The Low Sulfated Chondroitin Sulfate Proteoglycans of Human Placenta Have Sulfate Group-clustered Domains That Can Efficiently Bind Plasmodium falciparum-infected Erythrocytes*

Plasmodium falciparum infection in pregnant women results in the chondroitin 4-sulfate-mediated adherence of the parasite-infected red blood cells (IRBCs) in the placenta, adversely affecting the health of the fetus and mother. We have previously shown that unusually low sulfated chondroitin sulfate proteoglycans (CSPGs) in the intervillous spaces of the placenta are the receptors for IRBC adhesion, which involves a chondroitin 4-sulfate motif consisting of six disaccharide moieties with ∼30% 4-sulfated residues. However, it was puzzling how the placental CSPGs, which have only ∼8% of the disaccharide 4-sulfated, could efficiently bind IRBCs. Thus, we undertook to determine the precise structural features of the CS chains of placental CSPGs that interact with IRBCs. We show that the placental CSPGs are a mixture of two major populations, which are similar by all criteria except differing in their sulfate contents; 2–3% and 9–14% of the disaccharide units of the CS chains are 4-sulfated, and the remainder are nonsulfated. The majority of the sulfate groups in the CSPGs are clustered in CS chain domains consisting of 6–14 repeating disaccharide units. While the sulfate-rich regions of the CS chains contain 20–28% 4-sulfated disaccharides, the other regions have little or no sulfate. Further, we find that the placental CSPGs are able to efficiently bind IRBCs due to the presence of 4-sulfated disaccharide clusters. The oligosaccharides corresponding to the sulfate-rich domains of the CS chains efficiently inhibited IRBC adhesion. Thus, our data demonstrate, for the first time, the unique distribution of sulfate groups in the CS chains of placental CSPGs and that these sulfate-clustered domains have the necessary structural elements for the efficient adhesion of IRBCs, although the CS chains have an overall low degree of sulfation.

A distinctive feature of Plasmodium falciparum compared with the other three human malaria parasites is its ability to express adherent protein(s) on the surfaces of the infected red blood cells (IRBCs) 1 and thereby sequester in the microvascu-lar capillaries of various organs by adhering to endothelial cell surfaces (1)(2)(3)(4)(5). The extensive accumulation of IRBCs in vital organs causes capillary blockage with deprivation of oxygen and nutrients and production of toxic levels of proinflammatory cytokines (3, 6 -10), damaging the endothelial lining and causing organ dysfunction and severe pathological conditions. A number of studies have shown that the adherent protein expressed on the surfaces of IRBCs to be P. falciparum erythrocyte membrane protein 1 (EMP1), a multidomain, antigenic var gene family protein (11)(12)(13)(14)(15)(16)(17). P. falciparum EMP1 can bind, in a domain-specific manner, CD36, intercellular adhesion molecule-1, vascular cell adhesion molecule-1, E-selectin, platelet endothelial cell adhesion molecule-1/CD31, and thrombospondin on vascular endothelial cell surface (18 -24). In addition, P. falciparum EMP1 can also bind complement receptor (25), heparan sulfate (20), and chondroitin 4-sulfate (C4S) (22)(23)(24)(25)(26)(27)(28). Thus, the parasite, by its ability to express divergent P. falciparum EMP1s using the var gene repertoire, can adhere to various organs. However, over a period of time, the host develops antibodies against the exposed P. falciparum EMP1 that are able to inhibit adhesion of IRBC adhesion and aid clearance of infection (29 -34). To overcome this defensive mechanism, the parasite constantly switches, at low frequency, to various adherent phenotypes by expressing P. falciparum EMP1s with different receptor specificity (35,36). This ability of the parasite to express P. falciparum EMP1, for which the host has not yet developed adhesion-inhibitory antibodies, enables it to selectively adhere through a different receptor. In this manner, when one adherent phenotype of parasite is eliminated by the host, another phenotype continues to thrive. In endemic areas, people by adulthood acquire a broad spectrum protective immunity against P. falciparum, including antibodies to P. falciparum EMP1s (37,38). Therefore, in immuneprotected people, the IRBCs cannot adhere in the vascular capillaries, limiting the parasite growth.
In pregnant women, however, the placenta provides a new opportunity for IRBC adhesion, because women lack immunity against placenta-adherent parasites prior to pregnancy (39). Extensive adherence of IRBCs in the placenta and infiltration of mononuclear cells in response to the infection results in impaired placental function, leading to poor fetal outcome and maternal morbidity and mortality (40 -46). However, women acquire placental malaria-specific immunity, including adhesion-inhibitory antibody response, during the first and second pregnancies (29 -34). Therefore, primigravidas are at highest risk of placental malaria, and the susceptibility diminishes with increasing gravidity (44,45).
C4S mediates the adhesion of IRBCs in the human placenta (39, [47][48][49]. Previously, we have shown that the CSPGs localized in the intervillous spaces of the placenta are the receptors for the adherence of IRBCs in the placenta (50). These CSPGs were found to be unusually low sulfated; on an average, only ϳ8% of the disaccharide repeating units of the CS chains of placental CSPGs are 4-sulfated, and the remainder are nonsulfated (50). In previous studies, we have also shown that IRBC adhesion involves the participation of both nonsulfated and 4-sulfated disaccharide repeating units and the optimal binding requires ϳ30% 4-sulfated and ϳ70% nonsulfated disaccharide repeats (51). Further, we established that a C4S motif having six disaccharide repeating units (6-mer) with two 4-sulfated and four nonsulfated disaccharide units is the minimum structural motif required for optimal binding of IRBCs (51). A recent study confirmed most of our findings (52), except that four or five rather than two of the disaccharide repeating units of the binding motif containing six-disaccharide repeating units needs to be 4-sulfated for effective IRBC binding. However, it should be noted that these investigators measured C4S-IRBC interactions by immobilizing the commercially available bovine trachea C4S/C6S copolymer (52), a nonrelevant glycosaminoglycan, rather than the placental CSPGs, the natural receptor used in our study (50).
