Structural Basis for the ABO Blood-Group Dependence of Plasmodium falciparum Rosetting

The ABO blood group influences susceptibility to severe Plasmodium falciparum malaria. Recent evidence indicates that the protective effect of group O operates by virtue of reduced rosetting of infected red blood cells (iRBCs) with uninfected RBCs. Rosetting is mediated by a subgroup of PfEMP1 adhesins, with RBC binding being assigned to the N-terminal DBL1α1 domain. Here, we identify the ABO blood group as the main receptor for VarO rosetting, with a marked preference for group A over group B, which in turn is preferred to group O RBCs. We show that recombinant NTS-DBL1α1 and NTS-DBL1α1-CIDR1γ reproduce the VarO-iRBC blood group preference and document direct binding to blood group trisaccharides by surface plasmon resonance. More detailed RBC subgroup analysis showed preferred binding to group A1, weaker binding to groups A2 and B, and least binding to groups Ax and O. The 2.8 Å resolution crystal structure of the PfEMP1-VarO Head region, NTS-DBL1α1-CIDR1γ, reveals extensive contacts between the DBL1α1 and CIDR1γ and shows that the NTS-DBL1α1 hinge region is essential for RBC binding. Computer docking of the blood group trisaccharides and subsequent site-directed mutagenesis localized the RBC-binding site to the face opposite to the heparin-binding site of NTS-DBLα1. RBC binding involves residues that are conserved between rosette-forming PfEMP1 adhesins, opening novel opportunities for intervention against severe malaria. By deciphering the structural basis of blood group preferences in rosetting, we provide a link between ABO blood grouppolymorphisms and rosette-forming adhesins, consistent with the selective role of falciparum malaria on human genetic makeup.


Introduction
The ABO blood group system of carbohydrate antigen expression on the surface of human red blood cells (RBCs) is critically important in transfusion medicine. Several associations have been reported between the ABO blood group phenotype and relative risk of infectious diseases, including malaria [1][2][3][4][5][6]. In the case of Plasmodium falciparum malaria, recent studies have indicated that blood group O confers a protective effect against severe malaria [7][8][9]. The best-documented parasite determinant associated with the ABO blood group is rosetting, the capacity of infected RBCs to bind uninfected RBCs, which is consistently associated with severe malaria in African children [10][11][12] and is reduced inblood group O individuals [9,[12][13][14][15][16]. The hypothesis that group O protects against severe malaria by virtue of reduced rosetting has received strong support in a case-control study in Mali [9].
Although the ABO blood group preference of rosetting has been long known, understanding of its molecular basis is still fragmentary. Rosetting is caused by a sub-group of PfEMP1 adhesins encoded by the large var gene family. The extracellular region of PfEMP1 comprises multiple adhesion domains called Duffy Binding-Like (DBL) and Cysteine-Rich Interdomain Region (CIDR) [17]. DBL and CIDR domains are classified into different major classes (a to e) and sub-classes by sequence criteria, while the var genes can be classified into specific subfamilies that possess distinctive upstream and downstream flanking regions [18][19][20][21]. Efforts to unravel the molecular basis of PfEMP1-mediated rosetting are complicated by the mosaic structure of the var genes and the population diversity of var repertoires [20,22]. Nevertheless, the rosette-forming PfEMP1 adhesins described so far, namely IT4/R29 [23], Palo Alto 89F5 VarO [24], 3D7/PF13_0003 [25] and IT4/var60 [26], belong to a specific sub-group called groupA/UpsA var genes and, interestingly, all four present a specific DBL1a 1 -CIDR1c double domain Head region [19]. Analysis of pseudo-rosettes formed on the surface of COS cells or baculovirus-infected insect cells expressing individual PfEMP1 domains have mapped the RBC adhesion region to the N-terminal DBL1a 1 domain [23,24,26].
Here, we sought to understand the structural basis of the blood group preference in PfEMP1-VarO rosetting. We show that Palo Alto 89F5 VarO parasites bind to RBCs with a marked preference for blood group A compared to blood groups B and O. The binding preference of the PfEMP1-VarO N-terminal domain NTS-DBL1a 1 (termed hereafter DBL1a 1 , as we have shown that NTS is a structural component of this domain [27]), and the DBL1a 1 -CIDR1c double domain (called hereafter Head) mirror the ABO blood group preference of VarO parasites. Direct binding of the Head region to blood group trisaccharides was demonstrated by surface plasmon resonance. A more detailed blood group analysis showed that polymorphisms influence binding, with much stronger binding to subgroup A 1 than to subgroup A 2 , and minimal binding to A x (weak variant of A) RBCs. We determined the crystal structure of the double domain Head protein, the first structure of a multiple domain segment of PfEMP1, which reveals numerous interactions between the DBL1a 1 and CIDRc domains, and shows specific structural features of the CIDRc class that differ from the published CIDRa class [28]. Importantly, we clarified the structure of the NTS-DBL1a 1 hinge region that was lacking in our previous structural analysis of the DBL1a 1 -VarO domain [27]. We show that this region is surface-exposed and critical for RBC binding. The structural information obtained from the functional Head protein was used for computer docking and site-directed mutagenesis in order to localize the RBC-binding site. This was mapped to a specific region of the DBL1a 1 domain, which is structurally conserved between different rosetting variants and is located on the face opposite to that of the major heparin-binding site. This work identifies the interaction with the ABO group as pivotal in rosetting and associates, for the first time, P. falciparum rosetting with the A subgroups, consistent with a contribution of virulent malaria to the selection of ABO blood group polymorphisms. The molecular description of the RBC-binding site provides novel perspectives for the development of preventive or therapeutic measures to combat severe malaria.

VarO rosetting and ABO blood group preference
VarO-mediated rosetting was dependent on the presence of a minimum of 5% human serum (Figure S1A), consistent with findings with other laboratory lines [29][30][31][32] and field isolates [33]. VarO rosetting is not CR1-dependent, as the rosetting rate was not correlated with the CR1 expression level on the recipient RBCs ( Figure 1A), was unaffected by the anti-CR1 mAb J3B11 (which reduces rosetting of the R29 line [34]) ( Figure 1B), and, importantly, was unchanged when CR1 was cleaved with trypsin or chymotrypsin ( Figure 1C). Likewise, immunoglobulin binding, implicated in rosetting in some lines [32,35,36], apparently does not come into play in VarO rosetting, as VarO-iRBC rosetting was unimpaired in Ig-depleted human serum ( Figure S1B) and no IgG or IgM binding could be shown on the surface of VarO-iRBCs( Figure S1C).
Incubation of VarO-iRBCs with varying ratios of recipient group A and group O RBCs, differentially labeled using lipophilic fluorescent probes PKH67 or PKH26, showed preferential binding to the former ( Figure 1D). Blood group A was also preferred to blood group B but rosette formation was more efficient with group B than with group O RBCs ( Figure 1D). This ABO blood group dependence, in particular the reduced binding to group O RBCs, is in line with previous observations with rosetting lines and field isolates [9,13,14,16]. Although VarO rosettes could be readily disrupted by low concentrations of sulphated glycosaminoglycans such as heparin [37], they could not be disrupted by soluble blood group A or B trisaccharides used in the 20-40 mM range (data not shown), consistent with reports on other rosetting lines [13,14].Because VarO rosettes were resistant to mechanical disruption, we could not test the capacity of trisaccharides to inhibit the reformation of rosettes as done for some rosetting lines [13,14].

