Regulation of Complement and Contact System Activation via C1 Inhibitor Potentiation and Factor XIIa Activity Modulation by Sulfated Glycans – Structure-Activity Relationships

The serpin C1 inhibitor (C1-INH) is the only regulator of classical complement activation as well as the major regulator of the contact system. Its importance is demonstrated by hereditary angioedema (HAE), a severe disease with potentially life-threatening attacks due to deficiency or dysfunction of C1-INH. C1-INH replacement is the therapy of choice in HAE. In addition, C1-INH showed to have beneficial effects in other diseases characterized by inappropriate complement and contact system activation. Due to some limitations of its clinical application, there is a need for improving the efficacy of therapeutically applied C1-INH or to enhance the activity of endogenous C1-INH. Given the known potentiating effect of heparin on C1-INH, sulfated glycans (SG) may be such candidates. The aim of this study was to characterize suitable SG by evaluating structure-activity relationships. For this, more than 40 structurally distinct SG were examined for their effects on C1-INH, C1s and FXIIa. The SG turned out to potentiate the C1s inhibition by C1-INH without any direct influence on C1s. Their potentiating activity proved to depend on their degree of sulfation, molecular mass as well as glycan structure. In contrast, the SG had no effect on the FXIIa inhibition by C1-INH, but structure-dependently modulated the activity of FXIIa. Among the tested SG, β-1,3-glucan sulfates with a Mr ≤ 10 000 were identified as most promising lead candidates for the development of a glycan-based C1-INH amplifier. In conclusion, the obtained information on structural characteristics of SG favoring C1-INH potentiation represent an useful elementary basis for the development of compounds improving the potency of C1-INH in diseases and clinical situations characterized by inappropriate activation of complement and contact system.


Introduction
These may also be obstacles to the application of C1-INH in further indications, although clinical studies showed beneficial effects of C1-INH also in other disease states like sepsis, gram-negative endotoxic shock, vascular leak syndrome, transplant rejection, ischemia-reperfusion injury, myocardial infarction, and emergency coronary artery bypass surgery [24,25].
Consequently, there is still a need for improving the C1-INH replacement therapy and for alternatives to C1-INH. At this point, the potentiating effect of heparins and some other sulfated compounds on C1-INH, which is known for a long time [26], leads to the idea that sulfated glycans (SG) may be candidates to enhance the efficacy of therapeutically applied C1-INH or to increase the activity of endogenous C1-INH.
The C1-INH potentiating effect of unfractionated heparin (UFH) [26] was confirmed in vivo by increased C1rs-C1-INH complexes in the blood of patients undergoing coronary artery bypass grafting using heparin-coated surfaces [27]. Furthermore, a limited number of other SG was analyzed in vitro, yet, with varying results. In one study dextran sulfates (DexS) with different molecular masses potentiated the C1s inhibition by C1-INH even stronger than UFH [28], whereas others found a C1-INH potentiating effect smaller than that of UFH [26]. The glycosaminoglycans (GAGs) heparan sulfate (HS) and chondroitin sulfate (CS) were consistently less active than UFH or even inactive [28][29][30], whereas oversulfated chondroitin sulfate (OSCS) was shown to increase the binding of C1-INH to C1s stronger than UFH [30]. Furthermore, a fucoidan extracted from the brown alga Ascophyllum nodosum turned out to influence the C1s inhibition by C1-INH only to a small extent [31]. Rossi et al. demonstrated that the C1-INH potentiation by heparin oligomers is due to binding to both C1-INH and C1s [32]. Just recently, the effects of UFH and low-molecular weight heparins (LMWHs) on CP, LP and AP activation in absence and presence of C1-INH were compared [33].
Investigations on the contact system proteases targeted by C1-INH revealed that UFH, HS, DexS, dermatan sulfate and LMWHs enhanced the inhibition of FXIa by C1-INH, but had no effect on kallikrein and even protected FXIIa from inhibition by C1-INH [34][35][36]. In contrast, Gozzo et al. found a slightly increased kallikrein inhibition by C1-INH in presence of UFH, HS and CS [37].
In view of these limited data, we examined more than 40 SG for their potentiating effect on the C1s inhibition by C1-INH. By evaluating structure-activity relationships, we aimed to characterize the structural features of SG principally suitable for use as C1-INH amplifier. Moreover, the structure-dependent effects of SG on FXIIa and its inhibition by C1-INH were investigated, since this key serine protease of the contact system additionally triggers the CP complement activation and is involved in the pathophysiology of HAE [38,11].
Besides various heparins and further glycosaminoglycans, native and partially depolymerized SG isolated from four different algae were included in the study. Two series of semi-synthetic, structurally defined β-1,3-glucan sulfates enabled us to examine the impact of molecular mass (M r ) and degree of sulfation (DS) in more detail independently of the basic glycan structure.
The study revealed that the potentiating effect of SG on the C1s inhibition by C1-INH depends on their degree of sulfation (DS), their molecular mass (M r ) as well as their glycan structure, whereby the impact of the M r is closely linked to the DS of the SG. In contrast, the SG had no effect on the inhibition of FXIIa by C1-INH, but structure-dependently either stimulated or inhibited the activity of FXIIa directly.

