Structural requirements for the adherence of Plasmodium falciparum-infected erythrocytes to chondroitin sulfate proteoglycans of human placenta.

Plasmodium falciparum infection during pregnancy results in the accumulation of infected red blood cells (IRBCs) in the placenta, leading to poor pregnancy outcome. In the preceding paper (Achur, R. N., Valiyaveettil, M., Alkhalil, A., Ockenhouse, C. F., and Gowda, D. C. (2000) J. Biol. Chem. 275, 40344-40356), we reported that unusually low sulfated chondroitin sulfate proteoglycans (CSPGs) in the intervillous spaces of the placenta mediate the IRBC adherence. In this study, we report the structural requirements for the adherence and the minimum chondroitin 4-sulfate (C4S) structural motif that supports IRBC adherence. Partially sulfated C4Ss with varying sulfate contents were prepared by solvolytic desulfation of a fully sulfated C4S. These and other nonmodified C4Ss, with different proportions of 4-, 6-, and nonsulfated disaccharide repeats, were analyzed for inhibition of IRBC adherence to the placental CSPG. C4Ss containing 30-50% 4-sulfated and 50-70% nonsulfated disaccharide repeats efficiently inhibited IRBC adherence; C6S had no inhibitory activity. Oligosaccharides of varying sizes were prepared by the partial depolymerization of C4Ss containing varying levels of 4-sulfation, and their ability to inhibit the IRBC adherence was studied. Oligosaccharides with six or more disaccharide repeats inhibited IRBC adherence to the same level as that of the intact C4Ss, indicating that a dodecasaccharide is the minimum structural motif required for optimal IRBC adherence. Of the C4S dodecasaccharides, only those with two or three sulfate groups per molecule showed maximum IRBC inhibition. These data define the structural requirements for the IRBC adherence to placental CSPGs with implications for the development of therapeutics for maternal malaria.

Plasmodium falciparum, the deadliest among four species of malaria parasite that infect man, has developed a mechanism for efficient survival in the host by imparting an adherence property to infected erythrocytes through the surface expression of antigenic variant proteins (1,2). Since the host develops an antibody response against these parasite proteins over a period of time, the parasite constantly switches to different adherent phenotypes by expressing proteins with different receptor specificity through the use of its extensive and diverse var gene repertoire (3)(4)(5)(6)(7)(8)(9)(10)(11). Thus, in pregnant women, P. falciparum exploits the chondroitin 4-sulfate (C4S) 1 chains in an opportunistic manner to colonize in the placenta (12)(13)(14), which leads to poor fetal outcome and severe health problems in the mother (15)(16)(17).
A number of studies have shown that endothelial cell surface adhesion molecules and C4S can function as receptors for the sequestration of IRBCs to various organs of the host (13,18). Although one or more of the endothelial cell receptors are involved in this process, C4S specifically mediates the accumulation of IRBCs in the intervillous spaces of human placenta (12)(13)(14). Although in vitro cultured endothelial cells can express CSPGs on their surfaces, and thus bind C4S-adherent IRBCs as reported (19,20,21), in vivo, endothelial cells express mainly HSPGs and they either lack or contain low levels of CSPGs. Thus, C4S-adherent IRBCs are present in very low levels in individuals other than pregnant women (22,23). As shown in the preceding paper (24), the placental intervillous spaces contain relatively high levels of CSPGs, which can efficiently bind IRBCs. Apparently, the placenta provides an opportunity by presenting CSPGs for P. falciparum IRBCs to selectively accumulate in the intervillous spaces, and thus this phenotype multiplies to high density, causing maternal malaria.
It is well known that pathogenic microorganisms use carbohydrate-protein mediated adhesion mechanisms for cell invasion and tissue colonization in the host. A number of bacteria, parasites, and viruses have been reported to use HS moieties of cell surface HSPGs as receptors for attachment and/or invasion of host cells (25,26). In most cases, the GAG chain structural requirements for adherence have not been studied, and inhibition by highly charged heparin and lack of inhibition by CS or DS have been attributed to HS chain-specific binding (25). Although a question remains as to whether those observations merely represent nonspecific charge interactions by heparin, it is increasingly becoming clear that microbes use GAG chains of host cells for adherence (25)(26)(27)(28)(29). Recently, it has been shown that a heparin dodecasaccharide is involved in Chlamydia trachomatis attachment and infectivity to the host cells (29).
Compared with the large number of microbes that appear to use HS as adhesion receptors, the evidence for pathogens exploiting CS or DS chain-specific binding is scarce (25). Among the microbes that have been studied so far, only two, a subset of herpes simplex virus (25) and erythrocytic stage P. falciparum (12)(13)(14)24), are known to use CS chains of CSPG for adherence and tissue colonization, respectively. Of these, P. falciparum is the only protozoan microbe for which specific CS chain adherence has been established (12-14, 19 -23, 30).
