Xinkang Wang, Department of Thrombosis Research, Bristol-Myers Squibb Company, 311 Pennington-Rocky Hill Road, Pennington, NJ 08534, USA. Tel.: 609 818 5551; fax: 609 818 7877; e-mail: email@example.com
Summary. Factor XI (FXI) and factor IX (FIX) are zymogens of plasma serine proteases required for normal hemostasis. The purpose of this work was to evaluate FXI and FIX as potential therapeutic targets by means of a refined ferric chloride (FeCl3)-induced arterial injury model in factor-deficient mice. Various concentrations of FeCl3 were used to establish the arterial thrombosis model in C57BL/6 mice. Carotid artery blood flow was completely blocked within 10 min in C57BL/6 mice by application of 3.5% FeCl3. In contrast, FXI- and FIX-deficient mice were fully protected from occlusion induced by 5% FeCl3, and were partially protected against the effect of 7.5% FeCl3. The protective effect was comparable to very high doses of heparin (1000 units kg−1) and substantially more effective than aspirin. While FXI and FIX deficiencies were indistinguishable in the carotid artery injury model, there was a marked difference in a tail-bleeding-time assay. FXI-deficient and wild-type mice have similar bleeding times, while FIX deficiency was associated with severely prolonged bleeding times (>5.8-fold increase, P < 0.01). Given the relatively mild bleeding diathesis associated with FXI deficiency, therapeutic inhibition of FXI may be a reasonable strategy for treating or preventing thrombus formation.
The plasma serine protease zymogens factor IX (FIX) and factor XI (FXI) are key components of a pathway that is thought to sustain thrombin production at a wound site to maintain fibrin clot integrity [1,2]. Fibrin formation is initiated when vessel injury exposes FVII or its activated form (FVIIa) in plasma to the integral membrane protein tissue factor (TF) [1–3]. The factor VIIa/TF complex converts FX to the active form, FXa, through limited proteolysis, and FXa subsequently converts prothrombin to the key coagulation protease thrombin. FVIIa/TF also activates FIX to active FIXa (sometimes referred to as the ‘extrinsic’ pathway for FIX activation), which contributes to hemostasis through the activation of additional FX. Thrombin catalyzes multiple reactions important to hemostasis, including the transformation of soluble fibrinogen molecules to fibrin and activation of platelets. Furthermore, thrombin amplifies its own generation by activating the plasma cofactors FV and FVIII. In addition, thrombin can activate FXI in a reaction that may occur on the surface of activated platelets [4–6]. Activated FXI (FXIa) contributes to sustained thrombin generation by activating FIX [7–9]. This ‘intrinsic’ pathway for FIX activation appears to be particularly important for maintaining fibrin clot integrity in tissues with high intrinsic fibrinolytic activity .
The contribution of FXIa-mediated activation of FIX to pathologic thrombosis has generated considerable interest. Epidemiologic work has demonstrated that individuals with high FIX or FXI levels have a modestly increased risk of venous thromboembolism compared to the general population [11,12]. Studies in mice demonstrate that FXI deficiency can prevent perinatal death caused by absence of the key coagulation regulatory protein, protein C, allowing protein-C-deficient mice to reach adulthood .
Ferric chloride (FeCl3)-induced thrombus formation is a widely used experimental model of both arterial and venous thrombosis . In this model, a blood vessel is briefly exposed to high concentrations (10–50%) of FeCl3[15–18]. Using genetically modified mice, Rosen et al. demonstrated that complete FXI deficiency protected mice from carotid artery occlusion induced by exposure to 20% FeCl3, a concentration that uniformly caused vessel occlusion in wild-type mice. Interestingly, it was also noted that FXI deficiency was more protective than FIX or FVII deficiency in that particular model . The concentration of FeCl3 used may influence thrombus formation and response to different antithrombotic agents in this model . In the present study, we refined the FeCl3 injury model by determining the threshold concentration of FeCl3 required for inducing formation of an occlusive thrombus in the carotid arteries of normal mice, and used this information to carefully control the severity of injury to evaluate pharmacologic responses to the commonly used antithrombotic agents, aspirin and heparin. The well-defined thrombogenic stimuli were also used to study the effects of FIX and FXI deficiency on carotid artery occlusion, and results were compared to the protective effects of aspirin and heparin. In addition, a tail-bleeding-time assay was used to evaluate the effects of these treatments and deficiencies on hemostasis.
