Research Article

Transferrin receptor 1 is a reticulocyte-specific receptor for Plasmodium vivax

See allHide authors and affiliations

Science  05 Jan 2018:
Vol. 359, Issue 6371, pp. 48-55
DOI: 10.1126/science.aan1078

Vivax malaria host receptor

Human malaria is caused by half a dozen species of Plasmodium protozoan parasites, each with distinctive biology. P. vivax, which causes relapsing malaria, specifically parasitizes immature red blood cells called reticulocytes. Gruszczyk et al. identified TfR1 (host transferrin receptor 1) as an alternative receptor for P. vivax. TfR1 binds to a specific P. vivax surface protein. However, the parasite that causes cerebral malaria, P. falciparum, does not share TfR1 as a receptor: P. falciparum could still infect cells in which TfR1 expression was knocked down, but P. vivax could not. Monoclonal antibodies to the P. vivax protein successfully hindered P. vivax infection of red blood cells.

Science, this issue p. 48


Plasmodium vivax shows a strict host tropism for reticulocytes. We identified transferrin receptor 1 (TfR1) as the receptor for P. vivax reticulocyte-binding protein 2b (PvRBP2b). We determined the structure of the N-terminal domain of PvRBP2b involved in red blood cell binding, elucidating the molecular basis for TfR1 recognition. We validated TfR1 as the biological target of PvRBP2b engagement by means of TfR1 expression knockdown analysis. TfR1 mutant cells deficient in PvRBP2b binding were refractory to invasion of P. vivax but not to invasion of P. falciparum. Using Brazilian and Thai clinical isolates, we show that PvRBP2b monoclonal antibodies that inhibit reticulocyte binding also block P. vivax entry into reticulocytes. These data show that TfR1-PvRBP2b invasion pathway is critical for the recognition of reticulocytes during P. vivax invasion.

Of the hundreds of Plasmodium species, only P. falciparum, P. vivax, P. ovale curtisi, P. ovale wallikeri, P. malariae, and P. knowlesi are known to infect humans. Within the human host, malaria parasites invade liver and red blood cells for replication and transmission. Blood stage infection is the major cause of all clinical symptoms in malaria, and therefore the therapeutic prevention of parasite entry into red blood cells could alleviate malarial disease. Entry into red blood cells depends on the interactions between parasite invasion ligands and their cognate red blood cell receptors, of which only a handful have been identified (17). These ligand-receptor interactions initiate a cascade of molecular events that progress from initial attachment, recognition, commitment, and last, penetration of the parasite into red blood cells (8, 9).

P. vivax is the most widely distributed human malaria parasite. This parasite has a strict preference for invasion into reticulocytes, which are very young red blood cells that are formed in the bone marrow after enucleation and released into the circulation. The reticulocyte-specific receptor involved in P. vivax entry has not been identified (10). Most studies have focused on the interaction between the P. vivax Duffy binding protein (PvDBP) and the red blood cell Duffy antigen receptor for chemokines (DARC) because individuals from western and central Africa lacking DARC are resistant to P. vivax invasion (11). However, recent reports have highlighted the presence of P. vivax in apparently DARC-negative individuals, suggesting that P. vivax may enter reticulocytes by binding to other receptors (1214). Furthermore, DARC is present on both normocytes and reticulocytes, and therefore this ligand-receptor interaction cannot govern selective entry into reticulocytes (15). To identify other parasite proteins involved in reticulocyte recognition, we focused on the P. vivax reticulocyte-binding protein family (PvRBP). This protein family comprises 11 members, of which several have been shown to bind reticulocytes; however, their cognate receptors have not been identified (1619).

PvRBP2b binds transferrin receptor 1 to mediate recognition of reticulocytes

P. vivax preferentially invades reticulocytes that express high levels of transferrin receptor 1 (TfR1 or CD71) (20). TfR1 is an essential housekeeping protein involved in cellular transport of iron into cells through binding of iron-loaded transferrin (Tf) (21). On circulating red blood cells, TfR1 is expressed only on reticulocytes and is progressively lost from their membranes as they mature into erythrocytes (22, 23). TfR1 is a type II transmembrane glycoprotein that forms a dimer, and its ectodomain consists of three subdomains: a “protease-like domain” resembling the structure of zinc metalloproteinases, an “apical domain,” and a “helical domain” responsible for dimerization (24). TfR1 is also a cellular receptor for New World hemorrhagic fever arenaviruses, including Machupo (MACV), Junin, Guanarito, and Sabiá viruses (25, 26). Residues 208 to 212 of the TfR1 apical domain provide a critical recognition site for these viruses (25, 26).

