Factor XII promotes the thromboinflammatory response in a rat model of veno-arterial extracorporeal membrane oxygenation

Abstract

Objective:

Factor XII (FXII) is a multifunctional protease capable of activating thrombotic and inflammatory pathways. FXII has been linked to thrombosis in extracorporeal membrane oxygenation (ECMO), but the role of FXII in ECMO-induced inflammatory complications has not been studied. We used novel gene-targeted FXII deficient rats to evaluate the role of FXII in ECMO-induced thromboinflammation.

Methods:

FXII deficient (FXII^−/−) Sprague-Dawley rats were generated using CRISPR/Cas9. We used a minimally invasive veno-arterial (VA) ECMO model to compare wild-type (WT) and FXII^−/− rats in 2 separate experimental cohorts – rats placed on ECMO without pharmacological anticoagulation, and rats anticoagulated with argatroban. Rats were maintained on ECMO until circuit failure or 1 hour. Comparisons were made with unchallenged rats and rats that underwent a sham surgical procedure without ECMO.

Results:

FXII^−/− rats were maintained on ECMO without pharmacological anticoagulation with low resistance throughout a 1-hour experiment. In contrast, WT rats placed on ECMO without anticoagulation developed thrombotic circuit failure within 10 minutes. Argatroban provided a means to maintain WT and FXII^−/− rats on ECMO for the 1-hour timeframe without thrombotic complications. Analyses of these rats demonstrated that ECMO resulted in an increase in neutrophil migration into the liver that was significantly blunted by FXII deficiency. ECMO also resulted in increases in high molecular weight kininogen cleavage and complement activation that were abrogated by genetic deletion of FXII.

Conclusion:

FXII initiates hemostatic system activation and key inflammatory sequelae in ECMO, suggesting that therapies targeting FXII could limit both thromboembolism and inopportune inflammatory complications in this setting.

INTRODUCTION

Extracorporeal membrane oxygenation (ECMO) represents a life-saving treatment for patients with severe refractory cardiopulmonary failure. Thromboembolic complications, including thrombi in the circuit itself and embolic strokes, as well as bleeding complications are common in pediatric and adult ECMO.¹ ECMO is also complicated by a complex, rapid inflammatory response characterized by increased levels of circulating inflammatory cytokines, complement activation and leukocyte activation within hours of ECMO initiation.² However, the mechanisms driving the pathological immune responses in ECMO remain poorly defined. Although classically viewed separately, the hemostatic and inflammatory systems are highly integrated with extensive cross talk.³ The coagulation contact protease factor XII (FXII) is particularly exciting in this context as it lies at the interface of a key nexus between the hemostatic and inflammatory systems. FXII-mediated activation of FXI ultimately results in thrombin generation. FXII proteolytic functions as well as FXII zymogen have been linked to a variety of immune functions, including neutrophil activation, complement activation, and regulation of vascular leak.^{4, 5} Moreover, FXII plays no role in surgical hemostasis,⁶ making it an attractive therapeutic target. Previous animal studies have linked FXII to ECMO-associated thrombosis,^7–9 but the role of FXII in ECMO-induced pro-inflammatory mechanisms has not been evaluated. In this study, we used novel gene-targeted FXII deficient rats (FXII^−/−) and a novel minimally invasive model of veno-arterial extracorporeal membrane oxygenation (VA-ECMO) to investigate the role of FXII in ECMO-induced thrombo-inflammation.

METHODS

Generation of gene-targeted FXII deficient rats.

The CRISPR/Cas9 system was used to target the FXII gene (f12) in Sprague Dawley rats. Founder animals were interbred with wildtype Sprague Dawley rats purchased from Charles River to generate rats heterozygous for the FXII mutation (FXII^+/−). FXII^+/− rats were interbred generating FXII^+/+ (WT), FXII^+/− and FXII^−/− animals. See Supplement for details.

Rat ECMO model.

Adult (ages 8 – 10 weeks) male WT and FXII^−/− Sprague-Dawley rats weighing ~300–600 g were used as ECMO recipients. A rat of the same genotype was used as a blood donor to prime the circuit. All animal protocols were approved by Cincinnati Children’s Hospital Research Foundation Institutional Animal Care and Use Committee.