Test compounds
Structural characteristics of all test compounds including their molecular mass (M r ) and degree of sulfation (DS) are indicated in Table 1. All anionic glycans were used in form of the sodium salt, if not indicated otherwise.
Other glycosaminoglycans. A series of both naturally occurring and chemically modified glycosaminoglycans (GAGs) was included in the study: Chondroitin sulfate A from bovine trachea (CS-A, Sigma-Aldrich, St. Louis, Missouri, USA), chondroitin sulfate B from porcine intestinal mucosa, synonym dermatan sulfate (CS-B, Sigma-Aldrich, St. Louis, Missouri, USA), and chondroitin sulfate C from shark cartilage (CS-C, Sigma-Aldrich, St. Louis, Missouri, USA) were included in the study.
Algae-derived sulfated glycans and sodium alginate. The fucose-containing sulfated polysaccharides from Saccharina latissima (S.l.-SP) were extracted from dried material of the brown alga Saccharina latissima L. as previously described and identified as sulfated galactofucan with about 60% fucose and 15% galactose as main monosaccharides [44]. We examined two batches deriving from the North Atlantic Ocean harvested either in May (S.l.-SP#2) or in September (S.l.-SP#1). The two batches mainly differ in their DS, since both habitat and harvest time have an impact on the sulfate content of the S.l.-SP [44].
Further, two batches of the commercially available fucose-containing sulfated polysaccharides from Fucus vesiculosus L., named fucoidan and containing about 82% fucose, (F.v.-SP#a/ b, Sigma-Aldrich, St. Louis, Missouri, USA) were included in the study.
Additionally, we included series of S.l.-SP, F.v.-SP and C.t.-SP samples, which had been subjected to partial hydrothermal depolymerization ( Table 2).

Methods
Testing of the C1s inhibition by C1 inhibitor and the influence of sulfated glycans. The effect of C1-INH on the activity of C1s esterase in the presence and absence of test compounds was measured by a chromogenic substrate assay performed in microplates (nunc™ 269620, Thermo Fisher Scientific, Langenselbold, Germany). The C1-INH stock solution was diluted with TRIS buffer A (50 mmol/l TRIS, 257 mmol/l NaCl, pH 7.4) to a working concentration of 15 μg/ml (143 nmol/l). C1s esterase was diluted 1:4 with TRIS buffer A and stored on ice until use. The chromogenic substrate was diluted 1:6 with TRIS buffer B (50 mmol/l TRIS, 257 mmol/l NaCl, pH 8.3). The test compounds were diluted to the working concentration of 50 μg/ml with veronal buffer saline (VBS, 4.94 mmol/l 5,5-diethylbarbituric acid, 145 mmol/l NaCl, 0.83 mmol/l magnesium chloride, 0.25 mmol/l calcium chloride dihydrate, pH 7.3). Moreover, dilution series of the following compounds were prepared in VBS: UFH, PS3, OSHep: 0.1, 1.0, 10.0, 50.0, 100.0 μg/ml; CurS and PhyS series: 2.5, 5.0, 10.0, 50.0 μg/ml. Initially, the microplate was loaded with 20 μl sample solution per well and incubated at 37°C for 10 min. Then, 20 μl of each, C1s and C1-INH were added. After incubation at 37°C under shaking for 5 min, 100 μl of prewarmed chromogenic substrate (final concentration: 0.625 mmol/l) were pipetted into each well. The plate was further incubated at 37°C under shaking for 15 min. The reaction was stopped by addition of 40 μl acetic acid 50% per well. The absorbance was measured at 405 nm against blank. Control values were (1) VBS instead of test compound and TRIS buffer A instead of C1-INH (100% C1s activity), (2) VBS instead of test compound (100% C1-INH potency) and (3) VBS instead of test compound, TRIS buffer A instead of C1-INH and C1s (blank). All samples and controls of a test run were tested in duplicate on each microplate and each test run was repeated at least three times on different days.
Testing of direct influences of sulfated glycans on the C1s activity. According to the procedure described above, the direct influence of all test compounds on the C1s activity was Testing of the FXIIa inhibition by C1 inhibitor and the influence of sulfated glycans. The effect of C1-INH on the activity of FXIIa in the presence and absence of UFH, PS3, CurS3 and S.l.-SP#1 was measured by a chromogenic substrate assay performed in microplates (nunc™ 269620, Thermo Fisher Scientific, Langenselbold, Germany).
Initially, the microplate was loaded with 20 μl sample solution per well and incubated at 37°C for 10 min. Then, 20 μl of each, FXIIa and C1-INH were added. After incubation at 37°C under shaking for 10 min, 100 μl of S-2302 1 mmol/l in TRIS buffer (50 mmol/l TRIS, pH 7.8) were pipetted into each well. The kinetics of the absorbance at 405 nm was measured at 37°C for 30 min, whereby the data points after 10 min served as values to compare the test compounds. Control values were (1)  Testing of direct influences of sulfated glycans on the FXIIa activity. According to the procedure described above, the direct influence of the test compounds on the FXIIa activity was examined. For this, the C1-INH was replaced by 20 μl TRIS buffer. All samples were tested in duplicate per microplate on at least two different days.
Data analysis. Linear and non-linear regressions and statistics were calculated using Sig-maPlot TM 11.0 (Systat Software Inc.). P-values of 0.05 were considered significant and were calculated with a one-way analysis of variance followed by multiple comparisons versus control (Holm-Sidak method) unless stated otherwise.

