Molecular determinants of SR-B1-dependent Plasmodium sporozoite entry into hepatocytes

Sporozoite forms of the Plasmodium parasite, the causative agent of malaria, are transmitted by mosquitoes and first infect the liver for an initial round of replication before parasite proliferation in the blood. The molecular mechanisms involved during sporozoite invasion of hepatocytes remain poorly understood. Two receptors of the Hepatitis C virus (HCV), the tetraspanin CD81 and the scavenger receptor class B type 1 (SR-B1), play an important role during the entry of Plasmodium sporozoites into hepatocytes. In contrast to HCV entry, which requires both CD81 and SR-B1 together with additional host factors, CD81 and SR-B1 operate independently during malaria liver infection. Sporozoites from human-infecting P. falciparum and P. vivax rely respectively on CD81 or SR-B1. Rodent-infecting P. berghei can use SR-B1 to infect host cells as an alternative pathway to CD81, providing a tractable model to investigate the role of SR-B1 during Plasmodium liver infection. Here we show that mouse SR-B1 is less functional as compared to human SR-B1 during P. berghei infection. We took advantage of this functional difference to investigate the structural determinants of SR-B1 required for infection. Using a structure-guided strategy and chimeric mouse/human SR-B1 constructs, we could map the functional region of human SR-B1 within apical loops, suggesting that this region of the protein may play a crucial role for interaction of sporozoite ligands with host cells and thus the very first step of Plasmodium infection.


Results
Blockade of CD81 does not prevent P. berghei infection in SR-B1-deficient primary mouse hepatocytes. P. berghei sporozoites infect human hepatocyte cell lines using CD81 or SR-B1 as alternative entry routes 4 . Previous studies have shown that mice deficient for either CD81 or SR-B1 remain susceptible to P. berghei sporozoite infection 2, 3, 5 , which could be explained by the mutual functional compensation between the two entry routes 4 . To test whether CD81 and SR-B1 are the only host factors permitting the entry of the parasite in murine hepatocytes, we analyzed the effect of CD81 neutralization in primary hepatocytes isolated from wild type (WT) or transgenic C57BL/6 J mice harboring a Cre-mediated SR-B1 gene inactivation specifically in the liver 10 . We used the anti-CD81 monoclonal antibody MT81 to neutralize the CD81-dependent entry pathway 11 . CD81 inhibition did not impede P. berghei infection of SR-B1-deficient hepatocytes, but, paradoxically, substantially increased the infection rate, similarly to WT hepatocytes (Fig. 1a). This enhancing effect of anti-CD81 antibodies on P. berghei-infection has been reported before in C57BL/6 mouse hepatocyte cultures, yet the underlying mechanism remains unknown 12 . MT81 efficiently blocked P. berghei sporozoite infection in Hepa1-6 hepatoma cells (Fig. 1b), confirming the neutralizing activity of the antibody 12 . These data indicate that P. berghei sporozoites can infect primary mouse hepatocytes in a CD81-and SR-B1-independent manner and suggest the role of additional entry factors. However, it is difficult to conclude from these experiments on the contribution of mouse SR-B1 in P. berghei sporozoite entry.

CRISPR-Cas 9 mediated inactivation of CD81 abrogates P. berghei infection in Hepa1-6 cells.
