Structure and Anticoagulant Activity of Sulfated Galactans

We have characterized the structure of a sulfatedd-galactan from the red algae Botryocladia occidentalis. The following repeating structure (-4-α-d-Galp-1→3-β-d-Galp-1→) was found for this polysaccharide, but with a variable sulfation pattern. Clearly one-third of the total α-units are 2,3-di-O-sulfated and another one-third are 2-O-sulfated. The algal sulfated d-galactan has a potent anticoagulant activity (similar potency as unfractionated heparin) due to enhanced inhibition of thrombin and factor Xa by antithrombin and/or heparin cofactor II. We also extended the experiments to several sulfated polysaccharides from marine invertebrates with simple structures, composed of a single repeating structure. A 2-O- or 3-O-sulfatedl-galactan (as well as a 2-O-sulfatedl-fucan) has a weak anticoagulant action when compared with the potent action of the algal sulfated d-galactan. Possibly, the addition of two sulfate esters to a single α-galactose residue has an “amplifying effect” on the anticoagulant action, which cannot be totally ascribed to the increased charge density of the polymer. These results indicate that the wide diversity of polysaccharides from marine alga and invertebrates is a useful tool to elucidate structure/anticoagulant activity relationships.

In contrast with the algal polysaccharides, invertebrate galactans have simple structures, composed of a single repeating unit. The specific pattern of sulfation and the position of the glycosidic linkage varies among different species. The sea urchin (Echinodermata Echinoidea) Echinometra lucunter contains a linear polysaccharide composed of 2-O-sulfated, 3-linked ␣-L-galactose ( Fig. 1A) (13), while in the tunicate (Chordata Ascidiacea) Herdmania monus a similar polysaccharide is composed of 3-O-sulfated, 4-linked ␣-L-galactose ( Fig. 1B) (9). The sea urchin Strongylocentrotus franciscanus contains a related polysaccharide composed of 2-O-sulfated, 3-linked ␣-L-fucose units (Fig. 1C) (15). In other species of tunicate, the sulfated L-galactans have more complex and branched structures (4,6,7,9). In Styela plicata, non-sulfated L-galactose occurs as branched units linked to position O-2 of the central core ( Fig. 1D) (6,7). One way to determine the relationship between structure and biological activity of sulfated polysaccharides is to compare their activity in various assays where the contributions of the polysaccharide backbone, and the extent and position of sulfation have been fully characterized. In this line of work, new sulfated galactans and the sulfated fucan from invertebrates constitute a valuable tool.
Anticoagulant and antithrombotic activities are among the most widely studied properties of sulfated polysaccharides. The anticoagulant glycosaminoglycan heparin is an important therapeutic agent for prophylaxis and treatment of thrombosis (17); dermatan sulfate is also anticoagulant, although of lower potency than heparin (18,19). Other sulfated polysaccharides, either extracted from marine brown (16, 20 -23) and red alga (24,25), or obtained by chemical sulfation of natural polysaccharides (26), have been described as anticoagulant. In contrast with heparin and dermatan sulfate, the structural components of these algal sulfated polysaccharides have not been characterized for anticoagulant activity. Most of the difficulties arise from their heterogeneous chemical structures.
In the present study we isolated and characterized the structure of a sulfated D-galactan from the red algae Botryocladia occidentalis. This polysaccharide has the following repeating structure (-4-␣-D-Galp-133-␤-D-Galp-13). Besides its variable sulfation pattern, it clearly contains 2,3-di-O-sulfated D-galactose residues (approximately one-third of the total ␣-galactose units).
This algal sulfated galactan has a potent anticoagulant activity. Its action was compared with the sulfated galactans and the sulfated fucan shown in Fig. 1. Our results suggest that the 2,3-di-O-sulfated galactose residues have an "amplifying effect" on the anticoagulant activity of sulfated galactans.

Sulfated Galactans and a Sulfated Fucan from Invertebrates
Sulfated galactans were extracted from the tunics of the ascidians H. monus and S. plicata and from the egg jelly coat of the sea urchin E. lucunter. A sulfated fucan was extracted from the egg jelly coat of the sea urchin S. franciscanus. The sulfated polysaccharides were purified by anion exchange and/or gel filtration chromatography, as described (9,13,15).