PvRBP2b is expressed in late-stage P. vivax parasites, and recombinant PvRBP2b (residues 161 to 1454; PvRBP2b161–1454) binds preferentially to reticulocytes that express TfR1 (19, 27). We observed that binding by recombinant PvRBP2b was abolished when reticulocytes were treated with trypsin and chymotrypsin (fig. S1, A and B). We confirmed that the combination of these proteases cleaves TfR1 and complement receptor 1 (CR1) from the surface of reticulocytes, with other known malaria receptors—including glycophorin A, basigin, and DARC—being susceptible to different sets of protease treatment (fig. S1, A and B). The profile of PvRBP2b binding is strikingly similar to the TfR1 surface expression on reticulocytes (Fig. 1A, bottom), and we show that the level of PvRBP2b binding is directly correlated with the levels of TfR1 on the surface of reticulocytes (fig. S1, C and D).

Fig. 1 PvRBP2b binds TfR1 on the reticulocyte surface.

(A) PvRBP2b161–1454 binding in the presence of anti-TfR1 mAbs analyzed by means of flow cytometry. (Left) Dot plots of PvRBP2b161–1454 binding (y axis) to reticulocytes stained with thiazole orange (TO; x axis). (Right) Normalized binding results in which PvRBP2b161–1454 binding in the absence of mAbs was arbitrarily assigned to be 100%. (B) PvRBP2b161–1454 and PfRh428–766 binding were evaluated by means of flow cytometry with the addition of anti-TfR1 mAb OKT9 or CCP 1–3. PvRBP2b161–1454 and PfRh4 binding in buffer were arbitrarily assigned to be 100%. (C) Eluates of individual or mixtures of proteins immunoprecipitated with anti-PvRBP2b mAb analyzed by means of SDS-PAGE. Plus and minus signs indicate protein present and absent, respectively. M, molecular weight marker. (D) Anti-TfR1 mAbs inhibit PvRBP2b-TfR1 complex formation in the FRET-based assay. The FRET signal was relative to “no mAb” control. (E) Binding of PvRBP2b161–1454 and PfRh428–766 in the presence of anti-TfR1 mAb MEM-189, CCP 1–3, and MACV GP1. (Left) Dot plots showing PvRBP2b161–1454 (top) and PfRh428–766 binding (bottom). (Right) Normalized binding results in which PvRBP2b161–1454 and PfRh428–766 binding in the presence of buffer was arbitrarily assigned to be 100%. (F) MACV GP1 inhibits PvRBP2b161–1454–TfR1 complex formation monitored by means of FRET assay. For (A), (B), (D), (E), and (F), mean ± SEM, n ≥ 3 biological replicates; open circles represent biological replicates. Mann-Whitney test was used for (A) and (D), where MEM-75 was considered noninhibitory, and t tests were used for (B), (E), and (F). *P ≤ 0.05, **P ≤ 0.001.

To determine whether PvRBP2b161–1454 binds to the population of reticulocytes that express TfR1 on their surfaces, we tested a panel of commercially available anti-TfR1 monoclonal antibodies (mAbs) for their ability to block recombinant PvRBP2b binding. Indeed, anti-TfR1 mAbs 23D10, L01.1, LT71, M-A712, MEM-189, and OKT9 inhibited PvRBP2b binding to reticulocytes by 78, 76, 33, 75, 92, and 90%, respectively (Fig. 1A). M-A712 also prevents MACV pseudovirus entry (25, 28). Anti-TfR1 mAbs 2B6, 13E4, and MEM-75 did not inhibit PvRBP2b binding; although their epitopes have not been mapped, we propose that these three antibodies may bind to a site on TfR1 that is not involved in the PvRBP2b interaction (Fig. 1A). To determine whether this inhibition was specific to PvRBP2b161–1454 binding, we analyzed the binding of P. falciparum reticulocyte binding protein-like homolog 4 (PfRh4) to its cognate receptor CR1 (4). Whereas addition of the first three complement control protein modules of CR1 (CCP 1–3) inhibited PfRh428–766 binding as expected (29), addition of anti-TfR1 mAb OKT9 did not significantly reduce PfRh4 binding (Fig. 1B). Because anti-TfR1 did not affect PfRh4 binding, these results show that TfR1 is a specific reticulocyte receptor for PvRBP2b.

