Heparin microarray-based specific interaction assay of marine sulfated polysaccharides with antithrombin Ⅲ

Abstract

Marine sulfated polysaccharides (SPs) are promising candidates for new anticoagulant drug development, necessitating efficient method to evaluate their anticoagulant activity at the molecular level. In this work, a heparin microarray-based competitive strategy was developed to investigate the specific interactions between SPs and antithrombin Ⅲ (AT). The strategy is based on the principle that SPs competitively bind to the active domain of AT, reducing the signal of AT binding to immobilized heparin on microarrays. Validation was performed using established anticoagulants heparin and enoxaparin as model analytes. After optimizing key experimental conditions, the method successfully determined the IC₅₀ values for three SPs: fucoidan derived from Ascophyllum nodosum (AnF, 50.55 ± 2.79 μg‧mL⁻¹), fucosylated chondroitin sulfate derived from Holothuria tubulosa (FCS_ht, 44.18 ± 4.05 μg‧mL⁻¹), and its selectively degraded sulfation product (S-dFCS_ht, 16.99 ± 6.56 μg‧mL⁻¹). The reliability of these results was confirmed by surface plasmon resonance assay. The strategy’s versatility was further demonstrated by assessing SP interactions with side effect-related proteins, providing valuable insights into both efficacy and safety profiles. Although currently applicable primarily to heparin-interacting anticoagulant targets, this strategy can be extended to other targets by fabricating specialized glycan microarrays, enabling comprehensive evaluation of SP interactions with various anticoagulant targets.

Marine sulfated polysaccharides hold promise for anticoagulant drug development, yet efficient evaluation methods are lacking. Here, the authors develop a heparin microarray-based competitive strategy to assess interactions with antithrombin III, successfully determining IC₅₀ values and demonstrating versatility for broader anticoagulant target evaluation.

Introduction

Heparin is a linear sulfated polysaccharide (SP) and composed of the repeating disaccharide units of 1,4-linked uronic acid (α-L-iduronic or β-D-glucuronic acid) and D-glucosamine¹. It has been used as a clinical anticoagulant since the 1930s. Heparin exerts its anticoagulant activity through a unique pentasaccharide sequence that specifically binds to antithrombin Ⅲ (AT), significantly enhancing its activity². AT is the primary natural inhibitor of the coagulation, binding to two pivotal proteins in coagulation cascade, thrombin and Factor Ⅹa, to form complexes that are rapidly degraded by the circulation system. Due to heparin promoting the activity of AT by approximately 1000 times, it is widely used as an efficient anticoagulant drug³. However, heparin is associated with several side effects, including thrombocytopenia and an increased risk of bleeding^4,5. Furthermore, heparin is mainly extracted from pig intestines and bovine lungs, the risk of prion-related diseases and land-sourced contamination limits its use⁴. Marine SPs, which are extensively distributed in various marine organisms (from seaweeds and algae to fish, sea cucumbers, shrimp, and sea urchins), are vital resources for discovering and developing new anticoagulant drugs^6,7. Marine SPs with demonstrated anticoagulant activity include glycosaminoglycans (GAGs) such as chondroitin sulfate, dermatan sulfate, heparin, heparan sulfate, and fucosylated chondroitin sulfates (FCSs), as well as GAG-like polysaccharides such as sulfated fucans and sulfated galactans. For example, FCS_hm isolated from Holothuria mexicana, possesses a unique structural motif: a fucosyl branch at the O6 position of N-acetylgalactosamine (GalNAc), in addition to the typical fucosyl unit at the O3 of Glucuronic Acid (GlcA). The GlcA-linked fucose displayed Fuc2S4S or Fuc4S sulfation, whereas the GalNAc-linked fucose exhibits Fuc4S and Fuc3S4S patterns. It demonstrates anticoagulant activity by effectively inhibiting AT-mediated anti-FIIa and anti-FXa functions⁸. Other SPs, including alginates, sulfated rhamnan, and chitins also exhibited anticoagulant activity. For instance, propylene glycol alginate sodium sulfate, which is a sulfated derivative of alginic acid, possesses potent anticoagulant activity⁹. Nevertheless, the anticoagulant activity of SPs is related to their charge density and structure, which means that not all SPs possess anticoagulant activity. Therefore, establishing efficient methods to evaluate the anticoagulant activity of various SPs is crucial for accurately identifying the unique structures in specific organisms that are associated with the desired anticoagulant activity.

As known, the coagulation cascade comprises extrinsic, intrinsic, and common pathways, involving a series of sequentially activated coagulation factors. Thrombin and factor Xa, located at key junctions of the coagulation cascade, are common targets for anticoagulant drug development⁵. Anticoagulants can be categorized as indirect or direct inhibitors based on their dependence on AT. Studies indicate that many marine SPs, such as fucoidan and chondroitin sulfate, primarily exert indirect anticoagulant effects by binding to serine protease inhibitors (such as AT and heparin cofactor Ⅱ), significantly enhancing their activity². However, anticoagulation achieved by inhibiting thrombin and/or Xa activity inevitably interferes with physiological hemostasis, leading to a significant risk of bleeding. With advances in coagulation research, novel anticoagulant compounds that directly target specific factors within the intrinsic pathway have been continuously discovered. For instance, sulfated glycosaminoglycan mimetics exert anticoagulant effects by inhibiting coagulation factor XIa⁵. Thus, evaluating the anticoagulant activity of marine SPs involves multiple targets, necessitating the establishment of a versatile, high-throughput evaluation method to efficiently assess their anticoagulant properties at the molecular level.