PGs are macromolecules formed by the attachment of GAG chains to proteins, and they are ubiquitously present in eukaryotic cells and tissues, mainly as extracellular or cell surface molecules. PGs play a key role in several biological functions, including the normal physiology of cartilage tissues, and in the regulation of biological processes such as cell migration and proliferation, cell-cell recognition and adhesion, extracellular matrix organization, cell-matrix and substratum adhesion, and morphogenesis (31)(32)(33)(34)(35)(36). PGs are also involved in mobilizing growth factors, enzymes, cytokines, and protease inhibitors at the sites where they are secreted, to create reservoirs for efficient biological interactions and to protect proteins from proteolytic degradation (33)(34)(35)(36)(37)(38). Many of the biological functions of PGs are mediated by the interactions of specific structural motifs of GAG chains (31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45), e.g. the binding of distinct heparin domains by type IV collagen (45), a specific heparin pentasaccharide by antithrombin III (41), DS hexasaccharide by heparin cofactor II (46), and heparin dodecasaccharide by chlamydial protein (29), HS dodecasaccharide by tumor-derived angiomodulin (47), and as shown in this study, a partially sulfated C4S dodecasaccharide by intraerythrocytic P. falciparum. The characterization of the fine structural requirements of GAG for ligand binding in microbial interactions with the host has implications for understanding these basic biological processes as well as for pharmacological applications.
GAG chains contribute predominantly to the characteristic properties and functions of PGs. Structurally, GAGs share a common feature in that all have a linear, O-sulfated polysaccharide backbone consisting of alternating residues of uronic acid (GlcA in CS, IdoUA in DS, and IdoUA and GlcA in HS and heparin) and hexosamine (GalNAc in CS and DS, and GlcNAc or GlcNSO 3 2Ϫ in HS and heparin) (31,32). Thus, C4S and C6S are made up of -GlcA␤1-3GalNAc␤1-4 disaccharide repeats with sulfate groups at O-4 or O-6 of GalNAc, respectively (31). GAGs in general are extremely heterogeneous with respect to their size, as well as levels, positions, and distribution of sulfate groups. C4Ss from different sources vary in distribution of nonsulfated and 4-sulfated disaccharide repeats, and some may also contain significant amount of 6-, 2,6-, and 4,6-disulfated units (31). Therefore, detailed studies using defined CS structures could furnish a complete understanding of the molecular interactions involved between C4S and the P. falciparum protein-mediated placental IRBC adherence.
Although the results of several studies suggest that C4S mediates IRBC adherence to the placenta (12)(13)(14), the precise nature of the CS chains and the structural requirements for the adherence of IRBCs to the placental CSPGs remain unclear. Previous studies have reported contradictory information regarding the minimum CS chain length requirement for IRBC binding (48 -50). This could be due to the use of nonrelevant receptors in those studies. In the preceding paper (24), we showed that the placental intervillous spaces contain high levels of CSPGs, and that these are the receptors for placental IRBC adherence. In this study, we established the IRBC binding specificity to the C4S chains of the placental CSPGs, determined the minimum CS chain motif involved, and defined structural requirements for IRBC adherence. Polystyrene Petri dishes (Falcon 1058) were from Becton-Dickinson Labware. Dowex 50W-X8, Bio-Gel P-2, and Bio-Gel P-6 were from Bio-Rad.
Selection of C4S Adherent IRBCs-The C4S-adherent parasites (3D7 clone) were originally selected from the NF-54 laboratory strains by the panning procedure (52). Tissue culture plastic Petri dishes were coated overnight with a sterile 10 g/ml solution of bovine trachea C4S in PBS, pH 7.2, at room temperature in a tissue culture hood, and then blocked with a 2% sterile solution of BSA in PBS for 1 h. Synchronized cultures of the parasites with 10 -20% parasitemia at the late trophozoite or mid-schizont stage were harvested, washed two times with RPMI 1640 medium without serum, suspended at 2% hematocrit, layered on C4Scoated Petri dishes, and incubated at room temperature. After 1 h, unbound RBCs were aspirated, plates were washed gently three times with RPMI 1640 medium, and the bound cells were detached by vigorous pipetting several times with the culture medium. These parasites were cultured, the panning procedure repeated two more times, and the final isolate designated as the 3D7 clone. The selected IRBCs adhered at high density to CSPGs of placental intervillous spaces. The adherent property of IRBCs significantly decreases over a period of continuous culture; therefore, the parasites were panned every 6 -8 weeks on placental CSPG-coated plates.
Enrichment of IRBCs-The IRBCs were enriched by gelatin floatation as described by Jensen (53). Briefly, the parasite cultures with 10 -20% parasitemia were harvested at the late trophozoite stage. The cells were suspended in 0.65% gelatin in PBS, pH 7.2, and incubated at 37°C for 20 min. Most uninfected erythrocytes and infected erythrocytes containing rings and early trophozoites settled to the bottom, and the IRBCs with mid-stage and late trophozoites remained in the supernatant. The supernatant was removed, centrifuged, and the enriched IRBCs (50 -70% parasitemia) were washed twice with sterile PBS, pH 7.2. A 2% suspension of enriched IRBCs was used for the cytoadherence assays. In some experiments, parasite cultures with 20 -25% parasitemia were used without further enrichment.