Regardless of this discrepancy, it remained a puzzle how IRBCs are able to efficiently bind the unusually low sulfated CS chains of the placental intervillous space CSPGs. A full understanding of the structural requirements for IRBC binding to placenta is important for developing therapeutics or vaccine for placental malaria (53). Therefore, in this study, we investigated in detail the structure of the CS chains of placental CSPGs, particularly the pattern of sulfate group distribution and its correlation to IRBC binding. The placental CSPGs were fractionated into two differentially sulfated proteoglycan populations and their CS chains isolated. The polysaccharide chains were degraded with an endoenzyme that specifically cleaves the nonsulfated regions of the CS into disaccharides, and the oligosaccharide products thus obtained were purified and examined for their ability to inhibit IRBC binding to intact CSPGs. The data demonstrate that, although the overall sulfate content of the CS chains of placental CSPGs are markedly lower than that required for optimal binding, the sulfate groups in the CS chains are clustered in uniquely size-defined domains. These sulfate-rich CS domains have the requisite structural features for the efficient binding of IRBCs.
Isolation of CSPGs from Human Placenta-The low sulfated CSPGs of the placental intervillous spaces were isolated as described previously with minor modification (50). Briefly, the placentas were cut into small pieces and extracted with PBS, pH 7.2, containing protease inhibitors, and the extract was applied onto DEAE-Sephacel columns (2.5 ϫ 22 cm). The columns were washed with 25 mM Tris-HCl, 150 mM NaCl, 10 mM EDTA, pH 8.0, and then equilibrated with 50 mM NaOAc, 100 mM NaCl, pH 5.5. The bound material was eluted with a linear gradient of 0.1-0.9 M NaCl in 50 mM NaOAc, pH 5.5. 10-ml fractions were collected, and aliquots were analyzed for uronic acid content (54). Uronic acid-containing fractions corresponding to BCSPG-2, the major CSPG of the placental intervillous spaces (50), were pooled and dialyzed against water. The dialysates were adjusted to 50 mM NaOAc, 100 mM NaCl, pH 5.5, and applied onto DEAE-Sepharose columns (2.5 ϫ 17 cm). The columns were washed with 50 mM NaOAc, 0.15 M NaCl, pH 5.5, and eluted with a linear gradient of 0.15-0.55 M NaCl in 50 mM NaOAc, pH 5.5. Fractions of 10-ml were collected, absorption at 260 and 280 nm was measured, and aliquots were analyzed for uronic acid content (54).
Cesium Bromide Density Gradient Centrifugation of CSPGs-The crude CSPG fractions obtained by DEAE-Sepharose chromatography were dissolved (1 mg/ml) in 25 mM sodium phosphate, pH 7.2, containing 50 mM NaCl, 0.02% NaN 3 , 4 M guanidine hydrochloride, and 42% (w/w) CsBr. The solutions were centrifuged in a Beckman 50 TI rotor at 44,000 rpm for 65 h (55). Gradients were collected from the bottom of the centrifuge tubes into 15 equal fractions and analyzed for uronic acid content (54) and for proteins by measuring the absorption at 260 and 280 nm.
Size Exclusion Chromatography of CSPGs-The CSPG fractions obtained from the CsBr density gradient centrifugation step were further purified by chromatography on columns of Sepharose CL-6B (1 ϫ 49 cm) and/or Sepharose CL-4B (1 ϫ 48 cm) in 20 mM Tris-HCl, 150 mM NaCl, pH 7.6, containing 4 M guanidine hydrochloride. Fractions were collected and monitored for absorption at 260 and 280 nm and for uronic acid content (54).
Isolation of the CS Chains of CSPGs-The purified CSPGs were treated with 0.1 M NaOH, 1 M NaBH 4 for 18 -20 h under nitrogen atmosphere at 45°C (56). The solutions were cooled in ice bath, neutralized with 1 M cold acetic acid, and then dried in a rotary evaporator. Boric acid was removed by repeated evaporation in a rotary evaporator by the addition of methanol, 0.1% acetic acid. The residue was applied onto a DEAE-Sepharose column (1 ϫ 10 cm) in 20 mM Tris-HCl, pH 7.8, washed with 20 mM Tris-HCl, 0.15 M NaCl, pH 7.8, and then eluted with a linear gradient of 0.15-0.6 M NaCl in the same buffer. Fractions (2.5 ml) were collected, and aliquots were assayed for uronic acid content (54). Uronic acid-positive fractions were pooled, dialyzed against distilled water, and lyophilized.
Gel Electrophoresis-The CS oligosaccharides were dissolved in 100 mM Tris base, 100 mM boric acid, 2 mM EDTA, pH 8.3, containing 5% glycerol (58). The solutions were electrophoresed on 10% polyacrylamide gels (15 ϫ 16 cm) in 100 mM Tris base, 100 mM boric acid, 2 mM EDTA, pH 8.3. The gels were stained with 0.03% Alcian Blue in 25% ethanol, 10% aqueous acetic acid for 4 h and destained with 25% ethanol, 10% aqueous acetic acid. For silver staining, the gels were treated with 10% aqueous glutaraldehyde for 30 min and washed with water three times for 30 min each. The gels were then treated with freshly prepared ammoniacal silver for 15 min, washed with water two times for 30 s each, developed with 0.005% citric acid and 0.019% formaldehyde in water, and then washed with water (59).