To evaluate whether PvRBP2b161–1454 interacts directly with TfR1, we performed immunoprecipitation experiments using purified recombinant TfR1, Tf, and PvRBP2b161–1454 proteins (Fig. 1C) (30). Using an anti-PvRBP2b mAb, we immunoprecipitated PvRBP2b in complex with TfR1 and Tf. PvRBP2b and TfR1 also formed a binary complex in the absence of Tf, demonstrating that PvRBP2b binds directly to TfR1 (Fig. 1C). The interaction between PvRBP2b and TfR1 is specific; immunoprecipitation of PvRBP1a, PvRBP1b, or PvRBP2a did not show evidence of complex formation with TfR1 (fig. S2A).

We developed a fluorescence resonance energy transfer (FRET)–based assay to monitor PvRBP2b-TfR1 complex formation in which TfR1 labeled with DyLight-594 could be shown to interact with PvRBP2b161–1454 labeled with DyLight-488 (fig. S2B). The addition of 10-fold molar excess of unlabeled PvRBP2b161–1454 and TfR1 competed out the labeled proteins and reduced the signal of the PvRBP2b-TfR1 FRET pair. By contrast, proteins that were unable to bind TfR1, such as PfRh4, had no effect on the FRET signal. Using this assay, we observed that anti-TfR1 mAbs 23D10, M-A712, MEM-189, and OKT9 that inhibited PvRBP2b161–1454 reticulocyte binding also blocked PvRBP2b-TfR1 complex formation (Fig. 1D).

MACV GP1 and PvRBP2b bind to the apical domain of TfR1

The arenavirus envelope glycoprotein is the only protein on the virion surface and, during maturation, is processed into three subunits: the stable signal peptide, GP1, and GP2. The GP1 subunit interacts with cellular receptors, and the structure of a MACV GP1–TfR1 complex shows that MACV GP1 binds to the apical domain of TfR1 (31, 32). To determine whether PvRBP2b interacts with a similar surface on TfR1, we examined whether soluble MACV GP1 competes with PvRBP2b161–1454 for binding to TfR1 on reticulocytes (Fig. 1E). Indeed, the addition of MACV GP1 reduced PvRBP2b161–1454 binding to reticulocytes, albeit at a lower level of inhibition as compared with the addition of anti-TfR1 mAb MEM-189. This inhibition was specific; PfRh4 binding was unaffected by addition of MACV GP1 or MEM-189 but clearly reduced with the addition of CCP 1–3 (Fig. 1E). The addition of MACV GP1 inhibited PvRBP2b-TfR1 complex formation and reduced the FRET signal to a similar extent as unlabeled PvRBP2b161–1454, whereas addition of CCP 1–3 had negligible effect (Fig. 1F). These results indicate that MACV GP1 and PvRBP2b161–1454 bind to an overlapping site on TfR1.

Crystal structure of the N-terminal domain of PvRBP2b

PvRBP2b is a 326-kDa protein with a putative red blood cell–binding domain and a C-terminal transmembrane region (Fig. 2). We determined the crystal structure of the N-terminal domain of PvRBP2b (residues 169 to 470; PvRBP2b169–470), refined to 1.71-Å resolution (Fig. 2A; fig. S3, A to D; and table S1). The surface of the domain is mostly positively charged (Fig. 2B). It is predominantly an α-helical protein, comprising 10 α-helices and two very short antiparallel β-sheets, each comprising two β-strands. The crystal structure of PvRBP2b169–470 has two disulfide bonds: one between Cys312 and Cys316 and the other between Cys240 and Cys284. This structure closely resembles the homologous domain of PvRBP2a and PfRh5, with a root mean square deviation of 1.7 and 3.7 Å over 268 and 225 aligned Cα atoms, respectively (Fig. 2C and fig. S4) (18, 33, 34). The theoretical x-ray solution scattering pattern calculated from the PvRBP2b169–470 crystal structure coordinates shows excellent agreement with the experimental small-angle x-ray scattering (SAXS) data (χ = 0.35) (fig. S5, A to F, and table S2), with concordance between the crystal and solution conformations apparent from the overlay of the crystal structure and the ab initio calculated molecular envelope (Fig. 2D, left). We also obtained SAXS data for a longer fragment of PvRBP2b including residues 169 to 652 (fig. S6 and table S2). The reconstructed molecular envelope has a rodlike shape, with a C-terminal part forming a continuous extension of the N-terminal domain (Fig. 2D, right). SAXS data for a larger fragment of PvRBP2b encompassing residues 161 to 969 indicate that the molecule adopts an elongated, boomerang-like shape, similar to that previously reported for PvRBP2a (figs. S6, A to F, and S7, A to D, and table S2) (18).

Fig. 2 Crystal structure of the N-terminal domain of PvRBP2b and its functional requirement.