RBC binding of soluble recombinant PfEMP1-VarO domains
All six individual PfEMP1-VarO domains and the Head protein (DBL1a 1 -CIDR1c, corresponding to residues 2-716) were produced as soluble recombinant proteins in Escherichia coli ( Figures  S2A and S2B). All proteins used in this study were monomeric, as judged by gel permeation chromatography (not shown). Correct protein folding was ascertained by CD spectroscopy, analytical ultra-centrifugation ( Figure S2C) and the capacity to induce surface-reacting antibodies (Ab) (Table S1). RBC-binding capacity was monitored by immunoblot and flow cytometry using specific polyclonal sera at dilutions that did not disrupt VarO rosettes. To monitor binding of the Head protein, we used mAb G8.49, which reacted by ELISA with CIDR1c ( Figure S3). Equal molar amounts of protein were used in all assays. Binding was specific, as no signal was detected when the recombinant domain, the anti-VarO Abs or the secondary anti-IgG Abs were omitted. Of all six individual PfEMP1-VarO domains, only constructs containing the N-terminal DBL1a 1 bound RBCs (Figures 2A and 2B). This confirmed previous findings with PfEMP1-VarO domains expressed on the surface of infected Spodoptera frugiperda (SF9) cells [24] and is consistent with reports by other groups [23,38]. TheHead protein bound more efficiently than DBL1a 1 , the mean MFI being enhanced by about 1 log unit and the immunoblot signal consistently higher for the Head protein than for DBL1a 1 . Comparison of binding efficiency of DBL1a 1 constructs of varying length showed essentially similar binding for an 18-Cys construct ending with the canonical Cys(11)xxCys(12) doublet (residues 2-471) and for the 20 Cys constructs (residues 2-487) with two additional Cys residues at the C-terminus (corresponding to the structurally complete DBL1a 1 domain). Presence of a hexa-His tag

Author Summary
Rosetting, the capacity of infected red blood cells (RBCs) to bind uninfected RBCs, is a Plasmodium falciparum virulence factor. Rosetting is influenced by the ABO blood group, being less efficient with O RBCs. Although this preference may account for protection against severe malaria afforded by the O blood group, its understanding is fragmentary. We identify the ABO blood group as the main receptor for the rosetting Palo Alto VarO parasites, which display a marked preference for blood group A. Rosetting is caused by a sub-group of PfEMP1 adhesins. PfEMP1-VarO shares with other rosetting lines a specific NTS-DBL1a 1 -CIDR1c Head region. We show that the Head region binds RBCs more efficiently than NTS-DBL1a 1 and that ABO blood group polymorphisms influence binding of both domains. The 2.8 Å resolution crystal structure of the Head region reveals extensive contacts between the DBL1a 1 and CIDR1c domains, and shows structural features of the NTS-DBL1a 1 hinge region essential for RBC binding. We localize the RBC-binding site to the face opposite to the heparin-binding site of NTS-DBL1a 1 and document direct binding of the Head region to A and B trisaccharides These findings provide novel insights into the interactions established by malaria parasites with a prominent human blood group.
at the C-terminus did not substantially modify binding (data not shown). As the various recombinant domains had their putative Nglycosylation sites removed (NxT/S mutated to NxA), we also produced recombinant DBL1a 1 and Head with a wild-type (wt) coding sequence (DBL1a 1 (wt) and Head(wt), respectively). Seroreactivity was similar for the wt and the corresponding mutated constructs ( Figure S4) but DBL1a 1 (wt) and Head(wt)bound slightly more efficiently than their mutated constructs ( Figure 2). RBC binding by mutated and wt constructs was heparinsensitive and inhibited by rosette-disrupting mAbs ( Figures S5A  and S5B). Presence of human serum moderately enhanced RBC binding, an effect that was more marked for the Head constructs (2.9-4 fold increase) than the DBL1a 1 constructs (1.4-2 fold increase) ( Figures S5C and S5D). A similar enhancement was observed when foetal calf serum was added ( Figure S5C). This indicates that the serum-enhancing activity is not species-specific and, as such, is different from the requirement of human-specific serum for VarO rosetting in human cells.
ABO blood group preference of DBL1a 1 and Head domains ABO blood group preference was explored for DBL1a 1 , DBL1a 1 (wt), Head and Head(wt) using a panel of RBC donors.
All four proteins bound more efficiently to blood group A (N = 5) than to blood group B (N = 3) RBCs and, in turn, to blood group B more efficiently than to blood group O (N = 5) RBCs ( Figure 3A and data not shown). This indicates that the terminal a-1,3-linked N-acetylgalactosamine (GalNAc) of blood group A and, to a lesser extent, galactose (Gal) of blood group B are key determinants of the interaction. The ABO blood group dependence was specific as no such dependence was observed for binding of the P.vivax Duffy Binding Protein PvDBP, known to bind Duffy-DARC on the RBC surface ( Figure 3A). We next explored binding to subtypes of group A RBCs, which differ mainly in the quantity of the terminal a-1,3-linked Nacetylgalactosamine (GalNAc) antigen displayed: subgroup A 1 RBCs express up to four times as many A epitopes as subgroup A 2 , while A x (weak variant of A) RBCs express very low amounts of terminal (GalNAc) [39]. Binding of DBL1a 1 (wt) and Head(wt)was subgroup-dependent, with binding to A 1 (N = 5) being substantially higher than binding to A 2 (N = 5), while binding to A x (N = 3) was minimal ( Figure 3B). Similar results were obtained when binding was monitored by immunoblot assays ( Figure 3C). Binding strongly correlated with the amount of A antigen displayed on the recipient cell (Spearman coefficient correlation, rho = 0.96 (p,0.0001) and rho = 0.93 (p,0.0001) for DBL1a 1 (wt) and Head(wt) respectively) ( Figure S6A). Furthermore, when A 2 RBCs were treated with a-N-acetylgalactosaminidase, which cleaves the GalNAc groups, binding was reduced even further (data not shown). Binding showed an inverse relationship to the amount of H antigen displayed on the RBC (Spearman correlation coefficient, rho = 20.67 (p = 0.001) and rho = 20.70 (p = 0.0006) for DBL1a 1 (wt) and Head(wt)respectively) ( Figure S6B). Levels of binding to A 2 and B RBCs, which displayed similar amounts of H antigen, were similar. Consistent with this, removal of the terminal galactose from blood group B RBCs by treatment with galactosidase reduced their binding to DBL1a 1 (wt)(data not shown).
Direct binding of the recombinant Head(wt)domain to the blood group A and B trisaccharides was demonstrated using a surface plasmon resonance assay with covalently immobilized trisaccharide-conjugated-Bovine-Serum-Albumin as ligands. In line with the observations made on red blood cells, the Head (wt) region appears to bind more efficiently to group A than to group B conjugates ( Figure 4). This was consistently observed across a large range of protein concentration ( Figure S7). Furthermore, pre-mixing Head(wt) with a 2-fold excess of heparin totally abolished binding to both blood group BSA-conjugated trisaccharides ( Figure 4).
Together, these data show that binding is influenced by the amount and type of terminal glycan displayed on the RBC surface and is heparin-sensitive. The N-terminal DBL1a 1 domain was identified as the minimal binding unit of the ABO blood group and the presence of CIDR1c enhanced binding. In order to understand the molecular basis of these interactions, we solved the crystal structure of the double domain Head protein, since we had already determined the structure of the DBL1a 1 domain [27].

