Novel interaction of properdin and coagulation factor XI: Crosstalk between complement and coagulation

Abstract Background Evidence of crosstalk between the complement and coagulation cascades exists, and dysregulation of either pathway can lead to serious thromboinflammatory events. Both the intrinsic pathway of coagulation and the alternative pathway of complement interact with anionic surfaces, such as glycosaminoglycans. Hitherto, there is no evidence for a direct interaction of properdin (factor P [FP]), the only known positive regulator of complement, with coagulation factor XI (FXI) or activated FXI (FXIa). Objectives The aim was to investigate crosstalk between FP and the intrinsic pathway and the potential downstream consequences. Methods Chromogenic assays were established to characterize autoactivation of FXI in the presence of dextran sulfate (DXS), enzyme kinetics of FXIa, and the downstream effects of FP on intrinsic pathway activity. Substrate specificity changes were investigated using SDS‐PAGE and liquid chromatography–mass spectrometry (LC‐MS). Surface plasmon resonance (SPR) was used to determine direct binding between FP and FXIa. Results/Conclusions We identified a novel interaction of FP with FXIa resulting in functional consequences. FP reduces activity of autoactivated FXIa toward S‐2288. FXIa can cleave FP in the presence of DXS, demonstrated using SDS‐PAGE, and confirmed by LC‐MS. FXIa can cleave factor IX (FIX) and FP in the presence of DXS, determined by SDS‐PAGE. DXS alone modulates FXIa activity, and this effect is further modulated by FP. We demonstrate that FXI and FXIa bind to FP with high affinity. Furthermore, FX activation downstream of FXIa cleavage of FIX is modulated by FP. These findings suggest a novel intercommunication between complement and coagulation pathways.


| INTRODUC TI ON
The complement and coagulation pathways play a central role in thromboinflammation. Both systems are descended from shared ancestry, 1 and although the interactions between the two are yet to be completely defined, the two systems should not be considered as separate entities. 2 Complement is a tightly regulated protease cascade and is a key player in host defense against microbial infection. This cascade can be activated by three pathways: the classical, lectin, and alternative pathways. These three pathways result in a common terminal pathway, culminating in membrane attack complex (MAC) formation. 3 The classical and lectin pathways are inducible through conformational changes in the primary proteases through antibody-antigen complexes 4 or when in contact with polysaccharides at microbial surfaces. 5 The alternative pathway is different from the inducible classical and lectin pathways, constantly undergoing a process known as "tick-over," the spontaneous hydrolysis of the thioester bond of C3 creating C3(H 2 O), an analogue of C3b. 6 C3(H 2 O) will either bind to pathogen surfaces to recruit further complement components or will remain soluble and be quickly degraded. Properdin (factor P [FP]) and factor B (FB) are recruited, forming a complex with surface bound C3(H 2 O), with FB binding in a Mg 2+ -dependent manner. FB is then activated by factor D, forming the initial C3 convertase. FP is a cofactor for the C3 convertase complex, increasing the half-life 10-fold, 7 and is the only known positive regulator of complement, necessary for alternative pathway activation, and initiates a positive feedback loop, amplifying the terminal pathway and MAC generation culminating in lysis of pathogenic cells. 8 FP is a highly positively charged 53-kDa monomer made of seven thrombospondin type-1 repeats. These monomers associate head to tail to create dimers, trimers, and tetramers in serum. 9 FP circulates in the plasma at a concentration of 4 to 25 µg/mL and is constitutively released from a number of cells 10 including monocytes, 11 dendritic cells, 12,13 endothelial cells, 10 mast cells, 14 and adipocytes. 15,16 FP is also released from stimulated granulocytes, including neutrophils, in the local microenvironment, promoting complement activation; stimuli include tumor necrosis factorα and C5a. 17,18 FP deficiency often displays a phenotype of recurring meningococcal infections, with higher mortality rates when compared to healthy individuals. 19 The intrinsic pathway of coagulation is initiated through contact activation. Prekallikrein (PK) and factor XII (FXII) can undergo a reciprocal activation process enhanced by the presence of artificial and physiological negatively charges surfaces 20 to generate kallikrein (PKa) and activated FXII (FXIIa). FXIIa generated from contact activation initiates the intrinsic pathway of coagulation through the cleavage of FXI to form FXIa. FXIa, 21 along with PKa, [22][23][24][25] can then cleave factor IX (FIX), subsequently leading to factor X (FX) activation, thrombin generation, and fibrin clot formation. It has been demonstrated that in the presence of surfaces such as sulfatides and glycosaminoglycans (GAGs), FXI also can autoactivate and induce coagulation. 26 FXI is a zymogen that circulates at a concentration of 30 nM in the blood, most often in association with its cofactor, high-molecularweight kininogen (HK). 27 FXI is composed of two identical 80-kDa monomers linked by a disulfide bond. Each monomer contains four apple domains with similar structural properties to PK, which also circulates in complex with HK. 28 Deficiency of FXI results in a mild to moderate bleeding disorder known as hemophilia C. FXI can be activated by FXIIa 29,30 and thrombin, 31,32 and FXIa has been shown to have numerous natural substrates including FIX, factor V, FX, prochemerin, and complement regulatory protein factor H (FH). [33][34][35] Proteases from both the complement and contact activation systems have similar structural functions and characteristics, and it has been demonstrated that the classical pathway of the complement system can be initiated by FXIIa. 36 Also, the common pathway of complement can be initiated through the cleavage of C3 by PKa. 37 There is evidence that the lectin pathway of complement may interact with the kallikrein-kinin system through mannose-binding lectin-associated serine protease-1 cleavage of HK leading to bradykinin release. 38 The primary inhibitor of the complement classical pathway, C1 esterase inhibitor (C1-INH) can also inhibit the intrinsic pathway of coagulation 39 Figure S1).