IRBC Adherence Assay-The solutions (10 -15 l) of purified CSPGs in PBS, pH 7.2, were spotted on 150 ϫ 15-mm plastic Petri dishes and allowed to coat overnight at 4°C (52). The spots were then blocked with 20 l of 2% BSA for 2 h at room temperature. The spots coated only with PBS (controls) were similarly blocked with BSA. After aspirating BSA, 15 l of a 2% suspension of enriched parasite culture (50 -70% parasitemia) in PBS, pH 7.2, was overlaid on each spot and incubated at room temperature. Uninfected RBCs layered on CSPG-coated plates were used as separate negative controls. After 40 min, the dishes were washed three times with PBS, pH 7.2. The bound cells were fixed with 2% glutaraldehyde and stained with either 1% Giemsa or SYBR green fluorescent dye at 1:10,000 dilution of the stock solution as recommended by the supplier. The bound IRBCs were counted under light or fluorescent microscopy and photographed. All assays were carried out either in duplicate or triplicate.
Enzyme Treatments in Adhesion Assays-The CSPG was coated as circular spots on plastic Petri dishes (3.5 ϫ 1 cm) as described above, and the entire inside surface was blocked with BSA. The plates were then incubated for 2 h with 1 ml of chondroitinase ABC (50 milliunits/ ml) or testicular hyaluronidase (50 units/ml) at 37°C, or heparitinase (20 milliunits/ml) at 43°C in buffers containing protease inhibitors as described (54 -56). The CSPG-coated plates were also treated with S. hyalurolyticus hyaluronidase (40 units/ml) at 60°C for 2 h (57). The plates were washed three times with PBS, pH 7.2, and IRBC adhesion was assayed as described above.
Cytoadherence Inhibition Assay-The solutions of oligosaccharides or polysaccharides at twice the indicated concentrations in PBS, pH 7.2, were mixed with equal volumes of 4% suspension of IRBCs in PBS, pH 7.2, in 96-well microtiter plates. The suspensions were incubated at room temperature for 30 min with intermittent mixing, and then layered on CSPG-coated spots on Petri dishes. After 40 min at room temperature, the unbound cells were washed and the bound cells were fixed with 2% glutaraldehyde, stained with either Giemsa reagent or SYBR green, and measured as outlined above.
Polyacrylamide Gel Electrophoresis of C4S Oligosaccharides-Aliquots containing 3-4 g of CS oligosaccharides obtained by Bio-Gel P-6 chromatography or 30 -50 g of total digests were dissolved in 100 mM Tris base, 100 mM boric acid, 2 mM EDTA, pH 8.3, containing 2 M sucrose; 0.2% bromphenol blue in the above buffer was used as tracking dye. The solutions were electrophoresed on 1.5-mm-thick 10% polyacrylamide gels (14 ϫ 23.5 cm) in 100 mM Tris base, 100 mM boric acid, 2 mM EDTA, pH 8.3 (55). The gels were stained with 0.3% Alcian blue in 3% aqueous acetic acid containing 50 mM MgCl 2 for 4 h, and destained with the same solution (59). The gels were further stained with ammoniacal silver (60). Briefly, gels were treated with 10% glutaraldehyde for 30 min, washed three times with water for 30 min. The gels were treated with ammoniacal silver, washed two times with water for 30 s, and then developed with 0.005% citric acid and 0.019% formaldehyde and washed with water.
Solvolytic Desulfation of Glycosaminoglycans-The sodium salt of sturgeon notochord C4S (20 mg) was converted into free acid using Dowex 50W-X8 (H ϩ ), neutralized with pyridine, and lyophilized (61). The pyridinium salt of C4S was dissolved in 10 ml of 10% aqueous Me 2 SO, divided into 10 equal parts, and placed in separate screwcapped glass vials. The vials were heated at 80°C, and each was removed at 5, 15, 30, 45, 60, 75, 90, 105, 120, and 140 min and cooled in an ice bath. The reaction mixture in each vial was diluted to 2 ml with water, the pH adjusted to 9.0 with 0.1 M NaOH, dialyzed (molecular weight cut-off 3,500) against water, and lyophilized. Experiments were repeated several times with slightly altered time periods to obtain C4Ss with sulfate contents ranging from 3% to 89% (see Table II).
Regioselective 6-O-Desulfation of GAGs-C4Ss from bovine trachea and whale cartilage (100 mg each) were separately desalted on Dowex 50W-X8 (H ϩ ), converted into the pyridinium salts, lyophilized, and dissolved in dry pyridine (20 ml). N,O-Bis(trimethylsilyl)acetamide (4 ml) was added to a final concentration of 20% and heated in a screwcapped glass tube at 80°C for 4 h (62). The reaction mixtures were cooled in an ice bath, and ice-cold water (25 ml) was added to decompose excess sialylating reagent, dialyzed against water, and lyophilized.