Conventional methods for the study of SP-involved molecular interactions typically based on enzyme-linked immunosorbent assay (ELISA), such as the reported SP-coating-ELISA¹⁰. While ELISA-based methods offer high throughput and specificity, the lack of enzyme-labeled primary or secondary antibodies in some cases hinders their applicability. In recent years, label-free methods such as surface plasmon resonance (SPR) and biolayer interferometry (BLI) have been developed to evaluate the interactions between SPs and AT, as well as other immunogenicity-related proteins^2,11. However, the aforementioned methods require expensive equipment and skilled operators, resulting in high operational costs for investigating SP-involved molecular interactions. Carbohydrate microarrays, featuring high-throughput detection and the ability to employ various detectors for carbohydrate-mediated binding analysis, have become cutting-edge tools for studying carbohydrate-protein interactions^12–14. They are prepared by adopting different noncovalent or covalent immobilization approaches to site-specifically or site-nonspecifically attach carbohydrates to solid surfaces with regulated density and orientation¹⁵. Carbohydrate microarrays have been widely used in biological research. For instance, Linhardt and colleagues constructed a three-dimensional heparin microarray chip with a high signal-to-noise ratio by covalently immobilizing heparin on poly-L-lysine layers, in order to assess the affinity of heparins to AT¹⁶. In our previous work, the poly(2-hydroxyethyl methacrylate) (pHEMA)-cyanuric chloride (CC)-modified glass substrates were adopted for covalent immobilization of natural carbohydrates including monosaccharides, oligosaccharides, polysaccharides, neutral saccharides, acidic saccharides, and alkaline saccharides¹⁷. The pHEMA-CC-based carbohydrate microarrays were used to directly analyze the interactions between marine algin-derived acidic oligosaccharides and influenza A viral hemagglutinin¹⁸. Despite the already established direct-assay methods were extensible to analyze interactions of target marine SPs and serpins, SPs frequently non-specifically bind to target proteins due to electrostatic adsorption. This leads to false positive results, as the SPs bind to regions other than their specific binding domains.

Using the specific interaction between SPs and AT as a model, we developed a heparin microarray-based competitive assay to characterize their binding in this work. SPs in solution and immobilized heparin on microarrays competed for binding to AT, and the specific binding of SPs to the active domain of AT resulted in the decreased signals of AT binding to heparin on microarrays (Scheme 1). Heparin and enoxaparin, known anticoagulant drugs targeting AT, were used as model analytes to verify the feasibility of the heparin microarray-based competitive strategy. After proper experimental condition optimization, the established method was applied to assess the specific interactions and binding potency of marine SPs with AT by measuring the values of half maximal inhibitory concentration (IC₅₀). Furthermore, the strategy was not confined to the specific interaction assay of SPs with anticoagulant-related proteins, it was also applicable to the interaction assay between SPs and side effect-related proteins, such as heparin-induced thrombocytopenia (HIT)-related protein platelet factor 4 (PF4)¹⁹, in anticoagulant drug development.

Scheme 1.

The heparin microarray-based competitive strategy for investigating the interactions between SPs and AT.

Results and Discussion

Establishment of heparin microarray-based competitive strategy

A competitive strategy for investigating the specific interactions between SPs and AT was designed and realized by adopting heparin microarrays. Heparin was covalently and site-nonspecifically immobilized on pHEMA-CC-modified glass substrates. In order to confirm the specific recognition of immobilized heparin on microarrays with AT was maintained, heparin, dextran, and water were spotted on pHEMA-CC-modified glass substrates to form microarrays. The microarrays were then incubated with AT-His tag and labeled with Ab-His tag-FITC, and strong binding signals were observed only at heparin spots (Fig. 1a). Differently, no signals were detected at water spots, which were used as blank control of spotting solvent, or at the dextran spots, which were used as negative control. When the microarrays were only incubated with Ab-His tag-FITC, no binding signals were detected at any of the spots (Fig. 1b). This indicated that the specific binding of heparin and AT-His tag resulted in the strong binding signals in Fig. 1a. To further confirm the specificity of the binding, the microarrays were sequentially incubated with NA-His tag and Ab-His tag-FITC. As shown in Fig. 1c, no binding signals were observed at any of the spots. It indicated that the binding signals between immobilized heparin and AT-His tag resulted from the specific binding of AT to heparin rather than electrostatic interaction between cationic polyhistidine chains and heparin. These results demonstrated that the binding ability of immobilized heparin on microarrays with AT was highly conserved.

Fig. 1. Heparin microarray-based competitive strategy for investigating SPs and AT specific interactions was feasible.