Sulfated Galactans from the Red Algae B. occidentalis 1
Extraction-The marine red algae B. occidentalis was collected at Pacheco beach, Caucaia, Ceará , Brazil, separated from other species and sun-dried. The dried tissue (5 g) was cut in small pieces, suspended in 250 ml of 0.1 M sodium acetate buffer (pH 6.0) containing 510 mg of papain (E. Merck, Darmstadt, Germany), 5 mM EDTA, and 5 mM cysteine, and incubated at 60°C for 24 h. The incubation mixture was then filtrated and the supernatant saved. The residue was washed with 138 ml of distilled water, filtered again, and the two supernatants were combined. Sulfated polysaccharides in solution were precipitated with 16 ml of 10% cetylpyridinium chloride solution. After standing at room temperature for 24 h, the mixture was centrifuged at 2,560 ϫ g, for 20 min, at 5°C. The sulfated polysaccharides in the pellet were washed with 610 ml of 0.05% cetylpyridinium chloride solution, dissolved with 172 ml of a 2 M NaCl, ethanol (100:15, v/v) solution, and precipitated with 305 ml of absolute ethanol. After 24 h at 4°C, the precipitate was collected by centrifugation (2,560 ϫ g for 20 min at 5°C), washed twice with 305 ml of 80% ethanol, and once with the same volume of absolute ethanol. The final precipitate was dried at 60°C overnight and ϳ200 mg (dry weight) of crude polysaccharide was obtained after these procedures.
Purification-The crude polysaccharide (10 mg) was applied to a Mono Q column FPLC 2 (HR 5/5) (Amersham Pharmacia Biotech), equilibrated with 20 mM Tris-HCl buffer (pH 8.0). The column was developed by a linear gradient of 0 -3.0 M NaCl in the same buffer. The flow rate of the column was 0.50 ml/min and fractions of 0.5 ml were collected and assayed by metachromasia using 1,9-dimethylmethylene blue (27) and by the phenol-H 2 SO 4 reactions (28).
Chemical Analyses-Total hexose was measured by the method of Dubois et al. (28). After strong acid hydrolysis (6.0 M trifluoroacetic acid, 100°C for 5 h) of the polysaccharides, total sulfate was estimated by the BaCl 2 gelatin method (29). Standard curves for hexose and sulfate were constructed from galactose and Na 2 SO 4 . The hexose in the acidic hydrolysates was identified by gas-liquid chromatography of derived alditol acetates (30) and by paper chromatography in isobutyric acid, 1.0 M NH 4 OH (5:3, v/v) for 24 h on Whatman No. 1 paper, followed by staining with silver nitrate.
Determination of the D or L Configuration of Galactose-The enantiomeric form of the galactose was assigned based on analysis of the trimethylsilylated (Ϫ)-2-butyl glycosides, as described (31). The polysaccharide from B. occidentalis (1 mg) was mixed with 0.5 ml of (Ϫ)-2-butanol, 1 M HCl (Aldrich). After butanolysis for 18 h at 80°C, the solution was neutralized with Ag 2 CO 3 , the supernatant concentrated, and dissolved in 50 l. Thereafter we added 50 l of bis(trimethylsilyl) trifluoro acetamide (Sigma) and kept the solution for 30 min at room temperature. The butanolyzed and trimethylsilylated derivatives were analyzed on a DB-5 GLC column. The temperature was programmed from 120 to 240°C at 2°C/min. The injector and detector temperatures were 220 and 260°C, respectively. Appropriate controls of trimethylsilylated (Ϫ)-2-butyl-D-and L-galactosides were analyzed under the same conditions.
