Phosphorylation of myosin A regulates Plasmodium sporozoite motility and is essential for efficient malaria transmission

Malaria-causing parasites rely on an actin-myosin based motor for the invasion of different host cells as well as tissue traversal in mosquitoes and vertebrates. The unusual myosin A of Plasmodium spp. has a unique N-terminal extension which is important for red blood cell invasion by P. falciparum merozoites in vitro and harbors a phosphorylation site at serine 19. Here, using the rodent-infecting P. berghei we show that serine 19 is essential for efficient transmission of Plasmodium by mosquitoes as S19A mutants show defects in mosquito salivary gland entry and migration of salivary gland sporozoites in both 2D and 3D environments. Our data suggests that entry into salivary glands represents the strongest barrier in parasite transmission and hence is the key determinant for evolution of the motility and invasion machinery of these parasites. Highlights The unusual N-terminal extension of Plasmodium myosin A is important for efficient gliding motility Altering the kinetics of the myosin A power stroke impacts Plasmodium life cycle progression and sporozoite motility Myosin A phosphorylation at serine 19 is important for malaria transmission by mosquitoes Salivary gland invasion emerges as key selection step for evolution of the parasite motor


34
Apicomplexan parasites rely on an actin-myosin-based motor for migration on and through tissues 35 and for cell invasion (Heintzelman, 2015;Frénal et al, 2017). Compared with canonical actin-36 myosin motors, however, apicomplexans have evolved an intriguingly divergent machinery reliant 37 on short and highly dynamic actin filaments and unique class XIV myosins, which lack the 38 extended cargo-binding tail domain (Herm-Götz et al, 2002;Bookwalter et al, 2017;Douglas et al, 39 2018; Robert-Paganin et al, 2019). As key parts of the motor, actin 1 and myosin A (MyoA) are 40 essential in Plasmodium spp., the causative agents of malaria in vertebrates (Bushell et al, 2017). Anopheles mosquitoes as disease vectors (Douglas et al, 2015). Indeed, Plasmodium parasites 46 likely evolved in insects before they spread to vertebrate hosts suggesting that the core machinery 47 of the parasite has initially adapted to a life in insects (Poinar, 2016). Within the mosquito, two 48 highly motile extracellular forms of the parasite also rely on the actin-myosin motor. In the 49 mosquito gut, ookinetes are formed and transmigrate across the midgut epithelium to transform 50 into oocysts. In those oocysts, sporozoites are formed (Angrisano et al, 2012b). Sporozoites need 51 motility to egress from oocysts, enter salivary glands, migrate in the skin, enter and exit the 52 bloodstream and to infect hepatocytes (Fig 1A). This formidable journey likely requires a higher 53 level of regulation of the actin-myosin motor compared to merozoites. Indeed, several actin 54 regulatory proteins are only found to be important in sporozoites and mutations in actin or actin 55 binding proteins could be identified that only impact sporozoite motility with little impact on 56 merozoites or ookinetes (Ganter et al, 2009;Bane et al, 2016;Moreau et al, 2017;Douglas et al, 57 2018). Furthermore, MyoA and other components of the gliding machinery are phosphorylated in 58 blood stages, ookinetes and sporozoites, suggesting that phosphorylation regulates motility 59 (Sebastian et al, 2012;Alam et al, 2015;Lasonder et al, 2015;Swearingen et al, 2017). 60 Many insights into the function of the apicomplexan actin-myosin motor and its associated proteins 61 have been gained from the study of Toxoplasma gondii, a highly successful parasite circulating 62 between cats and their prey but also infecting about one third of the world's human population 63 (Frénal et al, 2017). For practical reasons, most studies on T. gondii are limited to tissue culture 64 migration and invasion assays and it is important to note that the common ancestor of T. gondii 65 and Plasmodium spp. split about 350-820 Mio years ago (Sato, 2011). Yet, both parasites, as well 66 as most apicomplexans, feature similarly designed highly polarized invasive forms featuring a 67 namesake complex apical end where vesicles fuse with the plasma membrane to secrete proteins 68 or deposit them into the plasma membrane (Frénal et al, 2017). The plasma membrane is 69 subtended by a membrane organelle called the inner membrane complex (IMC), which is the 70 defining structure of the alveolates, the superphylum of organisms the apicomplexans belong to. 71 On the cytoplasmic face of the IMC, a complex and stable membrane-associated network, the 72 subpellicular network, is giving the parasites their shape (Harding & Frischknecht, 2020). In the 73 narrow space (30 nm) between plasma membrane and IMC, the actin-myosin motor is located 74 (Heintzelman, 2015;Frénal et al, 2017). Actin filaments are likely polymerized by formins located 75 at the apical end of the parasites (Baum et al, 2008;Douglas et al, 2018;Tosetti et al, 2019). 76 MyoA is anchored into the IMC by light chains and so-called gliding associated proteins (GAPs) 77 and can move actin filaments from the front to the rear (Heintzelman, 2015;Frénal et al, 2017). 78 As the actin filaments bind receptors spanning the plasma membrane these receptors are moved 79 rearwards.  1B). 84 Conditional deletion of myoA or actin in T. gondii showed a dramatic decrease in both tachyzoite 85 motility and cell invasion (Meissner et al, 2002;Andenmatten et al, 2013;Egarter et al, 2014).