Crystal structure of the Head protein
The DBL1a 1 structure we previously determined [27] was cleaved after residue 69, which turned out to abolish RBC binding ( Figure S8, lane 2). The Head construct studied here, however, was prepared with the native, intact sequence. The Head crystallised only in the presence of heparin, as found earlier for DBL1a 1 , with the best crystals diffracting to 2.8 Å resolution. The Head structure (residues 12-487) was solved by molecular replacement and the polypeptide chain in the final model was traced for the entire sequence, with the exception of the first 18 Nterminal residues (see Table 1 for refinement statistics).
The overall shape of the Head is rather compact, with the CIDR1c domain folding back upon the DBL1a 1 domain, thus burying a significant surface at the interface ( Figure 5A). The structure of the Head protein allows a detailed description of all segments that were missing in our previous analysis of the single DBL1a 1 domain, in particular the stretch of residues 53 to 71 and 83 to 94 ( Figure 5B). The NTS region contains two more helices, followed by a long surface-exposed loop that includes canonical cysteines Cys (21) and Cys(1) of subdomain 1 ( Figure 5B and 6). An antiparallel b-sheet then connects to aH1 at the conserved PPR motif, which is very close to the previous DBL1a 1 structure. In the latter structure, we were able to trace only a short peptide D72-F82 in the hinge region, which has a different conformation to that of the intact Head protein analysed here. The structure of the major heparin-binding site is essentially unchanged and all of the typical DBL motifs of subdomain 2 (in green) and part of subdomain 3 (in blue) are very close in the Head and the single domain DBL1a 1 ( Figure 5B). Importantly, the structure of the PEXEL-like sequence of NTS is identical to that observed earlier [27], thus confirming that this sequence is buried and important for the stability of the protein. There are also differences in the region in contact with the CIDR1c domain. The C-terminal moiety of helix aH6 and the N-terminal region of helix aH7, are displaced by up to 14 Å , while loop C357-D372 (including helix aH69) follows a different path. The RMS difference between the DBL1a 1 Ca coordinates of the Head and single domain structures is 2.9 Å for the complete domain, but only 1.3 Å when the NTS-DBL1a 1 junction and the C-terminal region in contact with CIDR1c are excluded. The density connecting DBL1a 1 and CIDR1c is rather poor and allowed modelling the polypeptide backbone atoms only. From D504 onwards, however, the electron density becomes clear and the rest of the CIDR1c domain could be traced unambiguously.
The CIDR1c-VarO domain (220 residues) forms a rather compact structure, in contrast to the CIDR1a-MC179 structure [28], which lacks the N-terminal part of the domain. In CIDR1c-VarO, this region (residues 496-570) contains a twisted antipar-allel b-sheet composed of four strands (b1-4) as well as three short a-helices (H-2, 3 10 H1, H-1) ( Figure 5A and Figure 6), forming an N-terminal subdomain. The latter part of CIDR1c is largely ahelical, with a pattern of helices similar to that of MC179, though helix a is shorter and helices b and c together span the length of helix b in MC179. The relative position of the three-helix bundle H1, H2, H3 and helices a, b, c is, however, quite different, giving CIDR1c-VarO a much more compact shape ( Figures 7A and 7B). The first Cys residue of CIDR1c-VarO (C534, canonical Cys(1)) forms a disulfide bond with C614 (canonical Cys (8)) in helix H2. In our construct, CIDR1c-VarO contains 11 Cys residues (31 for the entire Head protein) and thus there must be at least one free sulfhydryl group per molecule. The protein was therefore purified in the presence of 25 mM cystamine, which prevented the formation of covalent dimers. All Cys residues are engaged in disulfide bridges except one free Cys (C604, canonical Cys(5) of CIDR1c) (Table S2). This free Cys is indeed quite exposed, but the electron density is not sufficiently well defined to show a covalent modification. The MC179 protein included 17 additional residues at the C-terminal end, including C168 (canonical Cys(10)), which paired with canonical Cys(5) [28]. In the VarO sequence there is a Cys doublet (C722 and C723) in the corresponding site and since it was not obvious which of the two would form a disulfide with Cys(5), we terminated our construct upstream of it at residue 716. The polypeptide chain ends in close proximity of C604 (canonical Cys (5)), however, and could conceivably form a disulfide bridge. We have been able to express a longer Head construct ending at V787 in soluble form but did not succeed in obtaining crystals.

Analytical ultracentrifugation
The crystals were prepared in the presence of heparin, which, although not visible in the electron density, may have induced the domains to come together or interfered with any dimer formation. We therefore compared the overall shape of the individual domains and the double domain using analytical ultracentrifugation(AUC) in the absence of heparin. When the hydrodynamic characteristics of the double domain were calculated from the crystal coordinates [40,41], the results showed that the shape of the double domain in solution was very similar to that found in the crystal. Furthermore, the overall shape of the isolated CIDR1c domain in solution corresponds to that in the crystal structure of the double domain (Table S3). Both the CIDR1c and the Head protein are monomeric in solution and in the crystal unit cell, as was the case for the single DBL1a 1 domain [27].