Evaluation of incubation times in the C1-INH/C1s assay
In the course of establishing the C1-INH/C1s assay, the effect of various incubation procedures was examined, whereby PS3 was used as exemplary C1-INH amplifier. The aims were (1) to determine the extent and velocity of C1s inhibition by C1-INH, (2) to estimate the extent of C1-INH potentiation by PS3, and (3) to evaluate the influence of preincubation of PS3 with C1-INH alone as well as of incubation with both C1-INH and C1s on its C1-INH potentiating effect.
Influence of preincubation of test compound and C1 inhibitor on C1s inhibition. After a preincubation of C1-INH in absence and presence of PS3 (final concentration: 0.625 μg/ml) for 0, 2, 5, 10 or 15 min at 37°C the residual C1s activity was determined as described in materials and methods. The C1s inhibition was calculated in relation to 100% C1s activity without C1-INH and PS3.
Influence of the incubation time of test compound with C1 inhibitor and C1s on C1s inhibition. To evaluate the velocity of the C1s inhibition by C1-INH and the extent of C1-INH potentiation by SG in dependence on the C1-INH-C1s reaction time, we determined the C1s inhibition after incubation of C1-INH with C1s in absence and presence of PS3 for 0, 2, 5, 10, 15, and 30 min. The inhibition of the amidolytic activity of C1s by C1-INH turned out to be a slow process (Fig 1). After 30 min, it amounted to 88 ± 6% with 50% enzyme inhibition after 4 min. In presence of PS3 (final concentration: 1.25 μg/ml), about 90% C1s inhibition was achieved already after an incubation time of 5 min (Fig 1). Apparently, the reaction between C1-INH and C1s was accelerated in presence of PS3 with 50% enzyme inhibition already after 1 min. However, the calculated C1-INH potentiation related to the C1s inhibition by C1-INH alone decreased with increasing incubation time. After 2 min, PS3 increased the C1s inhibition from 35 ± 3% to 71 ± 3%, corresponding to a potentiation of about 100%, whereas that after 15 min amounted to 15 ± 7%. Although the potentiation after 2 min was even stronger, the following experiments were performed with an incubation time of 5 min for practical reasons. It should be mentioned that the increase of the C1s inhibition from 55 ± 6% to 87 ± 3% by PS3 after 5 min corresponded to a potentiation of 65%, which was nearly the maximal possible one.

C1s inhibition by C1 inhibitor and its potentiation by sulfated glycans
Before investigating the structure-dependent C1-INH potentiation by SG, we checked whether any of the test compounds has a direct influence on the C1s activity. As none of the tested SG inhibited C1s directly (S1 Fig), the amplified C1s inhibition in the presence of C1-INH was exclusively due to their potentiating effect on C1-INH (Fig 2). As an example, Fig 2 presents the concentration-dependent reduction of the C1s activity by the three SG UFH, OSHep and PS3 (Fig 2A) and their corresponding C1-INH potentiation ( Fig 2B). PS3 was superior to OSHep and UFH; 1.25 μg/ml PS3 led to a C1-INH potentiation of about 65%, which was the highest possible one under the used experimental conditions. To evaluate the C1-INH potentiation by SG in dependence on their structural characteristics, their effects were primarily compared in a final concentration of 6.25 μg/ml.
Heparins. Testing a series of heparins differing in their M r revealed that their C1-INH potentiating effect depends on their M r (S2 Fig). The smallest compound (PENTA, M r 1 500) was completely inactive, whereas UFH was the most active compound among the heparins with similar DS. But the DS showed to play a role as well (Fig 3). Despite their high M r , the two desulfated heparins deNsUFH and deOsUFH were inactive, whereas the highly sulfated pentasaccharide fondaparinux (FPX) exhibited at least a small C1-INH potentiation (3.3 ± 5.3% at a final concentration of 6.25 μg/ml). OSHep led to an even stronger C1-INH potentiation than UFH: about 66% vs. 55% at a final concentration of 6.25 μg/ml (Fig 2, Fig 3).
Other glycosaminoglycans. Among the tested sulfated GAGs, CS-A, CS-B, CS-C and HS33-04, which all have a DS 0.4, were inactive. In contrast, we found a modest C1-INH potentiating activity for the higher sulfated heparan sulfate HS11-08 (DS 0.8) as well as for danaparoid, which also consists mainly of heparan sulfate. Danaparoid has a mean DS of about 0.5, but additionally contains about 4% high-sulfated antithrombin-binding heparan sulfate [46]. Comparing the three heparan sulfate species (HS33-04, HS11-08 and danaparoid) and heparin (e.g. TINZA), which can be considered as a higher sulfated heparan sulfate, the DS turned out to be the crucial parameter for their C1-INH potentiating activity (Fig 4A). Its important role also becomes obvious with OSCS that was highly active whereas the low-sulfated CS-A, -B and -C rather attenuated the C1s inhibition by C1-INH ( Fig 4B). Interestingly, hyaluronic acid (HA, M r 915 000), which is negatively charged due to uronic acids, but not sulfate groups, as well as fragments of HA (oligoHA M r 3 600) slightly potentiated the C1-INH by about 8 ± 2% at a final concentration of 6.25 μg/ml.