Disaccharide Composition Analysis of GAG Chains-The disaccharides released by the chondroitinase ABC digestion of CS or DS were analyzed on a 4.6 ϫ 250-mm amine-bonded silica PA03 column (YMC Inc., Milford, MA) using 600E HPLC system (Waters, Milford, MA) as described (63). The enzyme digests corresponding to 10 -15 g of GAGs were injected and eluted with a linear gradient of 16 -530 mM NaH 2 PO 4 over 70 min at a flow rate of 1 ml/min at room temperature. The elution of disaccharides was monitored by measuring the absorption at 232 nm using a Waters 484 variable wavelength UV detector. The data were processed with the Millennium 2010 chromatography manager using NEC PowerMate 433 data processing system. (24), we reported that the unusually low sulfated CSPGs localized in the intervillous spaces of human placenta mediate the in vivo adherence of IRBCs. In this paper, we present the data that establishes the specificity of IRBC adherence to the placental CSPGs and define structural requirements for the adherence. In an in vitro cytoadherence assay, when a mixture of infected and uninfected RBCs overlaid on CSPG-coated plastic plates, only IRBCs but not RBCs bound to the CSPGs of the placental intervillous spaces (data not shown). The adherence of IRBCs was completely abolished upon prior treatment of the CSPGcoated plates with chondroitinase ABC or testicular hyaluronidase; the binding was not affected when the plates were treated with heparitinase or S. hyalurolyticus hyaluronidase (data not shown). These results clearly demonstrate that CS chains of the placental CSPGs mediate the IRBC adherence. The structural requirement for IRBC adherence to the placental CSPGs was studied by the inhibition of cytoadherence using CSs from different sources, which contain variable structural features (see below). In all the studies described here, highly purified BCSPG-2 fraction (24) was used for the adherence assay.

Specificity of P. falciparum-IRBC Adherence to the CS Chains of Placental CSPGs-In the preceding paper
Information on the level of 4-sulfation required for the optimal IRBC binding and the effect of sulfate groups at other positions within C4S chains is lacking. Because the CS chains of the CSPGs of the placental intervillous spaces, BCSPG-1 and BCSPG-2, contain only ϳ2% and ϳ8% 4-sulfated groups, respectively (24), it is possible that the IRBC adherence could be mediated by one or more factors such as unique distribution of sulfate groups, unusual GAG structural features within the chondroitin chain, or nonspecific charge interactions. To examine these possibilities, a variety of GAGs including completely nonsulfated chondroitin, HA, fully sulfated C4S, chondroitin sulfates with varying amounts of 4-, 6-, and nonsulfated disaccharide repeats, CS chains that contain significant levels of 2,6and 4,6-disulfated disaccharide repeats (C2,6diS and C4,6diS), DS, and heparin were analyzed for their compositions and ability to inhibit IRBC adherence (Table I, Fig. 1). Whereas all variously 4-sulfated CSs inhibited IRBC adherence, albeit to different degrees (see below), nonsulfated chondroitin, HA, C6S, C2,6diS, and heparin were completely non-inhibitory ( Fig.  1). Highly sulfated dextran sulfate, HS, and pentosan polysulfate were also completely non-inhibitory (data not shown). These data indicated that the IRBC adherence was not because of nonspecific interactions with the strongly charged anionic sulfate groups but is mediated by a specific 4-sulfated chondroitin structure. Lack of inhibition by HA and HS suggest that the activity was not due to low levels of other GAG structural features within C4S chains. The inhibitory activity exhibited by DS from porcine intestinal mucosa was comparable to that of the fully 4-sulfated CS from sturgeon notochord, and it is likely because of the presence of significant amounts of C4S structural features in the DS chains. The complete inactivity of C2,6diS in contrast to the efficient inhibition by C4,6diS was interesting considering that both GAGs contain similar levels of 4-sulfated disaccharide units. A comparison of the structural differences between these two GAGs (Table I) suggests that C-2 hydroxyl of GlcA should be free for interaction with IRBCs. The efficient inhibition of IRBC adherence to BCSPG-2 by C4,6diS demonstrates that the presence of high levels of 6-sulfate groups in addition to 4-sulfate groups has no significant effect on IRBC binding.
Inhibition of IRBC Adherence to Placental CSPG by C4Ss with Varying Sulfation Pattern-Many CSs that are classified as C4S contain variable amounts of 6-sulfate groups, and they differ significantly with respect to the levels of 4-sulfate groups, and thus they are likely to differ in the distribution of sulfate groups within their repeating disaccharide units (see Table I).
To investigate how these varying structural elements within C4S chain contribute to IRBC adherence, three distinctly different C4Ss, almost fully 4-sulfated C4S from sturgeon notochord, C4S from whale cartilage with 69%, 27%, and 4% of the disaccharide repeats, respectively, 4-, 6-, and nonsulfated, and C4S from bovine trachea with 53%, 39% and 8%, respectively, 4-, 6-, and nonsulfated, were studied for their ability to inhibit IRBC binding to the placental CSPG (Fig. 1). Although the C4S chains from all the three sources inhibited IRBC binding in a dose-dependent manner, contrary to the expected results, the completely 4-sulfated C4S was about 2.5 times less efficient in inhibiting IRBC binding compared with the C4Ss from the whale cartilage and bovine trachea (Fig. 1). This is despite the presence of significant amounts of 6-sulfation in whale cartilage and bovine trachea C4Ss. These results suggest that the C4S structures containing both 4-sulfated and 4-nonsulfated disaccharide repeats support efficient IRBC binding. C6S was completely non-inhibitory even at Ͼ100 g/ml, suggesting that 6-O-sulfation does not contribute to IRBC adherence. This conclusion agrees with the unaltered inhibitory activities of C4Ss from bovine trachea and whale cartilage after the regioselective removal of sulfate groups at C-6 (see below).

