TRSP is dispensable for the Plasmodium pre-erythrocytic phase

Plasmodium sporozoites deposited in the skin following a mosquito bite must migrate and invade blood vessels to complete their development in the liver. Once in the bloodstream, sporozoites arrest in the liver sinusoids, but the molecular determinants that mediate this specific homing are not yet genetically defined. Here we investigate the involvement of the thrombospondin-related sporozoite protein (TRSP) in this process using knockout Plasmodium berghei parasites and in vivo bioluminescence imaging in mice. Resorting to a homing assay, trsp knockout sporozoites were found to arrest in the liver similar to control parasites. Moreover, we found no defects in the establishment of infection in mice following inoculation of trsp knockout sporozoites via intravenous and cutaneous injection or mosquito bite. Accordingly, mutant sporozoites were also able to successfully invade hepatocytes in vitro. Altogether, these results suggest TRSP may have a redundant role in the completion of the pre-erythrocytic phase of the malaria parasite. Nonetheless, identifying molecules with paramount roles in this phase could aid in the search for new antigens needed for the design of a protective vaccine against malaria.

Malaria is a vector-borne infectious disease caused by apicomplexan parasites belonging to the genus Plasmodium. It remains one of the most concerning public health problems with 216 million cases reported in 2016 leading to nearly half a million deaths, most of which were children under the age of 5 in Sub-Saharan Africa 1 .
A considerable effort has been made to generate a vaccine against malaria. In fact, several projects are currently underway 2 and, most notably, the circumsporozoite protein (CSP)-targeting RTS,S/AS01 has completed Phase 3 clinical trials 3 . Nonetheless, the protection conferred by RTS,S/AS01 against malaria in young children is only partial and, therefore, more effective options are required.
The pre-erythrocytic phase of malaria is initiated when an infected female anopheline mosquito injects sporozoites in the skin of a mammalian host while taking a blood meal. Only a few sporozoites eventually leave the skin and travel through the bloodstream to reach the liver. Inside hepatocytes they differentiate and multiply into merozoites, the red blood cell-infective forms. These are subsequently released into the bloodstream to initiate the symptomatic erythrocytic phase 4 .
Prior to hepatocyte invasion, sporozoites passively transported in the blood must arrest in the liver sinusoids, a crucial step for the ensuing liver infection. This arrest has been proposed to be enabled by both slower blood circulation speed in the sinusoids and the interaction of sporozoite surface molecules such as CSP and the thrombospondin-related anonymous protein (TRAP) with the highly-sulphated heparan sulphate proteoglycans (HSPGs) of liver cells [5][6][7][8] . It is noteworthy that the current model explaining this critical step is based on the interpretation of experiments where the binding of recombinant proteins to hepatocytes was tested or the outcome of a liver infection by sporozoites was assessed. Using in vivo bioluminescence imaging, we have recently demonstrated that Plasmodium sporozoites home to the liver in the first minutes following their intravascular inoculation in mice 9 . Thus, the identification of the parasite determinants involved in this stage-specific behaviour could be possible by generating transgenic parasites lacking candidate surface molecules and following their capacity to home to the liver by direct imaging in vivo.
The thrombospondin-related sporozoite protein (TRSP), a protein found on the parasite surface and previously implicated in the pre-erythrocytic phase, is among those candidates 10,11 . TRSP possesses an adhesive thrombospondin type 1 repeat (TSR) domain that is conserved throughout Plasmodium species and shows great similarity to those of TRAP and CSP 10,12 . TSRs are present in a diverse group of proteins, promote cell-cell and cell-matrix interactions in a wide array of organisms, and can bind to proteoglycans [13][14][15] . It has been proposed that TRAP and CSP TSRs bind specifically to HSPGs in the liver 5,7,16,17 . In the absence of TRSP, Plasmodium berghei sporozoites have been shown to present compromised hepatocyte entry in vitro, which correlated with lower parasite burden in the liver 10 . However, the role of this protein in mediating the arrest of sporozoites in the liver, a step that precedes hepatocyte infection, has never been addressed.
In this work, we have engineered luciferase-expressing P. berghei sporozoites lacking TRSP and evaluated their capacity to home to the liver using bioluminescence imaging. Moreover, we found no defects in the infectivity of trsp knockout sporozoites suggesting this molecule has a redundant role in the pre-erythrocytic phase of malaria.