Materials and methods
All genotypes studied have been backcrossed to the C57BL/6 strain through at least seven generations. Male and female mice weighing 18–25 g were used. Mice homozygous for a null FXI allele (FXI–/–) have been described . These animals do not exhibit obvious signs of a coagulopathy, and breeding pairs of homozygous null animals are fertile and produce normal-sized litters. FIX-deficient mice were a generous gift from Dr Darrell Stafford (University of North Carolina, Chapel Hill, NC, USA) . For propagation of FIX null mice, FIX–/– females were crossed with FIX– males. There is little maternal mortality, and few pups were lost to bleeding complications. Wild-type C57BL/6 mice, used to establish the thrombosis model and as controls for all experiments, were purchased from Charles River Laboratories (Wilmington, MA, USA). All animals were housed in microisolator cages on a constant 12-h light/12-h dark cycle with controlled temperature and humidity and were given access to food and water ad libitum. Mice were housed and cared for in accordance with the Guide for the Care and Use of Laboratory Animals [DHEW (DHHS) Publication No. (NIH) 85–23, revised 1996, Office of Science and Health Reports, DRR/NIH, Bethesda, MD 20205]. The Institutional Animal Care and Use Committees of Bristol-Myers Squibb Company and/or Vanderbilt University approved all the procedures using mice.
FIX and FXI genotypes were confirmed by polymerase chain reaction using DNA extracted from tail clip or ear punch samples. For FXI, the oligonucleotide 5′-TTGCAGCAAAGATGAGTACGTGAAC, located at the 5′ end of exon 5, served as a common primer for the wild-type and null alleles. The 3′ primers for the wild-type and null alleles were 5′-ATGGTCGACACTGGGAAAATACCC (located at the 3′ end of exon 5), and 5′-ATTCGCAGCGCATCGCCTTCTATC (neomycin selection marker of knockout cassette), respectively. For the FIX wild-type allele, the 5′-and 3′ primers were 5′-AGCTATGGAATTACATGAAC (intron 1) and 5′-AACAGGGATAGTAAGATTGTTCC (intron 3). For the FIX null allele, the 5′ and 3′ primers were 5′-GCAGCCAACCATTGGGCTGA (promoter region) and 5′-CCAGTGAATCTTTGTCAGCAG (hypoxanthine phosphoribosyl transferase (HPRT) portion of knockout cassette), respectively.
Mice were anesthetized with pentobarbital (50 mg kg−1 administered by intraperitoneal injection). An incision was made with a scalpel directly over the right common carotid artery, and a segment of the artery was exposed using blunt dissection. Using this technique, blood loss in FIX- and FXI-deficient animals was similar (minimal) to that of wild-type mice. A miniature Doppler flow probe (Model 0.5 VB, Transonic System Inc., Ithaca, NY, USA) was attached to the carotid artery to monitor blood flow as previously described [20,23]. Thrombus formation was induced by applying two pieces of filter paper (Gel Blot Paper, GB003, Schleicher & Schuell, Keene, NH, USA) (1 × 1.5 mm) saturated with FeCl3 solutions ranging in concentration from 2.5 to 10%. The pieces of filter paper were placed on opposite sides of the carotid artery in contact with the adventitial surface of the vessel. After 3 min exposure, the filter paper was removed and the vessel was washed with sterile normal saline. Carotid blood flow was continuously monitored for 30 min after FeCl3 application, and flow rates were recorded at selected time-points (0, 5, 10, 20 and 30 min postexposure) to provide convenient data-processing without losing key information. Mice were killed by cervical dislocation immediately after conclusion of the experiment when the animals were still under deep anesthesia.