Fluorescence images of carbohydrate microarray containing heparin, dextran, and water spots before (a1, b1, and c1) and after recognition with AT-His tag/Ab-His tag-FITC (a2), Ab-His tag-FITC (b2), or NA-His tag/Ab-His tag-FITC (c2). The images were in pseudocolor, and the default color was set according to the wavelength of excitation laser. d AT binding ratio on heparin microarrays after AT preincubating with heparin, enoxaparin, or dextran solutions at different concentrations. e IC₅₀ values of heparin or enoxaparin on AT-immobilized heparin binding events. The data were expressed as means ± standard deviation (SD) and determined by t-test. NS not significant; **P < 0.01.

Heparin and enoxaparin, possessing known AT binding ability, were used as model analytes to establish the competitive strategy for investigating SPs and AT specific interactions using heparin microarray. A series of heparin or enoxaparin solutions in different concentration were preincubated with AT-His tag and incubated with heparin microarray, then the AT-His tag bound to heparin microarray was labelled with Ab-His tag-FITC. The binding signals of immobilized heparin on microarrays with remaining AT-His tag decreased as the increase of concentration of heparin or enoxaparin solutions. The typical fluorescence images are shown in Figure S1.

The curves of AT binding ratio formed downward curves along with the increase of the concentration of heparin or enoxaparin solutions (Fig. 1d). Differently, when dextran solutions at different concentrations, used as negative control, were preincubated with AT-His tag and then incubated with heparin microarray, the fluorescence binding signals resulted from immobilized heparin on microarrays and remaining AT-His tag remained basically unchanged after labelled with Ab-His tag-FITC (Figure S1). It indicated that heparin or enoxaparin in solutions competed with immobilized heparin on microarrays to bind to specific binding domains of AT. In this work, the binding ability of target SPs with AT was evaluated by determining IC₅₀ values in AT-immobilized heparin binding events. The IC₅₀ values for heparin in solution was 56.89 ± 7.22 μg‧mL⁻¹, while the IC₅₀ for enoxaparin in solution was 225.30 ± 38.45 μg‧mL⁻¹. The experimental data revealed that heparin exhibited stronger binding potency for AT compared to enoxaparin, which was consisted with reported literature²⁰. These results demonstrated the feasibility of adopting heparin microarray-based competitive strategy to investigate SP-AT specific interactions. It was worth noting that mass concentration was frequently used instead of molar concentration in this work for two primary reasons. First, the structural heterogeneity of polysaccharides precludes the accurate determination of their molecular weights. Second, as the pentasaccharide fragment is known to be responsible for heparin’s anticoagulant activity, our molecular docking results in Supporting Information (SI) confirmed that only specific oligosaccharide fragments (degree of polymerization, DP ≤ 6) could fully enter the binding pocket of the AT active site. Although these key fragments themselves have a molecular weight distribution, the variation is negligible compared to the vast (often orders-of-magnitude) differences in the molecular weights of the different polysaccharides. Therefore, using mass concentration effectively corresponds to a molar concentration normalized to the active oligosaccharide unit, thereby avoiding potential biases in activity assessment caused by disproportionately large molecular weights.

Key conditions

The key factors that impacted the performance of heparin microarray-based competitive strategy for investigating SPs and AT specific interactions were optimized by adopting control variate method. Firstly, the heparin spotting concentration for heparin microarray fabrication was optimized by maintaining constant the concentrations of AT-His tag and Ab-His tag-FITC. The obtained fluorescence binding intensity along with the increase of heparin spotting concentration formed upward asymptotic curves (Fig. 2a). The binding signals of heparin immobilized on microarrays with AT were detectable when the spotting concentration of heparin was 10 μM, and the signals plateaued when the concentration exceeded 1 mM. The density of heparin immobilized on the microarray was determined by wash-off measurement (Figure S2 and Table S1). Increasing the spotting concentration from 0.1 mM to 1 mM resulted in a proportional increase in immobilization density, from 1.09 × 10¹⁵ to 1.02 × 10¹⁶ molecules cm⁻². This tenfold enhancement is accounted for by the three-dimensional spatial structure of the surface-modified pHEMA-CC framework, which provides a high loading capacity. However, the substantial increase in heparin density did not correspondingly enhance the fluorescence signal from heparin-AT binding. This apparent discrepancy is likely due to steric hindrance at higher surface densities, which impedes AT access and binding. Therefore, considering the combined effects of spot inhomogeneity induced by the coffee-ring effect and overall cost-effectiveness, a heparin concentration of 1 mM was selected as the standard for spotting in this work.

Fig. 2. Key conditions for heparin microarray-based competitive strategy for investigating SP-AT interactions.

a Fluorescence images (inset) and fluorescence intensity of heparin at different spotting concentrations binding with AT-His tag (50 μg‧mL⁻¹) then labelled with Ab-His tag-FITC (100 μg‧mL⁻¹). b Fluorescence images (inset) and fluorescence intensity of immobilized heparin binding with AT-His tag at different concentrations then labelled with Ab-His tag-FITC (100 μg‧mL⁻¹). c Fluorescence images (inset) and fluorescence intensity of immobilized heparin binding with AT (50 μg‧mL⁻¹) then labelled with Ab-His tag-FITC at different concentrations. d Fluorescence images (inset) and fluorescence intensity of immobilized heparin binding with remaining AT-His tag then labelled with Ab-His tag-FITC after AT-His tag preincubating with heparin solution (50 μg‧mL⁻¹) for different time.