Briefly, rats were anaesthetized with isoflurane, intubated orotracheally and connected to a ventilator. The right external jugular vein and left common carotid artery were cannulated and connected to a peristaltic pump (Harvard Peristaltic Pump P-230, Harvard Apparatus Inc.) and an oxygenator with a priming volume of 1.6 ml (MICRO-1 oxygenator-Dongguan Kewei Medical Instrument Co) (Fig. 2A). The left femoral artery was cannulated for invasive blood pressure monitoring. The oxygenator inlet and outlet pressures were monitored via pressure transducers and used to calculate resistance by dividing the inlet outlet pressure difference by the flow rate. Circuit dysfunction was considered when blood flow resistance reached three times it’s baseline resistance with impairment of gas exchange.

Figure 2. Description of rat VA-EMCO model.

A) Shown is a typical example of a rat on the VA-ECMO circuit. Note the venous line (1), peristaltic pump (2), oxygenator (3), arterial line (4), femoral arterial line used for blood pressure monitoring (5), and the pre- and post-pressure transducers (*). Also shown are summaries of the timelines for the two experimental paradigms used. The first (B) compared WT and FXII^−/− rats without anticoagulation The second (C) compared WT and FXII^−/− rats anticoagulated with argatroban. FXII^−/−-FXII deficient, WT-Wild type, CaCl₂-Calcium Chloride, ABG-Arterial blood gases.

Rats were enrolled onto ECMO in two separate experimental cohorts. The first consisted of WT and FXII^−/− rats (n = 5 per genotype) placed on ECMO with no pharmacological anticoagulation. The second group of WT and FXII^−/− rats (n = 5 per genotype) was placed on ECMO with argatroban administration. We chose argatroban (Exela Pharma Sciences, LCC, USA) as this short-acting thrombin inhibitor has no effect on FXII activity. Argatroban dosing was determined by pilot studies performed in WT rats. We choose an argatroban dosing schema that resulted in an activated clotting time of > 350 seconds prior to ECMO initiation and provided sufficient anticoagulation to maintain WT rats on ECMO for the predetermined 1 hour timeframe without complications. A continuous infusion of argatroban (1mg/mL) was initiated through a femoral vein catheter at 50 mcg/kg/min starting 30 minutes prior to ECMO initiation. This was followed by administration of bolus of argatroban (3 mg/kg) just before ECMO initiation.

Once cannulated, the ECMO recipient received either no anticoagulation or argatroban anticoagulation as described above. The circuit was primed with 7 ml of whole blood from a rat of the same genotype containing sodium citrate 3.2% (blood:citrate ratio 9:1). ECMO was initiated and flow rate increased to 60–80 mL/kg/min. Once the circuit was running for ~5 minutes, calcium chloride was administered to normalize the calcium, reversing citrate anticoagulation. Animals were euthanized at the end and harvested immediately. The sham surgery consisted of intubation, surgical cannulation, and argatroban administration for the same timeframes as the rats undergoing ECMO, with the exclusion of ECMO. See Supplement for details.

Statistics

Statistical analyses were performed using GraphPad Prism software v.9.

Refer to the Supplement for details regarding the following assays: thrombin generation assay, activated partial thromboplastin time, quantitation of FXII mRNA and protein, FXII activity, plasma fibrinogen, histological analyses, measurement of cleaved HMW kininogen, and analyses of complement activation.

RESULTS

Genetic deletion of factor XII limits thrombus formation and blood flow resistance in ECMO.

To define the role of FXII in ECMO-associated pathologies, we generated FXII-deficient Sprague Dawley rats using the CRISPR/Cas9 system that carry a premature stop codon in exon 4 the of F12 gene (Fig. 1A). Paralleling FXII-deficient mice,^{10, 11} rats homozygous for this mutation (FXII^−/−) are viable, appear healthy, and exhibit normal fecundity (data not shown). Liver-derived FXII mRNA was below the detection level of the assay in FXII^−/− rats (Fig. 1B). ELISA-based quantitation of plasma FXII protein in FXII^−/− rats was ~1% of that in FXII sufficient rats (FXII^+/+) (Fig. 1C). Whether this small amount of protein represents a truncated amino acid fragment of FXII or rare readthrough events of the premature stop codon remains to be determined. FXII^−/− rats exhibited significantly prolonged aPTTs (Fig. 1D), and plasma FXII activity that was below the lower limit of detection based on a 1-stage clotting assay (Fig. 1E). Consistent with previous studies showing that FXII plays no role in tissue factor (TF) initiated hemostasis, analyses of TF-initiated thrombin generation in plasma from FXII^−/− rats was indistinguishable from that in plasma from FXII^+/+ rats (Fig. 1F). CBCs and plasma fibrinogen levels were also similar in FXII^−/− rats and controls (Supplemental Fig. 1)

Figure 1. Generation and characterization of gene-targeted FXII deficient rats.