86
Replacing the myoA promoter with a promoter that is only active in blood stages in the rodent-87 infecting malaria parasite P. berghei showed that ookinetes are still formed but cannot migrate in 88 the absence of MyoA (Siden-Kiamos et al, 2011). Conditional mutations in MyoA and a myosin 89 light chain in P. falciparum revealed an orchestrated need for merozoite force production to enter 90 red blood cells (Blake et al, 2020). Crystal structures of P. falciparum MyoA showed that 91 interaction sites formed by E6 and phosphorylated S19 in the N-terminal extension are important 92 for MyoA kinetics (Robert-Paganin et al, 2019;Moussaoui et al, 2020). Mutations of these amino 93 acids leading to the disruption of the respective interactions reduced the speed at which purified 94 MyoA transports actin filaments in vitro but increased maximal force production (Robert-Paganin 95 et al, 2019;Moussaoui et al, 2020). This suggests that phosphorylation of S19 is required for 96 maximum myosin velocity and might hence play a role in ookinetes or sporozoites, that rely on 97 motility for much longer periods than merozoites. In laboratory settings, the major human infecting 98 P. falciparum is usually studied in cultured blood cells, while for the study of mosquito transmission 99 the rodent malaria parasite P. berghei is used for reasons of experimental ease, ethical limitations 100 and safety (Matz & Kooij, 2015). In order to probe the role of the N-terminal extension of MyoA for 101 ookinete and sporozoite motility, we sought to generate parasite lines of P. berghei that harbour 102 different mutations in MyoA and analyse their effects in the mosquito stages. We show that the 103 endogenous 3' untranslated region (UTR) of myoA is important for expression in sporozoites 104 limiting gene modifications by classic homologous recombination. Consistent with data from P. 105 falciparum, we find that the N-terminal extension is essential for blood stages. However, point 106 mutations of E6 and S19 yielded viable parasite lines that can infect mosquitoes at normal levels.

107
Strikingly, parasites expressing S19A but not the phosphomimetic S19D showed reduced 108 mosquito salivary gland colonization and liver infection suggesting that phosphorylation of S19 is 109 essential for efficient transmission. 110 111

113
Exchange of 3'myoA by 3'dhfs leads to reduced myoA expression levels resulting 114 in a defect in salivary gland invasion 115 To understand how single amino acid residues contribute to the function of MyoA, we first 116 attempted to use conventional gene editing of the myoA locus which results in integration of the 117 3'UTR of dhfs and a selection cassette downstream of myoA (Supplementary Fig 1). This would 118 allow insertions of mutations similar to what was achieved in our study on the actin binding proteins 119 coronin and profilin (Bane et al, 2016;Moreau et al, 2017). Along with the generation of the control 120 line, we also tested whether the P. falciparum orthologue can complement the function of P. 121 berghei MyoA (PbMyoA) by replacing PbMyoA with P. falciparum myoA (PfMyoA). In addition, we 122 C-terminally tagged the protein with GFP for localization studies (Fig 2A and Supplementary Fig  123  1). To test if the clonal lines proceed through the life cycle, a prerequisite for the mutagenesis 124 study, we analyzed all life cycle stages. While the blood stage growth rate of the PbMyoA 3'dhfs 125 and PbMyoA-GFP 3'dhfs parasite lines were similar to wild type (wt), it was significantly reduced 126 from 9-fold per day in wt to 6-fold per day in the PfMyoA 3'dhfs line (Fig 2B). To assess the ability 127 of parasites to infect mosquitoes, we counted the numbers of oocysts that formed in the mosquito 128 midguts 10 to 14 days after the insects took a blood meal on infected mice. All parasite lines could 129 form oocysts, but the PbMyoA 3'dhfs and PfMyoA 3'dhfs lines showed significantly reduced 130 numbers of oocysts (Fig 2C). In line with the very low number of oocysts, the number of 131 sporozoites that formed in midgut oocysts was most reduced in the PfMyoA 3'dhfs line ( Table 1).

