THE RATIONALE FOR TARGETING ACTIVATED PROTEIN C IN HEMOPHILIA
APC is activated from its precursor zymogen protein C by thrombin bound to the cofactor thrombomodulin on the surface of endothelial cells . This event is accelerated 20-fold by the binding of protein C to the same cell surface via an interaction between its γ-carboxyglutamic acid (Gla) domain and endothelial protein C receptor (EPCR)  (Fig. 2). Once formed, APC exerts its anticoagulant activity by proteolytically inactivating the cofactors of the intrinsic Xase (fVIIIa) and prothrombinase (fVa) complexes (Fig. 1). Its potency is well established, but was recently demonstrated anew in a family with an unexplained bleeding disorder . We found a c.1611C>A mutation in the thrombomodulin gene (p.Cys537Stop), causing a truncation within the C-terminal transmembrane helix and resulting in highly elevated plasma levels of soluble thrombomodulin. The elevation in circulating thrombomodulin caused systemic protein C activation, and resulted in trauma-induced and some spontaneous bleeding in affected family members. However, one carrier of the thrombomodulin mutation had no history of bleeding, and sequencing revealed that he was also a carrier of the fVLeiden mutation, which causes APC resistance.
The Leiden mutation, c.1691G>A (Arg506Gln), is present in 5% of whites and is the most common cause of thrombophilia . The mutation is at the primary APC cleavage site in fVa and confers partial APC resistance by slowing the rate of inactivation by seven-fold (three-fold when the cofactor protein S is present) . It is sufficiently common that fVLeiden has been found coincidentally in hemophilia patients, and in some cases the bleeding severity appears milder than what would be expected based solely on factor levels (for review, see ). These clinical observations were supported by the rescue of hemostasis in hemophilia mice crossed with fVLeiden mice; however, the effect appeared to be limited to small vessels . The magnitude of the effect of fVLeiden in ameliorating the clinical manifestations of hemophilia is somewhat controversial. However, observations are limited to fVLeiden heterozygotes in whom half of the fVa generated will be inactivated by APC at the normal rate, and half at a rate only 3–7-fold slower. Overall, the data suggest that if fVa can be protected from degradation by APC, normal hemostasis can occur in the absence of the intrinsic Xase complex (i.e. in hemophilia).
AN ACTIVATED PROTEIN C-SPECIFIC SERPIN NORMALIZES BLEEDING IN HEMOPHILIA MODELS
APC is a trypsin-like serine protease, and its principal physiological inhibitor is the serpin (serine protease inhibitor) protein C inhibitor (PCI). However, PCI is highly promiscuous and is a better inhibitor of procoagulant proteases, such as thrombin, fXa, fXIa, fVIIa-TF, than of the anticoagulant APC . Indeed, adding PCI to plasma to a final concentration of 5 μM extends the activated partial thromboplastin time (aPTT) from 1 min to unclottable after 5 min . The challenge, therefore, was to alter the specificity of PCI so as to reduce the inhibition of procoagulant proteases while maintaining or improving inhibition of APC. The serpin mechanism of protease inhibition is well established (Fig. 3a), and the principal determinant of serpin specificity is the amino acid composition of the reactive center loop (RCL). The RCL of the serpin acts as a pseudo-substrate for serine proteases by forming a recognition complex before hydrolysis begins. In the commonly used nomenclature, the bond that is targeted by a protease is between the P1 and P1’ residues, and amino acids to either side are numbered sequentially (with a ’ toward the C-terminus). Proteases typically utilize the sequences from P4 to P3’ to recognize their substrates, with specificity primarily determined by the residue at the P1 position. All clotting proteases, including APC, require a P1 arginine in their substrates, so to confer our desired specificity profile we mutated residues on either side of P1, guided by published structures of thrombin and APC [11▪] (Fig. 3b).
The active sites of thrombin and APC were most different at the S2 and S1’ sites (where P2 and P1’ side chains fit), with APC predicted to be more accommodating of large positively charged side chains. We found that mutating P2 and P1’ to lysine conferred a high degree of specificity for APC over thrombin, and when the mutated PCI was added to plasma at 5 μM the aPTT was unaffected. However, the mutated PCI was not an attractive drug candidate because of its low rate of APC inhibition (280 M−1 s−1, a two-fold reduction relative to wild-type) and short half-life (24 h).