To further analyze the role of human and mouse SR-B1, we took advantage of the fact that the murine hepatoma Hepa1-6 cells do not express SR-B1 4 and generated a variant cell line deficient for murine CD81 (CD81 knockout (KO) Hepa1-6 or CD81KOH16) using the CRISPR-Cas9 system. The loss of CD81 expression in CD81KOH16 cells was confirmed by flow cytometry (Fig. 2a) and western blot (Fig. 2b). Confirming previous results using either antibodies or small interfering RNA (siRNA), a dramatic reduction of the percentage of P. berghei infected cells was observed in the CD81KOH16 cell line as compared to parental cells (Fig. 2c). PV quantification by microscopy after staining of UIS4, a PV membrane marker 13 , revealed a complete inhibition of productive infection in CD81KOH16 cells (Fig. 2d). Intranuclear UIS4-negative parasites were observed in the CD81-deficient cells, contrasting with the well-developed EEFs with a strong UIS4 staining found in the parental Hepa1-6 cells (Fig. 2e). We have shown before that intranuclear parasites result from sporozoites arrested during  Human and murine SR-B1 differ in their ability to support P. berghei sporozoite infection. We have previously shown that the ectopic expression of human SR-B1 can restore P. berghei infection in Hepa1-6 cells where CD81 expression has been previously silenced with siRNA 4 . Here, we compared the functionality of SR-B1 proteins from human and mouse origins (hereinafter referred to as hSR-B1 and mSR-B1, respectively) during P. berghei infection after genetic complementation of CD81KOH16 cells. After transient cell transfection with plasmids encoding hSR-B1 or mSR-B1, we observed a similar expression of the two proteins by western blot (Fig. 3a) and flow cytometry (Fig. 3b). The transfected cells were then infected with GFP-expressing P. berghei sporozoites (PbGFP). In agreement with our previous observations in CD81-silenced cells 4 , the transfection of hSR-B1 in CD81KOH16 cells restored their susceptibility to P. berghei infection (Fig. 3c). Unexpectedly, despite similar protein expression, mSR-B1 was not as efficient as hSR-B1 in restoring P. berghei infection (Fig. 3c). We performed similar transfection experiments in the parental Hepa1-6 cell line after CD81 silencing with siRNA, which confirmed the lower functionality of mSR-B1 protein during P. berghei sporozoite infection as compared to hSR-B1 (Fig. 3d).
Human and mouse SR-B1 protein sequence analysis and structure-homology modeling. We next investigated the structural basis that could explain the differential functionality of human and mouse SR-B1 during P. berghei sporozoite invasion. hSR-B1 (isoform 1) contains 509 amino acids (AA) and presents a large extracellular domain (404 AA) flanked by two transmembrane domains (both 23 AA) and two cytoplasmic tails (N-terminal: 12 AA; C-terminal: 47 AA) 6 . The modeling of hSR-B1 using CD36 as a template (PDB ID: 5lgd) 7 shows that the extracellular part of the receptor can be divided into three regions: a N-terminal region (AA 36-136) harboring a thrombospondin-binding domain in the homologous CD36 protein 15 , an apical region (AA www.nature.com/scientificreports/ 137-214) consisting of four alpha helices (α4, 5, 6 and 7), and a large C-terminal region (AA 215-439) contributing to the hydrophobic channel (Fig. 4a,c). The pairwise sequence alignment of hSR-B1 and mSR-B1 showed that the N-terminal and C-terminal extracellular regions were the most similar, with 81.1% and 85.7% identity, respectively, whilst the apical domain is more divergent, with 66.2% identity (Fig. 4c,d). The hSR-B1 protein harbors 9 N-glycosylation sites, against 11 sites for mSR-B1 (Fig. 4c), which likely explains the different migration pattern observed in western blots 16 (Fig. 3a). The superposition of hSR-B1 and mSR-B1 structural models suggested differences for two loops at the very top of the apex, between the α4 and α5 helices and after the α7 helix (Fig. 4b). The model also suggested differences in the electrostatic surface potentials in this area (Fig. 4e). When the structure is orientated in a side view to present its hydrophobic tunnel entrance, the apex lateral surface of mSR-B1 seems to be mainly electropositive whereas electronegativity is predominant in the human  www.nature.com/scientificreports/ model (Fig. 4e). Remarkably, whilst in the model the top of the apical surface is strictly neutral to electropositive in hSR-B1, mSR-B1 displays a dense electronegative region (Fig. 4e), notably due to the presence of an aspartate residue at position 197 (Fig. 4d).

The apical domain of SR-B1 plays a crucial role during P. berghei infection.