132
In contrast, PbMyoA 3'dhfs and PbMyoA-GFP 3'dhfs sporozoite numbers within oocysts were 133 similar to wt. However, sporozoite numbers from salivary glands were strongly reduced, 134 suggesting a defect in salivary gland invasion ( Table 1, Fig 2D). Surprisingly, the ratio of salivary 135 gland to midgut sporozoites was similar to wt in the PfMyoA 3'dhfs parasite line. Next, we 136 examined whether the defect in salivary gland invasion could be due to reduced motility of 137 sporozoites floating in the hemolymph, the circulating fluid of the mosquito. Indeed, a much lower 138 fraction of isolated hemolymph sporozoites displayed gliding motility on glass in these parasite 139 lines (Fig 2E). We hypothesized that the defect in in vitro motility and salivary gland invasion might 140 be due to a difference in myoA expression resulting from the replaced 3'UTR. Indeed, 141 quantification of myoA mRNA levels via qPCR revealed a 5 to 10-fold reduction in the PbMyoA 142 3'dhfs line in early oocysts, while there was increased myoA expression in schizonts (Fig 2F).

143
Hence a different strategy for mutagenesis was needed that kept the endogenous 3'UTR of myoA 144 intact.

146
Phosphorylation of MyoA at serine 19 is important for salivary gland invasion of 147 sporozoites and efficient malaria transmission 148 In order to leave the 3'UTR within the myoA locus intact, we next used a cloning strategy that 149 resulted in selection marker free parasite lines (Lin et al., 2011). In a first step, we generated a 150 clonal ama1 promoter swap parasite line (Supplementary Fig 2). In a second step, we replaced 151 the ama1 promoter by the endogenous myoA promoter and the myoA ORF carrying the mutation 152 of interest at the 5' end of myoA. As a control, we reconstituted the wt myoA locus. This resulted 153 in control parasites proceeding through the life cycle as anticipated (Table 1, Fig 3). To assess 154 the role of the unique N-terminus of MyoA on parasite motility, we focused on two amino acids, 155 E6 and S19, which are thought to be especially important as they interact with switch I/II and the 156 converter, respectively and are required to transport actin filaments at maximum speed (Robert-157 Paganin et al, 2019; Moussaoui et al, 2020) (Fig 3A). To this end, we introduced a reverse charge 158 at position 6 (mutant E6R) and replaced serine at position 19 with either a phosphorylation 159 incompatible alanine (mutant S19A) or a phosphomimetic aspartate (mutant S19D) (Fig 3B). 160 Previous data suggests that the S19A mutant would only have a minor impact on merozoites but 161 affect the fast gliding stages (Robert-Paganin et al, 2019;Blake et al, 2020). To study the overall 162 importance of the N-terminal extension, we also generated a construct aiming to delete the first 163 19 amino acids from the N-terminus of MyoA (Fig 3B). While we obtained parasites with the point 164 mutations after transfection, we failed to delete the N-terminal extension in four attempts, 165 suggesting that this part of the protein is essential. Interestingly, E6R and S19A mutants grew as 166 well as wild type while S19D parasites grew significantly slower in the blood (Fig 3C). In contrast, 167 mutagenesis had no effect on mosquito infection as assessed by the numbers of oocysts present 168 in mosquitoes fed on infected mice (Fig 3D). This indicates that the mutations had no major effect 169 on ookinete formation and migration. Oocysts produced normal numbers of sporozoites in all lines, 170 which were able to exit into the hemolymph. However, sporozoite numbers in the hemolymph were 171 slightly enhanced in the S19A parasite line, resulting in an increase in the ratio of hemolymph to 172 midgut sporozoites ( Table 1, Fig 3E). Far fewer sporozoites were found in salivary glands of this 173 S19A parasite line than in the other lines (Table 1) as evidenced by a significantly lower ratio of 174 salivary gland to midgut sporozoites (Fig 3F). This suggests a defect of S19A sporozoites to enter 175 into salivary glands. 176 Next, we tested whether the mutations within the N-terminal extension of MyoA have an effect on 177 parasite transmission back to mice. To this end, we allowed infected mosquitoes to feed on mice.

