Factor XII Silencing Using siRNA Prevents Thrombus Formation in a Rat Model of Extracorporeal Life Support

Heparin anticoagulation increases the bleeding risk during extracorporeal life support (ECLS). This study determined whether factor XII (FXII) silencing using short interfering RNA (siRNA) can provide ECLS circuit anticoagulation without bleeding. Adult male, Sprague-Dawley rats were randomized to four groups (n = 3 each) based on anticoagulant: (1) no anticoagulant, (2) heparin, (3) FXII siRNA, or (4) nontargeting siRNA. Heparin was administered intravenously before and during ECLS. FXII or nontargeting siRNA were administered intravenously 3 days before the initiation of ECLS via lipidoid nanoparticles. The rats were placed on pumped, arteriovenous ECLS for 8 hours or until the blood flow resistance reached three times its baseline resistance. Without anticoagulant, mock-oxygenator resistance tripled within 7 ± 2 minutes. The resistance in the FXII siRNA group did not increase for 8 hours. There were no significant differences in resistance or mock-oxygenator thrombus volume between the FXII siRNA and the heparin groups. However, the bleeding time in the FXII siRNA group (3.4 ± 0.6 minutes) was significantly shorter than that in the heparin group (5.5 ± 0.5 minutes, p < 0.05). FXII silencing using siRNA provided simpler anticoagulation of ECLS circuits with reduced bleeding time as compared to heparin. http://links.lww.com/ASAIO/A937

therapy for patients with cardiac and/or respiratory failure. 1 Unfortunately, ECLS is plagued by a high rate of both bleeding and thrombotic complications. [1][2][3] Currently, unfractionated heparin is the most widely used anticoagulant. 4 Although its use can delay thrombotic complications, such as oxygenator failure, it also increases the risk of intracerebral, pulmonary, and surgical sites bleeding. 1,5,6 Therefore, major improvements in anticoagulant are needed to reduce these complications and the need for careful coagulation monitoring.
Thrombus formation during ECLS is initiated primarily by three synergistic processes: (1) activation of factor XII (FXII) and the intrinsic coagulation cascade, (2) platelet binding and activation to fibrinogen that is adsorbed to the artificial surfaces of the circuit, and (3) shear-induced activation of platelets caused by high-resistance circuit components and the pump. 7 Each of these modes of activation accelerates activation of the common coagulation cascade and, ultimately, fibrin formation. Heparin anticoagulation inhibits fibrin formation by inhibiting thrombin and activated factor X of the common coagulation cascade. Direct inhibition of FXII could serve a similar role, with the advantage of not inhibiting tissue hemostasis. FXII has little to no role in normal hemostasis, and thus, people who do not produce FXII live normal lives without excessive bleeding. 8,9 This makes FXII or activated FXII (FXIIa) inhibition an attractive, safe alternative to heparin during ECLS.
To date, several FXII and FXIIa inhibitors have been developed including (1) natural inhibitors, (2) small-molecule inhibitors, (3) monoclonal antibodies, (4) antisense oligonucleotides (ASO), and (5) aptamers. [10][11][12][13][14][15][16][17][18] Among these, the recombinant fully human FXIIa-blocking antibody 3F7 and small-molecule inhibitor FXII900 have shown that FXIIa inhibition can significantly reduce thrombus formation without bleeding complications during ECLS. 11,17,18 An alternative, long-acting means is to eliminate hepatic FXII synthesis for days to weeks with a single administration using oligonucleotide-based drugs, including ASO and short interfering RNA (siRNA). In one study, a FXIIspecific ASO attenuated catheter-induced thrombosis up to 35 days in rabbits. 19 However, this ASO has a delayed onset of action (4 weeks) that renders it unusable for ECLS. In contrast, a single siRNA dose can reduce plasma protein concentrations within one day. siRNA is a synthetic RNA duplex consisting of two unmodified annealed 21-23-mer oligonucleotides. Once siRNA enters the cytosol, it binds to the RNA-induced silencing complex and degrades target messenger RNA (mRNA), preventing mRNA translation into a specific protein. 20 This study determined whether FXII silencing using siRNA can provide sufficient anticoagulation for ECLS, maintain normal tissue coagulation, and reduce the need for careful anticoagulation monitoring.