We therefore decided to switch the template serpin to α1-antitrypsin (α1AT) which has a 5–7 day half-life. The P1 Arg ‘Pittsburgh’ variant has been reported in several people, most of whom had associated bleeding due to the change in specificity from neutrophil elastase to coagulation factors [12,13]. The Pittsburgh variant of α1AT also happens to be an excellent inhibitor of APC, with a rate constant of ∼1 × 105 M−1 s−1. When the flanking Lys mutations were made at the P2 and P1’ positions, a similar gain in specificity was obtained with no increase in aPTT observed when spiked into plasma. Importantly, when the mutated α1AT (SerpinPC) was added to normal or hemophilia plasma (in the presence of thrombomodulin to activate the protein C system), it rescued thrombin generation in a dose-dependent manner. The effect of SerpinPC on hemostasis was assessed in a hemophilia mouse model, monitoring clot formation by intravital microscopy after laser-induced injury and blood loss after tail clip. With both challenges, we observed correction of hemostasis. These results demonstrated that inhibition of the protein C anticoagulant pathway by an APC-specific serpin is effective at restoring hemostasis in vivo in the absence of the intrinsic Xase complex, and may be useful in the treatment of hemophilia.
ACTIVATED PROTEIN C SIGNALING IS DEPENDENT ON ENDOTHELIAL PROTEIN C RECEPTOR AND PROTEASE ACTIVATED RECEPTOR-1
In addition to its anticoagulant function, APC also exerts important anti-inflammatory, antiproliferative and cytoprotective activities that have been demonstrated in vitro and in animal models. Indeed, recombinant APC was approved as a treatment for sepsis (marketed by Eli Lilly as Xigris), although it was later withdrawn after a placebo-controlled clinical trial (PROWESS-SHOCK) failed to demonstrate efficacy . It is therefore important to consider whether inhibition of APC might have unwanted proinflammatory effects by reviewing what is known about how APC exerts its signaling activities.
As mentioned above, EPCR binding to protein C accelerates the formation of APC, thereby exerting an anticoagulant function. However, EPCR is absolutely essential for all of the signaling activities attributed to APC, by localizing APC to the endothelial cell surface where it can cleave and activate the G-protein coupled receptor, protease activated receptor 1 (PAR-1) (Fig. 2). The dependence of APC signaling on both EPCR and PAR-1 was shown in vitro and in mouse models . APC cleavage of PAR-1 produces anti-inflammatory and cytoprotective effect, but paradoxically thrombin cleavage of PAR-1 (25 000-times faster than APC) has proinflammatory, proapoptotic effects and reduces endothelial barrier function (reviewed in ). Since thrombin, as the protein C activator, is necessarily present at the time APC is formed on endothelial cells, it is unclear which PAR-1 signaling pathway is dominant with endogenous proteins in the normal physiological setting. This issue was partially resolved by some recent studies by Ray Rezaie's group, demonstrating that occupancy of EPCR by protein C or APC is sufficient to elicit protective signaling, even if cleavage of PAR-1 is mediated by thrombin [18,19,20▪▪]. It was concluded that in the physiological setting where EPCR is occupied by protein C or APC, PAR-1 cleavage by either thrombin or APC will elicit anti-inflammatory and cytoprotective effects.
PROTEIN C DEFICIENCY
Protein C deficiency in humans is associated with a 10-fold increase in risk of early and recurrent venous thrombosis, and homozygous deficiency causes neonatal purpura fulminans, a life-threatening condition characterized by microvascular thrombosis. Similar effects are seen in animal models, with full deficiency in mice leading to neonatal death caused by disseminated intravascular coagulation (DIC) . The importance of the protein C levels in thrombosis is therefore well established. To investigate the importance of endogenous protein C in inflammation, partial knockout mice were made with levels ranging from 1 to 18% of normal [22,23]. Mice with levels below 3% were prone to DIC, similar to complete knockouts. Surviving mice with low protein C levels were found to be highly susceptible to lipopolysaccharide (LPS) challenge, exhibiting increased inflammatory markers and signs of DIC. In contrast, mice with levels approximately 18% of normal were protected from low-dose LPS challenge. It is interesting to note that warfarin treatment typically reduces the amount of correctly processed Gla domain to about 20% for vitamin K-dependent coagulation factors, including protein C, and its use has not been associated with any inflammatory disease.