The blood group A binding site of DBL1a 1
Given the preference of VarO iRBC, as well as the DBL1a 1 domain and the Head, for binding to blood group A, we sought to localize its binding site in the two recombinant proteins. We initially used the collection of mutant constructs used for defining the heparin-binding site [27] (Table 2). Mut1 and Mut3, whose affinity for heparin is reduced by more than 100 fold, bind to RBCs, whereas Mut2 and Mut4, whose heparin binding is unaltered or only slightly affected, do not bind ( Figure S9A), indicating that the heparin-binding and RBC-binding sites do not overlap.
In some protein preparations, spontaneous cleavage of the polypeptide occurred after residue R64 or R69 upon storage (depending on the protein). Detailed analysis showed that only the full-length protein retained the ability to bind ( Figure S8, lane 1), thus indicating that the integrity of a surface-accessible region of the NTS moiety was essential for RBC binding. This is the region with a large difference between the crystal structures of the intact  Head and the cleaved, single DBL1a 1 domain. We therefore used the Head structure to guide additional site-directed mutagenesis. Mutation of residues from the hinge region between NTS and DBL1a 1 (Mut 10, 11,12 and 13, see Table 2 and Figure S9B) disrupted binding, confirming its essential structural role. We used computer docking to localize the potential binding site(s) of the terminal trisaccharide of blood group A within the Head protein. The lowest energy solutions of the docked trisaccharide were very tightly clustered about a site that included residues K87, T88, D147 and K149, corresponding to the region that is significantly different in the cleaved DBL1a 1 domain ( Figures 5B and 8). We therefore prepared a series of single or double mutants exploring this region of the structure. Because of the poor yields of the Head protein, these mutants were derived from the single DBL1a 1 domain. This identified several key residues: K95 (Mut17), F145, K216 (Mut19), whose mutation completely disrupted binding, and T88 (Mut15) and D147, K149 (Mut18), whose mutation reduced binding by more than 90%. Single mutations,R64 (Mut14, localized within the NTS domain) and K87 (Mut16, located in the hinge region), reduced RBC binding by 50% (see Figure S9C). Overall, this demonstrates that the site predicted by computer docking was indeed critical for binding. The binding site includes the NTS segment and residues from subdomains 1 and 2 ( Figure 6). Significantly, this region is conserved between rosetting strains and, moreover, is located on the face opposite to the major heparin-binding site ( Figure 8A).

