The structures of the O-acylated derivatives were further studied by ¹H and ¹³C NMR spectroscopy. The 1D and 2D NMR spectra of the original dFuCS and its synthesized O-propionoyl derivative were shown in Figure 1 and Figure 5. The chemical shifts in Table 2 were based on the interpretations of the COSY spectra (Figure 5).

In the Figure 1C,D, two signals were present at 1.0–1.2 ppm and 2.3–2.6 ppm region, corresponding to the methyl and methylene protons of O-propionoyl FuCS. Interestingly, the H2 and H3 signals of β-D-glucuronic acid residue after the O-acylated process are at 3.82 ppm and 3.91 ppm (Figure 1, Table 2), respectively, shifted downfield by 0.2~0.3 ppm, according to trace the corresponding spin systems using the ¹H-¹H COSY (Figure 5). The result suggested that this may be attributed to effects of carbonyl of the propionoyl group at 2-position of β-D-glucuronic acid residues (GlcUA2Pr).

In addition, the ¹H-¹H COSY spectrum (Figure 5) showed that three signals with ~6.0–6.1 ppm in 1D NMR (Figure 1C) can be readily assigned to the proton H-4 of 4,5-unsaturated glucuronic acids (ΔUA). The proportion ∆4,5-glucuronic acid residues in the non-reducing end to fucose residues was about 7:100, according to integration of H4 resonances of the ΔUA and H1 resonances of the fucose residues in the one-dimensional spectrum (Figure 1C). The results suggested that the degradation may take place by β-elimination mechanism.

In the ¹H spectrum (Figure 1), the signals of synthesized O-propionoyl derivatives are not easy to find due to overlap of H signals among three sugar residues, but the carbonyl carbon signals is easily seen in the ¹³C spectrum. The carbonyl carbon signal of ester linkage was observed at the shift of ~178–179 ppm. There are signals at ~10.7 and ~29.9 ppm, respectively, corresponding to the methyl and methylene carbons of the propionoyl residue. Additionally, the signals at 140–150 ppm, ~110 ppm can be assigned to the unsaturated C of 4,5-unsaturated glucuronic acids, and C1 in their reducing ends, respectively. The data further confirmed that β-elimination had taken place.

Overall, comparison with the structures of their parent compound indicated that the structural characters of synthesized derivatives may not significantly change, such as the structure of the polyanionic backbone, but the degradation took place by the β-elimination mechanism.

Confirmation of acylated substitution at position is not easy due to the complex structure of the dFuCS with distinguishable patterns and proportions of sulfate substitution in the fucose branches [1,2,21]. However, priority of substitution of different OH in sugar unit could be elucidated. The chemical shifts of the anomeric protons of the 2,4-di-sulfated fucose (α-L-Fuc2S4S), 3-, and 4-mono-O-sulfated fucose (α-L-Fuc3S, α-L-Fuc4S) residues can be readily assigned according to comparison of the signals before and after the modification (Figure 5).

In the ¹H-¹H COSY spectrum of the O-propionoyl dFuCS with ~40% degree of substitution (Figure 5B), the chemical shifts of the anomeric protons of the α-L-Fuc2S4S and α-L-Fuc3S residues are 5.68 and 5.38 ppm (Table 2, Figure 5A), respectively, which were in good agreement with those of original dFuCS in our previous study [2]. In addition, the others signals of the two kinds of fucose residues were not shifted, which clearly indicated that the OH groups of α-L-Fuc2S4S and α-L-Fuc3S residues were not introduced into propionoyl groups. In other words, the 3-OH groups of the α-L-Fuc2S4S residues and the 2-OH, 3-OH groups of the α-L-Fuc3S residues may be difficult to be acylated, possibly due to their special sulfation patterns. However, after the modification, the H-2 signal of the α-L-Fuc4S residues (δ_H-2 = 4.34) was shifted downfield by ~0.5 ppm, which resulted from the propionoyl groups at the 2-position of α-L-Fuc4S residues (α-L-Fuc2Pr4S). Additionally, in the Figure 5B, the signals of the propionoyl groups at the 4-position of α-L-Fuc2S residues (α-L-Fuc2S4Pr) were observed, although the anomeric protons of α-L-Fuc2S and α-L-Fuc2S4S residues before the modification were overlap to not be discerned. Partial acylation of 2-OH of β-D-GlcUA residue in the chondroitin sulfate-like backbones took also place, but the degree of substitution were slight. These results suggested that the propionoyl groups may be preferentially introduced to the OH of the α-L-Fuc4S residues, slightly to those of α-L-Fuc2S in the fucose branches under above experimental conditions (Figure 5B).

However, the COSY spectrum of the O-propionoyl dFuCS with ~95% degree of substitution (Figure 5C) showed that the signals of Fuc and GlcUA residues became more complex. These H-2 and H-3 signals of GlcUA residues in the product were shifted downfield by ~0.2 ppm, which indicated that their 2-OH may be completely acylated. Besides remaining of OH in minor Fuc2Pr4S and Fuc2Pr3S, others hydroxyl groups were acylated, whose corresponding signals were significantly shifted downfield (Figure 5C).

Anticoagulant activity of O-acylated derivatives of the dFuCS from the sea cucumber T. ananas were assessed by measuring the activated partial thromboplastin time (APTT), and compared with the same activities of the dFuCSs, unfractionated heparin, and a low molecular weight heparin (LMWH, enoxaparin) from mammalian. The APTT assays (summarized in Table 1, Figure 6) indicated that five control dFuCS have significant anticoagulant activity (Figure 6), comparable with that of LMWH (32 U/mg), but lower than that of unfractionated heparin (212 U/mg). The Figure 6 showed that the anticoagulant activity varies of the five control dFuCS in proportion to the molecular weight following a logarithmic-like function according to an empirical formula [3]. The TT and PT assays were also carried out but none of the dFuCS and its O-acylated derivatives showed any effects on TT and PT at the concentrations (4–64 μg/mL) tested. The results were consistent with that from previous studies [2,3].