SELECTIVE INHIBITON OF ACTIVATED PROTEIN C ACTIVITIES
To investigate the relative importance of the anticoagulant and anti-inflammatory effects of protein C/APC in mice, anti-APC monoclonal antibodies were generated by the Esmon group . MAPC1591 was found to bind specifically to APC over protein C and inhibit only its anticoagulant function, presumably by sterically interfering with binding to its substrate fVa. MAPC1591 did not, however, inhibit endothelial cell binding, consistent with an epitope on the catalytic domain of APC. In contrast, MPC1609 did not distinguish between protein C and APC, and completely blocked binding to endothelial cells, suggesting an epitope on the Gla domain. Both antibodies inhibited the anticoagulant activity of APC in a one-stage coagulation assay. However, only MPC1609 blocked the cytoprotective effect of APC in vitro, consistent with previous studies showing the dependence of EPCR binding for APC signaling. These antibodies were then given to mice at a dose of 10 mg/kg, followed by administration of a sublethal dose of LPS. All MPC1609-treated mice died within a 3-day window, whereas all of the MAPC1591 and vehicle-treated mice survived. Serum interlukin-6 (IL-6) and kidney function markers [blood urea nitrogen (BUN) and creatinine] were elevated with MPC1609 treatment, whereas none were elevated in MAPC1591-treated group. Indeed, IL-6, creatinine and BUN levels were even lower in the MAPC1591 group than vehicle-treated controls, indicating a possible protective effect of inhibiting the anticoagulant activity of APC in this model. Thrombin-antithrombin complex levels were higher than control in the MAPC1591-treated animals, confirming effective inhibition of APC anticoagulant activity. Similar results were obtained with these antibodies in mouse models of gram-negative pneumosepsis  and renal ischemia reperfusion injury [26▪]. The conclusion from these studies was that the anticoagulant function of endogenous APC is not required for its anti-inflammatory or cytoprotective activities. Treatment of hemophilia by inhibiting APC should therefore be safe, provided that the inhibitor selectively targets its anticoagulant function.
COMPARTMENTALIZATION OF ACTIVATED PROTEIN C ACTIVITIES
The signaling activities of APC are dependent on binding to EPCR on endothelial cells, and the anticoagulant activity of APC requires release from EPCR and diffusion to the site where clotting is occurring. If PAR-1 cleavage by APC (and not thrombin) is essential for its signaling activities, administration of an APC inhibitor would, in theory, inhibit both its anticoagulant and anti-inflammatory functions. However, provided that APC inhibition is slower than its release from EPCR on which it was formed, then only its anticoagulant function can be affected.
Protein C and APC bind to EPCR with identical affinity (Kd of ∼30 nM) , and as protein C circulates at a higher concentration than APC (70 nM vs. 40pM) , once APC dissociates from EPCR it will not reassociate, limiting its activity to anticoagulation. So, we can think of APC as existing in two compartments: the EPCR-bound signaling pool and the circulating anticoagulant pool. The key to ensuring selective inhibition of the anticoagulant pool of APC is to tailor the rate of inhibition so that it is slower than the rate of dissociation from EPCR. The dissociation rate of APC from EPCR is reported to be 0.09 s−1, giving newly formed APC a residence half-life of less than 8 s. SerpinPC inhibits APC with a second-order rate constant of approximately 15 000 M−1 s−1, and would require a concentration of 5 μM (equivalent to a dose of more than 10 mg/kg) to achieve a t1/2 of inhibition of 8 s. Therefore, provided that efficacy can be demonstrated in hemophilia at doses below 10 mg/kg, there should be no effect on APC-mediated signaling. The LPS studies using MAPC1591 were conducted at a dose of 10 mg/kg (about 1 μM peak concentration), and antibodies typically bind their targets with rate constants of approximately 105 M−1 s−1, which translates to a t1/2 of inhibition of 7 s. It may, therefore, be possible to safely exceed doses of an inhibitor that would theoretically inhibit APC before it dissociates from EPCR. This makes sense because cleavage of PAR1 must occur faster than dissociation of APC from EPCR. Indeed, a dose of 20 mg/kg MAPC1591 was ‘safe’ in the renal ischemia reperfusion model [26▪]. On the basis of studies from the Rezaie group, it is also possible that inhibition of the proteolytic activity of APC will have no effect on signaling, even at infinitely high doses of an inhibitor, as protective signaling can be achieved by thrombin cleavage of PAR-1 so long as protein C/APC is bound to EPCR.
