Evolutionarily distant I domains can functionally replace the essential ligand-binding domain of Plasmodium TRAP

The importance of the I domain for TRAP function was tested using the trapΔI parasite line (Figure 2A, Figure 2—figure supplement 1). In several independent experiments in which mosquitoes fed on infected mice, only few trapΔI or trap(-) sporozoites could be observed within the mosquito salivary glands, whereas for trap wild type ~10,000 sporozoites were observed per mosquito (Figure 2B, Table 1). Thus, trapΔI sporozoites are similarly impaired in salivary gland invasion as trap(-) sporozoites. To determine whether mutant parasites retained the ability to migrate steadily on microscope slides, that is to glide in circles (Vanderberg, 1974), sporozoites were isolated from hemolymph and activated by addition of 3% bovine serum albumin (BSA). Hemolymph sporozoites, like midgut sporozoites, were present in all mutants studied here (Table 1). Sporozoites were defined as gliding and productively motile if they were able to complete at least one circle within 3 or 5 min, depending on the experiment. Sporozoites exhibiting other types of motion were classified as unproductively motile, while sporozoites that were attached but were not moving or were not attached were classified as non-motile (Figure 2—figure supplement 2; Münter et al., 2009). Hemolymph sporozoites were ~19% productively motile in wild type while none were motile in trapΔI (Figure 2C) and trap(-) mutants (Münter et al., 2009; Hegge et al., 2010; Figure 2C). These results show that the I domain is required for productive motility.

Figure 2 with 3 supplements see all

Download asset Open asset

Infectivity of mutant sporozoites was tested by exposing naive mice to infected mosquitoes or by intravenously injecting sporozoites obtained from the midgut, hemolymph or even salivary glands (Table 1). Upon infection with wild type sporozoites, the first blood stage parasites were visible as expected after 3 to 8 days; in contrast, no infections could be observed for trapΔI and trap(-) parasites (Figure 2—figure supplement 3, Table 2, Table 3). Immunofluorescence assays on midgut sporozoites using an antibody recognizing the repeat region of TRAP showed specific fluorescence in most sporozoites at one end with no recognizable difference between trapΔI and wild type sporozoites. This suggests that the mutated TRAP is also localized in the secretory micronemes. In contrast, TRAP-specific fluorescence was absent in trap(-) sporozoites. TRAP could also be observed on the surface of unpermeabilized trapΔI sporozoites indicating that micronemal secretion is not abolished in these parasites (Figure 2D).

We next tested whether the lack of productive motility and infectivity as well as the severely impaired salivary gland invasion rate in trapΔI parasites could be rescued by replacing the deleted TRAP I domain with an I domain from a foreign species. We selected structurally characterized I domains of MIC2 from Toxoplasma gondii and the I domains of the human integrin α-subunits αX (CD11c) and αL (CD11a) (Figure 1C–E). Sequence identity among I domains is 36% between αL and αX, 18% between these integrins and P. berghei TRAP, and 28% between TRAP and MIC2 (Supplementary file 2). The I domains from TRAP and αX are basic, with pI values of 9.7 and 8.9, respectively, while those of MIC2 and αL are acidic, with pI values of 6.1 and 5.8, respectively (Song et al., 2012; Song and Springer, 2014). Furthermore, the αX I domain is poly-specific as shown by binding to multiple glycoproteins and proteolytic fragments as well as heparin (Vorup-Jensen et al., 2005; Vorup-Jensen et al., 2007), while the αL I domain is highly specific for the ligand intercellular adhesion molecule (ICAM-1) and its homologues ICAM-2, ICAM-3, and ICAM-5 (Grakoui et al., 1999).

The fluo line, with eGFP constitutively expressed in all parasite stages and mCherry specifically expressed in sporozoites, was used to generate some of the respective transgenic parasite lines to simplify analysis of TRAP I domain replacements throughout the parasite life cycle (Bane et al., 2016). Parasite lines expressing either codon modified wild type TRAP (TRAP-I) or codon modified TRAP with foreign I domains (MIC2-I, αX-I or αL-I) replacing the P. berghei TRAP I domain (Supplementary file 2) were generated with both fluo and non-fluo trap(-) parasite lines by homologous recombination (Figure 3A and Figure 3—figure supplement 1). TRAP shows the same localization in MIC2-I, αX-I and αL-I midgut sporozoites as in TRAP-I midgut sporozoites (Figure 3B). Western blots showed similar expression levels of TRAP in hemolymph sporozoites of TRAP-I, trapΔI, αX-I and αL-I while no TRAP could be detected for trap(-) (Figure 3C). Similar localization and expression of TRAP was also observed for MIC2-I and TRAP-I salivary gland sporozoites (Figure 3D,E). MIC2-I fluo and non-fluo lines entered salivary glands at similar rates as wild type sporozoites (Figure 3F,G,H, Table 1). αX-I and αL-I sporozoites were capable of invading the salivary glands, albeit at very low rates (Table 1 and Figure 3F and G, note the log scale of the y axis). The numbers for αX-I ranged from ~100 (fluo) to ~200 (non-fluo) and for αL-I from ~30 (non-fluo) to ~90 (fluo) sporozoites per gland (Figure 3F,G and Table 1) compared to 0–200 for trap(-) sporozoites and 0–800 for trapΔI. These small numbers might also be due to contaminants from the hemolymph and hence need to be treated with caution. Yet in all (8/8) αX-I mosquito infections sporozoites were detected in salivary glands, whereas this was only the case in 50% (4/8) of αL-I infections. In contrast for mosquitoes infected with trap(-) or trapΔI sporozoites were only detected in 13% (1/8) and 21% (3/14) experiments, respectively. (Table 1). In line with this observation, αX-I but not αL-I sporozoites could be detected in isolated salivary glands by live fluorescence microscopy (Figure 3H and Video 1) that indeed sporozoites expressing αX-I can enter this organ more efficiently than those expressing αL-I.