A) Shown are the nucleotide and amino acid sequences of the wild-type (WT) and mutant FXII allele. Note that the mutant animals carry a premature stop codon in exon 4 (*) of the f12 gene. Shown are quantitation of liver mRNA expression normalized to FXII^+/+ (WT) rats (B), and ELISA quantitation of plasma FXII protein (C). The aPTT was significantly prolonged in FXII^−/− rats (D) and FXII activity based on a one-stage clotting assay was below the lower limit of detection (E). F) Thrombin generation assays initiated by 1 pM tissue factor were similar between genotypes. (Data represents mean ± SEM.) FXII^−/− FXII deficient, FXII^+/+ FXII sufficient, FXII^+/− FXII Heterozygous mutants.

To define the role of FXII in ECMO-related thrombosis, we placed 5 FXII^−/− and 5 FXII^+/+ (hereafter referred to as WT) rats on VA-ECMO without any anticoagulation for a predetermined 1-hour timeframe. See Figure 2A&B for details of the ECMO circuit and protocol. All 5 WT rats experienced circuit failure with 10 minutes of ECMO initiation (Fig. 3A), as defined by an exponential increase in the blood flow resistance across the oxygenator (Fig. 3B). Consistent with this there was substantial visible clot burden at the oxygenator inlets. In contrast, FXII deficient rats all maintained low resistance flow without pharmacological anticoagulation (< 5mmHg/(mL/min) during the 1-hour ECMO timeframe.

Figure 3. FXII promotes thrombosis in ECMO.

A) FXII-deficiency significantly prolonged circuit life compared to rats with normal FXII expression (WT) in ECMO without any pharmacological anticoagulation. B) Circuit failure in WT rats was preceded by a rapid exponential rise in blood flow resistance across the oxygenator. C) Shown is the blood flow resistance across the oxygenator from a cohort of 5 WT and 5 FXII^−/− rats placed on ECMO with argatroban anticoagulation. Note that blood flow resistance was relatively low and stable in both cohorts, but significantly lower in the FXII^−/− rats. D) Platelet counts and fibrinogen were measured at the end of the 1 hour ECMO and sham procedures in the argatroban anticoagulated cohort. The dotted line denotes the mean obtained from analyses of unchallenged rats. Note that ECMO, but not the sham surgery, resulted in an ~50% diminution in circulating platelet count that was similar between genotypes. WT-Wild type, FXII^−/− FXII deficient.

The rapid circuit failure observed in WT rats on ECMO in the absence of pharmacological anticoagulation made meaningful comparisons between genotypes impossible. To address this, we pursued a separate experiment where 5 WT and 5 FXII^−/− rats were placed on ECMO following the same protocol with the addition of argatroban. We also performed “sham” procedures on 3 WT and 5 FXII^−/− rats that included intubation, cannulation and argatroban administration for the same timeframe as the ECMO cohort. All 10 rats placed on ECMO were maintained for the 1 hour timeframe without complications. Hemodynamic and metabolic parameters, including mean arterial pressure, heart rate, PaO₂, PaCO₂, pH and lactate remained stable throughout the 1-hour ECMO run and were similar between genotypes (Table 1). Blood flow resistance during ECMO was overall low in both genotypes anticoagulated with argatroban, but was lower in the FXII^−/− rats relative to the WT cohort (Fig. 3C). Consistent with clinical observations,¹ CBCs obtained at the end of ECMO showed an ~50% diminution in platelet count relative to that in unchallenged rats that was similar between genotypes. No diminution in platelet count was observed in the sham cohort (Fig. 3D). Neither ECMO nor the sham surgery resulted in a significant change in plasma fibrinogen relative to unchallenged rats (Fig. 3D).

Table 1. Hemodynamic and gas exchange variables.

Shown are measures of MAP (mean arterial blood pressure), HR (heart rate), PaO₂ (partial pressure of oxygen), PaCO₂ (partial pressure of carbon dioxide), pH, and lactate measured prior to (baseline) and during sham surgeries and VA-ECMO (veno-arterial extracorporeal membrane oxygenation) experiments. Note that all rats in these analyses were anticoagulated with argatroban. (Data represent mean ± SD).