DEAE-Sepharose Chromatography of Oligosaccharides-The oligosaccharides (fractions 20 -35, 50 g) obtained by the Bio-Gel P-6 chro-2 R. N. Achur and D. C. Gowda, unpublished results. matography of the S. dysgalactiae hyaluronidase digest of chondroitin were dissolved in 20 mM NaOAc, 50 mM NaCl, pH 5.0, and applied onto a DEAE-Sepharose microcolumn (0.1-ml bed volume). After washing with 0.5 ml of the above buffer, the bound oligosaccharides were eluted stepwise with buffer containing 0.1, 0.2, 0.4, and 0.6 M NaCl (0.5 ml each). Fractions (0.1 ml) were collected, and aliquots were analyzed for uronic acid (54). The oligosaccharide-containing fractions were pooled and digested with chondroitinase ABC, and the disaccharides formed were analyzed by HPLC.
Disaccharide Composition Analysis-The C4Ss or C4S oligosaccharides (10 -15 g) were digested with chondroitinase ABC (10 -20 milliunits) in 50 l of 0.1 M Tris-HCl, pH 8.0, containing 30 mM NaOAc and 0.01% BSA at 37°C for 12-15 h (60). The released, unsaturated disaccharides were analyzed on an amine-bond Microsorb-MV column (4.6 ϫ 250 mm; Varian) using Waters 600E HPLC system (Milford, MA) (61). The enzyme digests were injected, and the column was eluted with a linear gradient of 16 -530 mM NaH 2 PO 4 over 70 min at room temperature at a flow rate of 1 ml/min. 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.
Carbohydrate Composition Analysis-The CSPGs or CS chains (5-10 g) were hydrolyzed with 4 M HCl at 100°C for 6 h. The hydrolysates were dried in a Speed-Vac and analyzed on a CarboPac PA1 high pH anion exchange HPLC column (4 ϫ 250 mm; Dionex) (62). The column was eluted with 20 mM sodium hydroxide, elution of sugars was monitored by pulsed amperometric detection, and the response factors for sugars were determined using standard sugar solutions.
Other Analytical Procedures-The uronic acid contents in various column chromatography fractions were determined by the carbazolesulfuric acid method (54). Protein contents were measured using the Micro BCA Protein Assay Reagent kit from Pierce (63).
P. falciparum Cell Culture-The C4S adherent P. falciparum, selected by panning of 3D7 laboratory parasite clones on placental CSPGcoated plastic plates, were used in this study. The parasites were cultured using type O-positive human red blood cells at 3% hematocrit in RPMI 1640 medium supplemented with 25 mM HEPES, 29 mM sodium bicarbonate, 0.005% hypoxanthine, p-aminobenzoic acid (2 mg/ liter), gentamycin sulfate (50 mg/liter), and 10% O-positive human serum. The cultures were incubated at 37°C in an atmosphere of 90% nitrogen, 5% oxygen, and 5% carbon dioxide (51).
IRBC Adhesion and Adhesion-Inhibition Assays-The adherence of IRBCs was performed by coating solutions (10 -15 l) of purified CSPGs as circular spots on 150 ϫ 15-mm plastic Petri dishes as described previously (51). The specificity of IRBC binding to CSPGs was ascertained by incubating the CSPG-coated plates with chondroitinase ABC (50 milliunits/ml) as well as by competitive inhibition with various C4Ss.
For adhesion-inhibition assays, IRBCs were incubated with various C4Ss or C4S oligosaccharides at the indicated concentrations in PBS, pH 7.2, in 96-well microtiter plates at room temperature for 30 min with intermittent mixing (51). The IRBC suspension was 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 Giemsa, and counted using a light microscope.

Fractionation and Characterization of Differentially Sulfated
CSPGs of Human Placental Intervillous Spaces-The major low sulfated CSPG fraction (previously designated as BCSPG-2) (50) was isolated by one-step DEAE-Sephacel chromatography of the isotonic buffer extract of placentas. This CSPG fraction represents about 93-94% of the total low sulfated CSPGs in the intervillous spaces (50). When subjected to DEAE-Sepharose chromatography, using a 0.15-0.55 M NaCl gradient, the CSPG was partially resolved into two fractions (designated BCSPG-2a and BCSPG-2b) that are distinct in their sulfation levels (Fig. 1). The proportions of BCSPG-2a and BCSPG-2b varied considerably from one placenta to another, in the range 40 -65% and 35-60%, respectively. These CSPG fractions were further purified and characterized, and their ability to bind IRBCs was studied.
On CsBr density gradient centrifugation, BCSPG-2a and BCSPG-2b sedimented to the middle of the gradients (average ϭ 1.43 g/ml) separating from protein and nucleic acid contaminants (not shown). The sedimentation patterns of BCSPG-2a and BCSPG-2b were indistinguishable from each other and very similar to that previously reported for the total BCSPG-2 (50). In both cases, fractions of the gradient containing significant levels of CSPGs ( ϭ 1.35-1.5 g/ml density regions) were pooled, and the material was recovered.