Methods
This study was approved by the Allegheny Health Network Institutional Animal Care and Use Committee (Project No. 1071). Adult male Sprague-Dawley rats (Taconic Biosciences, Germantown, NY) were used and received humane care in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. 21 FXII siRNA was purchased from Sigma-Aldrich (SASI_ Rn01_00106775_s, St. Louis, MO). MISSION siRNA Universal Negative Control #1 (SIC001, Sigma-Aldrich) was used as an untargeted siRNA control. Lipidoid nanoparticles (LNPs) (Method 1, Supplemental Digital Content 1, http://links.lww. com/ASAIO/A933) were used as the delivery vehicle for siRNA, as they have been previously identified as an effective and safe siRNA delivery material. [22][23][24][25] Rats were randomized to four groups (n = 3 each) based on the means of anticoagulant: (1) no anticoagulant, (2) heparin, (3) FXII siRNA, or (4) nontargeting siRNA. In group one, no anticoagulant was administered. In group two, intravenous 50 IU/kg heparin was administered to achieve an activated clotting time (ACT) of 180-250 seconds before initiating ECLS. Afterward, heparin was administered continuously at 50 IU/ kg/h for 1 hour and then at 25 IU/kg/h for 7 hours. 26 In groups three and four, 3 mg/kg of FXII or nontargeting siRNA was administered 3 days before ECLS. Before siRNA administration, a blood sample was taken to evaluate FXII concentration (0.3 ml), ACT (15 µl), and activated partial thromboplastin time (aPTT) (0.4 ml). The dose of 3 mg/kg and timing of ECLS were determined based on a preliminary (n = 2) kinetic study evaluating the effect of siRNA dose and time on FXII concentration, ACT, and aPTT ( Figures 1 and 2, Supplemental Digital Content 2, http://links.lww.com/ASAIO/A934). For the presentation and statistical analysis, FXII concentration was normalized to the percentage of the baseline FXII concentration before siRNA delivery.
The rats were placed on pumped arteriovenous ECLS by cannulating the left carotid artery for access and the right jugular vein for return, with blood flow maintained at 2 ml/min. The ECLS circuit was described previously. 26 Briefly, it consists of small-bore tubing and a custom-designed mock-oxygenator (Figures 3 and 4, Supplemental Digital Content 3, http://links. lww.com/ASAIO/A935) and has a 2.5-ml priming volume. During ECLS, mean arterial pressure (MAP), heart rate (HR), and peripheral oxygen saturation (SpO 2 ) were monitored continuously by using the BIOPAC MP150A-CE data acquisition system (BIOPAC Systems, Inc., Santa Barbara, CA) and a Heska VetOx Plus 4800 Vital Signs Monitor (Heska, Loveland, CO). MAP was maintained over 60 mm Hg by administering norepinephrine (0-0.5 µg/kg/min). The targeted HR was 250-350 beats/min, and targeted SpO 2 was > 94%. Any rats not meeting these criteria were euthanized. In total, two rats in the FXII siRNA group were euthanized because of air embolism during right jugular vein cannulation. The mock-oxygenator inlet and outlet pressures were monitored continuously using the BIOPAC MP150A-CE data acquisition system and recorded at 0.05 (baseline), 0.08, 0.16, 0.5, 1, 2, 3, 4, 5, 6, 7, and 8 hours after flow initiation. Blood flow resistance was calculated in standard fashion as (inlet pressure − outlet pressure)/(blood flow).
The bleeding time was also measured before and at the end of ECLS, as previously described. 26 Blood samples were also taken after cannulation for ECLS and at the conclusion of ECLS to measure ACT (15 µL), aPTT (0.4 ml), and platelet count (Plt) (0.3 ml). Arterial blood gases were not measured due to restrictions on blood sampling volumes. For presentation, the Plt was corrected for hemodilution during circuit priming using the formula corrected Plt = raw Plt × (hematocrit before ECLS)/ (hematocrit at the end of ECLS).
The experiment was continued for 8 hours or until the mock-oxygenator blood flow resistance reached three times its baseline resistance and was considered failed. The circuit was disconnected and gently flushed with heparinized saline (2 IU/ml), and the rats were euthanized using pentobarbital (175 mg/kg). The saline in the mock-oxygenator was flushed using air for 10 seconds before and after the experiment, and the amount of flushed volume was measured by weight (g) and converted to volume using the density of saline (1.0 mg/ml). Then the difference between the pre-and postexperimental volume was determined to be the clot volume. From the clot volume, the percent of the mock-oxygenator filled with the clot was calculated.