WT							FXII−/−
	Sham			VA-ECMO			Sham			VA-ECMO
Time Point	Baseline	30 mins	60 mins	Baseline	30 mins	60 mins	Baseline	30 mins	60 mins	Baseline	30 mins	60 mins
MAP (mm Hg)	66.5± 18	64.95± 14	59.05± 16	69.96± 21	99.9± 20	93.34± 20	67.6± 16	60.9± 14	53.7± 5	58.74± 10	108.2± 2	84.68± 2
HR (beat/min)	327.6± 31	338± 29	340± 27	327.5± 40	321± 17	330.8± 16	321± 45	326± 36	329± 25	323.4± 55	355± 21	345.2± 1
PaO2 (mm Hg)	297± 71	308± 91	272± 53	309± 44	448± 94	299± 50	267± 33	236± 73	208± 49	366± 61	444.6± 6	411± 139
PaCO2 (mm Hg)	43.9± 21	36± 12	32.8± 2	51.84± 2	23.86± 17	42.24± 12	49± 21	42± 12	40± 2	49.22± 14	25.42± 8	38.06± 15
pH	7.4± 0.1	7.4± 0.1	7.32± 0.1	7.32± 0.1	7.6± 0.1	7.47± 0.1	7.28± 0.1	7.36± 0.1	7.3± 0.1	7.39± 0.0	7.5± 0.2	7.44± 0.1
Lactate (mmol/L)	2.7± 1.0	3.2± 1.0	3.3± 0.9	2.2± 0.7	3.0± 0.2	3.0± 0.5	2.13± 0.4	2.32± 0.7	2.34± 0.8	1.6± 0.7	2.5± 0.6	3.3± 0.9

WT - Wild type; FXII^−/− - FXII-deficient.

FXII promotes neutrophil margination/migration into liver tissue in ECMO

Following completion of ECMO, we performed analyses of the oxygenators as it is well established that leukocytes quickly adhere to the oxygenator membrane. The oxygenators were flushed using a non-enzymatic cell dissociation buffer to remove adherent cells. The isolated cells were enumerated, and a portion was analyzed by cytospins stained with Wright Giemsa. We obtained ~10 – 20 million cells from each oxygenator, and the number of adherent cells was similar between genotypes (Fig. 4A). Microscopic review of the cytospins revealed that most of the adherent cells (~95%) were neutrophils, regardless of rat genotype (Fig. 4B).

Figure 4. FXII is not a determinant of early leukocyte adhesion to the oxygenator in ECMO.

A) Data represent the quantitation of total leukocytes adherent to the oxygenators. B) Shown are representative images of cytospins of leukocytes adherent to the oxygenators from WT and FXII^−/− rats harvested at the end of the ECMO run. Note that the majority of the leukocytes (~95%) were neutrophils. Data shown represent mean ± SEM. WT-Wild type, FXII^−/− FXII deficient.

We harvested multiple organs from the argatroban anticoagulated FXII^−/− and WT rats at the end of the ECMO and sham experiments. H&E-stained histological sections of lungs, livers and kidneys were reviewed with a pathologist blinded to animal genotype. Organs from both genotypes subjected to ECMO showed evidence of mild venous congestion, but no evidence of significant organ damage (Supplemental Fig. 2). We immunohistochemically stained tissues for myeloperoxidase (MPO), a marker of neutrophils MPO+ cells in the livers and kidneys were enumerated in 6 – 8 20X fields per tissue section by an investigator blinded to animal genotype. MPO+ pixels in the lungs were quantified using Aperios eSlide software as a function of the total tissue stained. MPO staining cells were few in the liver, lung and kidney tissues of the sham and unchallenged rats and similar between genotypes. ECMO resulted in a significant increase in MPO staining cells throughout the liver sinusoids that was significantly diminished by FXII deficiency (Fig. 5). ECMO also resulted in an increase in MPO+ staining cells in the kidneys and lungs, but in contrast to the liver, deletion of FXII did not have a significant impact on this relative increase (Supplemental Fig. 3).

Figure 5. FXII promotes neutrophil margination/migration into liver tissue in ECMO.

Shown are representative photomicrographs of myeloperoxidase-stained (MPO) liver sections from WT and FXII^−/− rats that were unchallenged or harvested following sham or ECMO experiments. Note that ECMO but not sham surgery resulted in a significant increase in MPO+ staining cells (brown). Deletion of FXII significantly attenuated MPO+ cells (a marker for neutrophils) into the liver. Size bars represent 25 μm. Quantitative data are shown to the right of the photomicrographs. (Data shown represent mean ± SEM. * P < 0.05, ** P < 0.005, *** P < 0.001. All comparisons were analyed, but only statistically significant P values are shown. P values were generated using a one-way Anova, multiple comparisons.) WT-Wild type, FXII^−/− FXII deficient.