(A) Structure of the N-terminal domain of PvRBP2b from amino acid 169 to 470 shown in two orthogonal views. (B) Electrostatic surface potential on the PvRBP2b structure. (C) Superimposition of the PvRBP2b structure (green) with PvRBP2a (purple) and PfRh5 (orange). The Protein Data Bank (PDB) ID codes for PfRh5 and PvRBP2a are 4WAT and 4Z8N, respectively. (D) Crystal structure of the N-terminal domain superimposed with SAXS ab initio bead model of PvRBP2b169–470 (left) and PvRBP2b169–652 (right). (E) Sliding window analysis showing nucleotide diversity (π) values and Tajima’s D statistic in PvRBP2b. The gray boxes refer to a highly polymorphic region at amino acid positions 169 to 470 that appears to be under balancing selection. (F) Schematic representation of full-length PvRBP2b and recombinant protein fragments (left). Signal peptide (SP), transmembrane domain (TM), and N-terminal domain (yellow) are indicated. (G) PvRBP2b binding results by means of flow cytometry, in which PvRBP2b161–1454 binding was arbitrarily assigned to be 100%. (H) Unlabeled recombinant PvRBP2b fragments or PfRh4 were mixed at 10-fold molar excess relative to the labeled PvRBP2b161–1454–TfR1 FRET pair. The FRET intensity was relative to buffer control. For (G) and (H), mean ± SEM, n = 4 biological replicates; open circles represent biological replicates. The Mann-Whitney test was used to calculate the P value by using the binding of 2b474–1454 that was considered no binding. *P ≤ 0.05, **P ≤ 0.001.

We calculated nucleotide diversity (π) and Tajima’s D within PvRBP2b using data from the MalariaGEN P. vivax Genome Variation project (35). There was a peak in both metrics between amino acid positions 169 and 470, suggesting balancing selection within the N-terminal domain (Fig. 2E). Such signatures of balancing selection are often associated with genes or proteins expressed on the surface of merozoites and are likely due to interaction with the immune system.

To determine the importance of the N-terminal domain for PvRBP2b function, we generated a series of purified recombinant PvRBP2b protein fragments (Fig. 2F and tables S3 and S4). All proteins were soluble and properly folded as indicated by high α-helical content in CD spectra, which is in agreement with the secondary structure predictions (fig. S8, A to D). We observed that all fragments with the N-terminal domain bound reticulocytes (PvRBP2b161–1454, PvRBP2b161–969, PvRBP2b169–813, and PvRBP2b169–652), whereas their corresponding fragments without the domain did not (PvRBP2b474–1454 and PvRBP2b474–969) (Fig. 2G). However, the isolated N-terminal domain PvRBP2b169–470 was unable to bind reticulocytes on its own (Fig. 2G), indicating that this fragment of PvRBP2b is necessary but not sufficient for reticulocyte binding. The shortest PvRBP2b fragment that showed binding to reticulocytes encompasses residues 169 to 652 (Fig. 2G). Our FRET-based assay showed that unlabeled recombinant fragments that bind reticulocytes inhibited PvRBP2b-TfR1 complex formation, whereas recombinant fragments that did not bind reticulocytes had a negligible effect (Fig. 2H). Collectively, our structural and functional analyses indicate that the N-terminal domain is necessary for binding but requires the presence of the elongated C-terminal fragment to form a fully functional binding site.

PvRBP2b, TfR1, and Tf form a stable complex at nanomolar concentrations

Using surface plasmon resonance, we found that PvRBP2b161–1454 interacts with TfR1 alone or with the binary complex of TfR1-Tf (Fig. 3A, top and bottom, respectively). We also observed similar results for the PvRBP2b161–969 fragment with TfR1 and TfR1-Tf (Fig. 3B, top and bottom, respectively). These results indicate that Tf was not required for the PvRBP2b-TfR1 complex formation because the addition of Tf resulted in similar binding responses than for TfR1 alone. We analyzed a PvRBP2b, TfR1, and Tf ternary complex using analytical size exclusion chromatography (SEC) and used SDS–polyacrylamide gel electrophoresis (SDS-PAGE) analyses to confirm comigration of complex components. The ternary complex was detected for PvRBP2b161–1454 and PvRBP2b161–969 (Fig. 3, C and D, top, respectively, and table S5). By contrast, their corresponding fragments without the N-terminal domain (PvRBP2b474–1454 and PvRBP2b474–969) did not form any observable ternary complexes (Fig. 3, C and D, bottom). The interaction between PvRBP2b and TfR1-Tf binary complex is similar in the presence of either the iron-depleted or iron-loaded form of human transferrin (fig. S2C). Furthermore, the homologous member of the same protein family, PvRBP2a, did not form a ternary complex with TfR1-Tf (fig. S2D).