Interestingly, compared with the activity of control dFuCS with the similar molecular mass, partial O-acylation of the dFuCS with about 40% degrees of substitution did not affect their anticoagulant potency (Figure 6), although the dFuCS were substituted with different groups, such as acetic anhydride (C2), propionic anhydride (C3) and succinic anhydride (C4). Furthermore, comparison of the NMR data before and after the modification suggested that the O-acylation of 2-OH groups of 4-O-sulfated fucose units did not affect its anticoagulant activity. However, as the degree of O-acylation of the dFuCS increased, the activity decreased (Table 1, Figure 6). At 95% degree of O-acylation, the product had much low anticoagulant potency (<5 U/mg), about lower 2-fold than the dFuCS with similar molecular mass (Figure 6). But it is still difficult to completely rule out other causes such as molecular size, since the dFuCS with the low molecular weight (<4000 Da) may also have very low anticoagulant activity [3].

A low molecular weight fragment of glycosaminoglycan from the sea cucumber T. ananas was prepared by free radical depolymerization [4]. The O-acylated derivatives of this depolymerized fragments (dFuCS) were prepared by treating their tetrabutylammonium salts with several anhydrides, such as acetic anhydride (C2), propionic anhydride (C3) and succinic anhydride (C4), according to modification of the method described previously [12].

In order to establish the protocol, several experiments were carried out as Table 1. The conditions of O-acylation are as follows, briefly, taking the O-propionoyl acylation as an example. The tetrabutylammonium salt (180 mg) was dissolved in anhydrous N,N-dimethylformamide (DMF), cooled to 0 °C, then 4-(N,N-dimethylamino) pyridine (DMAP) (9 mg), propionic anhydride (200 μL), and tributylamine (360 μL) were successively added in single portion. And the reaction was allowed to proceed under nitrogen at 40 °C for 48 h. After cooling to 0 °C, 5% NaHCO₃ was gradually added to remove the superfluous anhydride, and the solution was stirred at 25 °C for another 48 h. Excess NaHCO₃ was eliminated by the addition of 1 mol/L HCl until a pH 4.0 was reached, and then readjusted to pH 7.0 with 1 mol/L NaOH. 20% Cold absolute ethanol (v/v) was added with stirring. The sample was allowed to sit overnight at 4 °C to afford precipitate. The precipitate was recovered by centrifugation, which was dissolved in water, and passed through a column 35 × 2.5 (170 mL) of Dowex 50W × 8 (H⁺) cation exchange resin and 60 mL eluent was recovered. The acid was neutralized to pH 7.0 with 0.1 M NaOH. The raw products were purified by dialysis with MWCO 1000 Da against distilled water for 48 h. After lyophilization, the O-propionyl FuCS derivative (65 mg) was obtained as an off-white powder. In each case, selective O-acylation was performed in the presence of different carboxylic acid anhydrides with a catalytic amount of DMAP as the above process.

In order to investigate the changing of molecular weight during the process of acylation reaction, the molecular weight (weight-average molecular mass, Mw) of the products during the course of the reaction were examined by high-performance gel permeation chromatography (HPGPC) using an Agilent technologies1200 series (Agilent Co., USA) apparatus equipped with a Shodex OH-pak SB-804 HQ column (8 mm × 300 mm) as described previously [4] .Specifically, the molecular weights of products were calculated using the standard curve by GPC software, which was dependent on the elution times of standard dextran oligosaccharides.

The ultraviolet (UV) analyses were performed with an UV-2450 UV-Vis spectrophotometer (Shimadzu, Japan). All samples were dissolved in purified water at a ~0.8 g/L concentration.

NMR analyses were performed at 20 °C with a Bruker Avance 500 spectrometer of 500 MHz equipped with ¹³C/¹H dual probe in the FT mode as described previously [22]. All lyophilized samples were dissolved in deuterium oxide (D₂O, 99.9% D) at a 10–15 g/L concentration. The NMR experiments were recorded with a spectral width of 12,335.5 Hz, an acquisition time of 1.33 s, a pulse width of 7.8 s, a relaxation time of 2 s and a number of 16 scans. The HOD signal was presaturated by a presaturation sequence. All chemical shifts are relative to internal trimethylsilyl-propionic acid (TSP) sodium salt.

The activated partial thromboplastin time (APTT), prothrombin time (PT), and thrombin time (TT) were determined with a coagulometer (TECO MC-4000, Germany). All the reagents were purchased from TECO GmbH (Germany). PT reagent, TT reagent, and normal human plasma were reconstituted in 4 mL (or 1 mL for plasma) of distilled water, according to the instructions of the manufacturer. The glycosaminoglycans were dissolved in 20 mM Tris-HCl (pH 7.4) at various concentrations. For the APTT assay, 5 μL samples were mixed with 45 μL of normal human plasma, and incubated for 2 min at 37 °C. Then 50 μL of APTT reagent was added to the mixture, which was incubated for another 3 min at 37 °C. CaCl₂ (50 μL) was then added, and the clotting time was recorded. For the PT and TT assays, 5 μL samples (or 10 μL for TT) were mixed with 45 μL (or 90 μL for TT) of normal human plasma and incubated for 2 min at 37 °C; 100 μL of PT (or 50 μL of TT) reagent was then added and the clotting time was recorded.