BCSPG-2a and BCSPG-2b were further purified by Sepharose CL-6B and Sepharose CL-4B chromatography, which removed any residual protein contaminants. In each case, the CSPG fraction eluted as a single nonsymmetrical peak, and the chromatograms were similar to that previously reported for the total BCSPG-2 (not shown) (50). The yields and compositions of the purified BCSPG-2a and BCSPG-2b are shown in Table I. As in the case of total BCSPG-2, BCSPG-2a and BCSPG-2b each contained high and low molecular mass (ϳ1000-and ϳ570-kDa, respectively) proteoglycan species (50). The proportions of ϳ1000and ϳ570-kDa species in BCSPG-2a and BCSPG-2b were similar (not shown). Chondroitinase ABC degraded the glycosaminoglycan chains of both BCSPG-2a and BCSPG-2b completely into unsaturated disaccharides, as assessed by chromatography of the enzyme digests on Bio-Gel P-6 column (not shown), indicating the absence of hyaluronic acid and/or heparan sulfate in these fractions. Consistent with these results, the CS chains of the proteoglycan fractions contain predominantly galactosamine (Table II), and they were completely resistant to the action of S. hyalurolyticus hyaluronidase and heparitinase (not shown). As evident from the disaccharide compositions of the CS chains (see below; Table  II), BCSPG-2a and BCSPG-2b differ significantly in their sulfate contents. Thus, the differences in the overall charge density of the proteoglycan fractions, as indicated by their elution at different salt concentrations from DEAE-Sepharose columns (Fig. 1), is mainly due to the differences in sulfate contents of the CS chains.
SDS-PAGE analysis of the core proteins released after chondroitinase ABC treatment revealed that BCSPG-2a and BCSPG-2b each contain two distinct core proteins, a high  (50). The uronic acid-containing fractions corresponding to the major CSPG fraction (designated BCSPG-2 in Ref. 50) were pooled, dialyzed, and chromatographed on DEAE-Sepharose (2.5 ϫ 17 cm) in 50 mM NaOAc, 0.15 M NaCl, pH 5.5. The column was washed with the same buffer, and the bound CSPGs were eluted with a linear gradient of 0.15-0.55 M NaCl in 50 mM NaOAc, pH 5.5. 10-ml fractions were collected, absorptions at 280 and 260 nm were measured, and aliquots were assayed for uronic acid (54). The CSPG peaks were pooled as indicated by horizontal bars. Note that the previously reported minor BCSPG-1 (50), which represents 6 -7% of the total CSPGs of the intervillous spaces of the placenta, was not studied here.
(ϳ670-kDa) and a low (ϳ56-kDa) molecular mass species, respectively, in ϳ40 and ϳ60% proportion ( Table I). The proteoglycan fractions closely resembled one another in the amino acid compositions of their core proteins and the molecular weight (ϳ60,000) of the CS chains, which were similar to that reported previously for BCSPG-2 (50). The results, when considered together, indicate that BCSPG-2a and BCSPG-2b are similar to one another with respect to proteoglycan types but differ in the amounts of sulfate groups in their CS chains (Table II).
Adhesion of P. falciparum IRBCs to BCSPG-2a and BCSPG-2b-The purified BCSPG-2a and BCSPG-2b were assessed for their abilities to adhere IRBCs by an in vitro cytoadherence assay (Fig. 2). Both CSPGs efficiently bound IRBCs in a concentration-dependent manner. Significant levels of IRBC binding were observed at coating concentration as low as 12 ng/ml. At 50 ng/ml, both fractions efficiently bound IRBCs, and at 100 -200 ng/ml they exhibited saturated levels of binding. Thus, despite significant difference in the sulfate contents of the CS chains, BCSPG-2a and BCSPG-2b were indistinguishable from one another with regard to the number of IRBCs adhering when immobilized on solid surfaces ( Fig. 2A). It is possible that similar IRBC binding capacities of BCSPG-2a and BCSPG-2b might merely reflect the number of IRBCs bound, even if the IRBCs adhered to these proteoglycans with different affinities. Therefore, to determine the relative affinities of IRBC binding by BCSPG-2a and BCSPG-2b, we performed adhesion inhibition assays. BCSPG-2a and BCSPG-2b were coated on plastic plates and tested in parallel for competitive inhibition of IRBC binding by a regioselectively 6-O-desulfated bovine trachea C4S fraction with 36% 4-sulfate and a C4S 6-mer with 36% 4-sulfate groups. Both compounds were marginally (10 -20%) more inhibitory to IRBC binding to BCSPG-2a compared with BCSPG-2b, suggesting that BCSPG-2b binds IRBCs with slightly higher affinity than BCSPG-2a (Fig. 2B). This agrees with the difference in the sulfate content and the number of IRBC binding sites in the CS chains of BCSPG-2a and BCSPG-2b. However, the difference in the levels of inhibition is not as much as that expected based on the number of IRBC binding sites in the CS chains of BCSPG-  (61). The relative proportions of the disaccharides were calculated from the areas of the HPLC peaks by assuming that the two unsaturated disaccharides have similar molar extinction coefficients.