Fig. 3 PvRBP2b binds to TfR1-Tf to form a stable ternary complex.

(A) PvRBP2b161–1454 and (B) PvRBP2b161–969 were coupled covalently to a biosensor chip to probe binding of TfR1 (concentration range assayed, 2 μM to 7.5 nM, top) and TfR1-Tf complexes [concentration range of TfR1-Tf complexes assayed, 2:4 μM to 1.8:3.9 nM (A) and 2:4 μM to 7.5:15 nM (B), bottom]. (C and D) Complex formation between PvRBP2b, TfR1, and Tf analyzed by means of analytical SEC. PvRBP2b-TfR1-Tf ternary complex can be observed for PvRBP2b161–1454 [(C), top] and PvRBP2b161–969 [(D), top]. Two corresponding truncations of the N-terminal domain, PvRBP2b474–1454 [(C), bottom] and PvRBP2b474–969 [(D), bottom], do not interact with the TfR1-Tf binary complex. The exclusion volume (V0) of the columns and the elution volumes of selected marker proteins are indicated with black arrowheads. (Bottom) Coomassie blue–stained SDS-PAGE gels of the fractions obtained from SEC. (E and F) Continuous sedimentation coefficient distributions derived from fitting sedimentation velocity data to a c(s) sedimentation model. (E) c(s) distributions for TfR1 (black line), Tf (magenta line), and PvRBP2b161–969 (blue line). (F) c(s) distributions for the TfR1-Tf complex (red line) and PvRBP2b161-969–TfR1-Tf complex (green line).

Sedimentation velocity analyses of TfR1, Tf, and PvRBP2b161–969 indicated that the isolated proteins are homogenous, with weight-average sedimentation coefficients of 7.3, 4.9, and 3.6 S, respectively (Fig. 3E). These values are consistent with a stable dimer of TfR1 and monomeric forms of both Tf and PvRBP2b161–969. The empirically fitted shape parameter value (frictional ratio) calculated for PvRBP2b161–969 was ~1.8, which is consistent with a highly elongated structure in solution. Mixtures of TfR1-Tf and PvRBP2b161–969–TfR1-Tf yielded single symmetrical peaks with weight-average sedimentation coefficients of 11.5 and 11.2 S, respectively, with no peaks observed for the individual components in these samples (Fig. 3F and fig. S9). These results indicate that Tf and TfR1 form a stable binary complex in solution and that PvRBP2b161–969 binds to this binary complex. The frictional ratio (f/f0) for the ternary PvRBP2b161–969–TfR1-Tf was higher than for the binary TfR1-Tf complex, resulting in a reduction in the sedimentation coefficient on formation of the ternary complex and indicating that it has an elongated structure in solution.

Deletions in TfR1 generated via CRISPR/Cas9 abolishes PvRBP2b binding and P. vivax invasion

To investigate whether loss of TfR1 surface expression on red blood cells would affect PvRBP2b protein binding, we attempted to generate a knockout of the TFRC gene using CRISPR/Cas9 genome editing of the JK-1 erythroleukemia cell line. We obtained single-cell clones that displayed reduced expression of TfR1 and validated the mutation in two independent clones (TfR1 mut1 and TfR1 mut2) (fig. S10, A to C). Both clones contained an identical –3-bp deletion that resulted in the loss of amino acid Gly217 in the TfR1 apical domain but left the rest of the protein in-frame. TfR1 mut1 was homozygous for this deletion, whereas TfR1 mut2 has a –3-bp deletion, as described above, on one allele and a –11-bp deletion on the other allele, the latter leading to a premature stop codon. Deletion of TFRC in a mouse model is embryonic lethal and leads to severe disruption of erythropoiesis (36), suggesting that complete deletion of TFRC in erythroid-lineage cells, such as JK-1, may not be possible.

Differentiated polychromatic JK-1 cells (termed jkRBCs) express surface proteins (including TfR1) at levels comparable with those of differentiated CD34+ bone marrow–derived cultured red blood cells (cRBCs) (37). The jkRBCs, cRBCs, and differentiated jkRBCs with a knockout within the basigin receptor (ΔBSG) show normal levels of TfR1, whereas TfR1 mutant clones displayed an intermediate level of TfR1 surface staining, with a panel of anti-TfR1 mAbs (Fig. 4A and fig. S11). Levels of glycophorin A (GypA) and basigin (BSG) on these TfR1 mutant clones were similar to all control cells, showing that only TfR1 surface expression is affected on these cells (Fig. 4A). To determine whether deletion of Gly217 affects PvRBP2b binding, we generated a recombinant TfR1 protein that lacks this amino acid (TfR1ΔG217). Using SEC, we show that although TfR1ΔG217 was still able to bind Tf, its binding to PvRBP2b was completely abolished (Fig. 4B). Gly217, which resides on the lateral surface of the TfR1 apical domain, is close to the MACV GP1 interaction surface (fig. S10D) (31).