To determine whether the predicted structural differences at the apical domain of SR-B1 could explain the differential functionality of human and mouse SR-B1, we analyzed the functional properties of two chimeric constructs made of human and mouse sequences of SR-B1. The ApicalH chimera corresponds to a mSR-B1 backbone protein with a human apical region (AA 137-214) (Fig. 5a,b). Reciprocally, the ApicalM chimera corresponds to a hSR-B1 protein bearing a murine Apical region (Fig. 5a,b). The electrostatic surface potentials of ApicalH and ApicalM apex top are similar to human and mouse SR-B1, respectively, with only ApicalM showing a dense electronegative spot (Fig. 5c). CD81KOH16 cells were transiently transfected with plasmids encoding hSR-B1, mSR-B1, ApicalH or ApicalM. The two chimeras were expressed at the surface of transfected cells and detected by flow cytometry using anti-human and anti-mouse SR-B1 polyclonal antibodies (Fig. 5d). They were also detected by western blot analysis of whole cellular extracts (Fig S1). A slightly higher band was observed in the lanes corresponding to cells transfected with mSR-B1 and ApicalH constructs compared to hSR-B1 and ApicalM, likely explained by the differential glycosylation pattern of the mSR-B1 backbone 16 (Fig. 5a). Cells transfected with ApicalH and ApicalM constructs bound Cy5-labelled HDLs (Fig S2), similarly to hSR-B1 and mSR-B1, suggesting that both chimeras are correctly folded. Transfected cells were then incubated with P. berghei sporozoites, and the number of infected cells was determined at 24 h post-infection. These experiments revealed that replacement of the apex of mSR-B1 by that of hSR-B1 in ApicalH yielded a chimera with a 2-3 fold increase in P. berghei infection rates as compared to mSR-B1 (Fig. 5e). Reciprocally, replacement of the apex of hSR-B1 by that of mSR-B1 in the ApicalM chimera resulted in a loss of function, with infection levels similar to those observed after transfection of mSR-B1 (Fig. 5e). Altogether, these results demonstrate that the hSR-B1 apical helix bundle (AA 137-214) is functionally determinant during P. berghei sporozoite invasion of hepatocytic cell lines.
A short portion of the apical domain of hSR-B1 facilitates P. berghei infection. We then sought to define more precisely the functional regions implicated in P. berghei infection within the apex domain. We designed three new chimeras made of a mSR-B1 backbone harboring short hSR-B1 sequences, based on both the amino acid differences between the mouse and the human sequences, and the putative interacting sites in other CD36 family receptors. The D1 chimera (AA 150-164) includes the loop between the α4 and α5 helices, where the Enterovirus 71 interacting site is located in the SR-B1 homolog LIMP-2, and encompasses a large part of the α5 helix including the PfEMP1-interacting site in CD36 (Fig. 6a,b). The D2 chimera (AA 193-203) comprises the external tip of the α7 helix but also three phenylalanine residues in the downstream loop, which are exclusive to the human sequence (Fig. 6a,b). The D3 chimera (AA 201-211) includes only one of these phenylalanine residues (Fig. 6a,b). The predicted electrostatic surface potential of D1 and D3 apex top is similar to mSR-B1 (Fig. 6c), whereas D2 apex is mostly electropositive, like hSR-B1, with no mark of electronegativity.