178
We then determined the appearance and numbers of infected red blood cells starting three days 179 after infection to evaluate the success of parasite transmission. The S19A parasite line only 180 infected 60% of mice (5 out of 8) via natural transmission by mosquito bite, while all other parasite 181 lines infected 100% of mice ( Table 2). Those mice that developed a blood stage infection with the 182 S19A mutant showed a one-day delay in infection as compared to the wt control ( Table 2, Fig  183  3G). A one-day delay without decreased numbers of infected mice is generally considered a 90% 184 reduction of infectivity (Vanderberg, 1975). These findings suggest that phosphorylation at S19 is 185 required for efficient salivary gland invasion and efficient malaria transmission. Such a defect in 186 natural transmission can be due to reduced numbers of sporozoites in the salivary glands, which 187 leads to fewer transmitted sporozoites (Aleshnick et al, 2020), a defect in motility in the skin or a 188 defect in liver cell invasion (Frischknecht & Matuschewski, 2017). To test for a role in transmission 189 and transmigration of the skin versus liver cell invasion, we intravenously injected 1000 salivary 190 gland derived sporozoites. In this experiment all mice, including those infected with S19A 191 sporozoites, became infected. However, there was still a delay in S19A infection ( Table 2, Fig  192  3H). While not excluding that S19A mutants have a defect in skin migration, this indicates that 193 S19A parasites have a reduced capacity to invade the liver. 194

Mutated MyoA affects sporozoite motility
195 Plasmodium sporozoites can migrate at more than 1 µm/s, an order of magnitude faster than 196 neutrophils (Vanderberg, 1974). Some even reach peak instantaneous velocities of over 5 µm/s 197 (Münter et al, 2009). To investigate whether the defect of S19A parasites in salivary gland infection 198 can be explained by reduced motility, we probed hemolymph and salivary gland sporozoite motility 199 of all generated parasite lines on glass. This showed that around 20% of wt hemolymph 200 sporozoites could migrate in the typical circular path of sporozoites or in a back-and-forth manner 201 (termed patch gliding) while a reduced fraction of just 3% of S19A hemolymph sporozoites were 202 motile (Fig 4A). The S19D mutant also showed a slight reduction of motile hemolymph 203 sporozoites. We determined the speed of the migrating sporozoites isolated from the hemolymph 204 and found that both S19A and S19D hemolymph sporozoites migrated significantly slower as 205 compared to the reconstituted wt sporozoites. We next isolated sporozoites from salivary glands 206 and imaged their migration behavior. Here, the fraction of motile E6R and S19A salivary gland 207 sporozoites was slightly lower but not significantly reduced compared to wt. However, the speed 208 of all mutant parasites was significantly reduced as compared to wt sporozoites (Fig 4B). As some 209 sporozoites expressing mutated actin-binding proteins alter their migration paths (Montagna et al, 210 2012; Bane et al, 2016), we next compared the trajectories of the migrating sporozoites from the 211 different parasite lines. This showed no apparent difference between the mutant hemolymph and 212 salivary gland sporozoites and the wt control parasites (Fig 4C, D). 213 Changing MyoA kinetics affects sporozoite force production 214 Parasite motility results from a complex interplay between adhesion and force generation. Yet, 215 force generation of sporozoites has so far only been investigated in parasites lacking either 216 surface receptors or actin-binding proteins without a direct function in force generation ( by-side. Nevertheless, while about 70% of wt or wt-like sporozoites were able to pull the bead out 231 of the trap, only 18% of the E6R mutant and 25% of the S19D mutant were capable of pulling the 232 bead out of the trap (Fig 5A). We were not able to perform these experiments with the 233 phosphodeficient mutant due to the low number of S19A sporozoites in the mosquito salivary 234 gland. 235 S19A mutation diminishes 3D migration of sporozoites 236 We next investigated 3D motility of the mutants in polyacrylamide-based hydrogels as a model to 237 study their migration through the skin (Ripp et al, 2021). To this end, dissected salivary glands 238 were sandwiched between two hydrogels and the sporozoites released from the gland entered 239 into the gel on corkscrew-like paths. This allowed us to also image the S19A salivary gland derived 240 sporozoites. Previous experiments revealed that mutant sporozoites with severe defects in 2D 241 motility due to changes in their substrate adhesion capacity showed much improved migration 242 capacity in 3D (Ripp et al, 2021). In contrast, the speed of all sporozoites expressing myosin 243 mutants was significantly reduced compared to the wt-like control line (Fig 5B). The trajectories 244 from sporozoites tracked for three minutes of observation were shorter, especially for the S19A 245 mutant highlighting the lower displacement of these sporozoites (Fig 5C). This, along with the in 246 vivo transmission data, is suggestive of a migration defect in the skin. 247 Taken together, these data suggest that ablating stabilizing interactions in the rigor state of MyoA 248 results in a defect in force generation and motility at the sporozoite stage. While the E6R and 249 S19D mutations only have a minor impact, the S19A mutation fails to efficiently enter into salivary 250 glands. Entry into salivary glands might therefore constitute the strongest barrier in parasite 251 transmission and hence represent the main obstacle against which the parasites needed to 252 primarily evolve their migration and invasion machinery. of MyoA. Subtle effects as have been seen in P. falciparum invasion assays were not detected 274 during the blood stage growth of our lines suggesting that they play comparatively minor roles in 275 vivo in the rodent model. Similarly, we observed a measurable reduction in force generation by 276 two mutants in P. berghei sporozoites. Yet this defect did not translate into a significant loss in 277 mosquito-to-mouse transmission efficiency of the parasite lines. Only the S19A mutant line, which 278 cannot be phosphorylated at S19 and hence alters the kinetics of the MyoA power stroke (Robert-279 Paganin et al, 2019) showed a dramatic reduction in the ability to transmit from mosquito to 280 mammal (Fig 6). This block was mainly due to the incapacity of the parasites to colonize the 281 salivary glands of the mosquito, although the parasite also showed defects in gliding in a skin-like 282 gel and in liver infection. These results indicate that sporozoites depend on very fast myosin 283 dynamics for efficient transmission. efficiency followed by a reduction in salivary gland invasion and transmission to the mammal, while 300 ookinete motility and the infection of mosquitoes is not yet affected. Strikingly, the efficiency of 301 mosquito colonization as measured by the presence of oocysts was not affected by specifically 302 tuning the force generation capacity of myosin. Instead, a similar hierarchy appears with myosin 303 mutants lacking some force generation capacity still being able to transmit efficiently, while the 304 S19A mutant showed the largest defects in sporozoite migration and salivary gland colonization.