Results
trsp knockout sporozoites efficiently target the liver. In order to investigate the potential role of TRSP in the homing of salivary gland sporozoites to the liver, we resorted to a homing assay using live imaging 9 . Briefly, making use of 2D bioluminescence imaging in mice it is possible to quantify the specific accumulation of luciferase-expressing sporozoites in the liver in the first minutes following their intravascular inoculation 9 . Thus, we generated a trsp knockout line on a bioluminescent background via targeted gene deletion by double crossover homologous recombination ( Supplementary Fig. 1A). Transfer populations were cloned to select isogenic lines hereby designated TRSP -. The two clonal parasite lines used throughout this work, TRSP -I21 and I26, were confirmed to have integrated the selectable marker to replace the trsp locus via PCR ( Supplementary Fig. 1B) and Southern blot ( Supplementary Fig. 1C), and shown to lack trsp mRNA by RT-PCR ( Supplementary Fig. 1D).
In the homing assay, we detected no difference in the bioluminescent signal in the liver 7 minutes after mice being inoculated with control versus TRSPsporozoites either in percentage (Fig. 1A,E) or absolute values (Fig. 1B,F). This indicates that sporozoites lacking TRSP efficiently home to the liver. In addition to the short 7-minute time point, bioluminescence was also quantified 2 to 4 hours post-infection and also no significant differences in the bioluminescent signal either in percentage (Fig. 1C,G) or absolute values (Fig. 1D,H) among the different groups were observed. Altogether, these observations demonstrate that trsp knockout parasites are able to successfully reach the liver.
TRSP is dispensable for the completion of the P. berghei pre-erythrocytic phase. To evaluate the capacity of intravenously-injected sporozoites to infect hepatocytes, bioluminescence in the liver was quantified at days 1 and 2 post-infection ( Fig. 2A,C) and the percentage of infected red blood cells was calculated on the following days (Fig. 2B,D). No differences in bioluminescent signal in the liver or in the percentage of infected red blood cells were observed between mutant and control parasites. Since intravenous inoculation bypasses the critical skin stage of the sporozoite journey to the mammalian liver, mice were also infected intradermally ( Fig. 2E-H). Similarly, TRSPparasites did not show any defects in the liver (Fig. 2E,G) and blood (Fig. 2F,H) infection. In conclusion, loss of TRSP does not impact the ability of sporozoites to reach and infect the liver, regardless of the infection route.
Although intravenous and intradermal infections are helpful experiments to scrutinize the behaviour of the parasites in specific stages of the pre-erythrocytic phase, these circumvent the deposition of sporozoites and mosquito saliva into the skin through the mosquito proboscis. To study the importance of TRSP in the context of natural transmission, infected mosquitos were allowed to probe on naive mice and the bioluminescent signal in the liver was measured two days after infection (Fig. 3A,C). Data obtained on day 1 were not considered because infections following mosquito bite typically lead to low parasite burdens in the liver when compared to intravenous ( Fig Overall, these data demonstrate that TRSPare as capable as control sporozoites of accomplishing the pre-erythrocytic phase of their life cycle. Furthermore, mutant parasite growth during the blood stage was indistinguishable from that of control parasites (Figs 2B,D,F,H and 3B,D). In addition, comparable numbers of salivary gland sporozoites were observed ( Supplementary Fig. 2), revealing that no problems in transmission to or infection of the mosquito occurred. This is further evidence that TRSP is not necessary for the P. berghei life cycle in standard conditions. TRSPsporozoites exhibit no impairment in hepatocyte entry. A particular phenotype relating to defective sporozoite entry into hepatocytes was previously described in vitro for trsp knockout parasites 10 . We inspected the ability of the TRSPclonal lines generated in this study to invade the HepG2 hepatoma cell line using an automated counting method to examine anti-CSP double staining, which distinguishes between intracellular and extracellular sporozoites 18 . Images were acquired using an IN Cell Analyzer 2000 and the numbers of host cell nuclei, total and intracellular sporozoites were counted either manually or using the IN Cell Developer Toolbox v.1.9.2 software ( Fig. 4A-E). The two counting methods yielded indistinguishable results ( Fig. 4B-E), validating the use of the automated method to determine the percentage of infected cells (Fig. 4F). No statistically significant differences were observed between TRSPand control sporozoites, indicating TRSPsporozoites have no major defects in host cell entry. These results also differ from the former study on the P. berghei trsp knockouts, where the authors described a failure in sporozoite entry into HepG2 cells that culminated in an incomplete entry phenotype 10 . We failed to observe this phenotype with our TRSPlines ( Fig. 4G-I). In fact, the partial entry of sporozoites was an extremely rare event in both control and mutant parasites (Fig. 4G). At most, only approximately 0.1% of host cells exhibited such events using any of the parasite lines (Fig. 4H) and these events did not constitute more than 2.1% of total intracellular parasites (Fig. 4I) in the selected representative experiments.