Figure 3 with 4 supplements see all

Download asset Open asset

Video 1

Download asset

We next analyzed motility of the parasite lines in vitro using the classification scheme shown in Figure 2—figure supplement 2. Gliding assays of hemolymph sporozoites revealed that ~24% of TRAP-I but only ~4% of MIC2-I sporozoites were productively motile (Figure 4A). As expected, a higher proportion of salivary gland sporozoites were productively motile;~53% for TRAP-I and ~15% for MIC2-I sporozoites (Figure 4B). Among hemolymph sporozoites,~1% of αX-I (Video 2) but none of the >3000 observed αL-I sporozoites showed productive movement (Figure 4A). Motile TRAP-I and MIC2-I salivary gland sporozoites moved with a similar speed of ~1.5 µm/s (Figure 4C), showed similar trajectories (Figure 4D), and showed similarly persistent gliding (Figure 4E). Owing to the low numbers of αX-I and αL-I sporozoites in the glands (Figure 3), similar quantitation of sporozoite motility was not possible.

Video 2

Download asset

Figure 4

Download asset Open asset

To probe the infectivity of mutant sporozoites, mice were exposed to infected mosquitoes. This revealed infection rates of 100% with a similar prepatent period (time until an infection could be detected in the blood) for TRAP-I and MIC2-I sporozoites (Table 2, Table 3). In contrast, no transmission could be observed for αX-I and αL-I sporozoites (Figure 3—figure supplement 2, Table 2). Intravenous injection of 10,000 TRAP-I or MIC2-I salivary gland sporozoites also infected all mice with a prepatency of three days. Unfortunately, the numbers of αX-I and αL-I salivary gland sporozoites were too low for comparative tests (Table 1). We therefore injected 10,000 hemolymph sporozoites, which again resulted in similar infection rates as seen for TRAP-I and MIC2-I sporozoites with a slightly longer prepatency when compared to salivary gland sporozoites (Figure 3—figure supplement 3, Table 2). Interestingly, 50% (4/8) of mice injected with 10,000 αX-I hemolymph sporozoites became blood stage patent after a delayed prepatency of >5 days, while no infection was observed for the same number of αL-I hemolymph sporozoites (0/8) (Figure 3—figure supplement 3, Table 2). When injecting 25,000 hemolymph sporozoites, 5 of 8 mice injected with αX-I sporozoites became blood stage patent and additionally, 1 of 6 mice injected with αL-I sporozoites became infected (Figure 3—figure supplement 3, Table 2, Table 3). Finally, we injected mice with 500,000 midgut sporozoites to compare αX-I and αL-I side by side with trap(-) and trapΔI. Injections of αX-I midgut sporozoites infected 2 out of 4 mice with a prepatent period of 8 days. In contrast no infections in mice could be observed when αL-I (0/2), trap(-) (0/4) or trapΔI (0/6) midgut sporozoites were injected (Table 2). Mice infected with the same number of wild type or wild type-like midgut sporozoites become blood stage patent after 6–8 days (Table 2; Klug and Frischknecht, 2017). To exclude spurious results from contamination with other parasite lines, some of the transmitted parasites were propagated in mice and analyzed via PCR for the correct genotype (Figure 3—figure supplement 4). Additionally, the TRAP locus of these parasites was sequenced to ensure that the correct I domain was present. In all tested cases, the expected αX-I and αL-I genotype was confirmed. Thus, TRAP-I and MIC2-I sporozoites are comparably infectious to mice, αX-I sporozoites are moderately infectious and αL-I, while nearly completely deficient in infectivity, could in one case still cause an infection.

While Plasmodium sporozoites can infect different types of cells they have a strong tropism for the liver. In contrast, Toxoplasma gondii can infect any nucleated cell from a warm-blooded animal (Boothroyd, 2009). Hence, we tested whether an exchange of the I domain affected host cell invasion or tissue tropism of sporozoites. First, we tested the capacity of TRAP-I and MIC2-I salivary gland sporozoites to invade HepG2 cells. After exposure for 1.5 hr, 2-fold fewer MIC2-I sporozoites than TRAP-I sporozoites were intracellular (Figure 5A). At 48 hr after infection, 5-fold fewer MIC2-I sporozoites had developed into liver stage parasites when compared to TRAP-I sporozoites (Figure 5B). However, the size of liver stage parasites after 48 hr was comparable implying that an exchange of the I domain affects parasite invasion but not intracellular development (Figure 5C,D).

Figure 5 with 1 supplement see all

Download asset Open asset

To test in vivo tropism of sporozoites, mice were infected by intravenous injection of 20,000 TRAP-I and MIC2-I salivary gland sporozoites and after 42 hr liver, spleen, lung and a part of the small intestine were harvested. Parasite load was measured by quantitative RT-PCR. Liver tropism of both TRAP-I and MIC2-I salivary gland sporozoites was pronounced, with ~16 fold more parasites localizing to the liver than to any other organ (Figure 5E, Figure 5—figure supplement 1). However, the liver burden of MIC2-I parasites was reduced by ~40% relative to TRAP-I parasites, reflecting the similar decrease observed during in vitro infection experiments.