305
This strongly suggests that the actin-myosin motor machinery was shaped and tuned to highest 306 efficiency by the need of the sporozoites for their long and elaborate journey from the oocysts in 307 the mosquito midgut to the hepatocyte in a mammalian liver. Strikingly, of all the barriers the entry 308 into the salivary gland appears as the most formidable. 309 On a molecular engineering note: we first sought to use our established mutagenesis approach to 310 alter myoA but found that the necessary change of the endogenous 3'UTR for the 3'UTR of dhfs 311 led to decreased expression of myoA in early oocysts, which impacted the capacity of the 312 sporozoites to enter salivary glands. A similar approach for generation of MyoA-GFP readily 313 yielded salivary gland invasion (Green et al, 2017). However, the produced line was not clonal 314 and the presence of wt myoA in the oocysts might have produced enough MyoA to compensate 315 for the loss due to the unnatural 3'UTR. We previously used a gene-in-marker-out approach to 316 mutate actin, which was also based on the generation of an initial recipient line where the 3'UTR 317 of actin 1 was replaced by the 3'UTR of dhfs to facilitate gene replacement (Douglas et al, 2018). 318 While this parasite line acted as a recipient for further modifications in the blood stages and was 319 thus not required for mosquito infection, a strongly reduced salivary gland entry by this modified 320 line was observed as well, suggesting that expression of not only myoA but also actin 1 is affected 321 by a modified 3'UTR (Ross Douglas, unpublished). To avoid similar problems with other proteins 322 it might be advisable to use a gene-in-marker-out approach that leaves the 3'UTRs unaffected or 323 a CRISPR/Cas9 system for subtle mutagenesis (Shinzawa et al, 2020). 324 Finally, we noted that it was possible to replace the P. berghei myoA gene for that of myoA from 325 P. falciparum. However, the resulting parasites grew much slower in the blood and had deficits in 326 infecting the mosquito. Multiple explanations are possible for these observations. Firstly, the myoA 327 gene from P. falciparum was codon modified and hence the gene might not have been transcribed 328 in sufficient amounts, as we noted recently for the gene encoding alpha-tubulin 1 (Spreng et al, 329 2019). Alternatively, the subtle differences in the amino acid sequence of the two myosins might 330 be enough to cause the phenotype. Due to the additional problems associated with the 3'UTR we 331 chose not to follow-up on these differences. 332 In conclusion, the presented data suggests that sporozoites need S19 phosphorylation to  and MyoA S19D, myoA was amplified with primers that were designed to introduce the mutations 362 and cloned into vector Pb262.