Fig. 4 Deletions in TFRC reduce TfR1 surface expression, abolish PvRBP2b binding, and inhibit P. vivax invasion.

(A) Expression of TfR1, BSG, and GypA on the surface of jkRBCs, TfR1 mutants, ΔBSG null, and cultured erythrocytes (cRBCs) as measured with flow cytometry. (Right) Cytospin analysis of cells stained with May-Grünwald Giemsa staining technique. (B) TfR1ΔG217 mutation in TfR1 abrogates PvRBP2b binding as observed by using analytical SEC. (C) Quantitative surface proteomics demonstrate specific reduction in TfR1 protein levels in TfR1 mutants compared with wild-type jkRBCs. Levels of Tf, the binding partner for TfR1, are similarly reduced. Significance A was used to estimate P values, and a minimum of two peptides were required for protein quantitation. (D) Binding of recombinant PvRBP2b fragments to jkRBCs, TfR1 mutants, ΔBSG, and cRBCs are shown in blue. Negative controls of unstained cells and isotype control stained cells are shown in the gray and orange lines, respectively. (Right) Compilation of results from PvRBP2b fragment binding to jkRBCs, TfR1 mutants, ΔBSG, and cRBCs. Mean ± SEM, n = 3 biological replicates. (E) Comparison of invasion efficiency between jkRBCs and TfR1 mutant cell lines with either P. vivax or P. falciparum. The data shown are averages and SEM from between four to five biological replicates shown as open circles. P value was calculated by using a paired, two-tailed t test. ****P ≤ 0.0001; ns, nonsignificant.

To confirm that the mutation in TFRC did not result in changes in expression of other red blood cell proteins, we compared the abundance of cell surface proteins between wild-type jkRBCs and the two TfR1 mutants using tandem mass tag–based quantitative surface proteomics (Fig. 4C). Out of 237 quantified surface proteins, only TfR1 and Tf were significantly modified, confirming the specificity of the TFRC mutations.

We next wanted to determine whether PvRBP2b binding was affected in the TfR1 mutant clones. PvRBP2b161–1454 and PvRBP2b161–969 bound jkRBCs and cRBCs, whereas recombinant fragments PvRBP2b474–1454 and PvRBP2b474–969 that lacked the N-terminal domain did not (Fig. 4D). By contrast, we did not detect any PvRBP2b161–1454 and PvRBP2b161–969 binding to TfR1 mutant cells. This abolition of binding was specific to deletions in TFRC because PvRBP2b binding was unaffected on ΔBSG null cells or on cRBCs (Fig. 4D, right). We also compared the invasion efficiency between jkRBCs and TfR1 mutant cell lines with either Brazilian P. vivax isolates or P. falciparum 3D7 (fig. S10E). A significant (>10-fold) reduction in invasion efficiency was observed in the TfR1 mutant line compared with the jkRBCs line with P. vivax, whereas no significant difference was observed with P. falciparum (Fig. 4E). These results validate TfR1 as the cognate receptor for PvRBP2b and that TfR1 is an essential host factor for P. vivax invasion.

Antibodies to PvRBP2b block reticulocyte binding and P. vivax invasion

To examine whether PvRBP2b antibodies could inhibit P. vivax invasion, we raised mouse monoclonal antibodies to PvRBP2b161–1454 and obtained four mAbs. 3E9, 6H1, and 10B12 bound epitopes within the N-terminal domain present in PvRBP2b169–470 with high affinities (Fig. 5A; fig. S12, A to C; and table S6), whereas mAb 8G7 recognized an epitope outside the N-terminal domain within amino acids 813 to 969 (Fig. 5A). Competition enzyme-linked immunosorbent assay (ELISA) experiments using mAbs directly conjugated to horseradish peroxidase (HRP) show that each mAb only competed with itself for binding to PvRBP2b, showing that 3E9, 6H1, and 10B12 bind to distinct epitopes in the N-terminal domain (Fig. 5B). Neither polyclonal nor monoclonal antibodies to PvRBP2b recognize recombinant PfRh4 and five other recombinant PvRBPs, indicating that these antibodies are specific to PvRBP2b (fig. S12B) (19). Using flow cytometry, we show that addition of anti-PvRBP2b mAbs 3E9, 6H1, and 10B12 abolished the PvRBP2b161–1454 binding to reticulocytes, whereas anti-PvRBP2b mAb 8G7 and anti-PvRBP2a mAb 3A11 had no effect (Fig. 5C).