FIG. 2. Adhesion of P. falciparum IRBCs to the differentially sulfated CSPG fractions of human placenta and inhibition of adhesion by partially sulfated C4S. The solutions of purified CSPG fractions in PBS, pH 7.2, were coated at the indicated concentrations as 4-mm circular spots on plastic Petri dishes overnight at 4°C. The nonspecific sites were blocked with 2% BSA for 2 h at room temperature. A, the blocked spots were overlaid with a 2% suspension of IRBCs in PBS, pH 7.2, for 40 min at room temperature. The unbound cells were washed, and the bound cells were fixed with 2% glutaraldehyde, stained with Giemsa, and counted under light microscopy. Assays were carried out three times each in duplicate. The spots coated only with PBS and the uninfected RBCs layered on CSPG-coated spots were used as negative controls. Shown is the dose-dependent binding of IRBCs to CSPG-coated plates (mean of all three assays). q, total BCSPG-2; E, BCSPG-2a; ‚, BCSPG-2b. B, a 2% suspension of IRBCs in PBS, pH 7.2, were incubated with various concentrations of the indicated C4S for 30 min at room temperature and then overlaid onto the CSPG-coated spots. After 40 min at room temperature, the unbound cells were washed, and the bound cells were fixed, stained with Giemsa, and measured using light microscopy. Assays were performed two times each in duplicate, and mean values were plotted. q and E, inhibition of IRBC binding to BCSPG-2a and BCSPG-2b, respectively, by C4S with 36% 4-sulfate prepared by regioselective 6-O-desulfation of bovine trachea C4S/C6S copolymer; OE and ‚, inhibition of IRBC binding to BCSPG-2a and BCSPG-2b, respectively, by 6-mer prepared from C4S with 36% 4-sulfate.
The Level and Distribution Pattern of Sulfate Groups in the CS Chains of Placental CSPGs-The CSPGs of the intervillous spaces can efficiently bind IRBCs despite markedly lower total sulfate contents of their CS chains than that required for the optimal IRBC binding (51). This suggested that the sulfate groups may be clustered in their CS chains and that these sulfate-rich regions could efficiently bind IRBCs. To investigate this further, we performed a detailed structural characterization of the CS chains of placental CSPGs. The CS chains of BCSPG-2a and BCSPG-2b were isolated by the alkaline ␤-elimination of the proteoglycans followed by chromatography on DEAE-Sepharose columns (Fig. 3). In the case of BCSPG-2a, about 68% of CS chains (designated as CS-2a) eluted as a single peak at 0.28 M NaCl, and the remainder eluted as heterogeneous peaks at a mean NaCl concentration of 0.36 M. In the case of BCSPG-2b, on the other hand, only 28% of the CS chains eluted at 0.28 M NaCl, and the remainder (CS-2b) eluted as a single peak at 0.37 M NaCl. The presence of additional CS population (Fig. 3) in both BCSPG-2a and BCSPG-2b was probably due to the overlapping separation of the CSPG fractions on DEAE-Sepharose columns (see Fig. 1). Thus, these results demonstrate that BCSPG-2a and BCSPG-2b carry differentially sulfated CS chains. The presence of two intervillous space CSPGs with distinctively sulfated CS chains was also evident from the elution pattern of the CS chains released by alkaline ␤-elimination of the total placental intervillous space CSPGs (BCSPG-2) on the DEAE-Sepharose column (see inset in Fig. 3).
The CS chains of BCSPG-2a and BCSPG-2b, CS-2a and CS-2b, fractionated by DEAE-Sepharose chromatography, were recovered by pooling the fractions as shown in Fig. 3. The molecular sizes of the CS chains were assessed by chromatography on a Sepharose CL-6B column calibrated with glycosaminoglycans of known molecular weights (50). The CS-2a and CS-2b were eluted as single symmetrical peaks, indistinguishable from one another, with an estimated molecular weight of ϳ60,000 (not shown). Hexosamine compositional analysis indicated that both CS-2a and CS-2b have predominantly N-acetylgalactosamine (Table II). HPLC analysis of the unsaturated disaccharides released by the digestion with chondroitinase ABC revealed that CS-2a and CS-2b, obtained from CSPGs of various placentas, consist of, respectively, 2-3% and 9 -14% 4-sulfated and 97-98% and 86 -91% nonsulfated disaccharide repeating units. Together, these results suggest that CS-2a and CS-2b are similar in molecular sizes but differ in the levels of sulfate groups.
To determine the distribution of sulfate groups, the CS chains of placental CSPG fractions were digested with S. dysgalactiae hyaluronidase, an endo-␤-N-acetylhexosaminyl lyase. This enzyme degrades hyaluronic acid and chondroitin, but not chondroitin sulfate, to produce disaccharides with nonreducing 4,5-unsaturated uronic acid residue (57). Chromatography of the enzyme digests on Bio-Gel P-6 indicated that ϳ95 and 80%, respectively, of CS-2a and CS-2b were degraded predominantly into disaccharides with minor amount of tetrasaccharides. The remainders of the CS chains were converted into oligosaccharides larger than decasaccharide (5-mer) (Fig. 4). The disaccharide composition of the oligosaccharide fractions (see Fig. 4), determined by HPLC after digestion with chondroitinase ABC, are given in Table III and 78 -80% nonsulfated disaccharide repeating units, whereas Fractions I and II from CS-2b had 25-28% 4-sulfated and 72-75% nonsulfated disaccharide repeating units. The tetrasaccharides (Fraction III) from CS-2a were ϳ97% nonsulfated and ϳ3% sulfated, whereas those from CS-2b were ϳ95% nonsulfated and ϳ5% sulfated. As expected, based on the specificity of S. dysgalactiae hyaluronidase, the disaccharides (Fraction IV), formed by the action of this enzyme, were exclusively nonsulfated. Interestingly, however, hexa-to decasaccharides (3-to 5-mers) were not formed in appreciable amounts from either CS-2a or CS-2b. These results indicate that, in the CS chains of both BCSPG-2a and BCSPG-2b, the majority of sulfate groups are clustered such that these regions have 20 -28% or more of 4-sulfated disaccharide repeating units. The yields of the larger oligosaccharides (Fractions I and II) from CS-2a and CS-2b should correspond to the number of sulfate group-rich domains in the CS chains.