Based on the results with different I domains, we sought a common pattern explaining the observed phenotypes. One parameter important for protein-protein interactions is surface charge. Interestingly, the surface of the TRAP I domain is very basic, with a pI of 9.7 (Figure 6A), while the surface of the second best functioning I domain, MIC2, is acidic (pI 6.1) (Figure 6B). Of the human integrin I domains, the one from αX has a basic charge (pI 8.9), similar to the TRAP I domain (Figure 6C), while the one of αL is even slightly more negatively charged (pI 5.8) (Figure 6D) than the I domain of MIC2. To test if surface charge could be important for P. berghei TRAP I-domain function, we rendered it anionic (pI of 6.8) with seven charge reversal mutations (H56E, H62E, H123E, K164Q, K165D, R195E and K202E) around the perimeter of the putative ligand binding site, distal from the MIDAS (Figure 6E, Figure 6—figure supplement 1). Secretion of TRAP to the surface was not altered in these RevCharge sporozoites and no difference in protein sub-cellular distribution or expression was observed (Figure 6F,G). Strikingly, salivary gland invasion was also not affected (Figure 6H, Table 1). However, only ~1% of RevCharge salivary gland derived sporozoites were productively motile in gliding assays compared to ~53% of TRAP-I parasites (Figure 6I). Infection by mosquito bite revealed that only 2 out of 8 mice in two independent experiments became blood stage positive with a delay in prepatency of >2 days (Figure 6J, Table 2, Table 3) indicating a decreased infectivity of salivary gland sporozoites by over 99%. In contrast, injection of 10,000 salivary gland sporozoites in three independent experiments (nine mice in total) led to an infection of all mice with the prepatency period of the RevCharge mutant being delayed by 0.5–1 day compared to controls (wild type/TRAP-I) (Figure 6K, Table 2). This corresponds to a decreased infectivity of 50–90%.

Figure 6 with 2 supplements see all

Download asset Open asset

In vitro invasion of liver cells also showed a severe impairment of the RevCharge mutant compared to TRAP-I. While after 2 hr 54% of TRAP-I sporozoites showed an intracellular localization only 3% of RevCharge sporozoites were observed inside cells (Figure 6—figure supplement 2A). In line with this result very few growing liver stages of the RevCharge mutant were observed while on average >250 liver stages per well were counted for TRAP-I (Figure 6—figure supplement 2B). However, liver stages of the RevCharge mutant developed normally (Figure 6—figure supplement 2C). These results suggest that the basic charge around the MIDAS of the TRAP I domain is a key determinant of sporozoite motility and infectivity during natural transmission from mosquito to mammal. Most intriguingly, the introduced mutations appear to uncouple the capacity of the P. berghei sporozoite to infect insect salivary glands from efficiently infecting the rodent host.

Previous studies on the function of the TRAP I domain focused on invariant residues that coordinate the MIDAS metal ion in the I domain (Wengelnik et al., 1999; Matuschewski et al., 2002). The sidechains of these five invariant I domain residues coordinate the MIDAS Mg²⁺ ion either directly or indirectly through water molecules. In contrast to the closed conformation, in the open conformation of TRAP and integrin αI domains, neither of the two Asp residues directly coordinate the MIDAS metal ion. The metal ion is therefore thought to have high propensity in the open conformation to bind an acidic residue in the ligand. The MIDAS residues and their bound water molecules occupy five of the six coordination positions around the MIDAS Mg²⁺ ion. The remaining, sixth coordination position is occupied when a critical Asp or Glu sidechain in the ligand binds through a carboxyl oxygen to the MIDAS Mg²⁺ ion (Liddington, 2014). Mutation of single MIDAS residues or removal of the Mg²⁺ ion by chelation abolishes ligand binding by integrins (Michishita et al., 1993; Kern et al., 1994; Kamata et al., 1995). Similarly, mutation of the MIDAS motif of TRAP severely impairs salivary gland invasion and infectivity of the vertebrate host while gliding motility is decreased in salivary gland but not in hemolymph sporozoites (Matuschewski et al., 2002). In contrast, deletion of TRAP completely abrogates salivary gland invasion, infectivity and productive motility (Sultan et al., 1997; Münter et al., 2009) and severely affects substrate adhesion (Münter et al., 2009; Hegge et al., 2010).

To elucidate the role of the I domain in TRAP function, we generated the parasite line trapΔI expressing TRAP without its I domain. TRAPΔI was expressed and correctly localized in sporozoites as shown by western blotting and immunofluorescence. Interestingly, this line was severely deficient in salivary gland invasion and could not infect mice. Furthermore, we observed no productive movement and only some back-and-forth motility in trapΔI hemolymph sporozoites, similar to trap(-) hemolymph sporozoites. These results suggest that the biological functions of TRAP are dependent on its I domain. It was therefore remarkable that TRAP function was largely restored by replacement of its I domain by its homologue from Toxoplasma gondi with only 28% amino acid sequence identity. Compared to cell surface receptors that engage in typical protein-protein interactions, that is those that are not dependent on Mg²⁺ ions, these sequence identities are substantially below the level of 40% to 50% identity generally required for members of the same receptor family to recognize the same ligand. The strength of the Mg²⁺ ion bond to the oxygen in the ligand, which has a bond distance of only 2 Å and is partially covalent, may overcome the lack of complementarity in other regions of the ligand binding site, and explain the ability of I domains of such remarkably low sequence identity to function in TRAP. In addition to the low sequence identity, there are six to seven sequence positions where residues are inserted or deleted in Toxoplasma MIC2 or human integrin αI domains compared to TRAP (Figure 1F). Nonetheless, all regions involved in conformational change around the MIDAS and in the β6-α7 loop are highly conserved in conformation in both the open and closed conformations of the apicomplexan and human I domains (Song et al., 2012; Song and Springer, 2014). Western blotting showed that the αL-I, αX-I, and MIC2-I TRAP fusions were as well expressed as TRAP-I; furthermore, immunofluorescence showed normal localization in sporozoites, suggesting that trafficking was not impaired.

Although deletion of the TRAP I domain completely abolished mosquito salivary gland invasion and mouse infection by midgut and hemolymph sporozoites, the MIC2 I domain completely restored these functions. Moreover, no differences in mouse infection were observed between TRAP and MIC2 I domain replacements in infection by mosquito bite or intravenous injection with salivary gland sporozoites, including the length of the prepatent period. Fewer MIC2-I than TRAP-I sporozoites were able to glide productively in vitro; however, the gliding speed of motile sporozoites and persistence of gliding was similar. Infectivity of HepG2 liver cells was significantly decreased for MIC2-I compared to TRAP-I sporozoites; yet, quantitative RT-PCR showed that there was little difference in ability to infect liver cells in vivo between intravenously injected MIC2-I and TRAP-I sporozoites. Moreover, there was no change in in vivo tropism.