Secondly, the working concentration of AT-His tag was optimized while keeping the spotting concentration of heparin and Ab-His tag-FITC constant. The fluorescence binding intensity increased with the AT-His tag concentration, forming upward asymptotic curves, with a turning point at approximately 10 μg‧mL⁻¹ and a saturation point at 50 μg‧mL⁻¹ (Fig. 2b). Hence, the working concentration of 50 μg‧mL⁻¹ for AT-His tag was used in this work. Similarly, the working concentration of Ab-His tag-FITC was optimized while maintaining constant of the spotting concentration of heparin and AT-His tag. The fluorescence binding intensity along with the increase of Ab-His tag-FITC concentration also formed upward asymptotic curves (Fig. 2c). However, the fluorescence binding intensity rarely increased when Ab-His tag-FITC concentration exceeded 100 μg‧mL⁻¹. Consequently, a working concentration 100 μg‧mL⁻¹ for Ab-His tag-FITC was selected in this work.

Finally, the competitive time for heparin microarray-based competitive strategy was optimized by adopting heparin as model analyte. The binding intensity of immobilized heparin with AT-His tag/Ab-His tag-FITC was measured after preincubating AT-His tag with heparin solution for different time. As shown in Fig. 2d, when the preincubation time was more than 15 min, the binding intensity of immobilized heparin with remaining AT kept nearly unchanged. It indicates that the specific interactions of heparin and AT in solution reached equilibrium within 15 min. The preincubation time of 30 min was commonly used to ensure the recognition between AT and SPs under investigation reach equilibrium.

Detection of marine SPs and AT specific interactions

The established heparin microarray-based competitive strategy was used for detecting the specific interactions between AT and target marine SPs, including AnF, FCS_ht, and S-dFCS_ht. The physicochemical properties and typical structural features of AnF, FCS_ht, and S-dFCS_ht were shown in Table S2 and Figure S3. The weight-average molecular weights (Mw) of AnF, FCS_ht, and S-dFCS_ht were determined to be 176.4 kDa, 42.00 kDa, and 6.36 kDa, respectively, with corresponding number-average molecular weights (Mn) of 165.8 kDa, 38.92 kDa, and 6.12 kDa. This yielded polydispersity index (PDI) values of 1.064, 1.079, and 1.039, indicating relatively narrow molecular weight distributions for all three samples, albeit not perfect monodispersity. AnF is composed of xylose (10.64%) and fucose (89.36%), with a sulfate content of 35.67%. FCS_ht is composed of glucuronic acid, fucose, and N-acetylgalactosamine, with a sulfate content of 16.65%. Since S-dFCS_ht is derived from the selective degradation products of FCS_ht through sulfation modification, S-dFCS_ht and FCS_ht share the same monosaccharide composition, but the degree of sulfation is significantly increased to 33.58%. For polysaccharides, the chemical shifts of protons on non-anomeric carbons from different glycosyl residues tend to be very similar, resulting in substantial signal overlap in the ¹H NMR spectrum. Complete structural characterization is particularly challenging for polysaccharides with high molecular weights and multiple sulfate substitution sites. In this work, the structural features of AnF, and FCS_ht were preliminarily characterized. In ¹H NMR analysis, anomeric proton signals with chemical shifts above 5.0 ppm are generally attributed to the α-configuration, whereas those below 5.0 ppm correspond to the β-configuration. Based on the classification established by Cumashi et al., fucoidans containing α−1,3 linkages are classified as type I, while those with both α−1,3 and α−1,4 linkages are designated as type II²¹. As shown in Figure S3, within the anomeric proton region (δ 5.05–5.50 ppm), AnF shows two signals at δ 5.46 ppm and δ 5.30 ppm, which correspond to the anomeric protons of α-Fuc residues linked at the 3- and 4-positions, respectively, indicating a type II structure. FCS_ht exhibits anomeric proton signals at δ 5.66 ppm and 5.32 ppm, along with characteristic proton signals for GalNAc and Fuc at δ 2.02 ppm and δ 1.32 ppm, consistent with previous literature²².

As shown in Fig. 3a, the AT binding ratio decreased along with the increasing preincubation concentrations of AnF, FCS_ht, and S-dFCS_ht, respectively. It indicated that AnF, FCS_ht, and S-dFCS_ht competed with immobilized heparin for binding to specific domains of AT. The IC₅₀ values for AnF, FCS_ht, and S-dFCS_ht were 50.55 ± 2.79 μg‧mL⁻¹, 44.18 ± 4.05 μg‧mL⁻¹, and 16.99 ± 6.56 μg‧mL⁻¹, respectively. It indicated that S-dFCS_ht bound more strongly to AT than AnF and FCS_ht (P < 0.01), while there was no significant difference between AnF and FCS_ht in their binding to AT (Fig. 3b). It indicated that S-dFCS_ht may exhibit highest anticoagulant activity compared to AnF and FCS_ht. For comparison, the IC₅₀ value for heparin in solution was 56.89 ± 7.22 μg‧mL⁻¹ (Fig. 1e), demonstrating that the binding potency of S-dFCS_ht for AT was stronger than that of heparin (P < 0.01), while there was no significant difference in the binding potency of heparin, AnF and FCS_ht with AT (Fig. 3b). It indicated that S-dFCS_ht may exhibit stronger anticoagulant activity than heparin, while AnF and FCS_ht had similar anticoagulant activity with heparin. Due to the IC₅₀ value for enoxaparin in solution was 225.30 ± 38.45 μg‧mL⁻¹ (Fig. 1e), it showed that the binding ability of AnF, FCS_ht, and S-dFCS_ht with AT were stronger than that of enoxaparin (P < 0.01) (Fig. 3c). It indicated that AnF, FCS_ht, and S-dFCS_ht could exhibit stronger anticoagulant activity than enoxaparin. Therefore, AnF, FCS_ht, and S-dFCS_ht had promising potentials in marine SP-based anticoagulant drug development.