Since the oligosaccharide Fractions I and II (Fig. 4) obtained by the digestion of the CS chains of placental CSPGs with S. dysgalactiae hyaluronidase contained only 20 -28% sulfated disaccharides, it is possible that the oligosaccharides are a mixture of sulfated and nonsulfated species. Because the placental CS chains were available only in limited amounts, the specificity of the enzyme, under the conditions used for the placental CS chains, was studied using a commercially available chondroitin, which consisted of ϳ4% 6-and 4-sulfated disaccharide moieties and ϳ96% nonsulfated. Bio-Gel P-6 chromatography showed that the enzyme degraded ϳ90% of chondroitin into disaccharides and ϳ10% (ϳ6% in fractions 20 -25 and ϳ4% in fractions 26 -35; the pattern was similar to that of CS-2a in Fig. 4, not shown) into a mixture of oligosaccharides with Ͼ5 disaccharide repeating units. The oligosaccharides in fractions 20 -25 and 26 -35 had 12-15% 6-sulfated and 6 -8% 4-sulfated disaccharide moieties and the remainder nonsulfated. Ion exchange chromatography on DEAE-Sepharose using stepwise elution with NaCl and compositional analysis of the fractions showed the presence of predominantly sulfated oligosaccharides. These results suggested that S. dysgalactiae hyaluronidase readily degrades the nonsulfated regions of CS and slowly acts at the sulfated domains, forming partially sulfated oligosaccharides.
To determine the exact sizes of the larger oligosaccharides formed by the digestion of CS-2a and CS-2b with S. dysgalactiae hyaluronidase, Fractions I and II (see Fig. 4) were analyzed by polyacrylamide gel electrophoresis (Fig. 5). In both the cases (CS-2a and CS-2b), the oligosaccharides in Fractions I and II ranged in size from 8 to 14 and from 6 to 11 disaccharide units, respectively (Fig. 5). These results indicate that the sulfate groups in the CS chains of BCSPG-2a and BCSPG-2b are clustered in CS chain motifs composed of 6 -14 disaccharide units.

Inhibition of P. falciparum IRBC Adherence to the Placental CSPGs by the CS Chains of Placental CSPGs and Their
Oligosaccharides-The intact CS chains, CS-2a and CS-2b, and the oligosaccharides obtained by the digestion of CS chains with S. dysgalactiae hyaluronidase (see Fig. 4) were assessed for their ability to inhibit the adhesion of IRBCs to placental CSPGs. The CS chains as well as their oligosaccharide Fractions I and II inhibited the IRBC adhesion to BCSPG-2a and BCSPG-2b in a dose-dependent manner (Fig. 6). Consistent with the prediction based on the level of 4-sulfated disaccharide clustered domains (see Table II), the inhibition of IRBC binding by CS-2b was 2-3-fold higher than that by CS-2a. Further, oligosaccharide Fractions I and II, obtained from CS-2a and CS-2b, were significantly better inhibitors than the corresponding intact chains ( Fig. 6 and data not shown). The inhibitory capacity of Fractions I and II of CS-2a was only marginally lower than those of Fractions I and II from CS-2b. The inhibitory ability of the oligosaccharide from CS-2b was comparable with that of C4S, with 36% 4-sulfated disaccharide residues prepared by the regioselective 6-O-desulfation of bovine a By the area of the chromatographic peaks obtained by plotting uronic acid contents of the fractions (see Fig. 4).
b The values were calculated from the areas of HPLC peaks by assuming that the different unsaturated disaccharides have similar molar extinction coefficients. trachea C4S/C6S copolymer (Fig. 6). In contrast, the C4S with 3 and 11% 4-sulfate groups prepared by solvolytic desulfation of a fully 4-sulfated sturgeon notochord C4S were markedly less inhibitory compared with the CS chains of BCSPG-2a, which have only 2-3% sulfated disaccharide repeating units. These results agree with the finding described above that the sulfate groups in the CS chains of BCSPG-2a and BCSPG-2b are clustered in CS chain motifs consisting of 6 -14-disaccharide repeating units. The results also agree with our previous finding that optimal binding of IRBCs requires a 6-mer motif in which two of six disaccharide repeating units sulfated on C-4 of N-acetylgalactosamine. Since the oligosaccharides obtained by the digestion of CS-2a and CS-2b with S. dysgalactiae hyaluronidase are larger than six disaccharide repeating units, the sulfate content in the 6-mer IRBC-binding motif is likely to be ϳ30% or more. This satisfies the level of 4-sulfation required for optimal IRBC binding. These results indicate that the sulfate groups in the CS chains of placental CSPGs are uniquely distributed and provide the necessary structural elements for the efficient adhesion of IRBCs. DISCUSSION Recently, we showed that unusually low sulfated CSPGs of the intervillous spaces of human placenta can efficiently support the adherence of P. falciparum IRBCs in the placenta (50). The IRBC binding involves critical interactions by both 4-sulfated and nonsulfated repeating units of the CS chains, and the optimal binding requires a 6-mer motif with two 4-sulfated disaccharide repeating units (51). In this study, the placental intervillous space CSPGs and the structural features of their CS chain motifs that bind IRBCs were investigated. The new findings are as follows. 1) The placental CSPGs are a mixture of two distinct populations, which are similar with regard to proteoglycan type and sizes but different with respect to the levels of sulfation. 2) The majority of the sulfate groups in the glycosaminoglycan chains of both CSPGs are clustered in sizedefined domains that comprised 6 -14 disaccharide repeating units. Within these domains, about 20 -28% of the disaccharide repeating units are 4-sulfated, whereas the other regions of the CS chains have essentially no sulfate groups.