Dextrans and dextran sulfates. To confirm the importance of sulfate groups for the C1-INH potentiation, we tested two dextrans (Dex) and two high-sulfated dextran sulfates (DexS) with different M r . Whereas the non-sulfated compounds were inactive, both DexS exhibited a strong C1-INH potentiating effect (Fig 5). DexS-H was more active than DexS-L suggesting that the large difference in their M r exceeds the DS in importance.
β-1,3-glucan sulfates. To evaluate the M r and DS dependence of the C1-INH potentiating activity of SG in more detail, we tested two series of semisynthetic β-1,3-glucan sulfates with different DS, namely the low-molecular mass phycarin sulfates (PhyS) and the high-molecular mass curdlan sulfates (CurS).
These SG potentiated the C1-INH effect against C1s mainly DS-dependently, whereby the effect increased only up to a DS of 1.8 and then reached a kind of plateau ( Fig 6A). The Interactions of Sulfated Glycans with C1 Inhibitor, C1s and Factor XIIa influence of the M r was particularly notable in case of SG with relatively low DS: PhyS1 with the lowest DS of 0.75 was inactive whereas the larger CurS1 with an even lower DS of 0.64 exhibited a concentration-dependent C1-INH potentiation (Fig 6B). In contrast to comparable activities at the same gravimetric concentration, testing of approximately similar molar concentrations revealed a considerable activity increase also between the higher sulfated PhyS3 and CurS3a, but not between CurS3a (M r 70 000) and CurS3 (M r 160 000).
As further examples of SG having glycan structures completely different from GAGs and homoglucans, we examined several sulfated glycans isolated from two red alga and two brown alga species: (1)   Interactions of Sulfated Glycans with C1 Inhibitor, C1s and Factor XIIa effect of these samples in a final concentration of 6.25 μg/ml, however, only decreased at M r 151 000 and that of the sample with M r 77 000 was reduced by 22% (Table 2) being still higher than that of the sulfated xylogalactans. Analogously, the activity of gradually depolymerized samples of the some higher sulfated F.v.-SP#b (M r 38 000, DS 0.63) only decreased at M r 15 900 and was reduced by about 60% at M r 10 300 (Table 2)  FXIIa inhibition by C1 inhibitor and activity modulation by sulfated glycans C1-INH is not only the major inhibitor of the complement-related C1s, but also neutralizes FXIIa, FXIa, and kallikrein of the contact system. Since FXIIa additionally activates the classical pathway of the complement system by converting C1r to its active form [38], it was of special interest to check, whether sulfated glycans potentiate the FXIIa inhibition by C1-INH, too.
To compare the potential activities of SG on the FXIIa inhibition by C1-INH with those on C1s inhibition, an analogous chromogenic assay was used. In line with kinetic data described in literature [28,34], the reaction between FXIIa and C1-INH turned out to be slower than the C1s/C1-INH interaction with 23 min vs. 4 min needed for 50% enzyme inhibition (S3 Fig and  Fig 1). Therefore, the incubation time of 5 min in the C1s/C1-INH assay was prolonged to 10 min, which resulted in a FXIIa inhibition of about 30-35% by C1-INH (S3 Fig and Fig 9).
Initial experiments, performed with some SG differing in their glycan structure, namely UFH, S.l.-SP#1, PS3 and CurS3, revealed divergent effects (Fig 9). The two β-1,3-glucan sulfates PS3 and CurS3 improved the FXIIa inhibition by C1-INH, reaching a C1-INH potentiation of about 124% by 12.5 μg/ml PS3 and about 107% by 12.5 μg/ml CurS3, whereas UFH rather tended to attenuate the FXIIa inhibition by C1-INH. S.l.-SP led to an even more pronounced Examining the SG in absence of C1-INH resulted in similar direct effects on FXIIa (Fig 9). Whereas UFH only non-significantly improved the FXIIa activity, S.l.-SP displayed a bellshaped curve with a significant increase of the FXIIa activity at a concentration of 1.25 μg/ml. In contrast, both β-1,3-glucan sulfates (PS3 and CurS3)-apart from a non-significant activity increase at 0.125 μg/ml-concentration-dependently reduced the FXIIa activity by up to 56% and 52%, respectively (Fig 9C and 9D). To elucidate these divergent findings, we aimed to investigate the structure-dependent effects of SG on the FXIIa activity in more detail.
β-1,3-glucan sulfates. The high-molecular mass curdlan sulfates displayed opposing effects on the FXIIa activity in dependence on their DS. The lower sulfated compounds CurS1 (DS 0.64) and CurS2 (DS 1.34) improved the FXIIa activity similar to S.l.-SP, whereby the rather low concentration of 1.25 μg/ml CurS1 caused the strongest increase, i.e. by 25%. As opposed to this, the higher sulfated CurS3 (DS 1.74) and CurS3a (DS 1.78) led to a decrease of FXIIa activity concentration-dependently (Fig 10A). A similar DS-dependent modulation was observed with the low-molecular mass phycarin sulfates. The PhyS with the lowest DS of 0.75 enhanced the FXIIa activity-in contrast to CurS1 and CurS2 -concentration-dependently, whereas PhyS2 (DS 1.48) had nearly no effect. The three PhyS with DS ! 1.80 resulted in a concentration-dependent decrease of FXIIa activity (Fig 10B).