Fig. 5 Anti-PvRBP2b mAbs inhibit reticulocyte binding and P. vivax invasion in Brazilian and Thai clinical isolates.

(A) ELISA plates were coated with equimolar concentrations of each recombinant fragment, and detection with anti-PvRBP2b mAbs 3E9, 6H1, 8G7, and 10B12 are shown. (B) Competition ELISA by using immobilized PvRBP2b incubated with unconjugated anti-2b mAbs (x axis) and detected with 3E9-HRP, 6H1-HRP, 8G7-HRP, and 10B12-HRP as indicated. For (A) and (B), error bars represent range showing the variability of duplicate measures. (C) PvRBP2b161–1454 binding in the presence of anti-PvRBP2b mAbs 3E9, 6H1, 8G7, and 10B12 was analyzed by means of flow cytometry. Normalized binding results in which PvRBP2b161–1454 binding in the absence of mAbs was arbitrarily assigned to be 100%. The anti-PvRBP2a mAb 3A11 was used as a negative antibody control. Mean ± SEM, n = 5 biological replicates, open circles represent biological replicates. The Kruskal-Wallis test was used to calculate the P value by using 8G7 binding as no inhibition. *P ≤ 0.05, ***P ≤ 0.0001. (D) Invasion of P. vivax in Brazilian (blue open circles) and Thai (black open circles) clinical isolates in the presence of anti-PvRBP2b 3E9, 6H1, 8G7, and 10B12, pooled mAbs (each mAb at one-third of final concentration), mouse isotype control, purified rabbit prebleed IgG, purified total IgG of anti-PvRBP2b polyclonal antibody R1527, and camelid anti-Fy6 mAb. Antibodies were added in concentrations from 25 to 125 μg/ml, except for the camelid anti-Fy6 mAb, which was used at 25 μg/ml. Mean ± SD, n = 2 to 6 biological replicates; open circles represent biological replicates. For experiments with n > 2 biological replicates, we used the Kolmogorov-Smirnov test to compare 8G7 with 3E9, 6H1, and 10B12. *P ≤ 0.05, **P ≤ 0.001.

We tested the ability of the antibodies to PvRBP2b to inhibit P. vivax invasion into human reticulocytes, using a short-term P. vivax ex vivo assay with Brazilian and Thai clinical isolates (Fig. 5D, blue and black open circles, respectively). As a control, we used a camelid antibody to Fy6, which is a single monovalent VHH domain (15 kDa) (38, 39) that targets a surface-exposed epitope on DARC and blocks its interaction with PvDBP. The addition of the 25 μg/ml camelid antibody to Fy6 in ex vivo assays by using Thai isolates resulted in 85% inhibition of P. vivax invasion (Fig. 5D). Using four Thai isolates, the addition of inhibitory anti-PvRBP mAbs 3E9, 6H1, and 10B12 at 25 μg/ml resulted in 49, 45, and 42% inhibition of P. vivax invasion, respectively. To determine whether inhibition could be improved by increasing the concentration of anti-PvRBP2b mAbs to match the molarity and valency of the single VHH domain, we used 125 μg/ml of inhibitory anti-PvRBP2b mAbs. Under these conditions, we tested the invasion efficiency of two Brazilian isolates. We observed that addition of inhibitory anti-PvRBP mAbs 3E9, 6H1, and 10B12 at 125 μg/ml resulted in 68, 45, and 57% inhibition of P. vivax invasion in Brazilian isolates, respectively (Fig. 5D). To enable quantitative analyses of the ex vivo assays, we combined our initial results of Thai and Brazilian isolates at their respective mAb concentrations. Increased concentration of the inhibitory anti-PvRBP2b mAbs resulted in an equivalent or small increase in inhibition of P. vivax invasion (Thai at 25 μg/ml versus Brazilian at 125 μg/ml). Thus, our combined sample set underestimates the level of inhibition for antibody concentrations of 125 μg/ml. As additional controls, we included the noninhibitory anti-PvRBP2b mAb 8G7 and an immunoglobulin G1 (IgG1) mouse isotype control, which displayed only 8 and 9% inhibition of P. vivax invasion, respectively (Fig. 5D). These results show that addition of anti-PvRBP2b inhibitory mAbs 3E9, 6H1, and 10B12 resulted in significant reduction of P. vivax invasion compared with the noninhibitory anti-PvRBP2b mAb 8G7 (Fig. 5D).