Desulfation and Methylation of the Galactan-Desulfation of the sulfated galactan was performed as described previously (4,8). About 20 mg of each polysaccharide were dissolved in 5 ml of distilled water and mixed with 1 g (dry weight) of Dowex 50-W (H ϩ , 200 -400 mesh). After neutralization with pyridine, solutions were lyophilized. The resulting pyridinium salts were dissolved in 2.5 ml of dimethyl sulfoxide/ methanol (9:1, v/v). The mixtures were heated at 80°C for 4 h, and the desulfated products were exhaustively dialyzed against distilled water and lyophilized. The extent of desulfation was estimated by the molar ratio of sulfate/total sugars. This method allows us to detect desulfation up to a molar ratio Յ0.1 sulfate/total sugar. About 5 mg of each desulfated fraction was obtained. The native and desulfated galactans were subjected to three rounds of methylation, as described (32) with the modifications suggested by Patankar et al. (33). The methylated polysaccharides were hydrolyzed with 6.0 M trifluoroacetic acid for 5 h at 100°C, reduced with borohydride, and the alditols were acetylated with 1:1 acetic anhydride/pyridine (30). The alditol acetates from the methylated sugars were dissolved in chloroform and analyzed in a Hewlett-Packard gas chromatography/mass spectrometry unit, model 5987-A. Injection was made in the splitless mode in a DB-1 capillary column (25 m ϫ 0.3 mm). The column was programmed to run at 120°C for 2 min, then raised to 230°C at 2°C/min, and held for 5 min.
NMR Spectroscopy-1 H and 13 C spectra were recorded using a Bruker DRX 600 with a triple resonance probe. About 3 mg of each sample was dissolved in 0.5 ml of 99.9% D 2 O (CIL). All spectra were recorded at 60°C with HOD suppression by presaturation. COSY, TOCSY, and 1 H/ 13 C heteronuclear correlation (HMQC) spectra were recorded using states-time proportion phase incrementation for quadrature detection in the indirect dimension. TOCSY spectra were run with 4,096 ϫ 400 points with a spin-lock field of about 10 KHz and a mixing time of 80 ms. HMQC were run with 1,024 ϫ 256 points and globally optimized alternating phase rectangular pulses for decoupling. NOESY spectra were run with a mixing time of 100 ms. All chemical shifts were relative to external trimethylsilylpropionic acid and 13 C-methanol. Anticoagulant Action Measured by APTT-Activated partial thromboplastin clotting assays were carried out by the method of Anderson et al. (34). Normal human plasma (100 l) was incubated with 10 l of a solution of polysaccharide (0.05-5 g) at 37°C for 1 min. Then 100 l of activated partial thromboplastin time reagent (Celite Biolab) 1 We extracted sulfated polysaccharides from 54 species of marine alga and tested their anticoagulant activity using clotting assays. The sulfated polysaccharide from B. occidentalis showed the highest anticoagulant activity among the species tested and therefore it was chosen for a detailed structural and anticoagulant analysis. The results of this preliminary screening analysis will be published elsewhere. 2 The abbreviation used is: FPLC, fast protein liquid chromatography; APTT, activated partial thromboplastin time; COSY, correlation spectroscopy; TOCSY, total correlation spectroscopy; NOESY, nuclear Overhauser effect spectroscopy; HMQC, 1 H/ 13 C heteronuclear multiple quantum coherence spectra. were added and incubated at 37°C. After 2 min of incubation 100 l of 0.25 M CaCl 2 were added to the mixtures and the clotting time recorded in a coagulometer (Amelung KC4A). The activity was expressed as international units/mg using a parallel standard curve based on the 4th International Heparin Standard (193 international units/mg).
Effect of Sulfated Polysaccharides on the Inactivation of Thrombin and Factor Xa-These effects were evaluated by the assay of amydolytic activity of thrombin or factor Xa using chromogenic substrate, as described (35,36).
Effect of Sulfated Polysaccharides on the Inactivation of Thrombin by Antithrombin-Sulfated polysaccharide solution (10 l) and 5 l of 1 unit/ml purified human antithrombin were mixed with 19 l of 10 units/ml human purified thrombin in 66 l of 0.015 M Tris-HCl buffer (pH 7.4) containing 0.15 M NaCl and 1 mg/ml polyethylene glycol (Sigma). After a 1-min incubation (inhibition period), 500 l of 0.24 mM chromogenic substrate S-2238 from Chromogenix AB (Molndal, Sweden) was added, and the remaining thrombin activity recorded for 2 min at 405 nm.