Dextran sulfates. Also the two high-sulfated dextran sulfates inhibited the FXIIa activity concentration-dependently, whereby the high-molecular mass DexS-H was superior to the low-molecular mass DexS-L despite its some lower DS (1.54 vs. 2.02) (Fig 12). The maximum inhibition was again reached at 6.25 μg/ml.

Discussion
The physiological importance of C1-INH, the major regulator of complement and contact system [2], is proven by hereditary and acquired C1-INH-dependent angioedema, which is treated with plasma-derived or recombinant C1-INH [11]. Moreover, C1-INH exhibited therapeutic effects in other diseases and clinical situations [25,24] and may have therapeutic potential in numerous diseases, where the tightly cross-linked complement and contact system are concurrently activated [2]. Due to their C1-INH potentiating effect, sulfated glycans may offer an option to improve either the efficacy of therapeutically applied C1-INH or the activity of endogenous C1-INH. As an elementary step, series of SG were investigated for their C1-INH modulating effect to obtain more information on the structural characteristics of glycans useful for this purpose. Additionally, we examined the effect on a second exemplary serine protease targeted by C1-INH [47,48] to evaluate potential differences concerning the interactions of SG with different enzymes. We have chosen FXIIa, a central enzyme of the contact system, which is additionally involved in complement activation [2].
The following discussion concentrates on a detailed analysis of the found structure-dependent activities to elucidate the respective role of M r , DS and glycan structure for the investigated activities of SG.  Interactions of Sulfated Glycans with C1 Inhibitor, C1s and Factor XIIa C1s, C1 inhibitor and sulfated glycans C1 inhibitor potentiation in dependence on the molecular mass. Heparin was the first compound described to potentiate C1-INH [26,52]. Besides UFH, a number of LMWHs is meanwhile widely used for therapy and prophylaxis of venous thromboembolism. The various LMWHs have different M r profiles [53], but so far it was not known whether this impacts their potentiating effect on the C1-INH-C1s reaction. Testing of four LMWHs and four further heparins representing the M r range of about 1 500 to 35 000 revealed that the C1-INH potentiation by heparins improves with increasing M r and that LMWHs are only half as active as UFH at a rough estimate (S2 Fig). However, based on international units (IU/ml) (LMWHs:~100 IU/ mg, UFH:~200 IU/mg), the potentiating activity of LMWHs is similar to that of UFH and at therapeutic plasma concentrations (LMWHs:~1.0 IU/ml; UFH:~0.7 IU/ml), LMWHs may be even superior to UFH. An exception appears to be MMWH (M r 10 500), which showed only a potentiation similar to TINZA (M r 6 500) at equal gravimetric concentrations. This can be explained by their overlapping M r profiles: whereas MMWH contains 30% (m/m) chains with M r < 8 000, the mass percentage (m/m) of TINZA with M r > 8 000 amounts to 22-36%. Moreover, the markedly lower activity of MMWH compared to UFH (M r 15 000) is plausible considering that UFH consists of chains with M r up to 35 000, whereas MMWH contains only 2.5% of chains with M r 12 000-15 000. In contrast to the activity of the heparin oligosaccharides VLMWH (18.2 ± 2.7% C1-INH potentiation at 6.25 μg/ml), which was in line with the results by Rossi et al. [32], the inactivity of the PENTA oligosaccharides (-0.7 ± 2.8% C1-INH potentiation at 6.25 μg/ml) suggests the requirement of a minimum chain length or a minimum number of sulfate groups, respectively. The latter is concluded from the slight C1-INH potentiating effect (3.3 ± 5.3% at 6.25 μg/ml) of fondaparinux (FPX), the synthetic antithrombin-binding pentasaccharide with 8 (DS 1.60) instead of about 5 sulfate groups, which indicates additional relevance of the DS for the C1-INH potentiation. Since pentasaccharides are too short to act as template between serpins and proteases, e.g. antithrombin and thrombin, the slight activity of FPX supports the sandwich theory of Beinrohr et al. [49], but the found M r dependence questions their assumption that the chain length may be of minor importance.
Likewise, already the results of Wuillemin et al. [28] suggest that at least extremely differing M r impact the C1-INH potentiating activity of SG, since they found that high-molecular mass dextran sulfate (M r 500 000, DS 1.9) was significantly more active than a small dextran sulfate (M r 5 000, DS not indicated). This is confirmed by our findings with dextran sulfates: despite its lower DS, DexS-H (M r 500 000, DS 1.54) was still slightly more active than DexS-L (M r 5 000, DS 2.02) (Fig 5). At the level of molar concentrations, DexS-H was even more than two orders of magnitude more active than DexS-L.
The closely linked dependence of the C1-INH potentiating activity of SG on both M r and DS becomes obvious with the two series of β-1,3-glucan sulfates, namely CurS (DS 0.64-1.78) and PhyS (DS 0.75-3.00). The PhyS have a degree of polymerization of about 25 (23-27) and thus a chain length between that of LWMHs and the MMWH, whereas the CurS represent a M r range of 70 000-210 000.