Inhibition of IRBC Adherence to Placental CSPG by Partially
Desulfated CS-To study further the requirements of 4-sulfated and 4-nonsulfated disaccharide repeats within the CS chains for effective interaction with IRBCs, we prepared CSs containing various levels of 4-sulfate groups, by solvolytic desulfation of the sturgeon notochord C4S or regioselective 6-Odesulfation of whale cartilage and bovine trachea C4Ss (Table  II). The inhibitory capacities of these partially desulfated C4Ss were assessed ( Fig. 2 and data not shown). Although all C4Ss competitively inhibited IRBC adherence to the placental CSPG, they differed significantly in their inhibitory activity. At similar concentrations, the C4Ss containing 30 -80% sulfated disaccharides and 20 -70% nonsulfated disaccharides had the higher inhibitory capacity compared with the fully 4-sulfated sturgeon notochord C4S (Fig. 2). The inhibitory ability of C4Ss that contain 70% and 30%, 4-sulfated and nonsulfated disaccharide repeats, respectively, prepared from sturgeon notochord C4S was comparable to that of the untreated whale cartilage C4S (compare Fig. 1 with Fig. 2). Similarly, C4S from bovine trachea showed a comparable level of inhibition with that of the partially desulfated sturgeon notochord C4S with 52% 4-sulfated disaccharides (compare Fig. 1 with Fig. 2). Furthermore, the inhibitory capacities of these partially 4-sulfated CSs, BTC4S-46 and WCC4S-65, obtained from regioselective 6-O-desulfation of bovine trachea and whale cartilage C4Ss, respectively (Table II and Fig. 2), were comparable to that of partially desulfated SNC4S with similar 4-sulfate contents (not shown). These results confirm that sulfate groups at 6-O position of GalNAc of whale cartilage and bovine trachea C4S chains neither interact nor interfere with the IRBC binding.
Minimum C4S Structural Motif Involved in IRBC Adherence to Placental CSPG-To determine the minimum structural motif of the C4S chain involved in the IRBC adherence, we prepared oligosaccharides of varying sizes by partial enzymatic depolymerization of several C4Ss with varying levels of 4-sulfation. In preliminary experiments, whale cartilage C4S was partially depolymerized by treating with different amounts of ovine testicular hyaluronidase for varying time periods (64). The enzyme degraded ϳ40 -50-kDa C4S (Ͼ100 disaccharide repeating units) into a mixture of low molecular mass oligosac-  charides. Based on the results of these pilot experiments (64), optimal conditions for the conversion of polysaccharides to maximum levels of oligosaccharides with 2-20 repeating units were selected. Large quantities of sturgeon notochord, bovine trachea, and whale cartilage C4Ss were depolymerized, and the oligosaccharides were size-fractionated on Bio-Gel P-6 columns. Each column fraction was analyzed by polyacrylamide gel electrophoresis (Fig. 3), and oligosaccharides that are homogeneous in size containing 2-7 disaccharide repeats as well as non-homogeneous higher oligomers were analyzed for their ability to inhibit IRBC adherence to the placental CSPG ( Fig. 4 and data not shown). The oligosaccharides with 3 or more disaccharide repeats, irrespective of the C4S used for their preparation, were able to inhibit IRBC adherence in a dose dependent manner. At comparable concentrations, inhibition by C4S oligosaccharides was proportional to their chain lengths. Among various sized oligosaccharides obtained from different C4Ss, only the dodecasaccharides and higher oligomers (with six or more disaccharide repeats) inhibited IRBC adherence to the extents comparable to that of the correspond-ing intact C4S polymers at all concentrations tested including that required for complete inhibition (Fig. 4). In the case of bovine and whale C4S, the decasaccharides exhibited about 75-85% activity of the intact polymer in a dose-dependent manner, whereas the decasaccharides from sturgeon notochord showed 50 -60% of activity compared with the corresponding intact C4S. The octa-and hexasaccharides showed significantly lower inhibitory activities, and the tetrasaccharides showed low level of activity only at high concentrations (Ͼ100 g/ml). As in the case of intact GAGs, the oligosaccharides obtained from fully 4-sulfated C4S from sturgeon notochord were 2-3fold less efficient compared with those from whale cartilage and bovine trachea C4S (Fig. 4). A comparison of the inhibition of IRBC binding to the placental CSPG by dodecasaccharides prepared from different C4Ss with the respective intact polysaccharides (Fig. 4) clearly indicates that, in each case, the dodecasaccharide inhibited the IRBC adherence to the same level as that of the corresponding polysaccharide. Thus, the minimum CS chain motif required for efficient interaction of IRBCs is a dodecasaccharide.