Effect of Sulfated Polysaccharides on the Inactivation of Thrombin by Heparin Cofactor II-This assay was just as described above except that heparin cofactor II (100 g/ml, from Diagnostica Stago, Asnièris, France) instead of antithrombin was added to the incubation mixtures.
Effect of Sulfated Polysaccharides on the Inactivation of Factor Xa by Antithrombin-Sulfated polysaccharide solution (10 l) and 5 l of 1 unit/ml purified human antithrombin were mixed with 7 l of 4 units/ml purified bovine factor Xa (Chromogenix Molndal, Sweden) in 80 l of 0.015 M Tris-HCl buffer (pH 7.4) containing 0.15 M NaCl and 1 mg/ml polyethylene glycol. After a 1-min incubation (inhibition period), 500 l of 0.24 mM chromogenic substrate S-2222 from Chromogenix AB (Molndal, Sweden) was added and the remaining factor Xa activity recorded for 2 min at 405 nm. In the incubation periods used, no inhibition occurred when thrombin or factor Xa was incubated at 37°C with the sulfated polysaccharides or with the cofactors (antithrombin or heparin cofactor II) alone.

Purification and Structural Analysis of the Sulfated Galactans from the Red Algae B. occidentalis
Purification of the Algal Sulfated Galactans-Anion exchange chromatography on Mono Q-FPLC separated the sulfated polysaccharides from red algae into three major peaks F1, F2, and F3, eluted from the column with ϳ1.0, ϳ2.2, and ϳ3.0 M NaCl, respectively ( Fig. 2A). All three fractions had strong metachromasia produced with 1,9-dimethylmethylene blue (open circles), and high hexose content, as revealed by the method of Dubois et al. (28) (closed circles). Agarose gel electrophoresis analysis revealed an increased mobility from F1 to F3 (Fig. 2B). Chemical analysis of the purified fractions showed galactose as the only sugar constituent and an increasing sulfate content from F1 to F3 (Table I). The trimethylsilylated (Ϫ)-2-butyl galactosides obtained from the algal polysaccharides have the same retention times and peak area proportions on GLC column as standard D-galactose (not shown). Therefore, galactose occurs on the B. occidentalis galactan as a D-enantiomer. Overall, these results indicate that the red algae B. occidentalis contains three fractions of sulfated D-galactans that differ in their sulfate ester contents.
Fractions F2 and F3 Contain Approximately Equimolar Proportions of 3-and 4-Linked Units-When the three fractions of B. occidentalis-sulfated galactans were submitted to three rounds of methylation, a variety of methylated derivatives were obtained, mainly mono-and dimethylated derivatives (not shown). The methylation data were not consistent with known polysaccharide structures. Despite the observation that the proportions of methylated derivatives remain unchanged after an additional methylation cycle, we probably still have incomplete reaction. In fact, methylation of sulfated polysaccharides does not always yield reliable proportions of methylated alditols (4, 33, 37, 38). This may be a consequence of steric hindrance due to the sulfate esters, which does not allow complete methylation of these polymers. The more drastic conditions necessary to remove sulfate esters may also destroy some of the methylated derivatives.
An alternative method to obtain information about the structure of these polymers is methylation of the desulfated polysaccharide. In fact methylation of the desulfated F2 and F3 fractions yields almost similar proportions of 2,3,6-tri-O-and 2,4,6tri-O-methylgalactose, indicating 4-linked and 3-linked galactose residues, respectively (Table II). Small amounts of dimethyl derivatives were also obtained, especially from desulfated F2. They are not merely products from an incomplete reaction since their proportions remain unchanged after an additional methylation cycle (not shown). Possibly these derivatives come from minor structural components and/or from units which were not desulfated. Desulfated F2 and F3 still contain ϳ0.18 and ϳ0.08 sulfate/sugar unit, respectively, which are close to the limit for detection of sulfate by the method we used (ϳ0.1 sulfate/sugar, as molar ratio).   Fig. 2. b Hexose was identified by its retention time on a gas chromatography/mass spectrometry unit of derived alditol acetate. The concentrations of hexose and sulfate on the acid hydrolysates obtained from the polysaccharides were determined by the phenol/H 2 SO 4 and BaCl 2 /gelatin reactions, respectively. c See experiment shown in Fig. 2A.