After the transient transfection of CD81KOH16 cells, D1, D2, and D3 chimeras were all detected by flow cytometry on the cell surface using the anti-mouse "αM" antibody. Interestingly, only D2 was detected by the anti-human "αH" antibody, similarly to hSR-B1 and ApicalH proteins (Fig. 6d). Infection of the transfected cells with P. berghei sporozoites revealed that replacement of the AA 193-203 sequence of mSR-B1 by that of hSR-B1 in the D2 chimera resulted in a twofold increase in P. berghei infection in CD81KOH16 cells (Fig. 6e). In contrast, replacement of the AA 150-164 or AA 201-211 sequences in the D1 and D3 chimera, respectively, did not increase infection as compared to mSR-B1 (Fig. 6e). These results thus highlight the functional importance of a short 11 amino acid sequence within the hSR-B1 apical domain, which is sufficient to promote efficient P. berghei infection. The black circle highlights a differential electrostatic surface potential between the two chimeric constructs at the top of the "apex" region. (D) CD81KOH16 cells were transfected with hSR-B1, mSR-B1, ApicalH or ApicalM chimera plasmids, or no plasmid as a control (Mock). Protein surface expression was analyzed using anti-hSR-B1 ("αH", blue histograms) and anti-mSR-B1 ("αM", orange histograms), 24 h after transfection. The grey histogram represents untransfected cells stained with the cognate antibody. (E) CD81KOH16 cells were transfected with hSR-B1, mSR-B1, ApicalH or ApicalM constructs, or no plasmid as a control (Mock), and infected with PbGFP sporozoites 24 h after transfection. The number of infected cells (EEFs) was determined by microscopy after UIS4 staining, 24 h after sporozoite addition (mean control values for each experiment: 59, 139, 214, 245, 299, 315 and 383 EEFs/well in hSR-B1-transfected CD81KOH16 cells). ns, non-significant; ***p < 0.001 (one-way ANOVA followed by Tukey's multiple comparisons test).

Discussion
Previous studies highlighted the dual role of SR-B1 during Plasmodium sporozoite invasion and intracellular liver stage development 2, 5 . More recently, we have shown that SR-B1 is an important host factor for P. vivax but not for P. falciparum infection, and that P. berghei sporozoites can use hSR-B1 as an alternative entry route to the CD81-dependent pathway 4 . P. berghei is a rodent-infecting parasite, yet P. berghei sporozoites can readily infect human cells, using either CD81-or SR-B1-dependent pathways 4 . Here, we show that mouse SR-B1, in contrast to its human counterpart, does not support efficient P. berghei sporozoite invasion of murine hepatoma cells. We took advantage of the differential functionality between human and murine SR-B1 and employed a structure-guided strategy to investigate the SR-B1 molecular determinants involved during P. berghei infection. Using complementary chimeras, we demonstrate that the differential ability of human and mouse SR-B1 to support P. berghei infection is due to differences in the apex region. The swaps in the chimeras between hSR-B1 and mSR-B1 sequences were engineered at conserved regions and guided by the structure modeling, thus minimizing the risk of incorrect folding of the chimeras. In addition, reciprocal results were observed with the reciprocal chimeras, and both ApicalH and ApicalM constructs mediated HDL binding to cells, suggesting that the chimeric proteins were correctly folded.
The apical helical bundle of the other CD36 family proteins mediates the binding to a variety of ligands. For instance, the ß-glucocerebrosidase binds to LIMP-2 apical domain to be delivered into the lysosome 17 . Binding of Enterovirus 71 depends on a 7 amino acid sequence (AA 144-151) in LIMP-2 18,19 . Furthermore, an apical phenylalanine of CD36 (F153) binds to Plasmodium PfEMP1 7 . These sites can be mapped on the SR-B1 predicted structure at the intersection between the α4 and α5 helices, at the very top of the apex. By analogy with CD36 and LIMP-2, we speculate that the apical helical domain of SR-B1 may serve as a receptor for a hitherto unidentified sporozoite ligand. One candidate is the 6-cysteine domain protein P36, which is required for sporozoite productive invasion of hepatocytes, and is functionally linked to host receptor usage. In particular, we have shown that P. yoelii sporozoites genetically complemented with P36 protein from P. berghei can infect host cells through a SR-B1-dependent pathway 4 . Whether P36 protein from P. berghei or from the medically-relevant P. vivax binds to the apical helix bundle of SR-B1 remains to be determined.