We show that the inhibitory anti-PvRBP2b mAbs target a domain that appears to be under balancing selection (Figs. 2E and 5A), which may result in differences in inhibition between clinical isolates owing to the presence of polymorphic epitopes. To circumvent inter-isolate differences, we further tested the combination of all three inhibitory mAbs, 3E9, 6H1, and 10B12 pooled together (mAb pool) and polyclonal antibodies to PvRBP2b. The mAb pool resulted in significant 48% reduction in P. vivax invasion in both Thai and Brazilian isolates compared with that of anti-PvRBP2b mAb 8G7 (Fig. 5D). Addition of purified total IgG of polyclonal antibodies to PvRBP2b R1527 resulted in 53% reduction in P. vivax invasion, whereas the rabbit prebleed IgG showed only 5% inhibition (Fig. 5D). A previous study using rabbit antibodies to PvDBP shows that P. vivax invasion was reduced up to 64% (40), a level of inhibition comparable with what has been observed with our antibodies to PvRBP2b (Fig. 5D). These results show that anti-PvRBP2b mAbs that block binding to reticulocytes also inhibit P. vivax invasion and highlight the important role of the PvRBP2b-TfR1 invasion pathway in P. vivax field isolates.

Our results reveal a stable interaction between PvRBP2b and TfR1 and that antibodies to PvRBP2b that block binding to reticulocytes also inhibit P. vivax invasion into human reticulocytes. P. vivax invasion is significantly inhibited in the presence of TfR1 mutant cells, showing that TfR1 is a critical host factor for entry into reticulocytes. We propose that the PvRBP2b-TfR1 interaction is important for the initial recognition of the target reticulocyte cells, which results in the commitment of P. vivax parasites for reticulocyte invasion and the subsequent engagement of PvDBP-DARC in tight junction formation, leading to the successful completion of the invasion process. Identification of the molecular entities required for P. vivax invasion offer the possibility to target multiple invasion pathways for synergistic inhibition of P. vivax blood stage infection.

Supplementary Materials

Materials and Methods

Figs. S1 to S12

Tables S1 to S6

References (4183)

Data Set S1

References and Notes

Acknowledgments: We thank J. Newman from the Commonwealth Scientific and Industrial Research Organization Collaborative Crystallization Centre for assistance with setting up the crystallization screens, the Walter and Eliza Hall Institute’s Monoclonal Antibody Facility for production of antibodies, J. Williamson for assistance with mass spectrometry, and MX and SAXS beamline staff at the Australian Synchrotron for their assistance during data collection. We thank F. Nosten, the staff and patients attending the Mae Sot Malaria Clinic in Thailand, and clinics associated with the Shoklo Malaria Research Unit (SMRU), Tak Province, Thailand. We also thank Y. Colin and O. S. Bertrand (INSERM/University Paris 7) for the generous gift of the antibodies to DARC. W.-H.T. is a Howard Hughes Medical Institute–Wellcome Trust International Research Scholar (208693/Z/17/Z). This work was supported in part by the Australian Research Council Future Fellowships to W.-H.T. and M.D.W.G., a Speedy Innovation Grant to W.-H.T., and a National Health and Medical Research Council fellowship (1105754) to J.M.M. U.K. was supported by a Canadian Institutes of Health Research Postdoctoral Fellowship. R.D.P. is funded by Wellcome Trust 090770. M.P.W. was supported by a Wellcome Trust Senior Clinical Research Fellowship (108070/Z/15/Z). This study received funding from Singapore National Medical Research Council (NMRC) (NMRC/CBRG/0047/2013) and the Agency for Science, Technology and Research (A*STAR, Singapore). SMRU is sponsored by The Wellcome Trust of Great Britain as part of the Oxford Tropical Medicine Research Programme of Wellcome Trust–Mahidol University. Work in the M.T.D. laboratory was supported by National Institutes of Health grant 1R01HL139337. We also acknowledge the support of the B.R. laboratory from the Marsden Fund 17-UOO-241.The authors acknowledge the Victorian State Government Operational Infrastructure Support and Australian Government National Health and Medical Research Council Independent Research Institute Infrastructure Support Scheme. All data and code to understand and assess the conclusions of this research are available in the main text, supplementary materials, and via the following repositories: The atomic coordinates and structure factors for PvRBP2b have been deposited in PDB with accession number 5W53. Genotypes were derived from sequence data generated at the Wellcome Trust Sanger Institute (Wellcome Trust 206194 and 098051).

Stay Connected to Science

Navigate This Article