Methylation of desulfated F1 fraction revealed a more complex mixture of methyl derivatives, which are not consistent with known polysaccharide structures. We made several attempts to methylate this polysaccharide using different rounds of the reaction and a variety of hydrolysis procedures to prepare the methylated derivatives. None of these attempts yielded reliable proportions of methyl derivatives. We cannot explain the difficulty in obtaining reliable proportions of methylated alditols from this polysaccharide. However, preliminary qualitative information can be obtained about the structure of this compound. In contrast with the other two fractions reported in Table II, desulfated F1

NMR Analysis Reveals Alternating ␣(133) and ␤(134) Units with a Variable Sulfation Pattern
We employed NMR analysis to confirm the saccharide backbone structure of fractions F2 and F3 and also to determine the sulfation pattern of these polysaccharides. This last aspect was not possible to approach by methylation analysis, as discussed above.
The 1 H one-dimensional spectra of the native and desulfated galactan from B. occidentalis (fraction F2) are shown in Fig. 3. The chemical shifts in Tables III and IV are based on the interpretations of TOCSY, COSY, and HMQC spectra.
The desulfated galactan shows two main anomeric resonances, one at 5.2 ppm (␣ unit) and another at 4.4 ppm (␤ unit) (Fig. 3B). Peak integration demonstrates ϳ1:1 ratio. The ␣ and ␤ spin systems can be traced based on TOCSY (Fig. 4B), COSY (not shown), and 1 H/ 13 C HMQC (Fig. 5B) spectra, giving the values presented in Table III and IV (c and d). The C4 of ␣-galactose and C3 of ␤-galactose residues show strong downfield shifts (ϳ10 ppm) indicating that the two residues are 4and 3-linked, respectively (Table IV). Linkage information was also obtained in the NOESY spectrum (Fig. 6). A strong interresidue NOE is seen between ␤-H1 and ␣-H4 and from ␣-H1 and ␤-H3, as expected for ␣133 and ␤134 linkages. NOEs also appear from the anomeric protons that are close in space to the linkage, but their intensity is smaller. These results are compatible with a polysaccharide with the following repeating structure: -4-␣-D-Galp-133-␤-D-Galp-13 (Fig. 7).
The native polysaccharide has a very complex 1 H NMR spectrum (Fig. 3A) due to its heterogeneous sulfation pattern. At least six distinguishable anomeric resonances were observed.
Three of them (A, B, and C) are between 5.6 and 5.2 ppm in agreement with ␣-anomeric protons while the three other (D, E, and F) are between 4.66 and 4.55 as expected for ␤-anomers. Integration of the two groups of anomers gave a 1:1 ratio.
Two-dimensional assignment techniques (TOCSY and COSY) were used to trace the spin systems but only two of them could be partially identified. Residue A is both 2-O and 3-O sulfated as seen by the unambiguous cross-peaks in the TOCSY spectrum (Fig. 4A). Positions 2 and 3 of residue A are downshifted from the desulfated value both in the 1 H (Ϫ0.88 and Ϫ0.63 ppm, respectively) and 13 C dimension, confirming sulfation in these two positions. Residue B is also sulfated at position 2, since it shows a 1 H downshift of Ϫ0.77 ppm. The chemical shifts are presented in Tables III and IV. For residues C, D, E, and F, no unambiguous cross-peak could be found. In this region several overlaps exist hampering the assignment strategy.