We also examined replacement with two mammalian integrin I domains, from the promiscuous αXβ2 integrin and the much more selective αLβ2 integrin. Salivary gland invasion was decreased 100 to 400 and 500 to 1,000-fold in αX-I and αL-I compared to TRAP-I sporozoites, respectively. Infectivity of hemolymph αX-I sporozoites in mice was 100-fold deacreased compared to controls while αL-I hemolymph sporozoites showed nearly no infectivity as trap(-) sporozoites. The greater efficacy of αX-I than αL-I sporozoites in invasion of mosquito salivary glands, infection of mice, and productive motility in vivo confirmed our hypothesis that the polyreactive αX I domain would better support TRAP function than the highly specific αL I domain, albeit at low levels.

Multiple factors may account for the substantially lesser efficacy of mammalian integrin I domains than the MIC2 I domain. MIC2 has a similar function to TRAP in Toxoplasma and may be its orthologue. Each has an I domain that is inserted in an extensible β-ribbon in tandem with a TSR domain and a long-range disulfide bond that moves in allostery (Figure 1). Integrins, TRAP, and MIC2 connect to the actin cytoskeleton, which applies force to their cytoplasmic domains that is resisted by ligands and provides traction for motility and cell invasion. The force (F) transmitted through the I domain, times the difference in extension of the I domain in the closed and open states (Δx), gives an energy (F•Δx = E) that tilts the energy landscape toward the open, high affinity state (Li and Springer, 2017). Because of their extensible β-ribbons, Δx is substantially greater for TRAP and MIC2 than for integrin I domains, and cytoskeletal forces may also differ considerably. To compensate for these differences, the energy differences between the closed and open states, as well as the kinetics for crossing between them, may be tuned differently for TRAP and MIC2 than for integrin I domains. Their degree of polyspecificity is also likely to vary. Integrins αXβ2 and αMβ2 have each been reported to bind >30 ligands, including glycoproteins and heparan (Yakubenko et al., 2002). However, these integrin I domains can also bind specifically, as shown by binding to distinct high-affinity sites within their common ligand, iC3b (Xu et al., 2017).

The most polyreactive integrin I domains, αX and αM are basic with higher pI (8.9 and 9.5, respectively) than the highly specific αL (pI 5.6), αE (pI 4.6), and collagen-binding α1, α2, α10, and α11 I domains (pI 5.1–5.6). As the surfaces of all cells are negatively charged and almost all proteins are acidic, we tested whether the functions of the basic TRAP I domain (pI = 9.7) would be altered with seven substitutions that made it neutral (pI 6.8). This RevCharge mutant was well expressed and localized. RevCharge sporozoites accumulated well in mosquito salivary glands and were infectious when injected intravenously. However, gliding in vitro and infectivity by mosquito bite were greatly decreased. Although the pI of RevCharge of 6.8 was close to that of MIC2, its mutations were clustered around the periphery of the MIDAS site and the electrostatic surface of its ligand-binding face more closely resembled that of αL than MIC2, TRAP, or αX (Figure 6). These results suggest that the distribution of electrostatic charge on the TRAP I domain is an important variable. The difference between infection by mosquito bite and intravenous injection further suggests that electrostatics may have an important role in successful sporozoite exit from the dermis into the circulation.

The findings that the I domain from Toxoplasma MIC2 that encounters different ligands as TRAP can replace its function in mediating salivary gland invasion and infection of mice strongly suggest that the I domain functions to provide traction for motility and cell invasion rather than tissue tropism. Binding of the TRAP I domain to multiple distinctive ligands is required by its function in mosquitos in migration into salivary glands and in vertebrate hosts in emigration from the dermis into the circulation and from the circulation into the liver, followed by productive invasion of liver parenchymal cells. The ligands in the mosquito must differ from those in the vertebrate; those encountered in the vertebrate in the dermis, endothelium and liver may also differ. Furthermore, Toxoplasma lives in epithelia in the gastrointestinal system of cats and other mammals, and lacks an arthropod host, yet its MIC2 I domain functions robustly in mosquito salivary gland invasion. Our results are consistent with TRAP binding to the many ligands that have been identified for it, including liver-specific fetuin-A (Jethwaney et al., 2005), proteoglycans (Robson et al., 1995; Pradel et al., 2002), glycosaminoglycans (Matuschewski et al., 2002), and the integrin αV subunit (Dundas et al., 2018) in vertebrates and saglin in mosquitos (Ghosh et al., 2009).

Abolition of TRAP function by deletion of its I domain as shown here, together with the negligible effects of mutation or exchange of other domains (Matuschewski et al., 2002; Ejigiri et al., 2012) strongly suggests that the I domain is the sole ligand binding domain. A model in which TRAP does not provide liver tropism is also consistent with the expression of TRAP in sporozoites of Plasmodium species that infect birds and initially infect skin rather than liver cells (Böhme et al., 2018).

Our results support a model in which TRAP does not determine tropism for salivary glands in mosquitos or liver in mammals and instead acts as a poly-specific receptor that provides traction for sporozoite emigration into these tissues and infection of cells that is triggered by other receptors. Similarly, integrins, from which the TRAP I domain may have been borrowed, are activated by cytoskeletal activity, which is generally signaled by G protein-coupled receptors or receptors that couple to tyrosine kinases. This model has the advantage that the signaling receptors can be extremely sensitive and ATP-dependent cytoskeleton polymerization or actomyosin contraction through activation of adhesins can greatly amplify those signals and provide ultrasensitive regulation of adhesiveness (Li and Springer, 2017). This might be essential during the stick-and-slip gliding of sporozoites (Münter et al., 2009).