In the D2 chimera, a string of 11 residues (AA 193 to 203) in the mouse SR-B1 apical domain, at the end and downstream of the seventh helix, was replaced by the corresponding human sequence. This change significantly increased the ability of mSR-B1 to support P. berghei infection. The differential functionality of this small region is consistent with the observation that it is particularly poorly conserved (only 5/11 residues are identical) between the human and mouse sequences. It is possible that this sequence mediates or regulates the interaction with a putative sporozoite ligand. In this regard, and considering the role of CD36 F153 in the binding to Plasmodium PfEMP1 7 , it should be noted that the 11 residue string includes 3 phenylalanines present in the human SR-B1 sequence but not in the mouse SR-B1 sequence. In addition, our structure models suggest that the replacement of this sequence by the corresponding human sequence results in a loss of the dense electronegative spot at the apex of mSR-B1. As poorly functional chimeras (ApicalM, D1 and D3) harboured an electronegative apex, and given the apex of hSR-B1 and the functional chimeras (ApicalH and D2) maintained an electropositive site, we speculate that electronegativity conferred by the 11 residue string of mSR-B1 apex is unfavourable for parasite binding. This sequence may reduce the ability of mSR-B1 to support P. berghei infection either by directly down-modulating the interaction with a ligand, or indirectly by changing other properties of SR-B1 such as its conformation or the interaction with other host surface molecules.
P. berghei sporozoites readily infect CD81-deficient mouse hepatocytes in vivo and in vitro 3,12 , supporting the existence of alternative entry pathways. Whilst SR-B1 provides a CD81-independent route for P. berghei in human cells 4 , we show here that concomitant blockage of murine CD81 and SR-B1 receptors does not prevent P. berghei infection in primary mouse hepatocyte cultures. These results support the existence of alternative entry routes for the parasite, which still remain to be identified. Possible candidate host receptors include the SR-B1-related proteins CD36 and LIMP-2. Although LIMP-2 is predominantly expressed in lysosomes, a fraction of the protein pool localizes at the cell plasma membrane, where LIMP-2 acts as a receptor that mediates the Enterovirus 71 host cell entry 19,20 . LIMP-2 role during Plasmodium infection has not been investigated so far. In contrast, CD36 is known to play major roles during malaria infection. CD36 binds PfEMP1 variants expressed Figure 6. A key domain within the SR-B1 apex regulates P. berghei infection. (A) Mouse and human protein sequence alignment of the apical region AA 132-223 with the corresponding predicted human secondary structure (alpha helices in red and beta strand in blue). Identical, similar and different amino acids are represented in black, blue and red respectively. Short domains D1, D2 and D3 are delimited by boxes. (B) Predicted tertiary structure of D1, D2 and D3 chimeras by homology modeling, highlighting the segments of mouse (orange) or human (blue) origins. (C) Top views of the electrostatic surface potential of the D1, D2 and D3 chimera apices. Values are in units of kT/e at 298 K, on a scale of − 2 kT/e (red) to + 2 kT/e (blue). White color indicates a neutral potential. Black circles highlight differential electrostatic surface potentials between the different chimeric constructs at the top of the "apex" region. (D) CD81KOH16 cells were transfected with hSR-B1, mSR-B1, D1, D2, or D3 chimeric constructs. Protein surface expression was analyzed using anti-hSR-B1 ("αH", blue histograms) and anti-mSR-B1 ("αM", orange histograms), 24 h after transfection. The grey histogram represents untransfected cells stained with the cognate antibody. (E) CD81KOH16 cells were transfected with hSR-B1, mSR-B1, D1, D2, or D3 chimeric constructs, or no plasmid as a control (Mock), and then infected with PbGFP sporozoites 24 h after transfection. The number of infected cells (EEFs) was determined by microscopy after UIS4 staining, 24 h after sporozoite addition (mean control values for each experiment: 59, 139, 214, 245, 299, 315 and 383 EEFs/well in hSR-B1-transfected CD81KOH16 cells). ns, non-significant; **p < 0.01 (one-way ANOVA followed by Tukey's multiple comparisons test). www.nature.com/scientificreports/ at the surface of P. falciparum-infected erythrocytes, and contributes to the cytoadherence of P. falciparum to vascular endothelial cells [21][22][23] . It is also a major receptor for tissue sequestration of P. berghei-infected erythrocytes in mice 24 . A previous study investigated the contribution of CD36 during P. yoelii and P. berghei sporozoite infection, using CD36-deficient mice. The data showed that both parasites could still infect hepatocytes in the absence of CD36 25 . However, in these experiments, the presence of a functional CD81-entry pathway could have masked any important role of CD36. Hence the contribution of CD36 and LIMP-2 deserves further investigation.