Discussion
Our results identify the ABO blood group as the major VarO rosetting receptor on the host RBC and show that the presence of the CIDR1c domain to form the Head region results in enhanced RBC binding, mimicking the ABO blood group preference of VarO rosetting. The crystal structure of the Head (DBL1a 1 -CIDR1c) allowed mapping of the RBC-binding site to a structurally conserved region of rosette-forming PfEMP1 variants, involving residues from subdomains 1 and 2, with contributions from the neighbouring NTS-DBL1a 1 hinge region. The RBCbinding site is distal to the heparin-binding site, indicating that although heparin prevents binding of the adhesion domain to the RBCs, it does not directly compete with the ligand-receptor interaction to prevent rosette formation.
Previous analyses of interactions at play in VarO rosetting showed that none of the common receptors of P. falciparum cytoadherence, such as CD36, ICAM-1, CSA, nor other potential receptors (VCAM-1, HABP1, CD31/PECAM, E-selectin, Endoglin, CHO receptor ''X'', and Fractalkine) were implicated in the binding of RBC to VarO-iRBCs [42,43]. We show here that CR1/CD35, shown to be a receptor for some rosetting lines (including R29 [23,34]), is not involved in VarO rosetting. VarO rosetting shares with other rosetting lines three generic characteristics, namely an extreme sensitivity to sulphated glycosaminogly-cans, the need for human serum and a marked ABO blood group preference characterised by reduced binding to group O RBCs. Our data indicate that the major determinant affecting VarO rosetting efficiency is indeed the ABO blood group. We explored VarO-iRBC binding characteristics using a monovariant culture of the Palo Alto 89F5 clone, in which .90% of the iRBCs were positively selected to express PfEMP1-VarO [24]. VarO-iRBCs preferentially bind to blood group A compared to blood group B,  which itself is preferred to blood group O. An identical blood group preference profile was observed with both the DBL1a 1 and Head proteins.
Our dissection of blood group A preference using the recombinant domains provides, for the first time, a link between rosetting and common group A polymorphisms. The difference between the A 1 and A 2 subgroups is mainly quantitative, with A 2 RBC displaying 4-5 times fewer blood group A determinants than A 1 [44], although they have some qualitative differences as well [45,46]. Our observation that DBL1a 1 hardly bound A x RBCs is a strong indication that copy number variation of the terminal glycan is the main cause of differences in the binding behaviour between the three subgroups. Binding to A 2 was similar in intensity to binding to blood group B. As the A 2 RBCs expressed about 5-fold more H antigen than the B RBCs ( Figure S6), we conclude that the terminal a-1,3-linked N-acetylgalactosamine (GalNAc) of group A is preferred to the terminal galactose (Gal) of group B. This was further documented by surface plasmon resonance assays, in which we explored binding to trisaccharide-BSA conjugates immobilized on the surface of a sensor chip. Binding to trisacharide A was more efficient than to trisaccharide B (Figure 4), with apparent affinities in the micromolar range. This blood group preference is fully consistent with the VarO-IRBCs blood group preference although it should be mentioned that BSA-conjugates sub-optimally mimic the RBC-displayed blood group saccharides, which present multiple branched saccharides or tandem copies of the blood group determinant.
The structure of the DBL1a 1 -CIDR1c Head reported here is the first crystal structure of a multiple domain from PfEMP1, as well as of a complete CIDR1c domain. The structure of the DBL1a 1 domain published earlier [37] is confirmed and we further show that the CIDR1c-VarO domain forms extensive contacts with the DBL1a 1 region (Figures 9A and 9B). The interface between the DBL1a 1 and CIDR1c domains is stabilised by several charged interactions (D367-D654, N381-R518, E389-Y521, Y437-K550) and, furthermore, a significant hydrophobic surface on DBL1a 1 is buried within this interface. A significant part of the CIDR1c accessible surface (1367A 2 of a total of 12477A 2 ) is buried within the interface, including residues from its N-terminal half, as well as from the a-b-c helical motif (Figure 9). The interface is also highly conserved within the DBL1a 1 -CIDR1c family of PfEMP1 adhesins ( Figure 9C).
The three-dimensional structure of the Head allows a definitive assignment of interdomain boundaries. Thus the boundary between DBL1a 1 and CIDR1c domains lies within a rather flexible linker around residue 490, indicating that the DBL1a 1 domain terminates after the disulphide Cys13-Cys14 (residues C477-C483of PfEMP1-VarO sequence). The CIDR1c domain is more compact than the partial CIDRa structure published by [28], not only in our double-domain crystal structure but also in solution as a single domain (Figures 7 and S10). This difference could be due to a number of reasons: CIDR1c -VarOcontains additional 98 residues at the N-terminus missing in the MC179 construct (which also lacked the preceding DBL1a domain) [28], the extensive contacts between DBL1a 1 and CIDR1c, or the dimer formation in the case of MC179. The two CIDR domains are of different sequence classes, c for VarO and a for MC179. Moerover, the latter binds CD36, which is not the case for VarO [42,43]. Both CIDR classes share a conserved arrangement of Cys and Trp residues in most of their sequence, with an additional disulphide (canonical Cys(8a)-Cys(8b)) in the c class. The sequences are, however, much less well conserved in loop regions connecting both the helices and the subdomains, which could be another reason for the structural differences observed. The similarity between DBL and CIDR domains in their general architecture, noted earlier [28], is also present in the VarO structure. Superposition of the CIDR1c -VarO domain upon the DBL1a 1 -VarO domain matches not only helices H1, H2 and H3 (CIDR1 c) to aH6, aH7 and aH10 (DBL1a 1 ), but also helices a and b (CIDR1c) to aH8 and aH9 (DBL1a 1 ). Furthermore, strands b1 and b2 of CIDR1c lie quite close to strands b-1 and b-2 of DBL1a 1 ( Figure S10). The similarity extends to the disulfide pattern as well, as suggested earlier [28]: canonical disulfides Cys(10)-Cys (11) and Cys(7)-Cys(9) of DBL1a 1 overlap with disulfides Cys(7)-Cys(9) and Cys(4)-Cys(6) of CIDR1c, respectively, while disulfide Cys(5)-Cys(10) of CIDR1c lies close to Cys(8)-Cys(12) of DBL1a 1 . Indeed, the N-terminal half of CIDR could be classified as equivalent to subdomain 1 of DBL domains, while the second, helical domain of CIDR corresponds more closely to subdomain 3.
In the MC179 structure, the region implicated in CD36 binding [28] lies near the N-terminal end of helix b. In the CIDR1c -VarO structure, the equivalent region corresponds to the b-c connecting loop, which faces the H1-H2-H3 helical bundle. Within the triplet S662-I663-D664 of VarO, corresponding to the critical residues E108-I109-K110 of MC179, S662 and I663 are buried by H1 and D664 forms a salt bridge with K591 from H1. Interestingly, S662 and I663 are highly conserved among CIDRc sequences, suggesting that this loop may have a common conformation in this domain class. If these residues were implicated in CD36 binding, they would be poorly accessible in the rosetting strains of PfEMP1, which do not bind CD36 [43].
The structure of the Head provides critical information about the RBC-binding site. Computer docking and site-directed mutagenesis localized a blood group A binding site in a restricted area situated at the interface of subdomain1 and subdomain 2 in the vicinity of the NTS-DBL1a 1 hinge region. This differs from the CSA-binding site localized on VAR2CSA DBL3X domain, which lies within subdomain 3 [47][48][49], and is more in line with the location reported for P. knowlesi Duffy Binding Protein (also a site engaging residues from subdomains 1 and 2) [50] or some of the sialic acid-binding sites of P. falciparum EBA175 [51]. The NTS-DBL1a 1 hinge region, missing in our previous single domain structure, is highly exposed on the surface and proved to be crucial for the RBC-binding site. Indeed, cleavage of this sequence disrupted binding and mutations of this region reduced binding, without substantially affecting antigenicity (recognition of all mutants by ELISA was essentially unimpaired, data not shown). This reinforces the conclusion that NTS is an essential functional and structural component of the DBL1a domain [37]. Importantly, the blood group A binding site and the major heparinbinding site that we mapped previously [37]are distant from each other on the surface, indeed on opposite sides of the molecule. Therefore, direct competition with binding to the receptor cannot be the reason why heparin disrupts rosettes and inhibits the binding of the recombinant domains to RBC and trisaccharide-BSA conjugates, contrary to one of the previously suggested hypotheses [27]. The other possible mechanism, namely that heparin could provoke the formation of oligomers that are no longer competent for receptor binding, remains an interesting possibility. The RBC-binding site could become inaccessible in heparin-aggregated adhesins, or several PfEMP1 molecules need to bind simultaneously to the ABO antigens displayed on the RBC surface for an efficient interaction to occur and this is prevented in the heparin complexes.
We analysed the location of the RBC-binding site on the DBL1a 1 domain with respect to the position of molecular signature tags used to classify var genes and associate them with either severe or uncomplicated malaria. Conserved tags, called positions of limited variability 1 to 4 (PolV1-4), had been identified [52,53]. The relative combination of PoLV motifs appears characteristic of specific var gene subsets. Figures 6 and 10 show the localization of the four PolV sequence tags with respect to the identified RBC binding site of PfEMP1-VarO. All PolV tags, except for PolV1, are remote from the RBC-binding site ( Figure 10). Normark et al. [54] identified specific PfEMP1-DBL1aamino acid motifs correlated with rosetting and severe malaria.One of the sequence signatures associated with ''high rosetting'', namely H3, maps close to the binding site identified here. Palo Alto 89F5 VarO has a H3 motif (H3 K D K/A V E/Q K G) located at the beginning of aH4, which includes K216, a residue critical for RBC binding (Figures 6 and 10). This motif is surface-exposed and located in close proximity of the RBCbinding site.
The RBC surface displays several million copies of ABO blood group determinants carried on membrane glycoproteins and glycolipids. The ABH antigens lie on terminal branches of poly-Nacetylgalactosamines, each of which may carry several ABH determinants. Although the type of branching varies, the ABH determinants displayed on the RBC surface are very dense. It is possible that PfEMP1 binding involves interaction with more than one glycan per Head region.Furthermore, although both DBL1a 1 domain and the Head region are monomeric in solution, we do not know whether binding is associated with oligomerization of the adhesion domain, as reported for other RBC-binding proteins with DBL domains such as P. falciparum EBA175 [51] and the P. vivax Duffy Binding Protein [55].
The DBL1a 1 -CIDR1c Head is present in a small subset of var genes from group A, four of which are implicated in rosetting [23,25,26]. The RBC-binding site is conserved in other rosetteforming PfEMP1 variants such as R29, PF13_003 and IT-var60, indicating that data obtained here can be extrapolated to other lines and form the molecular basis of the extensively documented ABO blood group preference in rosetting [9,[14][15][16]. The presence of CIDR1c increases binding efficiency, as indicated by the approximately 1 log unit higher MFI in flow cytometry and the increased amount of protein bound, as visualised by immunoblotting. The exact role played by CIDR1c, however, is still unclear. It is possible that its folding back upon the DBL1a 1 domain provides a structural framework for more efficient binding and increased affinity, just as the multimodular PfEMP1-VAR2CSA has been shown to require a compact fold for activity [56,57]. The binding characteristics of the Head region resemble those of the infected red cells, except for the susbtantial residual binding in the absence of human serum and the similar enhancement by human and foetal calf serum. Serum enhancement of NTS-DBL1a 1 andHead binding to RBCs may reflect a need to buffer the highly negatively charged RBC surface. As the serum component(s) implicated in VarO rosetting and Head region binding are unknown, we carried out all binding assays in the presence of human serum. VarO rosetting has an absolute requirement for human serum ( Figure  S1A) that cannot be replaced by foetal calf serum. As we show that binding of NTS-DBL1a 1 and/orHead does not account for this human-specific serum dependency, we suppose that interaction of serum components with downstream PfEMP1-VarO domains might contribute to modulate binding of PfEMP1-VarO, possibly by increasing affinity and optimising binding characteristics, or that other RBC surface proteins (eg. rifins or stevors) come into play in rosetting as well. Further work is needed to clarify this question.
This work provides the molecular basis underpinning the blood group preference of rosetting. The association between the ABO groups, rosetting and severe malaria [9] is a strong indication that rosetting, as a contributor to severe malaria, has exerted a selective pressure that has shaped population polymorphisms at the ABO locus and has contributed to their varying geographic distribution. The data reported here expand this framework to subgroups within the susceptible blood group A. Although the genetic basis of the A 1 , A 2 and other rare A subgroups is well established, the physiological consequences of such phenotypes and the selective advantage they provide are unclear. The lower prevalence of A 1 blood group in populations of African descent compared to populations of Asian or Caucasian origin [58][59][60] is consistent with the hypothesis that P. falciparum rosetting has contributed to subgroup selection and gene spread of blood group A variants.