Statistical Analysis
All data are presented as mean ± standard error. A paired t-test was used to examine the effects of the siRNA and LNPs delivery vehicle on the FXII level, ACT, and aPTT. A one-way analysis of variance (ANOVA) was used to examine the effect of different anticoagulation groups on ECLS thrombus volume. A linear mixed model was used to assess differences in ACT, aPTT, blood flow resistance, and bleeding time during ECLS. The anticoagulant type (FXII siRNA, heparin, etc.), time, and the interaction of anticoagulant and time were used as fixed effects, and the animal ID was used as a random effect. For posthoc correction, Bonferroni correction was used for all pairwise comparisons. A one-way ANOVA was used to compare the Plt before ECLS and loss in platelets during ECLS between groups. The Kaplan-Meier method was performed to compare mockoxygenator failure rates using log-rank as the statistical test for significant differences in failure rate. All statistical analyses were performed using SPSS software (version 27.0; SPSS Inc., Chicago, IL). All statistical tests were two-sided, and a p value < 0.05 was considered statistically significant.

Results
The mean weight of the rats was 549 ± 5 g in the no anticoagulant group, 516 ± 25 g in the heparin group, 586 ± 24 g in the siRNA group, and 538 ± 6 g in the nontargeting siRNA group. The differences in weight were not statistically significant (p = 0.11).
In contrast, 3 mg/kg of nontargeting siRNA had no effect on plasma FXII concentration (102% ± 3% after treatment, p = 0.48), indicating that the reduction of FXII using FXII siRNA was specifically due to the siRNA and not the LNPs delivery vehicle ( Figure 1A). The nontargeting siRNA also had no effect on aPTT (20.5 ± 0.6 seconds before treatment and 20.7 ± 2.7 seconds after treatment, p = 0.91) ( Figure 1B). However, the nontargeting siRNA unexpectedly caused a statistically significant ACT prolongation from 75 ± 4 seconds before treatment to 123 ± 2 seconds 3 days after treatment (p < 0.05, Figure 1C). Therefore, the combination of no effect on FXII concentration or aPTT with an increase in the ACT suggests an effect on the coagulation system outside the coagulation cascade factors.
The aPTT during ECLS for each anticoagulant group is shown in Figure 2A. The aPTT before and at the end of ECLS in the FXII siRNA group was 93.6 ± 10.0 and 102.3 ± 8.9 seconds, respectively, which were statistically prolonged when compared to the no anticoagulation (22.2 ± 1.8 seconds before and 26.5 ± 3.5 seconds at the end of ECLS, p < 0.05) and the nontargeting siRNA groups (20.7 ± 2.7 seconds before and 23.3 ± 2.9 seconds at the end of ECLS, p < 0.05) (Figure 2A). Similarly, the use of heparin created a statistically significant increase in aPTT (135.2 ± 31.3 seconds before and 46.9 ± 5.2 seconds at the end of ECLS) compared to no anticoagulation (p < 0.05) and nontargeting siRNA (p < 0.05). There was no significant difference in the aPTT between the FXII siRNA and the heparin groups (p > 0.99). The nontargeting siRNA group showed no significant difference in aPTT when compared to the no anticoagulant group (p > 0.99).
The ACT for each anticoagulant group is shown in Figure 2B. The ACT for the heparin group was significantly prolonged (202 ± 8 seconds before and 170 ± 5 seconds at the end of ECLS) when compared to the FXII siRNA (139 ± 2 seconds before and 137 ± 11 seconds at the end of ECLS, p < 0.05), the nontargeting siRNA (123 ± 2 seconds before and 126 ± 3 seconds at the end of ECLS, p < 0.05), and the no anticoagulant groups (104 ± 9 seconds before and 120 ± 3 seconds at the end of ECLS, p < 0.05). The ACT before and at the end of ECLS in the FXII siRNA group was also prolonged compared to that in the nontargeting siRNA and the no anticoagulant groups, but without statistical significance (p > 0.99 and p = 0.14, respectively). Figure 3A shows mock-oxygenator blood flow resistance throughout the experiments. The no anticoagulant group exhibited a significantly greater mock-oxygenator blood flow resistance than the FXII siRNA (p < 0.05) and the heparin groups (p < 0.05). For all rats not receiving an anticoagulant, mock-oxygenator resistance rapidly increased to greater than three times the baseline