Tail-bleeding time was measured by a modification of a previously described technique . Mice were anesthetized as described above and placed on a 37 °C heating pad. The tail was transected with a sterile scalpel at a point where the tail diameter was approximately 1 mm wide (2–4 mm from the tip). After transection, the tail was immediately placed in a 50-mL falcon tube filled with 0.9% NaCl warmed to 37 °C, and the time it took for bleeding to stop was recorded. No animal was allowed to bleed for more than 30 min. The total bleeding time was illustrated, including those mice in which rebleeding occurred within 30 min. Mice were killed by cervical dislocation immediately after conclusion of the experiment and prior to recovery from anesthesia.
Aspirin (acetyl salicylic acid) was purchased from Sigma (St Louis, MO, USA) and dissolved in water (1, 3, or 10 mg mL−1). Aspirin was administered by intraperitoneal injection at either 10, 30, or 100 mg kg−1, 2 h prior to performing the FeCl3 injury or tail-bleeding-time assays. Sodium heparin (unfractionated, American Pharmaceutical Partners, Inc., Los Angeles, CA, USA) was administered by intravenous injection at doses of 50, 200, or 1000 units kg−1, 1 min prior to performing the FeCl3 injury or tail-bleeding-time assays.
Platelet aggregation assay
Mouse blood samples were collected by inferior vena caval puncture into a one-tenth volume of 3.8% sodium citrate 30 min after treatment with aspirin. Samples from two mice were pooled to provide sufficient volume for testing. Samples underwent centrifugation for 10 min at 150 × g to prepare platelet-rich plasma (PRP). After removal of PRP, the remaining blood was centrifuged for 15 min at 1500 × g, and the platelet-poor plasma (PPP) was removed into a separate tube. Platelet aggregation assays were performed on 250-μL samples of PRP on an Optical Aggregometer (Model 490, Chrono-Log, Havertown, PA, USA). PPP was used as a blank. Aggregation was induced by addition of arachidonic acid (Bio/Data Corp. Horsham, PA, USA) to a final concentration of 1.5 mmol L−1. Aggregation was monitored for 5 min, and the area under the aggregation curve was calculated. Normal saline was used as an agonist control for platelet aggregation.
Activated partial thromboplastin time (APTT) assay
Following intravenous heparin or saline infusion, blood samples were collected by inferior vena caval puncture into a one-tenth volume of 3.8% sodium citrate, followed by centrifugation at 10 000 g for 3 min in a Sorvall RMC14 centrifuge to generate PPP. The APTTs were determined at 37 °C on an Amelung KC4A microcoagulation analyzer (Sigma) using Dade Actin factor sensitive lupus (FSL) activated PTT reagent (Dade Behring, Deerfield, IL, USA).
Data in the text and figure legends are means ± standard errors for the indicated number (n) of animals. Statistical comparisons were made by analysis of variance (anova; Fisher's protected least squares difference) and differences were considered to be significant at P < 0.05.
Effect of ferric chloride concentration on carotid artery blood flow
Concentration-dependent effects of FeCl3 (2.5–10%) on blood flow in the carotid artery were established in wild-type C57BL/6 mice (Fig. 1). Thrombus formation was indicated by reductions in Doppler blood flow. No effect on blood flow was observed following 2.5% FeCl3 treatment, while five of eight animals became completely occluded within 30 min when 3% FeCl3 was used. Complete occlusion of the carotid artery was reproducibly achieved at FeCl3 concentrations of 3.5% or greater, within 10 min of exposure. Thus, a threshold thrombotic stimulus was established at 3.5% FeCl3 in wild-type C57BL/6 mice.
Mice homozygous for the FXI null allele (FXI–/–) or FIX null allele [FIX–/– for females, hemizygous (FIX–) for males] were uniformly resistant to carotid artery occlusion at FeCl3 concentrations of 3.5% (data not shown) and 5% (Fig. 2A). Complete carotid artery occlusion occurred in all wild-type C57BL/6 animal tested (n = 7) in response to 5% FeCl3 within 10 min, while FXI–/– (n = 6) and FIX–/– (n = 5) experienced no appreciable changes in blood flow (Fig. 2A). At 7.5% FeCl3 vascular patency was retained in four of eight FXI–/– mice and in four of six FIX–/– mice, while complete occlusion occurred in all animals at 10% FeCl3. The areas under the curve of the relative Doppler blood flow studies over 30 min (reflecting continuous monitoring from each animal) showed a significant difference between the null mice and wild-type animals with 5 and 7.5% FeCl3, but not 10% FeCl3 (Fig. 2B). There was no statistical difference in antithrombotic efficacy between FIX and FXI deficiency at 7.5% FeCl3 (P = 0.27). These data suggest that FIX or FXI deficiency provides similar protection in this model, but that this protection can be overcome at a sufficiently high FeCl3 concentration.
