DARC extracellular domain remodeling in maturating reticulocytes explains Plasmodium vivax tropism

Plasmodium vivax is the most prevalent parasite species outside sub-Saharan regions of Africa and accounts for 10 to 100 million malaria cases annually.^1-3  Malaria-associated symptoms are mainly due to blood stage infection, which causes morbidity and occasional mortality, especially in the elderly and children.² P vivax merozoites exclusively invade reticulocytes,⁴  which depend on the expression of DARC (Duffy antigen chemokine receptor). DARC (officially annotated as ACKR1, atypical chemokine receptor 1) is a transmembrane G protein–coupled chemokine receptor family member that lacks the DRY signaling domain.^5,6  DARC interacts with P vivax Duffy binding protein (DBP^7,8 ), leading to dimerization of DBP and subsequent P vivax invasion of reticulocytes.^5,9  Parasite-induced evolutionary pressure leads to DARC negativity in sub-Saharan Africa, conferring a degree of protection against P vivax (review, see Baird¹⁰ ). Cocrystallization of DBP and DARC revealed that DARC residues 19 to 30 (QLDFEDVWNSSY) form the primary DBP binding site.⁵  Accessibility of the DBP binding site in DARC during reticulocyte maturation is ill defined but relevant to understand P vivax tropism and important for eventual vaccine development.^11,12  Malleret et al showed that P vivax prefers immature CD71⁺ reticulocytes but not CD71⁻ cells and that the invasion is blocked upon anti-DARC treatment.¹³  However, the mechanism of P vivax reticulocyte tropism remains to be elucidated, especially as DARC is expressed on both reticulocytes and erythrocytes. We hypothesized that the DBP binding site in DARC paramount to P vivax–reticulocyte association is differentially exposed upon reticulocyte maturation. To investigate this, we studied the recognition of specific DARC epitopes during various stages of reticulocyte maturation defined by expression of CD71 and RNA.

For antibodies, see supplemental Table 1, available on the Blood Web site. Thiazole orange (TO) was acquired from Sigma-Aldrich (St. Louis, MO). Recombinant Pv-DBP (P vivax Duffy binding protein) and biotinylated DBP were produced as previously published.^9,12 

For human blood, informed consent and ethical approval were given in accordance with the Declaration of Helsinki and the Dutch national and Sanquin internal ethic boards. Red cells were isolated by Ficoll-paque (ρ = 1.077 g/mL; 526g for 30 minutes), washed twice with phosphate-buffered saline, and subjected to Percoll-Urografin continuous gradients (48 209.7g; 20 minutes) to obtain reticulocyte enriched fractions.

Cells were supplemented in phosphate-buffered saline–bovine serum albumin (1%) and incubated with antibodies (30 minutes, 4°C) or DBP-biotin (60 minutes, RT) as indicated in the figure legends, followed by secondary fluorochrome-coupled antibodies or Streptavidin-APC for 30 minutes, 4°C. When appropriate, cells were stained with anti-CD71 Pacific blue and TO (100 ng/mL). Cells were measured on FACSCanto II or sorted with a FACS-Aria II/III (BD Biosciences, Oxford, UK), and data were analyzed using FlowJo software (Tree Star). DBP surface plasmon resonance (SPR) was performed as previously described.^14,15 

Cells were lysed in CARIN lysis buffer (20 mM Tris-HCl pH 8.0, 138 mM NaCl, 10 mM EDTA, 100 mM NaF, 1% Nonidet P-40, 10% glycerol). Following protein quantification (detergent compatible; Bio-Rad Laboratories, Hercules, CA), lysates were boiled in Laemmli sample buffer (2% sodium dodecyl sulfate [SDS] wt/vol, 10% glycerol, 5% 2-mercaptoethanol, 60 mM Tris-HCl pH 6.8, brome-phenol blue; 3 minutes, 95°C), subjected to SDS-polyacrylamide gel electrophoresis , blotted using iBlot-PVDF blotting system (Thermo Fisher Scientific, Bleiswijk, The Netherlands), and stained as indicated in the figure legends.

Cord blood reticulocytes can be divided into different maturation stages based on CD71 and RNA content.¹⁶  However, this has never been demonstrated for adult peripheral blood. Using CD71 expression and RNA content, we defined 4 subpopulations of circulating reticulocytes in adult peripheral blood. Normalized to the average reticulocyte frequency of 1% in blood, we identified CD71^high/TO^high (R1) at ∼0.016%, CD71^low/TO^high (R2) at ∼0.059%, CD71^neg/TO^high (R3) at ∼0.37%, CD71^neg/TO^low (R4) at ∼0.55%, and CD71^neg/TO^neg erythrocytes (E; Figure 1A-C). These reticulocyte subpopulations, including the CD71^high/TO^high population susceptible to P vivax infection, expressed similar levels of band 3 or DARC total protein (present as a smear on western blot; Figure 1D; specificity of DARC antibodies in supplemental Figure 1A-B).

