imaging. A 5.4-mL portion of infected erythrocytes was added on a glass slide and covered with a coverslip. Nuclei were stained with 1 mg/mL Hoechst 33342 (Invitrogen). Microtubules were visualized by incubation of parasites in medium containing 1:1,000 TubulinTracker deep red (Thermo Fisher Scientific; dissolved in dimethyl sulfoxide [DMSO]), which labels polymerized tubulin, for 15 min at 37°C prior to imaging, as previously described (94). Images were processed using Fiji (91), and Adobe Photoshop CC 2021 was used for display purposes only.
P. falciparum blood stage growth assay. For growth assays of TGD cell lines, a flow cytometry assay, adapted from previously published assays (95,96), was performed to measure proliferation over 5 days. Each day, parasite cultures were resuspended, and 20-mL samples were transferred to an Eppendorf tube. Eighty microliters of RPMI containing Hoechst 33342 and dihydroethidium (DHE) was added to obtain final concentrations of 5 mg/mL and 4.5 mg/mL, respectively. Samples were incubated for 20 min (protected from UV light) at room temperature, and parasitemia was determined using an ACEA NovoCyte flow cytometer.
P. falciparum gametocyte quantification assay. Giemsa-stained blood smears at day 10 after induction of GDV1 expression were obtained, and at least 10 fields of view were recorded using a 63Â objective per treatment and time point. Erythrocyte numbers were then determined using the automated Parasitemia software (http://www.gburri.org/parasitemia/), while the number and morphology of gametocytes was determined manually in 1,096 to 2,632 (mean = 1793) (3D7-iGP-SPM3-GFP-glmS and 3D7-iGP-SPM3-GFP-glmSM9) or 925 to 2,742 (mean = 1,711) (3D7-iGP control and 3D7-iGP-SPM3-TGD) erythrocytes per sample. The counting was performed in a blind fashion.
P. falciparum GlmS-based gene knockdown. GlmS-based knockdown assay was adapted from previously published assays (53,97). To induce knockdown, highly synchronous early ring-stage parasites were split into two dishes, 2.5 mM GLCN was added to one of them, and parasite growth was measured by fluorescence-activated cell sorting (FACS) after two and four parasite replication cycles. Parasite cultures were inspected daily by Giemsa smears and, if necessary, diluted to avoid growth bias caused by high parasitemia. As an additional control, the same amount of glucosamine was also added to 3D7 wild-type parasites. For all analyses, medium was changed daily, and fresh glucosamine was added every day.
GlmS-based knockdown in gametocytes was performed as previously described (98). Briefly, synchronized ring-stage cultures were induced by the addition of Shield-1, as described above. At day 3 postinduction, the culture was spilt into two dishes, and one dish was cultured in the presence of 2.5 mM glucosamine for the remaining 10 days.
Phylogenetic analysis. For phylogenetic analysis, we identified sequences of potentially orthologous genes in other species by using a protein-to-protein BLAST search (blastp, default parameters) (99) with the PfSPM3 protein sequence as a query (GenBank accession no. XP_001350024.1). The blastp search was performed on the nonredundant (nr) database but excluding "Plasmodium falciparum" as a species. Selected representative sequences (PRCDC_1326300, PGABG01_1325400, PVX_116515, AK88_00350, PcyM_ 1231000, PKNH_1201300, PCOAH_00043100, PCHAS_1347100, C922_01771, PVVCY_1304510, PBANKA_ 1342500, PY17X_1347200, PocGH01_12035100, PmUG01_12037200, PRELSG_1200700, PGAL8A_00239500, and HEP_00010500) were aligned with the NCBI tool Cobalt multiple alignment, and a phylogenetic gene tree (fast minimum evolution method based on Grishin protein sequence distances) was built via the NCBI web tools. Mega X (100) was used for visualization of the gene tree.
Software Animal work and ethics statements. For in vivo experiments, 6-to 8-week-old female Swiss mice obtained from Janvier labs were used. These experiments included parasite propagation and mosquito infections. For transmission experiments used to determine parasite infectivity, 4-to 6-week-old female C57/BL6 mice from Charles River Laboratories were used. All animal experiments were performed under the authorization number G111/20 according to the FELASA-B guidelines and were approved by the appropriate authority, the Regierungspräsidium in Karlsruhe, Germany.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. MOVIE S1, AVI file, 1.