Blood samples
For ABO groups, blood donated by healthy volunteers was purchased from the Blood Bank Centre (EFS, Rungis). Fresh or cryo-preserved and thawed RBCs of A 1 , A 2 or A x subgroup were obtained from the reference cryobank reagents of the Centre National de Référence pour les Groupes Sanguins (CNRGS -INTS, Paris). Blood samples with CR1 copy number variation were left-overs (''fond de tubes'') from healthy volunteers recruited by the URCA EA3798 (Reims, France) for clinical studies. Specific written consent was provided by each donor to use the left-overs for research. The Comité Consultatif pour la Protection des Personnes se prêtant à des Recherches Biomédicales of Champagne Ardenne approved the protocol. Supply and handling of human red cells followed the guidelines of the agreement between Institut Pasteur and the Etablissement Français du Sang.

Determination of CR1, A and H antigen levels on RBCs
The quantification of CR1 copy number was assessed using biotinylated anti-CR1/CD35 mAb J3D3, followed by a sequential labelling with streptavidin-phycoerythrin (PE), biotinylated antistreptavidin and streptavidin-PEantibodies as described [61,62].The amount of A and H antigens displayed on RBC surface was determined by flow cytometry using the following monoclonal antibodies: mAb BRIC-145/9W2 (mouse anti-A antigen, IgG1) and mAb MR3-517 (mouse anti-H antigen, IgM), respectively, and confirmed using TransClone Anti-ABO1 (IgM) and TransClone anti-H1 (IgM) murine mAbs (Bio-Rad Laboratories), respectively. Secondary goat anti-mouse IgG or IgM Alexa fluor 488-conjugated (Molecular Probes, Invitrogen) antibodies were subsequently used. To reduce agglutination of antigen-positive cells, RBC samples were fixed for 10 min at room temperature with 0.1% glutaraldehyde as described [61] before staining. For each sample, 50,000 events were collected using a BD-LSR1 flow cytometer (Becton Dickinson) and expression levels were analysed with the FlowJo 9.4.7 software.

Parasite cultures and rosetting characteristics
Monovariant cultures of the Palo Alto 89F5 VarO and It4/R29 parasites, procedures for rosette enrichment on ice-cold Ficoll, rosette dissociation with dextran sulphate, magnetic selection of mature iRBCs were as described [24,25]. Rosette reformation assays were carried out in RPMI with 10% AB human serum (RPMI-HS) 1 h at 37uC and the rosetting rate was calculated, after addition of Hoechst 33342 dye (Molecular Probes) for parasite nuclei staining, by determining the percentageof rosetteforming iRBCs present in the mature parasite population.
For experiments exploring the parameters affecting rosetting, purified VarO-iRBCs were diluted in RPMI-HS in the presence of specific mouse mAbs or isotype controls, or diluted in a range of human serum concentrations or 10% Ig-depleted human serum. Ig depletion of the serum was achieved by incubating 3 mL of RPMI-HS with 1 mL of RMPI-equilibrated protein G Sepharose beads (GE Healthcare). Depletion was assessed by Coomassie blue staining (PageBlue Protein Staining Solution, Thermo Scientific) of sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting using anti-human Ig. Enzyme treatment of recipient RBCs was carried out by incubating 5610 7 RBCs (washed twice in PBS, Phosphate Buffer Saline 16, Invitrogen) for 30 min at 37uC with 10 mg trypsin (Sigma, T1005) or a-chymotrypsin (Sigma, C4129) in a final volume of 100 mL. Enzymatic digestion was stopped by three extensive washes in RPMI-HS. Surface expression of CR1/CD35 was detected using anti-CR1 mAb J3B11, followed by an Alexa fluor 488-conjugated goat anti-mouse IgG. Treated or untreated recipient RBCs were used in rosette reformation assays with magnetically enriched mature VarO-iRBCs.
For ABO blood group preference assays, red cell membranes were labelled according to the manufacturer's instructions with the lipophilic fluorescent probes PKH67 or PKH26 (Sigma Aldrich), both of which provide stable, clear, intense and reproducible fluorescent labelling of live red cells with no apparent loss of function [62] or antigenicity [25]. VarO-iRBCs were purified to .90% parasitemia by magnetic separation and diluted in the presence of varying ratios of recipient A, B and O RBCs differentiated by labelling alternately with PKH26 or PKH67. After incubation in RPMI-HS for 1 hr at 37uC, the rosetting rate was evaluated by fluorescence microscopy.
Protein expression in Escherichia coli was carried out for 20 h at 20uC (16uC for the Head) and purified by affinitychromatography (TALON, Clontech), followed by size exclusionchromatography (S200 or S75 16/60, GeHealthcare) as described [37]. In the case of MBP fusion proteins, MBP was cleavedusing Factor Xa (Novagen) and the recombinant domain purified as described [37]. Where necessary, the hexa-His tag was cleaved by thrombin (Novagen) as recommended by the manufacturer and the cleavage monitored by Western blot. For crystallisation, the Head protein was further purified on a HiTrap Heparin column (GE Healthcare), followed by size exclusion chromatography(S20016/60, GE Healthcare) in 20 mM Tris-HCl, 200 mM NaCl, pH8.0 and on a CM ionexchange column with an NaCl gradient from 50 mM to 1 M NaCl. The protein in a final buffer (20 mM Tris-HCl, pH 8, 200 mM NaCl) was then concentrated to 10 mg.mL 21 .
The PvDBP construct used as control for some experiments encompassed residues 199-515 of the P. vivax Duffy Binding protein. The domain was cloned in baculovirus, produced in HiFive cells, and purified as described [24].