FXII is a major determinant of high molecular weight (HMW) kininogen cleavage and complement activation in ECMO.

We measured plasma levels of cytokines known to be altered in ECMO.² ECMO resulted in significant increases in the proinflammatory cytokines IL-6 and TNFα relative to unchallenged rats independently of FXII (Supplemental Fig. 4). IL-6 and TNFα levels tended to higher in ECMO-challenged rats relative to shams, but this did not reach statistical significance. The anti-inflammatory cytokine IL-10 was also increased in ECMO rats relative to unchallenged animals and WT shams with no FXII dependence. These data suggest that FXII is not a significant determinant of the generation of these cytokines in the context of ECMO.

Increased circulating bradykinin, a cleavage product of HMW kininogen, has been reported in humans undergoing cardiopulmonary bypass,¹² but there are limited studies of this peptide in VA-ECMO. To evaluate the impact of ECMO and FXII on HMW kininogen cleavage, a biomarker of bradykinin generation, we measured the intact (HK) and cleaved (cHK) species of HMW kininogen in plasma by Western blot using an antibody that recognizes both (Fig. 6A). The proportion of cHK was relatively low in unchallenged rats and similar between FXII genotypes (Fig. 6B&C). ECMO resulted in a significant decrease in HK and an increase in the ratio of cHK/HK in WT rats relative to unchallenged animals. This ECMO-dependent increase was significantly attenuated in FXII^−/− rats (Fig. 6B&C). There was a trend toward increased cHK in WT rats after sham surgery relative to unchallenged animals, but there was no statistically significant difference between WT rats in the sham and ECMO cohorts. These data indicate that FXII is a major driver of HMW kininogen cleavage in ECMO. They also suggest that the surgical procedure and catheterization may contribute to HMW kininogen cleavage in addition to ECMO in this context.

Figure 6. FXII promotes HMW kininogen cleavage and complement activation in ECMO.

A) Intact (HK) and cleaved (cHK) HMW kininogen were quantitated using a Western blot with an antibody that recognizes both species of the protein. Shown are quantitation of intact HK (B) (expressed as a percentage of HK in WT unchallenged rats) and (C) the percentage of cleaved HK relative to total HK (intact HK and cleaved HK combined). Note that ECMO resulted in a significant increase in HMW kininogen cleavage relative to unchallenged rats that was attenuated by FXII deficiency. Sham surgery did not result in a significant increase in HK cleavage relative to unchallenged rats. Markers of complement activation were measured in plasma from unchallenged rats and rats following 1 hour of ECMO or sham surgery. Shown are analyses of C3 fragments C3b/iC3b/C3c (D) and Terminal Complement Complex (TCC) (E). ECMO resulted in a significant increase in both markers of complement activation that was not observed in animals following sham surgery. FXII deletion completely abrogated the increase in complement activation associated with ECMO. (Data shown represent mean ± SEM. * P < 0.05, ** P < 0.05, *** P < 0.005, **** P < 0.001. All comparisons were analyzed, but only statistically significant P values are shown. P values were generated using a one-way Anova, multiple comparisons). FXII^−/−-FXII deficient, WT-Wild type, HK-Intact high molecular weight kininogen, cHK-cleaved high molecular weight kininogen.

Previous studies have shown that ECMO is associate with rapid complement activation, which is thought to be caused by both the underlying disease processes and by blood contact with the ECMO circuit. We measured circulating markers of complement activation products; Terminal Complement Complex (TCC) and C3 fragments (C3b/iC3b/C3c) in unchallenged rats, and rats underwent the sham and ECMO procedures. Both complement activation markers were similarly low in WT and FXII^−/− rats from the unchallenged and sham cohorts. ECMO resulted in a rapid increase in circulating markers of C3 activation (C3b/iC3b/C3c) and the TCC (Fig. 6D&E), suggesting that these sequelae are predominantly caused by the ECMO circuit and not the surgical procedure itself. Importantly, FXII deficiency essentially abrogated the ECMO-induced increase in C3b/iC3b/C3c and TCC.