Discussion
This is the first study where the importance of a sporozoite surface protein in the homing of Plasmodium parasites to the liver was investigated using live imaging techniques. We generated TRSPbioluminescent clonal lines and made use of a homing assay to dissect the role of this protein in the specific retention of sporozoites in the liver. Strikingly, we found that TRSP not only is not required for the homing of sporozoites to the liver, but is also dispensable for the establishment of a successful infection in the mammalian host. These results challenge the findings of a previous study that advocated an involvement of TRSP in hepatocyte entry and P. berghei infectivity 10 . The paucity of the in vitro partial entry events could, to an extent, explain the disagreeing results. Nonetheless, the infrequency of incomplete hepatocyte entry events in our experiments suggests this phenomenon is most likely not biologically relevant. Moreover, different methods were used in both studies to evaluate the liver infectivity of the TRSP-deficient sporozoites, as in the previous study parasite burdens in the liver were assessed using RT-PCR 10 . We cannot exclude that the measurement of parasite burden using bioluminescence imaging is less sensitive than RT-PCR. However, no differences in the emergence of the blood infection following inoculation of TRSPor control sporozoites were found in our study, supporting a similar capacity for both parasites to infect the liver. Once again, this contrasts with the one-day delay in the pre-patent period reported in the former work 10 . Noteworthy, in this study we used the P. berghei ANKA strain rather than the NK65 strain 10 . Despite the high similarity between the genome sequences of several P. berghei laboratory lines 19 we cannot preclude that such phenotypic discrepancies may come from the fact that the trsp knockouts were engineered in distinct strains.
The processes in which TRSP may participate remain therefore undisclosed. The trsp gene in P. berghei codifies for a relatively small 159 amino-acid protein with a signal peptide, a transmembrane domain and a short cytoplasmic tail, in addition to the adhesive TSR 10 . TRSP has been demonstrated to have apical localization in Plasmodium yoelli 12 and has also been readily detected on the surface of purified and untreated P. falciparum and P. yoelii salivary gland sporozoites 11,20 . The TRSP exact subcellular localization and the mechanism that regulates its trafficking to the sporozoite surface remain to be elucidated. trsp transcripts are sporozoite-specific and are enriched in salivary gland sporozoites in comparison to their midgut-associated counterparts 12,21 . Hence, TRSP would be predicted to play a part in the initial stages of the Plasmodium life cycle in the mammalian host, in processes such as motility, adhesion or invasion 22,23 . On the other hand, the similarities shared between TRSP and other sporozoite surface proteins, namely those belonging to the TRAP family and other TSR-containing proteins, lead us to speculate that functional redundancy among these molecules may exist. This phenomenon has already been reported in similar contexts for other proteins 24,25 . It is possible that other adhesins compensate for the loss of TRSP and allow for the parasite to complete its life cycle without any major impairment.
Finally, the fact that TRSP appears to be dispensable for the parasite to successfully infect the mammalian host dampens the hope that this surface molecule could become an attractive target for neutralizing antibodies 11 . Nonetheless, the identification of sporozoite surface proteins with functions that are vital to the completion of the  sporozoite journey in the mammalian host remains a promising approach to design a novel vaccine. Stimulating a humoral response against these antigens would not only directly target the parasites and facilitate their elimination but also block crucial steps of the life cycle and, subsequently, impede the natural progression of infection.