Mutagenesis
Point mutations were made using polymerase chain reactionbased mutagenesis kits (Quikchange Lightning Site-Directed Mutagenesis kit or Quikchange Multi Site-Directed Mutagenesis kit, Stratagene) following the supplier protocols. Mutants Mut1-7 have been described previously [27]. Mutants 12 and 13 were custom made by GeneCust (Luxembourg) using DBL1a 1 as template. Mutants 10 and 14-19 were custom-made by GenScript (Hong Kong) using DBL1a 1 and DBL1a 1 (wt) as template, respectively. All mutants were produced as MBP fusion proteins and purified as above.

Immune reagents and assays
Antibodies to the individual PfEMP1 domains and the Head were produced by immunising OF1 mice, five mice per antigen, (6-8 weeks old mice, Charles River, France) with 10 mg recombinant protein in Freund's Complete Adjuvant for the first injection and Freund's Incomplete Adjuvant for subsequent injections done at days 21 and 42. Serum was recovered ten days after the third injection and stored at 220uC until used. Antibody titers were determined by ELISA on the cognate antigen [24]. Immunoblot reactivity with PfEMP1 was assessed on parasite extracts separated on 4-12% gradient SDS-PAGE (Biorad) under reducing and non-reducing conditions. Reactivity with the VarO-iRBC surface was assessed by fluorescence microscopy using a Leica DM5000B Fluorescence microscope [24] and flow cytometry using a BD-LSR1 cytometer [25]. Monoclonal anti-CIDR1c antibody, mAb G8-49, was isolated from an OF1 mouse (as described inNato et al. [63]) immunised with the Head protein by screening by ELISA and VarO-iRBC surface reactivity. Monoclonal IgG was precipitated with 50% saturated ammonium sulphate from ascitic fluid, centrifuged, and dialyzed against PBS.

Binding of recombinant PfEMP1-VarO proteins to RBCs
For each binding assay, 2.5610 211 mole protein was incubated at RT for 30 min with 2610 7 RBCs in 100 mL RPMI-HS. RBCs were separated from the incubation mixture by centrifugation through 200 mL 85% silicone DC550 (Serva) 15% Nujol (Alfa Aesar) as described [27]. After one wash in RPMI, RBCs were either processed for immunoblotting [37]or for flow cytometry analyses by incubating with a specific mouse polyclonal serum followed by Alexa fluor-488-conjugated goat anti-mouse IgG F(ab9) 2 (Molecular probes) [25]. For flow cytometry analysis, the Head region was pre-incubated with 50 mg mAb G8.49 before addition of RBCs in the assay. 50,000 events were recorded using LSR1 flow-cytometer (Becton Dickinson) and analysed with FlowJo software.

Surface plasmon resonance assays
Assays were performed on a Biacore2000 instrument (GE Healthcare) in PBS buffer at 25uC. Trisaccharide A or B-conjugated BSA (6 atom spacer, average of 19 and 21 sugar residues per protein molecule for group A or B, respectively; Dextra Laboratories Ltd, UK) and unconjugated BSA (Sigma-Aldrich) were covalently coupled to the linear polycarboxylate hydrogel surface of three independent flowcells of a HLC 200 m sensorchip (Xantec bioanalytics), using the Amine coupling kit (GE Healthcare). Immobilization densities of 2500 resonance units (RU; 1RU = 1 pg.mm 22 ) were attained. A duplicate range of concentrations (10 nM-1 mM) with or without a 2-fold excess of heparin (heparin 5000; Sigma-Aldrich) of Head(wt) was then flowed at 50 mL.min 21 over the three surfaces. The association and dissociation profiles were double referenced using the Scrubber 2.0 software (BioLogic Software), i.e. both the signals from the unconjugated BSA reference surface and from blank experiments using interaction buffer instead of protein were subtracted. The shape of the profiles was suggestive of a complex binding mechanism that was not explored further in the context of this study.

Crystallisation and structure determination
The protein (10 mg.mL 21 final concentration) was mixed with heparin 5000 (Sigma-Aldrich) at a slight molar excess. Crystals were grown using the hanging-drop vapour-diffusion method: 1 mL of protein solution was mixed with 1 mL of reservoir solution (10% PEG 3350, 200 mM NaCl, 100 mM sodium citrate, pH 8.3) and equilibrated against 0.5 mL of reservoir solution. Crystals appeared after 6-8 hours and continued to grow for about one week. The crystals were passed through a cryoprotectant solution (20% glycerol, 15% PEG 3350, 200 mM NaCl, 100 mM sodium citrate, pH 8.3) and frozen in liquid nitrogen.
Data were collected on the beamline PROXIMA1 at SOLEIL (St. Aubin, France). All data were treated with XDS [64], followed by SCALA from the CCP4 program suite [65]. The structure was solved by molecular replacement using the high-resolution structure of the DBL1a 1 domain [37]Protein Data Bank (PDB entry 2xu0) as a search model. Several cycles of refinement using Buster [66] and rebuilding the structure using Coot [67]allowed to identify density belonging to the CIDR1c domain. This was fitted using Buccaneer from the CCP4 programme suite and further refined with Buster. The conserved three-helix motif from the CIDRa domain of the MC strain [28] was used to search for a similar motif in the electron density map to allow matching the sequence of the protein and to build the remaining structure. The refined coordinates and structure factors have been deposited in the Protein Data Bank (PDB entry code 2yk0). Structural figures were prepared withCCP4MG [68].