DISCUSSION

These studies represent the first to evaluate the role of FXII in thrombotic and inflammatory pathologies related to ECMO. The genetic deletion of FXII resulted in a major prolongation in ECMO circuit life in the absence of pharmacological anticoagulation. Moreover, FXII^−/− rats anticoagulated with argatroban had significantly lower blood flow resistance values relative to WT rats similarly anticoagulated, suggesting that limiting FXII could further enhance ECMO circuit function when combined with traditional anticoagulation approaches. FXII was also shown to be a major factor promoting neutrophil margination/migration into the liver in ECMO. The data also indicate that FXII plays a key role in driving HMW kininogen cleavage and complement activation in ECMO. See Figure 7 for a graphical summary.

Figure 7. Summary of the experimental design and findings of the study.

In part 1 (thrombosis), a minimally invasive venoarterial (VA) extracorporeal membrane oxygenation (ECMO) model was used to compare wild-type (WT) and factor XII (FXII)–deficient (FXII^−/−) Sprague-Dawley rats under conditions of no pharmacologic anticoagulation. WT rats experienced circuit failure within 10 minutes, whereas FXII^−/− rats completed the 1-hour ECMO timeframe. In part 2 (inflammation), the study investigated the impact of FXII on ECMO-induced inflammatory responses by subjecting WT and FXII^−/− rats to ECMO with systemic argatroban administration. Comparison groups included unchallenged rats and rats that underwent a sham surgical procedure without ECMO. The analysis revealed that ECMO led to increased neutrophil migration into the liver, which was attenuated in FXII^−/− rats. Additionally, ECMO-induced high molecular weight kininogen (HK) cleavage and complement activation, both of which were suppressed in rats lacking FXII. CaCl2, cCalcium chloride; ABG, arterial blood gas; ACT, activated clotting time; HMW, high molecular weight.

Our data are consistent with previous studies in rabbit models of ECMO showing that an antibody targeting FXII, a peptide FXII inhibitor, and an mRNA approach that limits hepatic FXII synthesis prolong ECMO circuit lifespan and limit thromboembolism.^7–9 Although the use of protein and heparin covalent bonded circuit components have limited the need for anticoagulation in certain circumstances, thromboembolism remains a major problem during ECMO, particularly in infants and young children where thrombotic complications occur in up to 37.5% of patients.¹³ The antithrombotic efficacy of targeting FXII in ECMO remains to be determined in clinical settings, but given that FXII plays no role in surgical hemostasis, any improvement in antithrombotic efficacy that comes with targeting FXII would carry no bleeding risk. Moreover, our data indicate that targeting FXII could come with the dual benefit of limiting ECMO-associated inflammatory sequelae.

It is notable that all the preclinical studies of FXII in ECMO, including the studies presented here, were carried out in healthy animals. It remains to be determined how effective targeting FXII would be in preventing mechanical circulation associated thromboembolism when there are additional initiators of thrombin generation other than the ECMO circuit. Patients requiring ECMO are often critically ill, and factors such as circulating TF-expressing extracellular vesicles, cell free DNA, and endothelial cell activation are all potential contributors to the prothrombotic state present in these patients.¹⁴ While the mechanical circuit used in cardiopulmonary bypass is similar to an ECMO circuit, the sternotomy and manipulation of cardiac tissue are likely to result in significant release of circulating TF.¹⁵ In these contexts, targeting FXII alone may be insufficient to prevent thromboembolism. Future studies in animal models where inflammatory and/or additional surgical challenges are imposed are needed. The data presented here showing that FXII deficiency together with argatroban lowered blood flow resistance more than argatroban alone suggest that targeting FXII could augment the antithrombotic efficacy of standard anticoagulation approaches. Moreover, the anti-inflammatory mechanisms associated with targeting FXII could also limit thrombotic complications by limiting downstream immune-mediated hemostatic system activation. For example, both bradykinin generation and complement activation are associated with vascular changes known to promote thromboembolism.^{16, 17}

ECMO resulted in a significant increase in neutrophil margination/migration into the liver, lung, and kidneys. FXII significantly attenuated this phenomenon in the liver, but not in the kidneys or lungs. It is unclear why the role of FXII in ECMO-induced accumulation of neutrophils is most prominent in the liver. The mechanisms promoting neutrophil migration are likely tissue specific, with the key receptors promoting neutrophil migration being distinct between different organs.¹⁸ The selectin dependent mechanisms that primarily regulate neutrophil adhesion and migration into the lung and kidney appear to be less important in liver.¹⁸ An intriguing observation that might offer a clue is the finding that hyaluronan is disproportionately expressed in the liver versus other organs under both basal and inflammatory conditions.¹⁸ These studies suggested that hyaluronan promotes neutrophil migration via interaction with CD44. Interestingly, previous studies have also suggested a link between hyaluronan/CD44 interactions and upregulation of uPAR in other contexts.¹⁹ FXII zymogen has been shown to promote neutrophil migration by binding to uPAR.²⁰ Together, these data suggest the intriguing hypothesis that the relatively hyaluronan rich environment of the liver may upregulate neutrophil uPAR expression, making them more sensitive to FXII zymogen activation.