Chemoattractants have the advantage that they can drive directional migration through multiple layers of distinctive cell types as in the liver. In vertebrates, chemoattractants and their receptors cooperate with integrins and their ligands to enable highly specific leukocyte emigration from the vasculature into organs or sites of inflammation (Springer, 1994). Chemoattractant receptors evolved distinctly in prokaryotes and eukaryotes and perhaps take yet a different form in apicomplexans. The ability of chemoattractants to activate motility, termed chemokinesis, may be related to the use here of BSA, a known carrier of fatty acids and other hydrophobes, to activate sporozoite gliding. Our finding that the I domain of TRAP is required for gliding motility and invasion by sporozoites of mosquito salivary glands and mammalian liver, without a strong requirement for evolved specificity, suggests that new paradigms may be needed to understand sporozoite activation and homing of sporozoites to specific organs in their hosts.

Plasmodium sequences were retrieved from PlasmoDB (http://plasmodb.org/plasmo/, version 26–34) (Aurrecoechea et al., 2009) and multiple sequence alignments were performed with Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/) (Sievers et al., 2011). To change the codon usage of open reading frames (ORFs) we applied the tool OPTIMIZER (http://genomes.urv.es/OPTIMIZER/) (Puigbò et al., 2007).

TRAP knockout (trap(-)) parasites were generated with the PlasmoGem (Schwach et al., 2015) vector (PbGEM-107890) using standard protocols (Janse et al., 2006). Isogenic trap(-) parasites were subsequently negatively selected with 5-fluorocytosine (1.0 mg/mL in the drinking water) to give rise to selection marker free trap(-) parasites (Braks et al., 2006). For the generation of trapΔI parasites we made use of the Pb238 vector (Deligianni et al., 2011; Klug and Frischknecht, 2017). In a first step the trap 3’UTR (970 bp) was amplified with the primers P165/P166 and cloned (BamHI and EcoRV) downstream of the resistance cassette in the Pb238 vector. In a next step the coding sequence of the trap gene including the 5’ and 3’ UTR was amplified with the primers P508/P509 and cloned in the pGEM-T-Easy vector giving rise to the plasmid pGEM-TRAPfull. Subsequently the pGEM-TRAPfull plasmid was mutated with the primers P535/P536 and P537/P538 to introduce a restriction site for NdeI directly in front of the start codon ATG and a restriction site for PacI directly after the stop codon TAA. The mutated sequence was cloned (SacII and EcoRV) in the Pb238 intermediate vector that contained already the trap 3’UTR downstream of the selection marker and the resulting plasmid was named Pb238-TRAP-NdeI/PacI. The designed DNA sequence lacking the coding region of the I domain was codon modified for E. coli K12 and synthesized at GeneArt (Invitrogen). Subsequently the designed sequence was cloned (NdeI and PacI) in the Pb238-TRAP-NdeI/PacI by replacing the endogenous trap gene. Final DNA sequences were digested for linearization (NotI, PbGEM-107890; SacII and KpnI, Pb238-TRAP∆I), purified and transfected into wild type parasites (wt) using standard protocols (Janse et al., 2006; Figure 2—figure supplement 1).

To generate parasite lines expressing TRAP with different I domains, bp 115 to bp 696 (581 bp; I42 to V228; 194 aa in total) of the wild type trap gene (Plasmodium berghei ANKA strain) were exchanged with sequences from the micronemal protein 2 (MIC2) of Toxoplasma gondii (567 bp, L75 to V263, 189 aa), the integrin CD11c/αX (552 bp, Q150 to I333, 184 aa) and the integrin CD11a/αL (531 bp, V155 to I331, 177 aa) of Homo sapiens. Chimeric sequences as well as the coding sequence of the wild type trap gene that served as a control, were codon modified for E. coli K12 to prevent incorrect homologous recombination with the trap coding sequence downstream of the I domain coding region and to avoid changes of the codon usage within the open reading frame caused by the exchanged I domain coding sequence. This enabled also simple differentiation between wild type and transgenic parasites by PCR. For the generation of RevCharge parasites seven mutations of non-conserved amino acids (H56E, H62E, H123E, K164Q, K165D, R195E and K202E; P. berghei ANKA strain) were introduced into the codon modified wild type trap gene. These mutations were expected to shift the surface charge at the apical side of the I domain from a pI of 9.7 to 6.8 while leaving the MIDAS intact and the structural integrity of the domain unaffected. All designed sequences were synthesized at GeneArt (Invitrogen) and cloned in the Pb238-TRAP-NdeI/PacI vector (NdeI/PacI) that was already used to generate the trapΔI line. Constructs were digested for linearization (ScaI-HF) and transfected using standard protocols (Janse et al., 2006). Transfections were performed in the negatively selected TRAP knockout line trap(-) as well as in the fluorescent background line fluo to independently generate fluorescent (fluo) and non-fluorescent (non-fluo) sets of mutants (Figure 3—figure supplement 1).

Isogenic parasite lines were generated by serial dilution of parental populations obtained from transfections. Per transfection one mouse was infected by intraperitoneal injection of ~200 µL frozen parasites of the parental population. To increase the number of transfected parasites within infected mice pyrimethamine (0.07 mg/mL) was given within the drinking water 24 hr post injection (hpi). Donor mice were bled two to three days post injection once parasitemia reached 0.5–1%. Parasites were diluted in phosphate buffered saline (PBS) to 0.7–0.8 parasites per 100 µL and the same volume was subsequently injected into 6–10 naive mice. Parasites were allowed to grow for 8–10 days until parasitemia reached 1.5–2%. Blood of infected mice was taken by cardiac puncture (usually 600–800 µL) and used to make parasite stocks (~200 µL infected blood) and to isolate genomic DNA with the Blood and Tissue Kit (Qiagen).