Inhibition of IRBC Adherence to Placental CSPG by Differentially 4-Sulfated C4S Dodecasaccharides-To determine the extent of 4-sulfation required for efficient interaction of the CS dodecasaccharide motif with IRBCs, we partially depolymerized C4Ss containing various levels of 4-sulfate groups using testicular hyaluronidase. From each enzymatic digest, dodecasaccharides were isolated by gel filtration on Bio-Gel P-6 column. The disaccharide composition of these oligosaccharides, determined by HPLC after digestion with chondroitinase ABC (Table III), were generally comparable to those present in corresponding C4Ss. The dodecasaccharides were tested for their ability to inhibit IRBC binding to the placental CSPG (Fig. 5). Oligosaccharides with two or three sulfate groups per molecule inhibited IRBC adherence to the placental CSPG to significantly higher level when compared with those with either more than three or less than two sulfate groups. These results together with the chain length requirement for binding  2. Inhibition of IRBC adherence to placental CSPG by partially desulfated C4S. The inhibition of IRBC adhesion to BC-SPG-2 by the partially desulfated C4Ss was performed as described in legend to Fig. 1. q, BTC4S-46; E, SNC4S-3; Ⅲ, SNC4S-11; ϫ, SNC4S-30; छ, SNC4S-38; Ⅺ, SNC4S-52; f, SNC4S-60; ‚, SNC4S-78; OE, SNC4S-89; ϩ, SNC4S-98. SNC4S, BTC4S, and WCC4S refer to C4Ss from sturgeon notochord, bovine trachea, and whale cartilage, respectively. The numbers that follow the hyphen refer to the percentage of 4-sulfate contents. The inhibition of IRBC adherence by SNC4S-8, SNC4S-18, and WCC4S-65 (see Table II) was also assessed.
FIG. 3. Polyacrylamide gel electrophoresis of C4S oligosaccharides from Bio-Gel P-6 column fractions. C4S from bovine trachea was digested with ovine testicular hyaluronidase and chromatographed on Bio-Gel P-6 column (1.5 ϫ 70 cm) as described under "Experimental Procedures." Fractions were lyophilized and dissolved in water, and aliquots corresponding to 3-4 g of oligosaccharide were electrophoresed on a 1.5-mm-thick 10% polyacrylamide gel (14 ϫ 23.5 cm), stained successively with Alcian Blue and ammoniacal silver, and photographed. The Bio-Gel P-6 column fraction numbers are indicated at the bottom, and the oligosaccharide size (number of disaccharide repeats) are indicated to the left. Oligosaccharides from C4Ss from whale cartilage and sturgeon notochord were similarly prepared.
indicate that the minimum CS structural motif for the adherence of IRBCs by C4S chains is a dodecasaccharide with two or three 4-sulfate groups. DISCUSSION In the preceding paper (24), we reported the detailed structural characterization of various CSPGs purified from different regions of human placenta, and identified the CSPGs responsible for the accumulation of P. falciparum-IRBCs in the intervillous spaces of the placenta. In this study, we demonstrate that the binding of IRBCs to the placental CSPGs is CS chain-specific, and using these CSPGs (natural receptors) in a cytoadherence assay, we defined the key structural elements and minimum CS chain length required for effective IRBC binding. The important findings are as follows. 1) The C4S chains, in which all the disaccharide repeats are 4-sulfated, were significantly less efficient in binding IRBCs compared with C4S containing both sulfated and nonsulfated disaccharide units. 2) C4S chains with 1:1 to 1:2 ratios of 4-sulfated and nonsulfated disaccharide repeats effectively bind IRBCs. 3) Significant levels (up to 40%) of 6-sulfation within C4S have no detectable effect on the adherence capacity of the C4S chains. 4) The minimum C4S chain length required for optimal IRBC binding is a dodecasaccharide with two or three sulfate groups.
The results presented in this paper indicate that IRBC adherence to the CSPGs of placental intervillous spaces is mediated by low sulfated CS chains with a stringent specificity for 4-sulfation. The data also indicate that IRBC adherence is not due to either the presence of low levels of other GAG structural features in the CSPGs or nonspecific anionic charge interactions by the sulfate groups. Several lines of evidence support these conclusions. First, the binding of IRBCs to the placental CSPG is completely abolished when the coated plates were treated with CS-degrading enzymes but not when treated with DS-or HS-degrading enzymes; the enzyme that specifically degrade HA also has no effect. Second, only C4S but not C6S, the closely related GAG that differs only in the position of sulfate substitution, can inhibit IRBC adherence. Completely nonsulfated chondroitin chains, hyaluronic acid, and highly sulfated GAGs such as heparin, C2,6diS, dextran sulfate, and pentosan polysulfate were all non-inhibitory. Thus, these data are consistent with the results of previous studies, which reported that the adherence of IRBCs to the placenta is mediated specifically by C4S (12-14).