3) The oligosaccharides corresponding to these sulfate group-rich domains can efficiently inhibit IRBC binding to placental CSPGs. Thus, our data define, for the first time, the distribution of sulfate groups in the CS chains of the low sulfated placental CSPGs and establish that P. falciparum uses the sulfate group-clustered domains of the CS chains for IRBC adhesion in the placenta.
The data show that the low sulfated CSPGs of the intervillous spaces of placenta consist of two major, differentially sulfated proteoglycan populations, BCSPG-2a and BCSPG-2b. These CSPG populations resemble one another very closely with respect to their physical properties, including hydrodynamic size, buoyant density, and the core protein type, proportion, and composition. The CSPG species are also very similar with respect to the number and molecular sizes of the CS chains, but they differ significantly in the sulfate contents of the CS chains (2-3% and 9 -14%, respectively, depending on placentas).
A CSPG population with 2% sulfate groups in the CS chains has been previously identified as a minor CSPG in the placental intervillous spaces (50). This CSPG species, designated as BCSPG-1, accounted for only 6 -7% of the total CSPGs and was eluted from the DEAE-Sephacel column as a distinct peak, completely separated from the remainder of the intervillous space CSPG molecules, BCSPG-2 (50). A comparison of the data from this study with those of the previous study indicates that BCSPG-1 is very similar to BCSPG-2a by all criteria including overall charge density, core protein types, and the content of sulfate groups in CS chains. Therefore, BCSPG-1 and BCSPG-2a appear to represent similar CSPG populations. The results of the present study clearly show that these proteoglycan species represent 40 -65% of the total CSPGs of the placental intervillous spaces. Based on the results of the previous study, it is evident that only a portion of BCSPG-2a aggregates with the low amount of hyaluronic acid present in the intervillous spaces and eluted as a distinct peak during DEAE-Sephacel chromatography (50).
The results of this study conclusively establish that the majority of sulfate groups of the CS chains of the CSPGs of placental intervillous spaces are clustered in size-defined motifs, consisting of 6 -14 disaccharide repeating units. The sizes of the oligosaccharides formed by the digestion of the CS chains of BCSPG-2a and BCSPG-2b with S. dysgalactiae hyaluronidase support this conclusion. The enzyme converted Ͼ90% of the CS chains of BCSPG-2a and ϳ80% of the CS chains of BCSPG-2b into predominantly nonsulfated disaccharides and a minor amount of tetrasaccharides; only a small portion of the latter appears to be 4-sulfated (Table III). Based on the specificity of the enzyme (degrades chondroitin but not chondroitin sulfate), the amount of sulfate groups in the tetrasaccharide fraction must represent the level of 4-sulfated disaccharide units that occur as single residues randomly distributed in relatively large nonsulfated regions of the CS chains. This accounts for 3-5% of the total sulfate groups in the CS chains of BCSPG-2a and BCSPG-2b. These results clearly show that the majority of the sulfate groups in CS chains of placental CSPGs are clus-  Fig. 2. The CS chains and their oligosaccharides were incubated at the indicated concentrations with a 2% IRBC suspension in PBS, pH 7.2, for 30 min at room temperature and then overlaid onto the CSPG-coated spots. After a 40-min incubation at room temperature, the unbound cells were washed. The bound cells were fixed, stained with Giemsa, and measured using light microscopy. Shown is the inhibition of IRBC binding to BCSPG-2b-coated plates. q, CS-2a; E, CS-2b; OE, oligosaccharide FI from CS-2b; ‚, oligosaccharide FII from CS-2b; Ⅺ, bovine trachea C4S/C6S copolymer; ϫ, C4S with 36% 4-sulfate prepared by the regioselective 6-O-desulfation of bovine trachea C4S followed by DEAE-Sepharose fractionation; f, the CS chains obtained from total BCSPG-2 (nonfractionated mixture of BCSPG-2a and BCSPG-2b); ƒ and , C4Ss with 3 and 11% 4-sulfate prepared by solvolytic partial desulfation of sturgeon notochord C4S (51); ࡗ, chondroitin. The inhibition capacities of oligosaccharide Fractions I and II from CS-2a were either similar or only marginally lower compared with those from CS-2b (not shown). Inhibition of IRBC binding to BCSPG-2a was similar to that observed in the case of BCSPG-2b, except that a 10 -20% higher inhibition was observed for each inhibitor used. Note that the inhibitory capacity of the 6-mer prepared by the testicular hyaluronidase digestion of C4S with 36% 4-sulfated disaccharide was similar to that by the intact C4S with 36% 4-sulfate. tered in domains consisting of 6 -14 disaccharide repeating units.
Our data demonstrate that P. falciparum IRBCs bind to the sulfate group-clustered motifs but not to the nonsulfated regions of the CS chains of placental intervillous space CSPGs. Consistent with this conclusion, chondroitin, the nonsulfated glycosaminoglycan, has no inhibitory effect on the adhesion of IRBCs to placental CSPGs. The C4Ss with 3 and 11% 4-sulfate content, prepared by partial desulfation of fully sulfated sturgeon notochord C4S, could only marginally inhibit the IRBC binding (Fig. 6). In these C4Ss, the majority of the sulfate groups are likely to be distributed as single residues separated by a number of nonsulfated moieties. In contrast, the CS chains of BCSPG-2a and BCSPG-2b, containing clustered sulfate residues, are severalfold more inhibitory. The oligosaccharides corresponding to the sulfate group-clustered domains (Fractions I and II), obtained by the S. dysgalactiae hyaluronidase digestion of the CS chains of the placental CSPGs, are superior inhibitors compared with the corresponding intact chains (Fig.  6). This is especially evident in the case of the CS chains of BCSPG-2a. This is not surprising, because the major portion of CS-2a and significant amounts of CS-2b have no sulfate groups, and thus these portions are not able to bind IRBCs. Therefore, on weight basis, the oligosaccharide Fractions I and II should be much more active in inhibiting IRBC binding than the intact CS chains, CS-2a and CS-2b, from which they were obtained.