In case of a low DS (< 0.8), a high M r turned out to be crucial for a C1-INH potentiating effect (active CurS1 (DS 0.64, M r 210 000) vs. inactive PhyS1 (DS 0.75, M r 6 000)) ( Fig 6B). In contrast, the activity of SG with DS ! 1.34, i.e. higher than that of heparins, was independent of the M r (CurS2 vs. PhyS2; CurS3 vs. CurS3a vs. PhyS3) (Fig 6A) or only M r -dependent at a molar level, respectively.
In the present study, primarily equal gravimetric concentrations of the SG were compared, since exact molar concentrations cannot be determined due to the inherent polydispersity of glycans. However, this may be associated with an underestimation of the role of the M r as shown by additional experiments using approximately similar molar concentrations. According to these assays, also the activity of SG with DS ! 1.34 improved with increasing M r (up to at least M r 70 000). Nevertheless, such a discrepant M r -dependence confirms its subordinate role for SG with higher DS. It may partly be due to a changed stoichiometry of the molecular interaction, i.e. one long SG chain may bind to more than one pair of C1-INH/C1s, whereas the binding affinity does not further increase above a certain chain length.
Thus, the C1-INH potentiating activity improves with increasing M r of the SG, whereby its impact becomes crucial, if the DS is below a value in the range of 1.2 as suggested by the data on heparins and β-1,3-glucan sulfates. In this way, a high M r can compensate a low DS. Such a limited importance of the M r is in line with the proposed sandwich mechanism for C1-INH potentiation [49].
The compensatory effect of high M r is confirmed by the algae-derived SG having a DS < 1.0 (Fig 7): The C1-INH potentiation by S.l.-SP#1 (DS 0.58, M r 534 000) and S.l.-SP#2 (DS 0.35, M r 541 000) was similar to that of UFH, and that of the other algae-derived SG was in the range of that of the LMWHs. Notably, the high-molecular mass algae-derived SG with DS < 0.5 (S.l.-SP#2, C.t.-SP) exhibited activity, whereas the GAGs with DS < 0.5 and M r < 80 000 were inactive (Fig 4). Moreover, the activity of partially depolymerized fucose-containing sulfated polysaccharides improved with increasing M r . But in line with the findings for heparins and β-1,3-glucan sulfates, there is obviously a certain upper limit, which differs depending on both the DS and the basis of comparison (gravimetric vs. molar concentrations) (Fig 8, Table 2). In case of the very low-sulfated S.l.-SP, a plateau of the activity is only observed at M r > 200 000 (molar comparison).
The C1-INH potentiating activity of the tested SG depending on their M r is summarized in Fig 13. C1 inhibitor potentiation in dependence on the degree of sulfation. In line with earlier findings by Wuillemin et al. [28], the inactivity of the two dextrans (Fig 5) showed that the presence of sulfate groups is essential for C1-INH potentiation by glycans. The activities of the compounds within each SG series (heparins, heparan sulfates, chondroitin sulfates, PhyS, S.l. -SP, F.v.-SP, sulfated xylogalactans (D.s.-SP, C.t.-SP)) revealed a correlation between the DS and the activity (Fig 3, Fig 4, Fig 7). The similar activity of PhyS3 (DS 1.80), PhyS4 (DS 2.21), and PhyS5 (DS 3.00) suggests a certain upper limit for the DS-dependent activity increase, which is however additionally influenced by the M r and possibly by further structural parameters. The minimum DS for any activity turned out to depend on both the M r and the glycan structure and widely varied from 0.34 (S.l.-SP#2) to 1.6 (FPX) among the tested SG.
Strikingly, some SG with DS So far, there is hardly any data whether polyanions other than SG potentiate the C1-INH. Early experiments performed with serum instead of C1-INH revealed a moderate potentiation of C1s inhibition by polyvinyl sulfonate, whereas sodium polyanethol sulfonate (Liquoid 1 , Grobax 1 ) neutralized the C1s inhibiting effect of serum [26]. We were able to confirm that the glycan structure is not essential by testing suramin, a symmetric molecule consisting of four benzene units and two naphthalene substituted with six sulfonate groups (M r 1 429). But in light of its moderate activity of about 20% C1-INH potentiation at 6.25 μg/ml (4.37 μmol/l) compared to almost maximum potentiation by only 0.625 μg/ml (0.07 μmol/l) PhyS3, glycan structures may be more suitable to potentiate C1-INH. Polyphosphate, i.e. linear polymers of orthophosphate (NaPO 3 ) linked by phosphoanhydride bonds, was reported to bind to C1-INH [54], and just published data demonstrate the C1-INH potentiating activity of polyphosphate consisting of 130 units [55]. Based on the tested gravimetric and molar concentrations its effect was, however, considerably weaker than that of heparin. This is in line with our preliminary data on curdlan phosphate (-2.4 ± 1.8% C1-INH potentiation at a final concentration of 6.25 μg/ml) indicating that phosphate groups might be less prone than sulfate groups to build the sandwich complex for potentiation. Similarly, uronic acids, which constitute 50% of the monosaccharides of the GAGs, seem to make only a small contribution to the C1-INH potentiation as concluded from the slight potentiating activity by both HA and oligoHA. The apparently contradictory effect of sodium alginate (-5.1 ± 1.1% C1-INH potentiation at a final concentration of 6.25 μg/ml) could be explained by its ordered conformation due to complex formation (egg-box model) in the presence of Ca 2+ and Mg 2+ [56], which may prevent sandwich complex formation.