Fig. 3. The investigation of binding potency between marine SPs and AT by adopting heparin microarray-based competitive strategy.

a Relationship of AT binding ratio on heparin microarrays with the concentrations of marine SP (AnF, FCS_ht, and S-dFCS_ht) that preincubated with AT. b IC₅₀ values for heparin, AnF, FCS_ht, and S-dFCS_ht on binding events of AT and immobilized heparin on microarrays. c IC₅₀ values for enoxaparin, AnF, FCS_ht, and S-dFCS_ht on binding events of AT and immobilized heparin on microarrays. The data were expressed as means ± SD and determined by t-test. NS not significant; **P < 0.01.

Molecular docking was adopted to further investigate the active sites involved in the binding of AnF, FCS_ht, and S-dFCS_ht to AT. The detailed information regarding the docking process can be found in SI. The docking scores for the most stable conformations resulting from the binding of the AT active site to characteristic structural fragments of AnF, FCS_ht, and S-dFCS_ht are summarized in the Table S3.The results indicate that a hexasaccharide fragment of AnF (α-Fuc4S-(1 → 3)-α-Fuc4S-(1 → 4)-α-Fuc2S-(1 → 3)-α-Fuc2S-(1 → 4)-α-Fuc3S-(1 → 3)-α-Fuc4S), a pentasaccharide fragment of FCS_ht (α-Fuc2,3,4S-(1 → 3)-β-GlcA-(1 → 3)-β-GalNAc6S-(1 → 4)-β-GlcA-(1 → 3)-α-Fuc2,4S), and a pentasaccharide fragment of S-dFCS_ht (α-Fuc2,3,4S-(1 → 3)-β-GlcA2S-(1 → 3)-β-GalNAc4,6S-(1 → 4)-β-GlcA2S-(1 → 3)-β-GalNAc4,6S) form the most stable conformations with AT (Figure S4). Analysis of the intermolecular interactions between these active fragments and the AT active pocket further elucidates how distinct structural features of AnF, FCS_ht, and S-dFCS_ht lead to differences in their binding affinity for AT. Specifically, the active fragment of S-dFCS_ht is a highly sulfated pentasaccharide containing 9 sulfate groups. It forms 19 hydrogen bonds and 33 ionic bonds within the AT pocket. The sulfate groups on GlcA establish multiple hydrogen bonds and salt bridges with the side-chain amino groups of Lys11 and Lys114, while those on Fuc and GalNAc form multi-point anchor-like interactions with key residues, including Arg132, Arg47, and Lys125. These high-density, short-range cooperative interactions underlie its high binding affinity. In contrast, the active FCS_ht pentasaccharide contains 6 sulfate groups, forming 19 hydrogen bonds and 16 ionic bonds within the AT pocket. Although the sulfate groups on Fuc can form stabilizing salt bridges with residues such as Arg132 and Lys125, the limited number and spatial distribution of sulfate groups result in weaker overall electrostatic interactions and reduced capacity for multi-point cooperative anchoring. The active AnF hexasaccharide also contains 6 sulfate groups and forms 17 hydrogen bonds and 22 ionic bonds within the AT pocket. Despite its higher degree of polymerization, the absence of carboxyl groups and the conformational constraints typically provided by GlcA/GalNAc units confer greater backbone flexibility. This leads to more dispersed interactions and an inability to form a high-density cooperative binding network. Consequently, the binding affinity of the AnF-AT complex is comparable to that of FCS_ht-AT, and both are weaker than that of S-dFCS_ht-AT. This result not only aligns with the total number of hydrogen and ionic bond but is also consistent with experimental observations.

In addition, the anti-thrombin (Factor IIa, FⅡa) and anti-Factor Xa (FXa) activities of AnF, FCS_ht, and S-dFCS_ht were measured using coagulation factor activity inhibition assays, and compared with heparin²³ (Figure S5). The inhibition rates of SPs against FⅡa and FXa showed that neither FCS_ht nor S-dFCS_ht exhibited a significant difference in inhibitory activity compared to heparin. In contrast, AnF demonstrated stronger inhibition of both FIIa and FXa than heparin, suggesting that its anticoagulant activity is mediated not only through interaction with AT but also via direct action on the coagulation factors themselves. The activated partial thromboplastin time (APTT) was measured to assess the apparent anticoagulant activities of AnF, FCS_ht, and S-dFCS_ht (Figure S6). The results indicated that all three SPs prolonged APTT compared to the blank control, and more importantly, their effects surpassed that of enoxaparin. The linear correlation between functional anticoagulant activity and AT binding was analyzed by plotting APTT against the reciprocal of IC₅₀, which was determined via the heparin microarray-based competitive assay (Figure S7). A significant positive correlation (R² = 0.933, p < 0.05) was obtained, demonstrating that strong AT binding directly predicts greater functional anticoagulant activity.