Because DARC protein levels did not change between reticulocytes and erythrocytes, we investigated whether the accessibility of the DBP binding pocket within DARC is different between these populations. The reticulocyte populations were stained with antibodies recognizing different DARC epitopes, including the DBP binding site (Figure 2A). Binding of antibodies Fy6 and 2C3, designed as competitive antibodies overlapping the DBP interaction domain, to CD71^high/TO^high reticulocytes was >10-fold increased compared with more mature reticulocytes and erythrocytes (Figure 2B). Binding of Fy3 antibodies, recognizing the last extracellular loop located outside the DBP binding pocket, was modestly increased (<2.5-fold). This difference indicated that accessibility of the FY6/2C3 epitope rather than overall expression decreases during reticulocyte maturation. Antibodies against Duffy blood group polymorphisms (Fyb and Fya), located near the DBP binding site, revealed high binding in CD71^high/TO^high populations, gradually decreasing upon maturation to erythrocytes with Fya seemingly more sensitive to epitope availability than Fyb (Figure 2B). Of note, compared with the 2C3^high population, the 2C3^low population expressed less CD71 and RNA, whereas overall protein content was similar, indicating that the 2C3^high population represents young reticulocytes (supplemental Figure 1C-D). Antibody binding to erythrocyte membrane protein Glycophorin A (CD235a) remained unchanged between the subpopulations (Figure 2C).

Next, we tested binding of recombinant P vivax DBP-biotin (region II, the ligand for DARC^5,7,9 ) to the various reticulocyte maturation stages. Notably, the strongest interaction of DBP was observed in the most immature CD71^high/TO^high reticulocyte subset, whereas erythrocytes did not bind DBP (Figure 2D). By ImageStream, binding was only observed in the R1 fraction (Figure 2E). SPR quantifies the affinity of proteins (including antibodies) to specific epitopes.^14,15  Flowing CD71⁺ reticulocytes or erythrocytes over SPR chips coated with DBP or anti-GPA antibody revealed a 3 times higher binding in CD71⁺ compared with CD71⁻ reticulocytes, whereas the binding affinity of GPA changed minimally (Figure 2F). Of note, Fya-Fyb- reticulocytes/erythrocytes did not bind DBP, anti-Fya, anti-Fyb, 2C3, and Fy6 antibodies (supplemental Figure 1E).

In conclusion, DBP, responsible and required for P vivax reticulocyte invasion, selectively associates with the most immature reticulocytes. The CD71^highRNA^high fraction is estimated to be 0.016% of all enucleated erythroid cells. Assuming a linear age distribution and rate of clearance, this corresponds to reticulocytes during the first 30 minutes after bone marrow egress. Decreased exposure of the DBP epitope on DARC during reticulocyte maturation suggests that the DARC epitope recognized by DBP is remodeled or differentially exposed. A possible reason could be masking of specific epitopes due to protein complex formation. Indeed, protein 4.1 deficiency reduces DARC expression, indicating that DARC expression is dependent on the junctional complex within the erythrocyte membrane.^17,18  These protein complexes remodel during reticulocyte maturation, which may lead to DARC epitope masking. Using a monoclonal antibody raised against the whole protein of DARC with unknown epitope specificity, Malleret et al reported no change in anti-DARC binding to specific reticulocyte populations in cord blood samples. This finding can be reconciled with our data showing minimal changes in recognition outside the region recognized by DBP.¹⁶  However, the use of monoclonal antibodies to specific epitopes within DARC revealed major recognition differences. It has previously been shown that P vivax preferentially infects CD71^pos immature reticulocytes, which represents our CD71^high/TO^high (R1) and CD71^low/TO^high (R2) populations, but not CD71^neg cells.¹³  Here, we provide a reason for this preferential infection, which is due to increased exposure of the DBP binding pocket in DARC, allowing DBP association with CD71^high/TO^high reticulocytes but not in mature reticulocytes or erythrocytes. It provides a plausible explanation to P vivax tropism, potentially leading to new targeting approaches.

The authors thank the Sanquin Central Facility for flow cytometry/sorting expertise (Sanquin Central Facility, Amsterdam, The Netherlands).

This study was supported by funding from Sanquin (PPOC: 11-035) (M.v.L., E.v.d.A., E.O.), the European Union (F.A. H2020-MSCA-ITN-2015, “RELEVANCE”), and National Institute of Allergy and Infectious Diseases, National Institutes of Health grants R56 AI080792 (N.D.S., N.H.T.) and R01 AI064478 (N.D.S., N.H.T.).

Contribution: E.O. and F.A. performed the majority of the experiments; A.E.H.B. performed the SPR experiment and N.D.S. produced the recombinant DBP; E.v.d.A., M.v.L., G.V., N.H.T., F.A., and E.O. designed the experiments and analyzed the data; E.v.d.A., N.H.T., N.D.S., G.V., E.O., F.A., and M.v.L. wrote and modified the manuscript. The manuscript was critically revised by all authors.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Emile van den Akker, Sanquin Research, Department of Hematopoiesis, Plesmanlaan 125, 1066CX Amsterdam, The Netherlands; e-mail: e.vandenakker@sanquin.nl.