Although previous studies have shown that the adherence of IRBCs to human placenta is mediated by C4S, the details of the structural requirements, particularly with respect to the levels and the distribution of sulfate groups within the CS chains have not been reported. The data presented in this paper show that fully sulfated C4S from sturgeon notochord was markedly less active in inhibiting IRBC adherence to the placental CSPG compared with C4S containing significant levels of nonsulfated disaccharide repeats. Of the several CSs with varying levels of 4-sulfated disaccharide repeats tested, those in which 30 -50% of the disaccharide repeats with 4-sulfated and the remainder FIG. 4. Inhibition of IRBC binding to the placental CSPG by C4S oligosaccharides. The inhibition of IRBC adhesion to BCSPG-2 by the oligosaccharides purified on Bio-Gel P-6 column was performed as described in Fig. 1. A-C, oligosaccharides from bovine trachea C4S, sturgeon notochord C4S, and whale cartilage C4S, respectively. q, tetrasaccharide; f, hexasaccharide; OE, octasaccharide; E, decasaccharide; Ⅺ, dodecasaccharide; ‚, tetradecasaccharide; ϫ, polysaccharide.

TABLE III Disaccharide composition of partially desulfated dodecasaccharides
The dodecasaccharides (DDSs) with varying 4-sulfate contents were prepared by partial depolymerization of the variously 4-sulfated C4Ss from sturgeon notochord (see Table II) with testicular hyaluronidase and Bio-Gel P-6 chromatography as described under "Experimental Procedures." The DDSs were digested with chondroitinase ABC and the disaccharides formed were analyzed by HPLC.  Table III) prepared from partially desulfated sturgeon notochord C4Ss was carried out at 5 g/ml as described in Fig. 1. The numbers on the x axis indicate the percentage sulfate content of the dodecasaccharides. 4-nonsulfated showed maximum inhibition. C4S containing 18% of 4-sulfated disaccharide repeats had the similar levels of inhibitory capacity as those exhibited by C4S with 60% sulfate content, and those with 3% and 8% 4-sulfated disaccharide repeats showed only marginal inhibition. This is surprising considering the low levels of sulfation in the CS chains of placental CSPGs. One explanation for these results is that the 4-sulfated disaccharide repeats in the CS chains of placental CSPGs are clustered such that some locations within the CS chains contain optimal distributions of 4-sulfated and 4-nonsulfated disaccharide repeats. Alternatively, it is possible that the structural requirement for IRBC binding with respect to the level of 4-sulfation is different from that required for effective inhibition.
Two naturally occurring C4Ss, one from bovine trachea and the other from whale cartilage, which are 53% and 69%, respectively, 4-sulfated with the major portions of the remainder 6-sulfated efficiently inhibited IRBC adherence to the placental CSPG. Since the CS chains of placental CSPGs lack 6-sulfation, these results together with the superior inhibitory ability of C4S from bovine trachea suggest that the primary hydroxyl group of the GalNAc is not interacting significantly with IR-BCs. Consistent with this conclusion, the inhibitory abilities of C4Ss from bovine trachea and whale cartilage were unaltered after the selective removal of the majority of the 6-sulfate groups by regiospecific 6-O-desulfation. This conclusion is further supported by the observation that a highly sulfated C4,6diS containing 60%, 21%, 10%, and 9%, respectively, 4,6-di-, 4-, 6-sulfated, and nonsulfated disaccharide repeats showed ϳ2-fold better inhibitory capacity compared with the fully 4-sulfated CS.
This study conclusively establishes that the minimum CS chain length required for effective inhibition of IRBC binding to the placental CSPG, at levels similar to those exhibited by the intact C4S polysaccharides, is a dodecasaccharide. Among the series of oligosaccharides of varying sizes prepared by the partial enzymatic digestion of three different naturally occurring C4Ss (from sturgeon notochord, bovine trachea, and whale cartilage) containing different levels of 4-sulfation, in all cases, oligosaccharides with more than three disaccharide repeats inhibited IRBC binding in a dose-dependent manner. The level of inhibition was directly proportional to the oligosaccharide size. In each case, only dodecasaccharide or higher oligomers inhibited IRBC binding to the similar levels as those of the corresponding intact C4Ss at comparable concentrations. As in the case of intact C4Ss, inhibition by dodecasaccharide, in which all six disaccharide repeats are 4-sulfated, was markedly less compared with the dodecasaccharides containing both 4-sulfated and 4-nonsulfated disaccharides. Of the variously 4-sulfated dodecasaccharides studied, oligosaccharides containing two or three sulfate groups maximally inhibited IRBC binding compared with the same size oligosaccharides with less than two or more than three sulfate groups. Since oligosaccharides studied are likely to contain a mixed population with varying, albeit in a narrow range, number of sulfate groups, it is not possible to define the exact number and the precise distribution within the six disaccharide repeats. Because, naturally occurring GAGs are highly heterogeneous with respect to the distribution of sulfate groups, the fractionation of oligosaccharides of this size, based on the positions of sulfate distribution, is hard to achieve. However, chemically synthesized oligosaccharides with sulfate groups at defined positions should identify the precise structural requirements for the maximal inhibition of IRBC adherence.