Verification of marine SPs and AT specific interactions

SPR, the gold standard for molecular interaction detection, was employed to evaluate the binding events of AT and SPs, thereby verifying the reliability of heparin microarray-based competitive strategy in specific interaction assay of marine sulfated polysaccharides with AT. AT was immobilized on CM5 sensor chips, and SPs solutions flew through the chip surface. The typical SPR sensing curves for heparin, enoxaparin, AnF, FCS_ht, and S-dFCS_ht at different concentrations binding to AT was shown in Fig. 4a–e, respectively. All aforementioned SPR sensing curves included association and dissociation procedures with background signals deducted. The increased signals in association procedures indicated the specific binding of SPs (heparin, enoxaparin, AnF, FCS_ht and S-dFCS_ht, respectively) to AT. The corresponding binding affinity, characterized by the equilibrium dissociation constant (K_D) values, were obtained by fitting the curves to a 1:1 biomolecular reaction model. All fitting results met the standards after the quality control checks. K_D values for heparin-AT, enoxaparin-AT, AnF-AT, FCS_ht -AT, and S-dFCS_ht-AT specific interaction was 36.11 ± 3.92 ng‧mL⁻¹, 121.40 ± 15.60 ng‧mL⁻¹, 35.47 ± 5.77 ng‧mL⁻¹, 38.94 ± 2.64 ng‧mL⁻¹, and 13.27 ± 6.01 ng‧mL⁻¹, respectively. Given that the bioactivity of polysaccharides primarily stems from their numerous active fragments rather than the intact molecules, the molecular weight herein is calculated approximately based on the active fragments determined by molecular docking, in order to avoid potential biases in activity assessment caused by disproportionately large molecular weights. Therefore, the corresponding molar concentrations were calculated approximately based on the molecular weights of their active units (AnF fragment: 1368 g‧mol⁻¹, FCS_ht fragment: 1340 g‧mol⁻¹, S-dFCS_ht fragment: 1633 g‧mol⁻¹, heparin or enoxaparin fragment: 1406 g‧mol⁻¹). Consequently, the K_D values expressed in terms of molar concentration for heparin-AT, enoxaparin-AT, AnF-AT, FCS_ht -AT, and S-dFCS_ht-AT specific interaction was 25.68 ± 2.79 nM, 70.25 ± 9.03 nM, 25.93 ± 4.22 nM, 29.06 ± 1.97 nM, and 8.13 ± 3.68 nM, respectively. The comparison of above-mentioned binding affinity between AT and each SP was illustrated in Fig. 4f, g. Statistical analysis revealed that the binding affinity of S-dFCS_ht with AT was stronger than that of heparin, AnF, and FCS_ht, while the binding ability of AnF and FCS_ht with AT had no significant difference compared to that of heparin. In addition, the binding affinity of AnF, FCS_ht, and S-dFCS_ht with AT were stronger than that of enoxaparin. These statistical results were consistent with those obtained from heparin microarray-based competitive strategy, confirming the accuracy and reliability of heparin microarray-based assay.

Fig. 4. SPR analysis of the affinity between SPs and AT.

The typical SPR sensing curves (solid lines) and fitting curves (dotted lines) of heparin (a), enoxaparin (b), AnF (c), FCS_ht (d), and S-dFCS_ht (e) at different concentrations binding with AT. K_D values for heparin-AT, enoxaparin-AT, AnF-AT, FCS_ht-AT, and S-dFCS_ht-AT specific interaction (f, g). The data were expressed as means ± SD and determined by t-test. NS not significant; **P < 0.01, ***P < 0.001.

Extendibility of heparin microarray-based competitive strategy

SPs, similar to heparin, are negatively charged and can bind to positively charged PF4 to form heparin-like/PF4 complex, which consequently increase the risk of HIT²⁴. The established heparin microarray-based competitive strategy was extendible to evaluate SP-induced potential side effects. Firstly, a carbohydrate microarray containing heparin, dextran, and water spots was fabricated on pHEMA-CC-modified substrates to confirm that the specific recognition of immobilized heparin with PF4 was still maintained. The microarrays were then incubated with PF4-His tag and labeled with Ab-His tag-FITC, only strong binding signals were obtained from the heparin spots (Fig. 5a). Differently, no signals were obtained from water spots, which served as the blank control of spotting solvent, or dextran spots, which served as negative control. When the microarrays were incubated only with Ab-His tag-FITC, no binding signals were obtained from all spots (Fig. 5b). It indicated that the strong binding signals arose specifically from the interaction between heparin and PF4-His tag.

Fig. 5. The heparin microarray-based competitive strategy used for evaluating SP-induced side effects.