Effects of aspirin and heparin on carotid artery blood flow in wild-type C57BL/6 mice
To put the effects of FIX or FXI deficiency on carotid artery blood flow in the FeCl3 injury model into perspective, we studied the effects of two commonly used antithrombotic agents on the FeCl3 injury model. Aspirin inhibits platelet function by interfering with thromboxane production through irreversible inhibition of cyclo-oxygenase, and is widely used for the prevention and treatment of cardiovascular disease in humans. When murine carotid arteries were exposed to 3.5% FeCl3, aspirin at 10, 30 and 100 mg kg−1 demonstrated modest antithrombotic effects (P = 0.07 compared to vehicle; Fig. 3A,B, upper panels). Almost all the arteries were completely occluded after 30 min, except for one animal in the 30 mg kg−1 and two animals in the 100 mg kg−1 aspirin-treated group. Aspirin at 100 mg kg−1 was not effective at preventing arterial occlusion when arteries were exposed to 5% FeCl3 (Fig. 3A,B, lower panels). Ex vivo platelet aggregation studies demonstrated that the 10 mg kg−1 aspirin dose completely inhibited arachidonic acid-induced platelet aggregation in mice (data not shown). It is possible that a lower dose of ASA might be able to induce maximal inhibition of arachidonic acid-induced platelet aggregation in mice.
The anticoagulant heparin facilitates the inhibition of coagulation proteases by the serine protease inhibitor antithrombin, and is frequently used to treat patients with impending arterial occlusions (for example, unstable angina). At 50 units kg−1, heparin was able to prevent vascular occlusion in three of seven animals treated with 3.5% FeCl3 (Fig. 4A,B, upper panels). Heparin at 200 units kg−1 maintained normal carotid blood flow in the 3.5% FeCl3 model. However, this dose of heparin was able to prevent occlusion in only four of eight animals when the FeCl3 concentration was increased to 5%. A very large dose of heparin (1000 units kg−1) maintained normal blood flow and vascular patency when 5% FeCl3 was used (Fig. 4A,B, lower panels). Ex vivo analysis of PPP (Fig. 5) showed that heparin at 50 units kg−1 significantly elevated the APTT (12-fold increase over controls), while heparin at 200 units kg−1 resulted in APTT values that exceeded the maximum recordable value (500 s). The data suggest that FIX and FXI deficiency are considerably more effective than aspirin, and at least comparable to high doses of heparin, in their ability to protect mice from FeCl3-induced carotid artery occlusion.
The tail-bleeding-time was used to assess hemostasis. As shown in Fig. 6, mean tail-bleeding-times of untreated mice were 265 ± 68 s for wild-type animals, 287 ± 92 s for FXI-deficient mice (a 1.1-fold increase over wild-type, n = 19), and 1561 ± 125 s for FIX-deficient mice (a 5.9-fold increase over wild-type, n = 10. P < 0.01 compared to wild-type). Aspirin at 10 and 30 mg kg−1 increased the bleeding times 1.5- and 2.2-fold, respectively, in wild-type C57BL/6 mice, and a comparable effect was noted in FXI-deficient mice (1.5- and 2.6-fold increases, respectively, for 10 and 30 mg kg−1 aspirin treatment over vehicle). While 50 units kg−1 heparin administration in wild-type C57BL/6 mice only produced a mild increase in bleeding time (2.6-fold), 200 units kg−1 resulted in markedly prolonged bleeding (100% of animals bleeding off scale). Bleeding in FIX-deficient animals was severe even in the absence of aspirin. Not unexpectedly, aspirin increased bleeding in FIX-deficient mice, however, the limit we placed on the bleeding time (30 min) made it difficult to compare the effect in this setting to the other genotypes.