The integrals of the three ␣-H1 resonances in the 1 H spectrum suggest a ϳ1:1:1 ratio of residues A:B:C. These integrals are derived from poorly resolved signals, and thus require careful interpretation. Nevertheless, they suggest that 2,3-di-O-and 2-O-sulfated residues account for one-third each of the total ␣-units in the sulfated D-galactan from B. occidentalis. The possible occurrence of 3,6-anhydrogalactose, a common component of algal polysaccharides, was excluded in the B. occidentalis galactan due to the absence of strong downshift of H6 in this type of residue (compare value for H6 in c and d with literature values for 3,6-anhydrogalactose in n and o, Table III).
Fraction F-3 shows a 1 H NMR spectrum similar to that obtained for F2 but with slight differences in the proportions of the various signals. In contrast, F1 has a much more complex spectrum which could not be resolved with the two-dimensional techniques.
In conclusion, methylation and NMR analyses indicate that fractions F2 and F3 obtained from B. occidentalis are linear polysaccharides, containing alternating residues of ␣(133)and ␤(134)-D-galactopyranose. A variable sulfation pattern  confers high heterogeneity to these polysaccharides. Nevertheless, it is clear that 2,3-di-O-sulfated ␣-D-galactopyranose residues occur as ϳ30% of the total ␣ units of these galactans (Fig. 7). It was not possible to determine the structure of fraction F1. For this polysaccharide even high field NMR is at the limit of its powers and complete methylation was not achieved.

Anticoagulant Action of Sulfated Galactans and Sulfated
Fucan from the Red Algae B. occidentalis and from Marine Invertebrates    Tables III and IV are relative to external trimethylsilylpropionic acid at 0 ppm for 1 H and methanol for 13 C. The anomerics were identified by the characteristic carbon chemical shifts. fied sulfated D-galactans from the red algae B. occidentalis were measured, and compared with the same activities of sulfated L-galactans and a sulfated L-fucan from marine invertebrates, which have well defined structures (Fig. 1).
The APTT assay (summarized in Table V) indicates that crude B. occidentalis polysaccharides have anticoagulant action (Table V, a). Purification of fractions F2 and F3 results in increased anticoagulant potency (Table V, c and d) comparable with that of unfractionated heparin (j). Fraction F1 has only a mild effect (Table V, b).
Comparison with sulfated polysaccharides from inverte-brates 3 shows that the sulfated L-galactan from E. lucunter has a significant anticoagulant action (Table V, (Table V, g). Nevertheless, the sulfated galactans from B. occidentalis (fractions F2 and F3) are much more potent as anticoagulant than sulfated galactans/fucans from invertebrates. Possibly, this potent effect is related to the occurrence of 2,3-di-O-sulfated galactose units, since this type of unit is absent in the other galactans/fucan tested. The sulfate content is important for anticoagulant action, since the desulfated galactan (Table V, e) lost the activity. However, it is not the only requirement since highly sulfated polysaccharides such as dextran sulfates have lower anticoagulant activity than F2 and F3 (Table V, l and m). In addition, fractions F2 and F3 differ in sulfate content (Table I)     approximately the same potency as anticoagulant.
Thus, the data in Table V indicate marked differences in the anticoagulant effect of closely related sulfated galactans. This is not solely due to the sulfate content, but also to distinct structure and/or sulfation patterns of the polysaccharides.

Sulfated D-Galactans from B. occidentalis Enhance Thrombin and Factor Xa Inhibition by Antithrombin and/or Heparin
Cofactor II-The sulfated D-galactan from B. occidentalis (fraction F3) enhances thrombin inhibition by antithrombin with an IC 50 similar to that of unfractionated heparin (Fig. 8A). Replacement of antithrombin by heparin cofactor II results in a shift to the right of the F3 effect for thrombin inhibition, but this polysaccharide still has a IC 50 15-fold lower than that of mammalian dermatan sulfate (Fig. 8B). Finally, replacement of thrombin by factor Xa also decreases the inhibitory effect of fraction F3, compared with unfractionated heparin (Fig. 8C).
Similar experiments were done with other fractions of sulfated D-galactans from B. occidentalis (Table VI, a-c). Both F2 and F3 fractions are potent thrombin inhibitors in the presence of antithrombin and heparin cofactor II. F3 is more potent, especially for thrombin inhibition in the presence of antithrombin. In contrast, fraction F1 is only a weak inhibitor.