Computer docking
Computer docking of trisaccharides onto the Head structure was performed using Autodock version 4.2 [69]. Coordinates for trisaccharide A were taken from a complex in the Protein Data Bank (PDB entry code 2obs). The search was performed over the surface of the region of the DBL1a 1 moiety encompassing the NTS segment, subdomains 1 and 2, in order to allow fine grid sampling (0.375 Å ). Default values were usedfor all docking parameters, except the number of search runs, which was 100. The 10 best solutions were retained.

Analytical ultracentrifugation
The protein samples (0.5-2 mg.mL 21 ) were analysed in a Beckman Coulter XL-I analytical ultracentrifuge. Detection of the protein concentration as a function of radial position and time was performed by optical density measurements at a wavelength of 280 nm or 250 nm for high concentration samples. All samples were in a 50 mM NaCl and 20 mM Tris pH8 buffer. All experiments were carried out in an An-Ti 50 rotor at 20uC at a rotor speed of 42,000 rpm for 8 hours. Sedimentation velocity analysis was performed by continuous size distribution analysis c(s) using Sedfit 12.0 [70]. Partial specific volume 0.726 mL.g 21 and 0.738 mL.g 21 for the CIDR1c and DBL1a 1 -CIDR1c respectively, viscosity 0.01013 Poise and density 1.00093 g.mL 21 were calculated using Sednterp 1.09 and used to analyze experimental data. Sedimentation coefficients at zero concentration were obtained by linear extrapolation to zero concentration of the sedimentation measured for each protein sample at different concentrations. Sedimentation coefficients were corrected for viscosity and expressed as values in water at 20uC. Theoretical sedimentation values were calculated with the programmes HYDROPRO 7c [41] and US-Somo [40].

Circular Dichroism (Far UV)
Far-UV CD spectra were measured (Aviv215 spectropolarimeter, Aviv Biomedical)using a cylindrical cell with a 0.01 cm path length as described [37]. The spectra, corrected using buffer baselines measured under the same conditions, were normalised to the molar peptide bond concentration and path length as mean molar differential coefficient per residue.

Ethics statement
Blood donation in France includes an optional consent for use of part or all of the donation for teaching, research purposes or for preparation of specific reagents -(see documenthttp://www. dondusang.net/content/medias/media1837_FsPssqwYyfXZQcm. pdf?finalFileName = Document_dinformation_pré-don.pdf). All donors participating to this study had given their written consent for this use. Healthy volunteers donating RBCs with known CR1 copy number were recruited by the URCA EA3798 in Reims as part of an ongoing clinical study on CR1. The samples used here were left-overs (''fond de tubes''), for which specific written consent was provided by each donor for use in research on other diseases.

Statistical analysis
Differences in mean fluorescence binding of DBL1a 1 (wt) or Head(wt) between ABO RBCs or A subgroup RBCs were tested using the non-parametric Kruskall-Wallis test. The associations between recombinant protein binding and blood group A or H antigens expression levels found to be significant using the Spearman's correlation analysis were investigated by linear regression analysis. The statistical analyses were done with STATA software (STATA Corp. Release 9.0). For all tests, P values of less than 0.05 were considered statistically significant.

List of accession numbers/ID numbers for genes and proteins
VarO cDNA sequence: GenBank database accession number EU908205 High-resolution structure of the DBLa 1 domain: Protein Data Bank entry 2xu0 Refined coordinates and structure factors of Head: Protein Data Bank entry 2yk0 Supporting Information Figure S1 Rosetting characteristics of the monovariant Palo Alto 89F5 VarO parasite culture. (A) VarO Rosette formation is human serum dependent. VarO rosettes were prepared from monovariant cultures [24], dissociated with dextran sulphate and washed twice in RPMI without human serum. Rosette reformation assays were performed in RPMI medium in the presence of increasing concentration of AB human serum (1 to 10%). Rosetting rate was counted by microscopic examination after incubation at 37uC for 1 hour. Results of three independent assays. (B) VarO rosetting is IgG-and IgM-independent. Nonimmune human serum was depleted from Ig by protein-G Sepharose chromatography. Rosette reformation assays were performed as in (A). Results show the rosetting rate of VarO iRBCs cultivated under standard conditions (1) and VarO rosettes formed in unfractionated serum (2) or Ig-depleted serum (3). Results are from three independent assays. (C) VarO rosettes do not bind non-immune IgG or IgM. VarO rosettes cultivated in RPMI-10% non-immune AB human serum (NIS) were resuspended in PBS supplemented with 2% fetal calf serum and incubated with goat anti-human IgG (Molecular Probes, A11013) or IgM (Molecular Probes, A21215) Alexa fluor 488 conjugated antibodies. Hoechst dye was added in each sample to detect iRBCs (FL4 fluorescence). After an incubation of 30 min at 37uC, samples were washed twice in PBS-FCS and surface immunostaining was analysed by flow cytometry. Representative results (of at least three independent experiments) including the background labelling (no antibodies added) and a positive control labelling obtained after incubation of the rosette enriched samples with a pool of hyper-immune sera (HIS) collected from Senegalese adults [24] [37]. The recombinant domains are colour-coded as indicated. The table on the right presents the secondary structure estimations derived from the normalized spectra using the CDSSTR method included in the CDPro software [72]. Histogram representation of flow cytometry analysis of protein bound to blood group A 1 RBCs. Assays for RBC binding to DBL1a 1 (wt) and the Head(wt) were conducted in RPMI supplemented with 10% human serum in the presence or absence of 1 mg.mL 21 heparin (which disrupts VarO rosettes) [37](A) or after pre-incubation of the protein with 50 mg of the VarO rosettedisrupting mAb D15-50 [24](B). Residual protein binding was assessed by flow cytometry. (C) & (D). RBC binding to DBL1a 1 (wt) and the Head(wt) domains is potentiated by serum. Frequency histogram of flow cytometry analysis of protein bound to blood group A 1 RBCs in the absence of human serum or in the presence of 10% AB human serum (green) or 10% foetal calf serum (blue) (C). Immunoblot analysis of the bound protein of the samples incubated in the presence (+) or absence (2) of 10% human AB + serum, using mouse polyclonal antibodies to DBL1a 1 , revealed using alkaline phosphatase (D). The right panel shows an immunoblot analysis of binding of PvDBP (38 kDa), which is serum-independent. Scanning of the immunoblot (lower panel) concurred with the results of flow cytometry. (TIF) Figure S6 Relationship between RBC binding to DBL1a 1 (wt)and the Head(wt) and amount of A or H antigen displayed on the RBC.