Naive mice were infected with 100–200 µL frozen parasite stocks and parasites allowed to grow for four to five days. Infected mice were either directly fed to mosquitoes or used for a fresh blood transfer of 2 × 10⁷ parasites by intraperitoneal injection into two to three naive mice. Parasites within recipient mice were allowed to grow for further three to four days, depending on the number of exflagellation events observed. To determine the number of exflagellation events, and subsequently the number of male gametocytes, a drop of tail blood was placed on a microscope slide, covered with a coverslip and incubated for 10–12 min at room temperature (20–22°C). The number of exflagellation events was counted with a light microscope (Zeiss) and a counting grid by using 40-fold magnification with phase contrast. If at least one exflagellation event per field could be observed mice were fed to mosquitoes. Mosquitoes were starved overnight prior to the feeding to increase the number of biting mosquitoes. Per mosquito cage (approximately 200–300 female mosquitoes) two to three mice were used for feeding.

To estimate the number of sporozoites, infected Anopheles stephensi mosquitoes were dissected on day 13, 14, 18, and 22 post infection. For the collection of hemolymph sporozoites, infected A. stephensi mosquitoes were cut with a needle to remove the last segment of the abdomen. Subsequently the thorax was pierced with a finely drawn Pasteur pipette filled with RPMI/PS solution. While gently pressing the pipette the haemocoel cavity was flushed with solution which dripped off the abdomen. The sporozoite solution was collected on a piece of foil and transferred into a plastic reaction tube. To determine the number of midgut sporozoites, the abdomen of infected mosquitoes was dissected with two needles and the midgut was extracted. Isolated midguts were transferred into a plastic reaction tube containing 50 µL RPMI/PS solution. For the isolation of salivary gland sporozoites, the head of infected mosquitoes was gently pulled away with a needle while fixing the mosquito in place with a second needle. Ideally the salivary glands stayed attached to the head and could be easily isolated. Salivary glands were transferred into a plastic reaction tube with 50 µL RPMI/PS. To release sporozoites, pooled midguts and salivary glands were homogenized using a plastic pestle for 2 min. To count the parasites in each sample, 5–10 µL of the sporozoite solution (1:10 dilution) was applied on a hemocytometer. Sporozoites were counted using a light microscope (Zeiss) with 40-fold magnification and phase contrast. For each counting experiment at least 10 mosquitoes were dissected. However, the number was adapted depending on the infection rate of the mosquitoes and the experiment that was performed.

To determine the prepatency of parasite lines during sporozoite transmission two different routes of infection were tested. Female C57BL/6 mice were either exposed to infected mosquitoes or infected by intravenous injection of midgut, hemolymph or salivary gland sporozoites. To infect mice by mosquito bites, mosquitoes infected with fluorescent parasite lines were pre-selected for fluorescence of the abdomen by using a stereomicroscope (SMZ1000, Nikon) with an attached fluorescence unit. Subsequently parasite positive mosquitoes were separated in cups to 10 each and allowed to recover overnight. Approximately six hours prior to the experiment mosquitoes were starved by removing salt and sugar pads. Mice were anesthetized by intraperitoneal injection of a mixture of ketamine and xylazine (87.5 mg/kg ketamine and 12.5 mg/kg xylazine) and placed with the ventral side on the mosquito cups. Mosquitoes infected 17–24 days prior to the experiment were allowed to feed for approximately 15 min before mice were removed. During this time eyes of mice were treated with Bepanthen cream (Bayer) to prevent dehydration of the cornea. After the experiment mice were allowed to recover and tested for blood stage parasites on a daily basis by evaluation of Giemsa stained blood smears. If non-fluorescent parasite lines were tested infected mosquitoes were not pre-sorted. In these experiments midguts of mosquitoes that had taken a blood meal were dissected after the experiment and the number of midgut sporozoites was counted as described previously. Mice that were bitten by mosquitoes that contained no midgut sporozoites were excluded from the analysis.

For injections hemolymph or salivary gland sporozoites were isolated either 13–16 days (hemolymph) or 17–24 days (salivary gland) post infection. Sporozoite solutions were diluted to the desired concentration with RPMI/PS (either 10,000 or 25,000 sporozoites) and injected intravenously into the tail vein of naive mice. The presence of blood stage parasites was evaluated on a daily basis.

To analyze speed and movement pattern of sporozoites, in vitro gliding assays were performed in glass-bottom 96-well plates (Nunc). Hemolymph and salivary gland sporozoites were obtained by dissecting infected A. stephensi mosquitoes. To free the sporozoites from salivary glands, samples were grounded with a pestle. Subsequently salivary gland samples were centrifuged for 3 min at 1,000 rpm (Thermo Fisher Scientific, Biofuge primo) to separate sporozoites from tissue. Afterwards ~40 µL of the supernatant was transferred into a new 1.5 mL plastic reaction tube and diluted with a variable volume of RPMI/PS depending on the planned number of assays and the sporozoite concentration resulting in a minimum of 50,000 sporozoites per well. For each assay about 50 µL of the sporozoite suspension was mixed with 50 µL RPMI medium containing 6% bovine serum albumin (BSA) to initiate activation. Subsequently sporozoites were allowed to attach to the bottom by centrifugation at 800 rpm for 3 min (Heraeus Multifuge S1). Using fluorescence microscopy (Axiovert 200M) with a 10x objective movies were recorded with one image every three seconds for 3 to 5 min depending on the experiment. Movies were analyzed manually using the Manual Tracking Plugin from ImageJ (Schindelin et al., 2012) to determine speed and trajectories of moving sporozoites. Sporozoites that were able to glide at least one full circle during a 3 min movie were considered to be productively moving while all other sporozoites were classified as non-productively moving (moving less than one circle) or non-motile.

For the imaging of sporozoites within salivary glands infected mosquitoes were dissected 17–24 days post infection as described previously. Isolated salivary glands were transferred with a needle to a microscope slide containing a drop of Grace’s medium (Gibco) and carefully sealed with a cover slip. Samples were imaged with an Axiovert 200M (Zeiss) using 63x (N.A. 1.3) and 10x (N.A. 0.25) objectives.