Direct thrombin inhibition by highly branched sulfated fucans from the brown algae Laminania brasiliensis and Fucus vesiculosus was recently reported (15), but in this case the amydolytic assays were carried out at 25°C. Under the same conditions the sulfated galactans from B. occidentalis also directly inhibit thrombin activity (not shown).
For comparison, we extended this set of experiments to the sulfated L-galactans/L-fucan from invertebrates (Table VI, d-f). The sulfated L-galactan from E. lucunter inhibits thrombin activity in the presence of either antithrombin or heparin cofactor II, but requires much higher concentrations than fractions F2 and F3 from B. occidentalis. Replacement of ␣-Lgalactose by ␣-L-fucose (as in the polysaccharide from S. franciscanus; Table VI, f), sulfation in a different position, or the position of the glycosidic linkage (as in H. monus; Table VI, d), abolishes the inhibitory effect on thrombin.
The structural feature distinguishing B. occidentalis sulfated D-galactans is the occurrence of 2,3-di-O-sulfated D-galactose units, which are absent in the other galactans or fucan tested in the experiment shown in Table VI. Possibly, this type of residue accounts for the potent thrombin inhibitory effect observed with these galactans.

Major Conclusions
Our results indicate that the red algae B. occidentalis contains two fractions (F2 and F3) of regular sulfated D-galactans  composed of alternating ␣(133) and ␤(134) residues, but with a heterogeneous sulfation pattern. Clearly, 2,3-di-O-sulfated and 2-O-sulfated units account for approximately one-third each of the total ␣-residues. The structure of fraction F1 could not be determined. Fractions F2 and F3 are potent anticoagulant polysaccharides due to enhanced thrombin/factor Xa inhibition by antithrombin and heparin cofactor II. Comparison with several well defined sulfated polysaccharides from marine invertebrates allows determination of some structure/biological activity relationships. Thus, galactans sulfated at either 2-O or 3-O positions of ␣-galactose residues (but not at both positions) have weak anticoagulant activity compared with B. occidentalis galactan. Presumably, the addition of two sulfate esters to a single ␣-galactose has an amplifying effect on anticoagulant activity. This is not merely a consequence of increased charge density. The anticoagulant activity increased 15-fold when comparing the 2-O-sulfated galactan of E. lucunter with B. occidentalis polysaccharide, fraction F2 (measured either as clotting time, Table V; or as IC 50 for thrombin inhibition,  Table VI). On the other hand, the equivalent increase in sulfate content is ϳ50% (Table I). In addition, highly sulfated dextrans (ϳ3 sulfate esters per glucose unit) have lower anticoagulant potency than the B. occidentalis galactan (Table V). Finally, the three linear and repetitive polysaccharides from invertebrates ( Fig. 1, A-C) show dramatic differences in anticoagulant activity, despite their similar charge density. Therefore, the structural requirements for interaction of sulfated polysaccharides with coagulation cofactors are stereospecific. In agreement with this structural stringency, oversulfated dermatan sulfates show only discrete, selected sites competent for interaction with heparin cofactor II (14). A similar observation was extended for sulfated fucans (16).
The conformational analysis of these sulfated polysaccharides is an important route to follow. The differences in chemical structure may in fact determine spacing between sulfate groups required to match the interval between basic amino acid residues in the protein chain. Conformational analysis may explain the drastic differences in biological activity between a sulfated galactan and a sulfated fucan despite the same positions of sulfation and glycosidic linkage. Similarly, changes in biological activity may reflect dramatic modifications in the conformation of the polysaccharide as a consequence of the 2,3-di-O-sulfated ␣-galactose units.
Our results indicate that combining structural analysis of sulfated polysaccharides from marine algae and invertebrates with specific biological assays is a useful tool to investigate anticoagulant activity in mammals. These studies may help to delineate a closer relationship between structure and anticoagulant activity of sulfated polysaccharides, as already reported for heparin. New compounds with obvious practical applications may be found. Finally, the sulfated galactans from B. occidentalis are natural candidate molecules for testing in experimental thrombosis.