Fluorescence images of carbohydrate microarray containing heparin, dextran, and water spots before (a1, b1) and after recognition with PF4-His tag/Ab-His tag-FITC (a2), and Ab-His tag-FITC (b2). The images were in pseudocolor, and the default color was set according to the wavelength of excitation laser. c PF4 binding ratio on heparin microarrays after PF4 preincubating with heparin, enoxaparin, and dextran solutions at different concentrations. d IC₅₀ for heparin, enoxaparin, AnF, FCS_ht, and S-dFCS_ht on binding event of PF4 and immobilized heparin. The data were expressed as means ± SD and determined by t-test., **P < 0.01, ***P < 0.001, and ****P < 0.0001.

In this work, heparin and enoxaparin, both known to pose HIT risks, were used as model analytes, while dextran served as the negative control. As shown in Fig. 5c, the PF4 binding ratio on heparin microarrays was as dependent variable, and the concentration of model analytes (heparin and enoxaparin) was as independent variable. The curves formed downward curves along with the increase in the concentration of heparin or enoxaparin solutions (Fig. 5c), Differently, when dextran solutions of different concentrations were preincubated with PF4-His tag and then incubated with heparin microarray, the PF4 binding signals obtained from heparin spots with remaining PF4-His tag remain basically unchanged. It indicated that heparin or enoxaparin in solutions competed with immobilized heparin to specifically bind to PF4. The binding ability of model analytes (heparin and enoxaparin in solutions) with PF4 were evaluated by determining their IC₅₀ values on the binding events of PF4 and immobilized heparin. The IC₅₀ value was 0.65 ± 0.22 μg‧mL⁻¹ for heparin in solution, while IC₅₀ value was 361.06 ± 30.07 μg‧mL⁻¹ for enoxaparin in solution (Fig. 5d). These results revealed that heparin exhibited stronger binding ability to PF4 and a higher HIT risk compared to enoxaparin, which was consistent with literatures²⁵. It demonstrated the feasibility of adopting heparin microarray-based competitive strategy for investigating SPs and PF4 specific interaction. The binding potency of AnF, FCS_ht, and S-dFCS_ht to PF4 were subsequently investigated to assess their potential for triggering HIT. As shown in Figure S8, the typical competitive binding curves displayed a concentration-dependent decrease in response to increasing concentrations of the SP solutions, yielding IC₅₀ values of 1.93 ± 0.03 μg‧mL⁻¹ for AnF, 127.77 ± 10.08 μg‧mL⁻¹ for FCS_ht, and 6.06 ± 0.12 μg‧mL⁻¹ for S-dFCS_ht. These results indicate that all three SPs possess a lower HIT risk than heparin but a higher risk than enoxaparin. These findings offer valuable in vitro data that informs the clinical safety evaluation of these candidate compounds. Consequently, the established heparin microarray-based competitive strategy offers a dual capability: assessing anticoagulant activity targeting AT and evaluating the risk of HIT through PF4 interactions, making it a valuable tool for broader applications in anticoagulant drug development.

Conclusions

In summary, a heparin microarray-based competitive strategy was developed for investigating the specific interactions between SPs and AT, in order to evaluate the anticoagulant activity of SPs. Heparin and enoxaparin, both with known AT binding ability, were used as model analytes to establish this strategy. Key factors that impacted the performance of this strategy including the heparin spotting concentration for heparin microarray fabrication, the working concentration of AT-His tag and Ab-His tag-FITC, and the competitive time were optimized. The established strategy was applied to assess the specific interactions of marine SPs with AT, and the reliability was verified by SPR analysis. Furthermore, the established strategy was extended to evaluate the interaction between SPs and side effect-related proteins, facilitating the simultaneous assessment of potential side effect in anticoagulant drug discovery and development. Although currently applicable primarily to heparin-interacting anticoagulant targets, this strategy can be extended to other targets by fabricating specialized glycan microarrays, enabling comprehensive evaluation of SP interactions with various anticoagulant targets.

Methods

Chemicals and materials

(3-Aminopropyl)trimethoxysilane (APTMS, 97%), α-bromoisobutyryl bromide (BIBB, 98%), L-ascorbic acid (AA, 99%), 2-hydroxyethyl methacrylate (HEMA, 96%), and dextran (40 kDa) were purchased from Aladdin Chemistry Co., Ltd (Shanghai, China). 2,2’-bipyridine (BPY, 99.0%), CuBr (99.0%), cyanuric chloride (CC), and N, N-diisopropylethylamine (DIPEA) were purchased from Energy Chemical Co., Ltd (Shanghai, China). FITC labelled anti-6X His tag antibody (Ab-His tag-FITC) was purchased from Abcam Trading Co., Ltd (Shanghai, China). Triethylamine (TEA) and ethanolamine (EOA) were purchased from Beijing Chemical Works (Beijing, China). Polyhistidine-tagged recombinant human antithrombin protein (AT-His tag) and polyhistidine-tagged influenza A H7N9 (A/Shanghai/1/2013) neuraminidase (NA-His tag) were purchased from Sino Biological Inc. (Beijing, China). Polyhistidine-tagged recombinant human platelet factor 4 protein (PF4-His tag) was purchased from Absin Bioscience Inc. (Shanghai, China). Series S sensor chip CM5, MgCl₂, sodium acetate solution (pH = 5), 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), N-hydroxysuccinimide (NHS), ethanolamine-HCl solution (1.0 M pH = 8.5), and 10 × HBS-EP buffer containing HEPES (100 mM, pH 7.4), NaCl (1.5 M), EDTA (30 mM), and Surfactant P20 were purchased from Cytiva (Shanghai, China). Heparin from porcine intestinal (Mw = 15 kDa) was obtained from Linhardt’s laboratory (Troy, New York). Enoxaparin was purchased from Wanbang Biopharmaceuticals (Jiangsu Province, China). Fucoidan from Ascophyllum nodosum (AnF), and fucosylated chondroitin sulfate from Holothuria tubulosa (FCS_ht) were prepared and characterized according to our previous work^22,26, detailed procedures were provided in SI. FCS_ht was selectively degraded by adopting nitrous acid degradation, then the products were modified by sulfation to obtain S-dFCS_ht^27,28. The physicochemical properties of AnF, FCS_ht, and S-dFCS_ht were shown in Table S2, and the ¹H NMR spectroscopy of AnF, FCS_ht in D₂O were shown in Figure S3. Pure water (18.2 MΩ∙cm) was used for the preparation of all aqueous solutions.