The C1-INH potentiating activity of the tested SG depending on their DS is summarized in Fig 14. C1 inhibitor potentiation in dependence on the glycan structure. The complexity of Fig  13 and Fig 14 illustrates that the C1-INH potentiating activity of the tested SG does not only depend on their DS and M r , but also on their glycan structure. The comparison of SG with different structures, but either similar activity or appropriate DS and M r (Table 3) suggests the following effects of the glycan structure on the C1-INH potentiating activity: (1) OSHep was more active than OSCS despite its some lower DS and M r . This may be due to the high iduronic acid content of heparins, whereas chondroitin sulfates exclusively contain glucuronic acid. A unique feature of iduronic acid is its conformational flexibility, which is considered to be advantageous for interactions with biomolecules [57]. (2) CurS1 was as active as DexS-H having a 2.4 timer higher DS and PhyS2 (DS 1.48) was more active than DexS-L (DS 2.02) so that a linear β-1,3-glucan structure seems to be more suitable to form sandwich complexes than the highly branched α-1,6-glucan structure of dextrans. (3) As indicated by the superiority of PhyS2 and PhyS3 over OSHep, the β-1,3-glucan structure is supposed to be also more favourable than that of heparins, whereby this could also be due to their homogenous composition, whereas heparins consist of a complex mixture of heterogeneous molecules [57]. (4) Among the algae-derived SG, the fucose-containing sulfated polysaccharides (syn. fucoidans) from brown algae turned out to represent better C1-INH modulators than xylogalactans from red algae. This was shown by the higher activity of the low-sulfated, depolymerized S.l.-SP#2 samples compared to C.t.-SP and D.s.-SP as well as by F.v.-SP#b, which was as active as D.s.-SP despite its lower M r and DS (Table 3) .

FXIIa, C1 inhibitor and sulfated glycans
State of research. The tested SG displayed highly divergent effects on the C1-INH-C1s interaction ranging from certain C1-INH neutralization up to potentiation stronger than that by dextran sulfates, which have so far been described as the most potent compounds [28]. Therefore, the question arose how these SG modulate the interactions of C1-INH with other serine proteases.
Of these, FXIIa, appeared to be the most interesting one, because it initiates not only contact activation, but also triggers the C1-mediated complement activation [2,58,38] and is predominantly inhibited by C1-INH like C1s [47,48]  Direct FXIIa activity modulation independent of C1 inhibitor. In part, our initial results seemed to confirm the distinct effects of SG on the FXIIa inhibition by C1-INH, namely no significant change by UFH and attenuation of the C1-INH activity by the very-high-molecular mass S.l.-SP up to complete recovery of FXIIa activity. Strikingly, however, the semi-synthetic β-1,3-glucan sulfates PS3 and CurS3 apparently potentiated C1-INH. Ultimately, this unique activity was due to direct inhibition of the amidolytic activity of FXIIa so that the improved FXIIa inhibition by C1-INH resulted from an additive effect. To our knowledge such a direct influence on FXIIa was not observed before. So far, dextran sulfate and GAGs have only been described to directly affect the amidolytic activity of both kallikrein and FXIa [34,35,59] and to reduce also the catalytic activity of kallikrein on the hydrolysis of plasminogen [37].
Moreover, the reduction of the FXIIa inhibition by C1-INH in the presence of S.l.-SP and other SG was shown to be not due to protection of FXIIa from inactivation as previously suggested [36,34,28] but due to direct stimulation of its amidolytic activity.
Beinrohr et al. [49] proposed that the C1-INH potentiating activity of polyanions correlates with the positive charges on the contact site of the serine proteases. Accordingly, the most significant effect is therefore observed with FXIa carrying the most positive charges, whereas the strikingly acidic FXIIa prevents the formation of the sandwich complex and even impairs the inactivation by C1-INH. However, this hypothesis conflicts not only with our observations of direct effects of SG on the FXIIa activity, but also with the identified binding site of FXII for negatively charged surfaces on the fibronectin type II domain of FXII, in particular residues 39 to 47 [60], and the found binding of OSCS on both FXII and FXIIa [61].
Structure-dependent FXIIa activity modulation by sulfated glycans. Although the mode of action of SG on FXIIa needs further investigation, e.g. studies in a more physiological system, testing of a considerable number of SG provided the following information about structure-activity relationships.
The type of modulation was decisively determined by the DS: (1) All the tested SG with DS ! 1.35 inhibited FXIIa with an optimum at DS 1.8-2.2, whereby maximum inhibition was observed at 6.12 μg/ml and did not exceed about 60%. (2) Heparins, i.e. SG with a DS of about 1.0, did not significantly modulate the FXIIa activity.
Besides the DS, the M r of the SG showed to play a role. The FXIIa stimulating SG with Mr 150 000, like HS11-08, exhibited a concentration-dependent activity increase (highest concentration 12.5 μg/ml), whereas larger ones, e.g. S.l.-SP#1, displayed a bell-shaped curve with maximum stimulation at 1.25 μg/ml. In the case of the FXIIa inhibiting high-sulfated compounds, those with high M r (DexS-H, CurS) showed a stimulating effect at the lowest concentration (0.125 μg/ml).