For immunofluorescence assays we made use of antibodies directed against the circumsporozoite protein (CSP) and the thrombospondin related anonymous protein (TRAP). In all assays the anti-CSP antibody mAb 3D11 (Yoshida et al., 1980) was applied as unpurified culture supernatant of the corresponding hybridoma cell line (1:5 diluted for immunofluorescence assays). TRAP antibodies were generated against the peptide AEPAEPAEPAEPAEPAEP by Eurogenetec and the purified antibody was applied as 1:100 dilution in immunofluorescence assays. Antibodies against the same peptide have been shown previously to specifically detect TRAP by immunofluorescence and western blotting (Ejigiri et al., 2012). Secondary antibodies coupled to AlexaFluor 488 or Cy5 (goat anti-mouse or goat anti-rabbit) directed against primary antibodies were obtained from Invitrogen and always used as 1:500 dilution.

To visualize the expression and localization of TRAP in sporozoites, infected salivary glands were dissected as described previously and pooled in plastic reaction tubes containing 50 µL PBS or RPMI/PS. Afterwards salivary glands were mechanically grounded with a plastic pestle to release sporozoites from tissue. Immunofluorescence assays were performed by two different methods either fixing sporozoites in solution or on glass cover slips. To fix the parasites on glass, salivary glands were dissected in RPMI/PS and treated as described. Sporozoite solutions were transferred into 24-well plates containing round cover slips, activated with an equal volume RPMI/PS containing 6% BSA and allowed to glide for approximately 30 min at RT. Subsequently the supernatant was discarded and sporozoites were fixed with 4% PFA (in PBS) for 1 hr at RT. Fixed samples were washed three times with PBS for 5 min each. If immunofluorescence was performed on sporozoites in solution, salivary glands were dissected in PBS and treated as described previously. Sporozoite solutions were directly fixed by adding 1 mL of 4% PFA (in PBS) for 1 hr at RT. After fixation samples were washed as described for samples fixed on glass while samples in solution had to be additionally pelleted after each step by centrifugation for 3 min at 10,000 rpm (Thermo Fisher Scientific, Biofuge primo). Subsequently sporozoites were blocked (PBS containing 2% BSA) or blocked and permeabilized (PBS containing 2% BSA and 0.5% Triton X-100) over night at 4°C or for 1 hr at RT, respectively. Samples were incubated with primary antibody solutions for 1 hr at RT in the dark and subsequently washed three times with PBS. After the last washing step, samples were treated with secondary antibody solutions and incubated for 1 hr at RT in the dark. Stained samples were washed three times in PBS and the supernatant was discarded. If the immunofluorescence assay was performed in solution, sporozoite pellets were resuspended in 50 μL of remaining PBS, carefully pipetted on microscopy slides and allowed to settle for 10–15 min at RT. Remaining liquid was removed with a soft tissue and samples were covered with cover slips which had been prepared with 7 μL of mounting medium (ThermoFisher Scientific, ProLong Gold Antifade Reagent). If the immunofluorescence assay was performed on sporozoites that were fixed on glass, cover slips were removed with forceps, carefully dabbed on a soft tissue and placed on microscopy slides that had been prepared with 7 μL of mounting medium. Samples were allowed to set overnight at RT and kept at 4°C or directly examined. Images were acquired with a spinning disc confocal microscope (Nikon Ti series) with 60-fold magnification (CFI Apo TIRF 60x H; NA 1.49).

Salivary glands of infected mosquitoes were dissected in 100 µL PBS on ice and subsequently smashed with a pestle to release sporozoites. For the isolation of hemolymph sporozoites infected mosquitoes were flushed with PBS as previously described. Sporozoite solutions were kept on ice, counted using a haemocytometer and distributed to 30,000 sporozoites per reaction tube. Samples were centrifuged for 1 min with 13,000 rpm (Thermo Fisher Scientific, Biofuge primo) at 4°C to pellet sporozoites. Subsequently the supernatant was discarded, and pellets were lysed in 20 µL RIPA buffer (50 mM Tris pH 8, 1% NP40, 0.5% sodium dexoycholate, 0.1% SDS, 150 mM NaCl, 2 mM EDTA) supplemented with protease inhibitors (Sigma-Aldrich, P8340) for ≥1 hr on ice. Lysates were mixed with Laemmli buffer, heated for 10 min at 95°C and centrifuged for 1 min at 13,000 rpm (Thermo Fisher Scientific, Biofuge primo). Samples were separated on precast 4–15% SDS-PAGE gels (Mini Protein TGX Gels, Bio-Rad) and blotted on nitrocellulose membranes with the Trans-Blot Turbo Transfer System (Bio-Rad). Blocking was performed by incubation in PBS containing 0.05% Tween20% and 5% milk powder for 1 hr at RT. Afterwards, the solution was refreshed and antibodies directed against TRAP (rabbit polyclonal antibody, 1:1000 diluted) or the loading control CSP (mAb 3D11, cell culture supernatant 1:1000 diluted) were added. Membranes were washed three times (PBS with 0.05% Tween20) and secondary anti-rabbit antibodies (Immun-Star (GAR)-HRP, Bio-Rad) or anti-mouse antibodies (NXA931, GE Healthcare) conjugated to horse radish peroxidase were applied for 1 hr (1:10,000 dilution) at room temperature. Signals were detected using SuperSignal West Pico Chemiluminescent Substrate or SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific). After the detection of TRAP, blots were treated with stripping buffer (Glycine 15 g/L, SDS 1 g/L, Tween20 10 ml/L, pH 2,2) for 15 min prior to incubation with anti-CSP antibodies used as loading control.

HepG2 cells were a gift from our virology department who had obtained the cells from ATCC. Cell line identity was regularly confirmed by SNP sequencing and visual observations of cell morphology. Cells were routinely tested for mycoplasma contamination using the MycoAlert mycoplasma detection kit (Lonza, Basel, Switzerland). We culture cells for a maximum of 10 passages.