We have previously shown that two of the six disaccharide residues of the CS 6-mer motif must be 4-sulfated for optimal IRBC binding (51). This translates to 33% 4-sulfated residues in the oligosaccharide. The level of 4-sulfate determined in this study for the sulfate group-clustered domains of the CS chains of placental CSPGs is 20 -28%. However, it should be noted that the sizes of the predominant oligosaccharides (Fractions I and II in Fig. 4) formed by the action of S. dysgalactiae hyaluronidase are larger than the 6-mer, the minimum chain length required for optimal inhibition. Therefore, it is possible that the 6-mer binding motifs in the CS chains of placental CSPGs have the level of 4-sulfation required for optimal IRBC adhesion.
Although the capacity of the CS chains of BCSPG-2a to inhibit the IRBC binding to placental CSPGs is relatively low, immobilized BCSPG-2a efficiently binds IRBCs. When coated on plastic plates, the density of IRBC binding by BCSPG-2a is remarkably similar to that by BCSPG-2b. Further, the CS chains of BCSPG-2b contain significantly higher 4-sulfate, and the inhibitory capacities of these chains were 4 -5-fold higher compared with those of BCSPG-2a. Based on the sulfate contents, size of the CS chains, and requirements of two sulfate groups per 6-mer IRBC-binding motif, it can be calculated that the CS chains of BCSPG-2a and BCSPG-2b have about 1-2 and 4 -10 binding sites, respectively. Thus, the difference in the inhibitory capacities of the CS chains of these proteoglycan species agree with that expected based on the number of available IRBC binding sites in the CS chains. However, when IRBC binding and inhibitory data considered together, it appears that the density of IRBCs adhered to CSPG-coated plastic plates merely represents the number of cells bound and does not reflect the binding strength. When IRBC binding affinity to immobilized BCSPG-2a and BCSPG-2b was measured by adhesion inhibition using C4S with 36% 4-sulfate, marginal differences were observed, only at very low inhibitor concentrations (Fig. 6). This was also the case with C4S 6-mer having 36% 4-sulfate or oligosaccharide Fractions I and II from the CS chains of BCSPG-2b. Therefore, it appears that, when CSPGs are immobilized on solid surface, because of clustering of molecules, the CS chains of BCSPG-2a, despite containing fewer binding sites, can provide a sufficient number of binding sites for IRBCs using adjacent CS chains. Thus, they are able to bind IRBCs with capacity almost similar to that of the CS chains of BCSPG-2b. Alternatively, it is possible that because of the steric constraints due to the one-dimensional disposition of IRBCs with respect to the immobilized CS chains, IRBCs might not be able to interact with all available binding sites in the CS chains of BCSPG-2b. Therefore, the CS chains of both proteoglycan fractions could provide a similar number of binding sites to interact with IRBCs, and thus the CS chains of both BCSPG-2a and BCSPG-2b bind with almost comparable capacity. In solution phase, on the other hand, the disposition of IRBCs to binding by CS chains is multidirectional, which might allow for effective interactions with all of the available binding sites in the CS chains. Therefore, the CS chains of BCSPG-2b can inhibit the IRBC adhesion more effectively than the CS chains of BCSPG-2a. However, in the placenta, since the CSPGs are immobilized in the intervillous spaces, it is likely that both proteoglycans are equally efficient in binding IRBCs.
The clustering of almost all of the sulfate groups present in CS chains of the placental CSPGs in size-defined oligosaccharide domains suggests a well regulated mechanism of biosynthesis. This clustering of sulfate groups in the CS chains of the CSPGs might have important roles in the function of placenta. Clearly, P. falciparum has evolved to exploit these sulfated group-clustered motifs of the CS chains for adherence in the placenta. Recent studies have demonstrated that versican, a high molecular weight CSPG expressed in a variety of cells and tissues, including fibroblasts, arterial smooth muscle cells, keratinocytes, kidney, brain, aorta, and skin, can bind L-selectin, P-selectin, CD44, and a number chemokines (64 -66). The last class of proteins include secondary lymphoid chemokines, macrophage inflammatory protein, stromal cell-derived factor 1, monocyte chemoattractant protein, regulated on activation normal T-cell-expressed and secreted proteins, ␥-interferoninducible protein 10, platelet factor 4, and liver and activationregulated chemokine (64 -66). It has been shown that CS-dependent binding of these proteins to versican is important in leukocyte trafficking, signal transduction to trigger inflammatory responses, and the regulation of chemokine functions (64 -66). These proteins have been shown to bind different structural features within the CS chains of versican. CD44, a hyaluronic acid-binding protein, was found to bind to the nonsulfated regions of the CS chains of versican, whereas L-and P-selectins and various chemokines have been shown to interact with motifs consisting of specific sulfated residues (66). Therefore, analogous to this, the nonsulfated regions of the CS chains of placental CSPGs may bind CD44, whereas sulfate group-clustered regions support binding of other functional proteins. The placental CSPGs have been identified as members of the aggrecan family of proteoglycans (50). Since versican and aggrecan exhibit many similarities with respect to both core protein and CS chain structural features, it is possible that the CS chains of placental CSPGs also have diverse biological functions.