Overall, only high-sulfated SG with a limited M r like PhyS clearly inhibited FXIIa up to a certain extent, whereas both low-sulfated SG and large SG seem to be able to exhibit FXIIa stimulating effects in dependence on their concentration.
Negatively charged surfaces are well-known to induce auto-activation of FXII, but a stimulating effect on the FXIIa activity has not been reported so far. Notably, a recent study demonstrated that FXII, in the presence of polyphosphates (polyP) (scFXII-polyP70), develops the ability to hydrolyze a chromogenic substrate as well as its natural substrates FXI and prekallikrein without becoming a two-chain enzyme and thus FXIIa [62]. Binding of polyP showed to be essential for this activity and a bell-shaped concentration response to polyP was evident, as is typical of surface-mediated reactions. Accordingly, it may be conceivable that also the binding of SG to FXIIa may induce or stabilize the conformation of FXIIa optimal for its enzymatic activity.
What impact the observed direct effects of SG on the activity of FXIIa have in more physiological systems, still needs to be investigated. This is difficult to estimate due to the (patho-) physiological complex role of FXII/FXIIa (coagulation, prekallikrein (bradykinin generation) and complement activation), its regulation by several endogenous inhibitors [47] and the multivalent interactions of SG with involved enzymes and inhibitors. Moreover, SG may induce autoactivation of FXII [63] (e.g. PhyS heparins < OSCS < DexS (unpublished data)). Consequently, it is very challenging to assess the overall effect of SG by any in vitro assays. This is illustrated by the OSCS, misused to falsify heparin, which exhibits both anti-and procoagulant and both complement activating and inhibiting effects in vitro [61,64,30]. The adverse clinical events observed with OSCS-contaminated heparin have been linked with the activation of kinin-kallikrein and complement pathways [64]. However, the majority of patients who received contaminated heparin did not experience an adverse event. To clarify this fact, a study was performed including the analysis of patient data and plasma samples [65]. It revealed that low C1-INH levels turned out as a risk factor for adverse events, whereas high levels were suggested to be protective, which once again emphasizes the importance of an appropriate C1-INH function.

Conclusion
In summary, the evaluated structure-activity relationships indicate that the potentiating effect of SG on the C1s inhibition by C1-INH correlates with their DS and M r , but is additionally influenced by their glycan structure. The dependence on DS and M r are closely linked, which means that the M r plays especially an important role, if the DS is below~1.2. Both the minimum M r needed for any potentiation by low-sulfated SG and the upper M r limit still having an impact, increases with decreasing DS of the SG. Accordingly, SG with a very low DS require a high M r to exhibit any C1-INH potentiation, whereas for SG with a very high DS a chain length of only 5 monosaccharides (FPX) turned out to be sufficient for a slight effect. These observations are consistent with the sandwich mechanism, whereby the polyanion interposes as a negatively charged bridge between positively charged surfaces of C1-INH and C1s [50,49]. Among the glycan structures examined in this in vitro study, both β-1,3-glucans as well as fucose-rich glycans turned out to be favourable structures for the C1-INH potentiation.
In contrast to the C1-INH-C1s interaction, SG do not influence the C1-INH-FXIIa interaction, but SG with DS and M r considerably differing from that of heparins can directly modulate the FXIIa activity. Whereas SG with DS < 1.0 turned out to stimulate the FXIIa activity in a bell-shaped manner, SG with DS > 1.0 concentration-dependently inhibited the FXIIa activity up to a certain extent. However, especially in the case of FXIIa with its complex functions and regulation, SG may additionally modulate other enzymes and inhibitors involved in this network. Further investigations are, therefore, needed, whereby the heparin contaminant OSCS demonstrated that the finally resulting in vivo effect is hardly conceivable from in vitro studies.
As usual in drug development, an essential precondition for the use of a SG as C1-INH amplifier is its safety. In this regard, a corresponding SG should neither stimulate FXIIa nor trigger the FXII auto-activation as it is known from polyanions with high M r [63]. Since SG usually act as multivalent biomodulators [53] and can exhibit additional activities in dependence on their structural parameters, the overall pharmacological profile has to be considered.
Generally, SG appropriate for use as C1-INH amplifiers should be less anticoagulant than heparins. Additional anti-inflammatory activities are assumed to be beneficial. Given the reduced adverse effects (e.g. heparin-induced thrombocytopenia) [66] and improved pharmacokinetics of LMWHs compared to UFH [53], a rather low M r is supposed to be advantageous for in vivo applicable SG.
Taking into account these aspects as well as the results of the present study, PhyS, i.e. lowmolecular mass β-1,3-glucan sulfates, may represent promising candidates for further investigations. Compared to heparins and other GAGs, PhyS have the advantage that they are chemically defined, semi-synthetic glycans of non-animal origin. Though PS3 proved to have pronounced in vivo anti-inflammatory effects [42,67], a PhyS with a lower DS may be more appropriate for use as C1-INH amplifier. Further, depolymerized, sulfated fucan-rich glycans seem to be worth to be evaluated as lead structures in more detail.
In conclusion, the results of this study represent basic pharmacological information, which may be useful for the initial steps towards developing C1-INH potentiating compounds for improving the current C1-INH replacement therapy or enhancing the potency of endogenous C1-INH in diseases and clinical situations characterized by inappropriate activation of complement and contact system.