Previously, four different studies including the one of our own reported contradictory information on the minimum CS chain length requirement for IRBC adherence (48 -50, 64). Using recombinant thrombomodulin in an in vitro cytoadher-ence assay, Beeson et al. (48) assessed the inhibition by oligosaccharide fractions with 1-10 disaccharide repeats prepared from bovine trachea C4S, and found that the minimum structural motif that supports the IRBC adherence was a tetradecasaccharide. In that study, inhibition by dodecasaccharides and decasaccharides were, respectively, ϳ60% and ϳ30% of that of the intact C4S. A second study assayed C4S oligosaccharides of molecular mass ranging from 1 to 9 kDa for inhibition of IRBC binding to in vitro cultured Saimiri brain microvascular endothelial cells (49). In this study, only oligosaccharides with size Ͼ4 kDa (with 8 or 9 disaccharide repeats) showed noticeable inhibitory activity and a 9-kDa polysaccharide (ϳ18 or 19 disaccharide repeats) was required for inhibition equivalent to that by intact C4S. These two studies used chondroitinase ABC, a lyase enzyme that forms an unsaturated uronic acid at the nonreducing end for the preparation of CS oligosaccharides. This might have rendered the nonreducing end of the disaccharide repeat unable to interact with IRBCs because of altered structural feature. In addition, the above studies used nonrelevant receptors for the assessment of oligosaccharide inhibitory activity. In a preliminary study using mixtures from partial hydrolysates of a C4S (64), we reported that oligosaccharides with Ͼ7 disaccharide repeats effectively inhibit IRBC adhesion. However, this study, using C4S oligosaccharides that are homogeneous in size with two to seven disaccharide repeats and relevant CSPG, unequivocally establishes that a dodecasaccharide is the minimum chain length required for optimal IRBC adherence.
The minimum C4S chain length of six disaccharide repeats required for efficient interactions with IRBCs appears to be too large for a binding site of a protein. Usually, oligosaccharide-or polysaccharide-binding clefts for proteins such as enzymes, antibodies, and lectins contain 1-4 sugar residues, if nonconformational determinants are involved. Therefore, the involvement of a long saccharide segment in binding of IRBCs by partially 4-sulfated CS strongly suggests the requirement of either a specific structural sequence (arrangements of sulfated and nonsulfated disaccharide repeats) for the binding event or a conformational epitope. Since our results demonstrate that both 4-sulfated and nonsulfated disaccharide repeats in the ratios of 1:1 to 1:2 are required for efficient interaction with IRBCs, it appears that a specific distribution pattern of the sulfate groups within the CS chain is involved in the binding of IRBCs. Although various sulfate distribution can also be accommodated in oligosaccharides with four or five disaccharide repeats, the requirement of a minimum CS chain length of six disaccharide repeats for maximal activity suggests the involvement of a specific conformational structure as well. It is likely that the required conformation is attainable only with a CS chain length of six disaccharide repeats having a specific distribution of 4-sulfated groups. The involvement of conformational epitopes formed from extended oligosaccharide segments has been shown in binding of antibodies or cell adhesion molecules to polysaccharides. For example, antibodies against meningococcal group B ␣-2,8-linked sialic acid polysaccharide recognize only oligosaccharides with Ͼ10 sugar residues, which can assume the required helical conformation (65,66). Similarly, antibodies against type 14 pneumococcal capsular polysaccharide and type B Haemophilus influenzae polysaccharide bind conformational epitopes in extended oligosaccharide segments (67); binding of neuronal cell adhesion molecule to polysialic acid also has been suggested to involve conformational determinants (68).
Alternatively, it is possible that the adherence of IRBCs to the placental CSPG may involve simultaneous interactions by two or more binding sites of parasite protein(s) with closely related specificity, each interacting with a partially sulfated trisaccharide within the contiguous CS chain. However, we believe this is less likely. In either case, it appears that the limited number of sulfate residues present in the CS chains of placental CSPG are closely spaced within the large 60-kDa CS chains so as to provide specific structural requirements for the adherence of IRBCs.
The C4S dodecasaccharide chain length required for maximal inhibition of IRBC adhesion to the placental CSPG parallels the heparin dodecasaccharide requirement for the effective binding and infectivity of C. trachomatis to the host cells (29), and HS dodecasaccharide binding to tumor-derived angiomodulin (47). The identical GAG chain length requirement by proteins expressed by three entirely different organisms despite that they use entirely different types of GAG chains for attachment, low sulfated CS chains by P. falciparum (24), highly sulfated heparin by C. trachomatis (29), and moderately sulfated HS chains by human carcinoma cellderived angiomodulin is surprising (47). Although it is premature to make a generalization based on these three known cases out of a large number of microbial as well as animal protein-GAG interactions, it is tempting to speculate that the structural motifs of the GAG-binding proteins of various organisms might be evolutionarily conserved. Furthermore, it is possible that the binding motifs that are conserved among proteins of various organisms may recognize a conformational epitope within the GAG chains, and a minimum of 12 sugar residues are required for a conformational structure, irrespective of the GAG type.
In this study, oligosaccharides prepared from C4Ss from other animal sources rather than those from the placental CSPGs were used for the cytoadherence inhibition assay. The data show that dodecasaccharides from these C4Ss were able to inhibit the adherence up to 90% at Ͼ20 g/ml concentrations. However, in the case of human placental CSPGs, the CS chain motif that binds IRBCs may have an unique sulfate group distribution pattern. The dodecasaccharides obtained from those regions of the CS chains may interact optimally with the parasite ligand to completely inhibit IRBC binding even at lower concentrations.