Two days prior to the experiment 50,000 HepG2 cells/well were seeded in an 8-well Permanox Lab-Tek chamber slide (Nunc). On day zero salivary glands were isolated from infected female mosquitoes and collected in RPMI medium within a 1.5 mL reaction tube. Sporozoites were released by mechanically disrupting the salivary glands with a polypropylene pestle. The solution was centrifuged in a tabletop centrifuge at 1,000 rpm for 3 min at RT and the supernatant was transferred to a new 1.5 mL reaction tube. The salivary gland pellet was resuspended in 100 µL RPMI medium, smashed again using a pestle, centrifuged (1,000 rpm, 3 min, RT) and the supernatant was pooled with the first one. A 1:10 dilution of the sporozoite solution was counted in a Neubauer counting chamber and 10,000 salivary gland sporozoites were used to infect HepG2 cells per well. After 1.5 hr wells were washed twice with complete DMEM medium and HepG2 cells were allowed to grow in complete DMEM medium supplemented with 1x Antibiotic-Antimycotic (Thermo Fisher Scientific). At 24 hr and 48 hr post infection cells were fixed using ice cold methanol for 10 min at RT, followed by blocking with 10% FBS/PBS overnight at 4°C. Staining with primary antibody α-PbHSP70 1:300 in 10% FBS/PBS for 2 hr at 37°C was succeeded by two washing steps with 1% FBS/PBS. Incubation with secondary antibody α-mouse Alexa Fluor 488 1:300 in 10% FBS/PBS was performed for 1 hr at 37°C. Hoechst 33342 was added and incubated for 5 min at RT followed by two washing steps with 1% FBS/PBS. The assay was mounted in 50% glycerol and sealed using a glass cover slip. Samples were imaged with an Axiovert 200M (Zeiss) microscope and subsequently analyzed using ImageJ (Schindelin et al., 2012). In brief, the perimeter of single liver stages was encircled, and the area measured using the internal measurement tool.

Two days prior to the experiment 180,000 HepG2 cells/well were seeded in an 8-well Permanox Lab-Tek chamber slide (Nunc). Sporozoites were isolated as described above and 10,000 sporozoites/well were used to infect HepG2. At 1.5 hr post infection cells were washed twice with complete DMEM medium and fixed using 4% PFA/PBS 20 min at RT. Blocking was performed with 10% FBS/PBS o/n at 4°C followed by incubation with primary antibodies α-PbCSP 1:100 in 10% FBS/PBS (2 hr 37°C), two washing steps with 1% FBS/PBS and incubation with secondary antibodies α-mouse Alexa Fluor 488 1:300 in 10% FBS/PBS (1 hr 37°C). After two washing steps with 1% FBS/PBS, cells were permeabilized by addition of ice cold methanol and incubation at RT for 10 min. Blocking with 10% FBS/PBS (4°C and overnight) was followed by an incubation with primary antibodies α-PbCSP 1:100 in 10% FBS/PBS (2 hr at 37°C), two washing steps with 1% FBS/PBS and incubation with primary antibodies α-mouse Alexa Fluor 546 1:300 in 10% FBS/PBS (1 hr at 37°C). The assay was mounted in 50% glycerol after two washing steps with 1% FBS/PBS.

To measure the parasite load of different organs C57BL/6 mice were infected by intravenous injection of 20,000 salivary gland sporozoites. At 42 hr after infection organs were harvested, homogenized and the RNA was isolated using Trizol according to the manufacturer's protocol. Isolated RNA was treated with DNase using the Turbo DNA-free Kit (Invitrogen). Subsequently RNA content of generated samples was measured using a NanoDrop and liver, intestine, spleen and lung samples were pooled in equal amounts of RNA to generate single samples for each parasite line and harvested organ. RNA pools were used to synthetize cDNA using the First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). Quantitative RT-PCR was performed in triplicates on an ABI7500 (Applied Biosystems) using a 2x SYBR green Mastermix (Applied Biosystems). Plasmodium berghei 18S rRNA was used to quantify parasites and mouse specific GAPDH was utilized as housekeeping gene for normalization. Subsequently the ∆cT was plotted as mean of all replicates per parasite strain and harvested organ (Schmittgen and Livak, 2008).

For all experiments female 4–6 week-old Naval Medical Research Institute (NMRI) mice or C57BL/6 mice obtained from Janvier laboratories were used. Transgenic parasites were generated in the Plasmodium berghei ANKA background (Vincke and Bafort, 1968) either directly in wild type or from wild type derived strains (e.g. trap(-) and fluo). Parasites were cultivated in NMRI mice while transmission experiments with sporozoites were performed in C57Bl/6 mice only.

Statistical analysis was performed using GraphPad Prism 5.0 (GraphPad, San Diego, CA, USA). Data sets were either tested with a one-way ANOVA or a Student’s t test. A value of p<0.05 was considered significant.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We thank Christian Sommerauer and Miriam Reinig for rearing Anopheles stephensi mosquitoes as well as Christine Hopp for helpful discussions at the BioMalPar conference. This work was funded by the Human Frontier Science Program (RGY0071/2011), the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – project number 240245660 – SFB 1129, NIH Grant CA 31798, and the European Research Council (ERC StG 281719). DK is currently funded by a DFG postdoctoral fellow ship (KL 3251/1–1) and is an alumnus of the Heidelberg Biosciences International Graduate School (HBIGS). FF is member of CellNetworks Cluster of excellence at Heidelberg University. We acknowledge the microscopy support from the Infectious Diseases Imaging Platform (IDIP) at the Center for Integrative Infectious Disease Research.

Animal experimentation: All animal experiments were performed according to the German Animal Welfare Act (Tierschutzgesetz) and executed following the guidelines of the Society of Laboratory Animal Science (GV-SOLAS) and of the Federation of European Laboratory Animal Science Associations (FELASA). All experiments were approved by the responsible German authorities (Regierungspräsidium Karlsruhe, G-134/14 and G266/16). All interventions were performed under ketamine/xylazine anesthesia, and every effort was made to minimize suffering.

© 2020, Klug et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.