| Michaelis-Menten kinetics
Three nanoMolar of FXIa was incubated with and without 0.6 µg/mL DXS, in the presence and absence of 25 µg/mL FP using a twofold serial dilution of S-2288 (0.05-3.0 mM). FXIa catalytic activity was determined by monitoring cleavage of S-2288.

| Substrate specificity assays
Substrate specificity was analyzed using reducing SDS-PAGE.
Incubations were performed and all samples were diluted twofold into a running buffer containing reducing agent, LDS sample buffer, and HBS. Samples were separated by reducing SDS-PAGE at 100 V for 52 minutes. The gel was stained with InstantBlue protein stain overnight at 21°C with gentle shaking, and washed three times with water with gentle shaking for 5 minutes. Gels were imaged using Syngene G:BOX Chemi and GeneSys software (Syngene, Bengaluru, India).

| FXI in the presence of DXS and FP
Reactions of 100 µg/mL FXIa and 200 µg/mL FP in the presence of 12.5 µg/mL DXS were incubated for 120 minutes in HBS and were diluted into the running buffer. Bands of interest were analyzed by liquid chromatography-mass spectrometry (LC-MS) (Appendix S1) by the Biomolecular Mass Spectrometry Facility at the University of Leeds.

| FXIa substrate specificity
Reactions of 100 µg/mL FXIa, 100 µg/mL FIX, and 200 µg/mL FP in the presence of 12.5 µg/mL DXS were incubated, and samples were taken over a course of 60 minutes and were diluted into the running buffer.

| Binding of FXI and FP using surface plasmon resonance
SPR was performed using the Pall/ForteBio Pioneer biosensor platform (Molecular Devices, LLC, San Jose, CA, USA) (Appendix S1). FP was immobilized to the sensor surface to around 1 × R MAX 79.55 RU using the amine coupling method as previously described. 45 After priming three times, 50 nM of FXI or FXIa was injected over the sensor surface for the first experiment using a OneStep 100% loop inject, using Taylor dispersion to create a concentration gradient through the capillary tube before entering the flow cell (FC), at a flow rate of 30 µL/min with a dissociation time of 300 seconds.
The sensor surface was regenerated using 5 µL of 1 M NaCl, 3 mM NaOH injected at 60 µL/min, with a dissociation time of 30 seconds.
To determine the binding kinetics, a OneStep assay was performed.
Response curves from FC2 were subtracted from FC1 and FC3. Buffer blanks were averaged and subtracted from the binding curves. The assay was run in triplicate, and the standard error of the mean was calculated using Prism 8 (GraphPad Software, San Diego, CA).

| Phospholipid vesicle preparation
Phospholipids were prepared as previously described. 46 Micelles were formed from phospholipids supplied in chloroform, using 20%

| Data analysis and statistics
All data figures were created using GraphPad Prism 8 unless otherwise stated.

| Statistical analysis
All analysis was performed in GraphPad Prism 8 unless otherwise stated. Statistical differences were analyzed by one-way analysis of variance (anova), and tests are highlighted in figure legends where appropriate.

| Amidolytic activity of FXI autoactivated by DXS is modulated in the presence of FP
This study was initiated by investigating the effect of FP on FXI autoactivation by the synthetic GAG DXS. In a purified system, physiological concentrations of FP and FXI were incubated in the presence and absence of DXS for 90 minutes. The presence of DXS rapidly induced amidolytic activity of FXI toward S-2288 via autoactivation to FXIa ( Figure 1A). The addition of FP (5-25 µg/ mL) to the autoactivation reaction drastically reduced cleavage of S-2288 in a dose-dependent manner ( Figure 1B). To determine if this effect was due to the cationicity of FP, the same experiment was repeated using a titration of the cationic protein protamine sulfate, a clinical reagent applied for the reversal of heparin anticoagulation. 47 It was observed that protamine sulfate reduced the cleavage of chromogenic substrate by FXI autoactivation in the presence of DXS, though this was not dose dependent with the employed concentrations ( Figure S1). To determine if the decreased cleavage of the chromogenic substrate was due to a change in substrate specificity, reactions of FXI, DXS, and FP were incubated for 120 minutes and were analyzed using reducing SDS-PAGE. Novel cleavage bands were revealed and were analyzed by LC-MS (Table S1). The cleavage products were found to be FP, suggesting a change in substrate specificity of FXIa away from the chromogenic substrate and toward FP. Similar reactions were performed and analyzed using SDS-PAGE; however, FXI was replaced with its activated form, purified FXIa (preactivated by FXIIa) ( Figure 1D). The cleavage products of FP were still visible, suggesting that FP can be cleaved by FXIa autoactivated by DXS, or by FXIa activated by FXIIa. These data suggest a surfacedependent interaction between FXI and FP.