resistance before 10 minutes had elapsed. However, the blood flow resistance in the FXII siRNA group did not change significantly during the experiment (p = 0.94), and no devices reached three times the baseline resistance. No heparin group mock-oxygenators reached three times the baseline blood flow resistance, and there was no significant difference between the FXII siRNA and the heparin groups' resistance (p > 0.99). Unexpectedly, the resistance of all devices in the nontargeting siRNA group did not increase as rapidly as in the no anticoagulant group. The mock-oxygenator failure curve exhibiting significant differences (p < 0.05) between the four groups is shown in Figure 3B. The mean time from initiation of ECLS until the oxygenator failure in the no anticoagulant and the nontargeting siRNA groups was 7 ± 2 and 113 ± 27 minutes, respectively. Figure 3C shows thrombus volume in the mock-oxygenator after circuit detachment. The thrombus volume in the FXII siRNA, heparin, no anticoagulant, and nontargeting siRNA groups were 5% ± 5%, 0% ± 2%, 49% ± 4%, and 15% ± 2%, respectively (p < 0.05). The thrombus volume in the FXII siRNA group was significantly lower than that in the no anticoagulant group (p < 0.05). There was no significant difference in the thrombus volume between the FXII siRNA and the heparin groups (p > 0.99). The thrombus volume in the nontargeting siRNA group was also significantly lower than that in the no anticoagulant group (p < 0.05), mirroring the unexpected ACT and blood flow resistance results. Last, the change in platelet counts during ECLS is shown in Figure 5, Supplemental Digital Content 4, http://links.lww.com/ASAIO/ A936. The bleeding time for each anticoagulant group is shown in Figure 4. The bleeding time for the heparin group was significantly prolonged (6.5 ± 0.3 minutes before and 5.5 ± 0.5 minutes at the end of ECLS) compared to the FXII siRNA (2.8 ± 0.4 minutes before and 3.4 ± 0.6 minutes at the end of ECLS), the nontargeting siRNA (1.7 ± 0.4 minutes before and 2.0 ± 0.3 minutes at the end of ECLS), and the no anticoagulant groups (0.7 ± 0.2 minutes before and 0.8 ± 0.2 minutes at the end of ECLS) (p < 0.05). The bleeding time before and at the end of ECLS in the FXII siRNA group was significantly prolonged compared to the no anticoagulant group (p < 0.05). There was no significant difference between the nontargeting siRNA and the no anticoagulant groups (p = 0.45) and between the FXII siRNA and nontargeting siRNA groups (p = 0.18).

Discussion
This study determined that FXII siRNA provided sufficient anticoagulation for at least 8 hours of ECLS. A single intravenous dose of siRNA (3 mg/kg) administered 3 days before the initiation of ECLS reduced plasma FXII concentration to 26% of the normal level, and this prevented thrombus formation within the mock-oxygenator over 8 hours, similarly to heparin anticoagulation. At the same time, this had a lesser effect on prolonging tissue bleeding when compared to heparin. It should also be noted that direct FXII inhibitors have no effect on tissue bleeding times. 11,17,18 Therefore, further work should be performed to optimize the siRNA sequence and dosing to eliminate any effect on tissue bleeding. Once optimized, replacing heparin with FXII silencing could reduce bleeding complications that occur in approximately 15% to 30% of ECLS cases. 1,5,6,27,28 Moreover, the ability of FXII siRNA to provide anticoagulation for multiple days would greatly simplify anticoagulant management. During clinical ECLS management, heparin is given continuously and titrated to balance the risk of thrombotic and bleeding complications. This requires regular measurement of clotting times and subsequent adjustment of heparin infusion rates. In the current study, a single intravenous dose of siRNA provided a significant effect that lasted for 3  days and did not require management or titration during ECLS. Clinically, siRNA could be repeated on some regular schedule with little or no required measurement of clotting times. The combination of sufficient ECLS anticoagulation and a reduction in bleeding complications will make ECLS safer to employ, and the simplicity in once every 3-4 days dosing would simplify patient care and reduce ECLS costs.