Using a refined FeCl3 injury model based on titration of the concentration of FeCl3 required to induce arterial occlusion we have demonstrated that the antithrombotic efficacy of FXI deficiency and FIX deficiency are comparable in mice. Our results are partially in agreement with a recent communication by Rosen et al. who demonstrated the importance of FXI to thrombus formation following carotid artery injury with 20% FeCl3. However, in contrast to the earlier report, it appears likely that the protective effect of FXI deficiency is related to a lack of FXIa-mediated activation of FIX, as FIX and FXI deficiency were indistinguishable in our model. The antithrombotic properties of low FIX activity were also noted in a rat model of FeCl3-induced arterial thrombosis  and a guinea-pig model of recurrent arterial thrombosis . In addition, our study showed a limited level of protection in FIX and FXI deficiency because it can be overcome by increasing FeCl3 concentration. The protective effects of FXI or FIX deficiency were comparable to a very high dose of heparin (1000 units kg−1) as demonstrated in the current study, and as previously reported for high-dose clopidogrel (100 mg kg−1) , and substantially more effective than aspirin (30 mg kg−1) in preventing arterial thrombosis.
A direct correlation between the concentration of FeCl3 applied to the arterial surface and the rate of thrombus formation has been previously demonstrated in rats by monitoring time to vascular occlusion  and in mice by monitoring changes in blood flow . Our technique is somewhat different in that it is based on establishing the threshold concentration of 3.5% FeCl3 required to reproducibly induce rapid vessel occlusion in wild-type mice. It should be noted that there are differences in the results obtained when anesthesia is induced with pentobarbital (current report) and isoflurane gas anesthesia (threshold at 2.5% FeCl3) . Thus, comparisons of data generated in different laboratories should be made with caution, particularly considering the wide range of FeCl3 concentrations (10–50%) used [14–18,23]. Primarily for this reason, we felt that it was important to directly compare the effects of FXI and FIX deficiency to the widely used antithrombotic agents aspirin and heparin in the same thrombosis model.
Both FIX and FXI deficiency provided protection from FeCl3-induced carotid artery obstruction that could not be completely overcome until the FeCl3 concentration reached 10%. While it is difficult to draw concrete conclusions from the comparison of the antithrombotic effects of a single dose of drug and a congenital coagulation factor deficiency state, the data indicate that the effects of either FIX or FXI deficiency are potent in this model. FeCl3 injury induces arterial thrombus formation in a manner that is dependent on platelets [14,17,18], von Willebrand factor , and thrombin generation [14,16]; and inhibitors of both platelet function and coagulation provide some degree of protection from occlusion in this model [14,16]. Perhaps fibrin formation through FVIIa/TF or platelet activation is enhanced at the higher FeCl3 concentration, over-riding the anticoagulant effect of the FXI or FIX deficiency.
During severe hemostatic challenges FIXa generated by FVIIa/TF, may be inadequate to achieve or maintain hemostasis and additional FIXa must then be generated through FXIa. Sustaining thrombin generation through FXIa may be critical for activities in addition to fibrin formation. Trace amounts of FXIa have been shown to induce clot resistance to fibrinolysis . This effect appears to be mediated, at least in part, by a thrombin-activated metalloproteinase, thrombin activatable fibrinolysis inhibitor (TAFI), which down-regulates fibrinolysis through proteolytic modification of fibrin [28–30]. Minnema and co-workers demonstrated an anti-fibrinolytic effect of FXIa by showing that inhibition of FXI with antibodies renders clots more susceptible to fibrinolytic therapy in a rabbit jugular vein thrombus model . The proposed role for FXI in thrombin generation and inhibition of fibrinolysis implies that dysregulation of FXI activation or FXIa activity may increase the risk of thrombosis. Data from the Leyden Thrombophilia Study indicate that the 10% of persons with the highest FXI levels have an ∼2–2.5-fold increased risk of venous thromboembolism compared to the general population . A high FXI level was an independent risk factor for thrombus formation in this study. A similar increase in risk has been reported for individuals with the highest FIX levels . Activated protein C (PC) inhibits coagulation by degradation of FVa and FVIIIa . Mice lacking PC die around the time of birth from a consumptive coagulopathy similar to the syndrome of purpura fulminans that occurs in human neonates lacking PC . Superimposing FXI deficiency on PC deficiency (PC–/–/FXI–/–) allows mice to survive the neonatal period and reach adulthood, demonstrating that the absence of FXI alters the phenotype of a lethal thrombotic disorder . These studies demonstrate an important role for FXI in pathologic coagulation, and suggest that therapeutic inhibition of FXIa may be a reasonable approach to the treatment or prevention of certain types of thromboembolic problems. Consistent with this notion, Gruber and Hanson demonstrated that neutralizing antibody against human FXI markedly reduced thrombus growth in a baboon arteriovenous shunt model .