Hemophilia treatment has remained essentially unchanged for over 100 years; blood transfusions became plasma infusions became factor infusions. Safety was greatly improved with the introduction of recombinant factors, but efficacy was still limited by short half-lives. Recently, there has been a surge in innovation in the hemophilia space, including gene therapy, an fVIIIa-mimicking bispecific antibody, a small interfering RNA molecule against antithrombin and antibodies against tissue-factor pathway inhibitor. The latter two approaches aim to rebalance hemostasis by reducing intrinsic anticoagulant activity. Why then has no APC inhibitor been developed? One explanation is the fear of affecting its anti-inflammatory activities. It may also have been unclear how effective an APC-inhibitor would be in treating hemophilia, as reports of amelioration of clinical severity with coinheritance of fVLeiden are mixed, and in mouse models fVLeiden only partially rescues hemostasis. We created an APC-specific serpin (SerpinPC) and found that it rescued hemostasis in an intravital microscopy experiment and in the more severe tail-clip model using hemophilia B mice. We have since produced mammalian-expressed SerpinPC and have demonstrated efficacy at very low doses in hemophilia A mice (unpublished data). The safety concerns, however, persist, and although the effect of long-term administration of an APC inhibitor on inflammation is difficult to predict, it is clear that the anticoagulant and anti-inflammatory functions of APC are executed by two separate pools, circulating and EPCR-bound. The template of SerpinPC, α1AT, does not bind to endothelial cells, and does not affect the binding of protein C/APC to EPCR. SerpinPC is therefore unlikely affect the anti-inflammatory functions of APC, even at high doses, but this will need to be tested in LPS-challenged mice.
S.G.I.P. was supported by a Wellcome Trust studentship.
Financial support and sponsorship
S.G.I.P. is currently supported by a grant from Cambridge University Hospital Trust.
Conflicts of interest
J.A.H., T.P.B. and S.G.I.P. have shares in ApcinteX Ltd, a company founded to develop SerpinPC.
REFERENCES AND RECOMMENDED READING
Papers of particular interest, published within the annual period of review, have been highlighted as:
- ▪ of special interest
- ▪▪ of outstanding interest
1. Esmon CT. The protein C pathway. Chest 2003; 124:S26–S32.
2. Mohan Rao LV, Esmon CT, Pendurthi UR. Endothelial cell protein C receptor: a multiliganded and multifunctional receptor. Blood 2014; 124:1553–1562.
3. Langdown J, Luddington RJ, Huntington JA, et al. A hereditary bleeding disorder resulting from a premature stop codon in thrombomodulin (p.Cys537Stop). Blood 2014; 124:1951–1956.
4. Van Cott EM, Khor B, Zehnder JL. Factor V Leiden. Am J Hematol 2016; 91:46–49.
5. Egan JO, Kalafatis M, Mann KG. The effect of Arg306 → Ala and Arg506 → Gln substitutions in the inactivation of recombinant human factor Va by activated protein C
and protein S. Protein Sci 1997; 6:2016–2027.
6. Franchini M, Lippi G. Factor V Leiden and hemophilia. Thromb Res 2010; 125:119–123.
7. Schlachterman A, Schuettrumpf J, Liu JH, et al. Factor V Leiden improves in vivo hemostasis
in murine hemophilia models. J Thromb Haemost 2005; 3:2730–2737.
8. Geiger M, Wahlmüller F, Furtmüller M. The Serpin
Family. Switzerland: Springer; 2015.
9. Huntington JA, Polderdijk SGI, Baglin TP. Modified serpins for the treatment of bleeding disorders. June 18, 2015. EP3080157A1.
10. Huntington JA, Read RJ, Carrell RW. Structure of a serpin
-protease complex shows inhibition by deformation. Nature 2000; 407:923–926.
11▪. Polderdijk SGI, Adams TE, Ivanciu L, et al. Design and characterization of an APC-specific serpin
for the treatment of hemophilia. Blood 2017; 129:105–113.