In conclusion, this study provides new insights into the function of SR-B1 during malaria infection, and paves the way towards a better characterization of the molecular interactions leading to parasite entry into hepatocytes. Our results may be particularly relevant to P. vivax malaria, as SR-B1 is the first and up to now only known host entry factor for P. vivax sporozoites 4 . The characterization of SR-B1 molecular function and the identification of interacting parasite ligands may lead to the development of novel intervention strategies to prevent P. vivax sporozoite entry, before the establishment of the liver stage and the hypnozoite reservoir.

Methods
Ethics statement. All animal work was conducted in strict accordance with the Directive 2010/63/EU of the European Parliament and Council on the protection of animals used for scientific purposes. Protocols were approved by the Ethical Committee Charles Darwin N°005 (approval #7475-2016110315516522).
Small interfering RNA silencing of CD81. The siRNA oligonucleotide against CD81 (5′-CGU GUC ACC UUC AAC UGU A-3′) was validated in previous studies 12 . Transfection of siRNA oligonucleotides was performed by electroporation in the presence of 10 µL of 20 µM siRNA, as previously described 14 . Cells were cultured for 48 h before infection or analysis by immunofluorescence. As negative controls, we used cells electroporated in the absence of siRNA oligonucleotide.
Generation of a CD81KOH16 cell line using CRISPR-Cas9. The day before transfection, Hepa1-6 cells were plated in 24 well plates at a density of 90 000 cells per well. Cells were transfected with 500 ng of Len-tiCrispR V2 (Addgene plasmid #52961) containing a guide RNA targeting mouse CD81 (GCA ACC ACA GAG CTA CAC CT) using Lipofectamine 2000 (11668027, Life Technologies). Puromycin selection was carried out 36 h after transfection using a 5 µg/ml solution (ant-pr-1, InvivoGen). Cells were exposed to puromycin for 48 h, then washed and expanded for two weeks in DMEM complete medium before analysis. Immunostaining was performed using the rat monoclonal antibody MT81 to label mouse CD81 11 . All incubations were performed at 4 °C for one hour. We used a 2 µg/ml final concentration of Alexa Fluor 488-conjugated Goat anti-rat antibody (A1106, Life technologies) as a secondary antibody. Cells were then fixed with 1% (w/v) formaldehyde solution and analyzed using a Guava EasyCyte 6/2L bench cytometer equipped with 488 nm and 532 nm lasers (Millipore).
Homology modeling of SR-B1 chimeras. The SR-B1 amino acid sequence of H. sapiens (Uniprot: Q8WTV0) was submitted to the HHpred interactive server for remote protein homology detection 29 . The server identified the X-ray structure of the scavenger receptor CD36 (PDB ID: 5lgd) at 2.07 Å resolution 7 as the best template to model the SR-B1 protein (probability: 100%, e-value: 2.3e−91). Sequences of SR-B1 chimeras were aligned and modeled using Swiss-Model through the ExPAsy molecular biology suite 30 . Each SR-B1 model was then subjected to loop refinement and energy minimization using GalaxyRefine 31 and YASARA 32 , respectively. SR-B1 models were validated for quality using MolProbity for local stereochemistry 33 , and Prosa II for global 3D quality metrics 34 . Additionally, we validated the structure by checking that all the N-glycosylation sites were solvent-exposed.