Previous studies of shorter-acting FXIIa inhibitors have provided similar results. Wilbs et al. 17 demonstrated that FXII900, a bicyclic peptide inhibitor of FXIIa, reduced thrombus formation and blood flow resistance in oxygenators over 4 hours in the rabbit veno-venous ECLS model when compared to no anticoagulation. In that study, the oxygenator thrombus volume was significantly lower (p < 0.05) for the FXII900-treated rabbits (10%) than for the untreated rabbits (37%). 17 Similarly, Larsson et al. found that the recombinant human FXIIa-blocking antibody (3F7) significantly reduced thrombus formation and increases in oxygenator pressure gradient in a rabbit veno-arterial ECLS model during 6-hour testing when compared to no anticoagulant use. 11 Of similar importance, FXII900 and 3F7 preserved normal tissue bleeding times, unlike heparin. 11,17,18 In contrast, FXII silencing also provides several days of anticoagulation with a single dose. The FXII siRNA used in this study provided at least 3 days of effective silencing in rats at a dose of 3 mg/kg. FXII900 had to be continuously administered as it was eliminated by the kidney, although it could be bound to long-chain polymers (e.g., polyethylene glycol, polycarboxybetaine) to increase its half-life to as long as a few days. In contrast, siRNA provides a long-lived reduction in FXII concentrations. In our preliminary study, siRNA showed a dosedependent reduction of plasma FXII and prolongation of aPTT on day 3, but these returned to nearly baseline values on day 7. Further optimization of the siRNA sequence and delivery vehicle could lead to more effective, longer-lasting silencing. Cai et al. observed that 0.01 mg/kg of siRNA provided an 82% reduction in FXII and 1 mg/kg of siRNA provided a 99% reduction in FXII on day 7 in rats. 29 Similarly, a single dose of GalNAc-siRNA targeting FXII (ALN-F12) provided a dosedependent maximal FXII suppression of 51-55% at 0.3 mg/kg and 93% at 1 mg/kg on day 10, lasting for 64 days. 30,31 This holds the promise for single dosing during traditional ECLS or infrequent dosing during extended ECLS.
Additionally, the current study demonstrated an increase in the bleeding time in rats treated with the FXII siRNA unlike FXIIa inhibition using the 3F7 and FXII900 and silencing via ALN-F12. 11,17,18,31 Similarly, Cai et al. also reported that rats treated with FXII siRNA delivered via cationic LNPs displayed a statistically significant increase in bleeding time. 29 Therefore, there may be direct side effects on platelet function or off-target gene silencing on other proteins in the hemostatic system when using this delivery method. [32][33][34] Ultimately, the time course and effectiveness of the FXII silencing will depend on the delivery vehicle, siRNA sequence, and potential chemical modification of the siRNA. In the current study, FXII siRNA was administered 3 days before ECLS because the half-life of FXII is 50-70 hours, and it would take at least 2 days for circulating FXII to disappear completely. 8 However, Stavrou et al. demonstrated very little difference in FXII levels between 1 and 3 days after treatment. 35 Therefore, FXII concentrations may decrease within 24 hours after treatment. Clinically, a reasonable use of FXII siRNA would be to administer FXII siRNA as soon as a decision has been made to put the patient on ECMO, and heparin or a short-acting FXII inhibitor would also be provided before cannulation to provide sufficient time for siRNA to decrease the FXII concentration. 17,18 In patients using ECMO as a bridge to lung transplantation, administration of FXII siRNA 1 day before cannulation may be feasible.
This study had several limitations. The ability to anticoagulate the ECLS circuit was evaluated in a miniaturized system with components different from a clinical ECLS system and for a period of only 8 hours in the rat model. An initial, short-term, small animal study was deemed necessary, however, to provide the proof of concept prior to embarking on significantly more expensive long-term studies in larger animals. Furthermore, we have not yet optimized the optimal delay between delivery and initiation of ECLS and fully evaluated the side effects of LNPs. The nontargeting siRNA showed mild prolonged ACT and a mild reduction in ECLS thrombus formation for approximately 2 hours. These LNPs demonstrated no toxicity in mice and rats. 22,23 However, a more detailed examination of their effects on the coagulation cascade might indicate some unknown mild effects, as seen in our study.
In future studies, the siRNA sequence, doses, and delivery vehicles should first be optimized for this application to increase efficacy and decrease off-target effects, and the time course of FXII decrease should be investigated. Thereafter, large-animal studies could be used to evaluate anticoagulation and anti-inflammatory efficacy over a typical course of ECLS for acute respiratory failure (5-7 days) or bridge to transplant (2-4 weeks) using a commercial ECLS system. FXII siRNA might also reduce inflammation by reducing contact activation and bradykinin production. 8,9 Additionally, the optimal use of anticoagulant monitoring (aPTT, etc.) should be studied, as less frequent monitoring would likely be necessary due to the longer duration of action and decreased bleeding risk. Last, the siRNA efficacy should be investigated during ECLS in patients with hypercoagulative and hyperinflammatory states.

Conclusions
FXII silencing can provide simpler, long-acting anticoagulation in ECLS circuits. Due to the reduced bleeding time and need for coagulation monitoring, FXII siRNA would enable simpler and safer ECLS.