363
The transfection vectors were linearized via restriction digest with KpnI or BamHI and SacII before 364 transfection. Plasmodium berghei ANKA parasites were used for genetic modifications. 365 Transfection and generation of isogenic parasite lines was carried out essentially as described 366 before (Janse et al, 2006;Klug & Frischknecht, 2017). DNA was prepared using ethanol 367 precipitation and electroporation of purified schizonts was carried out using Nucleofactor 368 technology (Lonza). Transfection mixtures were then injected intravenously into a mouse. 369 Transfected parasites were positively selected with 0.07 mg/mL pyrimethamine or negatively 370 selected with 1.5 mg/ml 5-FC added into the drinking water of the mice. Clonal parasite lines were 371 obtained by limiting dilution. Successful mutagenesis was verified via genotyping PCR with the 372 primers listed in Supplementary

Mosquito infection and sporozoite isolation 376
Anopheles stephensi mosquitoes were infected and sporozoites isolated from midguts, 377 hemolymph and salivary glands as described previously (Klug & Frischknecht, 2017). Briefly, for 378 mosquito infections, mice were infected by intraperitoneal injection of frozen stocks (150-200 ml). 379 After 3-5 days, the infected mice were checked for presence of gametocytes by placing a drop of 380 tail blood on a microscope slide followed by incubation at room temperature for 10-12 min. During 381 this time, flagellated gametes mature and exit in a process termed exflagellation. If 1-2 382 exflagellation events per field were observed at 40x magnification, mice were anesthetized and 383 fed to mosquitoes. Experiments with hemolymph and salivary gland sporozoites were performed 384 13 to 16 and 17 to 25 days post mosquito infection, respectively. Oocysts were counted from 385 dissected midguts following mercurochrome staining: Midguts were permeabilized for 20 min with 386 1% Nonidet P40 in PBS, stained with 0.1% mercurochrome in PBS for 30 min, mounted in a small 387 amount of PBS, covered with a coverslip, and counted using a 10× objective on a Zeiss 388 CellObserver microscope. Midguts and salivary glands were dissected with a pair of needles in 389 PBS and placed on ice until further use. Sporozoites were released by gently crushing the 390 collected organs with a plastic pestle. Sporozoites were isolated from the hemolymph by cutting 391 the last segment of the abdomen with a syringe and flushing with RPMI (supplemented with 50,000 392 units/L penicillin and 50 mg/L streptomycin) by inserting a long-drawn Pasteur pipette into the 393 lateral side of the thorax. The hemolymph was drained from the abdomen, collected on a plastic 394 foil and transferred to a reaction tube. Sporozoites were counted using a Neubauer chamber and 395 total numbers extrapolated according to the dilution. 396

397
Cell migration assays on glass or in polyacrylamide hydrogels were performed and analysed as 398 described before (Douglas et al, 2018;Ripp et al, 2021). Briefly, isolated sporozoites in RPMI 399 medium supplemented with 50,000 units/l penicillin, 50 mg/l streptomycin and 3% bovine serum 400 albumin (BSA) were pipetted into a 96-well optical bottom plate (Nunc) and centrifuged at 1,000 401 rpm for 3 min. For 3D hydrogel assays, whole infected salivary glands dissected into 30 µl of 402 medium were sandwiched between a glass coverslip (22 × 22 mm) placed on top of a microscope 403 slide and a 3% AA/0.03% BIS hydrogel. Imaging was performed at room temperature on an 404 inverted Zeiss CellObserver microscope. Images were recorded every second for hemolymph 405 sporozoites or every three seconds for salivary gland sporozoites for a total time of three minutes. The 18S rRNA gene was used as a reference, and across-run differences were normalized using 432 a calibrator sample. Relative copy numbers were calculated by applying the ΔΔCt methodology. 433 The sequences of the gene-specific primers used are shown in Supplementary Table 3