The protein electrostatic surface potential was calculated using Adaptive Poisson-Boltzmann Solver (APBS) 35 , after determining the per-atom charge and radius of the structure with PDB2PQR v.2.1.1 36 . The Poisson-Boltzmann equation was solved at 298 K using a grid-based method, with solute and solvent dielectric constants fixed at 2 and 78.5 respectively. We used a scale of −2 kT/e to + 2 kT/e to map the electrostatic surface potential in a radius of 1.4 Å. All molecular drawings were produced using UCSF Chimera 37 . www.nature.com/scientificreports/ SR-B1 chimeric construct design and plasmid transfection. Plasmids encoding human and mouse SR-B1 have been described previously 38,39 . The ApicalH and ApicalM chimeras were obtained by cloning a single insert amplified from chimeric synthetic genes (Eurofins Genomics) into the mSR-B1 and hSR-B1 plasmids, respectively. The D1, D2 and D3 chimeras were generated by inserting into the mSR-B1 plasmid two fragments amplified with primers containing hSR-B1 sequences. The sequence of all oligonucleotides used to amplify DNA inserts and the sequence of synthetic genes used as templates are indicated in the Supplementary Western blot. After cell lysis in 1% (v/v) NP-40, soluble fractions were analyzed by western blot under non-reducing conditions, using a Biorad Mini-Protean electrophoresis chamber for SDS-PAGE and transfer on polyvinylidene fluoride (PVDF) membranes. Membranes were probed with anti-mouse CD81 MT81 11 at 2 µg/ml, anti-mSR-B1 polyclonal antibody (Ab24603) diluted at 0.9 µg/ml, and anti-mouse GADPH (TAB1001) as a loading control (0.5 µg/ml). Chemiluminescence detection was performed using ECL Prime reagents (RPN2232,GE healthcare Life sciences) and an ImageQuant LAS 4,000 system (GE Healthcare).
Immunofluorescence assays. For the immunolabeling of SR-B1 and chimeric proteins, cells were harvested using an enzyme-free cell dissociation buffer (13151014, Thermofisher). All incubations were performed at 4 °C in PBS supplemented with 3% (v/v) BSA for one hour with either "αH" anti-SR-B1 polyclonal rabbit serum 39 or "αM" anti-SR-B1 polyclonal rabbit antibodies NB400-113 (Novus Biological). We used a 2 µg/ml final concentration of Alexa Fluor 488-conjugated Donkey anti-rabbit antibody (Ab150073, Life technologies) as secondary antibody with a 45 min incubation. After fixation in 1% (w/v) formaldehyde, cells were analyzed using a Guava EasyCyte 6/2L bench cytometer equipped with 488 nm and 532 nm lasers (Millipore). Flow cytometry plots are representative of at least three independent experiments.
in vitro infection assays. Hepa1-6 cells were seeded in 96 well plates (2 × 10 4 per well seeded the day before transfection) and incubated with 1 × 10 4 PbGFP sporozoites for 3 h, washed, and further cultured until 24 h post-infection. HepG2 and HepG2/CD81, plated in 96 well plates with 3 × 10 4 cells per well seeded the day before infection, were infected using 5 × 10 3 PbGFP sporozoites. In some experiments, anti-mouse CD81 MT81 at 20 µg/ml 11 was added to the cultures at the same time as sporozoites, and the mix was incubated for 3 h before washing the cells with fresh medium. After 24 h, infected cultures were either trypsinized for detection of GFPpositive cells by flow cytometry or fixed with 4% (w/v) paraformaldehyde for fluorescence microscopy. Flow cytometry was performed on a Guava EasyCyte 6/2L bench cytometer (Millipore), and a total of 10,000 cells was analyzed for each sample. For fluorescence microscopy, infected cultures were labeled with polyclonal goat antibodies specific for UIS4 (Sicgen) used at 2 µg/ml, secondary Alexa Fluor 594-conjugated Donkey anti-goat antibodies (A11058, Life technologies) at 2 µg/ml, and the nuclear stain Hoechst 33342. The total number of UIS4-positive EEFs was counted in each well.

Statistical analyses.
Statistical analyses were performed with GraphPad Prism on at least three independent experiments, each performed at least in triplicate, as indicated in the figure legends. All graphs show the mean ± SEM (unless otherwise indicated) expressed as percentage of control (cells without MT81, WT Hepa1-6 cells or CD81KOH16 cells transfected with hSR-B1, as indicated in the figure legends). Each dot represents the mean of the triplicate values of each experiment.