Heparin Microarray fabrication

Heparin microarray was fabricated on pHEMA-CC-modified glass slides^17,18. Heparin was dissolved in water to prepare solutions of different concentrations, and a 1 mM heparin solution was commonly used in this work. The pH of heparin solutions was adjusted to 7-8 with 1 M NaOH prior to use if needed. Heparin solutions, dextran solution, and water were manually spotted on pHEMA-CC-modified slides to form microarrays. The slides were then incubated in a humid chamber at 25 °C for at least 10 h, and washed three times with water. After washing, the slides were immersed in EOA (1 M, pH = 8.60) for 3 h at 25 °C to deactivate unreacted CC. After drying under a N₂ stream, the slides were used instantly or stored at -20 °C for later use.

Microarray imaging assay

AT-His tag and Ab-His tag-FITC were diluted with PBS buffer (10 mM, pH 7.40). To investigate the specific recognition between heparin and AT, heparin microarrays were incubated with AT-His tag, followed by labeling with Ab-His tag-FITC for at least 30 min. For the investigation of SPs and heparin competitively bound to AT, AT-His tag was pre-mixed with SPs at different concentrations for at least 30 min. The final concentration of AT-His tag in the mixed solution was 50 μg‧mL⁻¹, and the final concentration of SPs in the mixed solution varied from 0 to 500 μg‧mL⁻¹, respectively. Then, the mixed solution incubated with the heparin microarrays for 30 min. Each incubation chamber was sealed with polydimethylsiloxane strips as needed. The heparin microarrays were then rinsed with PBS buffer and water, and dried under a N₂ stream. Finally, all heparin microarrays were labeled with Ab-His tag-FITC for at least 30 min, subsequently rinsed with PBS buffer and water, and dried under a N₂ stream.

The microarrays were then scanned using a GenePix® 4300 A scanner equipped with a 488 nm excitation laser and 513-555 nm emission filters (Molecular Devices, Inc., USA). All the fluorescence images were recorded and analyzed using GenePix Pro 7 software. The obtained images were displayed with pseudocolor, and the default color was set according to the wavelength of excitation laser (488 nm). The AT binding ratio, which was calculated by the following equation,

$$\mathrm{AT}\mathrm{binding}\mathrm{ratio}=I/I_0$$

Here, I was the fluorescence binding intensity of the remaining AT after pre-incubation with SPs at different concentrations to heparin on microarrays; I₀ was the fluorescence binding intensity of AT without pre-incubating with SPs to heparin on microarrays.

Surface plasmon resonance (SPR) analysis

Biacore® T200 SPR system, equipped with a CM5 sensor chip, was adopted for SPR analysis. In the working channel, an EDC/NHS solution was injected into flow cell at a flow rate of 10 μL‧ min⁻¹ to activate carboxyl groups on CM5 sensor chip, according to the manufacturer’s instructions. Subsequently, AT (12.5 μg‧mL⁻¹) diluted in sodium acetate solution (pH 5.0) was injected to facilitate AT immobilization. The successful immobilization of AT was confirmed by observing a signal increase of 9000 resonance units (RU) on the sensor chip. Then, the sensor chip was blocked with ethanolamine-HCl. A flow cell that did not undergo AT immobilization served as the reference channel.

A series of SP solutions at different concentrations that diluted by HBS-EP buffer was injected through the chip at a flow rate of 30 μL‧min⁻¹ for 60 s, followed by washing the channels with HBS-EP buffer for 240 s. After each run, the sensor chip was regenerated by MgCl₂ solution (4 M) for 30 s. The equilibrium dissociation constant (K_D) was calculated by Biacore® affinity analysis software.

Author contributions

These authors contributed equally: Fei Wang, Qi Li F. Wang and Q. Li performed all the experiments and processed raw data. H. Zhou provided marine sulfated polysaccharides samples. C.J Liu conceived the concept, designed the experiments, and provided financial support. G.L. Yu provided financial support. G.Y. Li conceived the concept and provided financial support.

Peer review

Peer review information

Communications Chemistry thanks Mauro Sergio Goncalves Pavão, Fuming Zhang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Data availability

All key data supporting the findings of this work are included in the manuscript and Supporting Information, and Supplementary Data Files.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42004-026-01999-4.