A significant increase in HMW kininogen cleavage, a marker of bradykinin generation, was observed in our rat model of ECMO. ECMO-induced HMW kininogen cleavage was essentially abrogated by deletion of FXII. Clinical studies have shown that cardiopulmonary bypass (CPB) results in increased plasma bradykinin,¹² but this has not been specifically studied in ECMO. Given that the mechanical circuits used in ECMO and CPB are similar, it would seem likely that bradykinin generation is a factor in both. Consistent with the relatively short ECMO timeframe, we did not see any evidence of vascular leak or organ damage in our histological analyses. Previous studies from our group suggested that limiting FXII expression in a 4 hour long rabbit model of ECMO results in less pulmonary edema.⁹ Based on the data presented here, it is conceivable that FXII-mediated bradykinin generation may promote tissue edema in ECMO. Notably, circulation in complete VA-ECMO largely bypasses the pulmonary arterial system, which is responsible for inactivation of the majority of circulating bradykinin.²¹ Whether FXII plays any role in bradykinin clearance remains to be determined. Future studies of FXII-deficient rats in ECMO that span longer timeframes will better elucidate these mechanisms.

Previous clinical studies have shown that ECMO results in rapid complement activation that peaks within 1 – 2 hours.² This was recapitulated in our rat model and was entirely driven by FXII. FXII could drive complement activation in this context through multiple non-mutually exclusive mechanisms. FXII is a multifunctional protease that can proteolytically activate FXI, leading to thrombin generation. FXII also activates prekallikrein to the active protease, kallikrein. Kallikrein has been shown to activate the complement component C3.⁴ FXII has also been shown to interact with the globular C1q receptor and the complement regulators C1INH and FH.^22–24 Recently, FXI was also shown to neutralize the complement regulator, factor H²⁵. Future mechanistic studies can readily be envisioned where genetic or pharmacological targeting of FXI and/or prekallikrein are combined with FXII deficiency to better define the role of these pathways in ECMO related complement activation and other pathologies.

It is notable that ECMO resulted in a significant increase in tissue neutrophil infiltration and complement activation compared to shams. In contrast, proinflammatory cytokine elaboration and cleavage of HK were clearly increased in ECMO, but they were not statically significant compared to the sham group. This suggests that vascular access manipulation and catheter insertion may play a role in these specific inflammatory responses. These data highlight the fact that multiple factors can influence the inflammatory response to ECMO, including surgical tissue injury, circuit design, duration of support, and the catheterization materials and techniques used.

In this study we evaluated the impact of genetic FXII deletion on ECMO-induced inflammatory responses. It remains to be determined if targeting FXII clinically could replicate these findings. As noted, FXII proteolytic functions can activate FXI, cleave HMW kininogen and activate prekallikrein. Additionally, FXII zymogen has been shown to promote neutrophil activation through engagement of uPAR.²⁰ Conceivably, antibodies that clear FXII and mRNA-based approaches limiting FXII production could target all these mechanisms. FXII is made by hepatocytes and neutrophils, it is unclear to what extent emerging antibody therapies and ASO approaches would limit neutrophil-derived FXII. Comparison of these modalities together with genetic deletion of FXII and emerging small molecule inhibitors of FXII proteolytic function could be mechanistically informative.

It should be noted that the small amount of FXII protein detected by ELISA in FXII^−/− rats could represent a truncated FXII protein with no function, or rare readthrough events of the premature stop codon resulting in small amounts of full length FXII. It would seem unlikely that such small amounts of FXII would significantly contribute to thrombin generation. However, it is conceivable that even small amounts of zymogen FXII could significantly impact neutrophil functions. Analyses in rats with a large deletion of the FXII gene rather than a premature stop codon could be informative.