| DXS modulates the kinetics of FXIa, but this is further modified in the presence of FP
Subsequently, we explored FXIa activity in the presence and absence of DXS and FP ( Table 1). The data were analyzed using the k cat

| FXI and FXIa bind to FP with high affinity
Binding studies were performed using SPR to determine if there was a direct interaction between FXIa and FP. FP was immobilized to the sensor surface and FXI and FXIa were titrated over the surface using the OneStep protocol, with a maximum concentration of 50 nM. We found that both FXI and FXIa bind to FP with a K D (equilibrium dissociation constant) of 16.1 nM and 350 pM, respectively ( Figure 3A, B). These data suggest that FP can bind to both activated and zymogen FXI but with a preference to the active enzyme due to a higheraffinity interaction.

| FP modulates FX generation by the intrinsic pathway in purified reactions
The purpose of this experiment was to measure FXIa activation of FIX; however, due to the low catalytic activity of FIXa toward F I G U R E 1 FP acts as a substrate for FXIa in the presence of DXS. A chromogenic assay was employed to determine the effect of FP on the autoactivation of 30 nM FXI by 0.6 μg/mL DXS.

| DISCUSS ION
Complement-coagulation crosstalk has been a topic of research for over 80 years 48 ; however, the mechanisms behind this are still not fully understood, and many thrombotic and inflammatory diseases involve both cascades. Moreover, it is important to understand how agents targeting coagulation may modulate inflammatory pathways and vice versa. Polyanions play an important role within both systems, with several components interacting with these complex molecules. These molecules include polyphosphates, phospholipids, and GAGs. The net negative charge of GAGs has been shown to modulate contact activation of the intrinsic pathway of coagulation, through activation of FXII, 49 and it has been observed that FXI can also autoactivate in the presence of polyanionic surfaces such as DXS and sulfatides. 50 It has been shown that sufficient inhibition of contact knowledge, this is the first-ever report that FP is a substrate for FXIa.
It was also determined that although there was potential for FXIa to cleave FP, it was still able to cleave its natural substrate FIX into FIXa, suggesting that the interaction between FXIa and FP may be a mechanism involved in inflammatory pathways.
We next explored the effect of FP on FXIa amidolytic activity using Michaelis-Menten kinetics. FXIa activity was not affected by the presence of FP alone; however, FXIa activity was modulated by DXS, indicated by decreased V max , k cat , and K m . This modulatory effect of DXS was partially reversed by FP, demonstrated by the rescue of the k cat , and the increase in K m and V max caused by the presence of FP.
These data support the substrate specificity change of FXIa and suggest that FP can interact with FXIa in a surface-dependent manner.
We determined using SPR that there is a direct interaction between FXIa and FP, as both FXI and FXIa bind to FP with high affinity, with K D values in the nanometer and picometer ranges, respectively. Although this binding is determined in a purified system and many other important components are not present, the high af- It has been shown that FXI deficiency improves survival in a murine model of sepsis caused by cecal ligation and puncture 54,55 and that FXI deficiency alters the cytokine response. 56  It is most likely that FP, a highly positively charged protein, is interacting directly with DXS, coating the anionic surface. FP may undergo conformational changes, making it more susceptible to cleavage by FXIa. We demonstrated an interaction between FP and the mechanism of FXa generation via the intrinsic pathway, which was not surface dependent, however this was more obvious when PL were present in the reactions; thus, steric hindrance may be playing a role in the interaction of FP with intrinsic pathway activity, though it is not as simple as this ( Figure 5).
There are limitations to this study, including the availability of FP-deficient plasma and access to a FP knockout mouse model; it is therefore difficult, at this time, to determine whether these interactions are physiological. These observations lead to a range of important questions that need to be answered. This is the first report, to our knowledge, that FP can be cleaved, and therefore the physiological relevance of this mechanism is unknown. If cleavage products are produced, there is potential for implications in inflammatory diseases.
This study demonstrates a novel interaction between the intrinsic coagulation pathway and the alternative pathway of complement.
FXIa is a current target for novel anticoagulation therapeutics, and this study shows the potential for off-target effects of FXI inhibitors in the treatment of thrombotic disorders.

ACK N OWLED G M ENTS
This study was funded by a grant from the British Heart Foundation PG/16/6/31941 and a LARS studentship supported by the University of Leeds. The authors thank Rachel George of the Biomolecular Mass Spectrometry Facility at the University of Leeds for their support and assistance in this work.