The similar effects of FIX and FXI deficiency in the carotid artery injury model stand in stark contrast to the differences in severity of bleeding caused by the two deficiency states. Published data indicate that wild-type and FXI-deficient mice have similar tail-bleeding-times , while FIX deficiency is associated with marked prolongation of bleeding that can be fatal . The results of the tail-bleeding-time assay in mice mirror, in many respects, the bleeding problems associated with FIX and FXI deficiency in humans. Surgical procedures in humans with severe FIX deficiency (hemophilia B) are expected to be associated with significant bleeding problems in the absence of factor replacement. However, it should be recognized that in pre-clinical animal models, therapeutic inhibition of FIX was efficacious in preventing thrombus formation without producing significant bleeding [24,25,35]. Bleeding with surgery in FXI deficiency, on the other hand, is far less predictable and tends to be most pronounced when involving tissues with high fibrinolytic activity (urinary tract and oropharynx), but not necessarily other tissues [10,36]. Furthermore, in sharp contrast to patients with severe hemophilia (FVIII or FIX deficiency), it is rare for a FXI-deficient human to experience significant bleeding in the absence of trauma or surgery (so called ‘spontaneous’ bleeding) [10,36]. Given these observations, therapeutic inhibition of FXI/FXIa would possibly not be associated with the significant risks of bleeding that accompany therapy with currently used anticoagulants. Interestingly, administration of aspirin to FXI-deficient mice did not prolong bleeding in the tail-bleeding-time assay compared to wild-type mice (Fig. 6), suggesting that therapeutic inhibition of FXI/FXIa in combination with aspirin may be a safe combination therapy for treating thrombotic diseases.
Xinkang Wang – principle investigator for overall design of the work. He also performed some of the in vivo experiments on FXI and FIX knockout (KO) mice; Qiufang Cheng – designed breeding strategies for FXI and FIX mice including genotyping, and participated in in vivo studies using the KO mice; Lin Xu – established the FeCl3-induced arterial thrombosis model in mice and performed some of in vivo studies on FXI and FIX KO mice; Giora Feuerstein – involved in initial animal model designing and strategy for the work using FXI and FIX KO mice; critically reviewed/edited the entire manuscript; Mei-Yin Hsu – designed and performed the ex vivo studies on aspirin and heparin; Patricia Smith – re-established the FeCl3-induced thrombosis and bleeding time models at the Bristol-Myers Squib site in NJ; and performed the in vivo experiments on aspirin and heparin; Dietmar Seiffert – strategically planned FXI and FIX KO antithrombotic efficacy studies in comparison with ASA and heparin to highlight the levels of efficacy; critically reviewed/edited the manuscript; William Schumacher – designed the aspirin and heparin doses for in vivo work; data processing/analysis of area under curve for Doppler blood flow to quantify and compare the animal models following drug and/or KOs; critically reviewed/edited the manuscript; Martin Ogletree – designed the bleeding study of ASA on top of FXI or FIX KO, and analyzed the bleeding profile along with antithrombotic efficacy to highlight the therapeutic potential of FXI as a target; critically reviewed/edited the manuscript; David Gailani – co-principle investigator for overall experimental design; wrote part of the manuscript and critically reviewed/edited the entire manuscript.