The potential for FXII to limit multiple inflammatory pathways in ECMO raises the question of whether targeting FXII could increase infection risk or impair wound healing. Recent studies suggest that FXII evolved, at least in part, to deal with soil contamination of wounds.²⁶ Here FXII has been proposed to bridge activation of hemostasis and inflammatory pathways, leading to maintenance of vascular integrity and pathogen clearance. Interestingly, studies in mice showed that FXII deletion had no impact on survival following systemic infection with Streptococcus pneumoniae, and FXII deletion improved survival following infection with Klebsiella pneumoniae.²⁷ Furthermore, FXII deletion in mice resulted in impaired migration of neutrophils into skin wounds, but deletion of FXII resulted in more rapid wound closure.²⁰ Taken together, the available data suggest that targeting FXII in ECMO would not impair wound healing or fundamentally impair the response to bacterial infection, but additional preclinical studies and clinical trials are needed.

In summary, the data presented demonstrate, for the first time, that FXII plays a multifaceted role in ECMO, promoting both thrombotic complications that limit circuit life, and deleterious inflammatory events capable of damaging end organs. The current study suggests that pharmacological targeting of FXII in ECMO could have the dual benefit of limiting thromboembolism without incurring any bleeding risk and reducing inflammatory-mediated organ damage.

Supplementary Material

Supplementary Figure1

SUPPLEMENTARY FIGURE 1. Complete blood counts and plasma fibrinogen levels in FXII-deficient rats. The circulating white blood cells count (A), hemoglobin (B), platelet count (C), and plasma fibrinogen levels were similar in FXII-sufficient and -deficient rats. Note that there were no statistically significant differences in any comparisons. Data shown represent mean ± SEM. FXII^−/− FXII deficient, FXII^+/+ FXII sufficient, FXII^+/− FXII Heterozygous mutants.

Supplementary Figure2

SUPPLEMENTARY FIGURE 2. Histological analyses of lung, liver and kidney following ECMO. Shown are representative photo micrographs of lung, liver and kidney tissue harvested from WT and FXII^−/− immediately following sham and ECMO. There was no significant pathology in any of the organs analyzed. There was evidence of modest venous congestion in organs from EMCO challenged rats, but no evidence of organ damage or significant cell death. Size bars represent 50 μm. WT-Wild type, FXII^−/− FXII deficient.

Supplementary Figure3

SUPPLEMENTARY FIGURE 3. Myeloperoxidase staining of lung and kidney. Shown are representative photomicrographs of myeloperoxidase-stained (MPO) tissue sections of kidney and lung harvested from unchallenged WT and FXII^−/− rats as well as rats following sham or ECMO experiments. ECMO resulted in a significant increase in MPO+ staining cells (brown staining) in both organs relative to unchallenged and sham surgical rats, but there was no FXII genotype dependence. Size bars represent 25 μm. Quantitative data are shown to the right of the photomicrographs. (Data shown represent mean ± SEM. * P < 0.05, ** P < 0.005. P values were generated using a one-way Anova, multiple comparisons. All comparisons were analyzed, but only statistically significant P values are shown.) WT-Wild type, FXII^−/− FXII deficient.

Supplementary Figure4

SUPPLEMENTARY FIGURE 4. Plasma cytokine measurements. Shown are quantitations of plasma cytokines in unchallenged WT and FXII^−/− rats as well as rats following 1 hour sham or ECMO experiments. There were no genotype dependent differences in cytokine levels. (Data shown represent mean ± SEM. * P < 0.05, ** P < 0.005. P values were generated using a one-way Anova, multiple comparisons. All comparisons were analyzed, but only statistically significant P values are shown.) WT-Wild type, FXII^−/− FXII deficient.

Central Picture.

Central Message:

The present study suggests that targeting FXII in ECMO could have the dual benefit of limiting thromboembolism, and inflammatory-mediated organ damage.

Perspective Statement:

Previous studies have linked factor XII (FXII) to ECMO-induced thrombosis. A model using novel gene-targeted FXII deficient rats on VA-ECMO demonstrated FXII drives thrombosis, supports neutrophil migration into liver, and triggers HMW kininogen cleavage and complement activation Future studies are needed to better elucidate these mechanisms and reveal additional FXII-dependent pathologies.

Glossary of Abbreviations:

CPB

Cardiopulmonary bypass

ECMO

Extracorporeal membrane oxygenation

MPO

Myeloperoxidase

uPAR

Urokinase plasminogen activator receptor

cHK

Cleaved high molecular weight kininogen