This article describes the design and testing of the specific anti-APC serpin. It demonstrated remarkable efficacy, even in large vessels.
12. Lewis JH, Iammarino RM, Spero JA, et al. Antithrombin Pittsburgh: an alpha1-antitrypsin variant causing hemorrhagic disease. Blood 1978; 51:129–137.
13. Owen MC, Brennan SO, Lewis JH, et al. Mutation of antitrypsin to antithrombin. Alpha 1-antitrypsin Pittsburgh (358 Met leads to Arg), a fatal bleeding disorder. N Engl J Med 1983; 309:694–698.
14. Abraham E, Laterre P-F, Garg R, et al. Drotrecogin alfa (activated) for adults with severe sepsis and a low risk of death. N Engl J Med 2005; 353:1332–1341.
15. Riewald M, Petrovan RJ, Donner A, et al. Activation of endothelial cell protease activated receptor 1 by the protein C pathway. Science 2002; 296:1880–1882.
16. Cheng T, Liu D, Griffin JH, et al. Activated protein C
blocks p53-mediated apoptosis in ischemic human brain endothelium and is neuroprotective. Nat Med 2003; 9:338–342.
17. Popović M, Smiljanić K, Dobutović B, et al. Thrombin and vascular inflammation. Mol Cell Biochem 2012; 359:301–313.
18. Bae J-S, Rezaie AR. Protease activated receptor 1 (PAR-1) activation by thrombin is protective in human pulmonary artery endothelial cells if endothelial protein C receptor
is occupied by its natural ligand. Thromb Haemost 2008; 100:101–109.
19. Bae J-S, Rezaie AR. Thrombin inhibits HMGB1-mediated proinflammatory signaling responses when endothelial protein C receptor
is occupied by its natural ligand. BMB Rep 2013; 46:544–549.
20▪▪. Roy RV, Ardeshirylajimi A, Dinarvand P, et al. Occupancy of human EPCR by protein C induces (-arrestin-2 biased PAR1 signaling by both APC and thrombin. Blood 2016; 128:1884–1893.
This is one of several articles by the Rezaie group demonstrating that occupancy of EPCR by APC/protein C is sufficient for protective signaling, even if PAR-1 is cleaved by thrombin.
21. Jalbert LR, Rosen ED, Moons L, et al. Inactivation of the gene for anticoagulant protein C causes lethal perinatal consumptive coagulopathy in mice. J Clin Invest 1998; 102:1481–1488.
22. Lay AJ, Liang Z, Rosen ED, et al. Mice with a severe deficiency in protein C display prothrombotic and proinflammatory phenotypes and compromised maternal reproductive capabilities. J Clin Invest 2005; 115:1552–1561.
23. Lay AJ, Donahue D, Tsai M-J, et al. Acute inflammation is exacerbated in mice genetically predisposed to a severe protein C deficiency. Blood 2007; 109:1984–1991.
24. Xu J, Ji Y, Zhang X, et al. Endogenous activated protein C
signaling is critical to protection of mice from lipopolysaccaride-induced septic shock. J Thromb Haemost 2009; 7:851–856.
25. Kager LM, Wiersinga WJ, Roelofs JJ, et al. Endogenous protein C has a protective role during Gram-negative pneumosepsis (melioidosis). J Thromb Haemost 2013; 11:282–292.
26▪. Lattenist L, Jansen MPB, Teske G, et al. Activated protein C
protects against renal ischaemia/reperfusion injury, independent of its anticoagulant properties. Thromb Haemost 2016; 116:124–133.
This is the latest article testing the effect of the anti-protein C/APC antibodies on APC-mediate cytoprotection.
27. Fukudome K, Kurosawa S, Stearns-Kurosawa DJ, et al. The endothelial cell protein C receptor. Cell surface expression and direct ligand binding by the soluble receptor. J Biol Chem 1996; 271:17491–17498.
28. Gruber A, Griffin JH. Direct detection of activated protein C
in blood from human subjects. Blood 1992; 79:2340–2348.
29. Preston RJS, Villegas-Mendez A, Sun Y-H, et al. Selective modulation of protein C affinity for EPCR and phospholipids by Gla domain mutation. Febs J 2005; 272:97–108.
Keywords:Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
activated protein C; endothelial protein C receptor; hemostasis; protease activated receptor-1; serpin