Ethics statement. All experiments carried out on mice were approved by the IBMC.INEB Animal Ethics
Committees and the project was licenced by the Portuguese National Authority for Animal Health, in accordance to the statements on the directive 2010/63/EU of the European Parliament and Council.
Mice, parasites and mosquitoes. Four-to six-week-old C57BL/6 or NMRI mice used in this study were purchased from Charles River or the IBMC/i3S animal facility.
P. berghei ANKA strain clone 676cl1 expressing a GFP-luciferase fusion gene via the pbef1 promoter (GFP:LUC) 26 was used to generate mutant lines and was included in experiments the parental control.
Anopheles stephensi mosquitoes (Sda500 strain) were reared in the Centre for Production and Infection of Anopheles (CEPIA) at the Pasteur Institute using standard procedures. Female mosquitoes were fed on infected NMRI mice as previously described 27 and from 21 to 28 days after the infectious blood meal these were either dissected for collection of salivary gland sporozoites or used in transmission experiments. In the latter scenario, starvation by sucrose deprivation 1 day prior to experimentation was performed to enhance the rate of mosquito bite.
Generation of TRSPclonal lines. Targeted deletion of trsp by double crossover homologous recombination was achieved using a previously described strategy 10 , with modifications. To this end, the 3′ and 5′ UTR sequences used as homology regions were amplified from genomic DNA using a Taq DNA polymerase with proofreading activity (Takara) and primer pairs P1 + P2 and P3 + P4 (Supplementary Table 1), respectively. Amplicons were cloned into pGEM-T Easy Vector (Promega), sequenced and later subcloned into the KpnI/ HindIII or EcoRI/BamHI sites of pL0001 vector (MR4; 5′ or 3′ homology regions, respectively). The linearized targeting vector resulting from KpnI/BamHI digestion was used to transfect schizonts with the Nucleofector ® device (Amaxa) 28 . Briefly, electroporated merozoites were injected intravenously into two mice (parental populations) and selected by pyrimethamine treatment at 0.07 mg/L in drinking water, started the day following infection. Once parasitaemia was above 1%, blood from each animal was transferred to two naive mice (transfer populations) for a second round of selection. Parasites were then cloned by limiting dilution.

Genetic validation by PCR and Southern blot analysis.
Genomic DNA extraction. Blood collected from infected animals was filtered with a Plasmodipur leukocyte filter (EuroProxima), followed by erythrocyte lysis with saponin (0.15%). DNA was then extracted and purified from the resulting parasite pellet using the QIAamp DNA Blood Mini Kit (Qiagen).
PCR. Absence of the trsp open reading frame (ORF) and integration of the transfection construct at the correct locus were confirmed by standard PCR analysis with primer pairs P5 + P6 and P7 + P8 (Supplementary Table 1), respectively.
Southern blot. For Southern blot analysis, 5 µg of genomic DNA were digested with EcoRV, separated on a 0.8% agarose gel and transferred to a Hybond-N + nylon membrane (Amersham). The DNA sequence amplified by PCR using primer pair P1 + P2 (Supplementary Table 1) was used as a probe. Labelling of the probe and signal generation and detection were performed resorting to the AlkPhos Direct Labeling and Detection System with CDP-Star chemiluminescent detection reagent (Amersham).

Evaluation of gene expression by reverse transcription PCR (RT-PCR).
Total RNA was isolated from salivary gland sporozoites using the Nucleospin RNA II kit (Macherey-Nagel) and reverse transcription for cDNA production was achieved with the NZY First-Strand cDNA Synthesis Kit (NZYtech). Primer pair P5 + P9 (Supplementary Table 1) was utilised to detect trsp cDNA by PCR. Tubulin beta chain gene cDNA was used as a control, employing primer pair P10 + P11 (Supplementary Table 1).
Bioluminescence imaging. Liver parasite loads in C57BL/6 mice infected with luciferase-expressing sporozoites were assessed by bioluminescence imaging using the IVIS Lumina LT system (Perkin Elmer). Prior to image acquisition, animals had their ventral fur shaved with an appropriate clipper. Infected mice were anaesthetized with 2.5% isoflurane (O 2 flow of 1 L/min) and injected subcutaneously with D-luciferin potassium salt (2.4 mg, Perkin Elmer) five minutes before image acquisition. Mice were then transferred to the stage of an intensified charge-coupled device photon-counting video camera box where anaesthesia was maintained with 2.5% isoflurane (O 2 flow of 0.3 L/min). Signal acquisition was controlled by the Living Image software (Perkin Elmer). In the end, animals returned to their cage and recovered from the anaesthesia. The detection of bioluminescent signals by the system resulted in the generation of signal maps automatically superimposed to the grey-scale photograph of the mice. The quantifications were performed using the Living Image software (Perkin Elmer). The regions of interest (ROI) encompassing the liver and, in homing assays, the whole ventral views of the animal body were manually defined 9 . The total flux (photons/second) and average radiance (photons/second/cm 2 /steradian) within these ROIs were automatically calculated. In the homing assays, the percentage of bioluminescent signal in the liver ROI was calculated by dividing the total flux of the respective ROI by the total flux of the ROI of the animal body in the ventral position.
Parasitaemia. The percentage of infected red blood cells was evaluated by Giemsa-stained thin blood smear starting from the third day after sporozoite inoculation. At least 40 microscopic fields (~ 20,000 red blood cells; detection limit 0.005%) were analysed.